mycological research 113 (2009) 815–821
journal homepage: www.elsevier.com/locate/mycres
Morphology and development of Nigrosabulum globosum,
a cleistothecial coprophile in the Bionectriaceae (Hypocreales)
Marcie J. R. PLISHKA, Akihiko TSUNEDA, Randolph S. CURRAH*
Department of Biological Sciences, University of Alberta, Edmonton, Alberta T6G 2E9, Canada
article info
abstract
Article history:
Recent DNA sequence analyses indicated that Nigrosabulum globosum is a cleistothecial rep-
Received 11 September 2008
resentative of the Bionectriaceae in the Hypocreales, but morphological characters supporting
Received in revised form
this relationship are unknown. Using light and electron microscopy we followed the develop-
28 January 2009
ment of the ascomata of this species, from the formation of gametangia through to the de-
Accepted 18 February 2009
velopment of mature ascospores, and observed a series of characters that confirmed its
Published online 26 February 2009
hypocrealean affinities. These included the formation of a gel-filled centrum during early
Corresponding Editor:
stages of ascoma development, the subsequent appearance of hyaline peridial tissue en-
H. Thorsten Lumbsch
closed within a layer we interpret as representing a melanized uniloculate stroma, apically
derived paraphyses, and an ascogenous system that gives rise to asci that were both cylindri-
Keywords:
cal to clavate and globose. Ascospores, previously reported to be smooth, were ornamented
Ascomycota
with a honeycomb-like reticulum and were able to germinate within the ascoma. The carbo-
Evolution
naceous outer (stromatic) walls of the mature, grit-like cleistothecia indicate possible resis-
Pseudeurotiaceae
tance to UV radiation and desiccation. Furthermore, the complement of germinated
ascospores would enable mature ascomata to function as propagules that could quickly initiate new growth when transferred to fresh substrate. Our reexamination of N. globosum also
provides data that support the hypothesized close relationship with other bionectriaceous,
cleistothecial coprophiles, i.e., species of Hapsidospora, and Bulbithecium in particular.
Crown Copyright ª 2009 Published by Elsevier Ltd on behalf of The British Mycological Society. All
rights reserved.
Introduction
Nigrosabulum globosum is a cleistothecial species reported primarily from dung from Australia, Europe, Africa and North
America. Ascomatal structure and morphology, as portrayed
in the line drawings included in the description of the type
species (Malloch & Cain 1970), provide few hints that would
indicate a putative relationship to the Hypocreales as indicated
by recent studies based on DNA sequence analyses (Suh &
Blackwell 1999; Rossman et al. 2001). Detailed studies of its development, beyond the brief analysis accompanying the original description, have not been done. The species was
originally placed in the Pseudeurotiaceae, a family erected to accommodate cleistothecial genera having irregularly disposed,
evanescent asci and small, aporate ascospores (Malloch &
Cain 1970). Results of molecular phylogenetic analyses (Suh
& Blackwell 1999; Rossman et al. 2001), supported in some
cases by detailed morphological studies (Greif et al. 2004;
Plishka et al. 2008), showed that this family was heterogeneous and included taxa allied to both apothecial and perithecial lineages, i.e., the Leotiomycetes (e.g., Pseudeurotium and
Pleuroascus), the Sordariomycetes (e.g., Cryptendoxyla), and the
Hypocreales (e.g., Bulbithecium, Emericellopsis, Hapsidospora,
Mycoarachis and Nigrosabulum). These hypocrealean members
* Corresponding author. Tel.: þ1 780 492 7010; fax: þ1 780 492 9234.
E-mail address: rcurrah@ualberta.ca
0953-7562/$ – see front matter Crown Copyright ª 2009 Published by Elsevier Ltd on behalf of The British Mycological Society. All rights reserved.
doi:10.1016/j.mycres.2009.02.005
816
of the Pseudeurotiaceae form a distinct lineage within the
Bionectriaceae.
We recently reexamined an isolate of N. globosum that had
been obtained from carnivore (coyote) dung collected in
Alberta. However, we found it difficult to reconcile the morphology of the mature ascomata of our isolate with a position
among the Hypocreales. In this order, most species are perithecial and saprobic, or parasitic on plants and fungi rather than
cleistothecial and coprophilous. Hypocrealean fungi also generally have fleshy, light to brightly colored perithecia and exhibit a distinctive pattern of centrum development that
involves the formation of cylindrical asci that grow up
through a network of apical paraphyses that may disintegrate
at maturity. These features are not evident in squash mounts
of cleistothecia of N. globosum, which show only that the bilayered ascoma wall surrounds a mass of globose, apparently unordered ascospores, each often having one or two blunt
protuberances (Malloch & Cain 1970). Neither asci nor interascal tissue can be discerned reliably in such preparations, even
when younger specimens are examined.
Considering the putative phylogenetic position of N. globosum among the Hypocreales, we anticipated that a detailed
study of centrum development in the cleistothecia of this fungus would reveal a pattern typical of the order. Using a combination of light and electron microscopy we examined stages
in the formation of mature ascomata to derive data that could
be used to test this hypothesis.
Materials and methods
Nigrosabulum globosum (UAMH 4433) was grown on cornmeal
agar (CMA, Acumedia, Baltimore, MD), oatmeal agar (OA:
10 g oatmeal, 16 g agar, 1 L dH20) and tap water agar (TWA:
15 g agar, 1 L dH20) at 20 C under ambient lighting of the laboratory. Slide cultures were prepared using Pablum cereal
agar (CER: 25 g Mead Johnson Mixed Cereal Pablum (Mead
Johnson, Evansville, IN), 5 g agar, 250 ml dH20) and TWA. Samples were examined and photographed with an Olympus BX50
microscope with an Olympus DP12 camera. For fluorescence
microscopy, material was stained with 0.5 mg/mL Hoechst
33258 (Sigma–Aldrich, Canada), a bisbenzimide DNA intercalator that excites in the near UV and emits in the blue region,
in buffered saline (1.37 M NaCl, 36 mM KCl, 100 mM Na2HPO4,
17.6 mM KH2PO4, pH 7.4), sealed under a coverslip with nail
polish and kept at 5 C for 6 h before viewing with a Leica
DMRXA fluorescence microscope with a Leica HCX PL Fluotar
100/1.30 oil immersion objective. Images were prepared with
an Optronics DXM digital camera using UV laser illumination
with emission filters set at 425–475 nm (blue region).
For scanning electron microscopy (SEM), 5-mm disks of
agar bearing ascomata at different ages were cut from mature
colonies and fixed in unbuffered 2 % glutaraldehyde for 4 h,
placed in 2 % tannic acid–2 % guanidine hydrochloride overnight at 5 C, and postfixed in 2 % OsO4 for 3 h at room temperature. Fixed material was dehydrated in an ethanol series,
critical point dried in a Polaron E-3000 dryer using carbon dioxide, then coated with gold and examined at 10 kV with
a Hitachi S-510 electron microscope. For transmission electron microscopy (TEM), and thin sections observed by light
M. J. R. Plishka et al.
microscopy (LM), specimens were fixed in 2 % glutaraldehyde
and 2 % OsO4 in phosphate buffer at pH 7.3 and dehydrated in
an ethanol series. Samples were then embedded in Spurr’s
resin. Thin sections (0.5–1 mm) were stained with Toluidine
Blue and examined using the Olympus BX50 microscope (see
above). Ultrathin sections were stained with uranyl acetate
and lead citrate. Photomicrographs were taken at 75 kV with
a Hitachi H-7000 electron microscope.
Results
When transferred to CMA or OA, globose ascospores, 2.5–5.8 mm
in diameter, swelled and produced 1–3 germ tubes within 36 h
(Fig 1A). After 14 d at 22 C on these media, colonies were
glabrous to sparsely tufted and hyaline toward the periphery,
but olive green to charcoal towards the center where dense
aggregations of grit-like, melanized cleistothecia had formed.
Initiation of ascomata was first evident by the formation of
sessile to stalked, globose, lateral branches on prostrate
vegetative hyphae (Fig 1B). Subsequently, many of these initials
were each encircled by a swollen septate hypha (Fig 1C), which
then formed lateral buds that proliferated, branched and arched
over the enlarging primordium (Fig 1D, F). Relatively broad
hyphae arising from neighbouring vegetative mycelium
(Fig 1E) were also incorporated into the outer layers of the
primordium. As this structure continued to increase in
diameter, hyphae toward the periphery became closely septate
and formed a pseudoparenchymatous tissue of cells, 2–4 layers
deep, with thick melanized walls, while thin-walled cells
towards the interior swelled to fill the centrum (Fig 1F–G),
became vacuolated and deliquesced, leaving an amorphous
gel containing numerous pale yellowish, osmiophilic globules.
Hyaline branching hyphae developed from one side of the
centrum (Fig 1G–H) and formed a prosenchyma, 2–4(6) layers
thick, lining the peridium. A second prosenchyma, of relatively
thin-walled cells then arose from the opposite pole, presumably
representing the apical region, elongated, and filled the centrum
with a tissue composed of tightly packed, more or less parallel,
hyaline hyphae that were irregularly swollen between the septa
(Fig 2A–B).
Densely cytoplasmic ascogenous hyphae grew through
this prosenchyma and formed cylindrical to clavate asci
with eight ascospores that often appeared to be uni or biseriate in arrangement (Figs 2C–E, 3B), or globose to subglobose
(Fig 2C, E). With further maturation, asci and most of the peripheral and interascal prosenchyma tissues within the
thick-walled melanized outer cells of the ascoma deliquesced
leaving only a few remnants along with traces of ascus wall
material and delimiting membranes (Fig 3A–C, E–F).
By SEM and TEM, ascospores were globose to polyhedral,
up to 5.8 mm in diameter and with thick (0.1–0.6 mm) electron
lucent walls, and strongly osmiophilic contents (Fig 3A–C).
By LM, ascospores appeared smooth but both SEM and TEM
images showed a honeycomb-like network of ridges that
appeared to interdigitate between adjacent ascospores (Fig
3A–C). Fluorescence staining showed that ascospores each
had two nuclei, both positioned towards one side of the spore
(Fig 3D). By LM, some mature ascospores had one or two papilla-like structures protruding from the surface. Sections of
Morphology and development of Nigrosabulum globosum
817
Fig 1 – Ascoma development and early centrum structure in Nigrosabulum globosum (UAMH 4433). Fig C–H based on thin
sections stained with Toluidine Blue. (A) Light micrograph showing a globose ascospore that germinated (on agar surface)
forming 2 germ tubes. Unstained. Bar [ 12 mm. (B) Ascoma initials consisting of sessile to stalked, globose lateral branches.
Bar [ 7 mm. (C) Initials (antheridia) (arrowheads) encircled by swollen septate hyphae (ascogonia) (arrows) forming lateral
buds. Bar [ 7 mm. (D) Swollen hyphae branching and arching over the enlarging primordium. Bar [ 7 mm. (E) Broad hyphae
arising from vegetative mycelium that are being incorporated into the outer layers of the primordium. Bar [ 7 mm. (F) Coiled
gametangia in the center of the young ascoma, surrounded by several layers of pseudoparenchyma. Bar [ 7 mm. (G) A
developing ascoma showing thick-walled melanized cells of the ascomal wall surrounding thin-walled cells of the centrum
that are vacuolated and deliquescing (arrowheads), leaving an amorphous gel containing numerous lipid globules (arrow).
Bar [ 10 mm. (H) Thick-walled melanized cells of the ascomal wall surrounding thin-walled, deliquescing cells of centrum,
leaving an amorphous gel containing numerous lipid globules (arrow). Note the intact, densely cytoplasmic cells on one side
of the centrum (arrowheads). Bar [ 10 mm.
818
M. J. R. Plishka et al.
Fig 2 – Centrum development and ascosporogenesis in Nigrosabulum globosum (UAMH 4433). Fig A–E based on thin sections
stained with Toluidine Blue. (A) Young ascoma showing the development of the bilayered ascoma wall. Thick-walled hyaline
branching hyphae (arrow) are forming a prosenchyma lining the melanized outer layer of the wall, interpreted, respectively,
as the true peridium, and the surrounding stroma. A second prosenchyma, composed of thin-walled cells derived from apical
paraphyses, is beginning to develop in the centrum. Bar [ 10 mm. (B) Young ascoma showing the development of the
bilayered ascoma wall at a later stage. Prosenchyma of thin-walled cells is filling the centrum. Bar [ 10 mm. (C) Ascoma
showing two wall layers, and centrum prosenchyma. Asci are growing up through centrum prosenchyma. Bar [ 20 mm.
(D) Ascoma showing two wall layers. Asci are growing up between cells of centrum prosenchyma. Note some asci are
clavate (arrowheads). Bar [ 20 mm. (E) Ascoma showing two wall layers. Asci are growing up between cells of centrum
prosenchyma. Note that some asci are globose to subglobose in shape (arrowheads). Bar [ 20 mm.
older cleistothecia showed ascospores with well-developed
germ tubes that had grown into the remnants of tissue lining
the centrum and the peridium (Fig 3F). Remains of paraphyses
were sparse among the ascospores (Fig 3A, E–F).
Discussion
The mature, darkly pigmented, globose cleistothecia of Nigrosabulum globosum outwardly provide few structural clues to
a putative ancestry among the Hypocreales. Patterns of development associated with the centrum are compatible in most
respects with perithecial representatives of this order. Ascomatal initials were formed by the production of bulbous lateral branches that were encircled by swollen septate
hyphae. We interpreted these as a central antheridium and
an encircling ascogonium. These paired gametangia were
foci for the development of concentric masses of hyphae
that arose from the ascogonium or its immediate vicinity.
Similar initials occur in Bulbithecium hyalosporum (Udagawa &
Morphology and development of Nigrosabulum globosum
819
Fig 3 – Ascospores of Nigrosabulum globsum (UAMH 4433). Fig A SEM. Fig B–C TEM. Fig D Hoechst stained spores examined by
fluorescence microscopy (425–475 nm). Fig E–F based on thin sections stained with Toluidine Blue. (A) Ascospores with
honeycomb-like network of ridges on the surface. Some spores have germinated and formed germ tubes (arrowheads).
Arrow indicates remnants of centrum pseudoparenchyma. Bar [ 7 mm. (B) Ascospores with thick electron lucent wall and
osmiophilic contents. Spore wall has network of ridges that interdigitate between adjacent ascospores. Note remnants of
ascus wall (arrow) and biseriate arrangement of ascospores (stars). Bar [ 5 mm. (C) Ascospores with thick electron lucent wall
and osmiophilic contents. Spore wall has network of ridges that interdigitate between adjacent ascospores. Note: remnants
of spore delimiting membrane (arrowhead). Bar [ 1.7 mm. (D) Binucleate ascospores stained with Hoechst. Both nuclei positioned towards one side of the spore. Bar [ 7 mm. (E) Mature ascoma showing that the hyaline peridium lining the melanized stroma, centrum pseudoparenchyma, and asci have evanesced. Note ascospores appear unordered in the centrum.
Note structures between spores are remnants of centrum pseudoparenchyma (arrowhead). Bar [ 25 mm. (F) Mature ascoma
showing that the hyaline peridium lining the melanized stroma, centrum pseudoparenchyma, and asci have evanesced.
Ascospores have germinated, and germ tubes are penetrating the peridium (arrow). Note structures between spores are
remnants of centrum pseudoparenchyma (arrowhead). Bar [ 25 mm.
Muroi 1990), a monotypic genus phylogenetically close to
N. globosum (Suh & Blackwell 1999; Rossman et al. 2001).
As primordia differentiated they passed through a stage in
which central, lipid-rich cells disintegrated to form a gelatinous matrix that was subsequently replaced, first by basally
derived peridial tissue, and second by a dense mass of apically
derived paraphyses that grew downwards to fill the centrum
with a thin-walled prosenchyma. This preascogenous pattern
of development also occurs in Bionectria ochroleuca, and
Hydropisphaera peziza (Hanlin 1961, 1963; as Nectria gliocladioides and Neuronectria peziza respectively) both now assigned
to the Bionectriaceae (Suh & Blackwell 1999; Rossman et al.
2001).
While the central cells of the primordium were breaking
down, cells derived from vegetative hyphae making up the
outer wall of the ascoma became thick-walled and melanized.
This outer layer, which eventually became carbonaceous, appears to represent stromatic tissue that at later stages clearly
820
surrounded a peridium of basally derived, thick-walled hyaline prosenchyma. These two distinct strata, i.e., the dark
outer pseudoparenchyma and the hyaline inner prosenchyma, were considered originally to be constituents of
a bilayered peridium (Malloch & Cain 1970), but our observations of development, and the putative position of N. globosum
among the Bionectriaceae (Rossman et al. 2001), strongly suggest that the cleistothecium represents a uniloculate stroma
that encloses an ascoma with a distinct peridium. The stromatal nature of a similar pigmented outer layer was suggested
by Hanlin (1961) in the related but ostiolate species B. ochroleuca. A similar interpretation may also be applicable to the
melanized outer wall layers of the ascomata of Hapsidospora
irregularis, a bionectriaceous genus also placed originally in
the Pseudeurotiaceae, and to Bulbithecium hyalosporum. The hyaline peridium in N. globosum further supports its affiliation
with the Hypocreales which usually have light to brightly colored, rather than melanized, peridia.
Malloch & Cain (1970) depict asci of N. globosum originating
in the absence of crosiers in an almost palmate cluster at the
apex of an ascogenous hypha. From our observations, deeply
staining ascogenous hyphae presumably arose from the basal
portion of the ascoma in N. globosum and grew in between the
paraphyses that deliquesced as the asci developed. Asci
appeared to form in clusters at the apices of ascogenous hyphae with some elongating to become cylindrical to clavate
and others remaining globose, possibly because of space constraints within the crowded centrum. In some cases, asci were
so closely packed together that they appeared to have formed
in short chains. Wu & Kimbrough (1990) interpreted their TEM
sections of Emericellopsis microspora, a closely related genus in
the Bionectriaceae, to indicate that asci arose through ‘‘repetitive generation from crosiers.’’ We did not see crosiers in
N. globosum but the formation of asci in clusters near the
apices of ascogenous hyphae would explain why some asci
appeared butted up together in chain-like configurations.
Deliquescence of the ascus wall early in development left
the ascospores free within the centrum, where they were
intermingled with remains of the paraphyses and delimiting
membrane system. When squashed, mature cleistothecia did
not release intact eight-spored ascal clusters; ascospores often
appeared in pairs, or in short chains of three (or rarely four),
suggesting the presence of connecting material or mechanism.
SEM and TEM preparations showed that the ascospore wall
was thick, electron lucent and had a network of ridges and
pits or grooves that interdigitated between adjacent ascospores and these interlocking configurations may have been
sufficient to keep adjacent spores attached to each other.
Low to pronounced and wing-like ridges as in H. irregularis
and E. microspora, respectively, are common in ascospores of
the Bionectriaceae (Hanlin 1963, 1971; Malloch & Cain 1970;
Wu & Kimbrough 1990) and have also been observed to interdigitate when examined in TEM sections (Wu & Kimbrough
1990). The papillate protuberances on some ascospores of
N. globosum, observed here and by Malloch & Cain (1970) as
‘‘small particle-like attachments’’, are most likely the first
indications of the formation of germ tubes that are able to extend
into the tissues of the peridium and surrounding stroma
(Fig 3E–F). Germination in situ at least indicates that gut-passage
is not a required step in the dissemination of this coprophile.
M. J. R. Plishka et al.
Malloch & Cain (1970) noted that an Acremonium conidial
state, with single, long, tapering conidiophores bearing ellipsoid, fusoid, ovoid or allantoid phialospores in moist clusters,
was associated with at least some of the isolates they list under
examined material. Phialidic anamorphs, such as this typical
Acremonium state, are common in the Hypocreales and particularly in the Bionectriaceae but none was detected in UAMH 4433.
A second isolate that we examined briefly, UAMH 6868 from
elephant dung collected in Tanzania, produces an Acremonium
anamorph intermittently (UAMH, unpubl.).
Our morphological data support the hypothesis, advanced
on the basis of DNA sequence similarity, that there is a close
relationship between N. globosum and several other cleistothecial genera in the Bionectriaceae (i.e., Bulbithecium, Emericellopsis, Hapsidospora, Leucosphaerina, Mycoarachis) (Malloch & Cain
1970; Beliakova 1975; von Arx 1987; Udagawa & Muroi 1990).
Most isolates in these genera were obtained as coprophiles,
and with the exception of Aphysiostroma and Roumegueriella,
this is an otherwise rare category among the normally perithecial Hypocreales (Suh & Blackwell 1999; Samuels & Blackwell 2001). Presumably the rigors of the coprophilous
lifestyle resulted in a shift from perithecial to cleistothecial
morphology, possibly resulting from an interruption of the
normal growth of the apical meristem that generates the cells
of the neck, and eventually the ostiole (Rehner & Samuels
1995). The advantages conferred by the cleistothecial morphology could be related to the adoption of new dissemination
strategies. In Nigrosabulum it appears as though the cleistothecia rather than the ascospores could serve as the propagules.
The frequent observation of germinated ascospores, which
are binucleate, within the abundant melanized and grit-like
ascomata in N. globosum, may indicate that these reproductive
units not only have protection from UV and desiccation but
also are primed and ready to initiate growth when transferred
by some unknown means to fresh dung. The relative impermeability of the melanized stromatic layer was evident during
our attempts to fix whole ascomata for TEM.
In their recent critique of the value of the cleistothecium in
ascomycete systematics, Stchigel & Guarro (2007) state that
this type of ascoma is a poor taxonomic character above the
rank of genus. This maxim may be true when the character
gets only cursory consideration or is poorly known. We contend that detailed observations of anatomical and developmental features associated with cleistothecial ascomata can
yield useful information especially when applied in conjunction with phylogenetic data from DNA analyses. Further detailed and comparative studies of ascoma structure will
serve to strengthen hypotheses concerning the evolution of
form in the ascomycetes and the intricacies of their taxonomic relationships.
Acknowledgements
We are indebted to Ichiko Tsuneda and Ming Chen for their
technical assistance with the transmission electron microscopy.
Many thanks to Jocelyn C. Hall, Lynne Sigler and Stephen E.
Strelkov for comments on earlier versions of this manuscript.
This work was supported by a Discovery Grant to RSC from the
Natural Sciences and Engineering Research Council of Canada.
Morphology and development of Nigrosabulum globosum
references
von Arx JA, 1987. A re-evaluation of the Eurotiales. Persoonia 13:
273–300.
Beliakova LA, 1975. A new species of the genus Hapsidospora
(Pseudeurotiaceae). Mikologiya i Fitopathologia 9: 524–525 (in
Russian).
Greif MD, Tsuneda A, Currah RS, 2004. The peridial development
and dehiscence mechanism of Cryptendoxyla hypophloia,
a cleistothecial ascomycete isolated from the bodies of arthropods. International Journal of Plant Sciences 165: 957–964.
Hanlin RT, 1961. Studies in the genus Nectria. II. Morphology of N.
gliocladioides. American Journal of Botany 48: 900–908.
Hanlin RT, 1963. Morphology of Neuronectria peziza. American
Journal of Botany 50: 56–66.
Hanlin RT, 1971. Morphology of Nectria haematococca. American
Journal of Botany 58: 105–116.
Malloch D, Cain RF, 1970. Five new genera in the new family
Pseudeurotiaceae. Canadian Journal of Botany 48: 1815–1825.
Plishka MJR, Tsuneda A, Currah RS, 2008. Evidence of apothecial
ancestry in the cleistothecial ascomata of Pleuroascus nicholsonii. Mycological Research 112: 1319–1326.
821
Rehner SA, Samuels GJ, 1995. Molecular systematics of the
Hypocreales: a teleomorph gene phylogeny and the status
of their anamorphs. Canadian Journal of Botany 73:
S816–S823.
Rossman AY, McKemy JM, Pardo-Schultheiss RA, Schroers HJ,
2001. Molecular studies of the Bionectriaceae using large
subunit rDNA sequences. Mycologia 93: 100–110.
Samuels GJ, Blackwell M, 2001. Pyrenomycetes-fungi with perithecia. In: McLaughlin DJ, McLaughlin EG, Lemke PA (eds), The
Mycota VII Part A. Springer, Berlin, pp. 221–255.
Stchigel AM, Guarro J, 2007. A reassessment of cleistothecia
as a taxonomic character. Mycological Research 111:
1100–1115.
Suh SO, Blackwell M, 1999. Molecular phylogeny of the cleistothecial fungi placed in the Cephalothecaceae and Pseudeurotiaceae. Mycologia 91: 836–848.
Udagawa SI, Muroi T, 1990. Bulbithecium, a new genus of cleistocarpic coprophilous ascomycetes. Bulletin of the National Science
Museum, Tokyo, ser. B (Botany) 16: 13–19.
Wu CG, Kimbrough JW, 1990. Ultrastructural studies on the
cleistothecial development of Emericellopsis microspora
(Eurotiales, Ascomycetes). Canadian Journal of Botany 68:
1877–1887.