ISSN 1905-7873 © 2012 - Maejo International Journal of Science ...
ISSN 1905-7873 © 2012 - Maejo International Journal of Science ...
ISSN 1905-7873 © 2012 - Maejo International Journal of Science ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Journal</strong> Information<br />
<strong>Maejo</strong> <strong>International</strong> <strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology (<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong> © <strong>2012</strong>),<br />
the international journal for preliminary communications in <strong>Science</strong> and Technology is the<br />
first peerrefereed scientific journal <strong>of</strong> <strong>Maejo</strong> University (www.mju.ac.th). Intended as a<br />
medium for communication, discussion, and rapid dissemination <strong>of</strong> important issues in<br />
<strong>Science</strong> and Technology, articles are initially published online in an open access format,<br />
which thereby gives authors the chance to communicate with a wide range <strong>of</strong> readers in an<br />
international community.<br />
Publication Information<br />
MIJST is published triannually. Articles are available online and can be accessed free<br />
<strong>of</strong> charge at http://www.mijst.mju.ac.th. Printed and bound copies <strong>of</strong> each volume are<br />
produced and distributed to authors and selected groups or individuals. This journal and the<br />
individual contributions contained in it are protected under the copyright by <strong>Maejo</strong><br />
University.<br />
Abstracting/Indexing Information<br />
MIJST is covered and cited by <strong>Science</strong> Citation Index Expanded, SCOPUS, <strong>Journal</strong><br />
Citation Reports/<strong>Science</strong> Edition, Zoological Record, Chemical Abstracts Service (CAS),<br />
SciFinder Scholar, Directory <strong>of</strong> Open Access <strong>Journal</strong>s (DOAJ), CAB Abstracts, ProQuest,<br />
and Google Scholar.<br />
Contact Information<br />
Editorial <strong>of</strong>fice: <strong>Maejo</strong> <strong>International</strong> <strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology (MIJST), 1st<br />
floor, Orchid Building, <strong>Maejo</strong> University, San Sai, Chiang Mai 50290, Thailand<br />
Tel: +66-53-87-3880<br />
E-mails: duang@mju.ac.th
MAEJO INTERNATIONAL JOURNAL<br />
OF SCIENCE AND TECHNOLOGY<br />
Editor<br />
Duang Buddhasukh, <strong>Maejo</strong> University, Thailand.<br />
Associate Editors<br />
Jatuphong Varith, <strong>Maejo</strong> University, Thailand.<br />
Wasin Charerntantanakul, <strong>Maejo</strong> University, Thailand.<br />
Morakot Sukchotiratana, Chiang Mai University, Thailand.<br />
Nakorn Tippayawong, Chiang Mai University, Thailand.<br />
Editorial Assistants<br />
James F. Maxwell, Chiang Mai University, Thailand.<br />
Jirawan Banditpuritat, <strong>Maejo</strong> University, Thailand.<br />
Pr<strong>of</strong>. Dallas E. Alston<br />
Dr. Pei-Yi Chu<br />
Asst. Pr<strong>of</strong>. Ekachai Chukeatirote<br />
Pr<strong>of</strong>. Richard L. Deming<br />
Pr<strong>of</strong>. Cynthia C. Divina<br />
Pr<strong>of</strong>. Mary Garson<br />
Pr<strong>of</strong>. Kate Grudpan<br />
Assoc. Pr<strong>of</strong>. Duangrat Inthorn<br />
Pr<strong>of</strong>. Minoru Isobe<br />
Pr<strong>of</strong>. Kunimitsu Kaya<br />
Assoc. Pr<strong>of</strong>. Margaret E. Kerr<br />
Dr. Ignacy Kitowski<br />
Asst. Pr<strong>of</strong>. Andrzej Komosa<br />
Asst. Pr<strong>of</strong>. Pradeep Kumar<br />
Asst. Pr<strong>of</strong>. Ma. Elizabeth C. Leoveras<br />
Dr. Subhash C. Mandal<br />
Pr<strong>of</strong>. Amarendra N. Misra<br />
Dr. Robert Molloy<br />
Pr<strong>of</strong>. Mohammad A. Mottaleb<br />
Pr<strong>of</strong>. Stephen G. Pyne<br />
Pr<strong>of</strong>. Renato G. Reyes<br />
Dr. Waya Sengpracha<br />
Dr. Settha Siripin<br />
Pr<strong>of</strong>. Paisarn Sithigorngul<br />
Pr<strong>of</strong>. Anupam Srivastav<br />
Pr<strong>of</strong>. Maitree Suttajit<br />
Assoc. Pr<strong>of</strong>. Chatchai Tayapiwatana<br />
Emeritus Pr<strong>of</strong>. Bela Ternai<br />
Asst. Pr<strong>of</strong>. Narin Tongwittaya<br />
Asst. Pr<strong>of</strong>. Jatuphong Varith<br />
Assoc. Pr<strong>of</strong>. Niwoot Whangchai<br />
Editorial Board<br />
University <strong>of</strong> Puerto Rico, USA.<br />
Changhua Christian Hospital, Taiwan, R.O.C.<br />
Mae Fah Luang University, Thailand.<br />
California State University Fullerton, Fullerton CA<br />
Central Luzon State University, Philippines.<br />
The University <strong>of</strong> Queensland, Australia.<br />
Chiang Mai University, Thailand.<br />
Mahidol University, Thailand.<br />
Nagoya University, Japan.<br />
Tohoku University, Japan.<br />
Worcester State College, Worcester,MA<br />
University <strong>of</strong> Maria-Curie Sklodowska, Poland.<br />
University <strong>of</strong> Maria-Curie Sklodowska, Poland.<br />
Jaypee University <strong>of</strong> Information Technology, India.<br />
Central Luzon State University, Philippines.<br />
Jadavpur University, India.<br />
Fakir Mohan University, Orissa, India.<br />
Chiang Mai University, Thailand.<br />
Northwest Missouri State University, USA.<br />
University <strong>of</strong> Wollongong, Australia.<br />
Central Luzon State University, Philippines.<br />
Silpakorn University, Thailand.<br />
<strong>Maejo</strong> University, Thailand.<br />
Srinakharinwirot University, Thailand.<br />
College <strong>of</strong> Engineering and Technology, India.<br />
Naresuan University (Payao Campus), Thailand.<br />
Chiang Mai University, Thailand.<br />
La Trobe University, Australia.<br />
<strong>Maejo</strong> University, Thailand.<br />
<strong>Maejo</strong> University, Thailand.<br />
<strong>Maejo</strong> University,Thailand.<br />
Consultants<br />
Asst. Pr<strong>of</strong>. Chamnian Yosraj, Ph.D., President <strong>of</strong> <strong>Maejo</strong> University<br />
Assoc. Pr<strong>of</strong>. Thep Phongparnich, Ed. D., Former President <strong>of</strong> <strong>Maejo</strong> University<br />
Assoc. Pr<strong>of</strong>. Chalermchai Panyadee, Ph.D., Vice-President in Research <strong>of</strong> <strong>Maejo</strong> University
MAEJO INTERNATIONAL JOURNAL<br />
OF SCIENCE AND TECHNOLOGY<br />
The <strong>International</strong> <strong>Journal</strong> for the Publication <strong>of</strong> Preliminary<br />
Communications in <strong>Science</strong> and Technology<br />
http://www.mijst.mju.ac.th<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong> © <strong>2012</strong> by <strong>Maejo</strong> University
MAEJO INTERNATIONAL JOURNAL<br />
OF SCIENCE AND TECHNOLOGY<br />
Volume 6, Issue 1 (January - April <strong>2012</strong>)<br />
CONTENTS<br />
Page<br />
Editor's Note<br />
Duang Buddhasukh, Editor-in-Chief .................…………………….……………………... i<br />
Enhancement <strong>of</strong> carotenoid and chlorophyll content <strong>of</strong> an edible freshwater alga (Kai:<br />
Cladophora sp.) by supplementary inorganic phosphate and investigation <strong>of</strong> its biomass<br />
production<br />
Taweesak Khuantrairong and Siripen Traichaiyaporn*………...……….…………….… 1-11<br />
Performance <strong>of</strong> self-excited induction generator with costeffective static compensator<br />
Yogesh K. Chauhan *, Sanjay K. Jain and Bhim Singh…..…………………………….. 12-27<br />
Benthic diatoms <strong>of</strong> Mekong River and its tributaries in northern and north-eastern Thailand<br />
and their application to water quality monitoring<br />
Sutthawan Suphan*, Yuwadee Peerapornpisal and Graham J. C. Underwood…………28-46<br />
Fourteen new records <strong>of</strong> cercosporoids from Thailand<br />
Pheng Phengsintham, Ekachai Chukeatirote, Eric H. C. McKenzie, Mohamed A. Moslem,<br />
Kevin D. Hyde * and Uwe Braun..………………………………………………..………47-61<br />
On the security <strong>of</strong> an anonymous roaming protocol in UMTS mobile networks<br />
Shuhua Wu *, Qiong Pu and Ji Fu………………..……………………………….……. 62-69<br />
Influence <strong>of</strong> MeV H+ ion beam flux on cross-linking and blister formation in PMMA resist<br />
Somrit Unai *, Nitipon Puttaraksa, Nirut Pussadee, Kanda Singkarat, Michael W. Rhodes,<br />
Harry J. Whitlow and Somsorn Singkarat……………………….……………...…….… 70-76<br />
Degradation <strong>of</strong> bisphenol A by ozonation: rate constants, influence <strong>of</strong> inorganic anions, and<br />
by-products<br />
Kheng Soo Tay *, Noorsaadah Abd. Rahman and Mhd. Radzi Bin Abas….………….… 77-94<br />
Chitinase production and antifungal potential <strong>of</strong> endophytic Streptomyces strain P4<br />
Julaluk Tang-um and Hataichanoke Niamsup*………………………………………... 95-104<br />
Water quality variation and algal succession in commercial hybrid catfish production ponds<br />
Chatree Wirasith and Siripen Traichaiyaporn*...………….….……………………….105-118<br />
Determination <strong>of</strong> production-shipment policy using a two-phase algebraic approach<br />
Yuan-Shyi Peter Chiu, Hong-Dar Lin and Huei-Hsin Chang *……..…………….… 119-129<br />
IIS-Mine: A new efficient method for mining frequent itemsets<br />
Supatra Sahaphong* and Veera Boonjing……………………………..……………... 130-151<br />
Lipase-catalysed sequential kinetic resolution <strong>of</strong> -lipoic acid<br />
Hong De Yan, Yin Jun Zhang, Li Jing Shen and Zhao Wang*…................................. 152-158<br />
http://www.mijst.mju.ac.th<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong> © <strong>2012</strong> by <strong>Maejo</strong> University
MAEJO INTERNATIONAL JOURNAL<br />
OF SCIENCE AND TECHNOLOGY<br />
Volume 6, Issue 1 (January - April <strong>2012</strong>)<br />
Author Index<br />
Author<br />
Page<br />
Abas Mhd. R. B. 77<br />
Boonjing V. 130<br />
Braun U. 47<br />
Chang H.-H. 119<br />
Chauhan Y. K. 12<br />
Chiu Y.-S. P. 119<br />
Chukeatirote E. 47<br />
Fu J. 62<br />
Hyde K. D. 47<br />
Jain S. K. 12<br />
Khuantrairong T. 1<br />
Lin H.-D. 119<br />
McKenzie E. H. C. 47<br />
Moslem M. A. 47<br />
Niamsup H. 95<br />
Peerapornpisal Y. 28<br />
Phengsintham P. 47<br />
Pu Q. 62<br />
Pussadee N. 70<br />
Puttaraksa N. 70<br />
Rahman N. Abd. 77<br />
Author<br />
Page<br />
Rhodes M. W. 70<br />
Sahaphong S. 130<br />
Shen L. J. 152<br />
Singh B. 12<br />
Singkarat K. 70<br />
Singkarat S. 70<br />
Suphan S. 28<br />
Tang-um J. 95<br />
Tay K. S. 77<br />
Traichaiyaporn S. 1<br />
Traichaiyaporn S. 105<br />
Unai S. 70<br />
Underwood G. J. C. 28<br />
Wang Z. 152<br />
Whitlow H. J. 70<br />
Wirasith C. 105<br />
Wu S. 62<br />
Yan H. De 152<br />
Zhang Y. J. 152
Instructions for Authors<br />
A proper introductory e-mail page containing the title <strong>of</strong> the submitted article and certifying<br />
its originality should be sent to the editor (Duang Buddhasukh, e-mail : duang@mju.ac.th). The<br />
manuscript proper together with a list <strong>of</strong> suggested referees should be attached in separate files. The<br />
list should contain at least 5 referees with appropriate expertise. Three referees should be non-native<br />
from 3 different countries. Each referee's academic/pr<strong>of</strong>essional position, scientific expertise,<br />
affiliation and e-mail address must be given. The referees should not be affiliated to the same<br />
university/institution as any <strong>of</strong> the authors, nor should any two referees come from the same<br />
university/institution. The editorial team, however, retain the sole right to decide whether or not the<br />
suggested referees are approached.<br />
Failure to conform to the above instructions will result in non-consideration <strong>of</strong> the<br />
submission.<br />
Please also ensure that English and style is properly edited before submission. UK style <strong>of</strong><br />
spelling should be used. Authors who would like to consult a pr<strong>of</strong>essional service can visit<br />
www.pro<strong>of</strong>-reading-service.com, www.editage.com, www.enago.com, www.bioedit.co.uk<br />
(bioscience and medical papers), www.bioscienceeditingsolutions.com, www.scribendi.com,<br />
www.letpub.com, www.papersconsulting.com or www.sticklerediting.com.<br />
Important : Manuscript with substandard English and style will not be considered.<br />
Warning : Plagiarism (including self-plagiarism) may be checked for at the last stage <strong>of</strong> processing<br />
and, if detected, will result in a rejection and blacklisting.<br />
Manuscript Preparation<br />
Manuscripts must be prepared in English using a word processor. MS Word for Macintosh or<br />
Windows, and .doc or .rtf files are preferred. Manuscripts may be prepared with other s<strong>of</strong>tware<br />
provided that the full document (with figures, schemes and tables inserted into the text) is exported to<br />
a MS Word format for submission. Times or Times New Roman font is preferred. The font size<br />
should be 12 pt and the line spacing 'at least 17 pt'. A4 paper size is used and margins must be 1.5 cm<br />
on top, 2.0 cm at the bottom and 2.0 cm on both left and right sides <strong>of</strong> the paper. Although our final<br />
output is in .pdf format, authors are asked NOT to send manuscripts in this format as editing them is<br />
much more complicated. Under the above settings, a manuscript submitted should not be longer than<br />
15 pages for a full paper or 20 pages for a review paper.<br />
A template file may be downloaded from the <strong>Maejo</strong> Int. J. Sci. Technol. homepage.<br />
(DOWNLOAD HERE)<br />
Authors' full mailing addresses, homepage addresses, phone and fax numbers, and e-mail<br />
addresses homepages can be included in the title page and these will be published in the manuscripts<br />
and the Table <strong>of</strong> Contents. The corresponding author should be clearly identified. It is the<br />
corresponding author's responsibility to ensure that all co-authors are aware <strong>of</strong> and approve <strong>of</strong> the<br />
contents <strong>of</strong> a submitted manuscript.<br />
A brief (200 word maximum) Abstract should be provided. The use in the Abstract <strong>of</strong><br />
numbers to identify compounds should be avoided unless these compounds are also identified by<br />
names.<br />
A list <strong>of</strong> three to five keywords must be given and placed after the Abstract. Keywords may<br />
be single words or very short phrases.
Although variations in accord with contents <strong>of</strong> a manuscript are permissible, in general all papers<br />
should have the following sections: Introduction, Materials and Methods, Results and Discussion,<br />
Conclusions, Acknowledgments (if applicable) and References.<br />
Authors are encouraged to prepare Figures and Schemes in colour. Full colour graphics will be<br />
published free <strong>of</strong> charge.<br />
Tables and Figures should be inserted into the main text, and numbers and titles supplied for<br />
all Tables and Figures. All table columns should have an explanatory heading. To facilitate layout <strong>of</strong><br />
large tables, smaller fonts may be used, but in no case should these be less than 10 pt in size. Authors<br />
should use the Table option <strong>of</strong> MS Word to create tables, rather than tabs, as tab-delimited columns<br />
are <strong>of</strong>ten difficult to format in .pdf for final output.<br />
Figures, tables and schemes should also be placed in numerical order in the appropriate place within<br />
the main text. Numbers, titles and legends should be provided for all tables, schemes and figures.<br />
Chemical structures and reaction schemes should be drawn using an appropriate s<strong>of</strong>tware package<br />
designed for this purpose. As a guideline, these should be drawn to a scale such that all the details and<br />
text are clearly legible when placed in the manuscript (i.e. text should be no smaller than 8-9 pt).<br />
For bibliographic citations, the reference numbers should be placed in square brackets, i.e. [ ],<br />
and placed before the punctuation, for example [4] or [1-3], and all the references should be listed<br />
separately and as the last section at the end <strong>of</strong> the manuscript.<br />
Format for References<br />
<strong>Journal</strong> :<br />
1. D. Buddhasukh, J. R. Cannon, B. W. Metcalf and A. J. Power, “Synthesis <strong>of</strong> 5-n-alkylresorcinol<br />
dimethyl ethers and related compounds via substituted thiophens”, Aust. J. Chem., 1971, 24, 2655-<br />
2664.<br />
Text :<br />
2. A. I. Vogel, “A Textbook <strong>of</strong> Practical Organic Chemistry”, 3rd Edn., Longmans, London, 1956, pp.<br />
130-132.<br />
Chapter in an edited text :<br />
3. W. Leistritz, “Methods <strong>of</strong> bacterial reduction in spices”, in “Spices: Flavor Chemistry and<br />
Antioxidant Porperties” (Ed. S. J. Risch and C-T. Ito), American Chemical Society, Washington, DC,<br />
1997, Ch. 2.<br />
Thesis / Dissertation :<br />
4. W. phutdhawong, “Isolation <strong>of</strong> glycosides by electrolytic decolourisation and synthesis <strong>of</strong><br />
pentinomycin”, PhD. Thesis, 2002, Chiang Mai University, Thailand.<br />
Patent :<br />
5. K. Miwa, S. Maeda and Y. Murata, “Purification <strong>of</strong> stevioside by electrolysis”, Jpn. Kokai Tokkyo<br />
Koho 79 89,066 (1979).<br />
Proceedings :<br />
6. P. M. Sears, J. Peele, M. Lassauzet and P. Blackburn, “Use <strong>of</strong> antimicrobial proteins in the<br />
treatment <strong>of</strong> bovine mastitis”, Proceedings <strong>of</strong> the 3rd <strong>International</strong> Mastitis Seminars, 1995, Tel-Aviv,<br />
Israel, pp. 17-18.<br />
(Download Submission Checklist)<br />
Manuscript Revision Time<br />
Authors who are instructed to revise their manuscript should do so within 45 days. Otherwise<br />
the revised manuscript will be regarded as a new submission.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), i<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Editor’s Note<br />
The upkeep <strong>of</strong> the standard <strong>of</strong> a journal is a major and time-consuming job. At <strong>Maejo</strong>,<br />
however, we are willing to go to the extremes to do that kind <strong>of</strong> job. In trying to maintain the<br />
standards <strong>of</strong> an international journal, we subject every submitted article to a series <strong>of</strong> screening that<br />
may sometimes be inconvenient or plain unpleasant to a fair number <strong>of</strong> authors, to whom we<br />
sincerely apologise. As a matter <strong>of</strong> fact, some <strong>of</strong> their complaints have had a role in shaping several<br />
<strong>of</strong> our journal’s policies which will better serve future contributors. In so far as the language and<br />
academic standards <strong>of</strong> an article are not compromised, we thank them for their candid comments<br />
and always welcome more contributions from them.<br />
Duang Buddhasukh<br />
Editor-in-Chief
Communication<br />
<strong>Maejo</strong> Int. <strong>Maejo</strong> J. Sci. Int. Technol. J. Sci. Technol. <strong>2012</strong>, 6(01), <strong>2012</strong>, 1-11 6(01), 1-111<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Enhancement <strong>of</strong> carotenoid and chlorophyll content <strong>of</strong> an<br />
edible freshwater alga (Kai: Cladophora sp.) by supplementary<br />
inorganic phosphate and investigation <strong>of</strong> its biomass<br />
production<br />
Taweesak Khuantrairong and Siripen Traichaiyaporn*<br />
Algae and Water Quality Research Unit, Department <strong>of</strong> Biology, Faculty <strong>of</strong> <strong>Science</strong>, Chiang Mai<br />
University, Chiang Mai, 50200 Thailand<br />
* Corresponding author, e-mail: tsiripen@yahoo.com; tel +66 (0)53 943 346; fax +66 (0)53 852 259<br />
Received: 10 March 2011 / Accepted: 1 January <strong>2012</strong> / Published: 12 January <strong>2012</strong><br />
Abstract: Enhancement <strong>of</strong> the carotenoid and chlorophyll content <strong>of</strong> an edible freshwater alga (Kai:<br />
Cladophora sp.) by supplementary inorganic phosphate in canteen wastewater was investigated. The<br />
mass cultivation <strong>of</strong> the alga was conducted at ambient temperature and light intensity in diluted<br />
canteen wastewater enhanced with dipotassium hydrogen phosphate at 5-20 mg L -1 . An increase in<br />
total carotenoid and chlorophyll was observed. However, it had no effect on biomass production. As<br />
a result <strong>of</strong> the algal cultivation, there was a dramatic improvement in the quality <strong>of</strong> canteen wastewater.<br />
Keywords: freshwater alga, Cladophora, Kai<br />
________________________________________________________________________________<br />
INTRODUCTION<br />
Algae are a significant source <strong>of</strong> human food, especially in Asia. In Thailand an edible<br />
freshwater alga, Cladophora (known locally as Kai), consists <strong>of</strong> two species, namely C. glomerata<br />
and a to-be-identified Cladophora sp. [1]. Kai is abundant in Nan and Mekong Rivers in the northern<br />
part <strong>of</strong> Thailand. The local people around these rivers collect it for domestic consumption and it is<br />
sold in the local markets. Many reports on the nutritional value <strong>of</strong> Cladophora spp. show that they<br />
contain a significant amount <strong>of</strong> carotenoids that are nutritionally essential for humans and animals [1-<br />
5]. Carotenoids and chlorophylls, which are found in plants and algae, are extremely important in<br />
photosynthesis and growth [6-8]. They are also powerful antioxidants that are beneficial to human
2 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 1-11<br />
health and are now used for supplementary food and animal feed. Previous studies have suggested<br />
that they can prevent or delay cancer and degenerative diseases in humans and animals by<br />
contributing to antioxidative defenses against metabolic oxidative by-products [9-11].<br />
Mass cultivation <strong>of</strong> algae in canteen wastewater for use as feed for Mekong giant catfish [2]<br />
and Tuptim tilapia [12] had been done. It was found that the algal biomass production strongly<br />
correlated with water quality and environmental factors, e.g. temperature, light intensity and nutrient<br />
concentration, results which were similarly obtained elsewhere [13-15]. However, several researches<br />
have indicated that phosphorus is a main factor related to growth and production <strong>of</strong> Cladophora<br />
[13-24]. Other researches have also shown that phosphorus is effective for algal carotenoid and<br />
chlorophyll production [25-30]. In contrast, stimulation <strong>of</strong> carotenoid and chlorophyll production<br />
with phosphorus starvation has been reported [31-35].<br />
Culturing <strong>of</strong> algae in wastewater can also improve the water quality by decreaseing BOD,<br />
COD and nutrients <strong>of</strong> the wastewater [2,12,36]. In this research, mass cultivation <strong>of</strong> Cladophora sp.<br />
(Kai) in canteen wastewater with different phosphate concentrations was carried out to investigate<br />
the effects <strong>of</strong> supplementary phosphorus on carotenoid and chlorophyll content <strong>of</strong> this alga, its<br />
biomass production and physico-chemical characteristics <strong>of</strong> the culture water.<br />
MATERIALS AND METHODS<br />
Culture Media Preparation<br />
One thousand litres <strong>of</strong> canteen wastewater were collected from the canteen wastewater<br />
clarifier <strong>of</strong> <strong>Maejo</strong> University and left to settle in an open cement pond for 3 weeks to allow<br />
microorganisms to break down solid organic waste. The wastewater was then filtered through an 80-<br />
m plankton net filter [36]. The filtrate was diluted to 10% and analysed for pH, dissolved oxygen<br />
(DO), biochemical oxygen demand (BOD), total hardness, ammonia-nitrogen (NH 3 -N), nitratenitrogen<br />
(NO 3 - -N), nitrite-nitrogen (NO 2 - -N), total Kjeldal nitrogen (TKN) and orthophosphatephosphorus<br />
(PO 4 3- -P) by following the methods <strong>of</strong> APHA et al [37].<br />
Culture Condition<br />
Cladophora sp. (Kai) was obtained from the Algae and Water Quality Research Unit, Chiang<br />
Mai University. The alga was cultured for 12 weeks at ambient air temperature (21-28 o C) and light<br />
intensity (12,333-39,267 lux) by attachment on plastic nets (60 g m -2 ) in cement raceway ponds (size<br />
1.2×2.3×0.5 m) each containing diluted canteen wastewater 20 cm deep (552 L per experiment) with<br />
continuous pump-driven circulation <strong>of</strong> 0.15 m s -1 [2,4]. A complete randomised design (CRD) was<br />
carried out with addition <strong>of</strong> dipotassium hydrogen orthophosphate (K 2 HPO 4 ) at concentrations <strong>of</strong> 5,<br />
10, 15 and 20 mg L -1 (treatment 1, 2, 3 and 4 respectively) with diluted canteen wastewater without<br />
K 2 HPO 4 as control. The experiment was done in triplicate.<br />
Carotenoid and Chlorophyll Analysis<br />
The alga at 12 weeks <strong>of</strong> culturing was harvested, washed with tap water, air dried and freezedried.<br />
The total carotenoid content was determined according to the method <strong>of</strong> Britton [38]. The<br />
freeze-dried sample (0.4 g) was homogenised with 20 mL <strong>of</strong> 95% ethanol in an extraction tube. Two
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 1-11<br />
3<br />
mL <strong>of</strong> 5% KOH were then added, air was driven out <strong>of</strong> the tube with nitrogen gas and the tube was<br />
stored in the dark for at least 2 hours. Ether (3×5 ml) was added to extract the carotenoid portion.<br />
The ether supernatant was separated and its absorbance was read at 400-700 nm with a<br />
spectrophotometer. The total amount <strong>of</strong> carotenoid was calculated according to the following<br />
equation: total carotenoid = [(A max /0.25)×ether supernatant volume]/sample weight, where A max =<br />
maximum absorbance.<br />
Extraction <strong>of</strong> carotenoid components and chlorophylls was performed by a modified method<br />
<strong>of</strong> Yoshii et al. [3] and Dere et al. [39] as follows: the freeze-dried sample (0.5 g) was mixed with<br />
ice-cooled acetone (25 mL) in the dark, the mixture stored in the dark at -20 o C for 18 hours and the<br />
supernatant filtered. Carotene, xanthophyll and chlorophylls a and b (μg g -1 ) were determined<br />
spectrophotometrically at 470, 645 and 662 nm respectively by means <strong>of</strong> equations proposed by<br />
Lichtenthaler and Wellburn [40]:<br />
Chlorophyll a = 11.75A 662 – 2.35A 645<br />
Chlorophyll b = 18.61A 645 – 3.960A 662<br />
Carotene = (1000A 470 – 2.270C a – 81.4 C b ) / 227 (C a = chlorophyll a, C b = chlorophyll b)<br />
Xanthophyll = total carotenoid _ carotene<br />
Biomass Production<br />
The growth in terms <strong>of</strong> biomass (g m -2 wet weight) and specific growth rate (SGR) (% d −1 )<br />
were measured every week following the method <strong>of</strong> Premila and Rao [41]. The SGR was calculated<br />
as SGR (% d −1 ) = {[ln(m 1 /m 0 )]/t}100, where m 0 = initial weight, m 1 = final weight, and t = time <strong>of</strong><br />
culture in days.<br />
Water Quality Analysis<br />
Water samples from all the experimental ponds were collected once a week and analysed [37]<br />
for the physico-chemical properties, viz. temperature, pH, DO, BOD, total hardness, NH 3 -N, NO 3 - -<br />
N, NO 2 - -N, TKN and PO 4 3- -P.<br />
Statistical Analysis<br />
The data were presented as mean value ± standard deviation. Comparison <strong>of</strong> mean values was<br />
made by one-way analysis <strong>of</strong> variance (ANOVA), followed by Duncan's multiple range test (DMRT)<br />
at a significance level <strong>of</strong> p
4 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 1-11<br />
Table 1. Carotenoid and chlorophyll content <strong>of</strong> Cladophora sp. (Kai) cultured at different phosphate<br />
concentrations<br />
Amount <strong>of</strong> pigments (g g -1 dry weight)<br />
Pigment<br />
Control Treatment 1 Treatment 2 Treatment 3 Treatment 4<br />
K 2HPO 4(mgL -1 )= 0 K 2HPO 4(mgL -1 )= 5 K 2HPO 4(mgL -1 )= 10 K 2HPO 4(mgL -1 )= 15 K 2HPO 4(mgL -1 )= 20<br />
Total carotenoid 889±157 a 1,084±253 a 1,130±147 a 1,151±151 a 1,729±212 b<br />
Carotene 44±4 a 86±9 b 87±6 b 96±7 bc 103±9 c<br />
Xanthophyll 779±42 a 997±254 a 1043±145 a 1055±154 a 1626±220 b<br />
Chlorophyll a 148±13 a 270±30 b 309±20 bc 305±52 bc 348±17 c<br />
Chlorophyll b 56±31 a 173±11 b 245±15 bc 200±49 c 249±16 c<br />
Note: Each value is mean ± SD<br />
Data in the same row with different superscripts are significantly different (p0.05).<br />
Carotenoids in green algae are produced in two different compartments and by two different<br />
pathways, i.e. the acetate-mevalonate pathway and the phosphoglyceraldehyde-pyruvate pathway,<br />
and they are further synthesised from isopentenyl diphosphate and its isomers [42-43]. In the present<br />
study it was observed that phosphate increased carotenoid and chlorophyll production in the alga,<br />
which agrees with the findings by Khuantrairong and Traichaiyaporn [4-5] that total carotenoid, β-<br />
carotene, lutein and zeaxanthin in Cladophora sp. positively correlate with phosphate level. Celekli<br />
et al. [29] reported that the phosphate supply increased biomass and carotenoid production in a bluegreen<br />
alga (Spirulina platensis) while Latasa and Berdalet [25] deserved that the synthesis <strong>of</strong><br />
pigments in a din<strong>of</strong>lagellate Heterocapsa sp. stopped upon phosphorus limitation. In contrast,<br />
Buapet et al. [30] suggested that phosphorus enrichment had no significant effect on chlorophyll a<br />
production by Ulva reticulate seaweed. Subudhi and Singh [35] reported that a high concentration<br />
<strong>of</strong> phosphate-phosphorus reduced the chlorophyll content <strong>of</strong> Azolla pinnata blue-green alga.<br />
Enhancement <strong>of</strong> astaxanthin in a green alga (Haematococcus pluvialis) [31,34] and that <strong>of</strong> β-<br />
carotene, zeaxanthin and violaxanthin in a marine microalga (Nannochloropsis gaditana) [33] were<br />
observed upon phosphorus limitation. Leonardos and Geider [32] stated that the nitrate-tophosphate<br />
supply ratio was related to carotenoid and chlorophyll-a production in the cryptophyte<br />
Rhinomonas reticulata.<br />
Biomass Production<br />
For 12-week cultures, the biomass production was similar among the treatments and ranged<br />
between 3,406-3,464 g m -2 (wet weight) (Figure 1). The highest value, 4,167 g m -2 , was observed in<br />
treatment 3 after culturing for 10 weeks. However, statistical analysis indicated that Cladophora sp.<br />
(Kai) cultured at different phosphorus concentrations showed no significant difference in growth<br />
(p>0.05).<br />
The specific growth rate (SGR) was similar among treatments and ranged between -2.7 _<br />
38.9% d -1 . After one week <strong>of</strong> cultivation, the alga began to grow quickly and the highest SGR values
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 1-11<br />
5<br />
were observed (Figure 2). Statistical analysis showed that the biomass production rate was not<br />
significantly different among the treatments (p>0.05).<br />
Figure 1. Biomass production <strong>of</strong> Cladophora sp. (Kai) cultured at different phosphorus<br />
concentrations (C = control group, T1 = treatment 1 (+5 mg L -1 K 2 HPO 4 ), T2 = treatment 2 (+10<br />
mg L -1 K 2 HPO 4 ), T3 = treatment 3 (+15 mg L -1 K 2 HPO 4 ), and T4 = treatment 4 (+20 mg L -1<br />
K 2 HPO 4 ))<br />
Figure 2. Specific growth rate (SGR) <strong>of</strong> Cladophora sp. (Kai) cultured at different phosphorus<br />
concentrations (C = control group, T1 = treatment 1 (+5 mg L -1 K 2 HPO 4 ), T2 = treatment 2 (+10<br />
mg L -1 K 2 HPO 4 ), T3 = treatment 3 (+15 mg L -1 K 2 HPO 4 ), and T4 = treatment 4 (+20 mg L -1<br />
K 2 HPO 4 ))
6 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 1-11<br />
Thus, the biomass production and SGR <strong>of</strong> the alga cultured at different phosphorus<br />
concentrations showed no difference among all the treatments, indicating that the added phosphate<br />
had no effect on biomass production. Phosphorus is one <strong>of</strong> the main factors related to growth and<br />
production <strong>of</strong> freshwater Cladophora [13-24]. However, although Auer and Canale [24] observed<br />
that high dissolved phosphorus values in water induced an increase in stored phosphorus and growth<br />
rate <strong>of</strong> Cladophora, our results agreed with the findings that when phosphorus is above a certain<br />
concentration (0.01 mg L -1 ), it has no effect on the growth and biomass production <strong>of</strong> Cladophora<br />
[13,18,44]. The biomass production <strong>of</strong> Cladophora also strongly depends on environmental<br />
conditions, especially temperature and light intensity. High biomass production was observed in<br />
winter season and under high light intensity [2, 5, 21].<br />
Water Quality<br />
For 12-week cultures, the physico-chemical properties <strong>of</strong> water <strong>of</strong> all treatments ranged as<br />
follows: temperature 19-28 o C, pH 8.3-8.9, DO 8.47-11.75 mg L -1 , BOD 1.00-10.27 mg L -1 , total<br />
hardness 38.91-68.21 mg L -1 (as CaCO 3 ), NH 3 -N 0.07-1.07 mg L -1 , NO 3 - -N 0.19-1.58 mg L -1 , NO 2 - -<br />
N 0.002-0.012 mg L -1 , TKN 0.19-4.16 mg L -1 and PO 4 3- -P 0.004-14.780 mg L -1 (Figure 3). It was<br />
noted that the PO 4 3- -P levels in all treatments were higher than the critical phosphorus level <strong>of</strong> 0.01<br />
mg L -1 for Cladophora growth in the natural ecosystem [20].<br />
High DO values occurred in all experiments as a result <strong>of</strong> aeration by the air pump. Algae<br />
cultivation in canteen wastewater with high nutrients has been observed to dramatically reduce<br />
nitrogen and phosphorous levels and at the same time produce a useful product for animal feed [36].<br />
In this work, it was noted that culturing <strong>of</strong> Cladophora sp. (Kai) showed similar improvement in<br />
water quality with decreasing BOD and nutrient levels (NH 3 -N, NO 3 - -N, TKN and PO 4 3- -P).<br />
CONCLUSIONS<br />
This study demonstrates that phosphate added to the culture <strong>of</strong> Cladophora sp. (Kai)<br />
enhances production <strong>of</strong> carotenoids and chlorophylls but has no effect on biomass production. In<br />
addition, cultivation <strong>of</strong> the alga in canteen wastewater improves the water quality by decreasing<br />
BOD and nutrients levels.<br />
ACKNOWLEDGEMENTS<br />
Financial support from Thailand Research Fund (TRF) through the Royal Golden Jubilee<br />
Ph.D. Program (Grant No. PHD/0130/2548) is acknowledged.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 1-11<br />
7<br />
A<br />
A<br />
B<br />
B<br />
C<br />
C<br />
D<br />
D<br />
E<br />
F F<br />
Figure 3. Water quality <strong>of</strong> Cladophora sp. (Kai) culture: water temperature (A), pH (B), DO (C),<br />
BOD (D), total hardness (E), and NH 3 -N (F) (C = control group, T1 = treatment 1, T2 = treatment 2,<br />
T3 = treatment 3, and T4 = treatment 4)
8 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 1-11<br />
A<br />
A<br />
B<br />
B<br />
C<br />
D<br />
D<br />
Figure 3 (Continued). Water quality <strong>of</strong> Cladophora sp. (Kai) culture: NO 3 - -N (A), NO 2 - -N (B), TKN (C),<br />
and PO 3- -P (D) (C = control group, T1 = treatment 1, T2 = treatment 2, T3 = treatment 3, and T4 =<br />
treatment 4)<br />
REFERENCES<br />
1. Y. Peerapornpisal, D. Amornlertpison, C. Rujjanawate, K. Ruangrit and D. Kanjanapothi, “Two<br />
endemic species <strong>of</strong> macroalgae in Nan river, northern Thailand, as therapeutic agents”,<br />
<strong>Science</strong>Asia, 2006, 32 (supplement 1), 71-76.<br />
2. S. Traichaiyaporn, B. Waraegsiri and J. Promya, “Culture <strong>of</strong> a green alga genus Cladophora<br />
(Kai) as feed for the Mekong giant catfish (Pangasianodon gigas, Chevey)”, Final Report <strong>of</strong><br />
The Thailand Research Fund, Bangkok, Thailand, 2011.<br />
3. Y. Yoshii, T. Hanyuda, I. Wakana, K. Miyaji, S. Arai, K. Ueda and I. Inouye, “Carotenoid<br />
compositions <strong>of</strong> Cladophora balls (Aegagropila linnaei) and some members <strong>of</strong> the<br />
Cladophorales (Ulvophyceae, Chlorophyta): Their taxonomic and evolutionary implication”, J.<br />
Phycol., 2004, 40, 1170-1177.<br />
4. T. Khuantrairong and S. Traichaiyaporn, “Production <strong>of</strong> biomass, carotenoid and nutritional<br />
values <strong>of</strong> Cladophora sp. (Kai) by cultivation in mass culture”, Phycologia, 2009, 48<br />
(supplement), 60-66.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 1-11<br />
9<br />
5. T. Khuantrairong and S. Traichaiyaporn, “The nutritional value <strong>of</strong> edible freshwater alga<br />
Cladophora sp. (Chlorophyta) grown under different phosphorus concentrations”, Int. J. Agric.<br />
Biol., 2011, 13, 297-300.<br />
6. T. W. Goodwin, “The Biochemistry <strong>of</strong> the Carotenoids”, 2nd Edn., Chapman and Hall, New<br />
York, 1980, pp.207-256.<br />
7. A. F. H. Marker, “The use <strong>of</strong> acetone and methanol in the estimation <strong>of</strong> chlorophyll in the<br />
presence <strong>of</strong> phaeophytin”, Freshwat. Biol., 1972, 2, 361-385.<br />
8. R. B. Woodward, “The total synthesis <strong>of</strong> chlorophyll”, Harvard University (no date),<br />
http://www.iupac.org/publications/pac/1961/pdf/0203x0383.pdf (Accessed: 29 Dec. 2011).<br />
9. P. Burtin, “Nutritional value <strong>of</strong> seaweeds”, Electron. J. Environ. Agric. Food Chem., 2003, 2,<br />
498-503.<br />
10. G. S. Omenn, G. E. Goodman, M. D. Thornquist, J. Balmes, M. R. Cullen, A. Glass, J. P.<br />
Keogh, F. L. Meyskens, B. Valanis, J. H. Williams, S. Barnhart and S. Hammar, “Effects <strong>of</strong> a<br />
combination <strong>of</strong> beta carotene and vitamin A on lung cancer and cardiovascular disease”, New<br />
Engl. J. Med., 1996, 334, 1150-1155.<br />
11. H. Tapiero, D. M. Townsend and K. D. Tew, “The role <strong>of</strong> carotenoids in the prevention <strong>of</strong><br />
human pathologies”, Biomed. Pharmacother., 2004, 58, 100-110.<br />
12. J. Promya, “Assessment <strong>of</strong> immunity stimulating capacity and meat, egg qualities <strong>of</strong> hybrid<br />
Tumtim tilapia ND56 (Oreochromis sp.) fed on raw Spirulina”, PhD Thesis, 2008, Chiang Mai<br />
University, Thailand.<br />
13. B. A. Whitton, “Studies on the growth <strong>of</strong> riverain Cladophora in culture”, Arch. Microbiol.,<br />
1967, 58, 21-29.<br />
14. C. E. R. Pitcairn and H. A. Hawkes, “The role <strong>of</strong> phosphorus in the growth <strong>of</strong> Cladophora”,<br />
Water Res., 1973, 7, 159-162.<br />
15. S. L. Wong and B. Clark, “Field determination <strong>of</strong> the critical nutrient concentrations for<br />
Cladophora in streams”, J. Fish. Res. Board Can., 1976, 33, 85-92.<br />
16. P. B. Birch, D. M. Gordon and A. J. McComb, “Nitrogen and phosphorus nutrition <strong>of</strong><br />
Cladophora in the Peel-Hervey estuarine system <strong>of</strong> Western Australia”, Bot. Mar., 1981, 24,<br />
381-387.<br />
17. J. M. Graham, M. T. Auer, R. P. Canale and J. P., H<strong>of</strong>fmann, “Ecological studies and<br />
mathematical modeling <strong>of</strong> Cladophora in Lake Huron: 4. Photosynthesis and respiration as<br />
functions <strong>of</strong> light and temperature”, J. Great Lakes Res., 1982, 8, 100-111.<br />
18. J. P. H<strong>of</strong>fmann and L. E. Graham, “Effects <strong>of</strong> selected physicochemical factors on growth and<br />
zoosporogenesis <strong>of</strong> Cladophora glomerata (Chlorophyta)”, J. Phycol., 1984, 20, 1-7.<br />
19. J. R. Wharfe, K. S. Taylor and H. A. C. Montgomery, “The growth <strong>of</strong> Cladophora glomerata in<br />
a river receiving sewage effluent”, Water Res., 1984, 18, 971-979.<br />
20. D. S. Painter and G. Kamaitis, “Reduction <strong>of</strong> Cladophora biomass and tissue phosphorus in<br />
Lake Ontario, 1972-83”, Can. J. Fish. Aquat. Sci., 1987, 44, 2212-2215.<br />
21. W. K. Dodds and D. A. Gudder, “The ecology <strong>of</strong> Cladophora”, J. Phycol., 1992, 28, 415-427.
10 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 1-11<br />
22. S. N. Higgins, R. E. Hecky and S. J. Guildford, “Environmental controls <strong>of</strong> Cladophora growth<br />
dynamics in eastern Lake Erie: Application <strong>of</strong> the Cladophora growth model (CGM)”, J. Great<br />
Lakes Res., 2006, 32, 629-644.<br />
23. S. N. Higgins, S. Y. Malkin, E. T. Howell, S. J. Guildford, L. Campbell, V. Hiriart-Baer and R.<br />
E. Hecky, “An ecological review <strong>of</strong> Cladophora glomerata (Chlorophyta) in the Laurentian<br />
Great Lakes”, J. Phycol., 2008, 44, 839-854.<br />
24. M. T. Auer and R. P. Canale, “Ecological studies and mathematical modeling <strong>of</strong> Cladophora in<br />
Lake Huron: 3. The dependence <strong>of</strong> growth rates on internal phosphorus pool size”, J. Great<br />
Lakes Res., 1982, 8, 93-99.<br />
25. M. Latasa and E. Berdalet, “Effect <strong>of</strong> nitrogen or phosphorus starvation on pigment<br />
composition <strong>of</strong> cultured Heterocapsa sp.”, J. Plankton Res., 1994, 16, 83-94.<br />
26. N. Leonardos and I. A. N. Lucas, “The nutritional values <strong>of</strong> algae grown under different<br />
culture conditions for Mytilus edulis L. larvae”, Aquaculture, 2000, 182, 301-315.<br />
27. M. Menéndez, J. Herrera and F. A. Comin, “Effect <strong>of</strong> nitrogen and phosphorus supply on<br />
growth, chlorophyll content and tissue composition <strong>of</strong> the macroalga Chaetomorpha linum<br />
(O.F. Müll.) Kütz, in a Mediterranean coastal lagoon”, Sci. Mar., 2002, 66, 355-364.<br />
28. M. Orosa, D. Franqueira, A. Cid and J. Abalde, “Analysis and enhancement <strong>of</strong> astaxanthin<br />
accumulation in Haematococcus pluvialis”, Bioresour. Technol., 2005, 96, 373-378.<br />
29. A. Celekli, M. Yavuzatmaca and H. Bozkurt, “Modeling <strong>of</strong> biomass production by Spirulina<br />
platensis as function <strong>of</strong> phosphate concentrations and pH regimes”, Bioresour. Technol., 2009,<br />
100, 3625-3629.<br />
30. P. Buapet, R. Hiranpan, R. J. Ritchie and A. Prathep, “Effect <strong>of</strong> nutrient inputs on growth,<br />
chlorophyll, and tissue nutrient concentration <strong>of</strong> Ulva reticulata from a tropical habitat”,<br />
<strong>Science</strong>Asia, 2008, 34, 245-252.<br />
31. B. R. Brinda, R. Sarada, B. S. Kamath and G. A. Ravishankar, “Accumulation <strong>of</strong> astaxanthin in<br />
flagellated cells <strong>of</strong> Haematococcus pluvialis - cultural and regulatory aspects”, Curr. Sci.,<br />
2004, 87, 1290-1295.<br />
32. N. Leonardos and R. J. Geider, “Elemental and biochemical composition <strong>of</strong> Rhinomonas<br />
reticulata (Cryptophyta) in relation to light and nitrate-to-phosphate supply ratios”, J. Phycol.,<br />
2005, 41, 567-576.<br />
33. E. Forján, I. Garbayo, C. Casal and C. Vílchez, “Enhancement <strong>of</strong> carotenoid production in<br />
Nannochloropsis by phosphate and sulphur limitation”, in “Communicating Current Research<br />
and Educational Topics and Trends in Applied Microbiology” (Ed. A. Méndez-Vilas),<br />
Formatex Research Center, Badajoz (Spain), 2007, pp.356-364.<br />
34. P. He, J. Duncan and J. Barber, “Astaxanthin accumulation in the green alga Haematococcus<br />
pluvialis: Effects <strong>of</strong> cultivation parameters”, J. Intgr. Plant Biol., 2007, 49, 447-451.<br />
35. B. P. R. Subudhi and P. K. Singh, “Effect <strong>of</strong> phosphorus and nitrogen on growth, chlorophyll,<br />
amino nitrogen, soluble sugar contents and algal heterocysts <strong>of</strong> water fern Azolla pinnata”,<br />
Biol. Plantarum, 1979, 21, 401-406.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 1-11<br />
11<br />
36. J. Promya, S. Traichaiyaporn and R. Deming, “Phytoremediation <strong>of</strong> kitchen wastewater by<br />
Spirulina platensis (Nordstedt) Geiteler: Pigment content, production variable cost and<br />
nutritional value”, <strong>Maejo</strong> Int. J. Sci. Technol., 2008, 2, 159-171.<br />
37. APHA, AWW and WPCF, “Standard Method for Examination <strong>of</strong> Water and Waste Water”,<br />
16 th Edn., American Public Health Association, Washington DC, 1985.<br />
38. G. Britton, “Workshop on carotenoid: The qualitative and quantitative analysis”, Prince <strong>of</strong><br />
Songkla University, Thailand, 2005, pp.9-16.<br />
39. S. Dere, T. Günes and R. Sivaci, “Spectrophotometric determination <strong>of</strong> chlorophyll-a, b and<br />
total carotenoid contents <strong>of</strong> some algae species using different solvents”, Turk. J. Bot., 1998,<br />
22, 13-17.<br />
40. H. K. Lichtenthaler and A. R. Wellburn, “Determination <strong>of</strong> total carotenoids and chlorophyll a<br />
and b <strong>of</strong> leaf extract in different solvents”, Biochem. Soc. Trans., 1983, 11, 591-592.<br />
41. V. E. Premila and M. U. Rao, “Effect <strong>of</strong> crude oil on the growth and reproduction <strong>of</strong> some<br />
benthic marine algae <strong>of</strong> Visakhapatnam coastline”, Indian J. Mar. Sci., 1997, 26, 195-200.<br />
42. F. X. Cunningham, “Regulation <strong>of</strong> carotenoid synthesis and accumulation in plants”, Pure<br />
Appl. Chem., 2002, 74, 1409-1417.<br />
43. V. G. Ladygin, “Biosynthesis <strong>of</strong> carotenoids in the chloroplasts <strong>of</strong> algae and higher plants”,<br />
Russ. J. Plant Physiol., 2000, 47, 904-923.<br />
44. V. J. Bellis and D. A. McLarty, “Ecology <strong>of</strong> Cladophora glomerata (L.) Kütz. in Southern<br />
Ontario”, J. Phycol., 1967, 3, 57-63.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
12 <strong>Maejo</strong> <strong>Maejo</strong> Int. J. Sci. Int. Technol. J. Sci. Technol. <strong>2012</strong>, 6(01), <strong>2012</strong>, 12-27 6(01), 12-27<br />
Full Paper<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Performance <strong>of</strong> self-excited induction generator with costeffective<br />
static compensator<br />
Yogesh K. Chauhan 1, * , Sanjay K. Jain 2 and Bhim Singh 3<br />
1<br />
Department <strong>of</strong> Electrical Engineering, School <strong>of</strong> Engineering, Gautam Buddha University, G.B.<br />
Nagar-201310, India<br />
2<br />
Electrical and Instrumentation Engineering Department, Thapar University, Patiala-147004, India<br />
3<br />
Department <strong>of</strong> Electrical Engineering, Indian Institute <strong>of</strong> Technology, New Delhi-110016, India<br />
* Corresponding author, e-mail: chauhanyk@yahoo.com<br />
Received: 6 June 2010 / Accepted: 8 January <strong>2012</strong> / Published: 13 January <strong>2012</strong><br />
Abstract: The performance <strong>of</strong> a system consisting <strong>of</strong> a three-phase self-excited induction<br />
generator (SEIG) with static compensator (STATCOM) for feeding static resistive-inductive<br />
(R-L) and dynamic induction motor (IM) loads was investigated. The cost-effective<br />
STATCOM providing stable operation was designed by connecting additional shunt<br />
capacitance with the load. The STATCOM-controlled algorithm was realised by controlling<br />
the source current using two control loops with proportional-integral (PI) controller: one for<br />
controlling the SEIG terminal voltage and the other for maintaining the DC bus voltage. The<br />
SEIG-STATCOM performance was studied for two designs <strong>of</strong> STATCOM, namely costeffective<br />
STATCOM and full-rating STATCOM. The cost-effective SEIG-STATCOM<br />
system with the proposed control scheme exhibited improved performance with respect to<br />
starting time, voltage dip, generator current and total harmonic distortion under various<br />
transient conditions.<br />
Keywords: self-excited induction generator (SEIG), static compensator (STATCOM),<br />
voltage regulation, renewable energy source<br />
__________________________________________________________________________________<br />
INTRODUCTION<br />
The fast rate <strong>of</strong> fossil fuel depletion is drawing attention to explore the alternative energy<br />
sources, e.g. small hydropower, wind and tidal power [1, 2]. The induction generator [3, 4] which<br />
operates in grid or self-excited mode is a strong candidate to harness electrical energy from these
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
13<br />
sources. An isolated electric supply using a self-excited induction generator (SEIG) is an economical<br />
option for such small-capacity applications as lighting and small motor loads in remote locations.<br />
The SEIG consists <strong>of</strong> a cage induction machine excited through an externally connected<br />
capacitor bank. The primary advantages <strong>of</strong> the SEIG in small capacity are the simple, brush-less and<br />
rugged construction, lower maintenance cost, small size and improved transient performance. The<br />
terminal voltage <strong>of</strong> SEIG is governed by parameters such as capacitance, prime mover speed and<br />
speed-torque characteristic and load. A poor voltage regulation results when the SEIG is loaded [5, 6]<br />
due to the increasing difference between the volt-ampere reactive (VAR) supplied by the capacitor<br />
bank and that demanded by the generator and load. The series capacitors [7-9] have been used with the<br />
SEIG for improved voltage regulation. The application <strong>of</strong> a thyristor as a static switch has resulted in<br />
such schemes as saturable core reactors and switching shunt capacitors for achieving improved voltage<br />
regulation <strong>of</strong> SEIG [10-12].<br />
The modern power electronic converters are characterised by fast response, improved<br />
switching features and low cost. These converters are being applied as flexible alternating current<br />
transmission systems (FACTS) for various control and regulating purposes. The performance and cost<br />
<strong>of</strong> the shunt capacitor, static VAR compensator (SVC) and STATCOM were compared [13]. The<br />
STATCOM is normally a current controlled–voltage source inverter (CC-VSI) and has wide<br />
applications in the power system for improving power quality, harmonic elimination, reactive power<br />
compensation and load balancing [14,15]. The installation issue and capability <strong>of</strong> STATCOM have<br />
been demonstrated for the voltage limited feeder and industrial facility [16,17]. The concepts <strong>of</strong> static<br />
compensation for SEIG have been described for static loads [18,19]; the steps in designing a<br />
STATCOM for SEIG system have been summarised [20]. However, most <strong>of</strong> these attempts have been<br />
made to feed three-phase static loads and very little efforts are made for SEIG feeding the dynamic<br />
(motor) loads.<br />
In this paper, the performance <strong>of</strong> SEIG with STATCOM on feeding a static R-L load and<br />
induction motor is investigated. The dynamic model <strong>of</strong> the system is developed and a methodology to<br />
decide the ratings <strong>of</strong> STATCOM components such as the DC bus capacitor, AC side filter and<br />
insulated gate bipolar transistors (IGBT) is presented. The system performance is also studied for<br />
cost-effective STATCOM and full-rating STATCOM designs.<br />
SYSTEM DESCRIPTION AND CONTROL SCHEME<br />
The schematic description <strong>of</strong> the SEIG-STATCOM system is shown in Figure 1. A suitable<br />
capacitor bank is needed to obtain the rated voltage at no load. The STATCOM consists <strong>of</strong> an IGBTbased<br />
3- CC-VSI, an AC side-filter inductor, and a capacitor on the self-supporting DC bus. The<br />
scheme to control the STATCOM is depicted in Figure 2. In this scheme, two control loops are<br />
employed: one for controlling the terminal voltage <strong>of</strong> SEIG and other for maintaining the DC bus<br />
voltage. The proportional-integral (PI) controller is used with both control loops, which possess the<br />
advantage <strong>of</strong> being simple, effective and easy to tune.
14 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
Figure 1. SEIG-STATCOM system supplying the load<br />
Figure 2. Control scheme for STATCOM<br />
The control scheme is based on direct control <strong>of</strong> the source current, which consists <strong>of</strong> in-phase<br />
I sd<br />
and quadrature I <br />
sq<br />
components. I sd<br />
is the current drawn from SEIG to maintain the DC bus<br />
voltage by charging (or discharging) the DC bus capacitor. I <br />
sq<br />
is the reactive current required to<br />
maintain the SEIG terminal voltage. To regulate the DC bus voltage, the error signal for the PI<br />
controller is obtained from the sensed DC bus voltage V dc<br />
and its reference value V dc<br />
. The output <strong>of</strong><br />
this PI controller is taken as I sd<br />
. The 3- reference in-phase current i sdabc<br />
is obtained by multiplying I <br />
sd<br />
with the sinusoidal in-phase unit voltage template u sdabc<br />
. To regulate the SEIG terminal voltage, the<br />
error signal for PI controller is obtained from the amplitude <strong>of</strong> SEIG voltage V sm<br />
and its reference<br />
value V sm<br />
. The output <strong>of</strong> this PI controller is taken as I sq<br />
. The 3- reference quadrature current i <br />
sqabc<br />
is<br />
obtained by multiplying I <br />
sq<br />
with the sinusoidal quadrature unit voltage template u sqabc<br />
. The total<br />
reference source current i sabc<br />
is taken as the sum <strong>of</strong> i sdabc<br />
and i <br />
sqabc<br />
. The sensed and reference source<br />
currents are processed in a rule based carrier less hysteresis band current controller to generate gating<br />
signals (S a , S b , S c ) for IGBT <strong>of</strong> CC-VSI <strong>of</strong> STATCOM.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
15<br />
DESIGN METHODOLOGY<br />
The STATCOM is designed considering that the VAR rating <strong>of</strong> STATCOM is fixed, the source<br />
voltage is completely sinusoidal and the pulse width modulation (PWM) inverter operates in linear<br />
mode. The VAR rating <strong>of</strong> STATCOM, (VAR) STATCOM , is estimated using the capacitance needed for<br />
providing the rated voltage at no load (C NL ) and that needed for a stable operation <strong>of</strong> the induction<br />
motor while being loaded to its rated capacity (C FL ). These capacitances are computed using sequential<br />
unconstrained minimisation technique (SUMT) in conjunction with Rosenbroack’s direct search<br />
method [23]. The problem formulation is briefed in Appendix I. The (VAR) STATCOM is computed as<br />
<br />
2 1 1 <br />
(VAR) STATCOM = 3V<br />
ab <br />
(1)<br />
<br />
XC<br />
X<br />
FL CNL<br />
<br />
and STATCOM line current I st is calculated as<br />
( VAR)<br />
STATCOM<br />
Ist<br />
(2)<br />
3V<br />
ab<br />
The reference DC bus voltage Vdcr<br />
<strong>of</strong> the voltage source converter <strong>of</strong> STATCOM depends upon<br />
the AC voltage. The STATCOM does not provide adequate compensation during the transient<br />
condition <strong>of</strong> low V dcr<br />
, whereas high V dcr<br />
may stress the devices. The V dcr<br />
is calculated using the peak<br />
supply line voltage V m [15] as<br />
V dcr = (1.2-2.0)V m , Vdcr<br />
Vm<br />
(3)<br />
The DC bus capacitor C dc stores energy and maintains the DC bus voltage with small ripple.<br />
The C dc is calculated using an energy balanced equation characterised by appropriate response time t<br />
and actual DC bus voltage V dca :<br />
C<br />
dc<br />
<br />
<br />
V<br />
3V ab<br />
I st<br />
t<br />
2 2<br />
dcr<br />
Vdca<br />
<br />
The AC filter inductance L f is decided by the allowable ripple in the compensation current [20].<br />
L f is calculated by assuming a linear mode operation (modulation index m=1) and a switching<br />
frequency f s <strong>of</strong> 10 kHz:<br />
3 <br />
Vdc<br />
2<br />
Lf<br />
<br />
<br />
(5)<br />
6af K<br />
s<br />
rp<br />
where the range <strong>of</strong> factor a (transient current) =1.2-2.0 and K rp (peak-to-peak ripple) =0.05-0.1.<br />
The ratings <strong>of</strong> IGBT suitable for medium rating and high frequency operation are decided as<br />
follows:<br />
Vdev (1 KL Kt ) Vm<br />
(6)<br />
I 2 K ( K 1)<br />
I<br />
dev s rp st<br />
where the ranges <strong>of</strong> K L (factor for filter inductor drop), K t (factor for transient voltage) and K s (factor<br />
<strong>of</strong> safety) are taken as 0.05-0.1, 0.1-0.2 and 1.25-1.50 respectively.<br />
(4)
16 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
SYSTEM MODEL<br />
The complete system as shown in Figure 1 consists <strong>of</strong> SEIG, STATCOM with associated<br />
control, and loads. The dynamic model <strong>of</strong> each component is presented herewith.<br />
SEIG Model<br />
The induction generator model is developed in a stationary q-d reference frame considering the<br />
effect <strong>of</strong> both main and cross flux saturation [21, 22]. The model, i.e. the q-d axis stator and rotor<br />
currents and the rotor speed (ω r ), in state space form is expressed using equations (7) and (8)<br />
respectively:<br />
p<br />
1<br />
i L v<br />
<br />
r<br />
i<br />
G<br />
i<br />
P<br />
p r<br />
TP Tem<br />
(8)<br />
2J<br />
3P<br />
Tem<br />
Lm<br />
iqsidr<br />
idsiqr<br />
, and v , i , r , L , G and T P are defined in Appendix-II.<br />
4<br />
where <br />
Shunt Capacitor Model<br />
The q-d axis stator currents i qs and i ds are converted into 3- stator currents i ga , i gb and i gc<br />
using 2- to 3- transformation [22]. Kirchh<strong>of</strong>f’s current law (KCL) [22] is applied to obtain the<br />
capacitor equations governing the SEIG voltage as<br />
pv<br />
pv<br />
ab<br />
bc<br />
<br />
<br />
iga ilda ica igb ildb icb<br />
<br />
3C<br />
sh<br />
iga ilda ica2igb ildb icb<br />
3C<br />
sh<br />
and v<br />
ab<br />
vbc<br />
vca<br />
0<br />
(10)<br />
where i ga<br />
, i lda<br />
and i ca<br />
are line ‘a’ currents for the generator, load and STATCOM respectively.<br />
STATCOM Model<br />
The charging (or discharging) <strong>of</strong> the DC bus capacitor C dc using hysteresis current controller<br />
switching functions (S a , S b , S c ) is expressed as<br />
pV<br />
<br />
i S i S i S<br />
<br />
ca a cb b cc c<br />
dc<br />
(11)<br />
Cdc<br />
The DC bus voltage V dc is reflected as voltages e a , e b and e c on the AC side <strong>of</strong> the PWM<br />
inverter as<br />
ea 2 1 1Sa<br />
Vdc<br />
e<br />
<br />
b<br />
1 2 1<br />
<br />
S<br />
<br />
<br />
<br />
b<br />
3 <br />
<br />
<br />
e<br />
<br />
c 1 1 2 S<br />
<br />
c<br />
(7)<br />
(9)<br />
(12)
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
17<br />
The AC side filter equations in state space form are expressed as<br />
vbc 2vab ebc 2eab 3Rf ica<br />
<br />
pica<br />
<br />
3Lf<br />
vbc vab ebc eab 3Rf icb<br />
picb<br />
<br />
3L<br />
where eab ea eb<br />
and e bc<br />
e b<br />
e c<br />
.<br />
f<br />
(13)<br />
Static R-L Load<br />
The static load is considered as delta-connected. The phase currents for static R-L load (i prla ,<br />
i prlb and i prlc ) are expressed as<br />
piprla ( vab Rai prla)<br />
La<br />
piprlb ( vbc Rbi prlb)<br />
Lb<br />
(14)<br />
pi ( v R i ) L<br />
prlc ca c prlc c<br />
Correspondingly, the line currents (i rla , i rlb and i rlc ) can be obtained as follows:<br />
i ( i i<br />
)<br />
rla prla prlc<br />
irlb ( iprlb iprla)<br />
i ( i i<br />
)<br />
rlc prlc prlb<br />
Induction Motor Load<br />
The values for v qs and v ds are obtained from SEIG voltages (v ab , v bc , v ca ) using 3- to 2-<br />
transformation [22]. The state space model <strong>of</strong> an induction motor dynamic load is similar to the<br />
induction generator model. It is expressed using motor parameters as<br />
1<br />
pi<br />
m<br />
Lm<br />
v<br />
m<br />
<br />
rm<br />
im<br />
<br />
G<br />
m<br />
im<br />
<br />
(16)<br />
Pm<br />
p rm<br />
Temm<br />
TL<br />
<br />
(17)<br />
2 J<br />
m<br />
The equations (7 _ 17) represent the model <strong>of</strong> the complete system. These equations are solved<br />
by fourth-order Runge-Kutta integration method [22] in MATLAB.<br />
(15)<br />
RESULTS AND DISCUSSION<br />
An investigation was carried out on a 3.7-kW induction machine operated as SEIG, which was<br />
loaded with a static R-L load <strong>of</strong> 0.8 pf and a 1.5-kW induction motor load. The parameters <strong>of</strong> this<br />
machine are given in Appendix III. The capacitances C NL and C FL were calculated as 16.1 F and 26.5<br />
F respectively for the rated SEIG voltage at no load and at the rated motor load under steady state.<br />
It was observed that the motor was unable to start even with C FL and resulted in a voltage<br />
collapse because the capacitor bank was unable to meet the excessive VAR requirements <strong>of</strong> SEIG and<br />
the motor load during starting. The starting <strong>of</strong> the motor was successful with a capacitance <strong>of</strong> 36F.<br />
The performance during voltage buildup, the starting <strong>of</strong> motor at 2.0 sec., and the subsequent loading<br />
on motor at 4.0 sec. are shown in Figure 3. During starting, the terminal voltage dropped to 0.25 p.u.<br />
and the induction motor took 0.7 sec. to stabilise the start-up transients. The excessive dip in voltage<br />
level and the longer duration for start-up were <strong>of</strong> serious quality concern. In addition, the SEIG<br />
exhibited poor voltage regulation even for the static load.
18 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
v (V)<br />
(Rad/s)<br />
ab<br />
T (Nm) i (A) i (A)<br />
rm emm ma gla<br />
1000<br />
0<br />
-1000<br />
20<br />
0<br />
-20<br />
10<br />
0<br />
-10<br />
50<br />
0<br />
-50<br />
400<br />
200<br />
0<br />
<br />
2 2.5 3 3.5 4<br />
2 2.5 3 3.5 4<br />
2 2.5 3 3.5 4<br />
2 2.5 3 3.5 4<br />
2 2.5 3 3.5 4<br />
Time(sec)<br />
Figure 3. SEIG feeding motor load with successful starting and consequent loading<br />
(Load conditions: 1.5-kW induction motor load at 2.0 sec. with C sh = 36 μF)<br />
The application <strong>of</strong> STATCOM was investigated for effective control and improved<br />
performance. The STATCOM was designed to give a cost-effective operation by selecting the C dc after<br />
considering the VAR requirement carefully. For a cost-effective STATCOM, the required VAR rating<br />
<strong>of</strong> STATCOM was calculated using the limiting capacitance <strong>of</strong> C NL and (C FL +C NL )/2 while an<br />
additional (C FL -C NL )/2 shunt capacitance was switched on with the load. For a full-rating STATCOM,<br />
the VAR rating <strong>of</strong> STATCOM was calculated with limiting capacitances C NL and C FL . For both the<br />
cost-effective and full-rating STATCOM designs, the shunt capacitance C NL needed to achieve the<br />
rated SEIG voltage at no load was taken as 16.1F. The different values <strong>of</strong> C FL were obtained for the<br />
static and motor loads.<br />
The capacitance needed for successful starting <strong>of</strong> the induction motor, which was 36 F, was<br />
taken as C FL for the motor load. C FL for the static load was considered as the capacitance needed to<br />
obtain the rated voltage at steady state, which was 30.4F. The STATCOM parameters are<br />
summarised in Table 1 for both cost-effective and full-rating STATCOM designs. The IGBT <strong>of</strong><br />
reduced current ratings were needed for a cost-effective STATCOM, which reduced the cost, and<br />
therefore the operation should be cost-effective.<br />
Table 1. STATCOM design parameters<br />
Parameter Cost-effective STATCOM Full-rating STATCOM<br />
2.2kW static load 1.5 kW IM load 2.2kW static load 1.5 kW IM load<br />
(0.8 pf)<br />
(0.8 pf)<br />
STATCOM rating 1.21 kVAR 1.615 kVAR 2.42 kVAR 3.23 kVAR<br />
Current supplied by STATCOM 1.68 A 2.25 A 3.37 A 4.5 A<br />
Ref. DC bus voltage 700 V 700 V 700 V 700 V<br />
DC bus capacitance 86.3 F 118.1 F 172.6 F 236.3 F<br />
AC filter inductance 11.16 mH 8.16 mH 5.58 mH 4.08 mH<br />
Device selection IGBT IGBT IGBT IGBT<br />
Device voltage rating 1094 V 1094 V 1094 V 1094 V<br />
Device current rating 5.5 A 7.5 A 11.0 A 15.0 A
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
19<br />
On the basis <strong>of</strong> STATCOM design parameters, a comparison between the cost-effective and<br />
full rating STATCOM is summarised in Table 2.<br />
Table 2. Comparison <strong>of</strong> performance parameters for cost-effective and full-rating STATCOM<br />
Performance parameter Cost-effective STATCOM Full-rating STATCOM<br />
Performance (voltage regulation,<br />
starting time, THD, etc.)<br />
Energy consumption (mainly in<br />
IGBT, DC bus capacitor and filter<br />
inductor loss)<br />
Overall cost (IGBT, DC bus<br />
capacitor, filter inductor)<br />
Reasonably good<br />
Low (lower IGBT current rating,<br />
lower DC bus capacitance and<br />
higher filter inductor)<br />
Low<br />
Good<br />
High (higher IGBT current rating,<br />
higher DC bus capacitance and<br />
lower filter inductor)<br />
High<br />
Performance with Balanced Static R-L Load<br />
The performance characteristics <strong>of</strong> the systems with cost-effective and full-rating STATCOM<br />
are shown in Figure 4 and Figure 5 respectively for feeding balanced 0.8-pf static R-L load. A load <strong>of</strong><br />
2.2 kW was applied at 2.5 sec., which was changed to 3.0 kW at 3.0 sec. and 2.2 kW at 3.5 sec.<br />
With cost-effective STATCOM (Figure 4), the visible transients lasting for about 0.2 sec were<br />
observed in v ab , i ga and V dc due to the application <strong>of</strong> 2.2-kW load. The V dc momentarily decreased to<br />
480V and varied between 480-800V before returning to the reference value. With full-rating<br />
STATCOM (Figure 5), the transients settled down after 5-6 cycles and the DC bus voltage decreased<br />
to 500V momentarily with the application <strong>of</strong> 2.2-kW load. The increase in static load to 3.0 kW at 3.0<br />
sec. did not result in any appreciable transients due to the choice <strong>of</strong> C dc corresponding to the motor<br />
load. In both cost-effective and full-rating STATCOM, V dc dropped momentarily to 630V and<br />
gradually returned to the reference value with the PI controller action. When the load was reduced to<br />
2.2 kW at 3.5 sec., the dynamics <strong>of</strong> the system changed accordingly. V dc momentarily rised to 760V<br />
before returning to the reference mark. A steady state was achieved within 0.12 sec. and 0.25 sec. with<br />
the full-rating and cost-effective STATCOM respectively.<br />
Performance with Unbalanced Static R-L Load<br />
The system performance with the cost-effective STATCOM was studied for an unbalanced R-L<br />
load and the corresponding characteristics for three-phase generator voltage v abc , generator current<br />
i gabc , load current i rlabc , compensation current i cabc , and DC bus voltage V dc are shown in Figure 6.<br />
Initially, the 2.2-kW, 3.0-kW and 1.0-kW loads <strong>of</strong> 0.8 pf were applied on ‘a’, ‘b’ and ‘c’ phases<br />
respectively at 3.0 sec. Further, an acute unbalance was made by unloading <strong>of</strong> ‘c’ phase at 3.5 sec. The<br />
STATCOM responded satisfactory during the unbalanced load condition and maintained the balanced<br />
condition at SEIG terminals.
20 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
1000<br />
v ab<br />
(V)<br />
0<br />
-1000<br />
2.4 2.6 2.8 3 3.2 3.4 3.6 3.8<br />
20<br />
i ga<br />
(A)<br />
0<br />
-20<br />
2.4 2.6 2.8 3 3.2 3.4 3.6 3.8<br />
10<br />
i rla<br />
(A)<br />
0<br />
-10<br />
2.4 2.6 2.8 3 3.2 3.4 3.6 3.8<br />
10<br />
i ca<br />
(A)<br />
0<br />
-10<br />
2.4 2.6 2.8 3 3.2 3.4 3.6 3.8<br />
V dc<br />
(V)<br />
800<br />
600<br />
400<br />
2.4 2.6 2.8 3 3.2 3.4 3.6 3.8<br />
Time(sec)<br />
Figure 4. Performance characteristics <strong>of</strong> SEIG-STATCOM (cost-effective) system with 0.8-pf static R-L<br />
load (Load conditions : 2.2 kW at 2.5 sec., 3.0 kW at 3.0 sec. and 2.2 kW at 3.5 sec.)<br />
1000<br />
v ab<br />
(V)<br />
0<br />
-1000<br />
2.4 2.6 2.8 3 3.2 3.4 3.6<br />
20<br />
i ga<br />
(A)<br />
0<br />
-20<br />
2.4 2.6 2.8 3 3.2 3.4 3.6<br />
10<br />
i rla<br />
(A)<br />
0<br />
-10<br />
2.4 2.6 2.8 3 3.2 3.4 3.6<br />
10<br />
i ca<br />
(A)<br />
0<br />
-10<br />
2.4 2.6 2.8 3 3.2 3.4 3.6<br />
800<br />
V dc<br />
(V)<br />
600<br />
400<br />
2.4 2.6 2.8 3 3.2 3.4 3.6<br />
Time(sec)<br />
Figure 5. Performance characteristics <strong>of</strong> SEIG-STATCOM (full rating) with 0.8-pf static R-L load<br />
(Load conditions: 2.2 kW at 2.5 sec., 3.0 kW at 3.0 sec., and 2.2 kW at 3.5 sec.)
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
21<br />
1000<br />
v abc<br />
(V)<br />
0<br />
-1000<br />
20<br />
3 3.2 3.4 3.6 3.8 4 4.2<br />
i rlabc<br />
(A)<br />
i gabc<br />
(A)<br />
i cabc<br />
(A)<br />
V dc<br />
(V)<br />
0<br />
-20<br />
10<br />
0<br />
-10<br />
5<br />
0<br />
-5<br />
750<br />
700<br />
650<br />
3 3.2 3.4 3.6 3.8 4 4.2<br />
3 3.2 3.4 3.6 3.8 4 4.2<br />
3 3.2 3.4 3.6 3.8 4 4.2<br />
3 3.2 3.4 3.6 3.8 4 4.2<br />
Time(sec)<br />
Figure 6. Performance characteristics <strong>of</strong> SEIG-STATCOM (cost-effective) with unbalanced 0.8-pf R-L load<br />
(Load conditions: 3-φ load (2.2 kW, 3.0 kW, 1.0 kW) at 3.0 sec.; load (2.0 kW, 3.0 kW,<br />
unloading on ‘c’ phase) at 3.5 sec; and balanced 3-φ loading <strong>of</strong> 2.2.0 kW at 4.0 sec.)<br />
Performance with Motor Load<br />
The performance characteristics <strong>of</strong> the system during the starting and sudden loading <strong>of</strong> an<br />
induction motor are shown in Figures 7 and 8 for cost-effective and full-rating STATCOM<br />
respectively. The motor at no load was switched on at 3.0 sec. and was loaded at 4.0 sec. During the<br />
starting <strong>of</strong> the motor, excessive reactive VAR were drawn and therefore the V dc dropped to 300V level<br />
with both STATCOM. The V dc quickly rose above the reference 700V before returning to reference<br />
level after replenishing the charge on DC bus capacitor. At starting, the transients in v ab , i ga , i ma , T emm<br />
and V dc were more visible with cost-effective STATCOM due to its lower rating and the uncharged<br />
capacitor connection across the motor load terminals. Because <strong>of</strong> a lower compensation level, the i ca<br />
under steady state was lower compared to the corresponding value with full-rating STATCOM. With<br />
cost-effective STATCOM, the SEIG voltage decreased by 10% and the transients were settled in 0.5<br />
sec., whereas these values were 6% and 0.4 sec. respectively for full-rating STATCOM. Without<br />
STATCOM, as shown in Figure 3, the decrease in SEIG voltage was 30%. The motor was loaded with<br />
rated load torque at 4.0 sec. After small transients, the system response changed accordingly. The<br />
increase in i ga , i ma , and T emm and the decrease in rm were observed. The V dc decreased to 670V before<br />
returning to steady-state reference value <strong>of</strong> 700V.<br />
The performance <strong>of</strong> SEIG-IM configuration was compared in the presence and absence <strong>of</strong><br />
STATCOM and the key parameter indices <strong>of</strong> the system are summarised in Table 3. The total<br />
harmonic distortion (THD) was obtained through fast Fourier transform (FFT) using discrete Fourier<br />
transform (DFT) algorithm <strong>of</strong> MATLAB. The generator-side THD values were within a permissible<br />
level, which demonstrated a satisfactory overall system performance.
22 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
1000<br />
v ab<br />
(V)<br />
0<br />
-1000<br />
20<br />
3 3.5 4 4.5<br />
i ga<br />
(A)<br />
0<br />
-20<br />
20<br />
3 3.5 4 4.5<br />
i (A) (Rad/sec) rm T (N )<br />
i (A)<br />
ca emm m ma<br />
0<br />
-20<br />
50<br />
0<br />
500<br />
0<br />
20<br />
0<br />
-20<br />
800<br />
3 3.5 4 4.5<br />
3 3.5 4 4.5<br />
3 3.5 4 4.5<br />
3 3.5 4 4.5<br />
V dc<br />
(V)<br />
600<br />
400<br />
3 3.5 4 4.5<br />
Time(sec)<br />
Figure 7. Performance characteristics <strong>of</strong> SEIG-STATCOM (cost-effective) with induction motor load<br />
(Load conditions: T L = 0 at 3.0 sec. and T L = rated torque at 4.0 sec.)<br />
1000<br />
v ab<br />
(V)<br />
0<br />
-1000<br />
2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4<br />
20<br />
i ga<br />
(A)<br />
0<br />
-20<br />
2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4<br />
20<br />
(Rad/sec)<br />
i (A) rm T (Nm) i (A)<br />
ca emm<br />
ma<br />
0<br />
-20<br />
2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4<br />
50<br />
0<br />
2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4<br />
500<br />
0<br />
2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4<br />
50<br />
0<br />
-50<br />
2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4<br />
800<br />
V dc<br />
(V)<br />
600<br />
400<br />
2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4<br />
Time(sec)<br />
Figure 8. Performance characteristics <strong>of</strong> SEIG-STATCOM (full-rating) with induction motor load<br />
(Load conditions : T L = 0 at 3.0 sec. and T L = rated torque at 4.0 sec.)
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
23<br />
Table 3. Comparison <strong>of</strong> performance indices <strong>of</strong> the system with and without STATCOM<br />
Performance index<br />
Without<br />
STATCOM<br />
With STATCOM<br />
Cost-effective<br />
design<br />
Supply voltage dip at start-up 75% 10.0% 6.1%<br />
Start-up time 0.7 sec 0.18 0.16 sec<br />
Supply current THD at full motor load NA 1.4% 1.3%<br />
Rise/drop in DC bus voltage <strong>of</strong><br />
STATCOM at the time <strong>of</strong> sudden IM<br />
loading<br />
Drop in DC bus voltage with full torque<br />
loading <strong>of</strong> IM<br />
Full-rating design<br />
NA 840V/300V 790V/330V<br />
NA 625V 650V<br />
CONCLUSIONS<br />
The investigations on SEIG-STATCOM system have been presented for feeding static R-L and<br />
induction motor loads. The dynamic model <strong>of</strong> the system has been developed in state space form. The<br />
STATCOM parameters have been calculated for two design cases, namely the cost-effective and fullrating<br />
designs using the proposed design procedure. The cost-effective STATCOM was designed to<br />
provide a stable operation by connecting additional shunt capacitance with the load. The system<br />
performance has been presented for both the cost-effective and full-rating STATCOM. With a<br />
controlling algorithm, the system exhibited improved performance in terms <strong>of</strong> parameters such as<br />
starting time, voltage dip, generator current and total harmonic distortion in the supply current under<br />
various transient conditions. The full-rating STATCOM could be considered for critical applications<br />
having stringent performance consideration while the cost-effective STATCOM should be for remote<br />
applications where the cost is important.
24 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
Appendix I<br />
The per phase equivalent circuit <strong>of</strong> SEIG feeding induction motor load at steady state is shown<br />
in Figure 9. In the equivalent circuit, the SEIG parameters R s and X ls are stator resistance and leakage<br />
reactance respectively, R r and X lr are rotor resistance and leakage reactance respectively, and X m is the<br />
magnetising reactance. The corresponding parameters are presented with subscript m for motor load.<br />
X csh is the reactance <strong>of</strong>fered by the capacitor bank, and F and are per unit frequency and prime-mover<br />
speed respectively.<br />
jFX lr<br />
jFX m<br />
jFX ls<br />
R s<br />
-jX csh<br />
/F<br />
R sm<br />
jFX lsm<br />
jFX mm<br />
jFX lrm<br />
R rm<br />
/(F- m<br />
)<br />
R r<br />
/(F-)<br />
I s<br />
Figure 9. Per phase equivalent <strong>of</strong> SEIG feeding motor load<br />
Applying KVL on stator side loop <strong>of</strong> induction generator results in<br />
Z loop I s = 0<br />
Under steady state condition, I s cannot be zero and therefore Z loop<br />
should be zero. An<br />
optimisation problem has been formulated to obtain the unknown variables X csh and F. The objective<br />
function F n is expressed as<br />
Fn( X<br />
csh, F) abs( Zloop)<br />
The values <strong>of</strong> X csh and F should lie between the respective minimum and maximum limits:<br />
Fmn F Fmx , Xcmn Xcsh Xcmx<br />
<br />
The above optimisation problem is solved through SUMT in conjunction with the Rosenbroack<br />
method <strong>of</strong> direct search technique [23]. After the convergence, the capacitance is computed.<br />
Appendix II<br />
The induction machine model is developed in a stationary reference frame while incorporating<br />
the effects <strong>of</strong> both the main flux and cross-flux saturation. The forms <strong>of</strong> v, i, r,<br />
L<br />
and G are<br />
given as<br />
T<br />
T<br />
v v v v v i i i i i r diag r r r r<br />
qs ds qr dr ; qs ds qr dr ; <br />
L<br />
L L L L<br />
<br />
L L L L<br />
<br />
L L L L<br />
<br />
L L L L<br />
sq dq mq dq<br />
dq sq dq md<br />
mq dq rq dq<br />
dq md dq rd<br />
<br />
<br />
<br />
<br />
<br />
<br />
0 0 0 0 <br />
<br />
0 0 0 0<br />
<br />
G <br />
<br />
0 <br />
rLm 0 Lr<br />
<br />
<br />
<br />
rLm 0 Lr<br />
0 <br />
<br />
s s r r<br />
The air gap voltage <strong>of</strong> SEIG does not remain constant during loading. Therefore, the<br />
magnetising inductance is calculated by calculating the magnetising current as<br />
<br />
2 2<br />
m<br />
<br />
qs<br />
<br />
qr<br />
<br />
ds<br />
<br />
dr<br />
i i i i i<br />
The inductances in [L] are evaluated [21] as
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
25<br />
L / i , Ld / di ,cos i / i , sin i / i<br />
m m m m m dm m qm m<br />
L ( LL<br />
)cossin<br />
dq<br />
m<br />
L Lcos L sin , L Lsin L<br />
cos <br />
2 2 2 2<br />
mq m md m<br />
L L L , L L L , L L L , L L L<br />
sq ls mq sd ls md rq lr mq rd lr md<br />
The prime mover torque driving the induction machine is expressed as<br />
T 6200 20<br />
P<br />
r<br />
Appendix III<br />
Generator parameters<br />
3.7 kW, 415 V, -connected, 7.6 A (line), 50 Hz, 4 pole, J=0.0842 kg-m 2 cage induction machine.<br />
R s =0.0585 pu, R r =0.06196 pu, X ls =X lr =0.1015 pu, X m(unsat) =2.858 pu.<br />
Magnetisation characteristic <strong>of</strong> SEIG<br />
2<br />
L K i K i K<br />
m 1 m 2 m 3<br />
where K 1 0.0091, K 2 2.0024 , K 3 348.35.<br />
Motor parameters<br />
1.5 kW, 415 V, -connected, 3.2 A (line), 50 Hz, 4 pole, J=0.0205 kg-m 2 cage induction machine.<br />
R sm =0.0832 pu, R rm =0.0853 pu, X lsm =X lrm =0.1101 pu, X mm(unsat) =1.83 pu.<br />
Magnetisation characteristic <strong>of</strong> motor<br />
5 4 3 2<br />
Lmm Km 1imm Km2imm Km3imm Km4imm Km5imm Km6<br />
where Km<br />
1<br />
0.0072, Km2 0.0849 , Km3 0.4298 , Km4 0.9924 , Km5 0.6847 , Km6 1.171.
26 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
REFERENCES<br />
1. S. Major, T. Commins, and A. Noppharatana, “Potential <strong>of</strong> wind power for Thailand: An<br />
assessment”, <strong>Maejo</strong> Int. J. Sci. Technol. 2008, 2, 255-266.<br />
2. F. F. Li, J. Kueck, T. Rizy and T. King, “A preliminary analysis <strong>of</strong> the economics <strong>of</strong> using<br />
distributed energy as a source <strong>of</strong> reactive power supply”, Report for the U.S. Department <strong>of</strong><br />
Energy, 2006, www.localpower.org/documents/reporto_doe_reactivepowerandde.pdf.<br />
3. E. D. Bassett and F. M. Potter, “Capacitive excitation for induction generators”, AIEE Trans.<br />
(Elect. Eng.), 1935, 54, 540-545.<br />
4. J. M. Elder, J. T. Boys and J. L. Woodward, “Self-excited induction machine as a small low-cost<br />
generator”, IEE Proc. Gener. Transm. Distrib., 1984, 131, 33-41.<br />
5. S. S. Murthy, O. P. Malik and A. K. Tandon, “Analysis <strong>of</strong> self excited induction generators”, IEE<br />
Proc. Gener. Transm. Distrib., 1982, 129, 260-265.<br />
6. L. Shridhar, B. Singh, C. S. Jha and B. P. Singh, “Analysis <strong>of</strong> self excited induction generator<br />
feeding induction motor”, IEEE Trans. Ener. Convers., 1994, 9, 390-396.<br />
7. E. Bim, J. Szajner and Y. Burian, “Voltage compensation <strong>of</strong> an induction generator with longshunt<br />
connection”, IEEE Trans. Ener. Convers., 1989, 4, 526-530.<br />
8. M. H. Haque, “Selection <strong>of</strong> capacitors to regulate voltage <strong>of</strong> a short-shunt induction generator”,<br />
IET Gener. Transm. Distrib., 2009, 3, 257–265.<br />
9. L. Wang and J. Y. Su “Effects <strong>of</strong> long-shunt and short-shunt connections on voltage variations <strong>of</strong> a<br />
self-excited induction generator”, IEEE Trans. Ener. Convers., 1997, 12, 368-374.<br />
10. S. M. Alghuwainem, “Steady-state analysis <strong>of</strong> an induction generator self excited by a capacitor in<br />
parallel with saturable reactor”, Electr. Mach. Power Syst., 1998, 26, 617-625.<br />
11. S. C. Kuo and L. Wang, “Analysis <strong>of</strong> isolated self-excited induction generator feeding a rectifier<br />
load”, IEE Proc. Gener. Transm. Distrib., 2002, 149, 90-97.<br />
12. T. L. Maguire and A. M. Gole, “Apparatus for supplying an isolated DC load from a variablespeed<br />
self-excited induction generator”, IEEE Trans. Ener. Convers., 1993, 8, 468-475.<br />
13. A. S. Yome and N. Mithulananthan, “Comparison <strong>of</strong> shunt capacitor, SVC and STATCOM in<br />
static voltage stability margin enhancement”, Int. J. Electr. Eng. Edu., 2004, 41, 158-171.<br />
14. B. Singh, K. Al-Haddad and A. Chandra, “Harmonic elimination, reactive power compensation<br />
and load balancing in three-phase, four-wire electric distribution systems supplying non-linear<br />
loads”, Electr. Power Syst. Res., 1998, 44, 93-100.<br />
15. S. K. Jain, P. Agarwal and H. O. Gupta, “Fuzzy logic controlled shunt active power filter for power<br />
quality improvement”, IEE Proc. Electr. Power Appl., 2002, 149, 317-328.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 12-27<br />
27<br />
16. S. M. Ramsay, P. E. Cronin, R. J. Nelson, J. Bian and F. E. Menendez, “Using distribution static<br />
compensators (D-STATCOMs) to extend the capability <strong>of</strong> voltage-limited distribution feeders”,<br />
Proceedings <strong>of</strong> IEEE Rural Electric Power Conference, 1996, Fort Worth, TX, USA, pp. A4:1-<br />
A4:7.<br />
17. K. N. Choma and M. Etezadi-Amoli, “The application <strong>of</strong> a DSTATCOM to an industrial facility”,<br />
Proceedings <strong>of</strong> IEEE Power Engineering Society Winter Meeting, 2002, New York, USA, Vol. 2,<br />
pp. 725-728.<br />
18. B. Singh and L. B. Shilpakar, “Analysis <strong>of</strong> a novel solid state voltage regulator for a self-excited<br />
induction generator”, IEE Proc. Gener. Transm. Distrib., 1998, 145, 647-655.<br />
19. S. C. Kuo and L. Wang, “Analysis <strong>of</strong> voltage control for a self-excited induction generator using a<br />
current-controlled voltage source inverter (CC-VSI)”, IEE Proc. Gener. Transm. Distrib., 2001,<br />
148, 431-438.<br />
20. B. Singh, S. S. Murthy and S. Gupta, “Analysis and design <strong>of</strong> STATCOM-based voltage regulator<br />
for self-excited induction generators”, IEEE Trans. Ener. Convers., 2004, 19, 783-790.<br />
21. E. Levi, “Applications <strong>of</strong> the current state space model in analyses <strong>of</strong> saturated induction<br />
machines”, Electr. Power Syst. Res., 1994, 31, 203-216.<br />
22. P. C. Krause, “Analysis <strong>of</strong> electrical machinery”, McGraw Hill, New York, 1986.<br />
23. S. S. Rao, “Engineering optimization, theory and practice”, New Age <strong>International</strong>, New Delhi,<br />
1998.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
28<strong>Maejo</strong> <strong>Maejo</strong> Int. J. Int. Sci. J. Technol. Sci. Technol. <strong>2012</strong>, <strong>2012</strong>, 6(01), 6(01), 28-46 28-46<br />
Full Paper<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Benthic diatoms <strong>of</strong> Mekong River and its tributaries in<br />
northern and north-eastern Thailand and their application to<br />
water quality monitoring<br />
Sutthawan Suphan 1,* , Yuwadee Peerapornpisal 1 and Graham J. C. Underwood 2<br />
1 Department <strong>of</strong> Biology, Faculty <strong>of</strong> <strong>Science</strong>, Chiang Mai University, 50200, Thailand<br />
2<br />
Department <strong>of</strong> Biological <strong>Science</strong>s, University <strong>of</strong> Essex, Colchester, Essex, United Kingdom CO4<br />
3SQ<br />
* Corresponding author, e-mail: suttawan@hotmail.com<br />
Received: 7 February 2011 / Accepted: 6 May 2011 / Published: 26 January <strong>2012</strong><br />
Abstract: Biomonitoring <strong>of</strong> benthic diatoms was performed to assess the water quality <strong>of</strong><br />
Mekong River and its tributaries in northern and north-eastern Thailand. Fourteen<br />
sampling sites along the river and its tributaries were investigated. Two hundred and<br />
fifty-two species in 53 genera <strong>of</strong> diatoms were recorded. Each sampling site had distinct<br />
water chemistry and other physical variables. Cluster analysis identified 11 groups at<br />
80% similarity. The relationship between diatom community composition and water<br />
quality variables was determined by statistical techniques. A number <strong>of</strong> diatom species<br />
were found to be useful as indicators <strong>of</strong> some physico-chemical properties <strong>of</strong> water.<br />
Keywords: benthic diatoms, water quality, biomonitoring, Mekong River<br />
_______________________________________________________________________________________<br />
INTRODUCTION<br />
One <strong>of</strong> the most important natural resources for human life is water resource. Human use<br />
including household, industrial, agricultural and recreational activities affects water quality. There<br />
is an urgent need to develop a more sustainable practice for the management and efficient use <strong>of</strong><br />
water resources as well as the need to protect the ecosystems where these resources are located.<br />
Therefore, it is important to monitor the quality <strong>of</strong> our limited water supplies. Presently, there are<br />
several methods to monitor water quality. One <strong>of</strong> the methods that is successfully used for<br />
monitoring aquatic environments around the world is biological assessment <strong>of</strong> water quality. It is<br />
considered an essential part in the assessment <strong>of</strong> the ecological quality <strong>of</strong> running waters apart from
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
29<br />
the data obtained from other sources such as hydrology, eco-morphology, physico-chemical water<br />
analysis and eco-toxicological analysis [1].<br />
In rivers and streams, benthic diatoms, the most common and diverse primary producers [2],<br />
are regarded as bioindicators due to their sensitivity and strong response to many physical, chemical<br />
and biological changes [3]. Diatoms are successfully used for monitoring aquatic environments<br />
around the world, especially in Europe, USA and Japan [4-5].<br />
In Thailand, Mekong River flows through Chiang Rai province in the north. As it continues<br />
along its path in Laos, it once again flows back into Loei, Nong Khai, Nakhon Panom, Mukdaharn,<br />
Amnaj Charoen and Ubon Ratchadhani provinces in the north-east <strong>of</strong> Thailand. There are many<br />
smaller rivers which are tributaries that may affect the Mekong River. Thus, the water quality in the<br />
Mekong River and its tributaries should be monitored continuously.<br />
There have been few studies <strong>of</strong> diatoms in the Mekong River in the past [6-8]. In this study,<br />
the diversity and distribution <strong>of</strong> benthic diatoms in the Mekong River and its tributaries in northern<br />
and north-eastern Thailand was investigated. In addition the relationship between diatoms species<br />
and some water physico-chemical properties was studied. The basic ecological data that could be<br />
applied to develop sustainable water resources were also obtained. Furthermore, the study should<br />
provide more information on diatoms in the South-east Asian region where very few reports on<br />
benthic diatoms were documented.<br />
MATERIALS AND METHODS<br />
Study Area<br />
Fourteen sampling sites along the Mekong River and its tributaries in northern and northeastern<br />
Thailand were selected based on the distance and environmental impact. Nine sites were<br />
located along Mekong River and five sites along its tributaries. The detail <strong>of</strong> each sampling site is<br />
shown in Figure 1. Diatom samples and physico-chemical water quality were determined 3 times<br />
per year in each season during July 2005 – April 2007.<br />
Benthic Diatom Collection and Identification<br />
Ten replicates <strong>of</strong> benthic diatoms were collected at each sampling site. Diatoms were taken<br />
from stone surfaces using a toothbrush and a 10-cm 2 plastic sheet. The samples were put in plastic<br />
boxes and fixed with Lugol’s solution on site. Diatom samples were then taken to the laboratory and<br />
cleaned by concentrated acid digestion method [9]. Briefly, each sample was centrifuged at 3,500<br />
rpm for 15 minutes. The diatom cells were placed in an 18-cm core tube, added with concentrated<br />
nitric acid, heated in a boiler (70-80 o C) for 30-45 minutes, and rinsed 4-5 times with deionised<br />
water.<br />
Each cleaned diatom sample was mounted on a microscope slide with Naphrax ® , a mountant<br />
with a high refractive index [9-10]. Up to 300 diatom valves were counted and identified with an<br />
Olympus CH30 microscope at ×1000 magnification. Taxonomy and nomenclature was determined<br />
according to the relevant references [8, 11-17].
30 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
Sampling sites along Mekong River<br />
Sampling sites along tributaries<br />
Site 1 Golden Triangle, Chiang Rai province (GT)<br />
Site 2 Kok River, Chiang Rai province (KO)<br />
Site 3 Ban Had Krai, Chiang Rai province (HK)<br />
Site 4 Hueng River, Loei province (HG)<br />
Site 5 Kaeng Khood Khoo, Loei province (KK)<br />
Site 6 Ponpisai, Nong Khai province (PS)<br />
Site 7 Laung River, Nong Khai province (LG)<br />
Site 8 Mueng, Nakhon Phanom province (NP)<br />
Site 9 Songkram River, Sakonnakron province (SK)<br />
Site 10 Kaeng Ka Bao, Nakhon Phanom province (KB)<br />
Site 11 Had Hin Win Chai, Amnat Charoen province (HW)<br />
Site 12 Kaeng Hin Kan, Mukdahan province (KH)<br />
Site 13 Kaeng Sapue, Ubon Ratchathani province (KP)<br />
Site 14 Khong Jium, Ubon Ratchathani province (KJ)<br />
Figure 1. Map <strong>of</strong> Thailand showing the location <strong>of</strong> 14 sampling sites along Mekong River and its<br />
tributaries<br />
Physico-Chemical Data<br />
Water samples were collected in triplicate at each sampling site. The samples were put in<br />
polyethylene bottles and kept in a cool box at 5-7 o C. Water chemical and physical properties were<br />
determined by established methods [18]; soluble reactive phosphorus (SRP) was determined by<br />
ascorbic acid method, nitrate nitrogen (NO 3 - -N) by cadmium reduction method, ammonia nitrogen<br />
(NH 4 + -N) by Nesslerisation method, alkalinity (as mg/L CaCO 3 ) by phenolpthalein methyl orange<br />
indicator method, dissolved oxygen (DO) by azide modification <strong>of</strong> the Winkler method, and<br />
biochemical oxygen demand (BOD) by 5-day incubation and azide modification <strong>of</strong> the Winkler<br />
method. The pH was measured with a pH meter, conductivity with a conductivity meter, turbidity<br />
with a turbidity meter, water temperature with a thermometer, and water velocity with a velocity<br />
meter (Aquaflow Probe - Model 6900, Ricky Hydrological Company).<br />
Data Analysis<br />
ANOVA single factor was performed on water quality and data sets on benthic diatoms<br />
assemblage to determine any significant differences between groups. Pearson’s correlation was also<br />
calculated for certain variables. Water quality data and diatom species were analysed with canonical<br />
correspondence analysis (CCA), a multivariate direct gradient analysis method widely used in<br />
ecology, to determine the relationship between physico-chemical water quality and diatom species<br />
using multivariate statistical package (MVSP) s<strong>of</strong>tware. Cluster analysis was performed on the log<br />
<strong>of</strong> transformed water quality data using MVSP s<strong>of</strong>tware with unweighted pair-group method <strong>of</strong><br />
arithmetic averages (UPGMA) cluster method to show similarity percentage between samples.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
31<br />
Environmental variables <strong>of</strong> the clusters were then tested for significance between groups with<br />
ANOVA single factor. The number <strong>of</strong> benthic diatoms in each group was calculated by Minitab<br />
program to find significant differences and correlations between groups.<br />
RESULTS AND DISCUSSION<br />
Diatom Diversity and Distribution<br />
A total <strong>of</strong> 135,859 benthic diatom cells were counted. Two hundred and fifty-two species <strong>of</strong><br />
benthic diatoms were found and classified into 3 classes, 6 subclasses, 14 orders, 27 families and 53<br />
genera. The majority (88.5%) was in Bacillariophyceae class with the remaining 6.0% in<br />
Fragilariophyceae class and 5.5% in Coscinodiscineae class. Nitzschia was the genus with the<br />
highest number <strong>of</strong> species (30 species) followed by Navicula (25 species), Gomphonema (16<br />
species), Eunotia (14 species), Luticola (12 species) and Pinnularia (8 species). The number <strong>of</strong><br />
diatom species recorded was similar to that reported for other river systems <strong>of</strong> comparable size.<br />
Holmes and Whitton [19] listed 230 diatom species collected from Tees River in northern England<br />
while Archibald [20] recorded 310 diatom species from Sundays and Great Fish Rivers in South<br />
Africa. Similarly, 267 species <strong>of</strong> diatoms found in La Trobe River and tributaries in Australia was<br />
reported [21]. The distribution <strong>of</strong> the diatoms is shown in Figure 2. The lowest number <strong>of</strong> diatoms<br />
was recorded at HK in Chiang Rai, northern Thailand, during the fourth sampling in July 2006<br />
(rainy season).<br />
Figure 2. Number <strong>of</strong> benthic diatoms in Mekong River and its tributaries between July 2005 –April<br />
2007<br />
Among the 252 species, 29 were common species as shown in Table 1 and Figures 3.<br />
Nitzschia palea showed the highest percentage (36.0%) followed by Mayamaea atomus (35.4%),<br />
Eolimna minima (25.3%), Navicula cryptotenelloides (24.9%), Cymbella sp.1 (21.3%) and<br />
Achnanthidium minutissimum (20.5%).
32 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
Table 1. Twenty nine common species <strong>of</strong> benthic diatoms in Mekong River and its tributaries. The<br />
percentage <strong>of</strong> relative abundance is shown in bracket.<br />
Taxa<br />
Nitzschia palea (Kützing) Smith (36.0%)<br />
Planothidium frequentissimum (Lange-Bertalot) Round<br />
Mayamaea atomus (Kützing) Lange-Bertalot (35.4%)<br />
& Bukhtiyarova(11.9%)<br />
Navicula symmetrica Patrick (33.4%) Cymbella sumatrensis Hustedt (11.7%)<br />
Eolimna minima (Grunow) Lange-Bertalot (25.3%) Nitzschia filiformis (Smith) Hustedt (11.4%)<br />
Navicula cryptotenelloides Lange-Bertalot (24.9%) Ulnaria ulna (Nitzsch) Compère (10.8%)<br />
Gomphonema lagenula Kützing (24.6%) Eolimna subminuscula (Manguin) Gerd Moser (9.8%)<br />
Cymbella sp.1 (21.3%) Fragilaria bidens Heiberg (9.0%)<br />
Achnanthidium minutissimum (Kützing) Czarnecki (20.5%) Melosira varians Agardh (9.0%)<br />
Navicula cryptotenella Lange-Bertalot (19.9%) Navicula menisculus Schumann (6.9%)<br />
Nitzschia inconspicua Grunow (19.2%) Nitzschia dissipata (Kützing) Grunow (6.6%)<br />
Nitzschia supralitorea Lange-Bertalot (17.2%) Geissleria decussis (Østrup) Lange-Bertalot&Metzeltin (5.9%)<br />
Navicula rostellata Kützing (14.5%) Achnanthidium convergens (Kobayasi) Kobayasi (5.9%)<br />
Encyonema sp.1 (14.3%) Frustulia undosa Metzeltin & Lange-Bertalot (4.7%)<br />
Luticola goeppertiana (Bleisch) Mann in Round, Sellaphora pupula (Kützing) Mereschkovsky (3.1%)<br />
Crawford & Mann (13.7%) Nitzschia microcephala Grunow (2.9%)<br />
Nitzschia clausii Hantzsch (13.1%)<br />
According to Jüttner et al. [22], M. atomus, A. minutissimum and N. palea were common<br />
species in streams in agricultural catchments <strong>of</strong> Kathmandu valley, Nepal. Duong et al.[23] reported<br />
the impact <strong>of</strong> urban pollution in Hanoi area on benthic communities collected from Red, Nhue and<br />
Tolich Rivers in Vietnam. They also reported that the diatom assemblage at the Tolich site<br />
consisted mainly <strong>of</strong> Nitzschia umbonata, N. palea and Eolimma minima. Some <strong>of</strong> these diatoms<br />
were reported to have preference for tropical regions without being restricted to these latitudes [23-<br />
24]. Cymbella turgidula was recorded in many tropical countries such as Sri Lanka [25] and Río<br />
Savegre River in the central and southern parts <strong>of</strong> Costa Rica [26]. Cymbella tumida was also<br />
reported to be widespread but very <strong>of</strong>ten found in the tropics [11]. Patrick and Reimer [27] reported<br />
this species from New England. Diploneis subovalis was found in Iceland and Finland, but with<br />
higher frequency in tropical rivers [11]. Gomphonema parvulum var. lagenula was regarded as a<br />
tropical species [26].<br />
Common species, especially Cymbella turgidula, C. tumida and Gomphonema parvulum<br />
var. lagenula (currently regarded as a synonym <strong>of</strong> Gomphonema lagenula) were found in many<br />
streams and rivers in Thailand [8, 28-38].<br />
Water Physico-Chemical Properties<br />
The water physico-chemical properties <strong>of</strong> Mekong River and its tributaries between July<br />
2005 - April 2007 are shown in Table 2. Broad differences were apparent between sampling sites.<br />
The water temperature ranged between 25-32 o C, the temperatures at GT, KO and HK in the north<br />
being lower than those in the north-eastern areas. The water velocity was highest (8.30 m/s) at KB<br />
in the Mekong River. All sampling sites showed neutral pH with the highest (7.67) at GT and the<br />
lowest (6.75) at SK. Average alkalinity ranged between 23.3-67.8 mg/L. The highest value <strong>of</strong><br />
conductivity was recorded at SK in the north-eastern area. The highest DO and BOD values were<br />
observed at KO in the north-eastern area, as were the highest NO 3 - -N (1.57 mg/L) and SRP (0.24<br />
mg/L). The highest NH 4 + -N (0.53 mg/L) was recorded at KP and the lowest (0.21 mg/L) at GT.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
33<br />
Figure 3. Common species <strong>of</strong> benthic diatoms in Mekong River and its tributaries<br />
(scale bar = 10 m)<br />
(1) Nitzschia filiformis (W. Smith) Hustedt, (2) Nitzschia palea (Kützing) W. Smith,<br />
(3-4) Nitzschia inconspicua Grunow, (5-7) Nitzschia supralitorea Lange-Bertalot,<br />
(8-9) Nitzschia clausii Hantzsch, (10) Nitzschia dissipata (Kützing) Grunow,<br />
(11) Nitzschia microcephala Grunow, (12-13) Navicula menisculus Schumann,<br />
(14) Navicula symmetrica R.M. Patrick, (15) Navicula cryptotenelloides Lange-<br />
Bertalot, (16) Navicula cryptotenella Lange-Bertalot, (17) Navicula rostellata Kützing,<br />
(18-19) Fragilaria bidens Heiberg, (20) Ulnaria ulna (Nitzsch) P. Compère
34 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
Figure 3 (continued). Common species <strong>of</strong> benthic diatoms in Mekong River and its tributaries<br />
(scale bar = 10 m)<br />
(1) Frustulia undosa D. Metzeltin & H. Lange-Bertalot , (2-3) Geissleria decussis<br />
(Østrup) Lange-Bertalot&Metzeltin, (4-6) Gomphonema lagenula Kützing,<br />
(7-8) Sellaphora pupula (Kützing) Mereschkovsky, (9-10) Luticola goeppertiana<br />
(Bleisch) D.G.Mann in Round,Crawford&Mann, (11) Achnanthidium convergens<br />
(H. Kobayasi) H. Kobayasi, (12-14) Achnanthidium minutissimum (Kützing) Czarnecki,<br />
(15-16) Eolimna minima (Grunow) Lange-Bertalot, (17-18) Eolimna subminuscula<br />
(Manguin) Gerd Moser, (19-20) Mayamaea atomus (Kützing) H. Lange-Bertalot,<br />
(21-22) Cymbella sumatrensis Hustedt, (23) Cymbella sp.1, (24) Encyonema sp.1,<br />
(25-26) Planothidium frequentissimum (Lange-Bertalot) Round & L. Bukhtiyarova,<br />
(27) Melosira varians C. Agardh
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
35<br />
The water quality in the Mekong River and its tributaries was classified in the fourth<br />
category according to the standards for surface water quality <strong>of</strong> Thailand certified by the Pollution<br />
Control Department [39], and can be used for household consumption after a disinfection process<br />
and special water treatment.<br />
Table 2. Physicochemical properties <strong>of</strong> water at 14 sampling sites (average values and min-max<br />
values, n=14)<br />
Sampling<br />
Site<br />
Temperature<br />
( o C)<br />
Velocity<br />
(m/s)<br />
pH<br />
Conductivity<br />
(μS/cm)<br />
Turbidity<br />
(NTU)<br />
Alkalinity<br />
(mg/L as CaCO 3)<br />
DO<br />
(mg/L)<br />
BOD<br />
(mg/L)<br />
NO 3 - -N<br />
(mg/L)<br />
NH 4 + -N<br />
(mg/L)<br />
SRP<br />
(mg/L)<br />
GT<br />
25.6<br />
(18.1-30.8)<br />
5.4<br />
(1.8-9.2)<br />
7.7<br />
(7.3-8.0)<br />
218.8<br />
(153.0-312.0)<br />
137.1<br />
(54.0-301.0)<br />
55.70<br />
(9.90-103.00)<br />
8.00<br />
(5.40-10.00)<br />
2.60<br />
(1.80-4.10)<br />
0.78<br />
(0.10-1.90)<br />
0.21<br />
(0.04-0.47)<br />
0.14<br />
(0.06-0.55)<br />
KO<br />
26.0<br />
(20.5-29.2)<br />
5.8<br />
(4.0-7.8)<br />
7.2<br />
(6.9-7.8)<br />
138.5<br />
(78.9-198.2)<br />
199.1<br />
(57.0-305.0)<br />
41.05<br />
(6.90-70.20)<br />
8.20<br />
(5.20-11.00)<br />
3.10<br />
(2.00-5.00)<br />
1.57<br />
(0.80-3.00)<br />
0.37<br />
(0.29-0.46)<br />
0.24<br />
(0.10-1.19)<br />
HK<br />
25.3<br />
(18.5-29.8)<br />
7.8<br />
(3.0-12.3)<br />
7.6<br />
(7.2-8.1)<br />
221.0<br />
(162.0-270.0)<br />
212.6<br />
(48.0-419.0)<br />
60.80<br />
(17.60-97.00)<br />
7.80<br />
(5.20-9.80)<br />
1.50<br />
(0.20-3.10)<br />
1.55<br />
(0.20-3.40)<br />
0.31<br />
(0.13-0.67)<br />
0.15<br />
(0.02-0.67)<br />
HG<br />
30.2<br />
(23.3-35.8)<br />
6.7<br />
(0.5-14.3)<br />
7.6<br />
(7.2-8.1)<br />
180.4<br />
(64.0-389.0)<br />
82.5<br />
(3.0-438.0)<br />
57.80<br />
(19.00-101.00)<br />
7.00<br />
(4.20-8.40)<br />
2.70<br />
(0.40-6.40)<br />
0.88<br />
(0.00-1.60)<br />
0.42<br />
(0.15-0.72)<br />
0.22<br />
(0.10-0.61)<br />
KK<br />
28.9<br />
(23.9-32.8)<br />
3.5<br />
(0.1-5.6)<br />
7.6<br />
(7.0-8.7)<br />
192.8<br />
(49.0-325.0)<br />
205.2<br />
(105.0-453.0)<br />
67.80<br />
(35.00-104.00)<br />
5.90<br />
(4.80-8.60)<br />
2.30<br />
(0.00-5.60)<br />
1.20<br />
(0.20-2.00)<br />
0.28<br />
(0.14-0.43)<br />
0.19<br />
(0.14-0.29)<br />
PS<br />
30.0<br />
(26.0-33.0)<br />
2.74<br />
(0.0-6.2)<br />
7.4<br />
(7.0-7.8)<br />
249.7<br />
(147.0-349.0)<br />
183.4<br />
(96.0-367.0)<br />
65.80<br />
(25.00-107.00)<br />
5.60<br />
(4.40-8.20)<br />
1.20<br />
(0.20-4.40)<br />
1.12<br />
(0.60-1.80)<br />
0.38<br />
(0.16-0.58)<br />
0.19<br />
(0.07-0.26)<br />
LG<br />
31.5<br />
(26.5-34.1)<br />
1.3<br />
(0.0-6.3)<br />
7.2<br />
(6.4-8.1)<br />
292.0<br />
(146.0-401.0)<br />
53.3<br />
(12.0-125.0)<br />
36.00<br />
(8.00-85.00)<br />
4.40<br />
(2.60-5.80)<br />
1.30<br />
(0.10-3.00)<br />
0.79<br />
(0.50-1.10)<br />
0.41<br />
(0.20-0.74)<br />
0.20<br />
(0.07-0.40)<br />
NP<br />
30.6<br />
(25.1-32.9)<br />
1.4<br />
(0.0-4.4)<br />
7.4<br />
(6.7-7.8)<br />
195.7<br />
(91.0-263.0)<br />
127.4<br />
(28.0-266.0)<br />
61.20<br />
(20.50-95.00)<br />
5.70<br />
(4.00-8.20)<br />
1.70<br />
(0.20-4.90)<br />
0.76<br />
(0.00-1.30)<br />
0.36<br />
(0.17-0.55)<br />
0.13<br />
(0.00-0.33)<br />
SK<br />
31.4<br />
(27.0-34.8)<br />
3.7<br />
(0.1-10.4)<br />
6.8<br />
(6.0-7.5)<br />
341.1<br />
(127.0-759.0)<br />
70.1<br />
(10.0-288.0)<br />
23.30<br />
(4.10-45.00)<br />
4.60<br />
(2.20-7.00)<br />
1.60<br />
(0.00-3.60)<br />
0.99<br />
(0.20-1.80)<br />
0.40<br />
(0.18-0.73)<br />
0.09<br />
(0.01-0.20)<br />
KB<br />
30.4<br />
(24.5-32.6)<br />
8.3<br />
(2.0-16.5)<br />
7.5<br />
(6.8-8.0)<br />
193.2<br />
(133.5-255.0)<br />
134.1<br />
(23.0-271.0)<br />
51.40<br />
(17.00-92.00)<br />
6.00<br />
(4.60-8.40)<br />
1.90<br />
(0.30-4.60)<br />
0.68<br />
(0.10-1.90)<br />
0.30<br />
(0.14-0.54)<br />
0.15<br />
(0.04-0.29)<br />
HW<br />
31.0<br />
(27.1-34.8)<br />
6.9<br />
(0.1-14.9)<br />
7.4<br />
(6.8-7.9)<br />
166.3<br />
(66.0-255.0)<br />
136.0<br />
(16.0-303.0)<br />
49.20<br />
(18.50-87.00)<br />
6.10<br />
(4.60-8.60)<br />
1.10<br />
(0.00-4.90)<br />
0.55<br />
(0.10-1.20)<br />
0.31<br />
(0.10-0.60)<br />
0.15<br />
(0.00-0.37)<br />
KH<br />
30.3<br />
(22.4-36.4)<br />
2.7<br />
(0.0-7.1)<br />
7.5<br />
(6.8-7.8)<br />
134.9<br />
(43.0-250.0)<br />
144.1<br />
(22.0-325.0)<br />
52.50<br />
(20.00-90.00)<br />
5.89<br />
(4.00-8.40)<br />
1.40<br />
(0.10-4.80)<br />
0.55<br />
(0.10-1.10)<br />
0.31<br />
(0.16-0.61)<br />
0.17<br />
(0.00-0.70)<br />
KP<br />
32.1<br />
(28.0-36.0)<br />
5.2<br />
(0.1-8.8)<br />
6.9<br />
(6.2-7.9)<br />
175.7<br />
(80.0-265.0)<br />
72.6<br />
(13.0-119.0)<br />
43.90<br />
(10.00-73.20)<br />
4.86<br />
(3.60-7.20)<br />
1.44<br />
(0.00-5.20)<br />
0.65<br />
(0.00-1.90)<br />
0.53<br />
(0.17-0.85)<br />
0.10<br />
(0.02-0.30)<br />
KJ<br />
31.5<br />
(27.0-34.8)<br />
1.7<br />
(0.0-4.7)<br />
7.9<br />
(6.6-8.2)<br />
141.8<br />
(44.0-243.0)<br />
119.9<br />
(14.0-260.0)<br />
59.80<br />
(21.00-85.00)<br />
6.29<br />
(4.60-8.50)<br />
2.36<br />
(0.60-5.10)<br />
0.68<br />
(0.00-1.30)<br />
0.40<br />
(0.11-1.07)<br />
0.14<br />
(0.01-0.34)<br />
Correlation between Physico-Chemical Variables<br />
There were significant positive and negative correlations between some <strong>of</strong> the physicochemical<br />
variables <strong>of</strong> water as shown in Table 3. Significant positive correlation between SRP and<br />
NO 3 - -N was observed (P
36 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
correlation with NH 4 + -N (P
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
37<br />
Figure 4. CCA <strong>of</strong> the relationship between water quality and common species <strong>of</strong> diatom (% <strong>of</strong><br />
relative abundance > 1)<br />
There are many environmental conditions which influence the growth <strong>of</strong> algae. In a lotic<br />
ecosystem, almost all algae live in benthic forms [40] and they usually grow on any surface <strong>of</strong> the<br />
substratum. Each type <strong>of</strong> substratum, either rock, mud, sand or silt, affects the benthic behaviour<br />
[41]. The substratum type is related to the current velocity and water volume [33]. The substratum<br />
characteristic has a direct impact on the distribution <strong>of</strong> benthic algae [42]. In this study, the lowest<br />
amounts <strong>of</strong> diatoms at all sampling times were recorded at HK sampling site (Figure 2). This<br />
observation was probably due to a high level <strong>of</strong> water in the wet season and the substrata along the<br />
bank being s<strong>of</strong>t sediment <strong>of</strong> sand and silt, which is not suitable for diatom growth [43].<br />
Furthermore, blooms <strong>of</strong> macroalgae, namely Cladophora spp. and Microspora spp., appeared in<br />
cool dry season and there were fewer suitable substrata for the attachment <strong>of</strong> benthic diatoms.<br />
Mpawenayo and Mathooko [44] published the structure <strong>of</strong> the diatom assemblage associated with<br />
Cladophora macroalgae and sediment in a highland stream <strong>of</strong> Kenya. The dominant species <strong>of</strong><br />
diatoms were Nitzschia amphibia and Gomphonema parvulum. Similary, G. parvulum was found at<br />
HK where there was a full bloom <strong>of</strong> Cladophora spp. Jüttner et al. [45] reported that G. parvulum in<br />
particular preferred vegetation as its habitat. However, G. parvulum was found equally <strong>of</strong>ten on<br />
sediment because <strong>of</strong> transportation and contamination. Persistent stability and protection <strong>of</strong> diatoms<br />
attached on Cladophora could explain the existence <strong>of</strong> higher diatom species richness on<br />
Cladophora than in a more unstable sediment habitat.
38 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
In fast-flowing water, there are diatoms such as Achnanthes and Cocconeis that can attach<br />
on rocks and other hard surfaces [24]. Similarly, in this study, Cocconeis spp. were found as<br />
dominant species at HG during summer 2007 with high water velocity. In slower flowing water,<br />
Melosira varians and various species <strong>of</strong> Synedra, Gomphonema and Cymbella are found on hard<br />
substrata [24]. Throughout the sampling period, Gomphonema spp. were also observed at LG<br />
located about 100 metres downstream from a small dam, where the flow rate <strong>of</strong> water was<br />
extremely low.<br />
Species <strong>of</strong> Benthic Diatoms in Relation to Physico-Chemical Properties <strong>of</strong> Water by Cluster<br />
Analysis<br />
Twenty nine common diatom species were arranged in groups <strong>of</strong> sampling sites detected by<br />
cluster analysis <strong>of</strong> physico-chemical parameters <strong>of</strong> water quality at 80% similarity. The number <strong>of</strong><br />
benthic diatoms in each group was calculated by Minitab program to find the significant difference<br />
and significant correlation between groups. The relationship between diatom community<br />
composition and water quality variables was determined using statistical techniques. Each sampling<br />
site had distinct water physico-chemical properties. At 80% similarity, the dendrogram divided<br />
sampling sites into eleven distinct clusters <strong>of</strong> characteristic water quality types: groups A to I 2<br />
(Figure 5 and Table 4). It was found that all sampling sites in groups A (n=3) and B (n=9) and some<br />
in group C (n=3) were those during December 2006 (cool dry season). Group E, the biggest group<br />
(n=54), was composed <strong>of</strong> sampling sites during May 2006 (summer) and some sites in two cool dry<br />
seasons.<br />
The values <strong>of</strong> physico-chemical parameters in each group were calculated by Minitab<br />
program to find significant differences and correlations between groups A to I 2 as shown in Table 5.<br />
It was found that groups A, C, G and H showed a non-significant correlation with NO 3 - -N<br />
concentration. Group F 1 showed higher concentrations <strong>of</strong> NO 3 - -N (average <strong>of</strong> 1.33 mg/L) than<br />
groups B, D, E, F 2 and I 1 (P
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
39<br />
UPGMA<br />
UPGMA<br />
40 50 60 70 80 90 100<br />
Per cent. Similarity<br />
Percent Similarity<br />
SK 5<br />
SK 5<br />
SK 5<br />
KJ 5<br />
KJ 5<br />
KJ 5<br />
KH5<br />
KH5<br />
KH5<br />
KK 5<br />
KK 5<br />
KK 5<br />
HG5<br />
HG5<br />
HG5<br />
SK 4<br />
SK 4<br />
SK 4<br />
LG4<br />
LG4<br />
LG4<br />
KP 5<br />
KP 5<br />
KP 5<br />
HG4<br />
HG4<br />
HG4<br />
KP 4<br />
KP 4<br />
KP 4<br />
KP 1<br />
KB 3<br />
KB 3<br />
KH3<br />
KH3<br />
KH3<br />
KB 3<br />
HG3<br />
HG3<br />
HG3<br />
KO2<br />
NP 5<br />
NP 5<br />
NP 5<br />
KP 3<br />
KP 3<br />
HK 3<br />
HK 3<br />
HK 3<br />
KJ 2<br />
KH2<br />
HW 2<br />
KB 2<br />
KK 2<br />
NP 2<br />
PS 2<br />
KK 3<br />
KO3<br />
KO3<br />
KO3<br />
KP 2<br />
KB 5<br />
KB 5<br />
KB 5<br />
SK 3<br />
SK 3<br />
SK 3<br />
KP 3<br />
NP 3<br />
NP 3<br />
LG2<br />
HW 5<br />
HW 5<br />
HW 5<br />
KJ 3<br />
KJ 3<br />
KJ 3<br />
NP 3<br />
GT 3<br />
GT 3<br />
GT 3<br />
LG3<br />
LG3<br />
LG3<br />
LG6<br />
LG6<br />
LG6<br />
HG6<br />
HG6<br />
HG6<br />
PS 5<br />
PS 5<br />
PS 5<br />
KK 3<br />
HG2<br />
PS 3<br />
PS 3<br />
PS 3<br />
KK 3<br />
HK 6<br />
HK 6<br />
HK 6<br />
PS 6<br />
PS 6<br />
PS 6<br />
KK 6<br />
KK 6<br />
KK 6<br />
GT 6<br />
GT 6<br />
GT 6<br />
SK 1<br />
LG1<br />
LG5<br />
LG5<br />
LG5<br />
KP 6<br />
KP 6<br />
KP 6<br />
SK 6<br />
SK 6<br />
SK 6<br />
NP 6<br />
NP 6<br />
KJ 6<br />
KJ 6<br />
KJ 6<br />
HW 6<br />
HW 6<br />
HW 6<br />
KH6<br />
KH6<br />
KH6<br />
KB 6<br />
KB 6<br />
KB 6<br />
NP 6<br />
HK 2<br />
GT 5<br />
GT 5<br />
GT 5<br />
GT 2<br />
SK 2<br />
KO5<br />
KO5<br />
KO5<br />
KO1<br />
KH4<br />
KH4<br />
KH4<br />
HW 4<br />
HW 4<br />
KJ 4<br />
HW 4<br />
NP 4<br />
NP 4<br />
KJ 4<br />
KJ 4<br />
NP 4<br />
KO4<br />
KO4<br />
KO4<br />
PS 1<br />
HG1<br />
KK 1<br />
HK 1<br />
KK 4<br />
KK 4<br />
KK 4<br />
PS 4<br />
PS 4<br />
PS 4<br />
HK 4<br />
HK 4<br />
HK 4<br />
HW 1<br />
KB 4<br />
KB 4<br />
KB 4<br />
GT 4<br />
GT 4<br />
GT 4<br />
KH1<br />
KB 1<br />
KJ 1<br />
NP 1<br />
HK 5<br />
HK 5<br />
HK 5<br />
KO6<br />
KO6<br />
KO6<br />
GT 1<br />
I2<br />
I1<br />
H G F2<br />
F1<br />
E<br />
D<br />
C<br />
B<br />
A<br />
Figure 5. Dendrogram <strong>of</strong> similarities between investigated sites according to physico-chemical<br />
parameters <strong>of</strong> water; 252 cases, 11 variables
40 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
Table 4. Groups <strong>of</strong> sampling sites detected by cluster analysis <strong>of</strong> physico-chemical parameters <strong>of</strong><br />
water quality at 80% similarity<br />
Group Sampling site Description<br />
A SK5.1, SK5.2, SK5.3 Tributaries in north-eastern Thailand<br />
in cool dry season (second year)<br />
B KK5.1, KK5.2, KK5.3, KH5.1, KH5.2, KH5.3, KJ5.1, KJ5.2,<br />
KJ5.3<br />
Mekong River in north-eastern Thailand in cool<br />
dry season (second year)<br />
C HG5.1, HG5.2, HG5.3 Tributaries in north-eastern Thailand<br />
in cool dry season (second year)<br />
D HG4.1, HG4.2, HG4.3, LG4.1, LG4.2, LG4.3, SK4.1, SK4.2,<br />
SK4.3, KP5.1, KP5.2, KP5.3<br />
Tributaries in north-eastern Thailand<br />
in summer (second year) and cool dry season<br />
(second year)<br />
E GT3.1, GT3.2, GT3.3, KO2, KO3.1, KO3.2, KO3.3, HK3.1,<br />
HK3.2, HK3.3, HG3.1, HG3.2, HG3.3,KK2, KK3.2, PS2, LG2,<br />
Mekong River and its tributaries in summer (first<br />
year) and some <strong>of</strong> two cool dry seasons<br />
NP2, NP3.1, NP3.2,NP3.3, NP5.1, NP5.2, NP5.3, SK3.1, SK3.2,<br />
SK3.3, KB2, KB3.1, KB3.2, KB3.3, KB5.1, KB5.2, KB5.3,<br />
HW2, HW5.1, HW5.2, HW5.3, KH2, KH3.1, KH3.2, KH3.3,<br />
KP1, KP2, KP3.1, KP3.2, KP3.3, KP4.1, KP4.2, KP4.3, KJ2,<br />
KJ3.1, KJ3.2, KJ3.3<br />
F 1 GT6.1,GT6.2,GT6.3, HK6.1,HK6.2,HK6.3,HG2, HG6.1, HG6.2,<br />
HG6.3, KK3.1, KK3.3, K6.1, KK6.2, KK6.3, PS3.1, PS3.2,<br />
PS3.3,PS5.1,PS5.2,PS5.3, PS6.1, PS6.2, PS6.3,LG3.1, LG3.2,<br />
LG3.3,LG6.1, LG6.2, LG6.3<br />
Mekong River and its tributaries in two summers<br />
F 2 GT2, GT5.1, GT5.2, GT5.3, HK2, LG1, LG5.1,LG5.2, LG5.3,<br />
NP6.1, NP6.2, NP6.3, SK1, SK6.1, SK6.2, SK6.3, KB6.1,<br />
KB6.2, KB6.3, HW6.1, HW6.2, HW6.3, KH6.1, KH6.2, KH6.3,<br />
Mekong River and its tributaries in summer<br />
(second year) and some <strong>of</strong> cool dry season (second<br />
year)<br />
KP6.1, KP6.2, KP6.3, KJ6.1, KJ6.2,KJ6.3<br />
G SK2 Tributaries in north-eastern Thailand<br />
in cool dry season (first year)<br />
H KO1, KO5.1, KO5.2, KO5.3 Tributaries in northern Thailand<br />
in rainy season (first year) and cool dry season<br />
(second year)<br />
I 1 KO4.1, KO4.2, KO4.3, NP4.1, NP4.2, NP4.3,HW4.1, HW4.2,<br />
HW4.3, KH4.1, KH4.2, KH4.3, KJ4.1, KJ4.2, KJ4.3<br />
Mekong River and its tributaries in rainy season<br />
(second year)<br />
I 2 GT1, GT4.1, GT4.2, GT4.3, KO6.1, KO6.2, KO6.3, HK1,<br />
HK4.1, HK4.2, HK4.3, HK5.1, HK5.2, HK5.3, HG1, KK1,<br />
KK4.1, KK4.2, KK4.3, PS1, PS4.1, PS4.2, PS4.3, NP1,KB1,<br />
KB4.1, KB4.2, KB4.3, HW1, KH1, KJ1<br />
Mekong River and its tributaries in two rainy<br />
seasons and some <strong>of</strong> cool dry season (second year)<br />
and summer (second year)<br />
There were significant differences (P
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
41<br />
turbidity (average <strong>of</strong> 4.33 NTU) was recorded for group C while the highest (average <strong>of</strong> 314.16<br />
NTU) was recorded for group I 2 .<br />
_<br />
Table 5. Values <strong>of</strong> average (X) and standard error (se) <strong>of</strong> physico-chemical parameters (P
42 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
Table 5. (continued)<br />
X = 0.82<br />
Parameter F 1 F 2 G H I 1 I 2<br />
NO — 3 N X = 1.33<br />
X = 1.80<br />
X =1.38<br />
X =0.68<br />
X =1.04<br />
cor F 1>B,D,E,F 2,I 1 cor F 2 < F 1<br />
cor ns<br />
cor ns<br />
cor I 1 < F 1<br />
cor I 2>D<br />
(mg/L)<br />
se = 0.08<br />
n = 30<br />
se = 0.08<br />
n = 31<br />
se = 0<br />
n = 1<br />
se = 0.54<br />
n=4<br />
se = 0.23<br />
n=15<br />
se=0.15<br />
n=31<br />
NH + 4 -N<br />
(mg/L)<br />
X = 0.32<br />
se = 0.03<br />
n = 30<br />
cor F 1=F 2
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
43<br />
CONCLUSIONS<br />
The number <strong>of</strong> benthic diatom species found (252 species from 53 genera) in Mekong River<br />
and its tributaries in north and north-eastern Thailand is similar to those reported for other river<br />
systems <strong>of</strong> comparable size. Determination <strong>of</strong> the relationship between diatom community<br />
composition and water quality variables shows that many species, notably Achnanthidium<br />
minutissimum, A. convergens, Navicula menisculus, Nitzschia clausii, Luticola goeppertiana,<br />
Eolimna minima, Ulnaria ulna and Mayamaea atomus, can act as indicators <strong>of</strong> some <strong>of</strong> the riverwater<br />
qualities, viz. conductivity, alkalinity, NH 4 + -N, NO 3<br />
-<br />
-N, soluble reactive phosphorus, and<br />
BOD.<br />
ACKNOWLEDGEMENTS<br />
The authors would like to thank Thailand Research Fund (TRF) for its support in the form <strong>of</strong><br />
a research grant (Grant No. PHD/0129/2547) through the Royal Golden Jubilee PhD Program. They<br />
are also grateful to Dr. Ingrid Jüttner from National Museum Cardiff, U.K. for her suggestions and<br />
assistance in diatom identification.<br />
REFERENCES<br />
1. A. Chovanec, P. Jäger, M. Jungwirth, V. Koller-Kreimel, O. Moog, S. Muhar and S. Schmutz,<br />
“The Austrian way <strong>of</strong> assessing the ecological integrity <strong>of</strong> running waters – a contribution to<br />
the EU water framework directive”, Hydrobiol., 2000, 422/423, 445-452.<br />
2. F. E. Round, “Use <strong>of</strong> diatoms for monitoring rivers”, in “Use <strong>of</strong> Algae for Monitoring Rivers”<br />
(Ed. B. A. Whitton, E. Rott and G. Friedrich), Institut fur Botanik, Universitat, Innsbruck,<br />
1991, pp.25-32.<br />
3. B. A. Whitton, E. Rott and G. Friedrich, “Use <strong>of</strong> algae for monitoring rivers”, Proceedings <strong>of</strong><br />
<strong>International</strong> Symposium on Use <strong>of</strong> Algae for Monitoring Rivers I, 1991, Düsseldorf,<br />
Germany, pp.10-18.<br />
4. E. Rott, E. Pipp and P. Pfister, “Diatom methods developed for river quality assessment in<br />
Austria and a cross-check against numerical trophic indication methods used in Europe”, Algol.<br />
Stud., 2003, 110, 91-115.<br />
5. J. Prygiel, M. Coste and J. Bukowska, “Review <strong>of</strong> the major diatom-based techniques for the<br />
quality assessment <strong>of</strong> river-state <strong>of</strong> the art in Europe”, Proceedings <strong>of</strong> <strong>International</strong> Symposium<br />
on Use <strong>of</strong> Algae for Monitoring Rivers III, 1999, Douai, France, pp.224-238.<br />
6. M. Ohno, H. Fukushima and T. Ko-Bayashi, “Diatom flora <strong>of</strong> the Mekong water system,<br />
Cambodia”, Nat. Sci., 1972, 20, 1-11.<br />
7. Mekong River Commission, “Biomonitoring <strong>of</strong> the lower Mekong River and selected<br />
tributaries”, MRC technical paper No.13, Mekong River Comission, Vientiane, Lao PDR,<br />
2006.<br />
8. S. Pruethiworanon, “Diversity <strong>of</strong> macroalgae and benthic diatoms in Mekong River passing<br />
Thailand and their application for water quality”, MS Thesis, 2008, Chiang Mai University,<br />
Thailand.
44 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
9. I. Renberg, “A procedure for preparing large sets <strong>of</strong> diatom slides from sediment cores”, J.<br />
Paleolimnol., 1990, 4, 87-90.<br />
10. M. G. Kelly, A. Cazaubon, E. Coring, A. Dell’Uomo, L. Ector, B. Goldsmit, H. Guasch, J.<br />
Hürlimann, A. Jarlman, B. Kaweeka, J. Kwandrans, R. Laugaste, E.- A. Lindstrom, M. Leitao,<br />
P. Marvan, J. Padisak, E. Pipp, J. Prygiel, E. Rott, S. Sabater, H. van Dam and J. Vizinet,<br />
“Recommendations for the routine sampling <strong>of</strong> diatoms for water quality assessments in<br />
Europe”, J. Appl. Phycol., 1998, 10, 215-224.<br />
11. K. Krammer and H. Lange-Bertalot, “Bacillariophyceae Teil 1: Naviculaceae, Süsswasserflora<br />
von Mitteleuropa Band 2/1”, Gustav Fisher Verlag, Stuttgart, 1986.<br />
12. K. Krammer and H. Lange-Bertalot, “Bacillariophyceae Teil 2: Epithemiaceae Surirellaceae,<br />
Süsswasserflora von Mitteleuropa Band 2/2”, Gustav Fisher Verlag, Stuttgart, 1988.<br />
13. K. Krammer and H. Lange-Bertalot, “Bacillariophyceae Teil 3: Centrales Fragilariaceae<br />
Eunotiaceae, Süsswasserflora von Mitteleuropa Band 2/3”, Gustav Fisher Verlag, Stuttgart,<br />
1991a.<br />
14. K. Krammer and H. Lange-Bertalot, “Bacillariophyceae Teil 4: Achnanthaceae,<br />
Süsswasserflora von Mitteleuropa Band 2/4”, Gustav Fisher Verlag, Stuttgart, 1991b.<br />
15. H. Lange-Bertalot, “Diatoms <strong>of</strong> Europe”, A.R.G. Gantner Verlag K.G., Ruggell, 2001.<br />
16. T. Watanabe, T. Ohtsuka, A. Tuji and A. Houki, “Picture Book and Ecology <strong>of</strong> the Freshwater<br />
Diatoms”, Uchida-rokakuho, Tokyo, 2005.<br />
17. H. Lange-Bertalot, “Iconographia Diatomologica: Annotated Diatom Micrographs”, Vol.13,<br />
A.R.G. Gantner Verlag K.G., Ruggell, 2004.<br />
18. A. D. Eaton, L. S. Clesceri, E. W. Rice and A. E. Greenberg, “Standard Methods for<br />
Examination <strong>of</strong> Water and Wastewater: Centennial Edition”, 21 st Edn., American Public Health<br />
Association, Washington D. C., 2005.<br />
19. N. T. H. Holmes and B. A. Whitton, “Phytobenthos <strong>of</strong> the River Tees and its tributaries”,<br />
Freshwater Biol., 1981, 11, 139-163.<br />
20. R. E. M. Archibald, “The diatoms <strong>of</strong> the Sundays and Great Fish Rivers in the Eastern Cape<br />
Province <strong>of</strong> South Africa”, A.R.G. Gantner Verlag K.G., Ruggell, 1983.<br />
21. B. C. Chessman, “Diatom flora <strong>of</strong> an Australian river system: Spatial patterns and environment<br />
relationships”, Freshwater Biol., 1986, 16, 805-819.<br />
22. I. Jüttner, S. Sharma, B. M. Dahal, S. J. Ormerod, P. J. Chimonides and E. J. Cox, “Diatoms as<br />
indicators <strong>of</strong> stream quality in the Kathmandu Valley and Middle Hills <strong>of</strong> Nepal and India”,<br />
Freshwater Biol., 2003, 48, 2065-2084.<br />
23. T. T. Duong, M. Coste, A. Feurtet-Mazel, D. K. Dang, C. Gold, Y. S. Park and A. Boudou,<br />
“Impact <strong>of</strong> urban pollution from the Hanoi area on benthic diatom communities collected from<br />
the Red, Nhue and Tolich Rivers (Vietnam)”, Hydrobiol., 2006, 563, 201-216.<br />
24. R. Patrick, “Ecology <strong>of</strong> freshwater diatoms and diatom communities”, in “The Biology <strong>of</strong><br />
Diatoms” (Ed. D. Werner), Blackwell Scientific Publications, Oxford, 1977, pp.284-322.<br />
25. N. Foged, “Freshwater Diatoms in SriLanka”, Odense Publisher, Odense (Denmark), 1976.<br />
26. A. M. Silva-Benavides, “The epilithic diatom flora <strong>of</strong> a pristine and a polluted river in Costa<br />
Rica, Central America”, Diatom Res., 1996, 11, 105-142.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
45<br />
27. R. Patrick and C. W. Reimer, “The Diatoms <strong>of</strong> the United States”, The Academy <strong>of</strong> Natural<br />
<strong>Science</strong> <strong>of</strong> Philadelphia, Philadelphia, 1966.<br />
28. T. Pekthong, “Diversity <strong>of</strong> phytoplankton and benthic algae in Mae Sa stream, Doi Suthep-Pui<br />
National Park, altitude 330-550 meters”, MS Thesis, 1998, Chiang Mai University, Thailand.<br />
29. T. Pekthong, “Fifty one new record species <strong>of</strong> freshwater diatoms in Thailand”, Chiang Mai J.<br />
Sci., 2002, 28, 97-112.<br />
30. P. Waiyaka, “Diversity <strong>of</strong> phytoplankton and benthic algae in Mae Sa stream, Doi Suthep-Pui<br />
National Park, altitude 550-1050 meters”, MS Thesis, 1998, Chiang Mai University, Thailand.<br />
31. T. Kunpradid, “Biodiversity <strong>of</strong> phytoplankton and macroalgae in Mae Sa stream, Doi Suthep-<br />
Pui National Park, Chiang Mai province”, MS Thesis, 2000, Chiang Mai University, Thailand.<br />
32. T. Kunpradid, “Diversity <strong>of</strong> macroalgae and benthic diatoms and their relationship with<br />
nutrient compounds in Ping and Nan Rivers”, PhD Thesis, 2005, Chiang Mai University,<br />
Thailand.<br />
33. Y. Peerapornpisal, T. Pekthong, P. Waiyaka and S. Promkutkaew, “Diversity <strong>of</strong> phytoplankton<br />
and benthic algae in Mae Sa stream, Doi Suthep-Pui National Park, Chiang Mai”, Siam Soc.,<br />
2000, 48, 193-211.<br />
34. S. Suphan, “Diversity <strong>of</strong> macroalgae and benthic diatoms in the area <strong>of</strong> Golden Jubilee Thong<br />
Pha Phum Project, Thong Pha Phum district, Kanchanaburi province”, MS Thesis, 2004,<br />
Chiang Mai University, Thailand.<br />
35. T. Inthasotti, “Seasonal distribution <strong>of</strong> macroalgae and benthic diatoms in Mae Tang River,<br />
Wiang Haeng district, Chiang Mai province between March 2004-January 2005”, Independent<br />
Study for Master Degree, 2006, Chiang Mai University, Thailand.<br />
36. T. Inthasotti, “Diversity <strong>of</strong> macroalgae and benthic diatoms in Kham watershed, Chiang Rai<br />
province”, MS Thesis, 2006, Chiang Mai University, Thailand.<br />
37. P. Leelahakriengkrai, “Usage <strong>of</strong> macroalgae and benthic diatoms to monitor the water quality<br />
<strong>of</strong> Kok River and tributaries”, Independent study for Master Degree, 2006, Chiang Mai<br />
University, Thailand.<br />
38. P. Leelahakriengkrai, “Diversity and usage <strong>of</strong> macroalgae and benthic diatoms to monitor water<br />
quality <strong>of</strong> Ping River, 2004-2005”, MS Thesis, 2006, Chiang Mai University, Thailand.<br />
39. National Environmental Board, “Surface water quality standards” (in Thai), the Royal<br />
Government Gazette, Vol. 111, Part 16 (February 24, 1994).<br />
40. H. B. N. Hynes, “The Ecology <strong>of</strong> Running Waters”, Liverpool University Press, Liverpool,<br />
1970, pp.53-77.<br />
41. S. Panha, “A new species <strong>of</strong> Amphidromus from Thailand (Stylommatophora: Camaenidae)”,<br />
Malacol. Rev., 1996, 29, 131-132.<br />
42. V. J. Chapman and D. J. Chapman, “The Algae”, Macmillan, London, 1973, pp.388-395.<br />
43. F. E. Round, “The Biology <strong>of</strong> the Algae”, 2 nd Edn., Edward Arnold Ltd., London, 1973, pp.64-<br />
69.<br />
44. B. Mpawenayo and J. M. Mathooko, “The structure <strong>of</strong> diatom assemblages associated with<br />
Cladophora and sediments in a highland stream in Kenya”, Hydrobiol., 2005, 544, 55-67.
46 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 28-46<br />
45. I. Jüttner, H. Rothfritz and S. J. Ormerod, “Diatoms as indicators <strong>of</strong> river quality in the<br />
Nepalese Middle Hills with consideration <strong>of</strong> the effects <strong>of</strong> habitat-specific sampling”,<br />
Freshwater Biol., 1996, 36, 475-486.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
Full Paper<br />
<strong>Maejo</strong> <strong>Maejo</strong> Int. J. Sci. Int. Technol. J. Sci. Technol. <strong>2012</strong>, 6(01), <strong>2012</strong>, 47-61 6(01), 47-61<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Fourteen new records <strong>of</strong> cercosporoids from Thailand<br />
Pheng Phengsintham 1,2 , Ekachai Chukeatirote 1 , Eric H. C. McKenzie 3 , Mohamed A. Moslem 4 ,<br />
Kevin D. Hyde 1, 4, * and Uwe Braun 5<br />
1 School <strong>of</strong> <strong>Science</strong>, Mae Fah Luang University, Chiang Rai 57100, Thailand<br />
2 Department <strong>of</strong> Biology, Faculty <strong>of</strong> <strong>Science</strong>s, National University <strong>of</strong> Laos, Laos<br />
3<br />
Landcare Research, Private Bag 92170, Auckland, New Zealand<br />
4 King Saud University, College <strong>of</strong> <strong>Science</strong>, Department <strong>of</strong> Botany and Microbiology, P.O. Box:<br />
2455, Riyadh 1145, Saudi Arabia<br />
5 Martin-Luther-Universität, Institut für Biologie, Bereich Geobotanik und Botanischer Garten,<br />
Herbarium, Neuwerk 21 D-06099 Halle/S, Germany<br />
* Corresponding author: kdhyde3@gmail.com<br />
Received: 24 July 2010 / Accepted: 27 January <strong>2012</strong> / Published: 1 February <strong>2012</strong><br />
Abstract: Comprehensive examination <strong>of</strong> cercosporoid leaf-spotting hyphomycetes was carried out<br />
in northern Thailand. Fourteen species assigned to the genera Cercospora (5), Passalora (3),<br />
Pseudocercospora (5) and Zasmidium (1) are new records for Thailand. Cercospora verniciferae<br />
and Zasmidium cassiicola are poorly known species and are fully described.<br />
Keywords:<br />
records<br />
anamorphic fungi, cercosporoid hyphomycetes, South-East Asia, taxonomy, new<br />
INTRODUCTION<br />
Cercospora sensu lato is one <strong>of</strong> the largest genera <strong>of</strong> hyphomycetes and is almost<br />
cosmopolitan in distribution. It causes leaf-spots and other lesions on a wide range <strong>of</strong> host plants.<br />
Species <strong>of</strong> this genus are important pathogens responsible for severe damage to beneficial plants such<br />
as maize, rice, grasses, vegetables, forest trees and ornamentals [1-3].<br />
There have been several recent comprehensive accounts <strong>of</strong> the fungi <strong>of</strong> Thailand which are<br />
among the best documented in the region [4, and references therein]. In Thailand, the study <strong>of</strong><br />
Cercospora and allied genera can be traced back to 1980 [5, 6]. Sixty cercosporoids were listed,<br />
including 13 unidentified species <strong>of</strong> Cercospora in the host index <strong>of</strong> plant diseases in Thailand [5],
48 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
whereas 21 species <strong>of</strong> Cercospora were specified as plant pathogens [6]. In 1989, 49 cercosporoid<br />
species were further identified in southern Thailand [7]. These reports, however, were based on the<br />
old generic characters, i.e. Cercospora sensu lato. Subsequently, 112 species <strong>of</strong> Cercospora as well<br />
as their synonyms were recorded in ‘The host index <strong>of</strong> plant diseases in Thailand’ [8]. It should also<br />
be noted that, according to the list, species names used were ambiguous since the criteria used for<br />
classification were based on both old and new criteria. In 2007, three new species <strong>of</strong> Cercospora<br />
were discovered; these included 11 species that were new to Thailand [9]. Forty-three cercosporoid<br />
species were included in an annotated list <strong>of</strong> cercosporoid fungi in northern Thailand [10], and two<br />
taxa associated with necrotic leaflets <strong>of</strong> an areca palm (Areca cathecu) were reported [11]. A PhD<br />
thesis on “Diversity and phylogeny <strong>of</strong> true cercosporoids fungi from northern Thailand” available in<br />
2009 encompassed 166 cercosporoids [12]. Recently, many new Cercospora species have been<br />
discovered, including Cercospora cristellae, a new cercosporoid species associated with the weed<br />
Cristella parasitica from northern Thailand [13], and three new records <strong>of</strong> cercosporoid fungi<br />
(Cercospora diplaziicola, Pseudocercospora duabangae and Pseudocercospora trematicola) from<br />
Thailand [14].<br />
The genus Cercospora Fresen. s. lat., which is one <strong>of</strong> the largest genera <strong>of</strong> hyphomycetes,<br />
has been monographed with over 3000 names [15]. Similar to other fungal group, the classification<br />
and identification <strong>of</strong> Cercosporoid fungi are mainly based on morphological characteristics.<br />
Currently, an identification key <strong>of</strong> Cercosporoid proposed in 2003 has been widely accepted [16]. In<br />
this present study, we explored the diversity <strong>of</strong> Cercospora and allied genera by using this<br />
identification key. The objectives <strong>of</strong> this paper are to investigate the cercosporoid fungi <strong>of</strong> Thailand<br />
and Laos, and to provide data on Thailand’s fungi in comparison with the diversity <strong>of</strong> these fungal<br />
groups in neighbouring countries.<br />
MATERIALS AND METHODS<br />
Sample Collections and Examination <strong>of</strong> Fungal Structures<br />
Plants leaves with leaf spots or other lesions were collected during field trips in northern<br />
Thailand. Photographs <strong>of</strong> symptoms, including fungal colonies or fruiting bodies, were taken.<br />
Macroscopic characters were observed using a stereomicroscope to check lesions/leaf spots<br />
(shape, size, colour, margin) and colonies/caespituli (with details, i.e. amphigenous/epiphyllous,<br />
punctiform/pustulate/inconspicuous, effuse, loose, dense, brown/blackish, and others.)<br />
Microscopic examination, measurement, description and presentation <strong>of</strong> drawings followed<br />
the standard procedures [16,17]. In the illustrations, thin-walled structures were depicted by a single<br />
line, thick-walled ones by double lines, and stippling was used to accentuate shape and pigmentation.<br />
Identification <strong>of</strong> Fungi<br />
The species <strong>of</strong> cercosporoid hyphomycetes were determined using the key available in the<br />
current taxonomic publications cited in the list <strong>of</strong> references.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
49<br />
Dried specimens were prepared and stored at the herbarium <strong>of</strong> the School <strong>of</strong> <strong>Science</strong>, Mae<br />
Fah Luang University (MFU). Duplicates were preserved at the herbarium <strong>of</strong> the Institute <strong>of</strong><br />
Biology, Geobotany and Botanical Garden, Halle (Saale), Germany (HAL).<br />
RESULTS AND DISCUSSION<br />
Fourteen cercosporoid hyphomycetes were identified and assigned to species <strong>of</strong> the genera<br />
Cercospora (5), Passalora (3), Pseudocercospora (5) and Zasmidium (1). The cercosporoid species<br />
and their habitats are listed in Table 1.<br />
Table 1. Cercosporoid species examined in this study<br />
Fungal species F DD G U<br />
Cercospora malloti ×<br />
Cercospora senecionicola ×<br />
Cercospora sidicola ×<br />
Cercospora verniciferae ×<br />
Cercospora zizphigena ×<br />
Passalora broussonetiae ×<br />
Passalora fusimasculans × ×<br />
Passalora graminis ×<br />
Pseudocercospora cycleae ×<br />
Pseudocercospora malloticola ×<br />
Pseudocercospora olacicola ×<br />
Pseudocercospora paederiae ×<br />
Pseudocercospora polysciatis ×<br />
Zasmidium cassiicola ×<br />
Note: F = fallow land; DD = dry dipterocarp forest; G = garden; U = urban area<br />
Cercospora malloti [18]<br />
Notes: The collection no. MFLU10-0310 from Tadsak waterfall, Ching Rai province agrees<br />
well with Cercospora malloti as previously circumscribed [15,16] (conidiophores 10–50 × 3–5 μm<br />
and conidia 40–75 × 1.5–3 μm). C. malloti is part <strong>of</strong> the C. apii Fesen. complex from which it is<br />
morphologically barely distinguishable [16]. The collection from Thailand has conidiophores which<br />
are 32–140 × 5–6 μm, and conidia which are 20–146 × 2–4 μm.<br />
Known hosts: Mallotus apelta (Lour.) Müll. Arg., M. japonicus (L. f.) Müll. Arg., and M.<br />
repandus (Rotter) Müll. Arg. (Euphorbiaceae) [15,16].<br />
Known distribution: Asia - China, Japan [15,16], Thailand (this paper); North America -<br />
USA (MS) [15,16].
50 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
Material examined: Phengsintham (MFLU10-0310) on leaf <strong>of</strong> Mallotus repandus<br />
(Euphorbiaceae) (Thailand: Chiang Rai province, Wiang Chiang Rung district, Tadsak waterfall), 23<br />
December 2009.<br />
Cercospora senecionicola [19]<br />
Notes: The collection no. MFLU10-0318 from Sri Pangsang village, Chiang Rai province<br />
agrees with Cercospora senecionicola as circumscribed in Chupp [15]. Brief description <strong>of</strong> the<br />
collection from Thailand: Leaf spots/Lesions circular to slightly irregular, 2–3 mm diam., at first<br />
dark green, later becoming brown to dark brown in the centre, dark brown margin.<br />
Caespituli/Colonies amphigenous, conspicuous, scattered, dark brown. Conidiophores single or<br />
fasciculate, arising from stromata (1–8 per fascicle), 0–5-geniculate, cylindrical, straight to curved,<br />
67–170 × 5–6 μm, 0–8-septate. Conidia acicular to obclavate, straight to curved, 17–82 × 4–7 μm,<br />
0–8-sepate, slightly constricted at the septa, hyaline to subhyaline, smooth, wall 0.3–0.5 μm thick,<br />
apex acute, based truncate to subtruncate, hila 2–3 μm wide, wall <strong>of</strong> the hila 0.5 μm wide, darkened.<br />
Known hosts: Senecio aureus L., S. aureus var. balamitea Toir. & Gray, and S. walkeri Arn.<br />
(Compositae = Asteraceae) [15,16].<br />
Known distribution: Asia - China, Laos, Thailand [15].<br />
Material examined: 1) Phengsintham (MFLU10-0318) on leaf <strong>of</strong> Senecio walkeri<br />
(Asteraceae) (Thailand: Chiang Rai province, Muang district, Sri Pangsang village), 11 August 2009;<br />
2) Phengsintham (P567) on leaf <strong>of</strong> Senecio walkeri (Asteraceae) (Laos: Bokeo province, Phimonsine<br />
village), 20 February 2010.<br />
Cercospora sidicola [20]<br />
Notes: The collection no. MFLU10-0312 from Tadsak waterfall, Chiang Rai province agrees<br />
with Cercospora sidicola as circumscribed by Chupp [15], but differs in formation <strong>of</strong> leaf spots.<br />
Known hosts: Sida acuta Burm. F., S. cordifolia (DC.) Fryxell , S. mysorensis Wight & Arn.,<br />
S. rhombifolia L., S. spinosa L., and Sida sp. (Malvaceae) [15,16].<br />
Known distribution: Asia - China, India, Thailand (this paper); North America - Cuba,<br />
Dominican Republic, Panama, Puerto Rico, USA (FL, LA, TX), Virgin Islands; South America -<br />
Argentina, Brazil [16].<br />
Material examined: Phengsintham (MFLU10-0312) on leaf <strong>of</strong> Sida mysorensis (Malvaceae)<br />
(Thailand: Chiang Rai province, Wiang Chiang Rung district, Tadsak waterfall), 23 December 2009.<br />
Cercospora verniciferae [21] (Figure 1; Redescribed because this species is poorly known.)<br />
Leaf spots/Lesions small to medium, suborbicular to irregular, 1–4 mm in diam., brown in<br />
the centre, and with brown-yellow margin. Caespituli/Colonies hypophyllous, scattered, dark<br />
brown. Mycelium internal; hyphae branched, 3–5 μm wide (x = 3.57 μm, n = 7), septate,<br />
constricted at the septa, distance between septa 7–12 μm ( x = 8.42 μm, n = 7), brownish or greenhyaline,<br />
wall 0.5–0.8 μm wide (x = 0.54 μm, n = 7), smooth, forming plate-like plectenchymatous<br />
stromatic hyphal aggregations. Stromata developed, small to medium-sized, globular to subglobular,<br />
substomatal and intraepidermal, 16–33 μm in diam. (x = 23.6 μm, n = 8), dark brown to black in
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
51<br />
mass, composed <strong>of</strong> swollen hyphal cells, subglobose, rounded to angular in outline, 5–10 μm wide<br />
(x = 7.9 μm, n = 13), brown to dark brown, wall 0.5–0.8 μm wide (x = 0.68 μm, n = 13), smooth.<br />
Conidiophores fasciculate, arising from stromata (1–4 per fascicle), emerging through stomata, not<br />
branched, straight to curved, cylindrical, 45–89 × 5–7 μm ( x = 71.2 × 5.4 μm, n = 5), 2–5-septate,<br />
distance between septa 5–30 μm ( x = 16.2 μm, n = 16), medium brown, paler at the apex, wall 0.5–<br />
0.8 μm wide (x = 0.63 μm, n = 16), smooth, 0–2-times geniculate. Conidiogenous cells integrated,<br />
terminal, cylindrical, 16–30 × 4–5 μm (x = 24.5 × 4.5 μm, n = 4), pale brown; conidiogenous loci<br />
conspicuous, subcircular, 2–2.5 μm wide (x = 2.12 μm, n = 4), wall 0.5–0.8 μm thick (x = 0.57<br />
μm, n = 4), thickened and darkened. Conidia solitary, acicular to obclavate, straight to curved, 23–<br />
105 × 2–4 μm ( x = 64 × 2.8 μm, n = 5), 5–12-septate, hyaline to subhyaline, thin-walled 0.3–0.5<br />
μm (x = 0.31 μm, n = 5), smooth; tip acute; base truncate, hila thickened and darkened 1–1.5 μm<br />
wide (x = 1.1 μm, n = 5), wall <strong>of</strong> the hila 0.3–0.35 μm (x = 0.31 μm, n = 5) thick.<br />
Known hosts: Rhus vernicifera DC., Spondias dulcis Parkinson, and S. pinnata (L. F.) Kurz<br />
(Anacardiaceae) [15,16].<br />
Known distribution: Asia - Thailand (this paper); Oceania - American Samoa; South<br />
America - Brazil [15,16].<br />
Material examined: Phengsintham (MFLU10-0313) on leaf <strong>of</strong> Spondias pinnata<br />
(Anacardiaceae) (Thailand: Chiang Rai province, Muang district, Sri Pangsang village), 22 December<br />
2009.<br />
Notes: The collection from Thailand agrees well with Cercospora verniciferae as<br />
circumscribed by Chupp [15]. C. verniciferae has conidiophores that are 45–89 × 5–7 μm and<br />
conidia that are 23–105 × 2–4 μm. C. verniciferae is part <strong>of</strong> the C. apii Fesen. complex from which<br />
it is morphologically barely distinguishable [16].<br />
Cercospora ziziphigena [22]<br />
Notes: The collection no. MFLU11-0019 from Mae Puem National Park, Pha Yao province<br />
(conidiophores 40–320 × 4–6 μm and conidia 163–195 × 2.5–3 μm) having a long conidia, differs<br />
from the Cercospora ziziphigena previously described [22] (conidiophores 13.8–92.5 × 3.5–6.3 μm<br />
and conidia 17.5–76.8 × 3.1–5.3 μm). A true Cercospora s. str. is quite distinct from Cercospora<br />
apii s. lat. [16].<br />
Known hosts: Ziziphus incurva Roxb. and Ziziphus sp. (Rhamnaceae) [16, 22].<br />
Known distribution: Asia - China [16, 22], Thailand (this paper).<br />
Material examined: Phengsintham (MFLU11-0019) on leaf <strong>of</strong> Ziziphus sp. (Rhamnaceae)<br />
(Thailand: Pha Yao province, Mae Jai district, Mae Puem National Park), 22 August 2010.
52 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
Figure 1(a). Cercospora verniciferae on Spondias pinnata from leaf spots: 1. Stroma with attached<br />
conidiophores; 2. Conidiophore; 3–7. Conidia. (Scale bar = 10 μm)
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
53<br />
Figure 1(b). Cercospora verniciferae on Spondias pinnata from leaf spots: 1–2. Lesions on host<br />
leaves (1. upper surface and 2. lower surface); 3. Caespituli; 4. Internal mycelium; 5. Stroma with<br />
attached conidiophores; 6. Stroma; 7–9. Conidia. (Scale bar: 1–2. = 10 mm; 3. = 3 mm; 4–9. = 10<br />
μm)
54 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
Passalora broussonetiae [16]<br />
Notes: The collection no. MFLU10-0314 from Tadsak waterfall, Chiang Rai province agrees<br />
well with Passalora broussonetiae [16], but the hyphae are smooth to distinctly verruculose<br />
(described to be smooth by Hsieh and Goh [3]). Brief description <strong>of</strong> Passalora broussonetiae from<br />
Thailand: Leaf spots/Lesions irregular, 1–9 mm diam., at first reddish brown, later becoming dark<br />
brown in the centre, gray to reddish brown margin. Caespituli/Colonies amphigenous, conspicuous.<br />
Conidiophores 170–390 × 2–5 μm, 5–17-septate. Conidia 6–28 × 4–6 μm, 0–3-septate.<br />
Known host: Broussonetia papyrifera (L.) L'Hér. ex Vent.[3, 23].<br />
Known distribution: Asia - Taiwan [3,16], Thailand (this paper).<br />
Material examined: Phengsintham (MFLU10-0314) on leaf <strong>of</strong> Broussonetia papyrifera<br />
(Moraceae) (Thailand: Chiang Rai province, Wiang Chiang Rung district, Tadsak waterfall), 23<br />
December 2009.<br />
Passalora fusimaculans [16] Cercospora fusimaculans [24]<br />
Notes: The collection no. MFLU10-0315 from Sri Pangsang village garden and no.<br />
MFLU10-0316 from Huay Kang Pah National Park, Chiang Rai province agree well with Passalora<br />
fusimaculans as circumscribed previously [3, 15, 25], but differ in having rather long conidiophores<br />
(up to 130 μm) and shorter conidia (up to 85 μm).<br />
Known hosts: Agrostis, Brachiaria, Brachia, Beckeropsis, Chasmopodium, Digitaria,<br />
Echinochloa, Eleusine, Entolasia, Ichnanthus, Leptoloma, Oplismenus, Panicum, Paspalidium,<br />
Pennisetum, Rottboellia, Setaria, Sorghum, Stenotaphrum, Urochloa, and Zea (Poaceae) [16].<br />
Known distribution: Africa - Botswana, Ethiopia, Ghana, Guinea, Ivory Coast, Kenya,<br />
Malawi, Nigeria, Rwanda, Sierra Leone, South Africa, Sudan, Tanzania, Togo, Uganda, Zambia,<br />
Zimbabwe; Asia - Brunei, China, India, Japan, Malaysia, Philippines, Taiwan, Thailand (this paper);<br />
Europe - Azerbaijan, France, Georgia; North America - Costa Rica, Cuba, Dominican Rep., El<br />
Salvador, Guatemala, Honduras, Jamaica, Mexico, Nicaragua, Panama, Trinidad & Tobago, USA<br />
(AL, PL, IA, ID, KS, NC, ND, OK, OR, TX, VA, WI); Oceania - Australia, Fiji, New Zealand,<br />
Palau, Papua New Guinea, Samoa, Solomon Islands, Vanuatu; South America - Bolivia, Brazil,<br />
Colombia, Ecuador, Guyana, Peru, Venezuela [16].<br />
Material examined: 1) Phengsintham (MFLU10-0315) on leaf <strong>of</strong> Echinochloa esculenta<br />
(Poaceae) (Thailand: Chiang Rai province, Muang district, Sri Pangsang village), 15 September<br />
2009; 2) Phengsintham (MFLU10-0316) on leaf <strong>of</strong> Echinochloa esculenta (Poaceae) (Thailand:<br />
Chiang Rai province, Huay Kang Pah National Park), 4 December 2009.<br />
Passalora graminis [26]<br />
Notes: The collection no. MFLU10-0317 from Doi Tung National Park, Chiang Rai province<br />
agrees well with Passalora graminis as circumscribed previously [15, 23]. Passalora graminis has<br />
conidiophores <strong>of</strong> 10–52 × 3–5 μm and conidia <strong>of</strong> 18–38 × 1.5–2 μm.<br />
Known hosts: Agrobardeum, Agropyron, Agrositanion, Agrostis, Alopecurus, Ammophila,<br />
Anthoxanthum, Arctagrostis, Arrhenatherum, Arundinaria, Avena, Beckmannia, Bromus,
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
55<br />
Calamagrostis, Cinna, Cynodon, Cynosurus, Dactylis, Danthonia, Deschampsia, Digitaria, Elymus,<br />
Elysitanion, Elytrigia, Egagrostis, Festuca, Glyceria, Hierochloe, Hordeum, Hystrix, Koeleria,<br />
Leersia, Leucopoa, Lolium, Melica, Milium, Miscanthus, Muhlenbergia, Oryzopsis, Panicum,<br />
Pennisetum, Phleum, Phragmities, Poa, Puccinella, Roegneria, Secale, Sitanion, Sparlina,<br />
Stenotaphrum, Stipa, Trisetum, and Zea (Poaceae) [16].<br />
Known distribution: Asia - Thailand (this paper) and worldwide [16].<br />
Material examined: Phengsintham (MFLU10-0317) on leaf <strong>of</strong> Agrostis sp. (Poaceae)<br />
(Thailand: Chiang Rai province, Doi Tung National Park), 18 August 2009.<br />
Pseudocercospora cycleae [23]<br />
Notes: The collection no. MFLU10-0319 from Khun Korn waterfall, Chiang Rai province<br />
agrees with the previous description <strong>of</strong> this species [27], but differs in having longer conidiophores.<br />
Known hosts: Cyclea fissicalyx Dunn, C. peltata Hook. f. & Thomson, and Cyclea sp.<br />
(Menispermaceae) [27].<br />
Known distribution: Asia - China, India [27], Thailand (this paper).<br />
Material examined: Phengsintham (MFLU10-0319) on leaf <strong>of</strong> Cyclea peltata<br />
(Menispermaceae) (Thailand: Chiang Rai province, Khun Korn waterfall), 18 December 2009.<br />
Pseudocercospora malloticola [3]<br />
Notes: The collection no. MFLU10-0320 from Sri Pangsang village, Chiang Rai province<br />
(Thailand) and no. NUOL P588 from Naloumai village, Savannakhet province (Laos) are similar to<br />
Pseudocercospora malloticola from Taiwan previously described [3]. Brief description <strong>of</strong> the<br />
collection from Thailand: Leaf spots/Lesions discoid to irregular, 1–6 mm diam., at first yellowish,<br />
later becoming brown or dark brown, and with yellowish margin. Conidiophores fasciculate, arising<br />
from stromata (2–11 per fascicle), emerging through stomata, nearly straight, cylindrical,<br />
unbranched, 10–40 × 3–5 μm. Conidia formed singly, cylindric, straight to slightly curved, 33–75 ×<br />
3–4 μm.<br />
Known hosts: Mallotus barbatus (Wall.) Müll. Arg., M. japonicus (L. F.) Müll. Arg., and M.<br />
thorelii Gagnep [3].<br />
Known distribution: Asia - Laos (this paper), Taiwan [3], Thailand (this paper).<br />
Material examined: 1) Phengsintham (MFLU10-0320) on leaf <strong>of</strong> Mallotus barbatus<br />
(Euphorbiaceae) (Thailand: Chiang Rai province, Muang district, Sri Pangsang village), 30 August<br />
2009; 2) Phengsintham (P588) on leaf <strong>of</strong> Mallotus thorelii (Euphorbiaceae) (Laos: Savannakhet<br />
province, Vilaboury district, Naloumai village), 23 June 2010.<br />
Pseudocercospora olacicola [28]<br />
Notes: The collection no. MFLU11-0020 from Mae Puem National Park, Pha Yao province<br />
(conidiophores 12–35 × 3–5 μm and conidia 11–58 × 2–3 μm) agrees with Pseudocercospora<br />
olacicola previously described [28] (conidiophores 10.5–40.5 × 2.5–5 μm and conidia 16–30.5 ×<br />
2.5–4 μm).
56 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
Known hosts: Olax scandens Roxb., Olax wightiana Wall. ex Wight & Arn., O. zeylanica L.,<br />
Olax sp., and Ximenia sp. (Olacaceae) [16, 28].<br />
Known distribution: Asia - India [16, 28], Thailand (this paper).<br />
Material examined: Phengsintham (MFLU11-0020) on leaf <strong>of</strong> Olax scandens (Olacaceae)<br />
(Thailand: Pha Yao province, Mae Jai district, Mae Puem National Park), 22 August 2010.<br />
Pseudocercospora paederiae [3]<br />
Notes: The collection no. MFLU10-0321 from Tadsak waterfall, Chiang Rai province<br />
(conidiophores 5–20 × 3–5 μm and conidia 42–75 × 2–3 μm) is similar to the original description <strong>of</strong><br />
this species, based on material from Taiwan, but there are slight differences in the size <strong>of</strong> the<br />
conidiophores and conidia. The collection from Taiwan has conidiophores that are densely<br />
fasciculate, 20–120 × 3–4 μm, subhyaline to pale brown and conidia that are obclavate, straight to<br />
moderately curved, 30–80 × 3.5–5 μm and medium olivaceous [3, 15, 27].<br />
Known hosts: Paederia chinensis Hance, P. foetida L., P. scandens (Lour.) Merr., and P.<br />
tomentosa Blume (Rubiaceae) [3,27].<br />
Known distribution: Asia - China, Japan, Korea, Taiwan [3, 27], Thailand (this paper).<br />
Material examined: Phengsintham (MFLU10-0321) on leaf <strong>of</strong> Paederia tomentosa<br />
(Rubiaceae) (Thailand: Chiang Rai province, Wiang Chiang Rung district, Tadsak waterfall), 23<br />
December 2009.<br />
Pseudocercospora polysciatis [29]<br />
Notes: The collection no. MFLU10-0322 from Sri Pangsang village, Chiang Rai province<br />
agrees well with Pseudocercospora polysciatis described previously [3, 27] but differs in having<br />
distinct constriction at the septa.<br />
Known hosts: Polyscias balfouriana (André) L.H. Bailey, P. guilfoylei (W. Bull) L.H.<br />
Bailey, and Polyscias sp. (Araliaceae) [3, 16, 27].<br />
Known distribution: Africa - Mauritius, Ivory Coast; Asia - Brunei, Philippines, Taiwan,<br />
Thailand (this paper); North America - Cuba; Oceania - American Samoa, Cook Islands, Fiji,<br />
Kiribati, Marshall Islands, Micronesia, Niue, Samoa, Solomon Islands, Tonga [16].<br />
Material examined: Phengsintham (MFLU10-0322) on leaf <strong>of</strong> Polyscias balfouriana<br />
(Araliaceae) (Thailand: Chiang Rai province, Muang district, Sri Pangsang village), 16 January 2010.<br />
Zasmidium cassiicola [30] Stenella cassiicola [31] (Figure 2; Redescribed because this species<br />
is poorly known.)<br />
Leaf spots/Lesions variable, more or less irregularly orbicular, 1–8 mm diam., typically deep<br />
brown. Caespituli/Colonies hypophyllous, conspicuous. Mycelium external; hyphae branched, 2–4<br />
μm wide (x = 3.2 μm, n = 9), septate, constricted at the septa, distance between septa 8–30 μm ( x<br />
= 18.56 μm, n = 9), pale olivaceous-brown, thin-walled 0.3–0.5 μm wide (x = 0.41 μm, n = 9),<br />
verruculose. Stromata absent. Conidiophores borne on external mycelial hyphae, unbranched,<br />
cylindrical, 30 – 117 × 3 – 4 μm ( x = 65.9 × 3.17 μm, n = 18), 3 – 8-septate, distance between
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
57<br />
Figure 2(a). Zasmidium cassiicola on Cassia fistula: 1–3. External mycelia with attached<br />
conidiophores; 4–7. Conidia. (Scale bar = 10 μm)
58 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
Figure 2(b). Zasmidium cassiicola on Cassia fistula from leaf spot: 1–2. Lesions on host leaves (1.<br />
upper surface and 2. lower surface); 3–6. Conidiophores; 7. Apex with attached conidium; 8.<br />
External mycelium; 9–11. Conidia; 12. Living culture. (Scale bars: 1–2. = 10 mm, 3–11. = 10 μm)<br />
septa 7–25 μm ( x = 15.8 μm, n = 30), mid pale golden brown, wall 0.5–0.8 μm ( x = 0.48 μm, n =<br />
30), smooth. Conidiogenous cells integrated, terminal or intercalary, 7–20 × 2–4 μm ( x = 13 ×<br />
2.63 μm, n = 8), cylindrical, swollen and curved at the apex; conidiogenous loci forming minute,<br />
dark or refractive scars on lateral and terminal denticles, 1–2 μm diam. (x = 1.4 μm, n = 7), giving<br />
rise to branched conidial chains, wall 0.3–0.5 μm wide (x = 0.4 μm, n = 7), thickened, darkened.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
59<br />
Conidia solitary or catenate, sometimes ellipsoidal-ovoid or subcylindrical, but mostly slightly<br />
obclavate, straight or slightly curved or sinuous, 11–70 × 2–4 μm ( x = 34.16 × 2.9 μm, n = 24), 1–<br />
5-septate, pale olivaceous, wall 0.3–0.5 μm wide (x = 0.33 μm, n = 24), smooth or finely<br />
verruculose; apex rounded or subtruncate, 1–1.5 μm wide, wall 0.3–0.5 μm wide; base short tapered<br />
at the base to the hilum, 1–2 μm wide (x = 1.3 μm, n = 9), wall 0.3–0.5 μm wide (x = 0.41 μm, n =<br />
9), thickened and darkened.<br />
Known host: Cassia fistula L. (Leguminosae) [31].<br />
Known distribution: Asia - India [31], Thailand (this paper).<br />
Material examined: Phengsintham (MFLU10-0324) on leaf <strong>of</strong> Cassia fistula (Leguminosae)<br />
(Thailand: Chiang Rai province, Muang district, Sri Pangsang village), 16 January 2010.<br />
Cultural characteristics: Colonies on potato dextrose agar after three weeks at 25°C with<br />
spreading mycelium, surface ridged, black and wavy in the centre and gray margin, reaching 5–15<br />
mm diam.; hyphae <strong>of</strong>ten constricted at the septa, distances between septa 6–16 × 3–5 μm ( x = 10.5<br />
× 3.6 μm, n = 30), thin-walled, 0.3–0.5 μm wide (x = 0.45 μm, n = 30), hyaline, smooth or<br />
verruculose; Conidiophores and conidia not formed in culture.<br />
Notes: The collection from Thailand agrees well with the Indian Zasmidium cassiicola<br />
(Stenella cassiicola [13, 31]. This is the first record outside India. Zasmidium cassiae-fistulae is a<br />
similar species but differs in forming its conidia consistently singly [32].<br />
CONCLUSIONS<br />
Cercosporoid fungi are one <strong>of</strong> the largest groups <strong>of</strong> pathogenic hyphomycetes causing leaf<br />
spots on a wide range <strong>of</strong> crops, fruit trees and other plants. The damage to living leaves and fruits<br />
may cause reduced yield. Fourteen species assigned to the genera Cercospora (5), Passalora (3),<br />
Pseudocercospora (5) and Zasmidium (1) are new records for Thailand. Cercospora verniciferae<br />
and Zasmidium cassiicola are poorly known species and are fully described.<br />
ACKNOWLEDGEMENTS<br />
The authors would like to thank the Global Research Network for Fungal Biology and King<br />
Saud University and the Mushroom Research Foundation (MRF) for supporting this research.<br />
Special thanks also go to the Mushroom Research Foundation.<br />
REFERENCES<br />
1. G. N. Agrios, “Plant Pathology”, 5 th Edn., Academic Press, San Diego, 2005.<br />
2. P. W. Crous, “Taxonomy and phylogeny <strong>of</strong> the genus Mycosphaerella and its anamorphs”,<br />
Fungal Divers., 2009, 38, 1-24.<br />
3. W. H. Hsieh and T. K. Goh, “Cercospora and Similar Fungi from Taiwan”, Maw Chang Book<br />
Co., Taipei, 1990.<br />
4. E. B. G. Jones, M. Tantichareon and K. D. Hyde (Ed.), “Thai Fungal Diversity”, Biotec,<br />
Bangkok, 2005.
60 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
5. P. Giatgong, “Host Index <strong>of</strong> Plant Diseases in Thailand”, Ministry <strong>of</strong> Agriculture and<br />
Cooperatives, Bangkok, 1980.<br />
6. P. Sontirat, P. Pitakpraiwan, W. Choonbamroong and U. Kueprakone, “Plant Pathogenic<br />
Cercosporoids in Thailand”, Ministry <strong>of</strong> Agriculture and Cooperatives, Bangkok, 1980.<br />
7. V. Petcharat and M. Kajanamaneesathian, “Species <strong>of</strong> plant pathogenic Cercospora in southern<br />
Thailand”, Thai J. Phythopathol., 1989, 9, 23-27.<br />
8. P. Sontirat, P. Pitakpraivan, T. Khamhangridthrirong, W. Choonbamroong and U. Kueprakone,<br />
“Host Index <strong>of</strong> Plant Diseases in Thailand”, Ministry <strong>of</strong> Agriculture and Cooperatives, Bangkok,<br />
1994.<br />
9. C. Nakashima, K. Motohashi, J. Meeboon and C. To-anun, “Studies on Cercospora and allied<br />
genera in northern Thailand”, Fungal Divers., 2007, 26, 257-270.<br />
10. J. Meeboon, I. Hidayat and C. To-anun, “An annotated list <strong>of</strong> cercosporoid fungi in northern<br />
Thailand”, J. Agric. Technol., 2007, 3, 51-63.<br />
11. C. To-anun, J. Nguenhom, J. Meeboon and I. Hidayat, “Two fungi associated with necrotic<br />
leaflets <strong>of</strong> areca palms (Areca catechu)”, Mycol. Prog., 2009, 8, 115-121.<br />
12. J. Meeboon, “Diversity and phylogeny <strong>of</strong> true cercosporoid fungi from northern Thailand”, PhD<br />
Thesis, 2009, Chiang Mai University, Thailand.<br />
13. C. To-anun, I. Hidayat and J. Meeboon, “Cercospora cristellae, a new cercosporoid fungus<br />
associated with weed Cristella parasitica from northern Thailand”, J. Agric. Technol., 2010, 6,<br />
331-339.<br />
14. P. Phengsintham, E. Chukeatirote, K. A. Abdelsalam, K. D. Hyde and U. Braun, “Cercospora<br />
and allied genera from Laos 3”, Cryptogamie Mycol., 2010, 31, 305-322.<br />
15. C. Chupp, “A Monograph <strong>of</strong> the Fungus Genus Cercospora”, Charles Chupp, Ithaca (New<br />
York), 1954.<br />
16. P. W. Crous and U. Braun, “Mycosphaerella and Its Anamorphs: 1. Names Published in<br />
Cercospora and Passalora”, Centraalbureau voor Schimmelcultures, Utrecht, 2003.<br />
17. U. Braun, “A Monograph <strong>of</strong> Cercosporella, Ramularia and Allied Genera (Phytopathogenic<br />
Hyphomycetes)”, Vol. 1, IHW Verlag, Eching (Germany), 1995.<br />
18. J. B. Ellis and B. M. Everhart, “New species <strong>of</strong> fungi from various localities”, J. Mycol., 1888,<br />
4, 113-118.<br />
19. J. J. Davis, “Notes on parasitic fungi in Wisconsin XX”, Trans. Wisc. Acad. Sci., 1937, 30, 1-<br />
16.<br />
20. J. B. Ellis and B. M. Everhart, “New species <strong>of</strong> hyphomycetes fungi”, J. Mycol., 1889, 5, 68-72.<br />
21. A. P. Viegas, “Alguns fungos do Brasil, Cercosporae”, Bol. Soc. Bras. Agron., 1945, 8, 1-160.<br />
22. L. Xu and Y. L. Guo, “Studies on Cercospora and allied genera in China XIII”, Mycosystema,<br />
2003, 22, 6-8.<br />
23. F. C. Deighton, “Studies on Cercospora and allied genera. VI. Pseudocercospora Speg.,<br />
Pantospora Cif. and Cercoseptoria Petr.”, Mycol. Papers, 1976, 140, 1-168.<br />
24. G. F. Atkinson, “Some Cercosporae from Alabama”, J. Elisha Mitchel Sci. Soc., 1891, 8, 33-<br />
67.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 47-61<br />
61<br />
25. M. B. Ellis, “More Dematiaceous Hyphomycetes”, Commonwealth Mycological Institute, Kew,<br />
Surrey, 1976.<br />
26. P. Srivastava, “Recombinations in genus Passalora Fries”, J. Living World, 1994, 1, 112-119.<br />
27. Y. L. Guo and W. H. Hsieh, “The Genus Pseudocercospora in China”, <strong>International</strong> Academic<br />
Publishers, Beijing, 1995.<br />
28. Kamal, M. K. Khan and R. K. Verma, “New species and combinations in Pseudocercospora<br />
from India”, Mycol. Res., 1990, 94, 240-242.<br />
29. U. Braun and A. Sivapalan, “Cercosporoid hyphomycetes from Brunei”, Fungal Divers., 1999,<br />
3, 1-27.<br />
30. Kamal, “Cercosporoid Fungi <strong>of</strong> India”, Bishen Singh Mahendra Pal Singh, Dehradun (India),<br />
2010.<br />
31. S. Misra, A. K. Srivastava and Kamal, “Further additions to Stenella from India and Nepal”,<br />
Mycol. Res., 1999, 103, 268-270.<br />
32. U. Braun, P. Crous and Kamal, “New species <strong>of</strong> Pseudocercospora, Pseudocercosporella,<br />
Ramularia and Stenella (cercosporoid hyphomycetes)”, Mycol. Prog., 2003, 2, 197-208.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
62 <strong>Maejo</strong> <strong>Maejo</strong> Int. J. Sci. Int. Technol. J. Sci. Technol. <strong>2012</strong>, 6(01), <strong>2012</strong>, 62-69 6(01), 62-69<br />
Short Communication<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
On the security <strong>of</strong> an anonymous roaming protocol in UMTS<br />
mobile networks<br />
Shuhua Wu 1, 2, * , Qiong Pu 2, 3 and Ji Fu 1<br />
1 Department <strong>of</strong> Network Engineering, Information Engineering University, Zhengzhou, China<br />
2 State Key Laboratory <strong>of</strong> Information Security, Graduate University <strong>of</strong> Chinese Academy <strong>of</strong><br />
<strong>Science</strong>s, Beijing, China<br />
3 CIMS Research Centre, Tongji Univerity, Shanghai, China<br />
* Corresponding author, e-mail: pqwsh@yahoo.com.cn<br />
Received: 21 April 2011 / Accepted: 4 February <strong>2012</strong> / Published: 10 February <strong>2012</strong><br />
Abstract: In this communication, we first show that the privacy-preserving roaming<br />
protocol recently proposed for mobile networks cannot achieve the claimed security level.<br />
Then we suggest an improved protocol to remedy its security problems.<br />
Keywords: cryptanalysis, anonymous roaming, authentication, UMTS<br />
INTRODUCTION<br />
With the advancement and tremendous development <strong>of</strong> computer networks and<br />
telecommunications, user mobility has become a highly desirable network feature nowadays,<br />
especially in wireless networks (e.g. cellular networks [1-3]). This technology enables users to access<br />
services universally and without geographical limitations. In other words, they can go outside the<br />
coverage zone <strong>of</strong> their home networks, travel to foreign networks and access services provided by<br />
the latter as a visiting user or a guest. This capability is usually called roaming. Security is one <strong>of</strong> the<br />
major requirement in roaming networks. In addition to authentication, user’s privacy is equally<br />
important in such networks. To preserve this feature, not only should the user’s identity be protected<br />
(anonymity requirement), but also his location and the relation between his activities should be kept<br />
secret (untraceability requirement). The violation <strong>of</strong> either <strong>of</strong> the mentioned requisites can seriously<br />
endanger the user’s privacy. Samfat et al. [1] have proposed a comprehensive classification for<br />
different levels <strong>of</strong> privacy protection according to the knowledge <strong>of</strong> different entities about the user’s<br />
identification information. The classification is as follows:
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 62-69<br />
63<br />
• C1: Each user is anonymous to eavesdroppers and his activities are unlinkable to them.<br />
• C2: In addition to C1, each user is anonymous to the foreign servers and his activities are<br />
unlinkable to them.<br />
• C3: In addition to C2, the relationship between the user and servers (the home server and the<br />
foreign servers) is anonymous for eavesdroppers.<br />
• C4: In addition to C3, the home server <strong>of</strong> the user is anonymous to the foreign servers.<br />
• C5: In addition to C4, each user is anonymous and his activities are unlinkable to his home<br />
server.<br />
In the standard universal mobile telecommunication system (UMTS) [2], the home server<br />
must be always aware <strong>of</strong> the mobile user’s location in order to route the incoming calls towards the<br />
user. Moreover, the foreign server should know the identity <strong>of</strong> the home server for billing purpose.<br />
Therefore, it seems that the admissible level <strong>of</strong> privacy protection in this scenario is C3. In the last<br />
decades, several schemes addressed the privacy <strong>of</strong> users in mobile networks [3-15]. However, the<br />
most perfect and practical scheme that has been proposed so far only achieves the C2 class <strong>of</strong><br />
anonymity and the possible C3 class has not been provided in UMTS yet. To fill this blank, Fatemi et<br />
al. [2] recently proposed a privacy-preserving roaming protocol based on hierarchical identity-based<br />
encryption (IBE) [16] for mobile networks. This protocol was claimed to achieve the acceptable C3<br />
level <strong>of</strong> privacy. In this communication, we first show that it has some security weakness and thus<br />
the claimed security level is not achieved. Finally, we propose an enhanced protocol to remedy the<br />
existing security loopholes.<br />
PRELIMINARY<br />
In this section we recall the concept <strong>of</strong> Identity-Based Encryption (IBE) and hierarchical IBE<br />
(HIBE) schemes, upon which Fatemi et al.’s scheme builds. Here we just follow their description [2].<br />
At first, we introduce the concept <strong>of</strong> a bilinear map between two groups, which will be used in the<br />
IBE scheme. Let G 1<br />
be an additive group and G 2<br />
be a multiplicative group, both <strong>of</strong> order q ( q<br />
should be some large prime, e.g. 160 bits). We say that a map eG1G1 G2<br />
is an admissible<br />
bilinear map if all the three following conditions are satisfied:<br />
1) eaPbQ ( ) ePQ<br />
( ) ab for all ab Zq<br />
and PQ G1<br />
(bilinear condition);<br />
2) the map does not send all elements <strong>of</strong> G 1<br />
G 1<br />
to the identity element <strong>of</strong> G 2<br />
(non-degeneracy<br />
condition); and<br />
3) there is an efficient algorithm to compute ePQ ( ) for all PQ G1<br />
(computability condition).<br />
Throughout this communication, the Bilinear Diffie-Hellman (BDH) in G 1<br />
G 2<br />
e is<br />
believed to be hard (i.e. it is hard to compute ePP ( ) abc G2<br />
, given PaPbPcP<br />
for some<br />
abc Z q<br />
). Since the BDH problem is not harder than the computational Diffie-Hellman (CDH)<br />
problem in G<br />
1<br />
or G<br />
2<br />
, the CDH problem in G 1<br />
or G<br />
2<br />
is also believed to be hard. The CDH problem<br />
in G<br />
1<br />
is as follows: given random PaPbP<br />
for ab Zq<br />
, compute abP ; the CDH problem in<br />
G<br />
2<br />
is defined similarly. In addition, for a point Q G <br />
1<br />
, the isomorphism fQ<br />
G1 G2<br />
by<br />
f ( ) ( )<br />
Q<br />
P e P Q is considered as a one-way function ( P cannot be inferred from ePQ ( ) and Q )
64 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 62-69<br />
since an efficient algorithm for inverting f Q<br />
for some Q results in an efficient algorithm for solving<br />
CDH problem in G<br />
2<br />
.<br />
Now we begin to introduce the IBE system. An IBE is a public key cryptosystem in which<br />
the public key takes any arbitrary string such as a name or an e-mail address, and the private key<br />
generator (PKG) can produce a private key corresponding to each string. Hence, one can encrypt a<br />
message by a public key even if the public key’s owner has not yet set up his private key. An efficient<br />
IBE is presented [17], which is called a Boneh-Franklin scheme. Let P be a generator <strong>of</strong> G 1<br />
and<br />
s Z <br />
q<br />
be the PKG’s master key. Then in the Boneh-Franklin scheme, each user’s identity-based<br />
private key should be computed as ( )<br />
<br />
kU<br />
sH1 U , where H 1<br />
{01}<br />
G 1<br />
is a cryptographic hash<br />
function and U is the user’s identity. Then one can encrypt a message using the public key U , and<br />
U can decrypt the ciphertext using the private key k U<br />
. The BF scheme is resistant to the chosen<br />
ciphertext attack, assuming the hardness <strong>of</strong> the BDH problem [17].<br />
Similar to the public key cryptosystems, a hierarchy <strong>of</strong> PKGs is desirable in an IBE system to<br />
reduce the workload <strong>of</strong> the master servers. A two-level HIBE (2-HIBE) is presented [16]. There are<br />
three entities involved in a 2-HIBE scheme: a root PKG which possesses a master key s , the domain<br />
PKGs which gain their domain keys from the root PKG, and the users with private keys generated by<br />
<br />
their domain PKGs. The 2-HIBE scheme benefits from a linear one-way function hG1Zq<br />
G1<br />
with the following properties:<br />
<br />
1) For all PG1 axZ h( aPx) ah( P x)<br />
,<br />
q<br />
<br />
2) Given xx <br />
iZqP G1<br />
and xih( aPxi)<br />
for i 1 n, haPx ( ) cannot be computed with<br />
any probabilistic polynomial-time algorithm.<br />
The function h defined above is a one-way function with respect to its first argument, i.e. P<br />
cannot be inferred from hPx ( ) and x . Then the key for domain S is kS<br />
sH1( S)<br />
G1<br />
and the key<br />
<br />
for user U in domain S is kU<br />
h( kSH2( S U))<br />
G1, where H 2<br />
{01} Zq<br />
is a cryptographic<br />
hash function and denotes concatenation. Finally, one can encrypt a message by a public key<br />
SU<br />
and U can decrypt the ciphertext using k . U<br />
FATEMI ET AL.’ S ROAMING PROTOCOL<br />
Review <strong>of</strong> Protocol<br />
Here we just follow the description <strong>of</strong> Fatemi et al [2]. Like Wan et al.’s scheme [18], they<br />
also assume that a 2-HIBE is implemented in the system and the domain servers have received their<br />
private keys { KS<br />
sH1( S )}<br />
i<br />
i<br />
from a root server. Also, they suppose that the user U obtains his<br />
private key KU<br />
h( KHS H2( HS U))<br />
during the registration at his home domain server HS . In<br />
addition, a temporary key K e( h( h( H1( HS) H2( HS Nym))<br />
H2( HS)) sH1( HS))<br />
corresponding to<br />
a pseudonym Nym will be computed by the user during the roaming protocol and will be used for<br />
the authentication and key agreement purposes when he enters a foreign network domain.<br />
As shown in Figure 1, the protocol is as follows:
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 62-69<br />
65<br />
Step 1. When the foreign server ( FS ) detects a new user in his domain, it generates a nonce<br />
a random number r s<br />
(both from<br />
Z q<br />
) and computes rP. s<br />
Then it stores the values<br />
N<br />
s<br />
and<br />
N<br />
s<br />
and rP s<br />
in<br />
his database and sends the first message including his identity IDFS Ns<br />
and rP s<br />
to the user.<br />
Figure. 1. Fatemi et al.’s roaming protocol [2]<br />
Step 2. Similarly, the user U generates a nonce N u<br />
and a random number r u<br />
and computes<br />
k u rrP<br />
u s<br />
. Then he fetches the only unused pair <strong>of</strong> ( Nym K)<br />
from his memory and computes<br />
the session key to be shared with the foreign server as sk ( 1)<br />
u<br />
H4 K k<br />
u FS Nym Nu Ns<br />
and a verifier mac ( u<br />
H<br />
4<br />
K k u FS Nym Nu<br />
N<br />
s<br />
0) , where H 4<br />
is a hash function which<br />
*<br />
maps {0 1} to {0 1} l for some security parameter l . After that, he selects an arbitrary Nym<br />
next<br />
to<br />
be used in the next execution <strong>of</strong> the roaming protocol (either in the current FS or another FS ).<br />
In order to compute the corresponding key K<br />
next<br />
with the help <strong>of</strong> FS and HS , the user selects a<br />
<br />
random number a Zq<br />
and computes the following values:<br />
<br />
<br />
<br />
<br />
x1 hhaH ( (<br />
1( HS) H2( HS Nymnext<br />
)) H2( HS))<br />
, x2 hhaH ( (<br />
1( HS) H2( HS U)) H2( HS))<br />
.<br />
Also, he chooses random numbers ba1 a2 Z <br />
<br />
<br />
q<br />
and computes ax<br />
1 1 ax<br />
2 2<br />
and<br />
EHS ( bUE( KU U) IDFS<br />
), where ES<br />
( M ) denotes the ID-based encryption <strong>of</strong> message M with<br />
the public key S (e.g. HS or FS ), and EK (<br />
U<br />
U)<br />
denotes the symmetric encryption <strong>of</strong> U with<br />
<br />
the key K U<br />
. Next, he sends the values EFS ( NymIDHS ) Nuru PNsrs Pmacu<br />
x1a1x1<br />
ax <br />
2 2<br />
and<br />
EHS ( bUE( KU U) IDFS<br />
) to the foreign server.<br />
Step 3. Upon receiving the above values, the foreign server checks if N s<br />
and rP<br />
s<br />
exist in its<br />
database and aborts the connection if it does not find such values. Otherwise, it decrypts<br />
EFS<br />
( Nym IDHS<br />
) with its private key sH 1<br />
( FS ) and obtains the Nym and ID<br />
HS<br />
. Then it<br />
generates a random number c Z <br />
q<br />
and computes z h( h( cH1( HS) H 2( HS Nym)) H<br />
2( HS))<br />
.<br />
Subsequently, the FS sends z and E ( bUE( K U) ID ) to the HS .<br />
HS U FS
66 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 62-69<br />
Step 4. The home server decrypts the message E ( bUE( K U) ID ) with its private key<br />
HS U FS<br />
sH ( ) 1<br />
HS and checks whether it has received the messages from the server with the identity<br />
ID<br />
FS<br />
. Then it authenticates the user U by verifying the correctness <strong>of</strong> EK (<br />
U<br />
U)<br />
. The home<br />
server terminates the connection if any <strong>of</strong> these verifications fails. Otherwise, it computes<br />
y e( zsH1( HS))<br />
( sb) H1( HS) sH1( HS)<br />
bH ( HS)<br />
1<br />
and sends them back to the FS .<br />
1<br />
c<br />
Step 5. The FS computes k s rrP<br />
s u<br />
and the values K<br />
y <br />
<br />
, ( <br />
macu<br />
H4 K k<br />
s FS Nym<br />
Nu<br />
Ns 0) . The FS rejects the connection if the equality macu mac <br />
u<br />
does not hold.<br />
Otherwise, it accepts K as the user’s key corresponding to Nym and authenticates the user.<br />
The computed K together with the message EHS ( bUE( KU U) IDFS<br />
) are credentials by<br />
which the foreign server will be able to request the user’s home server for service charge.<br />
Indeed, these values become a pro<strong>of</strong> for payment request. In the next step the foreign server<br />
computes the session key ( <br />
sk H 1)<br />
4<br />
K k s FS Nym N Ns and the authenticator<br />
s<br />
( <br />
4 s<br />
s<br />
mac H K k FS Nym N Ns 2) . Moreover, the foreign server calculates<br />
u<br />
<br />
<br />
y1 e( x1( s b) H1( HS))<br />
and y2 e( a1x1 a2x2( s b) H1( HS))<br />
to make the computation <strong>of</strong><br />
K<br />
next<br />
feasible for the user. Finally, it returns macs y1<br />
and H4( y<br />
2)<br />
to the user.<br />
Step 6. When the user receives the messages from the foreign server, he computes<br />
mac ( 2)<br />
s<br />
H4 K k U FS Nym Nu<br />
Ns and checks the equality macs mac <br />
s<br />
. If it<br />
does not hold, the user aborts the connection. Otherwise, he authenticates the foreign server<br />
<br />
<br />
and computes the following values: y1 y1e( x1 bH1( HS))<br />
,<br />
<br />
1<br />
2<br />
(<br />
1) a<br />
<br />
y y [ e( h( a K<br />
U<br />
H<br />
2( HS )) H 1( HS )) e(<br />
x2<br />
( 1<br />
1 ))]a 2<br />
( )<br />
bH HS , (<br />
1<br />
) a <br />
<br />
Knext<br />
y . Afterwards, the<br />
user considers whether H4( y2) H4( y <br />
2)<br />
. If the equation holds, he accepts K next<br />
as the key<br />
corresponding to Nym<br />
next<br />
. If not, he rejects the connection.<br />
At the end <strong>of</strong> the protocol, sku sks<br />
is the key that the user and the home server have agreed<br />
upon to be used for security purpose.<br />
Weakness <strong>of</strong> Fatemi et al.’s Protocol<br />
We assume the adversary has totally controlled a mobile user U or equivalently he has<br />
revealed the secret keys K through side channel attacks [19]. We further assume the adversary has<br />
U<br />
corrupted one <strong>of</strong> foreign networks, e.g. FN . Let HS be the home server <strong>of</strong> U and FS be the<br />
server <strong>of</strong> FN . The adversary impersonated U to visit FN and initiated an execution <strong>of</strong> Fatemi et<br />
al.’s roaming protocol. He obtained from the corrupted network the message transmitted from HA<br />
to FS in Step 4: ( s<br />
b) H1( HS)<br />
, where b is the random number chosen by the adversary<br />
in Step 2. He could then compute HS ’s secret key K through<br />
HS<br />
K sH1( HS) ( s b) H1( HS) bH1( HS)<br />
. When the adversary knows K , the problem <strong>of</strong><br />
HS<br />
HS<br />
Fatemi et al.’s roaming protocol becomes evident:<br />
• Firstly, the adversary can reveal the real identity <strong>of</strong> any other subscriber <strong>of</strong> HS . When a mobile<br />
user U ( U ) performs the authentication procedure with a foreign server FS ( FS ), the<br />
adversary eavesdrops their communication and can easily get the message transmitted between<br />
u
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 62-69<br />
67<br />
U and FS : ID<br />
FS<br />
in Step 1 and EHS ( bUE( KU U) IDFS<br />
) in Step 2. Then the adversary just<br />
guesses that U is a subscriber <strong>of</strong> HS and attempts to decrypt the message<br />
EHS ( bUE( KU U) IDFS<br />
) with K sH ( )<br />
HS 1<br />
HS . If he can retrieve ID<br />
FS<br />
from the decrypted<br />
message, he confirms his guess is correct, i.e. HS HS , because otherwise the probability that<br />
he gets any meaningful results from decryption for verification is next to zero. Then he further<br />
retrieves item U from the decrypted message and thus knows the user’s real identity U . This<br />
even contradicts the C1 security requirements. However, if U is not a subscriber <strong>of</strong> HS , his<br />
attack cannot succeed, but he will always succeed for any subscriber <strong>of</strong> HS .<br />
• Secondly, the adversary can impersonate any other subscriber <strong>of</strong> HS (e.g. U ) because he may<br />
derive K<br />
U<br />
from K as follows: K (<br />
HS<br />
U<br />
h K H2( HS U))<br />
. In other words, the authentication<br />
HS<br />
mechanism <strong>of</strong> the protocol is completely compromised.<br />
IMPROVED ROAMING PROTOCOL<br />
The above demonstrated attacks show that Fatemi et al.’s protocol does not seem to achieve<br />
authentication or anonymity. In this section we present an enhanced protocol to remedy the security<br />
loopholes. As shown in Figure 2, our protocol is based on that <strong>of</strong> Fatemi et al. and it has the<br />
following changes:<br />
Figure 2. Improved roaming protocol<br />
<br />
• In Step 2 the computation <strong>of</strong> ax 1 1<br />
a 2<br />
x 2<br />
is not needed any longer and finally the user U sends<br />
the values EFS ( NymIDHS ) Nuru PNsrs Pmacu<br />
x <br />
1<br />
and EHS<br />
( bUE( KUU) IDFS<br />
) to the<br />
foreign server.<br />
• In Step 3 the FS also forwards x 1<br />
to the HS. That is, the FS sends z,<br />
E ( bUE( K U) ID ) and x 1<br />
to the HS.<br />
HS U FS
68 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 62-69<br />
• In Step 4 the computation <strong>of</strong> ( sb) H1( HS) sH1( HS) bH1( HS)<br />
is not needed. Instead, the<br />
<br />
<br />
home server computes a new item w E( KU<br />
e( x1 sH1( HS)) x1)<br />
to make the computation <strong>of</strong><br />
K<br />
next<br />
feasible for the user and finally sends w along with y back to the FS .<br />
• In Step 5 the computation <strong>of</strong> y 1<br />
and y<br />
2<br />
is not needed and the foreign server returns mac<br />
s<br />
and<br />
w to the user.<br />
• In Step 6 the computation <strong>of</strong> y 1<br />
and y 2<br />
is not needed. After the verification <strong>of</strong> mac<br />
s<br />
is passed<br />
and the foreign server is authenticated, the user U decrypts w using his own secret key K<br />
U<br />
to<br />
retrieve the two items ex ( <br />
<br />
1 sH1( HS))<br />
x1. If the decrypted x 1<br />
is the same as x 1<br />
computed in<br />
1<br />
( )<br />
Step 1, he proceeds to compute ( (<br />
1 1( ))) a <br />
<br />
Knext<br />
e x sH HS and accepts K<br />
next<br />
as the key<br />
corresponding to Nym<br />
next<br />
. If not, he rejects the connection.<br />
In our improved protocol, the item ex ( 1<br />
sH1( HS))<br />
is used to make the computation <strong>of</strong> K<br />
next<br />
feasible for the user. Given ex ( 1<br />
sH1( HS))<br />
, it is impossible for the adversary to compute sH ( HS ) 1<br />
since the isomorphism f Q<br />
(here Q x <br />
1<br />
) is a one-way function. Therefore, the attacks described<br />
previously will not work any more. Although the changes introduce some computation overhead on<br />
the side <strong>of</strong> HS due to the computation <strong>of</strong> w , the computation cost <strong>of</strong> FS or U is significantly<br />
reduced since both FS and U omit several costly operations (including bilinear map and<br />
exponentiation). In practice the device <strong>of</strong> the mobile user is much less powerful than the servers’.<br />
Our protocol, therefore, would be more practical.<br />
ACKNOWLEDGEMENTS<br />
This work was supported in part by the National Natural <strong>Science</strong> Foundation <strong>of</strong> China<br />
(Project No. 61101112) and China Postdoctoral <strong>Science</strong> Foundation (Project No. 2011M500775).<br />
REFERENCES<br />
1. D. Samfat , R. Movla and N. Asokan , “Untraceability in mobile networks”, Proceedings <strong>of</strong> the<br />
1st <strong>International</strong> Conference on Mobile Computing, 1995, Santa Barbara, California, USA,<br />
pp.26-36.<br />
2. M. Fatemi, S. Salimi and A. Salahi, “Anonymous roaming in universal mobile telecommunication<br />
system mobile networks”, IET Inf. Secur., 2010, 4, 93-103.<br />
3. W.-S. Juang and J.-L. Wu, “Efficient 3GPP authentication and key agreement with robust user<br />
privacy protection”, Proceedings <strong>of</strong> IEEE Wireless Communications and Networking<br />
Conference, 2007, Hong Kong, China, pp.2720-2725.<br />
4. B. Sattarzadeh, M. Asadpour and R. Jalili, “Improved user identity confidentiality for UMTS<br />
mobile networks”, Proceedings <strong>of</strong> 4th European Conference on Universal Multiservice<br />
Networks, 2007, Toulouse, France, pp.401-409.<br />
5. Y. Jiang, C. Lin, X. Shen and M. Shi, “Mutual authentication and key exchange protocols for<br />
roaming services in wireless mobile networks”, IEEE Trans. Wireless Comm., 2006, 5, 2569-<br />
2577.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 62-69<br />
69<br />
6. G. Yang, D. S. Wong and X. Deng, “Efficient anonymous roaming and its security analysis”,<br />
Proceedings <strong>of</strong> the 3rd <strong>International</strong> Conference on Applied Cryptography and Network<br />
Security, 2005, New York, USA, pp.334-349.<br />
7. C.-T. Li and C.-C Lee, “A novel user authentication and privacy preserving scheme with smart<br />
cards for wireless communications”, Math. Comp. Model., <strong>2012</strong>, 55, 35-44.<br />
8. S. Wu, Y. Zhu and Q. Pu, “Security analysis <strong>of</strong> a cocktail protocol with the authentication and<br />
key agreement on the UMTS”, IEEE Comm. Lett., 2010, 14, 366-369.<br />
9. S. Wu,. Y. Zhu and Q. Pu, “A novel lightweight authentication scheme with anonymity for<br />
roaming service in global mobility networks”, Int. J. Network Manage., 2011, 21, 384-401.<br />
10. C.-C. Chang and H.-C. Tsai, “An anonymous and self-verified mobile authentication with<br />
authenticated key agreement for large-scale wireless networks”, IEEE Trans. Wireless Comm.,<br />
2010, 9, 3346-3353.<br />
11. T.-Y. Youn and J. Lim, “Improved delegation-based authentication protocol for secure roaming<br />
service with unlinkability”, IEEE Comm. Lett., 2010, 14, 791-793.<br />
12. D. He, J. Bu, S. Chan, C. Chen and M. Yin, “Privacy-preserving universal authentication<br />
protocol for wireless communications”, IEEE Trans. Wireless Comm., 2011, 10, 431-436.<br />
13. D. He, M. Ma, Y. Zhang, C. Chen and J. Bu, “A strong user authentication scheme with smart<br />
cards for wireless communications”, Comp. Comm. , 2011, 34, 367-374.<br />
14. J. Xu, W-T. Zhu and D-G. Feng, “An efficient mutual authentication and key agreement<br />
protocol preserving user anonymity in mobile networks”, Comp. Comm., 2011, 34, 319-325.<br />
15. C. Chen, D. He, S. Chan, J. Bu, Y. Gao and R. Fan, “Lightweight and provably secure user<br />
authentication with anonymity for the global mobility network”, Int. J. Comm. Syst., 2011, 24,<br />
347-362.<br />
16. J. Horwitz and B. Lynn, “Toward hierarchical identity-based encryption”, Proceedings <strong>of</strong> the<br />
21st <strong>International</strong> Conference on the Theory and Applications <strong>of</strong> Cryptographic Techniques,<br />
2002, Amsterdam, Netherlands, pp.466-481<br />
17. D. Boneh and M. Franklin, “Identity-based encryption from the weil pairing”, Proceedings <strong>of</strong><br />
the 21 st <strong>International</strong> Cryptology Conference on Advances in Cryptology, 2001, Santa Barbara,<br />
California, USA, pp.213-229<br />
18. Z. Wan, K. Ren and B. Preneel, “A secure privacy-preserving roaming protocol based on<br />
hierarchical identity-based encryption for mobile networks”, Proceedings <strong>of</strong> the 1st ACM<br />
Conference on Wireless Network Security, 2008, New York, USA, pp.62-67.<br />
19. P. Kocher, J. Jaffe and B. Jun, “Differential power analysis”, Proceedings <strong>of</strong> the 19th Annual<br />
<strong>International</strong> Cryptology Conference, 1999, Santa Barbara, California, USA, pp.388-397.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
70 <strong>Maejo</strong> <strong>Maejo</strong> Int. J. Sci. Int. Technol. J. Sci. Technol. <strong>2012</strong>, 6(01), <strong>2012</strong>, 70-76 6(01), 70-76<br />
Full Paper<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Influence <strong>of</strong> MeV H + ion beam flux on cross-linking and blister<br />
formation in PMMA resist<br />
Somrit Unai 1, 2, * , Nitipon Puttaraksa 1, 3 , Nirut Pussadee 1, 2 , Kanda Singkarat 4 ,<br />
Michael W. Rhodes 5 , Harry J. Whitlow 3 and Somsorn Singkarat 1, 2<br />
1 Plasma and Beam Physics Research Facility, Department <strong>of</strong> Physics and Materials <strong>Science</strong>,<br />
Faculty <strong>of</strong> <strong>Science</strong>, Chiang Mai University, Chiang Mai 50200, Thailand<br />
2<br />
Thailand Centre <strong>of</strong> Excellence in Physics, CHE, 328 Si Ayutthaya Road, Bangkok 10400, Thailand<br />
3 Department <strong>of</strong> Physics, University <strong>of</strong> Jyväskylä, P.O. Box 35 (YFL), FIN – 40014, Finland<br />
4 Department <strong>of</strong> Physics and Materials <strong>Science</strong>, Faculty <strong>of</strong> <strong>Science</strong>, Chiang Mai University,<br />
Chiang Mai 50200, Thailand<br />
5 <strong>Science</strong> and Technology Research Institute, Chiang Mai University, Chiang Mai 50200, Thailand<br />
* Corresponding author, e-mail: somrit@fnrf.science.cmu.ac.th<br />
Received: 31 March 2011 / Accepted: 13 February <strong>2012</strong> / Published: 14 February <strong>2012</strong><br />
Abstract: In s<strong>of</strong>t lithography, a pattern is produced in poly(dimethylsiloxane) (PDMS)<br />
elastomer by casting from a master mould. The mould can be made <strong>of</strong> poly(methylmethacrylate)<br />
(PMMA) resist by utilising either its positive or negative tone induced by an ion beam. Here we<br />
have investigated the irradiation conditions for achieving complete cross-linking and absence <strong>of</strong><br />
blister formation in PMMA so that its negative characteristic can be used in making master<br />
moulds. PMMA thin films approximately 9 μm thick on Si were deposited by spin coating. The<br />
2-MeV H + ion beam was generated using a 1.7-MV tandem Tandetron accelerator. The beam<br />
was collimated to a 500×500 μm 2 cross section using programmable proximity aperture<br />
lithography system with a real-time ion beam monitoring system and a high precision current<br />
integrator. The irradiated areas were investigated by a standard scanning electron microscope<br />
and a pr<strong>of</strong>ilometer. It was found that both the ion beam flux and the stopping power <strong>of</strong> the ions<br />
in the polymer have a critical influence on the blister formation.<br />
Keywords: s<strong>of</strong>t lithography, H + ion irradiation, PMMA resist, blister formation
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 70-76<br />
71<br />
INTRODUCTION<br />
Numerous studies <strong>of</strong> ion-beam-induced modification <strong>of</strong> polymers have been performed in the<br />
last three decades because <strong>of</strong> its potential for technological applications [e.g. 1-3]. In recent years, ion<br />
beam lithography on a resist for the development <strong>of</strong> micr<strong>of</strong>luidic devices has been extensively<br />
investigated [4-6]. The resist that has the tendency to form cross-linkage <strong>of</strong> the main polymer chains or<br />
pendant side chains during irradiation is <strong>of</strong> the negative type (also called negative tone). This is<br />
because the irradiated area becomes less soluble in developing solutions compared to an unirradiated<br />
region. On the other hand, the predominant modification in a positive resist (also called positive tone)<br />
is chain scissioning <strong>of</strong> the main and side polymer chains, which enhances the solubility rate in the<br />
exposed region [7]. Typically, the master mould for s<strong>of</strong>t lithography is fabricated from either an ionirradiated<br />
negative resist such as SU-8 [8] or a positive resist such as poly(methyl methacrylate)<br />
(PMMA) [5]. In addition, Licciardello et al. [9] found that the PMMA would undergo a changeover<br />
from a positive to a negative tone by the action <strong>of</strong> high fluence irradiation, but the utilisation <strong>of</strong><br />
PMMA as a negative tone resist for master mould fabrication has been little used [10]. For both<br />
positive and negative resists, a good master mould used in a direct replica moulding technique must<br />
have smooth surfaces. However, blisters or craters have been observed on irradiated polymer surfaces<br />
[11-12]. These defects can spoil the mould since they may introduce blockages into<br />
poly(dimethylsiloxane) (PDMS) replicas for micr<strong>of</strong>luidic networks. Consequently, blisters and craters<br />
must be avoided. This paper reports on an experimental study to control defects from 2-MeV H + ion<br />
irradiation until the dominance <strong>of</strong> cross-linking in PMMA has been achieved. Some effects <strong>of</strong> 1-MeV<br />
H + ion beam are also included for comparison.<br />
PRINCIPLE<br />
According to the stopping and range <strong>of</strong> ions in matter (SRIM) simulation program, version<br />
SRIM-2008 [13], for the projected range <strong>of</strong> 65 μm, the tracks <strong>of</strong> 2-MeV H + ions in PMMA is straight,<br />
especially in the first 10 μm. An individual ion loses about 200 keV <strong>of</strong> kinetic energy during its<br />
passage through the first 10 μm within 0.5 ps. For light incident ions with a velocity much greater than<br />
the Bohr velocity (2.2×10 6 m/s) in thin polymer targets, the energy deposited by the ion mainly results<br />
in the excitation and ionisation <strong>of</strong> atoms and molecules <strong>of</strong> the polymer in the ion-track zone. These<br />
initial physical processes <strong>of</strong> ion-solid interaction can cause chain scission, cross-linking and gas<br />
evolution [14]. In the case <strong>of</strong> PMMA, several simple low-mass gas products such as H 2 , CO, CO 2 and<br />
CH 4 have been detected, with carbon monoxide as the dominant product [15-16]. This is indicative <strong>of</strong><br />
bond breaking, which is a precursor <strong>of</strong> both the chain scission and cross-linking. There were reports<br />
that the gas yield from the polymers is strongly dependent on the linear energy transfer (LET) <strong>of</strong> the<br />
ions [17] and on the incident ion fluence [18]. Although the fundamental mechanism is still not fully<br />
understood, the proposed models [11-12] agree that blisters and craters on the surface <strong>of</strong> a polymer are<br />
formed by the gas products as they pass into the surrounding vacuum.
72 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 70-76<br />
MATERIALS AND METHODS<br />
The PMMA used in this work had a molecular weight <strong>of</strong> 950 kDa (950 PMMA A11,<br />
MicroChem). All PMMA films were spin-coated on clean 1×1-cm 2 silicon substrates at a spin speed <strong>of</strong><br />
2,500 rpm for 45 sec. by using a self-made spin coater. Subsequently, the films were s<strong>of</strong>t-baked on a<br />
hot-plate at 160 o C for 2 min. This process was repeated three times to attain a total film thickness <strong>of</strong><br />
8.8±0.1 μm as measured with a stylus pr<strong>of</strong>ilometer (P-15 Pr<strong>of</strong>iler, KLA-Tencor, USA). The 1.7-MV<br />
tandem accelerator (1.7 MV high current Tandetron, High Voltage Engineering Europa B. V., the<br />
Netherlands) was used to irradiate the polymer with 1- and 2-MeV H + ion beams. The pressure during<br />
irradiation was about 5×10 -6 mbar. The irradiated area was 500×500 μm 2 for all experiments and was<br />
defined by two computerised L-shaped blade apertures <strong>of</strong> the programmable proximity aperture<br />
lithography (PPAL) system [10]. The experimental set-up when utilising the PPAL technique is shown<br />
in Figure 1. The two L-shaped blades were made <strong>of</strong> copper plates 100 μm thick with well-polished<br />
edges. Each <strong>of</strong> the L-shaped blades was mounted on a computerised micro-stepper motor that<br />
independently moved with high precision in either vertical or horizontal direction. In this manner, the<br />
PPAL system could produce any rectangular pattern with dimensions between 1-1000 μm 2 . The<br />
aperture was located at ~2 mm in front <strong>of</strong> the sample. The PMMA film on silicon substrate was<br />
mounted to the translation stage holder, which could move in both x and y directions with a resolution<br />
<strong>of</strong> 1 μm.<br />
x<br />
Electron<br />
suppressor<br />
Aperture<br />
y<br />
Sample<br />
Faraday cup<br />
A<br />
Pico-ammeter<br />
Sample holder<br />
~3mm <br />
2-MeV H +<br />
ion beam<br />
-200 V<br />
Alumina<br />
screen<br />
A<br />
Pico-ammeter<br />
Figure 1. Schematic illustration <strong>of</strong> an experimental set-up using the PPAL technique for H + ion beam<br />
irradiation (not to scale)<br />
The key parameters in this study were the values <strong>of</strong> ion beam fluence (ions/cm 2 ) and ion beam<br />
flux (ions/cm 2·s) at the irradiated areas. To ensure the accuracy <strong>of</strong> these measurements, the ion beam<br />
current at two different positions, as shown in Figure 1, was measured with high precision. A<br />
picoammeter with 4.8-pC resolution was specially designed for this purpose. To assure correctness in<br />
measuring, an electron suppressor with a 5-mm-diameter hole and a -200V potential was placed in<br />
front <strong>of</strong> the aperture to prevent secondary electrons from leaving the copper blades.<br />
The picoammeter was used to measure the ion beam current detected by an 8-mm-innerdiameter,<br />
65-mm-long Faraday cup. The Faraday cup was isolated and buried in the sample holder.<br />
The ion beam current was collected every second from start to stop. Together with the known hole area
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 70-76<br />
73<br />
at the aperture, the ion beam fluence and flux could be easily calculated. Moreover, the major part <strong>of</strong><br />
the incident ion beam blocked by the aperture was monitored also; this was used to fine-tune the<br />
accelerator for a stable ion beam current.<br />
After irradiation, the surface morphology <strong>of</strong> the irradiated areas was investigated by an optical<br />
microscope and a scanning electron microscope (SEM) (JSM-5410LV, JEOL, Japan), while the<br />
shrinkage <strong>of</strong> the irradiated areas was measured by the pr<strong>of</strong>ilometer.<br />
RESULTS AND DISCUSSION<br />
It was found that for 2-MeV H + ions, the cross-linking <strong>of</strong> PMMA within the entire pattern<br />
occurred when the fluence exceeded 3.5×10 14 ions/cm 2 , which is consistent with a previous report [19].<br />
Therefore, all the incident fluences used in this work were above this value. Accordingly, in Figure 2<br />
both a and d, b and e, and c and f patterns were irradiated by the 2-MeV H + ion beam with the same<br />
fluence, viz. 1.0×10 15 , 1.25×10 15 and 1.75×10 15 ions/cm 2 respectively, while the ion beam flux used<br />
for patterns a-c and d-f was 4.7×10 11 and 3.0×10 12 ions/cm 2·s respectively. It was clearly seen that<br />
flaws on the PMMA surface were strongly dependent on the ion beam flux. For the same irradiation<br />
fluence but with small values <strong>of</strong> ion current, the flaws were absent. Figure 3(a) shows that the flaws<br />
are blisters and not craters. Each isolated blister has a common appearance <strong>of</strong> a round shape with about<br />
the same diameter.<br />
Figure 2. Optical microscope images (13×) <strong>of</strong> the six 500×500 μm 2 patterns <strong>of</strong> irradiated areas on the<br />
PMMA thin film<br />
The surface pr<strong>of</strong>iles <strong>of</strong> patterns a, b and c in Figure 2 were measured by the pr<strong>of</strong>ilometer. As<br />
shown in Figure 4, the surface regions modified by the 2-MeV H + ion beam are remarkably lower than<br />
those in the unirradiated regions. Also, the compaction or shrinking occurs even at the smooth patterns<br />
and increases with ion fluence. Hnatowicz and Fink [20] reported that the compaction <strong>of</strong> a polymer is<br />
initiated by cross-linking and tends to increase the material density. Experimental evidence from this<br />
study suggests that the gas evolution always occurs during irradiation even though no blister is
74 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 70-76<br />
observed, as shown in Figure 2 (a-c). However, the gas-release mechanisms for low-flux and high-flux<br />
irradiation may be different. More information can be found in a comparison <strong>of</strong> the smooth pattern a<br />
with the blister-filled pattern d, for example. Both <strong>of</strong> them experienced the same fluence but the<br />
irradiation duration <strong>of</strong> pattern a was 1.72 times longer than that <strong>of</strong> pattern d.<br />
Figure 3. Tilted SEM images <strong>of</strong> the blisters created by (a) 2-MeV H + ions and (b) 1-MeV H + ions.<br />
The average blister diameters are (a) 61.0±5.8 μm and (b) 92.8±2.5 μm.<br />
Figure 4. The surface pr<strong>of</strong>ile <strong>of</strong> the 3 irradiated areas in the top row <strong>of</strong> Figure 2<br />
It is reasonable to draw a conclusion that each ion track can be considered isolated in the case<br />
<strong>of</strong> low flux. Degassing due to the pressure gradient is thought to be the main mechanism. Moreover,<br />
the very thin PMMA film would allow the gases to leave the polymer easily without much build-up.<br />
On the other hand, in the high ion flux case, the temporal interval between the closely-spaced incident<br />
ion tracks would allow a build-up <strong>of</strong> dense low-molecular-weight fragments from intense main-chain<br />
scissions. This might nucleate gas bubbles in the whole irradiated volume <strong>of</strong> the polymer and<br />
accumulate gases which eventually converted it into a ‘foamed’ structure [21]. In this case, the gas<br />
bubbles that reached the polymer surface would form blisters.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 70-76<br />
75<br />
It was also found that a 1-MeV H + ion beam creates blisters at lower ion beam flux than does a<br />
2-MeV H + ion beam. By comparing Figure 3(b) with Figure 3(a), we can see that although each<br />
isolated blister has the same shape, the blisters from 1-MeV irradiation are significantly larger in<br />
diameter than those from 2-MeV irradiation. From the SRIM code, the stopping power is 1.5 times and<br />
the displacement damage 2.4 times greater in 1-MeV than in 2-MeV proton irradiation.<br />
CONCLUSIONS<br />
At an ion irradiation fluence <strong>of</strong> 10 14 ions/cm 2 <strong>of</strong> 2-MeV H + ions, the cross-linking process in<br />
PMMA started to overcome the chain scission process. The full cross-linking began at an ion fluence<br />
<strong>of</strong> 6.6×10 14 ions/cm 2 . The formation <strong>of</strong> blisters was strongly dependent on the ion beam flux and<br />
stopping power <strong>of</strong> the polymer for the ions. A 2-MeV proton flux <strong>of</strong> less than 4.7×10 11 ions/cm 2·s<br />
achieved a blister-free condition for the PMMA film ~9 μm thick. This work has confirmed that<br />
blisters are created by gas evolution due to chain scissions induced by the ion beams, especially in the<br />
early stage <strong>of</strong> irradiation.<br />
ACKNOWLEDGEMENTS<br />
This work was partly supported by the CoE Program <strong>of</strong> Chiang Mai University and the<br />
<strong>International</strong> Atomic Energy Agency (IAEA, Vienna). S.U. gratefully acknowledges a scholarship<br />
from the ThEP Centre. N. P. (N. Puttaraksa) gratefully acknowledges financial support from Thailand<br />
Research Fund (TRF) in the form <strong>of</strong> an RGJ scholarship. H. J. W.’s and N. P.’s work was carried out<br />
under the Academy <strong>of</strong> Finland Centre <strong>of</strong> Excellence in Nuclear and Accelerator Based Physics (Ref.<br />
213503). H. J. W. is grateful for a senior researcher grant from the Academy <strong>of</strong> Finland (Ref. 129999).<br />
The authors also wish to thank Mr. Chome Thongleurm and Mr. Witoon Ginamoon for their technical<br />
support.<br />
REFERENCES<br />
1. T. M. Hall, A. Wagner and L. F. Thompson, "Ion beam exposure characteristics <strong>of</strong> resists:<br />
Experimental results", J. Appl. Phys., 1982, 53, 3997-4010.<br />
2. S. V. Springham, T. Osipowicz, J. L. Sanchez, L. H. Gan and F. Watt, "Micromachining using<br />
deep ion beam lithography", Nucl. Instr. Meth. Phys. Res. B, 1997, 130, 155-159.<br />
3. F. Menzel, D. Spemann, S. Petriconi, J. Lenzner and T. Butz, "Proton beam writing <strong>of</strong><br />
submicrometer structures at LIPSION", Nucl. Instr. Meth. Phys. Res. B, 2007, 206, 419-425.<br />
4. P. G. Shao, J. A. van Kan, K. Ansari, A. A. Bettiol and F. Watt, "Poly(dimethylsiloxane)<br />
micro/nanostructure replication using proton beam written masters", Nucl. Instr. Meth. Phys.<br />
Res. B, 2007, 260, 479-482.<br />
5. S. Gorelick, N. Puttaraksa, T. Sajavaara, M. Laitinen, S. Singkarat and H. J. Whitlow,<br />
"Fabrication <strong>of</strong> micr<strong>of</strong>luidic devices using MeV ion beam Programmable Proximity Aperture<br />
Lithography (PPAL)", Nucl. Instr. Meth. Phys. Res. B, 2008, 266, 2461-2465.<br />
6. C. Udalagama, E. J. Teo, S. F. Chan, V. S. Kumar, A. A. Bettiol and F. Watt, "Proton beam<br />
writing <strong>of</strong> long, arbitrary structures for micro/nano photonics and fluidics applications", Nucl.<br />
Instr. Meth. Phys. Res. B, 2011, 269, 2417-2421.
76 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 70-76<br />
7. M. J. Madou, "Fundamentals <strong>of</strong> Micr<strong>of</strong>abrication: The <strong>Science</strong> <strong>of</strong> Miniaturization", 2nd Edn.,<br />
CRC Press, Boca Raton, 2002, pp.6-7.<br />
8. V. Auzelyte, M. Elfman, P. Kristiansson, C. Nilsson, J. Pallon, N. A. Marrero and M. Wegdén,<br />
"Exposure parameters for MeV proton beam writing on SU-8", Microelectron. Eng., 2006, 83,<br />
2015-2020.<br />
9. A. Licciardello, M. E. Fragalà, G. Foti, G. Compagnini and O. Puglisi, "Ion beam effects on the<br />
surface and on the bulk <strong>of</strong> thin films <strong>of</strong> polymethylmethacrylate", Nucl. Instr. Meth. Phys. B,<br />
1996, 116, 168-172.<br />
10. N. Puttaraksa, S. Unai, M. W. Rhodes, K. Singkarat, H. J. Whitlow and S. Singkarat,<br />
"Fabrication <strong>of</strong> a negative PMMA master mold for s<strong>of</strong>t-lithography by MeV ion beam<br />
lithography", Nucl. Instr. Meth. Phys. B, <strong>2012</strong>, 272, 149-152.<br />
11. J. Kaur, S. Singh, D. Kanjilal and S. K. Chakarvarti, "Nano/micro surface structures by swift<br />
heavy ion irradiation <strong>of</strong> polymeric thin films on GaAs", Digest J. Nanomater. Biostruct., 2009,<br />
4, 729-737.<br />
12. D. He and M. N. Bassim, "Atomic force microscope study <strong>of</strong> crater formation in ion<br />
bombarded polymer", J. Mater. Sci., 1998, 33, 3525-3528.<br />
13. J. F. Ziegler and J. P. Biersack, “The stopping and range <strong>of</strong> ions in matter”, 2008,<br />
http://www.srim.org.<br />
14. M. E. Fragalà, G. Compagnini, A. Licciardello and O. Puglisi, "Track overlap regime in ionirradiated<br />
PMMA", J. Polym. Sci. B, 1998, 36, 655-664.<br />
15. Z. Chang and J. A. LaVerne, "The gases produced in gamma and heavy-ion radiolysis <strong>of</strong><br />
poly(methyl methacrylate)", Radiat. Phys. Chem., 2001, 62, 19-24.<br />
16. M. E. Fragalà, G. Compagnini, L. Torrisi and O. Puglisi, "Ion beam assisted unzipping <strong>of</strong><br />
PMMA", Nucl. Instr. Meth. Phys. B, 1998, 141, 169-173.<br />
17. M. B. Lewis and E. H. Lee, "G-values for gas production from ion-irradiated polystyrene", J.<br />
Nucl. Mater., 1993, 203, 224-232.<br />
18. F. Schrempel and W. Witthuhn, "Deep light ion lithography in PMMA __ a parameter study",<br />
Nucl. Instr. Meth. Phys. B, 1997, 132, 430-438.<br />
19. N. Puttaraksa, R. Norarat, M. Laitinen, T. Sajavaara, S. Singkarat and H. J. Whitlow,<br />
"Lithography exposure characteristics <strong>of</strong> poly(methyl methacrylate) (PMMA) for carbon,<br />
helium and hydrogen ions", Nucl. Instr. Meth. Phys. B, <strong>2012</strong>, 272, 162-164.<br />
20. V. Hnatowicz and D. Fink, "Macroscopic changes in ion-irradiated polymers", in<br />
"Fundamentals <strong>of</strong> Ion-Irradiated Polymers" (Ed. D. Fink), Springer-Verlag, Berlin, 2004,<br />
pp.349-375.<br />
21. A. Chapiro, "Chemical modifications in irradiated polymers", Nucl. Instr. Meth. Phys. B, 1988,<br />
32, 111-114.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
<strong>Maejo</strong> <strong>Maejo</strong> Int. J. Int. Sci. J. Technol. Sci. Technol. <strong>2012</strong>, 6(01), <strong>2012</strong>, 6(01), 77-94 77-9477<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
Full Paper<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Degradation <strong>of</strong> bisphenol A by ozonation: rate constants,<br />
influence <strong>of</strong> inorganic anions, and by-products<br />
Kheng Soo Tay * , Noorsaadah Abd. Rahman and Mhd. Radzi Bin Abas<br />
Environmental Research Group, Department <strong>of</strong> Chemistry, Faculty <strong>of</strong> <strong>Science</strong>, University <strong>of</strong> Malaya,<br />
50603 Kuala Lumpur, Malaysia<br />
* Corresponding author, e-mail: khengsoo@um.edu.my<br />
Received: 8 February 2011 / Accepted: 23 February <strong>2012</strong> / Published: 27 February <strong>2012</strong><br />
Abstract: The second-order rate constants for the reaction between bisphenol A (BPA) and ozone<br />
were evaluated over the pH range <strong>of</strong> 2-12. The rate constants showed minimum values (×10 4 M -1 s -1 )<br />
under acidic condition (pH < 4) and were <strong>of</strong> maximum values (×10 9 M -1 s -1 ) under basic condition<br />
(pH >10). From pH 4 to 7, the second-order rate constants were found to increase by a magnitude<br />
<strong>of</strong> almost 10 2 and this was due to the increase in anionic BPA species in the solution. The rate<br />
constants increased almost tw<strong>of</strong>old when pH increased from 9.6 to 10.2. The presence <strong>of</strong> common<br />
inorganic anions at levels commonly found in the environment did not affect the rate <strong>of</strong> degradation<br />
<strong>of</strong> BPA.<br />
The degradation by-products from the ozonation <strong>of</strong> BPA were identified as 4-(prop-1-en-2-<br />
yl)phenol, hydroquinone, 4-hydroxyacetophenone, 2-(2-(4-hydroxyphenyl)propan-2-yl)succinaldehyde,<br />
2-(1-(4-hydroxyphenyl)vinyl)pent-2-enal, 3-formyl-4-(4-hydroxyphenyl)-4-methylpent-2-<br />
enoic acid, monohydroxy-BPA and dihydroxy-BPA. In conclusion, ozonation was found to be an<br />
effective method for the removal <strong>of</strong> BPA even in the presence <strong>of</strong> common inorganic anions at<br />
environmental concentrations. However, incomplete treatment <strong>of</strong> BPA might produce a variety <strong>of</strong><br />
degradation by-products.<br />
Keywords: bisphenol A, ozonation, rate constant, anions, competitive kinetics, OH radical<br />
________________________________________________________________________________<br />
INTRODUCTION<br />
Bisphenol-A (BPA; 2,2-bis(4-hydroxylphenyl)propane) is a commonly used chemical in the<br />
synthesis <strong>of</strong> polymers, especially for food and beverage packages. BPA has been reported to have<br />
developmental toxicity, carcinogenicity, possible neurotoxicity [1] and estrogenic effects [2]. This
78 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
chemical can leach from plastic products under normal and high temperatures [2-3]. Thus, the<br />
leaching <strong>of</strong> BPA into the environment from disposed plastic materials is expected. So far, legislation<br />
on the use <strong>of</strong> BPA and its discharge into the environment is sorely lacking. Consequently, the<br />
occurrences <strong>of</strong> BPA in the environment have been widely reported [e.g. 4-6]. The presence <strong>of</strong> BPA<br />
in coastal waters and supermarket seafood from Singapore and mussels from South and South-east<br />
Asia [7-8] indicates that BPA pollution is severe not only in developed countries but also in the<br />
Asian region.<br />
Since BPA has the potential to cause undesirable ecological and human health effects, various<br />
treatment technologies have been developed for its removal from water. The oxidative degradation<br />
methods for BPA such as ozonation [9-14], photo-Fenton reaction [15], photocatalytic reaction by<br />
TiO 2 [16] and ultrasound-UV-iron(II) treatment [17] have been reported. Among the treatment<br />
methods, ozonation has been projected to be one <strong>of</strong> the fastest growing water disinfection<br />
technologies in the market [18-19]. It has been shown to be effective in the removal <strong>of</strong> organic<br />
pollutants from water and wastewater [20-24]. During ozonation, organic pollutants undergo a series<br />
<strong>of</strong> oxidation processes by ozone (O 3 ) and hydroxyl radical (•OH) formed by the decomposition <strong>of</strong> O 3<br />
in water [25]. In some cases, toxic by-products may be generated [22]. Therefore, evaluation and<br />
determination <strong>of</strong> degradation by-products from the ozonation <strong>of</strong> organic pollutants is an important<br />
consideration.<br />
Although the ozonation <strong>of</strong> BPA has been widely reported, based on our literature review the<br />
influence <strong>of</strong> inorganic anions on the removal <strong>of</strong> BPA by ozonation has not been evaluated. The<br />
influence <strong>of</strong> inorganic anions is important since chloride and phosphate ions, for example, can react<br />
with O 3 and •OH, thus affecting the rate <strong>of</strong> removal <strong>of</strong> the organic pollutants in water [26-27]. The<br />
increase in salt concentration has also been reported to affect O 3 solubility in water [28]. The main<br />
objective <strong>of</strong> this study, therefore, is to evaluate the effect <strong>of</strong> inorganic anions, namely phosphate,<br />
nitrate, sulphate and chloride ions. In addition, the second-order rate constants for the reaction<br />
between BPA and O 3 at pH 2-12 were determined. The variation in the rate constant at different pH<br />
values, especially within the two pK a values (9.6 and 10.2) <strong>of</strong> BPA, was studied in detail in order to<br />
compare the reactivity <strong>of</strong> anionic and dianionic species <strong>of</strong> BPA in aqueous solution towards O 3 . The<br />
degradation by-products (DBPs) <strong>of</strong> BPA were also identified. Some DBPs <strong>of</strong> BPA have already been<br />
determined by previous studies [11,13]. However, in this study, we managed to identify a few<br />
additional compounds and degradation pathways <strong>of</strong> BPA during ozonation were proposed.<br />
MATERIALS AND METHODS<br />
Chemicals<br />
BPA (>99% purity) was obtained from Aldrich and used without further purification. Its<br />
stock solution (200 mg/L) was prepared by dissolving in boiling ultrapure deionised water (Elcagan,<br />
UK). All chemicals were used without further purification. Sodium phosphate (96%) and sodium<br />
dihydrogen phosphate (99%) was obtained from Aldrich. Sodium chloride (≥99.5%), sodium nitrate<br />
(99%) sodium sulphate decahydrate (>99%), sodium phosphate monobasic (99%) and tert-butyl<br />
alcohol (t-BuOH) (99.5%) were purchased from Sigma. Sodium phosphate dibasic (99%) and<br />
disodium hydrogen phosphate (99%) were purchased from Riedel-de-Haën. A mixture <strong>of</strong> BSTFA
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
79<br />
(N,O-bis(trimethylsilyl)trifluoroacetamide) and TMCS (trimethylchlorosilane) in a ratio <strong>of</strong> 99:1 was<br />
obtained from Supelco. All solvents (HPLC grade) and phosphoric acid (85%) were obtained from<br />
Merck. Sodium hydroxide (>98%) was purchased from Fluka. Purified oxygen (99.8%) was<br />
obtained from MOX-linde (Malaysia). Phosphate buffer (0.5 mol/L) was prepared using sodium<br />
dihydrogen phosphate and/or disodium hydrogen phosphate and the pH was adjusted using either<br />
phosphoric acid or sodium hydroxide solutions. O 3 was produced from purified oxygen by an ozone<br />
generator (model OZX03K, Enaly Trade Co. Ltd., Canada). All tubing from the ozone generator<br />
was <strong>of</strong> ozone-inert silicone material.<br />
Ozonation <strong>of</strong> BPA<br />
Determination <strong>of</strong> second-order rate constant for reaction <strong>of</strong> BPA with O 3<br />
A detailed description <strong>of</strong> the rate constant determination was given in our previous study<br />
[29]. Briefly, a competitive kinetic method was applied using phenol as a reference compound.<br />
Experiments were performed at room temperature (27-30ºC) in 20-mL vials with solution containing<br />
equal concentration <strong>of</strong> BPA and phenol (4 μmol/L) as well as 20 mmol/L <strong>of</strong> t-BuOH. The pH <strong>of</strong> the<br />
solution was adjusted using 20 mmol/L <strong>of</strong> phosphate buffer. Ozone solutions with concentrations<br />
ranging from 1.5 to 7.5 μmol/L were added. The aqueous O 3 stock solution was prepared by<br />
sparging O 3 at a rate <strong>of</strong> 0.70 g/hr into deionised water placed inside a water-jacketed beaker at 2ºC.<br />
The concentration <strong>of</strong> O 3 was measured by the indigo method [30]. The final volume <strong>of</strong> the mixtures<br />
was 20 mL, which also minimised the headspace. Each vial was then shaken vigorously. The time <strong>of</strong><br />
total ozone consumption was estimated from the half-life <strong>of</strong> ozone in the BPA solution under<br />
different pH conditions. The half-lives for ozone in BPA solution were found to be 28, 9 and
80 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
hr. The silylated extract was dried with nitrogen stream and redissolved in 30 μL <strong>of</strong><br />
dichloromethane,1.2 μL <strong>of</strong> which was analysed by gas chromatography-mass spectrometry (GC-<br />
MS).<br />
Analytical methods<br />
The concentration <strong>of</strong> BPA in the reaction mixture was determined using HPLC (Thermo<br />
Separation Product HPLC System P2000, HiTech Trader, USA) equipped with a UV detector and a<br />
degasser. A 250×4.6mm RP-8 (5μm) Lichrospher-100 analytical column (Merck) was used for<br />
separation. Acetonitrile (65%) in deionised water with 0.1% trifluoroacetic acid was used as the<br />
mobile phase. The separation was carried out under isocratic condition. The separated components<br />
were detected at 230 nm. The flow rate was maintained at 1.0 mL/min for all runs and the sample<br />
volume for HPLC analysis was 20 μL.<br />
The analysis <strong>of</strong> the degradation by-products was carried out using a Hewlett-Packard HP<br />
6890 gas chromatograph coupled with HP5972 mass spectrometer. The column was HP-5 (5%<br />
phenylmethylpolysiloxane) column with the dimension <strong>of</strong> 30 m 0.25 mm and 0.25 μm <strong>of</strong> film<br />
thickness. Helium was used as the carrier gas with an average flow rate <strong>of</strong> 40 cm/sec., and the GC<br />
oven temperature was initially 60°C (maintained for 2 min.) and increased to 280°C at the rate <strong>of</strong><br />
6°C/min and maintained at this temperature for 2 min. The temperatures <strong>of</strong> the injection port and the<br />
transfer line were set at 290°C and 300°C respectively. The data for quantitative analysis was<br />
acquired in the electron impact mode (70 eV) with scanning in the range <strong>of</strong> 50-600 amu at 1.5<br />
sec./scan.<br />
RESULTS AND DISCUSSION<br />
Kinetics <strong>of</strong> Degradation <strong>of</strong> BPA by Ozonation<br />
The reaction <strong>of</strong> ozone with an organic compound has been reported to be <strong>of</strong> first order with<br />
respect to both ozone and the organic compound. Thus, the kinetics <strong>of</strong> ozonation can be expressed<br />
as a second-order reaction [32]. The determination <strong>of</strong> its rate constants was performed using a<br />
competitive kinetics method with phenol as reference compound [20, 29, 33-34]. Phenol was<br />
selected because it was expected to have a similar decomposition pathway and rate constant as BPA<br />
in the ozonation [34]. For phenol, the second-order rate constants for the reaction with O 3 at<br />
different pH values ( k<br />
app,Phenol-O<br />
) were calculated using equation 1:<br />
3<br />
-pH<br />
-pKa<br />
10 10<br />
kapp,Phenol-O k <br />
3 O 3,Phenol -pKa -pH <br />
k<br />
O 3,Phenolate<br />
<br />
-pKa -pH <br />
10 +10 10 +10 <br />
(1)<br />
where pK a <strong>of</strong> phenol was 9.9 and the intrinsic rate constants for phenol and phenolate were 1.3 ×10 3<br />
M -1 s -1 ( k<br />
O 3 ,Phenol) and 1.4 ×10 9 M -1 s -1 ( k<br />
O 3 ,Phenolate<br />
) respectively [9, 34]. The rate constants for the<br />
reaction between O 3 and BPA ( ) were estimated using competitive kinetics method derived<br />
from equation 2 [35, 36]:<br />
kO3<br />
BPA
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
81<br />
[BPA]<br />
k<br />
n O3<br />
BPA<br />
[Phenol]<br />
<br />
n<br />
ln <br />
ln <br />
[BPA]<br />
k [Phenol]<br />
<br />
0 app,O3<br />
-Phenol 0<br />
(2)<br />
where [BPA]<br />
0<br />
and [Phenol]<br />
0<br />
represent the initial concentration <strong>of</strong> BPA and phenol and [BPA] n<br />
and<br />
[Phenol] n<br />
represent the concentration <strong>of</strong> BPA and phenol after the ozonation reaction at different<br />
ozone dose, n.<br />
Ozonation <strong>of</strong> BPA was carried out at pH 2.0-12.0 in the presence <strong>of</strong> excess t-BuOH<br />
([t-BuOH] / [O 3 ] > 300). t-BuOH is a radical scavenger added to scavenge the •OH radical, thus<br />
allowing BPA to react only with O 3 during ozonation. Lee et al. [9] and Deborde et al. [34] reported<br />
k at 20°C = 1.68×10 4 and 1.30×10 4 M -1 s -1 at pH 2, and 1.11×10 9 and 1.60×10 9 M -1 s -1 at pH 12<br />
o3<br />
-BPA<br />
respectively. In this work, somewhat higher values <strong>of</strong> ko3<br />
-BPA<br />
were obtained, viz. (1.7±0.5)×10 4 and<br />
(9.0±0.5)×10 9 M -1 s -1 at pH 2 and 12 respectively. This difference might be due to a large error that<br />
can occur in the competitive kinetics method [20] and also to a higher reaction temperature that was<br />
selected in this study.<br />
According to Staples et al. [37], the two dissociation constants <strong>of</strong> BPA are 9.6 (pK a 1) and<br />
10.2 (pK a 2). BPA is therefore a weak organic acid which can dissociate in solution as either an<br />
anionic or dianionic species. Generally, when pH = pK a , the undissociated and ionic species exist at<br />
equal concentration in solution. When pH < pK a , the undissociated species is predominant and when<br />
pH > pKa, the ionic species is predominant [38]. Therefore, for BPA, when pH < pK a 1,<br />
undissociated BPA exists predominantly in water. When pK a 1 < pH < pK a 2, anionic BPA is the<br />
predominant species and when pH > pK a 2, dianionic BPA are predominant (Figure 1).<br />
Figure 1. Dissociation <strong>of</strong> BPA<br />
It has been previously shown that in the presence <strong>of</strong> the •OH radical, the rate <strong>of</strong> BPA<br />
degradation by ozone increases from pH 2 to 7 and then decreases at pH 10 due to •OH scavenging<br />
by carbonate and bicarbonate ions [14]. In this study, t-BuOH was added to scavenge the •OH<br />
radical from the beginning in order to study the rate constants for BPA degradation by O 3 only. The<br />
results are shown in Figure 2, indicating that the reactivity increases as the pH increases from 2 to<br />
12. Under acidic condition (pH < 7), the reaction would be mainly between O 3 and undissociated<br />
BPA. Between pH 4-7, the rate constant was observed to increase by a magnitude <strong>of</strong> almost 10 2 and<br />
this would be due to the increase <strong>of</strong> anionic species <strong>of</strong> BPA in the solution. The rate constant<br />
increased almost tw<strong>of</strong>old when the pH increased from 9.6 (= pK a 1) to 10.2 (= pK a 2).
82 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
10 9<br />
10 8<br />
Rate constants (M -1 s -1 )<br />
10 7<br />
10 6<br />
10 5<br />
10 4<br />
10 3<br />
pH<br />
2 4 6 8 10 12 14<br />
pH<br />
Rate constants (M -1 s -1 )<br />
1.0x10 9<br />
8.0x10 8<br />
pK a<br />
2<br />
6.0x10 8<br />
4.0x10 8<br />
2.0x10 8<br />
0.0<br />
pK a<br />
1<br />
6 8 10 12 14<br />
Figure 2. Variation <strong>of</strong> second-order rate constant <strong>of</strong> BPA-ozone reaction at different pH values<br />
In order to study the effect <strong>of</strong> the ratio <strong>of</strong> [BPA 2- ] / [BPA - ] on the rate constant, the obtained<br />
second-order rate constants were plotted against [BPA 2- ] / [BPA - ] values at pH between 9.6-10.3<br />
(Figure 3). The ratio <strong>of</strong> [BPA 2- ] / [BPA - ] was estimated using Henderson-Hasselbach equation as<br />
follows [39]:<br />
2<br />
[BPA ] <br />
pH pKa<br />
2 log [BPA<br />
]<br />
<br />
(3)<br />
<br />
9x10 8<br />
Second-order rate constnats (M -1 s -1 )<br />
8x10 8<br />
7x10 8<br />
6x10 8<br />
5x10 8<br />
4x10 8<br />
R 2 = 0.9973<br />
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1<br />
Estimated [ BPA 2- ]/[ BPA - ]<br />
Figure 3. Effect <strong>of</strong> [BPA 2- ] / [BPA - ] on second-order rate constant <strong>of</strong> BPA-ozone reaction at the<br />
pH 9.6-10.3<br />
From Figure 3, the rate constant increases proportionally with [BPA 2- ] / [BPA - ], which<br />
implies that the dianionic species <strong>of</strong> BPA (BPA 2- ) is more easily oxidised by O 3 compared to its<br />
anionic counterpart (BPA - ). From Figure 2, it can be clearly observed that when the pH value
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
83<br />
increases further from 10.2 to 12.0, the reactivity <strong>of</strong> BPA with O 3 remains almost constant. This is<br />
most likely due to the fact that the maximum fraction <strong>of</strong> BPA 2- has been reached at pH 10.2 and<br />
further increasing <strong>of</strong> pH does not influence the amount <strong>of</strong> BPA 2- anymore.<br />
Influence <strong>of</strong> Inorganic Anions<br />
Most wastewaters normally contain inorganic anions coexisting with organic pollutants. The<br />
effects <strong>of</strong> the anions on the rate <strong>of</strong> BPA degradation was studied at concentrations found in<br />
wastewaters [40-42]. Figure 4 shows plots <strong>of</strong> BPA degradation in the presence <strong>of</strong> phosphate, nitrate,<br />
chloride and sulphate ions. The results indicate that the presence <strong>of</strong> common inorganic anions at<br />
concentration levels in the wastewaters does not significantly affect the rate <strong>of</strong> BPA degradation.<br />
This might be due to the fast reaction between O 3 and BPA as indicated by the determined secondorder<br />
rate constants.<br />
(a)<br />
(b)<br />
1.0<br />
0.8<br />
without phosphate<br />
0.05 ppm phosphate<br />
0.5 ppm phosphate<br />
5.0 ppm phosphate<br />
1.0<br />
0.8<br />
without nitrate<br />
0.1ppm nitrate<br />
1.0 ppm nitrate<br />
10 ppm nitrate<br />
[BPA]/[BPA] 0<br />
0.6<br />
0.4<br />
[BPA]/[BPA] 0<br />
0.6<br />
0.4<br />
0.2<br />
0.2<br />
0.0<br />
0.0<br />
0 2 4 6 8 10<br />
Time (min)<br />
0 2 4 6 8 10<br />
Time (min)<br />
(c)<br />
(d)<br />
[BPA]/[BPA] 0<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
without chloride<br />
0.5 ppm chloride<br />
5 ppm chloride<br />
50 ppm chloride<br />
500 ppm chloride<br />
[BPA]/[BPA] 0<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
without sulfate<br />
0.2 ppm sulfate<br />
2 ppm sulfate<br />
20 ppm sulfate<br />
200 ppm sulfate<br />
0.2<br />
0.2<br />
0.0<br />
0.0<br />
0 2 4 6 8 10<br />
Time (min)<br />
0 2 4 6 8 10<br />
Time (min)<br />
Figure 4. Effect <strong>of</strong> (a) phosphate, (b) nitrate, (c) chloride and (d) sulphate ions on the degradation<br />
<strong>of</strong> BPA (temperature = 25ºC; O 3 dose = 0.69 g/h; pH = 6.5; [BPA] 0 = 100 mg/L)<br />
Degradation By-Products (DBPs) <strong>of</strong> BPA<br />
GCMS analyses were performed by comparing the chromatogram <strong>of</strong> BPA with those <strong>of</strong> the<br />
aliquots taken at different ozonation times. All samples were subjected to similar derivatisation
84 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
procedure as mentioned in the experimental section. A chromatogram showing the distribution <strong>of</strong><br />
DBPs <strong>of</strong> BPA is presented in Figure 5 and the proposed DBPs are presented in Table 1. All<br />
derivatised compounds occurred as trimethylsilyl derivatives characterised by the peak at m/z 73 in<br />
the mass spectrum. Identification <strong>of</strong> DBPs was carried out based on fragmentation patterns in the<br />
mass spectrum and/or by comparing the mass spectrum with the library available in the instrument<br />
database.<br />
BPA<br />
Relative abundance<br />
DBP 7<br />
DBP 6<br />
DBP 2 DBP 4 DBP 5<br />
DBP 1 DBP3<br />
DBP 8<br />
Retention time (min)<br />
Figure 5. Gas chromatogram <strong>of</strong> BPA after 4 min. <strong>of</strong> ozonation time ([BPA] 0 = 100 mg/L, pH = 6.5,<br />
temperature = 25 ºC, O 3 dose = 0.70 g/hr)<br />
The mass spectrum <strong>of</strong> BPA shows the molecular ion peak at m/z 372 (Figure 6a). Besides the<br />
peak at m/z 357 representing the loss <strong>of</strong> methyl group from trimethylsilyl group ((M-CH 3 ) + ), the<br />
other significant peak is at m/z 207 representing the loss <strong>of</strong> trimethyl(phenoxy)silane from silylated<br />
BPA. DBP 1 , DBP 2 and DBP 3 represent the breakdown products <strong>of</strong> BPA and their mass spectra (as<br />
silylated derivatives) are presented in Figures 6 (b-d). The formation <strong>of</strong> DBP 1 , DBP 2 and DBP 3 has<br />
also been detected in previous studies [13,15,16].<br />
The mass spectra <strong>of</strong> DBP 4 and DBP 5 , formed by aromatic-ring opening during the ozonation<br />
process, are presented in Figures 7a and 7b respectively. Silylated DBP 5 with a molecular weight <strong>of</strong><br />
292 amu shows a peak representing (M−1) +· at m/z 291 (Figure 7b). DBP 6 is a BPA degradation byproduct<br />
proposed by Deborde et al. [11]. The structure <strong>of</strong> this compound was derived based on its<br />
fragmentation pattern in the mass spectrum (Figure 7c). The peak at m/z 378 corresponds to the<br />
molecular ion peak <strong>of</strong> silylated DBP 6 . Other major peaks are at m/z 212 and 217. The peak at m/z<br />
212 represents the radical cation <strong>of</strong> silylated 3-formyl-4-methylpenta-2,4-dienoic acid. The peak at<br />
m/z 217 represents the loss <strong>of</strong> silylated carboxylic acid and aldehyde functional groups from the<br />
parent compound.<br />
Monohydroxylated (DBP 7 ) and dihydroxylated (DBP 8 ) BPA were also detected during the<br />
ozonation process. The molecular ion peaks <strong>of</strong> DBP 7 and DBP 8 are at m/z 460 (Figure 7d) and m/z<br />
548 (Figure 7e) respectively. Additional 88 amu and 176 amu over the molecular ion peak <strong>of</strong>
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
85<br />
silylated BPA (m/z 372) indicated the addition <strong>of</strong> one and two [(CH 3 ) 3 SiO] + groups to silylated BPA<br />
respectively.<br />
Table 1. Degradation by-products <strong>of</strong> BPA<br />
Compound identified<br />
Retention<br />
time (min.)<br />
(Label)<br />
Proposed structure <strong>of</strong><br />
compound<br />
(Molecular weight)<br />
Name<br />
(m/z 372)<br />
25.01<br />
BPA<br />
(228.3)<br />
Bisphenol A<br />
(m/z 206)<br />
15.71<br />
(DBP 1 )<br />
(134.1)<br />
4-(Prop-1-en-2-yl)phenol<br />
(m/z 254)<br />
16.23<br />
(DBP 2 )<br />
(110.0)<br />
Hydroquinone<br />
(m/z 208)<br />
17.15<br />
(DBP 3 )<br />
(136.1)<br />
4-Hydroxyacetophenone<br />
(m/z 292)<br />
19.23<br />
(DBP 4 )<br />
HO O O<br />
(220.1)<br />
2-(2-(4-<br />
Hydroxyphenyl)propan-<br />
2-yl)succinaldehyde<br />
(m/z 274)<br />
20.24<br />
(DBP 5 )<br />
(202.1)<br />
2-(1-(4-<br />
Hydroxyphenyl)vinyl)-<br />
pent-2-enal<br />
TMSO<br />
O O OTMS<br />
(m/z 378)<br />
25.09<br />
(DBP 6 )<br />
(234.1)<br />
3-Formyl-4-(4-<br />
hydroxyphenyl)-4-<br />
methylpent-2-enoic acid<br />
(m/z 460)<br />
26.20<br />
(DBP 7 )<br />
(244.1)<br />
Monohydroxy-BPA<br />
(m/z 548)<br />
27.01<br />
(DBP 8 )<br />
(260.1)<br />
Dihydroxy-BPA
86 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
(a)<br />
(b)<br />
m/z<br />
m/z<br />
(c)<br />
(d)<br />
m/z<br />
m/z<br />
Figure 6. Mass spectra <strong>of</strong> DBP 1 , DBP 2 , DBP 3 and BPA<br />
Figure 8 shows the time pr<strong>of</strong>iles <strong>of</strong> the major DBPs, most <strong>of</strong> which were successfully<br />
removed after 10 min. <strong>of</strong> ozonation, with the exception <strong>of</strong> DBP 3 , DBP 6 and DBP 8 . These three byproducts<br />
thus seemed to be more resistant to ozonation compared to others. In the degradation<br />
experiment, ozonation was performed without a radical scavenger, so BPA could react with both O 3<br />
and •OH. As compared to the reaction between O 3 and BPA [11], the results in this study show that<br />
the presence <strong>of</strong> •OH seemed to produce more species <strong>of</strong> the breakdown products such as DBP 1 ,<br />
DBP 3 and DBP 4 , which was most likely due to the non-selective behaviour <strong>of</strong> •OH, which can also<br />
react at the aliphatic chain and aromatic ring <strong>of</strong> BPA [33, 43].
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
87<br />
Mass spectrum<br />
Fragmentation pathway in mass<br />
spectrum<br />
(a) DBP 4<br />
m/z<br />
(b) DBP 5<br />
m/z<br />
(c) DBP 6<br />
m/z<br />
Figure 7. Mass spectra and fragmentation patterns <strong>of</strong> DBP 4 , DBP 5 , DBP 6 , DBP 7 and DBP 8
88 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
(d) DBP 7<br />
(e) DBP 8<br />
m/z<br />
m/z<br />
Figure 7. (continued)<br />
BPA<br />
DBP 1<br />
1.0<br />
DBP 2<br />
DBP 3<br />
Relative abundance<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
DBP 4<br />
DBP 5<br />
DBP 6<br />
DBP 7<br />
DBP 8<br />
0.0<br />
0 2 4 6 8 10<br />
Ozonation time (min)<br />
Figure 8. Time pr<strong>of</strong>iles <strong>of</strong> major BPA degradation by-products ([BPA] 0 = 100 mg/L, pH = 6.5,<br />
temperature = 25ºC, and O 3 dose = 0.69 g/h). Relative abundance <strong>of</strong> the degradation byproducts<br />
was calculated by normalising the peak areas at the defined ozonation time to the<br />
highest peak areas.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
89<br />
Formation <strong>of</strong> Degradation By-Products<br />
Formation <strong>of</strong> DBP 1 , DBP 2 , DBP 3 , DBP 7 and DBP 8<br />
Formation <strong>of</strong> DBP 7 and DBP 8 can occur through hydroxylation <strong>of</strong> BPA or via a direct<br />
reaction between BPA and O 3 (Figure 9). Hydroxylated BPA is proposed to be an important<br />
intermediate <strong>of</strong> by-products resulting from ring opening, especially DBP 4 , DBP 5 and DBP 6 .<br />
Formation <strong>of</strong> polyhydroxylated BPA through the reaction between BPA and •OH has also been<br />
reported during the degradation <strong>of</strong> BPA by ultrasound-UV-iron (II) treatment [17].<br />
The formation <strong>of</strong> DBP 1 , DBP 2 and DBP 3 are presented in Figure 9b. Since O 3 with its<br />
eletrophilic nature reacts selectively with an electron-rich reaction site, it is proposed that the<br />
reaction begins at the side chain <strong>of</strong> BPA with an initial attack <strong>of</strong> •OH via hydrogen abstraction and<br />
leads to the formation <strong>of</strong> radical A. Intramolecular rearrangement <strong>of</strong> radical A then leads to the<br />
cleavage <strong>of</strong> C-C bond to form DBP 1 (4-isopropenylphenol) and a phenol radical, B. B will then react<br />
with •OH to form DBP 2 (hydroquinone). DBP 1 can further react with either O 3 or •OH, leading to<br />
the formation <strong>of</strong> dihydroxylated DBP 1 (C). C would then react with •OH via hydrogen abstraction to<br />
form radical D, which undergoes a fragmentation to form DBP 3 (4-hydroxyacetophenone).<br />
(a)<br />
(b)<br />
Figure 9. Proposed pathways for the formation <strong>of</strong>: (a) DBP 1 , DBP 2 and DBP 3 ; (b) DBP 7 and DBP 8<br />
from BPA during ozonation
90 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
Formation <strong>of</strong> DBP 4 , DBP 5 and DBP 6<br />
The formation pathway <strong>of</strong> DBP 4 is presented in Figure 10. The formation <strong>of</strong> DBP 6 is<br />
proposed to begin with the initial attack <strong>of</strong> •OH on the monohydroxylated BPA giving a radical<br />
intermediate (F) (Figure 11a). Intra-molecular rearrangement <strong>of</strong> F and cleavage <strong>of</strong> C-C bond lead to<br />
the opening <strong>of</strong> an aromatic ring, resulting in the formation <strong>of</strong> radical G, which then rearranges to<br />
form radical H, which reacts with water to form I. Attack <strong>of</strong> •OH at the enol site <strong>of</strong> I affords radical<br />
J, which gives radical K on cleavage <strong>of</strong> a C-C bond. Radical K further reacts with •OH to form L,<br />
which upon hydrogen abstraction gives radical M. M then further reacts with •OH leading to the<br />
formation <strong>of</strong> DBP 6 . The formation pathway <strong>of</strong> DBP 5 is presented in Figure 11b.<br />
-<br />
H<br />
HO<br />
BPA<br />
OH<br />
OH<br />
HO<br />
H<br />
OH<br />
OH<br />
HO<br />
OH<br />
OH<br />
OH<br />
+ OH<br />
O<br />
OH<br />
OH<br />
OH<br />
HO<br />
HO<br />
H<br />
OH<br />
HO<br />
H<br />
OH<br />
O<br />
HO<br />
OH<br />
O<br />
H<br />
+H<br />
O<br />
O<br />
O<br />
OH<br />
-<br />
CH 2 OH<br />
O<br />
H<br />
- H<br />
O<br />
HO<br />
OH<br />
HO<br />
OH<br />
OH<br />
HO<br />
OH<br />
OH<br />
- CO 2<br />
O<br />
OH<br />
CH<br />
OH<br />
HO<br />
O<br />
HO<br />
OH<br />
HO<br />
OH<br />
DBP 4<br />
Figure 10. Proposed pathway for the formation <strong>of</strong> DBP 4
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
91<br />
(a)<br />
(b)<br />
O<br />
OH<br />
O<br />
OH<br />
O<br />
OH<br />
HO<br />
G<br />
O H<br />
HO<br />
O<br />
OH<br />
HO<br />
O<br />
OH<br />
O<br />
O<br />
O<br />
H<br />
O<br />
OH<br />
- HOO<br />
O<br />
OH<br />
HO<br />
- H<br />
HO<br />
+ H<br />
HO<br />
CH 2<br />
H OH<br />
H<br />
OH<br />
OH<br />
CH 2<br />
CH<br />
- CO 2<br />
- CH 3<br />
OH<br />
O<br />
HO<br />
HO<br />
HO<br />
- H<br />
CH 2<br />
O<br />
O<br />
HO<br />
DBP 5<br />
HO<br />
Figure 11. Proposed pathways for the formation <strong>of</strong>: (a) DBP 5 and (b) DBP 6<br />
CONCLUSIONS<br />
The rate constant for the reaction between BPA and ozone in aqueous solution in the<br />
presence <strong>of</strong> t-BuOH increased with increase in pH between pH 2-10, beyond which it was fairly<br />
constant between pH 10-12. The reactivity <strong>of</strong> BPA increased in the order: undissociated form <<br />
anionic form < dianionic form. The presence <strong>of</strong> common inorganic anions (chloride, sulphate, nitrate
92 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
and phosphate ions) at environmental levels did not significantly affect the rate <strong>of</strong> degradation <strong>of</strong><br />
BPA by ozone.<br />
The degradation products <strong>of</strong> BPA during ozonation were identified to be 4-(prop-1-en-2-<br />
yl)phenol, hydroquinone, 4-hydroxyacetophenone, 2-(2-(4-hydroxyphenyl)propan-2-yl)succinaldehyde,<br />
2-(1-(4-hydroxyphenyl)vinyl)pent-2-enal, 3-formyl-4-(4-hydroxyphenyl)-4-methylpent-2-<br />
enoic acid, monohydroxy-BPA and dihydroxy-BPA.<br />
ACKNOWLEDGEMENTS<br />
This research was financially supported by Malaysia Toray <strong>Science</strong> Foundation (MTSF),<br />
University <strong>of</strong> Malaya (Grant No. P0266-2007A, SF025-2007A, PS153-2007B and FS305-2007C)<br />
and Ministry <strong>of</strong> Higher Education Malaysia (FRGS).<br />
REFERENCES<br />
1. Y. M. Lee, M. J. Seong, J. W. Lee, J. K. Lee, T. M. Kim, S. Y. Nam, D. J. Kim, Y. W. Yun, T.<br />
S. Kim, S.Y. Han and J. T. Hong, “Estrogen receptor independent neurotoxic mechanism <strong>of</strong><br />
bisphenol A, an environmental estrogen”, J. Vet. Sci., 2007, 8, 27-38.<br />
2. J. A. Brotons, M. F. Olea-Serrano, M. Villalobos, V. Pedraza and N. Olea, “Xenoestrogens<br />
released from lacquer coatings in food cans”, Environ. Health Perspect., 1995, 103, 608-612.<br />
3. C. Dash, M. Marcus and P. D. Terry, “Bisphenol A: Do recent studies <strong>of</strong> health effects among<br />
humans inform the long-standing debate”, Mutat. Res., 2006, 613, 68-75.<br />
4. D. de A. Azevedo, S. Lacorte, P. Viana and D. Barcelό, “Occurrence <strong>of</strong> nonylphenol and<br />
bisphenol-A in surface waters from Portugal”, J. Braz. Chem. Soc., 2001, 12, 532-537.<br />
5. E. Z. Harrisson, S. R. Oakes, M. Hysell and A. Hay, “Organic chemicals in sewage sludges”,<br />
Sci. Total Environ., 2006, 367, 481-497.<br />
6. S. Zhang, Q. Zhang, S. Darisaw, O. Ehie and G. Wang, “Simultaneous quantification <strong>of</strong><br />
polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), and<br />
pharmaceuticals and personal care products (PPCPs) in Mississippi river water, in New Orleans,<br />
Louisiana, USA”, Chemosphere, 2007, 66, 1057-1069.<br />
7. C. Basheer, H. K. Lee and K. S. Tan, “Endocrine disrupting alkylphenols and bisphenol-A in<br />
coastal waters and supermaket seafood from Singapore”, Mar. Pollut. Bull., 2004, 48, 1161-<br />
1167.<br />
8. T. Isobe, H. Takada, M. Kanai, S. Tsutsumi, K. O. Isobe, R. Boonyatumanond and M. P.<br />
Zakaria, “Distribution <strong>of</strong> polycyclic aromatic hydrocarbons (PAHs) and phenolic endocrine<br />
disrupting chemicals in South and Southeast Asian mussels”, Environ. Monit. Assess., 2007,<br />
135, 423-440.<br />
9. J. Lee, H. Park and J. Yoon, “Ozonation characteristics <strong>of</strong> bisphenol A in water”, Environ.<br />
Technol., 2003, 24, 241-248.<br />
10. B. Xu, N. Gao, M. Rui, H. Wang and H. Wu, “Degradation <strong>of</strong> endocrine disruptor bisphenol A<br />
in drinking water by ozone oxidation”, Front. Environ. Sci. Eng. China, 2007, 1, 350-356.<br />
11. M. Deborde, S. Rabouan, P. Mazellier, J. P. Duguet and B. Legube, “Oxidation <strong>of</strong> bisphenol A<br />
by ozone in aqueous solution”, Water Res., 2008, 42, 4299-4308.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
93<br />
12. F. J. Rivas, Á. Encinas, B. Acedo and F. J. Beltrán, “Mineralization <strong>of</strong> bisphenol A by advanced<br />
oxidation processes”, J. Chem. Technol. Biotechnol., 2009, 84, 589-594.<br />
13. I. Gultekin, M. Valko and H. I. Nilsun, “Degradation <strong>of</strong> bisphenol-A by ozonation”, J. Adv.<br />
Oxidat. Technol., 2009, 12, 242-248.<br />
14. T. Garoma and S. Matsumoto, “Ozonation <strong>of</strong> aqueous solution containing bisphenol A: Effect<br />
<strong>of</strong> operational parameters”, J. Hazard Mater., 2009, 167, 1185-1191.<br />
15. H. Katsumata, S. Kawabe, S. Kaneco, T. Suzuki and K. Ohta, “Degradation <strong>of</strong> bisphenol A in<br />
water by the photo-Fenton reaction”, J. Photochem. Photobiol. A, 2004, 162, 297-305.<br />
16. K. Nomiyama, T. Tanizaki, T. Koga, K. Arizono and R. Shinohara, “Oxidative degradation <strong>of</strong><br />
BPA using TiO 2 in water, and transition <strong>of</strong> estrogenic activity in the degradation pathways”,<br />
Arch. Environ. Contam. Toxicol., 2007, 52, 8-15.<br />
17. R. A. Torres, C. Pétrier, E. Combet, F. Moulet and C. Pulgarin, “Bisphenol A mineralization by<br />
integrated ultrasound-UV-iron (II) treatment”, Environ. Sci. Technol., 2007, 41, 297-302.<br />
18. J. Sowmya, “European emerging trends and technologies include UV, Ozone”, Water and<br />
Waste Water <strong>International</strong> Magazine (April, 2008), PennWell Publishing Corporation.<br />
(http://www.waterworld.com/index/display/article-display/329727/articles/water-wastewaterinternational/volume-23/issue-2/regional-focus/middle-east-north-africa/european-emergingtrends-amp-technologies-include-uv-ozone.html)<br />
(Accessed on 28 th December 2011).<br />
19. A. Conner, “Reducing cooling tower costs with ozone technology”, CTI Bibliography <strong>of</strong><br />
Technical Papers __ Ozone, Cooling Technology Institute, 2005, http://www.cti.org/tech_papers/<br />
ozone.php (Accessed on 28 th December 2011).<br />
20. M. M. Huber, S. Canonica, G. Y. Park and U. von Gunten, “Oxidation <strong>of</strong> pharmaceuticals<br />
during ozonation and advanced oxidation processes”, Environ. Sci. Technol., 2003, 37, 1016-<br />
1024.<br />
21. F. J. Beltrán, “Ozone Reaction Kinetics for Water and Wastewater Systems”, Lewis Publishers,<br />
Boca Raton, 2004, pp.1-4.<br />
22. K. Ikehata, N. J. Naghashkar and M. G. El-Din, “Degradation <strong>of</strong> aqueous pharmaceuticals by<br />
ozonation and advanced oxidation process: A review”, Ozone Sci. Eng., 2006, 28, 353-414.<br />
23. S. A. Snyder, E. C. Wert, D. J. Rexing, R. E. Zegers and D. D. Drury, “Ozone oxidation <strong>of</strong><br />
endocrine disruptors and pharmaceuticals in surface water and wastewater”, Ozone Sci. Eng.,<br />
2006, 28, 445-460.<br />
24. S. Esplugas, D. M. Bila, L. G. Krause and M. Dezotti, “Ozonation and advanced oxidation<br />
technologies to remove endocrine disrupting chemicals (EDCs) and pharmaceuticals and<br />
personal care products (PPCPs) in water effluents”, J. Hazard Mater., 2007, 149, 631-642.<br />
25. M. D. H. Guil, M. Petrović, J. Radjenovic, A. R. Fernández-Alba, A. R. Fernández-Alba and D.<br />
Barceló, “Removal <strong>of</strong> pharmaceuticals by advanced treatment technologies”, in “Comprehensive<br />
Analytical Chemistry, Volume 50: Analysis, Fate and Removal <strong>of</strong> Pharmaceuticals in the Water<br />
Cycle”, (Ed. M. Petrović and D. Barceló), Elsevier, Amsterdam, 2007, pp.451-474.<br />
26. J. Staehelin and J. Hoigné, “Decomposition <strong>of</strong> ozone in water in the presence <strong>of</strong> organic solutes<br />
acting as promoters and inhibitors <strong>of</strong> radical chain reactions”, Environ. Sci. Technol., 1985, 19,<br />
1206-1213.<br />
27. M. S. Siddiqui, “Chlorine-ozone interactions: Formation <strong>of</strong> chlorate”, Water Res., 1996, 30,<br />
2160-2170.
94 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 77-94<br />
28. M. A. Boncz, H. Bruning, W. H. Rulkens, H. Zuilh<strong>of</strong> and E. J. R. Sudhölter, “The effect <strong>of</strong> salts<br />
on ozone oxidation processes”, Ozone Sci. Eng., 2005, 27, 287-292.<br />
29. K. S. Tay, N. A. Rahman and M. R. B. Abas, “Ozonation <strong>of</strong> parabens in aqueous solution:<br />
Kinetics and mechanism <strong>of</strong> degradation”, Chemosphere, 2010, 81, 1446-1453.<br />
30. H. Bader and J. Hoigné, “Determination <strong>of</strong> ozone in water by the indigo method”, Water Res.,<br />
1981, 15, 449-456.<br />
31. K. S. Tay, N. A. Rahman and M. R. B. Abas, “Degradation <strong>of</strong> DEET by ozonation in aqueous<br />
solution”, Chemosphere, 2009, 76, 1296-1302.<br />
32. J. Hoigné and H. Bader, “Rate constants <strong>of</strong> reactions <strong>of</strong> ozone with organic and inorganic<br />
compounds in water – II: Dissociating organic compounds”, Water Res., 17, 185-194.<br />
33. K. S. Tay, N. A. Rahman and M. R. B. Abas, “Removal <strong>of</strong> selected endocrine disrupting<br />
chemicals and personal care products in surface waters and secondary wastewater by ozonation”,<br />
Water Environ. Res., 2011, 83, 684-691.<br />
34. M. Deborde, S. Rabouan, J. P. Duguet and B. Legube, “Kinetics <strong>of</strong> aqueous ozone-induced<br />
oxidation <strong>of</strong> some endocrine disruptor”, Environ. Sci. Technol., 2005, 39, 6086-6092.<br />
35. J. Benner, E. Salhi, T. Ternes and U. von Gunten, “Ozonation <strong>of</strong> reverse osmosis concentrate:<br />
Kinetics and efficiency <strong>of</strong> beta blocker oxidation”, Water Res., 2008, 42, 3003-3012.<br />
36. F. J. Benitez, J. L. Acero, F. J. Real and G. Roldán, “Ozonation <strong>of</strong> pharmaceutical compounds:<br />
Rate constants and elimination in various water matrices”, Chemosphere, 2009, 77, 53-59.<br />
37. C. A. Staples, P. B. Dorn, G. M. Klecka, S. T. Oblock and L. R. Harris, “A review <strong>of</strong> the<br />
environmental fate, effects, and exposures <strong>of</strong> bisphenol A”, Chemosphere, 1998, 36, 2149-2173.<br />
38. C. Tian, J. T. Wang and X. L. Song, “Sediment-water interactions <strong>of</strong> bisphenol A under<br />
simulated marine conditions”, Water Air Soil Pollut., 2009, 199, 301-310.<br />
39. J. C. Kotz, P. M. Treichel and J. R. Townsend "Chemistry and Chemical Reactivity", 7th Edn.,<br />
Brooks/Cole, Belmont, 2010, pp.810-859.<br />
40. S. V. Karmarkar, “Analysis <strong>of</strong> wastewater for anionic and cationic nutrients by ion<br />
chromatography in a single run with sequential flow injection analysis”, J. Chromatogr. A, 1999,<br />
850, 303-309.<br />
41. I. R. Santos, R. C. Costa, U. Freitas and G. Fillmann, “Influence <strong>of</strong> effluents from a wastewater<br />
treatment plant on nutrient distribution in a coastal creek from southern Brazil”, Braz. Arch.<br />
Biol. Technol., 2008, 51, 153-162.<br />
42. O. A. Al-Khashman, “Chemical evaluation <strong>of</strong> Ma’an sewage effluents and its reuse in irrigation<br />
purpose”, Water Resour. Manage., 2009, 23, 1041-1053.<br />
43. K. S. Tay, N. A. Rahman and M. R. B. Abas, “Characterization <strong>of</strong> atenolol transformation<br />
products in ozonation by using rapid resolution high-performance liquid chromatography/<br />
quadrupole-time-<strong>of</strong>-flight mass spectrometry”, Microchem. J., 2011, 99, 312-326.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
Full Paper<br />
<strong>Maejo</strong> <strong>Maejo</strong> Int. J. Int. Sci. J. Technol. Sci. Technol. <strong>2012</strong>, 6(01), <strong>2012</strong>, 6(01), 95-104 95-10495<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Chitinase production and antifungal potential <strong>of</strong> endophytic<br />
Streptomyces strain P4<br />
Julaluk Tang-um and Hataichanoke Niamsup*<br />
Department <strong>of</strong> Chemistry and Centre <strong>of</strong> Excellence for Innovation in Chemistry, Faculty <strong>of</strong> <strong>Science</strong>,<br />
Chiang Mai University, Chiang Mai 50200, Thailand<br />
* Corresponding author, e-mail: hataichanoke.n@cmu.ac.th<br />
Received: 2 August 2011 / Accepted: 11 February <strong>2012</strong> / Published: 29 February <strong>2012</strong><br />
Abstract: The endophytic actinomycete P4 strain, previously isolated from sweet pea root, was<br />
identified as Streptomyces sp. by full 16S rRNA sequencing. It is mostly related to Streptomyces<br />
grise<strong>of</strong>lavus with a 99.7% identity score. The Streptomyces sp. P4 was tested for its hydrolytic<br />
activities by plate method. The result showed the presence <strong>of</strong> chitinase. The extent <strong>of</strong> chitinase activity<br />
was assessed by spectrophotometric method along with growth monitoring. Chitinase production was<br />
growth-associated and showed the highest activity on the fifth day. The dual culture method revealed<br />
that the strain was effective in restricting the radial growth <strong>of</strong> Fusarium oxysporum f.sp. lycopersici, an<br />
important phytopathogen <strong>of</strong> tomato. Scanning electronic microscopic analysis showed that the rupture<br />
<strong>of</strong> the F. oxysporum mycelial cell wall occurred at the area <strong>of</strong> interaction between F. oxysporum and<br />
Streptomyces sp. P4. This was possibly due to the chitinolytic activity <strong>of</strong> the P4. Thus, this<br />
actinomycete has the potential for being used as a biocontrol agent, thereby reducing the use <strong>of</strong><br />
chemical fungicides.<br />
Keywords: endophyte, streptomycete, chitinase, fusarium, wilt, antimicrobial activity<br />
__________________________________________________________________________________<br />
INTRODUCTION<br />
As worldwide concern for the natural environment and human health has increased, so has<br />
interest in organic farming. A research institute <strong>of</strong> organic agriculture (FiBL) and the international<br />
federation <strong>of</strong> organic agriculture movements (IFOAM) reported that organic agricultural lands have<br />
expanded globally from 11.0 million hectares in 1999 to 37.2 million hectares in 2009, accounting for<br />
0.85% <strong>of</strong> the total agricultural lands [1]. One <strong>of</strong> the criteria for organic farming is the avoidance <strong>of</strong><br />
chemical usage. Soilborne plant diseases can be controlled through agronomic practices and microbial
96 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 95-104<br />
biocontrol agents instead [2]. Biological controlling agents can replace chemical agents in controlling<br />
pathogenic insects, microbials and weeds. Several bi<strong>of</strong>ungicides are based on antibiotic metabolites and<br />
hydrolytic enzymes. For example, Streptomyces griseoviridis strain K61, a soilborne fungal antagonist<br />
which produces aromatic antibiotics with characteristic 7-membered rings in the molecules was<br />
commercialised as Mycostop by Verdera Oy, a Finnish company [3], and Streptomyces sp. Di-944 was<br />
formulated to suppress Rhizoctonia damping-<strong>of</strong>f [4].<br />
Streptomyces is a major genus <strong>of</strong> actinimycetes, the Gram-positive terrestrial or marine bacteria<br />
found in both colony and mycelium forms. Although Streptomyces species with a characteristic earthy<br />
smell may be thought <strong>of</strong> as pathogens, the antibiotics that they produce have been pr<strong>of</strong>itably exploited<br />
[5]. For example, S. clavuligerus produces the -lactam cephamycin C and clavulanic acid, a -<br />
lactamase inhibitor. A new stereoisomeric anthracyclin with anticancer activity was isolated from<br />
Sreptomyces sp. Eg23 [6]. Various hydrolytic enzymes, e.g. proteases/peptidases, chitinases/chitosanases,<br />
cellulases/endoglucanases, amylases, and pectate lyases, are produced by S. coelicolor [7].<br />
In this study, the hydrolytic enzyme production <strong>of</strong> the endophytic actinomycete P4 strain,<br />
previously isolated from sweet pea root [8], is investigated. Additionally, the potential use <strong>of</strong> this<br />
endophyte as a biocontrol agent is studied by assessing its antagonism against pathogenic fungi.<br />
MATERIALS AND METHODS<br />
Microorganism Strains and Growth Conditions<br />
The bacterial strain P4, generously provided by Asst. Pr<strong>of</strong>. Dr. Ampan Bhromsiri (Department<br />
<strong>of</strong> Soil <strong>Science</strong> and Conservation, Chiang Mai University) was previously isolated from sweet pea root.<br />
The bacteria was maintained at 30C on an IMA-2 agar medium consisting <strong>of</strong> 5 g glucose, 5 g soluble<br />
starch, 1 g beef extract, 1 g yeast extract, 2 g N-Z-case ® , 2 g NaCl and 1 g CaCO 3 per litre [9]. For the<br />
production <strong>of</strong> chitinase, the culture was transferred into a colloidal chitin medium, 1 litre <strong>of</strong> which<br />
consisted <strong>of</strong> 20 g colloidal chitin, 0.5 g yeast extract, 1 g (NH 4 ) 2 SO 4 , 0.3 g MgSO 4 . 7H 2 O and 1.36 g<br />
KH 2 PO 4 , with an adjusted pH <strong>of</strong> 7.0 [10]. Colloidal chitin was prepared according to the method<br />
described by Souza et al. [11]. The liquid culture was incubated at 30C with agitation at 160 revs/min.<br />
The pathogenic fungi Fusarium oxysporum f.sp. lycopersici, Corynespora cassiicola and<br />
Rhizoctonia solani were obtained from the Department <strong>of</strong> Agriculture, the Ministry <strong>of</strong> Agriculture and<br />
Cooperatives (Thailand). They were maintained on potato-dextrose agar (PDA).<br />
Identification by 16s rRNA Sequencing<br />
An approximately 1.5-kb polymerase chain reaction (PCR) product was amplified from genomic<br />
DNA <strong>of</strong> the bacterial strain P4 using two primers, 20F (5’-GAG TTT GAT CCT GGC TCA G-3’) and<br />
1500R (5’-GTT ACC TTG TTA CGA CTT-3’), directed to the 16S rRNA region. The following<br />
conditions were used: an initial denaturation step at 94C for 3 min., 25 cycles at 94C (1 min.), 50C<br />
(1 min.) and 72C (2 min.), followed by a final extension at 72C (3 min.). The purified PCR product<br />
was sequenced by an ABI PRISM® BigDye Terminator Ready Reaction Cycle Sequencing Kit<br />
(Applied Biosystems, USA) on an ABI Prism® 3730XL DNA Sequence (Applied Biosystems, USA).<br />
Homology was analysed using the BLAST program from the GenBank database [12].
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 95-104<br />
97<br />
Plate Screening <strong>of</strong> Hydrolytic Enzymes<br />
The bacterial strain P4 (identified as Streptomyces) was screened for its capacity to produce<br />
hydrolytic enzymes using the plate method. The bacterium was allowed to grow at 30C on nutrient<br />
agar plates supplemented with different substrates, i.e. colloidal chitin, gelatin, sodium carboxymethylcellulose<br />
and Tween20, for detection <strong>of</strong> chitinase, protease, cellulase and lipase respectively. The clear<br />
zone around the colonies observed after 7-14 days is an indication for enzyme production. In the cases<br />
<strong>of</strong> cellulase and protease, the plates were reacted with 0.2% Congo red and saturated (NH 4 ) 2 SO 4<br />
respectively prior to the observation <strong>of</strong> growth [13-14].<br />
Quantification <strong>of</strong> Extracellular Chitinase Activity<br />
The strain P4 was grown in a colloidal chitin medium broth at 30ºC with continuous shaking at<br />
160 rpm. The supernatant fluid was harvested every two days for 17 days by filtration through<br />
Whatman no.1 filter. Chitinase activity in the supernatant was assayed using 0.6% colloidal chitin as a<br />
substrate and was based on a procedure by Taechowisan et al. [13]. The supernatant fluid (700 μl) was<br />
added with 2% colloidal chitin (300 μl) in 0.1M potassium phosphate buffer at pH 7.0, and the mixture<br />
was incubated in a water bath at 40ºC for 3 hr. One mL <strong>of</strong> Somogyi’s reagent [15] was added and the<br />
reaction mixture was boiled at 100ºC for 10 min. and cooled to room temperature. Then Nelson’s<br />
reagent (1 mL) was added and the mixture cooled to room temperature for 20 min. After centrifugation<br />
<strong>of</strong> the reaction mixture, the amount <strong>of</strong> N-acetyl glucosamine (GlcNAc) released in the supernatant was<br />
spectrophotometrically measured by the method <strong>of</strong> Somogyi-Nelson [15]. The method is based on the<br />
520-nm absorbance given by a coloured complex formed between a copper-oxidised sugar and<br />
arsenomolybdate. One unit (U) <strong>of</strong> chitinase activity was defined as the amount <strong>of</strong> enzyme required to<br />
produce 1 mol <strong>of</strong> reducing sugar per min. under the conditions <strong>of</strong> the experiment.<br />
The cell growth was followed by measurement <strong>of</strong> the cells’ dry weight. After removing the<br />
supernatant for chitinase assay, the cell pellet was dried in an oven at 80C until a constant weight was<br />
obtained. All measurements were performed in triplicate.<br />
In Vitro Antagonism Tests against Fungi<br />
The identified Streptomyces P4 was evaluated for antagonistic activity towards three fungal<br />
phytopathogens, i.e. F. oxysporum, C. cassiicola and R. solani, by the dual culture method. An agar<br />
plug <strong>of</strong> 6 mm in diameter taken from a 7-day-old colony <strong>of</strong> each test fungus and that taken from a 14-<br />
day-old P4 bacterial strain were placed on PDA plates with 4-cm spacing (3 replicates each). The<br />
cultures were incubated at room temperature (30-32C) and the diameters <strong>of</strong> the fungal colonies in the<br />
direction <strong>of</strong> the actinomycete were measured every 2 days for 14 days. The test fungi were grown<br />
alone to serve as control. Data were statistically analysed for significance (p
98 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 95-104<br />
osmium tetroxide. After dehydration in ethanol, the samples were dried to their critical points with<br />
carbon dioxide [16], mounted on slides and coated with gold for observation by SEM.<br />
RESULTS AND DISCUSSION<br />
Identification <strong>of</strong> Bacterial Strain<br />
A sequence <strong>of</strong> 1438-bp in length was obtained from the 1.5-kb PCR product. The DNA<br />
sequence was then submitted to GenBank database under accession no. JN102356 and was blasted<br />
against non-redundant DNA sequences in the database [12]. The 16S ribosomal DNA sequence showed<br />
the highest similarity to Streptomyces grise<strong>of</strong>lavus gene (accession no. EU741217) with a 99.7%<br />
identity score, followed by 99.5% score when aligned with S. variabilis (DQ442551, AB184884,<br />
AB184763), S. vinaceus (AB184186) and S. griseoincarnatus (AB184207, AJ781321). The P4 strain<br />
could not be definitely identified down to species level. It is worth noting that the 16S ribosomal RNA<br />
sequences are not highly variable among Streptomyces species and that Streptomyces systematics are<br />
rather complex. Lanoot et al. [17] suggested that 16S-ITS RFLP fingerprinting had a higher taxonomic<br />
resolution than 16S rDNA sequencing. A transformation reaction <strong>of</strong> progesterone has also been<br />
proposed for Streptomyces taxonomic classification [18].<br />
S. grise<strong>of</strong>lavus was reported to produce many potent secondary metabolites, e.g. hormaomycin,<br />
a peptide lactone and a bacterial signalling metabolite and narrow-spectrum antibiotic [19];<br />
okilactomycin, a polyketide antibiotic against gram-positive bacteria [20]; bicozamycin, a cyclic peptide<br />
antibiotic [21]; desferrioxamine, a precursor <strong>of</strong> iron chelator [22]; and an alkaline protease inhibitor<br />
[23]. However, there has been no report on the hydrolytic enzyme production in S. grise<strong>of</strong>lavus.<br />
Chitinase Activity<br />
From a preliminary screening <strong>of</strong> enzymes by the plant method (Figure 1), a clear zone<br />
surrounding the bacterial colonies was observed in the plate containing colloidal chitin as shown in<br />
Figure 1(A), indicating that the Streptomyces sp. P4 produced chitinase, whose activity was assayed<br />
during cell growth. Figure 2 shows that the activity was at a maximum on the fifth day, followed by a<br />
decrease upon approaching a stationary phase <strong>of</strong> growth. The findings imply that chitinase production<br />
is growth-associated. It should be noted that the Streptomyces sp. P4 strain had been pre-cultured in<br />
colloidal chitin medium prior to this experiment. As a consequence, the lag phase <strong>of</strong> growth was not<br />
observed, while logarithmic phase continued until the 5 th day before reaching a stationary phase.<br />
This result <strong>of</strong> chitinase production <strong>of</strong> Streptomyces sp. P4, induced in a colloidal chitincontaining<br />
environment as previously reported [10, 24] and indicating growth-associated behaviour, is<br />
similar to that obtained from the study on S. hygroscopicus [25]. However chitinase production in S.<br />
hygroscopicus occurred 1-2 days before cell growth, while in our case chitinase production <strong>of</strong><br />
Streptomyces sp. P4 was closely associated with cell growth, which should be because the pre-cultured<br />
and pre-induced bacteria were used in our experiment. Closely paralleling growth, the chitinolytic<br />
enzyme production by Streptomyces may be for the purpose <strong>of</strong> hydrolysing chitin into monosaccharides<br />
to be used as carbon and nitrogen sources [26]. However, the chitinase activity <strong>of</strong> 0.00093 U/mL in<br />
Streptomyces sp. P4, which corresponds to a specific activity <strong>of</strong> 0.050 U/mg protein, was 4 times lower<br />
than that in S. viridificans (0.0038 U/mL) [10]. In nature chitinase is produced by actinomycetes in<br />
order to degrade complex nutrients from the soil. As fungal cell walls and insect structures largely<br />
contain chitin, chitinase produced from endophytes can be deleterious to pathogens and pests [27-28].
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 95-104<br />
99<br />
A<br />
B<br />
C<br />
D<br />
Figure 1. Plate screening tests for hydrolytic enzyme production <strong>of</strong> Streptomyces sp. P4. Agar plates<br />
contain the corresponding substrates for chitinase (A), protease (B), cellulase (C) and lipase (D). Each<br />
plate represents a duplicate experiment.<br />
Antifungal Activity<br />
The results in Figures 3-4 show that Streptomyces sp. P4 could effectively suppress the growth<br />
<strong>of</strong> Fusarium oxysporum f.sp. lycopersici, a fungus causing Fusarium wilt, a severe disease in tomato<br />
[29]. Maximal inhibition was observed on the 9th day with 12.50% inhibition (Figure 3). The radial<br />
growth diameter <strong>of</strong> F. oxysporum when grown on the same plate as Streptomyces sp. P4 (5.370.23<br />
cm) was statistically smaller than that when cultured alone (6.130.38 cm). However, Streptomyces sp.<br />
P4 did not suppress the growth <strong>of</strong> the other two tested fungi, i.e. Corynesopra cassiicola (which causes<br />
leaf spot [30]) and Rhizoctonia solani (which causes root rot [31]) during the 14 days <strong>of</strong> observation<br />
(Figure 3). Anitha and Rabeeth [28] also reported the different responses <strong>of</strong> fungi to S. griseus<br />
chitinase. They suggested that the protein composition in the cell walls <strong>of</strong> different pathogenic fungi<br />
might make some fungal cell walls more resistant to chitinolytic degradation. Thus, only the co-culture<br />
containing F. oxysporum and Streptomyces sp. P4 was selected for SEM experiment.
100 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 95-104<br />
Figure 2. Chitinase activity and cell dry weight during the growth <strong>of</strong> Streptomyces sp. P4 . Error bars<br />
represent the standard deviation <strong>of</strong> 3 replicates.<br />
*<br />
Figure 3. In vitro inhibitory activity <strong>of</strong> Streptomyces sp. P4 against three fungi: Fusarium oxysporum<br />
f.sp. lycopersici, Corynesopra cassiicola and Rhizoctonia solani. Dark blue and light blue bars<br />
correspond to radial growth diameters <strong>of</strong> the fungi cultured alone (control) and co-cultured with<br />
Streptomyces sp. P4 (dual culture) respectively on the 9 th day <strong>of</strong> growth on PDA plates. Error bars<br />
represent standard deviation <strong>of</strong> 3 replicates. The asterisk indicates that the value differs significantly<br />
from control (p
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 95-104<br />
101<br />
A<br />
B<br />
Figure 4. Growth <strong>of</strong> Fusarium oxysporum f.sp. lycopersici grown alone (A) and co-cultured with<br />
Streptomyces sp. P4 (B) on PDA for 14 days<br />
Results obtained from SEM showed the breakage <strong>of</strong> the cell walls <strong>of</strong> F. oxysporum mycelia<br />
growing towards Streptomyces sp. P4 (Figure 5B) as compared to a control region (Figure 5A). The<br />
findings suggest that extracellular secondary metabolites and/or hydrolytic enzymes including chitinase<br />
play a crucial role in fungal growth inhibition. Prapagdee et al. [25] reported that the antifungal activity<br />
<strong>of</strong> S. hygroscopicus during exponential growth was mainly due to hydrolytic enzymes, while in the<br />
stationary phase it was due to secondary thermostable compound(s). In addition, there was a report on<br />
a positive correlation between chitinolytic and antagonistic activities <strong>of</strong> Streptomyces against the fungi<br />
Collectotrichum sublineolum, Guignardia citricarpa, Rhizoctonia solani and Fusarium oxysporum, but<br />
not in the oomycetes Pythium sp. and Phytophthora parasitica, which contain cellulose as a major cell<br />
wall component [16].<br />
A<br />
B<br />
Figure 5. Scanning electron microscopic analysis <strong>of</strong> Fusarium oxysporum f.sp. lycopersici grown<br />
alone (A) and co-cultured with Streptomyces sp. P4 (B). Bars indicate 1 m.
102 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 95-104<br />
CONCLUSIONS<br />
The present study provides background information for the potential use <strong>of</strong> the endophytic<br />
Streptomyces sp. P4 strain as a biocontrol agent antagonistic to specific fungi. The chitinase production<br />
and its association with the growth <strong>of</strong> Streptomyces sp. P4 were demonstrated. The fungal growth<br />
inhibition <strong>of</strong> F. oxysporum f.sp. lycopersici by Streptomyces sp. P4 was observed and demonstrated to<br />
result from the disruption <strong>of</strong> the fungal cell walls.<br />
ACKNOWLEDGEMENTS<br />
This work was financially supported by the Centre <strong>of</strong> Excellence for Innovation in Chemistry<br />
(PERCH-CIC), Lanna Products Company and the Graduate School <strong>of</strong> Chiang Mai University. We<br />
thank S. Chantal E. Stieber (Department <strong>of</strong> Chemistry and Chemical Biology, Cornell University) for<br />
critical reading <strong>of</strong> the manuscript.<br />
REFERENCES<br />
1. H. Willer and L. Kilcher (Eds.), “The world <strong>of</strong> organic agriculture __ statistics and emerging trends<br />
2011”, IFOAM and FiBL, 2011, http://www.organic-world.net/ (Accessed: 1 August 2011).<br />
2. M. Shoda, “Bacterial control <strong>of</strong> plant diseases”, J. Biosci. Bioeng., 2000, 89, 515-521.<br />
3. D. R. Fravel, “Commercialization and implementation <strong>of</strong> biocontrol”, Annu. Rev. Phytopathol.,<br />
2005, 43, 337-359.<br />
4. S. Sabaratnam and J. A. Traquair, “Formulation <strong>of</strong> a Streptomyces biocontrol agent for the<br />
suppression <strong>of</strong> Rhizoctonia damping-<strong>of</strong>f in tomato transplants”, Biol. Control, 2002, 23, 245-253.<br />
5. G. Sykes and F. A. Skinner, “Actinomycetales: Characteristics and Practical Importance”, Academic<br />
Press, London, 1973, pp.64-75.<br />
6. M. S. Abdelfattah, “Screening <strong>of</strong> terrestrial Streptomyces leading to identification <strong>of</strong> new<br />
stereoisomeric anthracyclines”, World J. Microbiol. Biotechnol., 2008, 24, 2619-2625.<br />
7. P. Dyson, “Streptomyces”, in “Encyclopedia <strong>of</strong> Microbiology” (Ed. M. Schaechter), 3 rd Edn.,<br />
Academic press, London, 2009, pp.318-332.<br />
8. P. Thapanapongworakul, “Characterization <strong>of</strong> endophytic actinomycetes capable <strong>of</strong> controlling<br />
sweet pea root rot diseases and effect on root nodule bacteria”, MS Thesis, 2003, Chiang Mai<br />
University, Thailand.<br />
9. M. Shimizu, Y. Nakagawa, Y. Sato, T. Furumai, Y. Igarashi, H. Onaka, R. Yoshida and H. Kunoh,<br />
“Studies on endophytic Actinomycetes (I) Streptomyces sp. isolated from rhododendron and its<br />
antifungal activity”, J. Gen. Plant Pathol., 2000, 66, 360-366.<br />
10. R. Gupta, R. K. Saxena, P. Chaturvedi and J. S. Virdi, “Chitinase production by Streptomyces<br />
viridificans: Its potential in fungal cell wall lysis”, J. Appl. Bacteriol., 1995, 78, 378-383.<br />
11. C. P. Souza, E. M. Burbano-Rosero, B. C. Almeida, G. G. Martins, L. S. Albertini and I. N. G.<br />
Rivera, “Culture medium for isolating chitinolytic bacteria from seawater and plankton”, World J.<br />
Microbiol. Biotechnol., 2009, 25, 2079-2082.<br />
12. D. A. Benson. I. Karsch-Mizrachi, D. J. Lipman, J. Ostell and E.W. Sayers, “GenBank”, Nucleic<br />
Acids Res., 2011, 39 (Database issue), D32-37.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 95-104<br />
103<br />
13. T. Taechowisan, J. F. Peberdy and S. Lumyong, “Chitinase production by endophytic Streptomyces<br />
aure<strong>of</strong>aciens CMU Ac 130 and its antagonism against phytopathogenic fungi”, Ann. Microbiol.,<br />
2003, 53, 447-461.<br />
14. G. L. Maria, K. R. Sridhar and N. S. Raviraja, “Antimicrobial and enzyme activity <strong>of</strong> mangrove<br />
endophytic fungi <strong>of</strong> southwest coast <strong>of</strong> India”, J. Agric. Technol., 2005, 1, 67-80.<br />
15. F. Green, C. A. Clausen and T. L. Highley, “Adaptation <strong>of</strong> the Nelson-Somogyi reducing-sugar<br />
assay to a microassay using microtiter plates”, Anal. Biochem., 1989, 182, 197-199.<br />
16. M. C. Quecine, W. L. Araujo, J. Marcon, C. S. Gai, J. L. Azevedo and A. A. Pizzirani-Kleiner,<br />
“Chitinolytic activity <strong>of</strong> endophytic Streptomyces and potential for biocontrol”, Lett. Appl.<br />
Microbiol., 2008, 47, 486-491.<br />
17. B. Lanoot, M. Vancanneyt, B. Hoste, K. Vandemeulebroecke, M. C. Cnockaert, P. Dawyndt, Z.<br />
Liu, Y. Huang and J. Swings, “Grouping <strong>of</strong> Streptomycetes using 16S-ITS RFLP fingerprinting”,<br />
Res. Microbiol., 2005, 156, 755-762.<br />
18. F. M. Atta and A. A. Zohri, “Transformation reactions <strong>of</strong> progesterone by different species <strong>of</strong><br />
Streptomyces”, J. Basic Microbiol., 1995, 35, 3-7.<br />
19. I. Höfer, M. Crüsemann, M. Radzom, B. Geers, D. Flachshaar, X. Cai, A. Zeeck and J. Piel,<br />
“Insights into the biosynthesis <strong>of</strong> hormaomycin, an exceptionally complex bacterial signaling<br />
metabolite”, Chem. Biol., 2011, 18, 381-391.<br />
20. H. Imai, A. Nakagawa and S. Omura, “Biosynthesis <strong>of</strong> the antibiotic okilactomycin”, J. Antibiot.,<br />
1989, 42, 1321-1323.<br />
21. K. Ochi, Y. Tsurumi, N. Shigematsu, M. Iwami, K. Umeharaand and M. Okuhara, “Physiological<br />
analysis <strong>of</strong> bicozamycin high-producing Streptomyces grise<strong>of</strong>lavus used at industrial level”, J.<br />
Antibiot., 1988, 41, 1106-1115.<br />
22. R. Nazari, A. Akbarzadeh, D. Norouzian, B. Farahmand, J. Vaez, A. Sadegi, F. Hormozi, M. Kiani-<br />
Rad and B. Zarbaksh, “Applying intra-specific protoplast fusion in Streptomyces grise<strong>of</strong>lavus to<br />
increase the production <strong>of</strong> desferrioxamine B”, Curr. Sci., 2005, 88, 1815-1820.<br />
23. A. Kuramoto, A. Lezhava, S. Taguchi, H. Momose and H. Kinashi, “The location and deletion <strong>of</strong><br />
the genes which code for SSI-like protease inhibitors in Streptomyces species”, FEMS Microbiol.<br />
Lett., 1996, 139, 37-42.<br />
24. B. Mahadevan and D. L. Crawford, “Properties <strong>of</strong> the chitinase <strong>of</strong> the antifungal biocontrol agent<br />
Streptomyces lydicus WYEC108”, Enzyme Microb. Technol., 1997, 20, 489-493.<br />
25. B. Prapagdee, C. Kuekulvong and S. Mongkolsuk, “Antifungal potential <strong>of</strong> extracellular<br />
metabolites produced by Streptomyces hygroscopicus against phytopathogenic fungi”, Int. J. Biol.<br />
Sci., 2008, 4, 330-337.<br />
26. M. Charpentier and F. Percheron, “The chitin-degrading enzyme system <strong>of</strong> a Streptomyces species”,<br />
Int. J. Biochem., 1983, 15, 289-292.<br />
27. N. Dahiya, R. Tewari and G. S. Hoondal, “Biotechnological aspects <strong>of</strong> chitinolytic enzyme: A<br />
review”, Appl. Microbiol. Biotechnol., 2006, 71, 773-782.<br />
28. A. Anitha and M. Rabeeth, “Degradation <strong>of</strong> fungal cell walls <strong>of</strong> phytopathogenic fungi by lytic<br />
enzyme <strong>of</strong> Streptomyces griseus”, Afr. J. Plant Sci., 2010, 4, 61-66.<br />
29. A. Anitha and M. Rabeeth, “Control <strong>of</strong> fusarium wilt <strong>of</strong> tomato by bi<strong>of</strong>ormulation <strong>of</strong> Streptomyces<br />
griseus in green house condition”, Afr. J. Basic Appl. Sci., 2009, 1, 9-10.
104 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 95-104<br />
30. S. Kurt, “Genetic variation in Corynesopra cassiicola, the target leaf spot pathogen”, Pakistan J.<br />
Biol. Sci., 2005, 8, 618-621.<br />
31. K. L. Olmos, S. H. Delgado and N. M. Pérez, “AFLP fingerprinting for identification <strong>of</strong><br />
anastomosis groups <strong>of</strong> Rhizoctonia solani Kühn from common bean (Phaseolus vulgaris L.) in<br />
Mexico”, Rev. Mexicana Fitopatol., 2005, 23, 147-151.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
Full Paper<br />
<strong>Maejo</strong> <strong>Maejo</strong> Int. J. Sci. Int. Technol. J. Sci. Technol. <strong>2012</strong>, 6(01), <strong>2012</strong>, 105-118 6(01), 105-118 105<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Water quality variation and algal succession in commercial<br />
hybrid catfish production ponds<br />
Chatree Wirasith 1,2 and Siripen Traichaiyaporn 3,*<br />
1<br />
Program <strong>of</strong> Environmental <strong>Science</strong>, Department <strong>of</strong> Biology, Faculty <strong>of</strong> <strong>Science</strong>, Chiang Mai<br />
University, Chiang Mai 50200, Thailand<br />
2<br />
Program <strong>of</strong> Fisheries, Faculty <strong>of</strong> Agricultural Technology and Agro-Industry, Rajamangala<br />
University <strong>of</strong> Technology, Suvarnabhumi, Ayutthaya 13000, Thailand<br />
3<br />
Department <strong>of</strong> Biology, Faculty <strong>of</strong> <strong>Science</strong>, Chiang Mai University, Chiang Mai 50200, Thailand<br />
* Corresponding author, e-mail: tsiripen@yahoo.com<br />
Received: 8 October 2011 / Accepted: 23 March <strong>2012</strong> / Published: 30 March <strong>2012</strong><br />
Abstract: This study on water quality variation and algal succession in commercial hybrid catfish<br />
production ponds was conducted in 2007 in Bang Pa-In district, Ayutthaya province, Thailand. The<br />
study covered two fish crops, May-August and September-December. The physico-chemical water<br />
quality in the catfish ponds changed dramatically over the study period due to the practices <strong>of</strong> water<br />
changing, lime application and the culture duration before harvesting. Samples <strong>of</strong> algae collected<br />
during the first crop period contained 83 species belonging to the following divisions: Chlorophyta<br />
(34 species), Cyanophyta (28 species), Euglenophyta (12 species), Bacillariophyta (6 species),<br />
Chrysophyta (1 species), Pyrrhophyta (1 species) and Cryptophyta (1 species). Samples collected<br />
during the second crop contained 60 species <strong>of</strong> the following divisions: Chlorophyta (28 species),<br />
Cyanophyta (16 species), Euglenophyta (10 species) and Bacillariophyta (6 species). Cyanophyta<br />
was the most abundant in both crops, followed by Chlorophyta, Euglenophyta, Bacillariophyta,<br />
Chrysophyta, Cryptophyta and Pyrrhophyta. The blue-green algae Microcystis increasingly<br />
dominated the algal population during the course <strong>of</strong> the culture period. Pseudanabaena spp. were<br />
succeeded by Oscillatoria spp. and then Microcystis spp. in the first crop. Microcystis spp.<br />
dominated during the first two months <strong>of</strong> the second crop, and then was succeeded by<br />
Planktolyngbya spp. and Nitzschia spp. in the third and fourth months. In summary, water quality<br />
may account for algal proliferation resulting in algal blooms and influence algal succession in<br />
commercial catfish production ponds.<br />
Keywords: water quality, algal succession, commercial production pond, hybrid catfish
106 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
INTRODUCTION<br />
Hybrid catfish, the <strong>of</strong>fspring <strong>of</strong> Clarias macrocephalus crossed with Clarias gariepinus, is<br />
among the most popular freshwater fish cultured commercially in South-east Asia, especially in<br />
Thailand [1]. In 2006, the production <strong>of</strong> this hybrid catfish in Thailand was estimated at 149,000 tons<br />
and valued at about 4,998.9 million Baht [2]. Since they are air breathers, hybrid catfish can be pondcultured<br />
at extremely high density, up to 100 fish/m 2 , with production reaching up to 100 tons/ha<br />
[1]. However, the <strong>of</strong>f-flavour in the flesh <strong>of</strong> cultured catfish can be a problem, leading to market<br />
value reduction and/or making the fish unmarketable for a certain period <strong>of</strong> time, from a few days to<br />
weeks [3]. The <strong>of</strong>f-flavour problem in cultured catfish is caused by compounds produced by certain<br />
kinds <strong>of</strong> blue-green algae, which are absorbed by the catfish and impart a bad flavour to the flesh if<br />
the harvest is delayed [4]. These blue-green algae can be found growing in catfish production ponds<br />
where an excessive amount <strong>of</strong> waste nutrients are generated. High density <strong>of</strong> the fish stocks and<br />
intensive feed input can result in extreme quantities <strong>of</strong> waste nutrients entering the production pond,<br />
which may account for the algal proliferation and resulting algal blooms [5]. Catfish cultured entirely<br />
and intensively in production ponds are commonly fed with pellet feed, trash fish and ground chicken<br />
skeletons and <strong>of</strong>fal. Although the water in the production pond is normally changed completely<br />
during each production cycle <strong>of</strong> about 120-150 days, such feeding nevertheless causes a general<br />
deterioration <strong>of</strong> water quality and a decrease in dissolved oxygen in the pond water [6]. Low water<br />
quality also influences fish growth. In addition, the wastewater effluent from hybrid catfish<br />
production ponds can contain concentrated algal compounds and nutrients with a high nitrogen<br />
content, making it unsuitable for other pr<strong>of</strong>itable uses such as the culturing <strong>of</strong> other aquatic animals<br />
[7–9].<br />
The seasonal succession <strong>of</strong> nutrients and phytoplankton populations in temperate and<br />
tropical systems has been extensively documented [10,11]. In tropical shallow water systems the<br />
roles <strong>of</strong> wet/dry seasons and wind typically have a greater impact on phytoplankton biomass<br />
production than inter-seasonal variations [12]. However, many other different types <strong>of</strong> algae<br />
prevailing in other climates also exhibit these wide swings in population densities. Some possible<br />
causes <strong>of</strong> these fluctuations include changes in temperature, pH, carbon dioxide, light intensity,<br />
nutrient concentration and the release <strong>of</strong> toxins by other organisms including competing algae [3].<br />
Inthamjit et al. [9] reported a significant change in water quality during intensive culturing <strong>of</strong> hybrid<br />
catfish, where the value ranges <strong>of</strong> the parameters contributing to water quality were: dissolved<br />
oxygen (DO), 4.8-30.8 mg/L; biochemical oxygen demand (BOD), 24-90 mg/L; chemical oxygen<br />
demand (COD), 62-330 mg/L; chlorophyll a, 218-1,908 μg/L; total suspended solids (TSS), 378-<br />
1,490 mg/L; ammonia-nitrogen (NH 3 -N), 0.003-0.270 mg/L; nitrate-nitrogen (NO 3 - -N), 0.00-0.06<br />
mg/L; nitrite-nitrogen (NO 2 - -N), 0.01-0.03 mg/L; total phosphorus (TP), 1.50-5.81 mg/L;<br />
orthophosphate phosphorus (PO 4 3- -P), 0.00-2.11 mg/L; alkalinity, 91-388 mg/L; hardness (as<br />
CaCO 3 ), 300-580 mg/L; electrical conductivity, 800-1,900 μS/cm; and pH, 6.7-7.8. Stephens and<br />
Farris [13] compared the water quality from two channel catfish farms near Paragould, Arkansas<br />
(USA) during the summer <strong>of</strong> 2001 and found the following values: DO, 8.3 and 9.6 mg/L;<br />
chlorophyll a, 62 and 143 mg/L; TSS, 102 and 81 mg/L; NH 3 -N, 0.16 and 0.16 mg/L; NO 3 - -N,<br />
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
107<br />
phosphorus (SRP)), 1.188 and 2.38 mg/L; alkalinity, 118 and 167 mg/L; hardness, 93 and 192 mg/L;<br />
conductivity, 303 and 354 μS/cm; pH, 9.0 and 8.9; water temperature, 23 and 29 ºC; and fecal<br />
coliform bacteria, 603 and 433 CFU/100 mL.<br />
In Thailand, Ayutthaya province has many commercial hybrid catfish farms. Geographically,<br />
the province is mainly a lowland plain situated in the Chao Phraya River basin <strong>of</strong> central Thailand,<br />
where the soil is highly fertile and water is readily available year-round. Because <strong>of</strong> these natural<br />
advantages, the province is an important farming area for many other types <strong>of</strong> fish in addition to<br />
hybrid catfish. Based on Thailand’s fisheries statistics in 2005 [2], Ayutthaya, with a total land area<br />
<strong>of</strong> 2,412.8 ha, has a total <strong>of</strong> 3,623 farms engaged in pisciculture with a total yield <strong>of</strong> 2,176 tons. In<br />
Bang Pa-In district <strong>of</strong> the province (Figure 1) where this study was conducted, various forms <strong>of</strong> fish<br />
culture are practiced, including extensive, semi-intensive, intensive and integrated fish culture. For<br />
commercial hybrid catfish farming in this area, most <strong>of</strong> the culture are intensive systems.<br />
This study investigated the variations <strong>of</strong> water quality in terms <strong>of</strong> physical, chemical and<br />
biological aspects, as well as the succession <strong>of</strong> algae in commercial hybrid catfish production ponds.<br />
Sampling site<br />
Map <strong>of</strong> Ayutthaya<br />
Figure 1. Map <strong>of</strong> Thailand and location <strong>of</strong> the study area in Phra Nakhon, Sri Ayutthaya province<br />
MATERIALS AND METHODS<br />
Study Area<br />
The study area selected, Bang Pa-In district <strong>of</strong> Ayutthaya province (Figure 1), has a total land<br />
area <strong>of</strong> 488.8 ha, where 873 farmers were involved in fish culture [14]. Early in 2007, a field survey<br />
was conducted in the district. Three hybrid catfish farms, which were located near to one another and
108 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
which practiced similar fish culture systems, were selected as the sampling sites for this study. The<br />
study duration covered two fish crops in 2007, with the first crop from May to August and the<br />
second from September to December. Three replicates <strong>of</strong> water samples were collected monthly<br />
from the hybrid catfish production ponds at the three selected fish farms. In keeping with standard<br />
industrial practice for the culture <strong>of</strong> hybrid catfish, farmers released fingerlings into the production<br />
ponds (each 0.08 ha in size) at a density <strong>of</strong> 50 fingerlings/m2. The fingerlings were fed twice a day<br />
between 8-9 a.m. and 5-6 p.m. with pellet feed during the first and second months. Then during the<br />
third and fourth months they were fed chicken <strong>of</strong>fal mixed with cassava chips in a ratio <strong>of</strong> 95:5.<br />
Water changing and lime application were carried out occasionally to manage water quality in the<br />
production ponds. The fish were cultured for 120-130 days before being harvested. The average fish<br />
yields were 59.38 tons/ha and 57.50 tons/ha for the first and second crops respectively, with the food<br />
conversion rate reaching 3.8-4.2.<br />
Physico-Chemical Water Quality Analysis<br />
Three replicates <strong>of</strong> water samples were collected monthly from the production ponds during<br />
both fish crops: May-August and September-December, 2007. All water samples were collected at a<br />
depth <strong>of</strong> 0.3-0.4 m and then preserved in an icebox until further processing. Water temperature, pH<br />
and DO were measured in situ using a portable hand-held meter (Multi 350i; WTW, Germany).<br />
Water transparency and water depth were measured using a Secchi disk and a measuring tape<br />
respectively. The analyses <strong>of</strong> chemical parameters were then carried out using suitable methods [15-<br />
16]: BOD by azide modification method; NH 3 -N by Nesslerisation method; NO 3 - -N by<br />
phenoldisulphonic acid method; total Kjeldahl nitrogen (TKN) by macro-Kjeldahl method; total<br />
phosphate by persulphate digestion/stannous chloride method; and orthophosphate phosphorus<br />
(PO 4 3- -P) by stannous chloride method.<br />
Algal Analysis<br />
For the algae count, the water sample (500 mL) from each production pond was transferred<br />
to a 500-mL cylinder and fixed with 5 mL <strong>of</strong> Lugol’s solution (20 g glacial acetic acid, 20 g<br />
potassium iodide and 20 g iodine dissolved in 200 mL distilled water). The preserved sample was left<br />
to stand in the dark for 10 days to allow concentration by decantation. A 20-25 mL sample from the<br />
lower layer <strong>of</strong> the 500-mL cylinder, containing the sedimented algae, was obtained and transferred to<br />
a 50-mL cylinder. A second decantation was conducted after another 7 days in the dark; a 10-mL<br />
sample from the lower layer <strong>of</strong> the 50-mL cylinder, containing the sedimented algae, was put into a<br />
glass vial and stored in a dark cupboard [17]. This concentrated sample <strong>of</strong> algae was used for their<br />
identification [18-22] and counting under a compound light microscope [17].<br />
Statistical Analysis<br />
Collected data were statistically analysed using SPSS s<strong>of</strong>tware program, version 14.<br />
Differences in means <strong>of</strong> water quality and algal population were established using analysis <strong>of</strong> variance<br />
(ANOVA), and relationships between algae and water quality parameters were tested using Pearson<br />
product-moment correlation. The level <strong>of</strong> significance was set at 0.05.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
109<br />
RESULTS AND DISCUSSION<br />
Water Quality<br />
Water quality based on physico-chemical and biological parameters from production ponds at<br />
all three sampling sites and for both fish crops is presented in Table 1. The temperature optimum for<br />
aquaculture in Thailand is between 25-33ºC, depending on the species <strong>of</strong> fish being cultured; at<br />
temperatures above or below the optimum, fish growth is reduced [23]. There was a significant<br />
difference in temperature between the two study crops and a marked decline during the 4 th month<br />
(December) for the second crop. The optimal water transparency for aquaculture is 30 cm [24].<br />
Water transparency in the study ponds ranged between 2.3-25.5 cm for the first fish crop and 1.1-<br />
10.3 cm for the second crop __ a significant difference between the two crops. There was no<br />
significant difference in water depth between the two crops.<br />
The optimal pH range for water used in aquaculture is between 6.5-8.5; however this will<br />
vary slightly depending on the cultured species [25]. Ingthamjit et al.[9] reported a pH range <strong>of</strong> 6.8-<br />
7.9 in hybrid catfish production ponds. The pH in the study ponds ranged between 6.6-7.1 for the<br />
first fish crop and 6.7-7.4 for the second crop. Generally, it is recommended that alkalinity be<br />
maintained within 50-300 mg/L to provide a sufficient buffering (stabilising) effect against pH swings<br />
that occur in ponds due to the respiration <strong>of</strong> the aquatic flora [25, 26]. Alkalinity in the study ponds<br />
ranged between 115.7-145.7 mg/L and 114.4-160.0 mg/L in the first and second fish crops<br />
respectively, thus displaying a significant difference.<br />
DO is probably the most critical water quality variable in freshwater aquaculture ponds. To<br />
achieve optimal growth, a good rule <strong>of</strong> thumb is to maintain the DO level at saturation or at least 5<br />
mg/L [25, 26]. The ranges <strong>of</strong> DO values in the study ponds were between 0.8-4.8 mg/L during the<br />
first fish crop and 1.5-4.3 mg/L during the second crop, with a significant difference in the 4 th -month<br />
values <strong>of</strong> the two study crops. Also, in the 4 th month, BOD reached 78.3 and 85.8 mg/L for the first<br />
and second fish crops respectively. These values were significantly different from the 1 st - and 2 nd -<br />
month values for both fish crops.<br />
According to a study by Boyd [26] on unfertilised woodland ponds in Alabama, the average<br />
total NH 3 -N (NH 4 + plus NH 3 expressed in terms <strong>of</strong> N) was 0.052 mg/L and and that for NO 3 - -N was<br />
0.075 mg/L. In intensive fish culture ponds, much higher concentrations <strong>of</strong> inorganic N are common.<br />
Channel catfish culture ponds can contain up to 0.5 mg/L <strong>of</strong> total NH 3 -N and 0.25 mg/L <strong>of</strong> NO 3 - -N<br />
[26]. NH 3 -N concentrations in the study ponds showed no significant difference between the first and<br />
second fish crops (0.06-1.77 mg/L). NO 3 - -N concentrations varied between 0.01-0.24 mg/L and<br />
0.02-0.03 mg/L for the first and second fish crops respectively, a significant difference being found<br />
between the 1 st and 4 th months <strong>of</strong> the first crop. Nitrogen is also present as soluble organic<br />
compounds and as constituents <strong>of</strong> living and dead particulate organic matter. Concentrations <strong>of</strong><br />
organic nitrogen are usually well below 1 mg/L in unpolluted natural water [26]. In fish production<br />
ponds, phytoplankton blooms are normally heavy and the concentration <strong>of</strong> organic nitrogen may<br />
exceed 2-3 mg/L. In the study ponds, the TKN concentration reached a maximum level (5.6 mg/L) in<br />
the 4 th month <strong>of</strong> the second fish crop.
110 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
Table 1. Monthly means and standard deviations (mean ± SD) <strong>of</strong> physico-chemical characteristics <strong>of</strong> water samples from commercial hybrid catfish<br />
production ponds, Ayutthaya province, Thailand, 2007<br />
Parameter<br />
Month<br />
Crop 1 (mean ± SD) Crop 2 (mean ± SD)<br />
May June July August September October November December<br />
Water temperature (ºC) 29.4 a ± 0.1 31.8 a ± 0.1 31.0 a ± 0.1 30.8 a ± 0.1 29.0 b ± 0.1 28.8 b ± 0.1 28.4 b ± 0.1 27.5 b ± 0.3<br />
Water transparency (cm) 25.5 a ± 0.1 10.4 a ± 0.9 5.7 a ± 0.8 2.3 a ± 0.2 10.3 b ± 0.3 5.5 b ± 0.9 1.7 b ± 0.2 1.1 b ± 0.1<br />
Water depth (m) 1.25 ± 0.1 1.25 ± 0.1 1.25 ± 0.1 1.25 ± 0.1 1.25 ± 0.1 1.25 ± 0.1 1.25 ± 0.1 1.00 ± 0.2<br />
pH 6.7 ± 0.1 7.1 a ± 0.2 6.6 ± 0.1 7.0 b ± 0.1 6.7 ± 0.1 6.7 b ± 0.1 6.7 ± 0.1 7.4 a ± 0.2<br />
Alkalinity as CaCO3 (mg/L) 145.7 a ± 5.1 140.0 a ± 3.0 123.7 b ± 3.5 115.7 b ± 2.1 121.1 b ± 3.9 114.4 b ± 1.9 151.0 a ± 3.6 160.0 a ± 1.7<br />
DO (mg/L) 4.8 ± 0.6 3.2± 0.9 2.6 ± 0.4 0.8 b ± 0.2 4.3 ± 0.4 3.5 ± 0.3 2.4 ± 0.5 1.5 a ± 0.3<br />
BOD (mg/L) 24.2 b ± 1.4 38.2 b ± 2.8 65.1 ± 6.8 78.3 ± 6.0 49.7 a ± 3.2 61.1 a ± 4.6 75.0 ± 2.5 85.8 ± 5.2<br />
NH3-N (mg/L) 0.2 ± 0.1 0.06 ± 0.02 0.7 ± 0.1 1.7 ± 0.1 0.06 ± 0.04 0.1 ± 0.05 0.6 ± 0.2 1.8 ± 0.3<br />
NO3 - -N (mg/L) 0.01 b ± 0.01 0.02 ± 0.01 0.03 ± 0.02 0.2 a ± 0.04 0.03 a ± 0.01 0.02 ± 0.01 0.03 ± 0.01 0.03 b ± 0.01<br />
TKN (mg/L) 1.2 ± 0.1 1.0 ± 0.1 2.4 ± 0.4 2.9 b ± 0.3 1.3 ± 0.05 1.27 ± 0.2 2.0 ± 0.2 5.6 a ± 0.7<br />
Total P (μg/L) 4.1 ± 3.4 11.8 ± 2.9 11.9 ± 2.4 22.1 ± 4.3 6.0 ± 1.0 17.0 ± 3.8 12.7 ± 2.5 14.9 ± 3.5<br />
PO4 3- -P (μg/L) 0.1b ± 0.1 0.8 b ± 0.5 9.5 ± 1.8 16.6 a ± 0.9 4.5 a ± 1.2 8.6 a ± 1.2 8.6 ± 1.3 10.6 b ± 0.5<br />
Note: Values in the same row followed by different superscripts indicate significant difference (p
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
111<br />
The phosphorus concentration in water is usually quite low; dissolved orthophosphate<br />
concentration lies within 5-20 μg/L and seldom exceeds 100 μg/L even in highly eutrophic waters,<br />
while the concentration <strong>of</strong> total phosphorus seldom exceeds 1,000 μg/L [26]. Total phosphorus<br />
concentrations in the study ponds varied between 4.1-22.1 μg/L with no significant difference<br />
between the two crops, while PO 4 3- -P concentrations were significantly different between the two<br />
crops in the 1 st , 2 nd and 4 th months.<br />
The waste effluent from intensive fish culture as a source <strong>of</strong> pollution <strong>of</strong> natural bodies <strong>of</strong><br />
water has been a major concern [27]. In the present study, water quality deteriorated as the farming<br />
season progressed, an occurrence shared by several other findings [9, 13, 28-30]. However, there<br />
was a difference in the amount <strong>of</strong> nutrients added to the water in the fish production ponds owing to<br />
difference in the feed applied. In this study, hybrid catfish were fed chicken <strong>of</strong>fal mixed with cassava<br />
chips as fresh feed, which actually caused the water quality to deteriorate more rapidly. This was<br />
similar to the findings <strong>of</strong> Yi et al. [6], who showed that using trash fish and chicken <strong>of</strong>fal as feed for<br />
fish culture could lead to a rapid deterioration <strong>of</strong> water quality.<br />
Algal Succession<br />
Algae found in the water samples <strong>of</strong> the first fish crop were categorised into 83 species <strong>of</strong> 7<br />
divisions, namely Chlorophyta (34 species), Cyanophyta (28 species), Euglenophyta (12 species),<br />
Bacillariophyta (6 species), Chrysophyta (1 species), Pyrrhophyta (1 species) and Cryptophyta (1<br />
species), whereas those <strong>of</strong> the second fish crop were categorised into 60 species <strong>of</strong> 4 divisions,<br />
namely Chlorophyta (28 species), Cyanophyta (16 species), Euglenophyta (10 species) and<br />
Bacillariophyta (6 species) (Table 2).<br />
Abundance percentages <strong>of</strong> the algal divisions are shown in Table 3. There was no significant<br />
difference between the two fish crops in the number <strong>of</strong> algae <strong>of</strong> each division. Chlorophyta was most<br />
abundant in both fish crops, followed by Cyanophyta, Euglenophyta, Bacillariophyta, Chrysophyta,<br />
Cryptophyta and Pyrrhophyta. This confirmed the results obtained by Ingthamjit et al. [9] and Boyd<br />
[26] as well as several other reports [28-33]. Phytoplankton occurring in fish production ponds<br />
includes members <strong>of</strong> the following taxonomic divisions: green algae (Chlorophyta), blue-green algae<br />
(Cyanophyta), euglenophytes (Euglenophyta), yellow-green and golden-brown algae, diatoms<br />
(Chrysophyta) and din<strong>of</strong>lagellates (Pyrrhophyta) [26, 28-33]. The dominant algae in the first fish<br />
crop were Microcystis spp., followed by Pseudanabaena spp., Monoraphidium spp., Oscillatoria<br />
spp., Scenedesmus spp., Euglena spp., Merismopedia spp., Cyclotella spp., Coelastrum spp.,<br />
Tetrastrum sp. and Spirulina sp.(Table 4). The second fish crop was dominated by Microcystis spp.,<br />
followed by Planktolyngbya sp., Spirulina sp., Cyclotella spp., Pseudanabaena spp., Merismopedia<br />
spp., Nitzschia spp., Monoraphidium spp., Scenedesmus spp., Tetrastrum sp. and Phacus spp.<br />
(Table 5). Algae <strong>of</strong> the division Cyanophyta (Microcystis) grew densely and were the dominant<br />
division in both fish crops. As reported by Welker et al. [34], Microcystis is commonly present in<br />
eutrophic temperate lakes during summer. These findings are supported by the results <strong>of</strong> the present<br />
study: Microcystis species demonstrated better growth in summer and under eutrophic conditions<br />
with high concentrations <strong>of</strong> nutrients in the fish production pond water. Lin [32] and Chowdhury and<br />
Mamun[33] also reported that Cyanophyta dominated nutrient-rich channel catfish ponds.
112 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
Table 2. Diversity and classification <strong>of</strong> algae occurring in commercial hybrid catfish production<br />
ponds, Ayutthaya province, Thailand, 2007<br />
DIVISION CHLOROPHYTA<br />
Actinastrum hantzschii, Ankistrodesmus sp., Closteriopsis sp., Closterium sp. 1, Coelastrum astroideum, Coelastrum<br />
pseudomicroporum, Coelastrum sp.1, Cosmarium sp.1, Crucigenia crucifera, Crucigeniella rectangularis,<br />
Dictyosphaerium granulatum, Dictyosphaerium sp., Elakatothrix sp., Kirchneriella sp, Monoraphidium arcuatum,<br />
Monoraphidium caribeum, Monoraphidium contortum, Monoraphidium griffithii, Monoraphidium minutum, Oocystis<br />
sp., Pediastrum duplex, Pediastrum simplex, Scenedesmus bernardii, Scenedesmus disciformis, Scenedesmus<br />
microspina, Scenedesmus opoliensis, Scenedesmus pannonicus, Scenedesmus perforates, Scenedesmus velitaris,<br />
Scenedesmus sp. 1, Scenedesmus sp. 2, Staurastrum cingulum, Tetraedron caudatum, Tetrastrum heteracanthum<br />
DIVISION CYANOPHYTA<br />
Anabaena catenula, Aphanothece sp., Arthrospira sp., Chlor<strong>of</strong>lexus sp., Chroococcus minutes, Chroococcus sp,<br />
Cylindrospermopsis curvispora, Cylindrospermopsis helicoidea, Cylindrospermopsis raciborskii, Gomphosphaeria sp.,<br />
Komvophoron sp., Lyngbya sp., Merismopedia convulata, Merismopedia glauca, Merismopedia punctata, Microcystis<br />
aeruginosa, Microcystis wesenbergii, Microcystis sp., Oscillatoria agardhii, Oscillatoria limosa, Oscillatoria redekei,<br />
Planktolyngbya limnetica, Pseudanabaena catenata, Pseudanabaena sp.1, Pseudanabaena sp.2, Raphidiopsis sp.,<br />
Romeria sp., Spirulina sp.<br />
DIVISION EUGLENOPHYTA<br />
Euglena acus, Phacus orbicularis, Phacus triqueter, Phacus sp.1, Phacus sp.2, Phacus sp.3, Strombomonas sp.,<br />
Trachelomonas acanthostoma, Trachelomonas caudata, Trachelomonas cylindrical, Trachelomonas volvocina,<br />
Trachelomonas sp.<br />
DIVISION BACILLARIOPHYTA<br />
Aulacoseira granulata, Cyclotella sp., Fragilaria sp., Melosira sp., Nitzschia sp.1, Nitzschia sp.2<br />
DIVISION CHRYSOPHYTA<br />
Isthmochloron sp.<br />
DIVISION PYRRHOPHYTA<br />
Peridinium sp.<br />
DIVISION CRYPTOPHYTA<br />
Cryptomonas sp.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
113<br />
Table 3. Means and standard deviations <strong>of</strong> the percentages <strong>of</strong> abundance <strong>of</strong> each algal division in hybrid catfish ponds by month<br />
Division<br />
Abundance (%)<br />
Crop 1 (mean ± SD) Crop 2 (mean ± SD)<br />
May June July August September October November December<br />
Chlorophyta 16.26 ± 8.43 34.67 ± 2.61 19.42 ± 12.52 29.79 ± 10.30 12.29 ± 10.88 9.24 ± 7.15 10.56 ± 4.11 21.23 ± 7.18<br />
Cyanophyta 68.40 ± 14.84 41.26 ± 8.59 73.23 ± 5.91 53.57 ± 1.9 74.77 ± 15.68 76.03 ± 12.46 69.02 ± 16.76 49.40 ± 12.98<br />
Bacillariophyta 3.77 ± 0.52 9.51 ± 10.69 2.69 ± 1.29 15.59 ± 11.85 7.33 ± 4.96 9.51 ± 5.16 19.88 ± 14.21 28.36 ± 7.73<br />
Euglenophyta 8.61 ± 4.83 12.10 ± 11.52 4.66 ± 0.65 0.75 ± 0.07 5.60 ± 0.38 5.19 ± 1.6 0.54 ± 0.21 1.02 ± 0.5<br />
Chrysophyta 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00 0.31 ± 0.03 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00<br />
Pyrrhophyta 1.07 ± 0.38 2.47 ± 1.80 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00<br />
Cryptophyta 1.89 ± 1.63 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00 0.00 ± 0.00
114 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
Table 4. Means and standard deviations <strong>of</strong> the percentages <strong>of</strong> abundance <strong>of</strong> each algal genus in hybrid<br />
catfish ponds by month (Crop 1)<br />
Genus (Division) Abundance (%)<br />
May June July August<br />
Microcystis spp. (Cyanophyta) 0.00 ± 0.00 1.45 ± 0.37 74.66 ± 7.60 28.89 ± 3.27<br />
Pseudanabaena spp. (Cyanophyta) 47.96 ± 21.62 4.14 ± 0.91 3.38 ± 1.01 4.39 ± 3.80<br />
Monoraphidium spp. (Chlorophyta) 12.62 ± 8.20 27.43 ± 8.60 3.17 ± 0.34 11.50 ± 9.15<br />
Oscillatoria spp. (Cyanophyta) 4.92 ± 0.76 37.37 ± 3.89 3.59 ± 1.37 1.30 ± 0.71<br />
Scenedesmus spp. (Chlorophyta) 11.41 ± 0.24 6.86 ± 3.99 4.83 ± 1.63 9.10 ± 4.64<br />
Euglena spp. (Euglenophyta) 12.58 ± 6.01 10.75 ± 9.08 1.53 ± 0.42 0.62 ± 0.04<br />
Merismopedia spp. (Cyanophyta) 3.73 ± 6.46 0.68 ± 1.18 3.57 ± 6.18 13.18 ± 4.99<br />
Cyclotella spp. (Bacillariophyta) 0.00 ± 0.00 3.47 ± 2.81 2.15 ± 2.33 14.35 ± 10.03<br />
Coelastrum spp. (Chlorophyta) 6.79 ± 5.96 2.07 ± 0.46 1.48 ± 1.18 2.33 ± 1.99<br />
Tetrastrum sp. (Chlorophyta) 0.00 ± 0.00 4.76 ± 5.02 1.29 ± 0.64 5.14 ± 4.06<br />
Spirulina sp. (Cyanophyta) 0.00 ± 0.00 1.00 ± 0.57 0.67 ± 0.46 9.18 ± 9.54<br />
Table 5. Means and standard deviations <strong>of</strong> the percentages <strong>of</strong> abundance <strong>of</strong> each algal genus in hybrid<br />
catfish ponds by month (Crop 2)<br />
Genus (Division)<br />
Abundance (%)<br />
September October November December<br />
Microcystis spp. (Cyanophyta) 32.93 ± 4.57 30.03 ± 5.45 23.31 ± 11.12 16.22 ± 9.73<br />
Planktolyngbya sp. (Cyanophyta) 14.21 ± 4.32 9.56 ± 6.30 28.90 ± 23.35 6.51 ± 4.89<br />
Spirulina sp. (Cyanophyta) 17.68 ± 8.68 29.84 ± 11.36 3.37 ± 0.63 0.95 ± 0.63<br />
Cyclotella spp. (Bacillariophyta) 3.80 ± 2.04 7.04 ± 7.20 19.21 ± 18.01 16.65 ± 3.49<br />
Pseudanabaena spp. (Cyanophyta) 14.33 ± 7.13 10.83 ± 1.20 12.77 ± 20.85 6.51 ± 4.89<br />
Merismopedia spp. (Cyanophyta) 3.91 ± 4.15 5.17 ± 0.68 5.50 ± 4.33 15.93 ± 6.86<br />
Nitzschia spp. (Bacillariophyta) 1.20 ± 0.38 0.00 ± 0.00 0.52 ± 0.13 19.48 ± 2.33<br />
Monoraphidium spp. (Chlorophyta) 3.91 ± 4.15 1.42 ± 1.11 1.87 ± 1.49 8.94 ± 6.15<br />
Scenedesmus spp. (Chlorophyta) 3.09 ± 0.77 2.57 ± 3.28 2.14 ± 0.76 6.09 ± 5.99<br />
Tetrastrum sp. (Chlorophyta) 2.34 ± 3.01 0.22 ± 0.38 1.85 ± 0.80 1.25 ± 1.92<br />
Phacus spp. (Euglenophyta) 4.37 ± 4.42 1.91 ± 0.72 0.20 ± 0.20 0.18 ± 0.13
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
115<br />
Figure 2 shows the dominant species among the algal populations during the different months<br />
<strong>of</strong> the study. Pseudanabaena spp. were the dominant species during the 1 st month <strong>of</strong> the first fish<br />
crop, followed by Oscillatoria spp. and Microcystis spp. in the 2 nd and 3 rd months respectively. For<br />
the 2 nd fish crop, Microcystis spp. dominated during the first two months, followed by<br />
Planktolyngbya sp. in the 3 rd month and Nitzschia spp. in the 4 th month. The succession <strong>of</strong> algae<br />
seemed to be associated with nutrient accumulation and water changing. This could be determined<br />
from the results <strong>of</strong> the correlation analysis <strong>of</strong> algal populations and water quality parameters. The<br />
prevalence <strong>of</strong> Pseudanabaena spp. was found to be significantly associated with water transparency,<br />
which was at the highest level in the first month when these species were the dominant algae.<br />
Microcystis spp. showed a high correlation with nitrate-nitrogen levels, which increased in the 3 th<br />
and 4 th months <strong>of</strong> the first fish crop and in the 1 st and 2 nd months <strong>of</strong> the second fish crop. The<br />
decrease in nitrate-nitrogen level also coincided with a decline in the population <strong>of</strong> Microcystis and<br />
an increase <strong>of</strong> Planktolyngbya and Nitzschia in the succeeding months. Planktolyngbya sp. showed<br />
no significant correlation with any <strong>of</strong> the studied water quality parameters, but tended to grow better<br />
in water with less transparency and lower temperature. Nitzschia spp. were found to significantly<br />
correspond to TKN level. An increase in TKN during the last month <strong>of</strong> the second fish crop might<br />
contribute to the succession <strong>of</strong> Nitzschia spp., as also indicated in studies by Ingthmjit et al.[9],<br />
Brunson et al. [35], and Zimba et al [36]. However, the succession <strong>of</strong> algae in the catfish production<br />
ponds differs from that occurring in natural lakes, owing to the dense stock <strong>of</strong> fish in the ponds and<br />
the daily feeding which provides abundant nutrients. As reported by Stephens and Farris [13], algal<br />
density in commercial channel catfish production ponds is limited more by nutrient availability than<br />
by light. Many other different types <strong>of</strong> algae also exhibit these wide swings in population density.<br />
Possible causes <strong>of</strong> these fluctuations include changes in temperature, pH, carbon dioxide<br />
concentration, light intensity and nutrient concentration, and the release <strong>of</strong> toxins by other organisms<br />
including competing algae [3].<br />
CONCLUSIONS<br />
The physico-chemical water quality in commercial hybrid catfish production ponds located in<br />
Ayutthaya changed dramatically over the culture period. Certain water quality parameters influenced<br />
algal dominance and succession. While Microcystis <strong>of</strong> the division Cyanophyta dominated<br />
throughout, Pseudanabaena was succeeded by Oscillatoria, followed by Microcystis in the first fish<br />
crop period, whereas Microcystis was succeeded by Planktolyngbya and Nitzschia in the second fish<br />
crop period.
116 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
Figure 2. Percentages <strong>of</strong> abundance <strong>of</strong> algae in hybrid catfish ponds by month<br />
ACKNOWLEDGEMENTS<br />
Our sincere gratitude is extended to Rajamangala University <strong>of</strong> Technology Suvarnabhumi<br />
for financial support.<br />
REFERENCES<br />
1. S. Areerat, “Clarias culture in Thailand”, Aquaculture, 1987, 63, 355-362.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
117<br />
2. Fisheries Information Technology Center, “Fisheries Statistics <strong>of</strong> Thailand 2005”, Department<br />
<strong>of</strong> Fisheries, Ministry <strong>of</strong> Agriculture and Cooperatives, Bangkok, 2007.<br />
3. H. S. Killian, “Phytoplankton in catfish ponds”, Cooperative Extension Program, University <strong>of</strong><br />
Arkansas, http://www.uaex.edu/aquaculture2/FSA/FSA9070.htm (Accessed: 2011).<br />
4. P. V. Zimba, C. S. Tucker, C. C. Mischke and C. C. Grimm, “Short-term effects <strong>of</strong> diuron on<br />
catfish pond ecology”, N. Am. J. Aquaculture, 2002, 64, 16-23.<br />
5. P. V. Zimba, L. Khoo, P. S. Gaunt, S. Brittain and W. W. Carmichael, “Confirmation <strong>of</strong> catfish,<br />
Ictalurus punctatus (Rafinesque), mortality from Microcystis toxins”, J. Fish Diseas., 2001, 24,<br />
41-47.<br />
6. Y. Yi, C. K. Lin and J. S. Diana, “Hybrid catfish (Clarias macrocephalus x C. gariepinus) and<br />
Nile tilapia (Oreochromis niloticus) culture in an integrated pen-cum-pond system: Growth<br />
performance and nutrient budgets”, Aquaculture, 2003, 217, 395-408.<br />
7. P. R. Adler, F. Takeda, D. M. Glenn and S. T. Summerfelt, “Enhancing aquaculture<br />
sustainability through utilizing byproducts”, World Aquaculture, 1996, 27, 24-26.<br />
8. T. V. R. Pillay, “The challenges <strong>of</strong> sustainable aquaculture”, World Aquaculture, 1996, 27, 7-9.<br />
9. S. Ingthamjit, S. Areerat, P. Tientong and S. Wallie, “Variation <strong>of</strong> water quality, phytoplankton<br />
and bacteria in catfish ponds”, Academic document No. 129, 1992, Freshwater Fisheries<br />
Research Institute, Department <strong>of</strong> Fisheries, Ministry <strong>of</strong> Agriculture and Co-operatives,<br />
Bangkok.<br />
10. G. E. Hutchinson, “A Treatise on Limnology”, Vol. 1, Wiley, New York, 1957.<br />
11. W. S. Rast, V. H. Smith and J. A. Thornton, “Characteristics <strong>of</strong> eutrophication”, in “The<br />
Control <strong>of</strong> Eutrophication in Lakes and Reservoirs” (Ed. S. O. Ryding and W. Rast (Eds.),<br />
UNESCO and Parthenon Publishers, London, 1989, pp.37-64.<br />
12. P. V. Zimba, “The use <strong>of</strong> nutrient enrichment bioassays to test for limiting factors affecting<br />
epiphytic growth in Lake Okeechobee, Florida: Confirmation <strong>of</strong> nitrogen and silica limitation”,<br />
Archiv. Hydrobiol., 1998, 141, 459-468.<br />
13. W. W. Stephens and J. L. Farris, “Instream community assessment <strong>of</strong> aquaculture effluents”,<br />
Aquaculture, 2004, 231, 149-162.<br />
14. The Center for Fisheries Development <strong>of</strong> Phranakorn-Sri Ayutthaya, “Fisheries statistics <strong>of</strong><br />
Phranakorn-Sri Ayutthaya 2007” (unpublished document), Department <strong>of</strong> Fisheries, Ministry <strong>of</strong><br />
Agriculture and Co-operatives, Bangkok.<br />
15. APHA, AWWA and WPCF, “Standard Methods for Examination <strong>of</strong> Water and Wastewater”,<br />
20 th Edn., American Public Health Association, Washington, DC, 1998.<br />
16. S. Traichaiyaporn, “Water Quality Analysis”, Department <strong>of</strong> Biology, Faculty <strong>of</strong> <strong>Science</strong>,<br />
Chiang Mai University, Chiang Mai, 2000 (Textbook in Thai).<br />
17. S. Traichaiyaporn, “Applied Phycology”, Department <strong>of</strong> Biology, Faculty <strong>of</strong> <strong>Science</strong>, Chiang<br />
Mai University, Chiang Mai, 2000 (Textbook in Thai).<br />
18. L. Wongrat, “Phytoplankton”, Department <strong>of</strong> Fishery Biology, Faculty <strong>of</strong> Fisheries, Kasetsart<br />
University, Bangkok, 1999 (Textbook in Thai).<br />
19. G. W. Prescott, “How to Know the Freshwater Algae”, 3 rd Edn., W. C. Brown, Dubuque,<br />
1978.<br />
20. C. B. Harrold and J. W. Michaell, “Introduction to the Algae”, 2 nd Edn., Prentice-Hall,<br />
Englewood Cliffs, 1985.
118 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 105-118<br />
21. A. Shirota, “The Plankton <strong>of</strong> South Vietnam: Fresh Water and Marine Plankton”, Overseas<br />
Technical Cooperation Agency, Tokyo, 1966.<br />
22. X. N. Verlencar and S. R. Desai, “Phytoplankton Identification Manual”, National Institute <strong>of</strong><br />
Oceanography, Dona Panla, 2004.<br />
23. M. Duangsawasdi and J. Somsiri, “Water Quality and Analytical Methods for Fisheries<br />
Research”, Department <strong>of</strong> Fisheries, Ministry <strong>of</strong> Agriculture and Cooperatives, Bangkok, 1986<br />
(Textbook in Thai).<br />
24. Y. Musit, “Water Quality for Aquaculture”, Department <strong>of</strong> Aquaculture, Faculty <strong>of</strong> Fisheries,<br />
Kasetsart University, Bangkok, 1992 (Textbook in Thai).<br />
25. J. K. Buttner, R. W. Soderberg and D. E. Terlizzi, “An introduction to water chemistry in<br />
freshwater aquaculture”, Fact Sheet No.170, 1993, Northeastern Regional Aquaculture Center,<br />
Dartmouth, Massachusetts, USA.<br />
26. C. E. Boyd, “Water Quality in Ponds for Aquaculture”, Department <strong>of</strong> Fisheries and Allied<br />
Aquacultures, Auburn University, Auburn (USA), 1990.<br />
27. H. S. Egna and C. E. Boyd, “Dynamics <strong>of</strong> Pond Aquaculture”, CRC Press, Boca Raton, 1997.<br />
28. B. Sen and F. Sonmez, “A study on the algae in fish ponds and their seasonal variations”, Int. J.<br />
Sci. Technol., 2006, 1, 25-33.<br />
29. M. A. Burford and D. C. Pearson, “Effect <strong>of</strong> different nitrogen sources on phytoplankton<br />
composition in aquaculture ponds”, Aquat. Microb. Ecol., 1998, 15, 277-284.<br />
30. C. S. Tucker and M. van der Ploeg, “Seasonal changes in water quality in commercial channel<br />
catfish ponds in Mississippi”, J. World Aquaculture Soc., 1993, 24, 473-481.<br />
31. M. M. Littler and J. H. Graffius, “The annual distribution <strong>of</strong> phytoplankton communities in a<br />
south-eastern Ohio pond”, Ohio J. Sci., 1974, 74, 313-324.<br />
32. C. K. Lin, “Biological principles <strong>of</strong> pond culture: Phytoplankton and macrophytes”, in<br />
“Principles and Practices <strong>of</strong> Pond Aquaculture” (Ed. J. E. Lamann, R. O. Smitherman and G.<br />
Tchobanoglous), Oregon State University, Corvallis, 1983.<br />
33. A. H. Chowdhury and A. A. Mamun, “Physio-chemical conditions and plankton population <strong>of</strong><br />
two fishponds in Khulna”, Univ. J. Zool. Rajshahi Univ., 2006, 25, 41-44.<br />
34. M. Welker, M. Brunke, K. Preussel, I. Lippert and H. von Döhren, “Diversity and distribution<br />
<strong>of</strong> Microcystis (Cyanobacteria) oligopeptide chemotypes from natural communities studied by<br />
single-colony mass spectrometry”, Microbiology, 2004, 150, 1785-1796.<br />
35. M. W. Brunson, C. G. Lutz and R. M. Durborow, “Algae blooms in commercial fish production<br />
ponds”, Publication No. 466, 1994, Southern Regional Aquaculture Center, Stoneville,<br />
Mississippi, USA.<br />
36. P. V. Zimba, C. C. Mischke and S. S. Brashear, “Pond age-water column trophic relationships in<br />
channel catfish Ictalurus punctatus production ponds”, Aquaculture, 2003, 219, 291-301.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
Communication<br />
<strong>Maejo</strong> <strong>Maejo</strong> Int. J. Sci. Int. Technol. J. Sci. Technol. <strong>2012</strong>, 6(01), <strong>2012</strong>, 119-129 6(01), 119-129 119<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Determination <strong>of</strong> production-shipment policy using a two-phase<br />
algebraic approach<br />
Yuan-Shyi Peter Chiu 1 , Hong-Dar Lin 1 and Huei-Hsin Chang 2,*<br />
1<br />
Department <strong>of</strong> Industrial Engineering and Management, Chaoyang University <strong>of</strong> Technology,<br />
Wufong, Taichung 413, Taiwan<br />
2<br />
Department <strong>of</strong> Finance, Chaoyang University <strong>of</strong> Technology, Wufong, Taichung 413, Taiwan<br />
*Corresponding author, e-mail: chs@cyut.edu.tw<br />
Received: 31 March 2011 / Accepted: 31 March <strong>2012</strong> / Published: 7 April <strong>2012</strong><br />
Abstract: The optimal production-shipment policy for end products using mathematical<br />
modeling and a two-phase algebraic approach is investigated. A manufacturing system<br />
with a random defective rate, a rework process, and multiple deliveries is studied with the<br />
purpose <strong>of</strong> deriving the optimal replenishment lot size and shipment policy that minimises<br />
total production-delivery costs. The conventional method uses differential calculus on the<br />
system cost function to determine the economic lot size and optimal number <strong>of</strong> shipments<br />
for such an integrated vendor-buyer system, whereas the proposed two-phase algebraic<br />
approach is a straightforward method that enables practitioners who may not have<br />
sufficient knowledge <strong>of</strong> calculus to manage real-world systems more effectively.<br />
Keywords: manufacturing system, replenishment lot size, delivery, two-phase algebraic<br />
approach, random defective rate<br />
________________________________________________________________________________<br />
INTRODUCTION<br />
With the purpose <strong>of</strong> minimising total set-up and holding costs, inventory controllers in most<br />
companies need to address two basic issues for items they routinely stock: when to start<br />
replenishment and how much to refill. For items made in-house by manufacturing firms, production<br />
planners must, without exception, decide when to initiate a production run and how many items to<br />
produce in a run [1]. An inventory model that uses mathematical techniques to derive the most<br />
economical production lot was first proposed by Taft [2] several decades ago. This is also known as<br />
the economic production quantity (EPQ) model [3].
120 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 119-129<br />
The classic EPQ model assumes a continuous inventory issuing policy to satisfy product<br />
demand. However, in real-world vendor-buyer systems, multiple or periodic deliveries <strong>of</strong> end items<br />
are commonly adopted. Hence, the determination <strong>of</strong> the optimal number <strong>of</strong> shipments for a finished<br />
lot becomes a critical issue to such a vendor-buyer system in terms <strong>of</strong> production-delivery cost<br />
minimisation. Schwarz [4] examined a one-warehouse N-retailer deterministic inventory system with<br />
the objective <strong>of</strong> determining the stock policy that minimises the long-run average system cost per<br />
unit time. He derived optimal solutions along with a few necessary properties for a one-retailer and<br />
N-identical-retailer problems. Heuristic solutions for the general problem were also suggested. Goyal<br />
[5] studied an integrated single-supplier-single-customer problem and presented a method that is<br />
typically applicable to those inventory problems where a product is procured by a single customer<br />
from a single supplier using examples to demonstrate the proposed model. Studies related to various<br />
aspects <strong>of</strong> supply chain optimisation have since been extensively carried out [e.g. 6-13]. The classic<br />
EPQ model also assumes that all items produced are <strong>of</strong> perfect quality. However, in a real-life<br />
manufacturing environment, the generation <strong>of</strong> nonconforming items is almost inevitable due to<br />
process deterioration or various other factors. In the past decades, many studies have attempted to<br />
address the issues <strong>of</strong> defective products and quality assurance in production systems [e.g. 14-24].<br />
Shih [14] extended two inventory models to the case where the proportion <strong>of</strong> defective<br />
units in the accepted lot is a random variable with known probability distribution. Optimal solutions<br />
to the amended systems were developed and comparisons with the traditional models were also<br />
presented via numerical examples. Moinzadeh and Aggarwal [17] studied a production-inventory<br />
system that was subjected to random disruptions. They assumed that the time between breakdowns is<br />
exponential, the restoration times are constant, and excess demand is back-ordered. An (s, S) policy<br />
was proposed and the policy parameters that minimise the expected total cost per unit time were<br />
investigated. A procedure for finding the optimal values <strong>of</strong> the policy was also developed. Makis<br />
[18] investigated the optimal lot sizing and inspection policy for an economic manufacturing quantity<br />
(EMQ) model with imperfect inspections and assumed that the process could be monitored through<br />
inspections and that both the lot size and the inspection schedule were subjected to control. It was<br />
assumed that the in-control periods were generally distributed and the inspections imperfect. Using<br />
Lagrange's method and solving a non-linear equation, a two-dimensional search procedure was<br />
proposed for finding the optimal lot sizing and inspection policy. Rahim and Ben-Daya [19] studied<br />
the simultaneous effects <strong>of</strong> deteriorating product items and deteriorating production processes on the<br />
economic production quantity, inspection schedules, and economic design <strong>of</strong> control charts.<br />
Deterioration times for both product and process were assumed to follow an arbitrary distribution,<br />
and the product quality characteristic was assumed to be normally distributed. Numerical examples<br />
were provided to demonstrate the usage <strong>of</strong> their models. Chiu et al. [21] studied the optimal<br />
replenishment policy for the EMQ model with rework failure, backlogging and random breakdowns.<br />
Mathematical modelling and cost analysis were employed in their study, along with a renewal reward<br />
theorem for dealing with variable cycle length. They derived a long-run average cost function for<br />
their proposed model and proved that it was a convex function. Finally, they obtained an optimal<br />
replenishment policy for such an imperfect EMQ model.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 119-129<br />
121<br />
Recently, an algebraic method <strong>of</strong> determining the economic order quantity (EOQ) model with<br />
backlogging was introduced by Grubbström and Erdem [25]. They used algebraic derivation to find<br />
the optimal order quantity without reference to the first- or second-order derivatives. Similar<br />
methodologies have been applied to solve various aspects <strong>of</strong> supply chain optimisation [26-28]. This<br />
paper extends such an approach in order to re-examine a manufacturing system with a random<br />
defective rate, a rework process, and multiple deliveries <strong>of</strong> its end product [13].<br />
METHODS<br />
We present a two-phase algebraic approach [13] in order to re-examine a manufacturing<br />
system with a random defective rate, a rework process, and multiple shipments <strong>of</strong> finished items.<br />
Such a specific model is described as follows. Assume a production system has an annual production<br />
rate P and randomly produces a proportion x <strong>of</strong> defective items during its uptime at a production rate<br />
d. All manufactured items are screened and the inspection cost is included in the unit production cost<br />
C. Non-conforming products fall into two groups: the scrap (a proportion <strong>of</strong> θ) and the repairable<br />
(1-θ). The rework process has a rate <strong>of</strong> P 1 units per year and commences immediately after regular<br />
production in each cycle. A proportion θ 1 <strong>of</strong> reworked items fails during rework and is treated as<br />
scrap. Under regular supply, the constant production rate P must be larger than the sum <strong>of</strong> the<br />
demand rate λ and the production rate <strong>of</strong> defective items d, i.e. (P-d-λ) > 0, where d can be<br />
expressed as d = Px. Let d 1 denote the production rate <strong>of</strong> scrap during rework; d 1 can then be<br />
expressed as d 1 = P 1 θ 1 . Furthermore, the proposed system considers a multi-delivery policy for the<br />
end items with quality assurance. That is, the finished items can only be delivered to the customers if<br />
the whole lot is quality assured at the end <strong>of</strong> the reworking process. A fixed quantity <strong>of</strong> n<br />
installments <strong>of</strong> the finished batch is delivered to customers at fixed interval <strong>of</strong> time during production<br />
downtime t 3 (Figure 1). Other notations used in the proposed system are listed below.<br />
t 1 = regular production time in the proposed model,<br />
t 2 = time required to rework defective items,<br />
t 3 = time required to deliver all perfect-quality end products,<br />
t n = fixed interval <strong>of</strong> time between each installment <strong>of</strong> finished end products delivered<br />
during production downtime t 3 ,<br />
T = cycle length,<br />
Q = manufacturing batch size __ the decision variable,<br />
n = number <strong>of</strong> fixed-quantity installments <strong>of</strong> finished batch to be delivered to<br />
customers __ the decision variable,<br />
H 1 = maximum level <strong>of</strong> on-hand inventory when regular production ends,<br />
H = maximum level <strong>of</strong> on-hand inventory when the rework process finishes,<br />
I(t) = on-hand inventory <strong>of</strong> perfect quality end items at manufacturer’s end at time t,<br />
TC(Q,n) = total production-inventory-delivery costs per cycle,<br />
K = set-up cost per cycle,<br />
C = unit production cost,<br />
h = unit holding cost,<br />
C R = unit rework cost,<br />
= holding cost for each reworked item,<br />
h 1
122 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 119-129<br />
C S = disposal cost per scrap item,<br />
K 1 = fixed delivery cost per shipment,<br />
C T = delivery cost per item shipped to customers,<br />
φ = overall scrap rate per cycle (sum <strong>of</strong> scrap rates in periods t 1 and t 2 ),<br />
h 2 = holding cost for each item kept by customer,<br />
E[TCU(Q,n)] = long-run average cost per unit time.<br />
Figure 1. On-hand inventory <strong>of</strong> perfect end items in the proposed model with random defective<br />
rate, reworking and multi-delivery policy [13]<br />
With reference to Figure 1, the total production-inventory-delivery cost per cycle, TC(Q, n),<br />
consists <strong>of</strong> the following. (a) set-up cost and variable manufacturing costs per cycle; (b) total quality<br />
costs including variable repairing costs, holding costs for reworked items, and disposal costs for<br />
scrap items per cycle; (c) fixed and variable delivery costs per cycle; (d) total holding costs at the<br />
manufacturer’s end for all items produced in the periods t 1 , t 2 and t 3 ; and (e) total holding costs at the<br />
customer’s end for all items stocked in t 3 :<br />
P1<br />
t2<br />
TC Q, n K CQ CR x 1Qh 1 t2CSxQnK1<br />
2<br />
H1 dt1 H1<br />
H n<br />
1<br />
<br />
(1)<br />
CT Q1<br />
xh<br />
t1 t2 Ht3<br />
2 2 2n<br />
<br />
<br />
<br />
h2<br />
H <br />
t<br />
3<br />
T H <br />
t<br />
3<br />
2 <br />
n<br />
<br />
<br />
With further derivations, the long-run average cost per unit time E[TCU(Q,n)] for the<br />
proposed system can be written as follows (see mathematical modelling section in Chiu et al. [13]):
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 119-129<br />
<br />
, <br />
C<br />
<br />
1 <br />
CT<br />
ET 1<br />
Ex Q1<br />
<br />
Ex<br />
2 2<br />
1<br />
<br />
h1<br />
Ex Q1<br />
<br />
CSEx<br />
1Ex<br />
2P1<br />
1Ex<br />
1Ex<br />
hQ<br />
hQ<br />
2 2<br />
<br />
<br />
<br />
2E x E x E x 1<br />
<br />
<br />
1<br />
<br />
<br />
1 hQ 1 <br />
E x hQ<br />
hQEx1<br />
<br />
E TC Q n K nK<br />
ETCUQ,<br />
n<br />
<br />
C E x<br />
R<br />
<br />
<br />
2P 1 E x 2P 1 E x<br />
<br />
<br />
1 n<br />
<br />
<br />
<br />
2 2P 2P1<br />
<br />
2 2 2<br />
<br />
1<br />
<br />
1hQ 1hQ<br />
hQ 1E x<br />
1 Ex<br />
1 1<br />
<br />
n 2 n 2P 2 n<br />
P1<br />
<br />
<br />
<br />
123<br />
(2)<br />
Derivation <strong>of</strong> Optimal Policy using Two-phase Algebraic Approach<br />
Unlike the conventional method, which uses differential calculus on the cost function<br />
E[TCU(Q, n)] to find the optimal point [13], a straightforward two-phase algebraic approach to<br />
determining the optimal production-shipment policy for the proposed model is adopted here.<br />
Phase 1: Derivation <strong>of</strong> n*<br />
It can be seen that Eq.2 has two decision variables, namely Q and n. Moreover, there are<br />
several different forms <strong>of</strong> these decision variables in the right-hand side <strong>of</strong> Eq.2, e.g., Q, Q -1 , nQ -1<br />
and Qn -1 . Therefore, Eq.2 can be rearranged as<br />
or<br />
<br />
<br />
<br />
1<br />
<br />
<br />
1<br />
<br />
<br />
<br />
<br />
C<br />
CRE x CSE x<br />
ETCUQ,<br />
n<br />
CT<br />
<br />
1 E x 1E x 1E x<br />
2<br />
<br />
2<br />
<br />
h1<br />
E x h h<br />
<br />
<br />
2 2 <br />
<br />
<br />
<br />
2E x E x E x 1<br />
<br />
<br />
Q<br />
1 E x 2P1 2P 2P<br />
<br />
<br />
<br />
<br />
<br />
1<br />
<br />
<br />
<br />
<br />
<br />
K<br />
<br />
Q<br />
P P1<br />
<br />
1<br />
Ex<br />
h 1 E x E x 1 <br />
<br />
hh2<br />
<br />
<br />
2 2 2<br />
<br />
K1<br />
1<br />
<br />
nQ<br />
1 E x<br />
<br />
1<br />
Ex Ex1<br />
<br />
<br />
<br />
+<br />
<br />
<br />
<br />
<br />
<br />
1<br />
h h2<br />
n Q<br />
2 2P<br />
2P1<br />
<br />
<br />
1 1 1<br />
, <br />
1 2 3 4 5<br />
<br />
<br />
ETCU Q n Q Q nQ n Q<br />
(4)<br />
<br />
Q<br />
1<br />
(3)<br />
where α 1 , α 2 , α 3 , α 4 and α 5 denote:<br />
<br />
1<br />
C<br />
CREx1<br />
CSE x<br />
CT<br />
<br />
(5)<br />
1Ex<br />
1Ex<br />
1Ex
124 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 119-129<br />
<br />
<br />
2<br />
3<br />
<br />
4<br />
<br />
<br />
<br />
<br />
<br />
1<br />
E x<br />
<br />
1<br />
<br />
<br />
2 2<br />
h1<br />
E x h h<br />
<br />
<br />
2P1 2P 2P<br />
<br />
1<br />
<br />
<br />
h 1E x E x 1 <br />
<br />
hh2<br />
<br />
<br />
<br />
<br />
2 2P<br />
2P1<br />
<br />
K<br />
<br />
2 2<br />
2Ex<br />
Ex Ex<br />
1<br />
<br />
<br />
1<br />
Ex<br />
(7)<br />
5<br />
K <br />
1<br />
1<br />
Ex<br />
(8)<br />
1<br />
Ex Ex1<br />
<br />
hh<br />
<br />
<br />
(9)<br />
2<br />
<br />
2 2P<br />
2P<br />
With further rearrangements, Eq.4 becomes<br />
1<br />
<br />
<br />
<br />
<br />
<br />
(6)<br />
1 2 1 1<br />
ETCUQ,<br />
n 2<br />
1<br />
Q 2 Q 3 n Q 4<br />
nQ<br />
<br />
<br />
5<br />
<br />
<br />
2 2<br />
1 1 1<br />
, <br />
1 2 3 4 <br />
ETCU Q n Q Q n Q nQ <br />
<br />
5<br />
<br />
2 2<br />
<br />
2 3 4 5<br />
Eq.11 will be minimised if its second and third terms in it equal zero. That is:<br />
(10)<br />
(11)<br />
Q<br />
n<br />
<br />
3<br />
(12)<br />
2<br />
<br />
5<br />
Q<br />
(13)<br />
4<br />
Substituting Eq.6 and 7 into Eq.12, and then substituting Eq.8, 9 and 12 into Eq.13, the optimal<br />
number <strong>of</strong> shipments n* is<br />
n <br />
<br />
E x 1 <br />
K<br />
hh2<br />
1<br />
Ex <br />
5<br />
P P1 K<br />
3<br />
1<br />
<br />
<br />
2 2<br />
4 2<br />
<br />
h1<br />
Ex 1<br />
<br />
h h<br />
<br />
2 2<br />
<br />
<br />
<br />
<br />
2E x E x <br />
E x 1<br />
1 E x P1 P P <br />
<br />
1<br />
<br />
<br />
<br />
<br />
<br />
<br />
E x<br />
h1<br />
Exhh2<br />
<br />
<br />
<br />
P<br />
<br />
<br />
1 <br />
<br />
P1<br />
<br />
<br />
(14)<br />
It is noted that Eq.14 is identical to that obtained using the conventional differential calculus<br />
method [13]. We can also see that although in real-world situation the number <strong>of</strong> deliveries takes<br />
integer values only, Eq. 14 results in a real number. In order to locate the integer value <strong>of</strong> n* that<br />
minimises the long-run average cost for the proposed system, the two adjacent integers to n must be<br />
examined respectively for cost minimisation [11]. Let n + denote the smallest integer greater than or
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 119-129<br />
125<br />
equal to n (derived from Eq. 14) and n - denote the largest integer less than or equal to n. Because n*<br />
is either n + or n - , we can first treat E[TCU(Q,n)] (Eq. 4) as a cost function with a single-decision<br />
variable Q, and perform the following rearrangements.<br />
Phase 2: Searching for Q*<br />
First, the long-run cost function E[TCU(Q, n)] (i.e. Eq.4) can be rearranged as the following<br />
single-decision-variable function:<br />
where α 6 and α 7 denote:<br />
With further rearrangements, Eq.15 becomes<br />
1<br />
, <br />
<br />
E TCU Q n Q Q<br />
1 6<br />
<br />
7<br />
(15)<br />
<br />
(16)<br />
1<br />
6<br />
2 n 5<br />
n<br />
(17)<br />
7 3 4<br />
<br />
<br />
2<br />
1<br />
<br />
<br />
1 6 7 6 7<br />
ETCU Q, n Q <br />
Q<br />
<br />
2<br />
<br />
(18)<br />
Upon derivation <strong>of</strong> Eq.18, it can be noted that E[TCU(Q,n)] will be minimised if the second term in<br />
it equals zero. That is:<br />
Q*<br />
<br />
7<br />
(19)<br />
<br />
6<br />
Substituting Eq.16 and 17 into Eq. 19, the optimal production lot size is<br />
Q*<br />
<br />
<br />
1<br />
<br />
<br />
2 2<br />
1<br />
2 2 <br />
2<br />
<br />
2Ex<br />
Ex Ex<br />
1 <br />
1 h1<br />
Ex<br />
<br />
1<br />
<br />
2 1<br />
Exh h2<br />
h2<br />
Ex<br />
n<br />
P P1<br />
n<br />
2<br />
K nK<br />
h E x 1 <br />
h h<br />
1 <br />
<br />
P1 P P <br />
<br />
1<br />
n <br />
<br />
<br />
<br />
E x 1 <br />
1 1 1<br />
<br />
<br />
<br />
(20)<br />
It is noted that Eq.20 is identical to that obtained using the conventional differential calculus<br />
method [13].<br />
To find the optimal production-shipment (Q*, n*) policy, we substitute all related system<br />
parameters, along with n + and n - , into Eq.20. Then, applying the resulting (Q, n + ) and (Q, n - )<br />
respectively in Eq. 4, we choose the one that gives the minimum long-run average cost as the optimal<br />
production-shipment policy (Q*, n*). A numerical example to demonstrate the practical usage <strong>of</strong> this<br />
method is provided in the next section.
126 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 119-129<br />
NUMERICAL EXAMPLE<br />
The aforementioned two-phase algebraic approach and its resulting Eq.14, 20 and 4 are<br />
verified in this section using the same numerical example [13]. Suppose an end product can be<br />
produced at a rate <strong>of</strong> 60,000 units per year, its annual demand being estimated to be 3,400 units, and<br />
during the production process there is a random defective rate x that follows a uniform distribution<br />
over a range <strong>of</strong> [0, 0.3]. A proportion θ = 0.1 <strong>of</strong> the imperfect items is considered to be scrap and<br />
the other portion is assumed to be repairable with a rework rate <strong>of</strong> P 1 = 2,100 units per year. It is<br />
further estimated that there is a proportion θ 1 = 0.1 <strong>of</strong> reworked items that fail (become scrap) during<br />
the rework period. As a quality assurance policy, the finished items can only be delivered to<br />
customers if the whole lot is quality-assured after reworking. A fixed quantity <strong>of</strong> n installments <strong>of</strong> the<br />
perfect end items are shipped to customers at a fixed interval <strong>of</strong> time during delivery time t 3 (Figure<br />
1). Other selected parameter values in this example are as follows:<br />
C = $100 per item,<br />
C R = $60 for each reworked item,<br />
C S = $20 for each scrap item,<br />
K = $20,000 per production run,<br />
h = $20 per item per year,<br />
h 1 = $40 per reworked item per unit time,<br />
K 1 = $2,400 per shipment,<br />
C T = $0.1 per item delivered,<br />
h 2 = $80 per item kept at the customer’s end per unit time.<br />
Applying Eq.14, we obtain n=2.736. Because the number <strong>of</strong> deliveries has to be an integer,<br />
we have n + =3 and n - =2. Substituting all system parameters, along with n + and n - respectively, into<br />
Eq.20, we find two possible policies, namely (Q, n + )=(1735, 3) and (Q, n - )=(1579, 2). We then apply<br />
(Q, n + ) and (Q, n - ) in Eq.4 to obtain E[TCU(1735,3)]=$485,541 and E[TCU(1579,2)]=$487,071.<br />
Selecting that with the minimum cost, we find that the optimal policy (Q*, n*)=(1735, 3) and<br />
the long-run average cost E[TCU(Q*, n*)]=$485,541. The results are noted to be identical to those<br />
obtained by Chiu et al. [13].<br />
The effect <strong>of</strong> varying the lot-size Q on the long-run average cost function E[TCU(Q, n)] and<br />
on the components <strong>of</strong> E[TCU(Q, n)], for n* = 3, is depicted in Figure 2.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 119-129<br />
127<br />
Figure 2. Effect <strong>of</strong> varying lot size Q on the long-run average cost function<br />
E[TCU(Q, n)] and on the components <strong>of</strong> E[TCU(Q, n)] for n* = 3<br />
CONCLUSIONS<br />
This paper proposes a two-phase algebraic approach to determining the optimal productionshipment<br />
policy for an end product in an integrated supplier-customer system with quality assurance.<br />
Unlike the conventional method, which uses differential calculus on the system cost function<br />
to find the economic lot size and optimal number <strong>of</strong> deliveries, the proposed two-phase algebraic<br />
approach is a straightforward method that may enable practitioners with little or no knowledge <strong>of</strong><br />
differential calculus to understand and manage real-world systems more effectively. The research<br />
results were confirmed to be identical to those obtained by the traditional method<br />
ACKNOWLEDGEMENTS<br />
The authors would like to express their appreciation <strong>of</strong> the support <strong>of</strong> this research project by<br />
the National <strong>Science</strong> Council <strong>of</strong> Taiwan under grant number: NSC 99-2221-E-324-017.<br />
REFERENCES<br />
1. F. S. Hillier and G. J. Lieberman, “Introduction to Operations Research”, McGraw Hill, New<br />
York, 2001.<br />
2. E. W. Taft, “The most economical production lot”, Iron Age, 1918, 101, 1410-1412.<br />
3. S. Nahmias, “Production and Operations Analysis”, McGraw Hill, New York, 2005, pp.183-<br />
205.<br />
4. L. B. Schwarz, “A simple continuous review deterministic one-warehouse N-retailer inventory<br />
problem”, Manage. Sci., 1973, 19, 555-566.<br />
5. S. K. Goyal, “An integrated inventory model for a single supplier-single customer problem”, Int.<br />
J. Prod. Res., 1977, 15, 107-111.
128 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 119-129<br />
6. L. B. Schwarz, B. L. Deuermeyer and R. D. Badinelli, “Fill-rate optimization in a onewarehouse<br />
N-identical retailer distribution system”, Manage. Sci., 1985, 31, 488-498.<br />
7. A. Banerjee and S. Banerjee, “Coordinated order-less inventory replenishment for a vendor and<br />
multiple buyers”, Int. J. Tech. Manage., 1992, 7, 328-336.<br />
8. R. M. Hill, “The optimal production and shipment policy for the single-vendor singlebuyer<br />
integrated production-inventory problem”, Int. J. Prod. Res., 1999, 37, 2463-2475.<br />
9. D. J. Thomas and S. T. Hackman, “A committed delivery strategy with fixed frequency and<br />
quantity”, Eur. J. Oper. Res., 2003, 148, 363-373.<br />
10. K. Ertogral, M. Darwish and M. Ben-Daya, “Production and shipment lot sizing in a vendorbuyer<br />
supply chain with transportation cost”, Eur. J. Oper. Res., 2007, 176, 1592-1606.<br />
11. Y-S. P. Chiu, F-T. Cheng and H-H. Chang, “Remarks on optimization process <strong>of</strong><br />
manufacturing system with stochastic breakdown and rework”, Appl. Math. Lett., 2010, 10,<br />
1152-1155.<br />
12. C-C. Chern and I-C. Yang, “A heuristic master planning algorithm for supply chains that<br />
consider substitutions and commonalities”, Expert Syst. Appl., 2011, 38, 14918-14934.<br />
13. S. W. Chiu, H-D. Lin, M-F. Wu and J-C. Yang, “Determining replenishment lot size and<br />
shipment policy for an extended EPQ model with delivery and quality assurance issues”, Sci.<br />
Iran., 2011, 18, 1537-1544.<br />
14. W. Shih, “Optimal inventory policies when stockouts result from defective products”, Int. J.<br />
Prod. Res., 1980, 18, 677-686.<br />
15. M. J. Rosenblatt and H. L. Lee, “Economic production cycles with imperfect production<br />
processes”, IIE Trans., 1986, 18, 48-55.<br />
16. X. Zhang and Y. Gerchak, “Joint lot sizing and inspection policy in and EOQ model with<br />
random yield”, IIE Trans., 1990, 22, 41-47.<br />
17. K. Moinzadeh and P. Aggarwal, “Analysis <strong>of</strong> a production/inventory system subject to random<br />
disruptions”, Manage. Sci., 1997, 43, 1577-1588.<br />
18. V. Makis, “Optimal lot sizing and inspection policy for an EMQ model with imperfect<br />
inspections”, Naval Res. Log., 1998, 45, 165-186.<br />
19. M. A. Rahim and M. Ben-Daya, “Joint determination <strong>of</strong> production quantity, inspection<br />
schedule, and quality control for an imperfect process with deteriorating products”, J. Oper.<br />
Res. Soc., 2001, 52, 1370-1378.<br />
20. S. W. Chiu, D-C. Gong and H-M. Wee, “Effects <strong>of</strong> random defective rate and imperfect rework<br />
process on economic production quantity model”, Japan J. Ind. Appl. Math., 2004, 21, 375-<br />
389.<br />
21. S. W. Chiu, K-K. Chen and J-C. Yang, “Optimal replenishment policy for manufacturing<br />
systems with failure in rework, backlogging and random breakdown”, Math. Comput. Model.<br />
Dyn. Sys., 2009, 15, 255-274.<br />
22. A. Amirteimoori and A. Emrouznejad, “Input/output deterioration in production processes”,<br />
Expert Syst. Appl., 2011, 38, 5822-5825.<br />
23. Y-S. P. Chiu, H-D. Lin and H-H. Chang, “Mathematical modeling for solving manufacturing<br />
run time problem with defective rate and random machine breakdown”, Comput. Ind. Eng.,<br />
2011, 60, 576-584.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 119-129<br />
129<br />
24. B. Sarkar, S. S. Sana and K. Chaudhuri, “An imperfect production process for time varying<br />
demand with inflation and time value <strong>of</strong> money - An EMQ model”, Expert Syst. Appl., 2011, 38,<br />
13543-13548.<br />
25. R. W. Grubbström and A. Erdem, “The EOQ with backlogging derived without derivatives”,<br />
Int. J. Prod. Econ., 1999, 59, 529-530.<br />
26. S. W. Chiu, “Production lot size problem with failure in repair and backlogging derived without<br />
derivatives”, Eur. J. Oper. Res., 2008, 188, 610-615.<br />
27. H-D. Lin, Y-S. P. Chiu and C-K. Ting, “A note on optimal replenishment policy for imperfect<br />
quality EMQ model with rework and backlogging”, Comput. Math. Appl., 2008, 56, 2819-<br />
2824.<br />
28. Y-Y. Xiao, R-Q. Zhang and I. Kaku, “A new approach <strong>of</strong> inventory classification based on loss<br />
pr<strong>of</strong>it”, Expert Syst. Appl., 2011, 38, 9382-9391.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
130 <strong>Maejo</strong> <strong>Maejo</strong> Int. J. Int. Sci. J. Technol. Sci. Technol. <strong>2012</strong>, <strong>2012</strong>, 6(01), 6(01), 130-151 130-151<br />
Full Paper<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
IIS-Mine: A new efficient method for mining frequent itemsets<br />
Supatra Sahaphong * and Veera Boonjing<br />
Department <strong>of</strong> Mathematics and Computer <strong>Science</strong>, Faculty <strong>of</strong> <strong>Science</strong>, King Mongkut’s Institute <strong>of</strong><br />
Technology Ladkrabang, Bangkok 10520, Thailand<br />
* Corresponding author, e-mail: supatra@ru.ac.th<br />
Received: 11 September 2011 / Accepted: 4 March <strong>2012</strong> / Published: 23 April <strong>2012</strong><br />
Abstract: A new approach to mine all frequent itemsets from a transaction database is<br />
proposed. The main features <strong>of</strong> this paper are as follows: (1) the proposed algorithm<br />
performs database scanning only once to construct a data structure called an inverted<br />
index structure (IIS); (2) the change in the minimum support threshold is not affected by<br />
this structure, and as a result, a rescan <strong>of</strong> the database is not required; and (3) the<br />
proposed mining algorithm, IIS-Mine, uses an efficient property <strong>of</strong> an extendable itemset,<br />
which reduces the recursiveness <strong>of</strong> mining steps without generating candidate itemsets,<br />
allowing frequent itemsets to be found quickly. We have provided definitions, examples,<br />
and a theorem, the completeness and correctness <strong>of</strong> which is shown by mathematical<br />
pro<strong>of</strong>. We present experiments in which the run time, memory consumption and scalability<br />
are tested in comparison with a frequent-pattern (FP) growth algorithm when the<br />
minimum support threshold is varied. Both algorithms are evaluated by applying them to<br />
synthetics and real-world datasets. The experimental results demonstrate that IIS-Mine<br />
provides better performance than FP-growth in terms <strong>of</strong> run time and space consumption<br />
and is effective when used on dense datasets.<br />
Keywords: association rule mining, data mining, frequent itemsets mining, frequent<br />
patterns mining, knowledge discovering<br />
________________________________________________________________________________<br />
INTRODUCTION<br />
The objective <strong>of</strong> frequent itemset mining is to identify all frequently occurring itemsets using<br />
a support threshold. Decision-makers are interested in all itemsets associated with high frequencies.<br />
Association rule mining algorithms can be broken down into two major phases. The first phase finds<br />
all <strong>of</strong> the itemsets that satisfy the minimum support threshold, which are the frequent itemsets. The
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
131<br />
second phase is rule generation, in which all the high confidence rules from the frequent itemsets<br />
found in the previous phase are extracted [1]. Many previous investigations focused on the first<br />
phase. Early algorithms based on generated and tested candidate itemsets have two major defects.<br />
First, the database must be scanned multiple times to generate candidate itemsets, which increases<br />
the I/O load and is time-consuming. The search space <strong>of</strong> itemsets that must be explored grows<br />
exponentially. Second, enormous candidate itemsets are generated and calculated from their<br />
supports, which consumes a large amount <strong>of</strong> CPU time [2].<br />
To overcome the above-mentioned problems, a next generation <strong>of</strong> algorithms using a<br />
compact tree structure was proposed, called a frequent-pattern (FP) tree [3], which finds frequent<br />
itemsets directly from the data structure. However, most <strong>of</strong> the FP-tree-based algorithms have the<br />
following weaknesses. First, the mining <strong>of</strong> frequent itemsets from the FP-tree to generate a huge<br />
conditional FP-tree requires a large amount <strong>of</strong> run time and space. The best case is when a database<br />
has the same set <strong>of</strong> transactions; an FP-tree then contains only a single branch <strong>of</strong> nodes. The worst<br />
case is when a database has a unique set <strong>of</strong> transactions [3]. Second, when the users change to a new<br />
minimum support threshold for their new decision, the algorithm restarts the whole operation and<br />
scans the database twice.<br />
Many researchers have tried to solve the above problems using a vertical data layout.<br />
However, most <strong>of</strong> the algorithms have the drawback <strong>of</strong> increasing the run time and space<br />
consumption due to the following reasons. First, when the users change to a new minimum support<br />
threshold for their new decision, the algorithm restarts the whole operation more than one time to<br />
scan a database and construct their data structure. Rescanning the database for a new minimum<br />
support threshold wastes both run time and space. Second, all <strong>of</strong> the FP-tree-based algorithms<br />
generate a huge conditional FP-tree, which has a large number <strong>of</strong> recursive processing steps and<br />
requires a large amount <strong>of</strong> run time and space consumption.<br />
In this paper we present a new, efficient method to solve the above-mentioned problems by<br />
proposing both a data structure and a mining algorithm for decreasing the consumption <strong>of</strong> run time<br />
and space. First, the proposed method performs database scanning to construct a data structure<br />
called an inverted index structure (IIS) only once. In addition, changing the minimum support<br />
threshold does not affect the IIS; therefore, database rescanning is not required. Second, IIS-Mine is<br />
a new algorithm that mines all <strong>of</strong> the frequent itemsets without generating candidate itemsets and<br />
uses a new tree structure called the IIS item Tree. IIS-Mine employs an efficient property <strong>of</strong> the<br />
extendable itemset, which decreases the number <strong>of</strong> recursive processing steps when mining frequent<br />
itemsets. The completeness and correctness <strong>of</strong> the algorithm is proved using a mathematical pro<strong>of</strong>.<br />
Last, the efficiency <strong>of</strong> IIS-Mine is compared with that <strong>of</strong> FP-growth in terms <strong>of</strong> run time and space<br />
consumption through simulation experiments. Our experiments show that IIS-Mine is more efficient<br />
than FP-growth in run time and space consumption for dense datasets.<br />
RELATED WORK<br />
The first algorithm to generate all frequent itemsets is the AIS algorithm, which was first<br />
introduced by Agrawal et al [4]. However, this algorithm constructs a list <strong>of</strong> all <strong>of</strong> the possible
132 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
itemsets at each level <strong>of</strong> traversal, so infrequent itemsets that are not needed are also generated.<br />
Later, the algorithm was improved upon and renamed the Apriori algorithm by Agrawal et al [5]. The<br />
Apriori algorithm uses a level-wise and breadth-first search approach for generating association<br />
rules. Many efficient association mining techniques have been developed based on the Apriori<br />
algorithm. Vu et al. [6] proposed a rule-based location prediction technique to predict the user’s<br />
featured location, but this proposal generates more candidate itemsets than are required. These<br />
algorithms are also expensive in terms <strong>of</strong> I/O load and run time when the database must be scanned<br />
multiple times to generate candidate itemsets.<br />
The above-mentioned problems can be improved upon by using a compact tree structure and<br />
finding frequent itemsets directly from the data structure. The algorithms scan a database twice. The<br />
first scan <strong>of</strong> the database is to discard infrequent itemsets; the second is to construct a tree. The FPgrowth<br />
algorithm, developed by Han et al. [7], is the most popular method. It performs a depth-first<br />
search approach in a search space. It encodes a dataset using a compact data structure called an FPtree<br />
or prefix tree and extracts frequent patterns directly from the FP-tree. Many approaches have<br />
been proposed to extend and improve upon this algorithm. Pei et al. [8] developed the H-mine<br />
algorithm using array- and tree-based data structures to improve the main memory cost. The<br />
PatriciaMine algorithm [9] compressed Patricia tries to store datasets, which is space efficient for<br />
both dense and sparse datasets. The FP-growth algorithm [10] reduces the FP-tree traversal time<br />
using an array technique. Zhu [11] proposed a new method to compress a large database into an FPtree<br />
with a children table but not a header table, and applied a depth-first search with this tree for the<br />
mining step, which reduces both the run time and the space consumption. Sahaphong and Boonjing<br />
[12] proposed a new algorithm which constructs a pattern base using a new method that is different<br />
from the pattern base in the FP-growth and mined frequent itemsets using a new combination method<br />
without the recursive construction <strong>of</strong> a conditional FP-tree. An approach based on the FP-tree and<br />
co-occurrence frequent items (COFI) was proposed to find frequent items in multilevel concept<br />
hierarchy by using a non-recursive mining process [13]. A new data structure called improved FP<br />
tree was proposed, which can reduce space consumption and enhance the efficiency <strong>of</strong> an attribute<br />
reduction algorithm [14]. To maintain the anti-monotone property <strong>of</strong> approximate weighted frequent<br />
patterns, a robust concept was proposed to relax the requirement for exact equality between the<br />
weighted supports <strong>of</strong> patterns and a minimum threshold [15]. However, most <strong>of</strong> the FP-tree-based<br />
algorithms require a large amount <strong>of</strong> run time and space to generate the huge conditional FP-trees.<br />
Moreover, the algorithm restarts the whole operation and requires that a database be scanned twice<br />
when the minimum support threshold is changed.<br />
As mentioned above, most <strong>of</strong> the algorithms that mine frequent itemsets use a horizontal data<br />
layout. However, many researchers use a vertical data layout. The Eclat algorithm was proposed<br />
[16] to generate all frequent itemsets in a breadth-first search using the joining step from the Apriori<br />
property when no candidate items can be found. The Eclat algorithm is very efficient for large<br />
itemsets but is less efficient for small ones. The diffset technique [17] was introduced to improve the<br />
memory requirement. Chai et al. [18] detailed a data structure called large-item bipartite graph to<br />
accommodate the data when a database is scanned. Similar to the FP-growth algorithm, this method
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
133<br />
mines frequent patterns using the recursive conditional FP-tree. The BitTableFI algorithm [19] uses<br />
horizontal and vertical data layouts to compress a database. Yen [20] presented an algorithm based<br />
on an undirected itemset graph that finds frequent itemsets by searching undirected graphs. When the<br />
database and minimum support change, this algorithm requires that the graph structure be researched<br />
to generate new frequent itemsets. The Index-BitTableFI [21] was developed to reduce the<br />
cost <strong>of</strong> candidate generation and to support counting. Sahaphong and Boonjing [22] proposed a new<br />
algorithm that reduces the run time. The drawback <strong>of</strong> this algorithm is its large memory consumption<br />
from generation <strong>of</strong> many repeated nodes. The JoinFI-Mine algorithm [23] uses a sorted-list structure<br />
constructed from the vertical data layout and finds all frequent itemsets using a depth-first search for<br />
joining frequent itemsets. Therefore, this algorithm consumes time and space in its joining step.<br />
METHODS<br />
Frequent-Itemsets Mining Problem<br />
We introduce the basic concepts <strong>of</strong> mining frequent itemsets. All terminologies in this section<br />
are proposed by Han et al [2].<br />
Let I = { x 1,<br />
x 2 ,..., x m}<br />
be a set <strong>of</strong> items and DB = { T1 , T2<br />
,..., Tn}<br />
be a transaction database, where<br />
T 1 , T2<br />
,..., T n are transactions that contain items in I. The support, or supp (occurrence frequency), <strong>of</strong> a<br />
pattern A, where A is a set <strong>of</strong> items, is the number <strong>of</strong> transactions containing A in DB. A pattern A is<br />
frequent if A’s support is no less than a predefined minimum support threshold, minsup.<br />
Given a DB and a minimum support threshold minsup, the problem <strong>of</strong> finding a complete set<br />
<strong>of</strong> frequent itemsets is called the frequent-itemsets mining problem.<br />
For a greater understanding, we provide an example to illustrate the above definitions.<br />
Example 1. An example <strong>of</strong> the database by Han et al. [2] is used here. Table 1 is a DB. It<br />
consists <strong>of</strong> 5 transactions (T 1 , T 2 , T 3 , T 4 , and T 5 ) and 17 items (a, b, c, d, e, f, g, h, i, j, k, l, m, n, o,<br />
p, and s). For example, the first transaction is T 1 , which contains f, a, c, d, g, i, m, and p.<br />
Table 1. A transaction database<br />
Transaction<br />
T 1<br />
T 2<br />
T 3<br />
T 4<br />
T 5<br />
Item<br />
f, a, c, d, g, i, m, p<br />
a, b, c, f, l, m, o<br />
b, f, h, j, o<br />
b, c, k, s, p<br />
a, f, c, e, l, p, m, n<br />
IIS: Design and Construction<br />
We present a data structure that contains transaction data called an inverted index structure<br />
(IIS). The IIS is a structure that holds a relationship between items and the transactions included<br />
within. The IIS is constructed from one scan <strong>of</strong> the DB. This original IIS can support every minsup;
134 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
therefore, it does not need to rescan the DB when the minsup is changed. According to the<br />
definitions in the previous section, we present a new definition with an example and an algorithm to<br />
demonstrate how to construct this IIS.<br />
Definition 1 (IIS). Let DB be a transaction database and I be a non-empty finite set <strong>of</strong> all<br />
items in each transaction in the DB, where each transaction is a set <strong>of</strong> items in I associated with an<br />
identifier, and let S be a set <strong>of</strong> all non-empty subsets <strong>of</strong> DB. An IIS is the function f : I → S defined<br />
by f ( a)<br />
= S1<br />
if T contains a for eachT<br />
∈S1<br />
. This function can be identified as a table consisting <strong>of</strong> the<br />
attributes <strong>of</strong> the items in I and the corresponding transactions in the DB. That is, each row in the IIS<br />
contains an item in I as well as the transactions in the DB that contain that item. The set <strong>of</strong><br />
transactions are written in the order <strong>of</strong> their ascending identification numbers.<br />
With the above definition, the IIS represents the relationship between each item in I and its<br />
corresponding transactions; therefore, the IIS can apply to all minimum support thresholds, and a<br />
rescan <strong>of</strong> the database is not required. We demonstrate the steps to construct the IIS through the<br />
following example.<br />
Example 2. We use the example <strong>of</strong> a DB in Table 1. The DB is scanned once to create the<br />
IIS. The scan <strong>of</strong> the first transaction is T 1 , which consists <strong>of</strong> items f, a, c, d, g, i, m and p. The<br />
transaction T 1 will be inserted for each corresponding item sorted in ascending order (a, c, d, f, g, i,<br />
m, p). T 1 will be the first transaction inserted in the transactions <strong>of</strong> item a. The second examined item<br />
is c, so we insert T 1 in item c. Next, we examine item d; we then subsequently insert T 1 in item d. The<br />
remaining items (f, g, i, m and p) in T 1 can be similarly inserted. The remaining transactions (T 2 , T 3 ,<br />
T 4 and T 5 ) in the DB are performed in a similar manner.<br />
Algorithm 1 shows how to construct the IIS. Figure 1 shows all <strong>of</strong> the items <strong>of</strong> the IIS after<br />
scanning the DB once, and the bold items are all frequent items that have a support greater than or<br />
equal to the minsup, which is assumed to be 3.<br />
Algorithm 1 (IIS construction)<br />
Input: DB.<br />
Output: IIS.<br />
Method: The IIS is constructed as follows.<br />
1 Begin<br />
2 Create header that contains all items.<br />
3 For each transaction T in DB do // scanning DB once<br />
4 Sort items in T // ascending order<br />
5 Create transaction to each corresponding item<br />
6 End //For<br />
7 End //Begin
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
135<br />
Item Transaction<br />
a T 1 T 2 T 5<br />
b T 2 T 3 T 4<br />
c T 1 T 2 T 4 T 5<br />
d T 1<br />
e T 5<br />
f T 1 T 2 T 3 T 5<br />
g T 1<br />
h T 3<br />
i T 1<br />
j T 3<br />
k T 4<br />
l T 2 T 5<br />
m T 1 T 2 T 5<br />
n T 5<br />
o T 2 T 3<br />
p T 1 T 4 T 5<br />
s T 4<br />
Figure 1. An example <strong>of</strong> the IIS<br />
IIS-Mine Algorithm<br />
We present a new algorithm called IIS-Mine. This algorithm uses a new tree structure, called<br />
the IIS item Tree, to mine frequent itemsets. The main features <strong>of</strong> this algorithm are as follows: (1)<br />
every frequent itemset is found without generating candidate itemsets; (2) the algorithm reduces the<br />
recursion <strong>of</strong> mining steps using the property <strong>of</strong> extendable itemset; and (3) the algorithm supports<br />
the mining <strong>of</strong> frequent itemsets with any value <strong>of</strong> the minsup without needing to rescan the database.<br />
From the above features, we can quickly find the frequent itemsets and completely and correctly<br />
obtain them. We now introduce the terminologies <strong>of</strong> the IIS item Tree, its construct, the theorem, the<br />
examples and the algorithms to describe how to mine frequent itemsets.<br />
Definition 2 (Itemset-tree structure). An itemset-tree structure is a tree structure<br />
constructed from the IIS. It is a finite set <strong>of</strong> one or more nodes with the following structure:<br />
(i) It consists <strong>of</strong> the root which contains an item, a set <strong>of</strong> item subtrees as the children <strong>of</strong> the root,<br />
and a set <strong>of</strong> header tables.<br />
(ii) Each node in this tree comprises five fields: item-name, which registers which item this node<br />
represents; support, which registers the number <strong>of</strong> transactions represented by the portion <strong>of</strong> the path<br />
reaching this node; same-item, which represents a pointer that points to the node in the itemset-tree<br />
structure that carries the same item-name; parent, which represents a pointer that points to the<br />
previous node in the same path; and child, which represents a pointer that points to the child node.<br />
(iii) Each member <strong>of</strong> the header table consists <strong>of</strong> two fields, item-name and head <strong>of</strong> node link,<br />
where head <strong>of</strong> node link represents a pointer that points to the first node in the itemset-tree structure<br />
carrying the item-name.
136 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
Definition 3 (IIS item tree). Let x 0 be an arbitrary frequent item in a given transaction database<br />
and IIS be the inverted index structure <strong>of</strong> the transaction database. A tree T constructed from the<br />
IIS is called an inverted index structure- x 0 tree, denoted by IIS x tree , if it satisfies the following:<br />
0<br />
(i) Each node <strong>of</strong> T is <strong>of</strong> the form ( A : s),<br />
where A is a frequent itemset and s is its support. If<br />
( A : s)<br />
is a node <strong>of</strong> T and A = { a},<br />
where a is a frequent item, then ( A : s)<br />
is simply written by ( a : s).<br />
(ii) Let ( x 0 : s0<br />
) be its root, where s 0 is the support <strong>of</strong> x 0 .<br />
(iii) Let x 0 , x1,...,<br />
xk<br />
be frequent items in the IIS, and s i = supp { x 0 , x1,...,<br />
xi}<br />
for all i = 0,1,..., k.<br />
In<br />
this case, P = (( x0 : s0<br />
), ( x1<br />
: s1<br />
),...,( x k : sk<br />
)) is a path from the root ( x 0 : s0<br />
) to a leaf ( x k : sk<br />
) <strong>of</strong> the tree T<br />
if and only if s 0 ≥s<br />
1 ... ≥s<br />
k > 0 , x 0 < l x1<br />
< l ... < l xk<br />
, where <br />
l<br />
is the lexicographic order. If a is a<br />
frequent item in the IIS and if ( a : s)<br />
is a node <strong>of</strong> T such that x k < a or x i < l a < l xi+<br />
1 for all i = 0,1,..., k,<br />
then supp{ x0 , x1,...,<br />
xi , a} = 0 and ( a : s)<br />
is not a node <strong>of</strong> P.<br />
The header table <strong>of</strong> IIS x0<br />
tree is a set <strong>of</strong> all frequent items a <strong>of</strong> a node ( a : s)<br />
<strong>of</strong> this tree.<br />
Based on the above definition, we have the IIS item Tree construction algorithm, as shown in<br />
Algorithm 2. It is evident that if minsup>0 is a minimum support threshold, then every frequent<br />
itemset can be derived from an IIS item Tree.<br />
Algorithm 2 (IIS item Tree construction)<br />
Input: IIS.<br />
Output: IIS item Tree.<br />
Method: An IIS item Tree is constructed as follows.<br />
1Begin<br />
2 Create header table<br />
3 Read frequent item x in IIS<br />
4 Create root R and initial supp(R) to 1<br />
5 Link R to header table<br />
6 For each transaction T <strong>of</strong> root R where T = T1<br />
to Tn<br />
do<br />
7 While next frequent item≠(last frequent item)+1 do<br />
8 Read next frequent item (N) that has same T with R<br />
9 Call InsertTree (N,R)<br />
10 End//While<br />
11 End//For<br />
12End //Begin<br />
Procedure InsertTree (N,R)<br />
1Begin<br />
2 If IIS R Tree has a node C such that C.item-name = N.item- name then<br />
3 Increment supp(N) by 1<br />
4 Else<br />
5 Create new node N and initial supp(N) to 1<br />
6 Link N to N’s parent<br />
7 Link N to N’s header table<br />
8 Link N to same-item<br />
9 End //If<br />
10End //Begin<br />
If no confusion arises, then { x 1,<br />
x2<br />
,..., xn}<br />
and ({ x1 , x2<br />
,..., xn }: s)<br />
are replaced by x 1x2...<br />
xn<br />
, and<br />
x1 x2...<br />
xn : s respectively, where x 1 , x2<br />
,..., xn<br />
are items.<br />
Example 3. In this example, we describe the steps to construct all <strong>of</strong> the IIS item Trees except<br />
the last frequent item p using the IIS in Figure 1 and minsup = 3. The first frequent item in the IIS is
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
137<br />
a; therefore, we first construct the IIS a Tree, and item a is a root. We obtain two paths: (,<br />
, , , ) and (, , , , ). The first path consists <strong>of</strong><br />
frequent items (a,c,f,m,p) that appear twice in the DB. Similarly, the second path indicates frequent<br />
items (a,b,c,f,m) that are contained in only one transaction in the DB. These two paths share the<br />
frequent item a; thus, a appears three times in the DB. Other IIS item Trees, except the IIS p Tree, can be<br />
similarly constructed. In Figure 2, all <strong>of</strong> the IIS item Trees, except the last IIS p Tree, are illustrated.<br />
a:3<br />
c:2<br />
b:1<br />
f:2 c:1<br />
m:2<br />
f:1<br />
p:2 m:1<br />
(a) (b) (c) (d) (e)<br />
Figure 2. IIS item Trees<br />
Definition 4 (A 1 -tree). Let DB be a transaction database; let T be a tree such that each node<br />
<strong>of</strong> the tree is <strong>of</strong> the form ( A : s)<br />
, where A is a frequent itemset in DB and s is the support <strong>of</strong> A; and let<br />
A 1 be a frequent itemset in the DB. T is called an A 1 -tree if, for any path P <strong>of</strong> T, P is <strong>of</strong> the form<br />
P = (( A1 : s1),<br />
( A2<br />
: s2<br />
),...,( A k : sk<br />
)), where k is a positive integer; A i and A j are pairwise disjoint<br />
frequent itemsets for all i, j = 1,2,…,k with i j<br />
i<br />
; s i is the support <strong>of</strong> <br />
m<br />
Am for i =1,2,…,k; (A 1 :s 1 )<br />
= 1<br />
is the root <strong>of</strong> T; and ( A k : sk<br />
) is a leaf <strong>of</strong> T.<br />
Example 4. According to Figure 2 (a), the ac-tree is shown in Figure 3.<br />
Definition 5 (Prefix subpath). Let T be a tree, a 1 be the root <strong>of</strong> T , and P ( a 1<br />
,..., a m<br />
) be a<br />
path <strong>of</strong> T, where m is a positive integer. Every path Q = ( a 1 ,..., a k ) <strong>of</strong> T is then called a prefix subpath<br />
<strong>of</strong> P, where 1 ≤k ≤m<br />
.<br />
Example 5. According to Figure 2(a), the paths ((a:3),(c:2)) and ((a:3),(c:2),(f:2)) are prefix<br />
subpaths <strong>of</strong> ((a:3),(c:2),(f:2),(m:2)).<br />
Definition 6 (Subheader). Let T be an A-tree and (x:s(x)) be a node <strong>of</strong> T, where x is a<br />
frequent item in a given transaction database and s(x) is the support <strong>of</strong> x. Suppose that all <strong>of</strong> the<br />
nodes (<strong>of</strong> T) containing x are only in the paths P 1 ,…,P k <strong>of</strong> T from the root (A:s) to some leafs <strong>of</strong> T,<br />
and suppose that Q i is a prefix subpath (<strong>of</strong> P i ) from the root ( A : s)<br />
to ( x : s(<br />
Qi<br />
)) for all i = 1,..., k.<br />
The<br />
subheader <strong>of</strong> T, denoted by SH(A), is defined as the order set SH(A) = {x|x is a frequent item not<br />
k<br />
contained in A, ( x : s(<br />
Qi<br />
)) is a node in Q i for i = 1,…,k and ∑s ( Qi)<br />
≥minsup}.<br />
i = 1
138 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
Example 6. According to Figure 2(a), SH(a) = {c,f,m}, where SH(a ) is the subheader <strong>of</strong> the<br />
tree in Figure 2 (a), supp(c) = 3, supp(f) = 3, and supp(m) = 3.<br />
Definition 7 (Conditional itemset-tree). Let A 1 be a frequent itemset, T be an A 1 -tree, x 2 be<br />
a frequent item with x2 ∈SH(<br />
A1<br />
) - A 1,<br />
and A1 x2<br />
: = A1<br />
∪{ x2}<br />
be a frequent itemset. A conditional A 1 x 2 -<br />
tree, denoted by T ( A 1 x 2 ), is a tree that satisfies the following:<br />
(i) ( A 1x2<br />
: s2<br />
) is the root <strong>of</strong> T ( A 1 x 2 ), where s 2 is the support <strong>of</strong> A 1 x and 2 s 2 ≥ minsup.<br />
(ii) The number <strong>of</strong> items in SH ( A 1 ) > 1.<br />
(iii) All <strong>of</strong> the paths are derived from T in the following way: Q = (( A1 x2<br />
: s2<br />
), ( x3<br />
: s3<br />
),...,( x k : sk<br />
)) is a<br />
path <strong>of</strong> T ( A 1 x 2 ) if and only if a path P = (( A1 : s1<br />
), ( x2<br />
: s2<br />
),...,( x : l sl<br />
)) exists from the root ( A 1 : s1<br />
) to the<br />
leaf ( x l : sl<br />
) <strong>of</strong> T, where l ≥k<br />
and xk<br />
is in SH ( A 1 ); and if there is a node ( x r : sr<br />
) <strong>of</strong> P such that r > k<br />
and (( A1 : s1<br />
),( x2<br />
: s2<br />
),...,( x r<br />
: sr<br />
)) is a prefix subpath <strong>of</strong> P, then x r must not be in SH ( A 1 ).<br />
Example 7. According to Figure 2 (a), the conditional itemset-tree, or conditional ac-tree, is<br />
shown in Figure 4.<br />
Notably, every non-empty subset <strong>of</strong> a frequent itemset is also frequent. This fact leads to the<br />
following definition.<br />
Definition 8 (Frequent itemset * <strong>of</strong> length m derived from tree). Let A be a frequent<br />
itemset, x be a frequent item, and x A. Suppose that T is a conditional Ax-tree, m is a positive<br />
integer greater than 1, and Ax = A∪{<br />
x}<br />
has m elements. Ax is called a frequent itemset * <strong>of</strong> length m<br />
derived from T, denoted by FS * m ( Ax ) , if T contains precisely two nodes and Ax is a frequent itemset,<br />
or if SH ( A)<br />
= { x}.<br />
Example 8. According to Figure 5, the frequent itemset * <strong>of</strong> length 4 derived from the<br />
*<br />
conditional acf-tree is FS 4 ( acfm)<br />
= acfm.<br />
.<br />
Figure 3. ac-tree Figure 4. Conditional ac-tree Figure 5. Conditional acf-tree<br />
Definition 9 (Extendable frequent itemset * ). Let m be a positive integer greater than 1, T<br />
be a T (Ax)<br />
defined as in definition 7 with A 1 = A and x 1 = x,<br />
and Ax be a frequent itemset* <strong>of</strong> length<br />
*<br />
m derived from T. Each itemset in FS m ( Ax)<br />
is said to be extendable if m ≥3<br />
. For every k = 2, 3,…,<br />
m, we let * ( Ax<br />
*<br />
Ext k ) denote the set <strong>of</strong> all itemsets containing exactly k items <strong>of</strong> FS m ( Ax),<br />
where m ≥3<br />
.<br />
Each element <strong>of</strong> Ext * ( Ax k ) is called an extendable frequent itemset * (derived from T) <strong>of</strong> length k for<br />
all k ≥2<br />
. The set <strong>of</strong> all extendable frequent itemsets* <strong>of</strong> length k that is denoted<br />
by Ext* k<br />
= ∪{ Ext* k ( Ay)<br />
| Ay is a frequent itemset* <strong>of</strong> length at least k derived from a conditional Aytree}<br />
for all k ≥2<br />
.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
139<br />
*<br />
Example 9. According to the previous example, the length <strong>of</strong> FS 4 ( acfm ) is 4; we then find<br />
*<br />
that 2 ( acfm<br />
*<br />
Ext ) = {ac, af, am, cf, cm, fm} and Ext 3 ( acfm)<br />
= {acf, acm, afm, cfm}. Therefore, Ext * 2 =<br />
{ac, af, am, cf, cm, fm, and all members <strong>of</strong> other * *<br />
Ext 2 ( Ay)<br />
} and Ext 3 = {acf, acm, afm, cfm, and all<br />
members <strong>of</strong> other Ext * 3 ( Ay)<br />
}.<br />
Definition 10 (Frequent itemset * <strong>of</strong> length m). Let k be the maximum length <strong>of</strong> all frequent<br />
itemsets in a transaction database, FS * m ( Ax ) and Ext* k be given as in definitions 8 and 9 respectively;<br />
let Am = { FS * m ( Ax)<br />
| Ax being a frequent itemset * <strong>of</strong> length m derived from T } for m = 2, 3,…, k;<br />
*<br />
define FS 2 = A 2 ∪{ Ax | Ax being a frequent itemset <strong>of</strong> length 2}; and define FS * m = Am for m = 2, 3,…,<br />
k}. Let FI * *<br />
* * *<br />
m be defined by FI 1 = {{ x}<br />
| x being a frequent item} and FI m = FS m ∪Extm<br />
for m = 2, 3,…,<br />
k}. Any element <strong>of</strong> FI * m is called the frequent itemset * <strong>of</strong> length m.<br />
*<br />
Example 10. According to the previous example, we find that FI 2 = {ac, af, am, cf, cm,<br />
*<br />
*<br />
fm}, FI 3 = {acf, acm, afm, cfm}, and FI 4 = {acfm}.<br />
Definition 11 (Frequent itemset * ). Let FI * m be given as in definition 10, and let FI<br />
* denote<br />
k *<br />
∪m = 1 FI m , where k is the maximum length <strong>of</strong> all frequent itemsets in a transaction database. Any<br />
element <strong>of</strong> FI<br />
* is called a frequent itemset * .<br />
Example 11. According to the previous example, FI<br />
* is {ac, af, am, cf, cm, fm, acf, acm,<br />
afm, cfm, acfm}.<br />
On the basis <strong>of</strong> the above definitions and examples, Algorithm 3 presents the IIS-Mine<br />
algorithm to show how it can be used to mine all frequent itemsets.<br />
Algorithm 3: (IIS-Mine: Mining frequent itemsets using IIS item Tree)<br />
Input: IIS, IIS item Trees constructed according to Algorithm 2, and minsup<br />
Output: FI*<br />
Procedure AllFreqItemset (IIS, FI*)<br />
1 Begin<br />
2 For each frequent item x in the IIS do<br />
3<br />
*<br />
FI1 = {{ x | x ∈I<br />
, supp(x) ≥ minsup}<br />
4 Call IIS x Tree , which is constructed from Algorithm 2<br />
5 If Tree ≠{}<br />
6 Call IIS-Mine (Tree, x)<br />
7 End //If<br />
8 End //For<br />
9<br />
*<br />
Find FI ∪<br />
FI *<br />
// FI* is given in definition 11<br />
= m=1<br />
m<br />
10 End //Begin<br />
Procedure IIS-Mine(Tree, x)<br />
1 Begin<br />
2 Call SubHeader (A-tree, subheaderA)<br />
3 Generate all Ax with its support<br />
//All Ax are the frequent itemsets where A is the root <strong>of</strong> A-tree and<br />
4 For each | λ |> 1 do // λ is Ax<br />
5 Flag=1<br />
6 While Flag = 1 do<br />
7 Call SkipFreqItemset (subheaderA, λ , δ ,FlagRepeat, FlagTree)<br />
8 If FlagRepeat=0 and FlagTree=0 then // ( δ *<br />
FI m )<br />
9 Call CondItemsetTree (A-tree, δ , conditional δ -tree)<br />
x ∈subheaderA
140 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
10 Else<br />
11 Flag=0<br />
12 End // If<br />
13 If conditional Ax-tree contains greater than two nodes<br />
14 Call IIS-Mine(conditional δ -tree, x)<br />
15 Else Flag=0<br />
16 End //If<br />
17 End //While<br />
18<br />
*<br />
*<br />
If | FS (δ) |≥3<br />
and FS (δ)<br />
*<br />
FI then<br />
m<br />
m<br />
*<br />
m<br />
*<br />
k<br />
19 Call ExtFreqItemset( FS m (δ)<br />
, Ext ) // FS is given in definition 8<br />
20<br />
*<br />
* *<br />
*<br />
Save FS m (δ)<br />
and all Extk<br />
to FI m<br />
// FI m is given in definition 10<br />
21 Else<br />
22<br />
*<br />
If FSm<br />
(δ)<br />
*<br />
FI m then<br />
23<br />
*<br />
Save FS (δ)<br />
to<br />
m<br />
*<br />
FI m<br />
24 End //If<br />
25 End //If<br />
26 End //For<br />
27 End //Begin<br />
Procedure SubHeader (A-tree, subheaderA)<br />
1 Begin<br />
2 Each frequent item x <strong>of</strong> A-tree // SH(A) is given in definition 6<br />
3 Find {(x:s(x))|x∈SH(A), s(x) is the support <strong>of</strong> x in A-tree}<br />
4 End // Begin<br />
Procedure CondItemsetTree(Tree, δ , conditional δ -tree)<br />
1 Begin<br />
2 Scan tree once to collect the paths that have an association with root δ<br />
3 For all paths are derived from tree do<br />
4 Connect all paths to δ<br />
5 End //For<br />
6 End //Begin<br />
Procedure SkipFreqItemset (SubheaderA, λ , δ , FlagRepeat, FlagTree)<br />
1 Begin // To skip the construction <strong>of</strong> conditional item tree<br />
2 // x is the frequent item in subheaderA // λ is the root <strong>of</strong> tree; or a frequent itemset<br />
3 // FlagRepeat=0 means δ *<br />
FI m // FlagTree=0 means the conditional item-tree is constructed<br />
4 // n is the maximum number <strong>of</strong> elements <strong>of</strong> itemsets in subheaderA<br />
5 FlagRepeat =1, FlagTree=1<br />
6 β = λ, α = β<br />
7 While( FlagRepeat=1 and (order <strong>of</strong> x ≤ n )) do<br />
8 If β *<br />
FI m then<br />
9 FlagRepeat=0, FlagTree=0<br />
10 If x is the last item<br />
11 If | δ |= 2 then FlagTree=1<br />
12 End //If<br />
13 δ = α<br />
14 End //If<br />
15 Else<br />
16 If x is the last item then<br />
17 Increment order <strong>of</strong> item x<br />
18 Else<br />
19 Increment order <strong>of</strong> item x<br />
20 β = δ∪x<br />
21 α = δ<br />
22 End// If<br />
*<br />
m
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
141<br />
23 δ = β<br />
24 End //If<br />
25 End //while<br />
26 End //Begin<br />
*<br />
Procedure ExtFreqItemset( FS (δ)<br />
,<br />
m<br />
*<br />
Ext k )<br />
1 Begin // Ext is given in definition 9<br />
*<br />
*<br />
2 Generate all subsets <strong>of</strong> FSm<br />
(δ)<br />
and save to Ext k<br />
3 End // Begin<br />
*<br />
k<br />
Example 12. This example is given to demonstrate how the proposed IIS-Mine algorithm<br />
can be used to mine all frequent itemsets. Assume minsup = 3 for all <strong>of</strong> the definitions above, and for<br />
Examples 1-3, Figure 2 and Algorithm 3. For simplicity, this example is divided into five main steps<br />
ordered by five frequent items in the IIS. The proposed mining algorithm proceeds as follows.<br />
Let step 1 be the first main step. According to Procedure AllFreqItemset, the frequent item a<br />
is the first frequent item in the IIS that is mined. The IIS a Tree is constructed, as illustrated in Figure<br />
2(a). The algorithm calls Procedure IIS-Mine, so Procedure Subheader is called in order. The SH(a)<br />
is (c,f,m), where supp(c) = 3, supp(f) = 3 and supp(m) = 3. After line 3 in the procedure, IIS-Mine<br />
generates all <strong>of</strong> the frequent itemsets with its support, i.e. ac, af and am; all <strong>of</strong> these frequent<br />
itemsets have support equal to three. Let steps 1.1, 1.2 and 1.3 represent each <strong>of</strong> the frequent<br />
itemsets.<br />
The step 1.1, the first frequent itemset is ac, which has support equal to three. The algorithm<br />
*<br />
checks |ac|>1 and then calls Procedure SkipFreqItem. At this procedure, ac is not in FI 2 , so the<br />
algorithm rolls back to line 9 <strong>of</strong> Procedure IIS-Mine. The frequent itemset ac with its support is<br />
defined to be the root; then, the conditional ac-tree, which has root “”, is constructed using<br />
the input IIS a Tree. The conditional ac-tree is illustrated in Figure 4. The algorithm is iterated by<br />
calling Procedure IIS-Mine again because the conditional ac-tree contains more than two nodes; let<br />
this call be step 1.1.1.<br />
In step 1.1.1, the Procedure IIS-Mine is called; SH(ac) = (f,m), where supp(f) and supp(m)<br />
are then equal to 3. At line 3, the algorithm generates frequent itemsets, which are acf and acm,<br />
where supp(acf) and supp(acm) are then equal to 3. The first frequent itemset in this step is acf and<br />
*<br />
| acf |> 1, so the procedure SkipFreqItem in line 7 is processed, and it finds that acf is not in FI 3 .<br />
Next, the algorithm rolls back to line 9 to construct a conditional acf-tree that has a conditional actree<br />
as an input tree, which is illustrated in Figure 5. The condition <strong>of</strong> line 13 is that the conditional<br />
*<br />
acf-tree contains only two nodes, so the algorithm obtains FS 4 ( acfm)<br />
= acfm . At line 18, the size <strong>of</strong><br />
FS *<br />
4 ( acfm ) is greater than three and * *<br />
FS 4 ( acfm)<br />
FI 4 . Thus, at line 19, Procedure ExtFreqItemset is<br />
*<br />
called to find all <strong>of</strong> the subsets <strong>of</strong> 4 ( acfm<br />
*<br />
FS ) , which are Ext 2 ( acfm ) and Ext *<br />
3 ( acfm ) : *<br />
Ext 2 ( acfm)<br />
= {ac,<br />
af, am, cf, cm, fm} and Ext *<br />
3 ( acfm ) = {acf, acm, afm, cfm}; hence, *<br />
FI 2 = {ac, af, am, cf, cm, fm},<br />
*<br />
3<br />
*<br />
FI = {acf, acm, afm, cfm}, and FI = {acfm}.<br />
4<br />
In step 1.1.2, the next frequent itemset generated together with step 1.1.1 is acm. At line 7 <strong>of</strong><br />
Procedure IIS-Mine, the algorithm calls Procedure SkipFreqItem and obtains acm, which is already a
142 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
*<br />
member <strong>of</strong> FI 3 ; m is the last frequent item in SH(ac), so we exit from this step without the<br />
construction <strong>of</strong> a conditional acm-tree.<br />
In step 1.2, at line 4 <strong>of</strong> Procedure IIS-Mine, the next frequent itemset is af, and at line 7,<br />
*<br />
Procedure SkipFreqItem is called and obtains af, which is contained in FI 2 . In the loop in line 20 <strong>of</strong><br />
Procedure SkipFreqItem, after af is combined with the next frequent item in SH(a), which is m, afm<br />
*<br />
is obtained, which is contained in FI 3 , and item m is the last item in SH(a). The algorithm rolls back<br />
to line 8 <strong>of</strong> Procedure IIS-Mine, so the conditional af-tree is not constructed.<br />
In step 1.3, at line 4 <strong>of</strong> Procedure IIS-Mine, the next frequent itemset is am. At line 7,<br />
*<br />
Procedure SkipFreqItem is called and obtains am, which is contained in FI 2 , and there is no frequent<br />
item in SH(a). Therefore, the conditional am-tree is not constructed.<br />
Let step 2 be the second main step. According to line 2 <strong>of</strong> Procedure AllFreqItemset, b is the<br />
second frequent item in the IIS that is mined. After line 4, the IIS b Tree is constructed, as illustrated<br />
in Figure 2(b). The algorithm calls Procedure IIS-Mine at line 6. At line 2 <strong>of</strong> Procedure IIS-Mine,<br />
Procedure Subheader is called and obtains an empty SH(b), so the size <strong>of</strong> the frequent itemset<br />
generated with b is one. The processing <strong>of</strong> this step is terminated, and we return to line 2 <strong>of</strong><br />
Procedure AllFreqItemset.<br />
Let the third main step be step 3. According to line 2 <strong>of</strong> Procedure AllFreqItemset, c is the<br />
third frequent item in IIS that is mined. Line 4 is called to construct the IIS c Tree, as illustrated in<br />
Figure 2(c). At line 6, the algorithm calls Procedure IIS-Mine to mine frequent itemsets. At line 2 <strong>of</strong><br />
Procedure IIS-Mine, Procedure Subheader is called to obtain the SH(c) that is (f,m,p). Next, at line<br />
3, the algorithm generates frequent itemsets, which are cf, cm and cp, where supp(cf), supp(cm) and<br />
supp(cp) = 3. Let steps 3.1, 3.2 and 3.3 represent each <strong>of</strong> the frequent itemsets.<br />
In step 3.1, at line 4 <strong>of</strong> Procedure IIS-Mine, the first frequent itemset is cf, where |cf| > 1; line<br />
7 then calls Procedure SkipFreqItemset. At Procedure SkipFreqItemset, cf combines with the next<br />
frequent item in SH(c), which is m, so cfm with a support <strong>of</strong> 3 is obtained after processing lines 19-<br />
*<br />
23. Next, line 8 is checked, and as cfm is already obtained in FI 3 , lines 19-23 are checked again, and<br />
*<br />
cfmp is obtained. The frequent itemset cfmp is not a member in FI 4 , and p is the last frequent item in<br />
SH(c), so cfm is set to be a root for a conditional cfm-tree. The algorithm goes back to line 8 <strong>of</strong><br />
Procedure IIS-Mine to construct a conditional cfm-tree, where the IIS c Tree is an input tree, which is<br />
*<br />
illustrated in Figure 6. Supp(p) is less than minsup, hence FS 3 ( cfm)<br />
= cfm . Procedure ExtFreqItemset<br />
in line 18 is not called because FS *<br />
3 ( cfm ) is contained in *<br />
FI 3 . In this step, the algorithm skips the<br />
construction <strong>of</strong> the conditional cf-tree.<br />
In step 3.2, according to line 4 <strong>of</strong> Procedure IIS-Mine, the next frequent itemset is cm, and<br />
Procedure SkipFreqItemset in line 7 is called. Line 8 <strong>of</strong> Procedure SkipFreqItemset checks that cm is<br />
*<br />
a member in FI 2 . Next, lines 19-23 are checked, so cm combines with the next item in SH(c), which<br />
*<br />
is p, and we obtain cmp with a support <strong>of</strong> 3. The frequent itemset cmp is not in FI 3 , and p is the last<br />
frequent item in SH(c), so cm is set to be a root for a conditional cm-tree. The algorithm goes back<br />
to line 8 <strong>of</strong> Procedure IIS-Mine to construct a conditional cm-tree, where the IIS c Tree is an input<br />
*<br />
tree, which is illustrated in Figure 7. Supp(p) is less than minsup, hence FS 2 ( cm)<br />
= cm and FS *<br />
2 ( cm )<br />
*<br />
are already members in FI 2 .
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
143<br />
Figure 6. Conditional cfm-tree<br />
Figure 7. Conditional cm-tree<br />
In step 3.3, according to line 4 <strong>of</strong> Procedure IIS-Mine, the next frequent itemset is cp. Next,<br />
*<br />
at line 5, Procedure SkipFreqItemset is called, and cp is not a member in FI 2 . There is no frequent<br />
item in SH(c), so this procedure is terminated, and we return to line 8 <strong>of</strong> Procedure IIS-Mine. After<br />
lines 22-23 are checked, the new answer is contained in FI<br />
* 2 , which is cp. This step skips the<br />
construction <strong>of</strong> a conditional cp-tree.<br />
The remaining steps such as the fourth main step, which constructs the IIS f Tree and is<br />
illustrated in Figure 2(d), and the fifth main step, which constructs the IIS m Tree and is illustrated in<br />
Figure 2(e), are performed in the same way in sequence.<br />
The complete frequent itemsets are shown by item in Table 2 and by length in Table 3.<br />
Table 2. Complete frequent itemsets by item<br />
Item<br />
a<br />
b<br />
c<br />
f<br />
m<br />
p<br />
Frequent itemset<br />
a, acfm, ac, af, am, cf, cm, fm, acf, acm, amf, cfm<br />
b<br />
c, cp<br />
f<br />
m<br />
p<br />
Table 3. Complete frequent itemsets by length<br />
m-Length<br />
Frequent itemset<br />
1 a, b, c, f, m, p<br />
2 ac, af, am, cf, cm, cp, fm<br />
3 acf, acm, afm, cfm<br />
4 acfm<br />
The advantages <strong>of</strong> our algorithm are as follows. First, according to step 1.1.1 <strong>of</strong> Example 12,<br />
the frequent itemsets acfm are obtained, which are derived from a conditional acf-tree. This step<br />
shows the properties <strong>of</strong> an extendable frequent itemset * , which are given in Definition 9 and<br />
Procedure ExtFreqItemset in Algorithm 3. This step then finds all <strong>of</strong> the subsets <strong>of</strong> acfm, so we<br />
derive ten frequent itemsets, which are ac, af, am, cf, cm, fm, acf, acm, amf and cfm, without<br />
contributing more trees or using recursion to mine. Therefore, our algorithm can reduce many
144 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
*<br />
subsequent steps in mining frequent itemsets. It can be noticed that if | FS m<br />
( Ax)<br />
| is large, then the<br />
property <strong>of</strong> an extendable frequent itemset * is frequently used. Second, the method is also good for<br />
reducing run time and space consumption, and its performance will be shown in the experimental<br />
section. Last, according to step 3.1 <strong>of</strong> Example 12, the conditional cfm-tree is obtained, which<br />
reduces one node in the tree because <strong>of</strong> the step that sets frequent itemsets as the root node in<br />
Procedure SkipFreqItemset in Algorithm 3. The algorithm reduces the number <strong>of</strong> nodes, levels and<br />
size <strong>of</strong> the tree, thus reducing space consumption. In general, users change many minimum support<br />
thresholds to make their decision. Our method, which uses an IIS and the restart algorithm shown in<br />
Algorithm 3, supports the ability to make these changes without rescanning the database.<br />
Correctness<br />
The following theorem and pro<strong>of</strong> are given to demonstrate that the proposed IIS-Mine<br />
algorithm can mine frequent itemsets completely and correctly.<br />
Theorem: The set <strong>of</strong> all frequent itemsets * derived from IIS-Mine is the complete set <strong>of</strong> all <strong>of</strong><br />
the frequent itemsets.<br />
Pro<strong>of</strong>: Let I be the nonempty finite set <strong>of</strong> all items in a given transaction database, FI denote<br />
the set <strong>of</strong> all frequent itemsets in the transaction database, and minsup 0 . It must be proved that<br />
*<br />
FI = FI .<br />
First, we prove that FI ⊆ FI<br />
* . Let F = { a 1 ,..., ak } ∈FI<br />
with a 1 < l ... < l ak<br />
, and si<br />
= supp { a 1 ,..., a i }<br />
for all i 1,...,<br />
k.<br />
Then, s 1 ≥... ≥sk<br />
≥α<br />
, and an item b exists such that b a1<br />
, and IIS b tree contains<br />
a 1 ,...,a k . It can be assumed that b is the item in I having these properties because I is the nonempty<br />
'<br />
'<br />
finite set <strong>of</strong> all items. Because s i ≥...<br />
≥sk<br />
≥α<br />
for all i 1,...,<br />
k , there exists a path (( b 1 : s1<br />
),...( b l : sl<br />
)) <strong>of</strong><br />
IIS b tree containing a : s ),...,( a k<br />
: s ); that is, for all i 1,...,<br />
k , there exists j 1,...,<br />
l such that<br />
(<br />
1 1<br />
k<br />
( a i : si<br />
) = ( b j : s j ). It is obvious from the IIS-Mine algorithm that FI FI .<br />
*<br />
*<br />
We also show that FI ⊆FI<br />
. Let F = { a1,...,<br />
ak } ∈FI<br />
with a 1 < l ... < l ak<br />
. Then, for some<br />
positive integer m, ∈<br />
* *<br />
F FI m . It is evident from definition 10 that if m 1, F ∈FI<br />
m implies F ∈ FI .<br />
*<br />
Now, suppose m 2 ; then, F ∈FI<br />
m or F ∈ Ext<br />
* m.<br />
In the first case, F ∈ FS<br />
* m , and from Definition 8,<br />
SH{a 1 ,…,a k-1 }=(a k ) or the conditional { a<br />
1,...,<br />
a k 1}<br />
ak<br />
tree contains exactly two nodes,<br />
({ a<br />
1,...,<br />
a k 1}:<br />
sk<br />
1<br />
) and ( a<br />
k<br />
: sk<br />
) , where si<br />
= supp { a 1 ,..., a i } for i k 1,<br />
k . Then from definitions 7 and<br />
8, we obtain { a1,...,<br />
ak<br />
1}<br />
ak<br />
{<br />
a1,...,<br />
ak<br />
} F FI.<br />
In the other case, suppose that F ∈ Ext*<br />
m for m 2 .<br />
Then from definition 9, an Ax exists such that F ∈Ext*<br />
m ( Ax)<br />
, and Ax is a frequent itemset * <strong>of</strong> length at<br />
least m derived from the conditional Ax-tree. Again, from definition 9, a positive integer k greater<br />
than 2 exists such that F has itemsets containing precisely m items <strong>of</strong> FS * k ( Ax ) , and from definitions<br />
6 and 8, FS * k ( Ax ) is a frequent itemset, hence F ∈FI.<br />
The pro<strong>of</strong> is complete.<br />
⊆ *<br />
RESULTS AND DISCUSSION<br />
We have presented the experiments in which the run time, memory consumption and<br />
scalability are tested for the IIS-Mine algorithm and FP-growth algorithm with different datasets and<br />
varying minimum support thresholds. The experiments were performed on a Micros<strong>of</strong>t Windows XP
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
145<br />
Pr<strong>of</strong>essional Version 2002 Service Pack 3 operating system on a personal computer with 1 GB <strong>of</strong><br />
main memory and Pentium (R) CPU 3.00 GHz. All algorithms were coded using C language. Two<br />
groups <strong>of</strong> benchmark datasets, i.e. two synthetic datasets and two real datasets, were used.<br />
For the first group <strong>of</strong> datasets, we also presented the experimental results for two synthetic<br />
datasets generated by the IBM Almaden Quest research group [24-25]. The datasets serve as the<br />
FIMI repository, which is a result <strong>of</strong> the workshops on frequent itemset mining implementations [26,<br />
27]. The two original databases <strong>of</strong> synthetic datasets are T10I4D100K and T40I10D100K, which are<br />
sparse datasets. The notation TxIyDzK denotes a dataset where K is 1,000 transactions. Table 4 lists<br />
the parameters <strong>of</strong> the synthetic datasets, which vary in the number <strong>of</strong> transactions, i.e. 20%, 40%,<br />
60% and 80% <strong>of</strong> the original database.<br />
Table 4. Parameters <strong>of</strong> the synthetic datasets<br />
|T|<br />
|I|<br />
|D|<br />
Average number <strong>of</strong> items per transaction<br />
Average length <strong>of</strong> a frequent itemset<br />
Number <strong>of</strong> transactions<br />
For the second group <strong>of</strong> datasets, the real datasets from the UCI machine learning repository<br />
[28] were used to test the proposed method. The real datasets used in the experiment were Chess<br />
[29] and Mushroom [30], which are dense datasets with a great number <strong>of</strong> long frequent itemsets.<br />
The characteristics <strong>of</strong> the real datasets are shown in Table 5.<br />
Table 5. Characteristics <strong>of</strong> real datasets<br />
Real dataset<br />
Chess<br />
Mushroom<br />
Description<br />
Average number <strong>of</strong> items per transaction = 37, number <strong>of</strong> transactions = 3,196, and<br />
number <strong>of</strong> items = 75<br />
Average number <strong>of</strong> items per transaction = 23, number <strong>of</strong> transactions = 8,124, and<br />
number <strong>of</strong> items = 119<br />
Run Time<br />
Figures 8(a) and 8(b) show the performance <strong>of</strong> the algorithms on two synthetic datasets,<br />
T10I4D100K and T40I10D100K respectively. In Figure 8(a), IIS-Mine performs better than FPgrowth<br />
in every support threshold. The gap in the graph becomes larger as the support threshold<br />
decreases. In Figure 8(b), when the minimum support is set at 20%, 17.5%, 15% or 12.5%, the run<br />
time between the two algorithms is not very different. However, when the minimum support is set at<br />
10%, 7.5% or 5%, the run time <strong>of</strong> FP-growth increases significantly when compared to that <strong>of</strong> IIS-<br />
Mine, which confirms that IIS-Mine performs better than FP-growth. The results shown in Figures<br />
8(a) and 8(b) can be explained as follows. With sparse datasets, when the minimum support is high,<br />
the number <strong>of</strong> frequent itemsets is low. However, when the minimum support is low, many frequent<br />
itemsets are obtained. IIS-Mine is always faster than FP-growth method, especially when the
146 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
minimum support is low, because FP-growth constructs bushy and wide FP-trees when the minimum<br />
support is low. So FP-growth is computationally expensive for tree traversing the FP-trees. IIS-Mine<br />
has the step <strong>of</strong> finding the root node from previous frequent itemsets, which can reduce the number<br />
<strong>of</strong> nodes and levels <strong>of</strong> the conditional itemset-tree. Therefore, traversing in the conditional itemsettree<br />
is on a reduced tree, which can result in a low run time consumption. However, the run time <strong>of</strong><br />
both algorithms relies on the length <strong>of</strong> the transaction (as observed in a comparison <strong>of</strong> the graphs in<br />
Figures 8(a) and 8(b) with a minimum support <strong>of</strong> 5%); when the transaction is long, so it the run<br />
time <strong>of</strong> both algorithms.<br />
Run time (s)<br />
80<br />
60<br />
40<br />
20<br />
0<br />
T10I4D100K<br />
IIS-Mine<br />
FP-growth<br />
Run time (s)<br />
4000<br />
3000<br />
2000<br />
1000<br />
0<br />
T40I10D100K<br />
IIS-Mine<br />
FP-growth<br />
5 3 2 1<br />
Minimum support (%)<br />
20<br />
17.5<br />
15<br />
12.5<br />
10<br />
7.5<br />
5<br />
Minimum support (%)<br />
(a)<br />
(b)<br />
Run time (s)<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
90 80 70 60 50<br />
Minimum support (%)<br />
Chess<br />
IIS-Mine<br />
FP-growth<br />
Run time (s)<br />
8<br />
6<br />
4<br />
2<br />
0<br />
40 30 20 10<br />
Minimum support (%)<br />
Mushroom<br />
IIS-Mine<br />
FP-growth<br />
(c)<br />
(d)<br />
Figure 8. Run time <strong>of</strong> mining on: a) T10I4D100K; b) T10I4D100K; c) Chess; d) Mushroom<br />
Figures 8(c) and 8(d) show the performance <strong>of</strong> algorithms on two dense datasets: chess and<br />
mushroom. Figure 8(c) shows that the run time <strong>of</strong> IIS-Mine is better than that <strong>of</strong> FP-growth in every<br />
support threshold. The run time <strong>of</strong> both algorithms increases when the minimum support threshold is<br />
reduced to 50%. In Figure 8(d), IIS-Mine again performs better than FP-growth in every minimum<br />
support. The run time <strong>of</strong> FP-growth increases significantly compared with IIS-Mine when the<br />
minimum support is less than 30%. The results shown in Figures 8(c) and 8(d) can be explained as<br />
follows. In the two Figures, IIS-Mine is faster than FP-growth for dense datasets. The main work in<br />
FP-growth is traversing FP-trees and constructing new conditional FP-trees after the first FP-tree is<br />
constructed from the original database. For dense datasets, we have found from numerous<br />
experiments that the time spent on traversing FP-trees is very long. IIS-Mine improves this problem<br />
using the property <strong>of</strong> extendable itemsets to reduce the number <strong>of</strong> recursive mining steps so that the<br />
size and number <strong>of</strong> constructing and traversing the trees are reduced. The run time <strong>of</strong> IIS-Mine is<br />
then less than that <strong>of</strong> FP-growth.
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
147<br />
Memory Consumption<br />
Figures 9(a) and 9(b) show the memory consumption <strong>of</strong> the algorithms on the synthetic<br />
datasets. In Figure 9(a), FP-growth consumes more memory than IIS-Mine. The graphs clearly<br />
separate out when the minimum support is less than 3%. In Figure 9(b), there is no difference in<br />
memory consumption until the minimum support is less than 15%. Thus, we can see that IIS-Mine<br />
consumes less memory than FP-growth on synthetic datasets. The large memory consumption <strong>of</strong> FPgrowth<br />
when running on synthetic datasets can be explained by the fairly low minimum support and<br />
the presence <strong>of</strong> many single items in the datasets; therefore, FP-growth constructs wide and bushy<br />
trees to mine all frequent itemsets. However, IIS-Mine uses the property <strong>of</strong> extendable itemsets,<br />
which reduces the construction <strong>of</strong> conditional itemset-trees, and uses the step <strong>of</strong> finding the root<br />
node from previous frequent itemsets. Therefore, the node construction and tree sizes are reduced,<br />
resulting in a reduction in memory consumption.<br />
Main memory (K)<br />
Main memory(K)<br />
4000000<br />
3000000<br />
2000000<br />
1000000<br />
0<br />
10000000<br />
8000000<br />
6000000<br />
4000000<br />
2000000<br />
0<br />
5 3 2 1<br />
Minimum support (%)<br />
(a)<br />
90 80 70 60 50<br />
Minimum support (%)<br />
(c)<br />
T10I4D100K<br />
IIS-Mine<br />
FP-growth<br />
Chess<br />
IIS-Mine<br />
FP-growth<br />
Main memory (K)<br />
Main memory(K)<br />
15000000<br />
10000000<br />
5000000<br />
500000<br />
400000<br />
300000<br />
200000<br />
100000<br />
0<br />
0<br />
20<br />
15<br />
10<br />
5<br />
Minimum support (%)<br />
(b)<br />
40 30 20 10<br />
Minimum support (%)<br />
(d)<br />
T40I10D100K<br />
IIS-Mine<br />
FP-growth<br />
Mushroom<br />
IIS-Mine<br />
FP-growth<br />
Figure 9. Memory consumption <strong>of</strong> mining on: a) T10I4D100K; b) T40I10D100K; c) Chess;<br />
d) Mushroom<br />
Figures 9(c) and 9(d) show that the memory consumption <strong>of</strong> IIS-Mine is less than that <strong>of</strong> FPgrowth<br />
on dense datasets. Figure 9(c) shows that when the minimum support is less than 80%, the<br />
memory consumption <strong>of</strong> FP-growth increases significantly compared with that <strong>of</strong> IIS-Mine. Figure<br />
9(d) also shows that when the minimum support is less than 30%, the gap <strong>of</strong> the graphs clearly<br />
widens, which confirms that FP-growth consumes more memory than IIS-Mine. In both figures, FPgrowth<br />
consumes a great deal more memory when the minimum support is low because FP-growth<br />
has constructed large FP-trees for mining all frequent itemsets, whereas IIS-Mine uses the property<br />
<strong>of</strong> extendable itemsets, which perform better for dense datasets. Consequently, the recursion <strong>of</strong>
148 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
mining frequent itemsets in the next loops is reduced. Therefore, the construction <strong>of</strong> nodes and the<br />
sizes <strong>of</strong> the conditional item-trees are reduced.<br />
Scalability<br />
The scalability <strong>of</strong> the algorithms was tested by running them on datasets generated from<br />
T10I4 and T40I10. The number <strong>of</strong> transactions in the datasets ranged from 20K to 100K, where K is<br />
1000 transactions. In Figure 10(a) and Figure 11(a), the algorithms were run on all <strong>of</strong> the datasets<br />
generated from T10I4 at a minimum support <strong>of</strong> 1%. Both run time and memory consumption were<br />
recorded. Figure 10(a) shows the speed scalability, which means that the number <strong>of</strong> transactions<br />
increases as the run time increases. Figure 11(a) shows the memory scalability <strong>of</strong> the algorithms; the<br />
curve <strong>of</strong> FP-growth is over that <strong>of</strong> IIS-Mine, which means that FP-growth consumes more memory<br />
than IIS-Mine. The figure also shows that the memory consumption <strong>of</strong> the algorithms increases<br />
linearly with the size <strong>of</strong> the datasets.<br />
In Figures 10(b) and 11(b), the algorithms were run on all <strong>of</strong> the datasets generated from<br />
T40I10 at a minimum support <strong>of</strong> 5%. Both run time and memory consumption were recorded.<br />
Figure 10(b) confirms that the run time <strong>of</strong> both algorithms relies on the length and number <strong>of</strong><br />
transactions: if the length or number <strong>of</strong> transactions increases, so does the run time <strong>of</strong> both<br />
algorithms. However, the run time <strong>of</strong> IIS-Mine was better than that <strong>of</strong> FP-growth for every number<br />
<strong>of</strong> transactions. Figure 11(b) confirms that the memory consumption <strong>of</strong> the algorithms increases with<br />
the length and number <strong>of</strong> transactions. However, the memory consumption <strong>of</strong> IIS-Mine was better<br />
than that <strong>of</strong> FP-growth for every number <strong>of</strong> transactions.<br />
Run time (s)<br />
80<br />
60<br />
40<br />
20<br />
0<br />
T10I4<br />
IIS-Mine<br />
FP-growth<br />
Run time (s)<br />
4000<br />
3000<br />
2000<br />
1000<br />
0<br />
T40I10<br />
IIS-Mine<br />
FP-Growth<br />
20 40 60 80 100<br />
20 40 60 80 100<br />
Number <strong>of</strong> transactions (K)<br />
Number <strong>of</strong> transactions (K)<br />
(a)<br />
(b)<br />
Figure 10. Scalability <strong>of</strong> runtime on a) T10I4; b) T40I10<br />
Main memory(K)<br />
4000000<br />
3000000<br />
2000000<br />
1000000<br />
0<br />
20 40 60 80 100<br />
Number <strong>of</strong> transactions (K)<br />
T10I4<br />
IIS-Mine<br />
FP-growth<br />
Main memory(K)<br />
15000000<br />
10000000<br />
5000000<br />
0<br />
20 40 60 80 100<br />
Number <strong>of</strong> transactions (K)<br />
T40I10<br />
IIS-Mine<br />
FP-growth<br />
(a)<br />
(b)<br />
Figure 11. Scalability <strong>of</strong> memory consumption on: a) T10I4; b) T40I10
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
149<br />
CONCLUSIONS<br />
A data structure called inverted index structure (IIS) can store transaction data by scanning a<br />
database only once. Changing the minimum support does not affect the IIS and rescanning <strong>of</strong> the<br />
database is not required. A new algorithm called IIS-Mine can find frequent itemsets without<br />
generating candidate itemsets. It employs a more efficient use <strong>of</strong> the extendable-itemset property to<br />
reduce the number <strong>of</strong> recursive steps <strong>of</strong> mining. The node construction and the size <strong>of</strong> trees are then<br />
reduced, thereby reducing the run time and memory consumption. Although the proposed method<br />
accesses the IIS structure multiple times, experimental results demonstrated that for dense datasets<br />
the IIS-Mine algorithm is better than FP-growth algorithm in run time and space consumption.<br />
ACKNOWLEDGEMENTS<br />
This work was supported by the National Centre <strong>of</strong> Excellence in Mathematics, PERDO,<br />
Office <strong>of</strong> the Higher Education Commission, Thailand.<br />
REFERENCES<br />
1. J. Han and M. Kamber, “Data Mining: Concepts and Techniques”, Elsevier, Maryland Heights<br />
(MO), 2006, pp.227-231.<br />
2. J. Han, J. Pei, Y. Yin and R. Mao, “Mining frequent patterns without candidate generation: A<br />
frequent-pattern tree approach”, Data Mining Knowl. Discov., 2004, 8, 53-87.<br />
3. G. Grahne and J. Zhu, “Fast algorithms for frequent itemset mining using FP-Trees”, IEEE<br />
Trans. Knowl. Data Eng., 2005, 17, 1347-1362.<br />
4. R. Agrawal, T. Imielinski and A. Swami, “Mining association rules between sets <strong>of</strong> items in<br />
large databases”, Proceedings <strong>of</strong> ACM SIGMOD <strong>International</strong> Conference on Management <strong>of</strong><br />
Data, 1993, Washington, DC, USA, pp.207-216.<br />
5. R. Agrawal and R. Srikant, “Fast algorithms for mining association rules”, Proceedings <strong>of</strong> 20th<br />
<strong>International</strong> Conference on Very Large Data Bases, 1994, Santiago de Chile, Chile, pp.487-<br />
499.<br />
6. T. H. N. Vu, J. W. Lee and K. H. Ryu, “Spatiotemporal pattern mining technique for locationbased<br />
service system”, ETRI J., 2008, 30, 421-431.<br />
7. J. Han, J. Pei and Y. Yin, “Mining frequent patterns without candidate generation”, Proceedings<br />
<strong>of</strong> ACM SIGMOD <strong>International</strong> Conference on Management <strong>of</strong> Data, 2000, Dallas (TX), USA,<br />
pp.1-12.<br />
8. J. Pei, J. Han, H. Lu, S. Nishio, S. Tang and D. Yang, “H-mine: Hyper-structure mining <strong>of</strong><br />
frequent patterns in large databases”, Proceedings <strong>of</strong> IEEE <strong>International</strong> Conference on Data<br />
Mining, 2001, San Jose (CA), USA, pp.441-448.<br />
9. A. Pietracaprina and D. Zandolin, “Mining frequent itemsets using Patricia tries”, Proceedings<br />
<strong>of</strong> 3rd IEEE <strong>International</strong> Conference on Data Mining, 2003, Melbourne (FL), USA.<br />
10. G. Grahne and J. Zhu, “Efficiently using Prefix-Trees in mining frequent itemsets”, Proceedings<br />
<strong>of</strong> 3rd IEEE <strong>International</strong> Conference on Data Mining, 2003, Melbourne (FL), USA.
150 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
11. Q. Zhu and X. Lin, “Depth first generation <strong>of</strong> frequent patterns without candidate generation”,<br />
Proceedings <strong>of</strong> 11th Pacific-Asia Conference on Knowledge Discovery and Data Mining, 2007,<br />
Nanjing, China, pp.378-388.<br />
12. S. Sahaphong and V. Boonjing, “The combination approach to frequent itemsets mining”,<br />
Proceedings <strong>of</strong> 3rd <strong>International</strong> Conference on Convergence and Hybrid Information<br />
Technology, 2008, Busan, Korea, pp.565-570.<br />
13. V. K. Shrivastava, P. Kumar and K. R. Padasani, “FP-tree and COFI based approach for mining<br />
<strong>of</strong> multiple level association rules in large database”, Int. J. Comp. Sci. Inf. Secur., 2010, 7,<br />
273-279.<br />
14. L. Huang, J. Z. Liang, Y. Pan and Y. Xian, “A complete attribute reduction algorithm based on<br />
improved FP tree”, Proceedings <strong>of</strong> <strong>International</strong> Conference on Circuits, Communications and<br />
System, 2010, Beijing, China, pp.1-4.<br />
15. U. Yun and K. H. Ryu, “Approximate weighted frequent pattern mining with/without noisy<br />
environments”, Knowl.-Based Syst., 2011, 24, 73-82.<br />
16. M. J. Zaki, “Scalable algorithms for association mining”, IEEE Trans. Knowl. Data Eng., 2000,<br />
12, 372-390.<br />
17. M. J. Zaki and K. Gouda, “Fast vertical mining using diffsets”, Proceedings <strong>of</strong> 9th ACM<br />
SIGKDD <strong>International</strong> Conference on Knowledge Discovery and Data Mining, 2003,<br />
Washington, DC, USA, pp.236-355.<br />
18. D. J. Chai, L. Jin, B. Hwang and K. H. Ryu, “Frequent pattern mining using bipartite graph”,<br />
Proceedings <strong>of</strong> 18th <strong>International</strong> Conference on Database and Expert Systems Applications,<br />
2007, Regensburg, Germany, pp.182-186.<br />
19. J. Dong and M. Han, “BitTableFI: An efficient mining frequent itemsets algorithm”, Knowl.-<br />
Based Syst., 2007, 20, 329-335.<br />
20. W. Yen, “A new mining algorithm based on frequent item sets”, Proceedings <strong>of</strong> <strong>International</strong><br />
Workshop on Knowledge Discovery and Data Mining, 2008, Adelaide, Australia, pp.410-413.<br />
21. W. Song, B. Yang and Z. Xu, “Index-BitTableFI: An improved algorithm for mining frequent<br />
itemsets”, Knowl.-Based Syst., 2008, 21, 507-513.<br />
22. S. Sahaphong and V. Boonjing, “Mining <strong>of</strong> frequent itemsets by using the property <strong>of</strong><br />
extendable-itemset”, Proceedings <strong>of</strong> 7th <strong>International</strong> Joint Conference on Computer <strong>Science</strong><br />
and S<strong>of</strong>tware Engineering, 2010, Bangkok, Thailand, pp.168-173.<br />
23. S. Sahaphong and G. Sritanratana, “Mining <strong>of</strong> frequent itemsets with JoinFI-Mine algorithm”,<br />
Proceedings <strong>of</strong> 10th WSEAS <strong>International</strong> Conference on Artificial Intelligence, Knowledge<br />
Engineering and Database, 2011, Cambridge, United Kingdom, pp.73-78.<br />
24. Frequent Itemset Mining Dataset Repository, “T10I4D100K”, http://fimi.cs.helsinki.fi/data/,<br />
2003 (Accessed: January 11, 2010).<br />
25. Frequent Itemset Mining Dataset Repository, “T40I10D100K”, http://fimi.cs.helsinki.fi/data/,<br />
2003 (Accessed: January 11, 2010).<br />
26. Workshop on Frequent Itemset Mining Implementations, http://fimi.ua.ac.be/fimi03/, 2003<br />
(Accessed: January 2, 2010).
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 130-151<br />
151<br />
27. Workshop on Frequent Itemset Mining Implementations, http://fimi.ua.ac.be/fimi04/, 2004<br />
(Accessed: January 2, 2010).<br />
28. UCI Machine Learning Repository, http://archive.ics.uci.edu/ml/, 2007 (Accessed: March 15,<br />
2010).<br />
29. UCI Machine Learning Repository, “Chess”, http://archive.ics.uci.edu/ml/datasets.html, 1989<br />
(Accessed: July 19, 2010).<br />
30. UCI Machine Learning Repository, “Mushroom”, http://archive.ics.uci.edu/ml/datasets.html,<br />
1987 (Accessed: July 19, 2010).<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.
152 <strong>Maejo</strong> <strong>Maejo</strong> Int. J. Int. Sci. J. Technol. Sci. Technol. <strong>2012</strong>, <strong>2012</strong>, 6(01), 6(01), 152-158 152-158<br />
Communication<br />
<strong>Maejo</strong> <strong>International</strong><br />
<strong>Journal</strong> <strong>of</strong> <strong>Science</strong> and Technology<br />
<strong>ISSN</strong> <strong>1905</strong>-<strong>7873</strong><br />
Available online at www.mijst.mju.ac.th<br />
Lipase-catalysed sequential kinetic resolution <strong>of</strong> α-lipoic acid<br />
Hong De Yan 1 , Yin Jun Zhang 1 , Li Jing Shen 2 and Zhao Wang 1, *<br />
1 College <strong>of</strong> Biological and Environmental Engineering, Zhejiang University <strong>of</strong> Technology, Hang Zhou<br />
310014, People’s Republic <strong>of</strong> China<br />
2 Jiaxing Vocational and Technical College, Jiaxing 314036, Zhejiang, People’s Republic <strong>of</strong> China<br />
* Corresponding author, e-mail: hzwangzhao@163.com<br />
Received: 16 April 2011 / Accepted: 18 April <strong>2012</strong> / Published: 30 April <strong>2012</strong><br />
Abstract: Lipase from Aspergillus sp. WZ002 was employed to kinetically resolve racemic<br />
α-lipoic acid by a sequential esterification process. Though the remoteness <strong>of</strong> this substrate’s<br />
stereocentre from the reaction centre provided a significant challenge, the introduction <strong>of</strong><br />
sequential kinetic resolution dramatically enhanced the lipase’s enantioselectivity for<br />
esterification at the terminal carbonyl group, producing the desired (R)-enantiomer virtually<br />
enantiomerically pure. The enantiomeric excess <strong>of</strong> the (R)-enantiomer increased from 52% in<br />
the first step to 92% for in second step.<br />
Keywords: esterification, lipase-catalysed reactions, α-lipoic acid, sequential kinetic<br />
resolution<br />
__________________________________________________________________________________<br />
INTRODUCTION<br />
(R)-α-Lipoic acid is a naturally occurring c<strong>of</strong>actor <strong>of</strong> several α-keto acid dehydrogenases and a<br />
growth factor for a variety <strong>of</strong> microorganisms [1-3]. It has also been reported that α-lipoic acid and its<br />
derivatives are highly active as an anti-oxidant [4], anti-inflammatory agent [5], anti-HIV [6] and antitumour<br />
[7]. Generally, the (R)-enantiomer is much more active than the (S)-enantiomer [8], which has<br />
fostered significant interest in stereoselective synthesis <strong>of</strong> the pure enantiomers. Chemical synthesis <strong>of</strong><br />
(R)- and (S)-α-lipoic acid has been achieved either from a ‘chiral pool’ starting material [9] or by<br />
asymmetric synthesis [10-11]. Alternative methods involving enzyme catalysis include Bakers’ yeast<br />
reduction [12], mono-oxygenase catalysis [13] and lipase-catalysed kinetic resolution [14-15]. α-Lipoic<br />
acid provides a significant challenge due to the remoteness <strong>of</strong> stereocentre located four carbon atoms<br />
away from the reaction centre (carboxylic group). Lipases from Aspergillus oryzae WZ007 [14] and
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 152-158<br />
153<br />
Candida rugosa [15] show enantionselectivity towards the (S)-enantiomer, leaving the target (R)-αlipoic<br />
acid in unreacted form. In this study, the tested lipases show opposite enantionselectivity towards<br />
the (R)-enantiomer (Table 1) although with a low enantiomeric excess. In order to improve the<br />
enantiomeric purity, a sequential kinetic resolution was undertaken (Scheme 1). Herein we report on a<br />
successful application <strong>of</strong> sequential biocatalytic resolution <strong>of</strong> (R)-α-lipoic acid with high enantiomeric<br />
purity.<br />
(CH 2 ) 4<br />
COOH<br />
S<br />
S<br />
(CH 2 ) 4<br />
First esterification<br />
COOH<br />
n-Octanol / Lipase<br />
S S<br />
(S) -1<br />
(CH 2 ) 4<br />
COOOct<br />
Alkaline hydrolysis<br />
(CH 2 ) 4<br />
COOH<br />
(RS) -1<br />
S S<br />
S S<br />
(R) -2 (R) -1<br />
S<br />
S<br />
(CH 2 ) 4<br />
COOH<br />
Alkaline hydrolysis<br />
S<br />
S<br />
(CH 2 ) 4<br />
COOOct<br />
Second esterification<br />
n-Octanol / Lipase<br />
(R) -1<br />
(R) -2<br />
Scheme 1. Sequential kinetic resolution <strong>of</strong> α-lipoic acid ((RS)-1) with lipase<br />
MATERIALS AND METHODS<br />
Chemicals and Reagents<br />
α-Lipoic acid (purity≥98.0%) was purchased from Fluka BioChemika (Switzerland). (R)-αlipoic<br />
acid (purity≥98.0%) was purchased from Aladdin Reagent (China). Porcine pancreas lipase (Type<br />
II) was purchased from Sigma (USA). Nov 435 (an immobilised lipase from Candida antartica) and<br />
Lip TL (an immobilised lipase from Thermomyces lanuginosus) were purchased from Novozymes<br />
(Denmark). Lipase from Penicillium expansum was purchased from Shenzhen Leveking Bioengineering<br />
(China). All other chemicals were obtained from commercial sources and were <strong>of</strong> analytical<br />
reagent grade.<br />
Production <strong>of</strong> Lipase from Aspergillus sp. WZ002<br />
The strain WZ002 <strong>of</strong> Aspergillus sp., isolated from carrion and conserved in our laboratory, was<br />
maintained on potato dextrose agar (PDA) medium. A sequence analysis revealed that the internal<br />
transcribed spacer (ITS) DNA sequence <strong>of</strong> the strain WZ002 (GenBank accession no. JQ670919)<br />
showed high similarity (100% homology) to 47 strains <strong>of</strong> Aspergillus sp. The strain WZ002 was<br />
therefore primarily identified as a strain <strong>of</strong> Aspergillus sp. The culture was grown aerobically at 30°C<br />
and 200 rpm for 48 hr in cell growth medium consisting <strong>of</strong> glucose (10 g/L), peptone (5 g/L), KH 2 PO 4<br />
(1 g/L), MgSO 4·7H 2 O (0.5 g/L), FeSO 4·7H 2 O (0.01 g/L), KCl (0.5 g/L) and olive oil (10 mL/L). After<br />
harvesting by filtration, cells were washed with 100 mM Tris-HCl buffer (pH 7.3) and then freeze-dried.<br />
The lyophilised microbial cells (intracellular lipase) were used to catalyse the esterification reaction.
154 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 152-158<br />
Screening <strong>of</strong> Lipase<br />
Lipase from Aspergillus sp. WZ002 (ASL), porcine pancreas lipase (PPL), lipase from<br />
Penicillium expansum (PEL), lipase from Candida antartica (Nov 435) and lipase from Thermomyces<br />
lanuginosus (Lip TL) were investigated in the catalysis <strong>of</strong> the esterification <strong>of</strong> α-lipoic acid with n-<br />
octanol [14]. The reaction mixture was made <strong>of</strong> α-lipoic acid (206 mg, 1 mmol), n-octanol (0.79 mL, 5<br />
mmol), heptane (20 mL) and an appropriate amount <strong>of</strong> lipase. The reaction mixture was shaken at 200<br />
rpm and after a specified time the reaction was quenched by removing enzyme particles through<br />
centrifugation. Unreacted α-lipoic acid was extracted with 0.5% NaHCO 3 and recovered with<br />
dichloromethane after acidification with 20% HCl. Dichloromethane was removed by vacuum<br />
distillation and the recovered α-lipoic acid was analysed by high performance liquid chromatography<br />
(HPLC).<br />
Effect <strong>of</strong> Time on ASL-catalysed Esterifying Reaction<br />
To investigate the effect <strong>of</strong> time on the conversion ratio and enantiomeric excess <strong>of</strong> (R)-1 at the<br />
first step <strong>of</strong> the esterification reaction (Scheme 1), the reaction mixture consisting <strong>of</strong> α-lipoic acid (206<br />
mg, 1 mmol), n-octanol (0.79 mL, 5 mmol), heptane (20 mL) and ASL (400 mg) were shaken at 200<br />
rpm and 40°C for 48, 60, 72 and 84 hr. Unreacted α-lipoic acid was extracted with 0.5% NaHCO 3 , and<br />
the solvent <strong>of</strong> the resulting upper organic phase was removed by vacuum distillation to enrich ester (R)-<br />
2, which was hydrolysed to the corresponding (R)-1 by alkaline hydrolysis. The resulting enriched (R)-1<br />
was subjected to HPLC analysis.<br />
Sequential Kinetic Resolution<br />
The esterification reaction was carried out at 200 rpm and 40°C on a shaker. The reaction<br />
mixture was made up <strong>of</strong> α-lipoic acid (206 mg, 1 mmol), n-octanol (0.79 mL, 5 mmol), heptane (20<br />
mL) and ASL (400 mg). The reaction was quenched after 84 hr. Unreacted α-lipoic acid was extracted<br />
with 0.5% NaHCO 3 and the solvent <strong>of</strong> the resulting upper organic phase was removed by vacuum<br />
distillation to enrich ester (R)-2. The ester was dissolved in 20 mL <strong>of</strong> 95% ethanol and mixed with 150<br />
mg <strong>of</strong> NaOH. The mixture was stirred for 6 hr at room temperature. The solvent was removed by<br />
vacuum distillation and the residue was partitioned with 10 mL <strong>of</strong> heptane and 20 mL <strong>of</strong> distilled water.<br />
The resulting aqueous phase was acidified by 20% HCl. The enriched (R)-1 was obtained by extraction<br />
with dichloromethane from the acidified solution.<br />
(R)-1 was then subjected to a second enzymatic resolution by the same procedure.<br />
Analysis<br />
An Agilent 1100 HPLC with a Chiralpak AS-H column (250 mm×4.6 mm, 5 μm, Daicel) was<br />
used to analyse the conversion ratio and enantiomeric excess <strong>of</strong> α-lipoic acid. Hexane:2-<br />
propanol:trifluoroacetic acid (97:3:0.1) was used as eluent at a flow-rate <strong>of</strong> 0.8 mL/min. Absorbance <strong>of</strong><br />
column effluent was monitored at 220 nm [14]. The two enantiomers <strong>of</strong> α-lipoic acid were identified in<br />
the HPLC chromatogram by their different retention times using optically pure (R)-α-lipoic acid as<br />
reference compound. The enantiomeric ratios (E) were calculated using the following equations [16]: E<br />
= ln[(1−c)(1−ee S )]/ln[(1−c)(1+ee S )], c = [(c 0 -c e )/c 0 ]×100%, ee S = [([S]−[R])/([S]+[R])]×100%, ee R =
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 152-158<br />
155<br />
[([R]−[S])/([S]+[R])]×100%, where c is the conversion ratio <strong>of</strong> reaction, c 0 is the initial amount <strong>of</strong><br />
racemic α-lipoic acid, c e is the amount <strong>of</strong> residual α-lipoic acid at the end <strong>of</strong> reaction, ee S is the<br />
enantiomeric excess <strong>of</strong> the residual α-lipoic acid, and [R] and [S] are the peak areas for (R)-α-lipoic acid<br />
and (S)-α-lipoic acid respectively.<br />
RESULTS AND DISCUSSION<br />
Screening <strong>of</strong> Lipase<br />
The results were summarised in Table 1. All lipases tested showed (R)-stereopreference and<br />
considerable conversion ratio, but the enantiomeric ratios (E) <strong>of</strong> the transformations were very low.<br />
After 84 hr <strong>of</strong> transformation with ASL (59.2% conversion), the ester (R)-2 was separated from the<br />
unreacted substrate (S)-1; then the ester (R)-2 was hydrolysed under alkaline conditions to yield the<br />
corresponding (R)-1 enantiomer (ee R 52.4%) (Scheme 1, Figure 1). PPL and PEL afforded 64.3% and<br />
57% ee S <strong>of</strong> the unreacted substrate ((S)-1) at 90.1% and 79.8% conversion respectively, and their E-<br />
values were close to 2.0. When immobilised Nov 435 and Lip TL were used, a high activity (>72%<br />
conversion after 1 hr) was observed, but the unreacted enantiomer ((S)-1) was obtained in poor<br />
enantiomeric excess, which was also shown in their E-values, suggesting that both enantiomers might<br />
easily enter into the catalytic active site <strong>of</strong> the enzymes, leading to little enantioselectivity by the<br />
enzymes [17]. Thus, ASL with the highest enantiomer selectivity (E = 3.4-3.6) was chosen as the<br />
biocatalyst in sequential esterification to resolve α-lipoic acid.<br />
Table 1. Kinetic resolution <strong>of</strong> α-lipoic acid by different lipases<br />
Enzyme<br />
Temp.<br />
(°C)<br />
Time<br />
(hr)<br />
c (%) ee S (%) Preferred<br />
enantiomer<br />
ASL (400mg) 40 60 47.6 38.0 R 3.5<br />
40 72 52.7 43.6 3.4<br />
40 84 59.2 54.4 3.6<br />
PPL (400 mg) 37 4 63.6 32.8 R 1.9<br />
37 5 71.0 43.9 2.1<br />
37 6 90.1 64.3 1.8<br />
PEL (1000 mg) 37 3 39.6 17.0 R 2.0<br />
37 4 59.2 33.1 2.1<br />
37 5 79.8 57.0 2.1<br />
Nov 435 (50 mg) 26 0.5 54.8 9.3 R 1.3<br />
26 1 77.6 20.0 1.3<br />
Lip TL (50 mg) 26 0.5 46.7 8.7 R 1.3<br />
26 1 72.5 17.2 1.3<br />
E
156 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 152-158<br />
Effect <strong>of</strong> Time on ASL-catalysed Esterifying Reaction<br />
Figure 1 plotted the time course for esterification <strong>of</strong> racemic α-lipoic acid ((RS)-1) by ASL. The<br />
conversion ratio increased significantly with increase in reaction time. On the contrary, enantiomeric<br />
excess <strong>of</strong> the hydrolysate <strong>of</strong> ester (R)-2 (ee R ), namely enriched (R)-1, decreased a little and ranged<br />
between 52.4-57%. So the reaction time <strong>of</strong> 84 hr, with 59.2% c and 52.4% ee R , was selected in order<br />
to obtain a higher yield <strong>of</strong> the target (R)-enantiomer.<br />
Conversion ratio /%<br />
65<br />
60<br />
55<br />
50<br />
45<br />
60<br />
56<br />
52<br />
48<br />
44<br />
ee R<br />
/%<br />
40<br />
40<br />
48 56 64 72 80 88<br />
Time /hr<br />
Figure 1. Time course for ASL-catalysed esterification <strong>of</strong> α-lipoic acid ((RS)-1) at the first resolution<br />
step: □ = conversion ratio; Δ = ee R <strong>of</strong> (R)-1)<br />
Sequential Kinetic Resolution <strong>of</strong> α-Lipoic Acid<br />
As shown in Scheme 1, racemic α-lipoic acid ((RS)-1) was first subjected to enzymatic<br />
resolution to give ester (R)-2. After hydrolysis <strong>of</strong> the ester (R)-2, the resulting (R)-1 was then subjected<br />
to the second enzymatic resolution at about 65% conversion to furnish ester (R)-2. Subsequent<br />
hydrolysis <strong>of</strong> this ester (R)-2 furnished (R)-1 with an average ee <strong>of</strong> 92%, based on chiral<br />
chromatographic analysis (Table 2). Both the enantioselectivity <strong>of</strong> the lipase and the conversion ratio<br />
were clearly enhanced in the second resolution step, which could possibly be explained by the faster<br />
reacting (R)-enantiomer furnishing the major substrate in the second step.<br />
Table 2. Second resolution <strong>of</strong> enriched (R)-1 by ASL<br />
Reaction<br />
time (hr)<br />
c (%) ee R (%)<br />
48 59.3 92.8<br />
60 65.6 91.6<br />
72 66.7 93.2<br />
Note: Reaction condition: a mixture <strong>of</strong> enriched (R)-1 (52% ee, 1 mmol), n-octanol (5 mmol),<br />
heptane (20 mL) and ASL (400 mg) was shaken at 200 rpm and 40°C.<br />
CONCLUSIONS<br />
A new and convenient way to prepare (R)-α-lipoic acid has been developed through a sequential<br />
kinetic resolution by Aspergillus sp. WZ002. Even though the substrate has a stereogenic centre four
<strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 152-158<br />
157<br />
carbon atoms away from the reaction site and a low E-value was exhibited in the first step, high<br />
enantiomeric purity (ee > 91%) <strong>of</strong> the target (R)-enantiomer was obtained at a high conversion ratio in<br />
the second step. The success <strong>of</strong> this method confirms the value <strong>of</strong> sequential kinetic resolution as an<br />
important approach in enzymatic resolution, especially in cases where a remote stereocentre is present.<br />
REFERENCES<br />
1. L. J. Reed, “Multienzyme complexes”, Acc. Chem. Res., 1974, 7, 40–46.<br />
2. L. J. Reed, B. G. Debusk, I. C. Gunsalus and C. S. Jr. Hornberger, “Crystalline α-lipoic acid: A<br />
catalytic agent associated with pyruvate dehydrogenase”, <strong>Science</strong>, 1951, 114, 93–94.<br />
3. H. Sigel, “The hydrophobic and metal-ion coordinating properties <strong>of</strong> α-lipoic acid––an example <strong>of</strong><br />
intramolecular equilibria in metal-ion complexes”, Angew. Chem. Int. Ed. Engl., 1982, 21, 389–<br />
400.<br />
4. L. Packer, E. H. Witt and H. J. Tritschler, “Alpha-lipoic acid as a biological antioxidant”, Free<br />
Rad. Biol. Med., 1995, 19, 227–250.<br />
5. Y. S. Cho, J. Lee, T-H. Lee, E. Y. Lee, K-U. Lee, J. Y. Park and H-B. Moon, “α-Lipoic acid<br />
inhibits airway inflammation and hyperresponsiveness in a mouse model <strong>of</strong> asthma”, J. Allergy<br />
Clin. Immunol., 2004, 114, 429–435.<br />
6. A. Baur, T. Harrer, M. Peukert, G. Jahn, J. R. Kalden and B. Fleckenstein, “Alpha-lipoic acid is an<br />
effective inhibitor <strong>of</strong> human immuno-deficiency virus (HIV-1) replication”, Klin Wochenschr.,<br />
1991, 69, 722–724.<br />
7. P. M. Bingham and Z. Zachar, “Lipoic acid derivatives and their use in treatment <strong>of</strong> disease”, Int.<br />
Patent, WO/2000/024734 (2000).<br />
8. I. C. Gunsalus, L. S. Barton and W. Gruber, “Biosynthesis and structure <strong>of</strong> lipoic acid derivatives”,<br />
J. Am. Chem. Soc., 1956, 78, 1763–1766.<br />
9. S. P. Chavan, C. Praveen, G. Ramakrishna and U. R. Kalkote, “Enantioselective synthesis <strong>of</strong> R-(+)-<br />
α and S-(–)-α-lipoic acid”, Tetrahedron Lett., 2004, 45, 6027–6028.<br />
10. T. T. Upadhya, M. D. Nikalje and A. Sudalai, “Asymmetric dihydroxylation and hydrogenation<br />
approaches to the enantioselective synthesis <strong>of</strong> R-(+)-α-lipoic acid”, Tetrahedron Lett., 2001, 42,<br />
4891–4893.<br />
11. S. Zhang, X. Chen, J. Zhang, W. Wang and W. Duan, “An enantioselective formal synthesis <strong>of</strong> (+)-<br />
(R)-α-lipoic acid by an L-proline-catalyzed aldol reaction”, Synthesis, 2008, 3, 383–386.<br />
12. A. S. Gopalan and H. K. Jacobs, “Bakers’ yeast reduction <strong>of</strong> alkyl 6-chloro-3-oxohexanoates:<br />
Synthesis <strong>of</strong> (R)-(+)-α-lipoic acid”, J. Chem. Soc. Perkin. Trans., 1990, 1, 1897–1900.<br />
13. B. Adger, M. T. Bes, G. Grogan, R. McCague, S. Pedragosa-Moreau, S. M. Roberts, R. Villa, P.<br />
W. H. Wan and A. J. Willetts, “The synthesis <strong>of</strong> (R)-(+)-lipoic acid using a monooxygenasecatalyzed<br />
biotransformation as the key step”, Bioorg. Med. Chem., 1997, 5, 253–261.<br />
14. H. D. Yan, Z. Wang and L. J. Chen, “Kinetic resolution <strong>of</strong> α-lipoic acid via enzymatic differentiation<br />
<strong>of</strong> a remote stereocenter”, J. Ind. Microbiol. Biotechnol., 2009, 36, 643–648.<br />
15. N. W. Fadnivas and K. Koteshwar, “Remote control <strong>of</strong> stereoselectivity: Lipase catalyzed<br />
enantioselective esterification <strong>of</strong> racemic α-lipoic acid”, Tetrahedron Asym., 1997, 8, 337–339.
158 <strong>Maejo</strong> Int. J. Sci. Technol. <strong>2012</strong>, 6(01), 152-158<br />
16. C. S. Chen, Y. Fujimoto, G. Girdaukas and C. J. Sih, “Quantitative analyses <strong>of</strong> biochemical kinetic<br />
resolutions <strong>of</strong> enantiomers”, J. Am. Chem. Soc., 1982, 104, 7294–7299.<br />
17. O. Jiménez, M. P. Bosch and A. Guerrero, “Lipase-catalyzed enantioselective synthesis <strong>of</strong> methyl<br />
(R)- and (S)-2-tetradecyloxiranecarboxylate through sequential kinetic resolution”, J. Org. Chem.,<br />
1997, 62, 3496-3499.<br />
© <strong>2012</strong> by <strong>Maejo</strong> University, San Sai, Chiang Mai, 50290 Thailand. Reproduction is permitted for<br />
noncommercial purposes.