ELSEVIER Palaeogeography,Palaeoclimatology,Palaeoecology109 (1994) 345-356 Pollen evidence of late Quaternary vegetation and inferred climate changes in Congo H. Elenga a,b, D. Schwartz b, A. Vincens a " GOologie du Quaternaire CNRS, FacultO de Sciences de Luminy, Case 907, 13288 Marseille Cedex 9, France b Dynamique Historique des EcosystOrnes Intertropicaux, ORSTOM, B.P. 1286, Pointe Noire, Congo (Received May 6, 1993; revised and accepted September 1, 1993) Abstract A detailed palynological analysis of two cores recovered from a swamp in the Southern Bat6k6 Plateaux (Congo), provides information on the botanical history of this region during the last 24,000 yr B.P. Rich and well-diversified pollen counts exhibit changes in the vegetational communities related to hydrological and climatic fluctuations. Around 24,000 yr B.P., the occurrence of hydromorphous forest elements gives evidence of a humid phase. Between 24,000 and around 13,000 yr B.P., swampy herbaceous communities expanded, indicating slightly drier climatic conditions. From approximately 13,000 yr B.P. onward, the beginning of a humid episode is reflected by the development of mesophilous forests. The decrease of these forests is documented since 3000 yr B.P. Grasslands are found to extend locally whereas Elaies guineensis (oil palm) spreads out, indicating both climatic change but also anthropogenic activities. I. Introduction In West Equatorial Africa, from a large forested area included in the Guineo-Congolian region (White, 1983), very few pollen data are reported concerning the history of its vegetation and the climatic fluctuations during the Quaternary. The only continental pollen sequences studied in this region have been recovered from Cameroun: Lake Barombi Mbo (4°40'N, 9°24'E; Brenac, 1988; Maley et al., 1990a) and Mboandong (Richards, 1986). From Congo, in the absence of pollen data, the reconstruction of Quaternary vegetational environments has been primarily based on indirect arguments inferred from geomorphological, pedological and archaeological studies and on remains of macroflora (De Ploey, 1963, 1965; De 0031-0182/94/$7.00© 1994 ElsevierScienceB.V. All rights reserved SSDI 0031-0182(93)E0182-S Ploey and Van Morsel, 1963; Lanfranchi, 1979; Giresse et al., 1981; Delibrias et al., 1983; Schwartz, 1988; Dechamps et al., 1988a,b). The major climatic and vegetation changes of the last 40,000 yr based on these works are summarized as follow: 40,000-30,000 yr B.P.: humid period characterized by forest development; --30,000-12,000 yr B.P.: arid period with an extension o f wooded savanna; --12,000-3000 yr B.P. : humid period with a new forest development; ----ca. 3000 yr B.P.: the climate and vegetation reach their present characteristics. Palynological data obtained in the Bat6k6 region and on coastal swamps give new information for the dynamics of vegetational and floristical environments in Congo (Elenga, 1992). The occurrence of montane species on the Bat6k6 plateaux 346 H. Elengaet al./Palaeogeography,Palaeoclimatology,Palaeoecology109 (1994) 345-356 ca. 11,000 yr B.P. was interpreted as an effect of temperature drop (Elenga et al., 1991). Such migrations to the tropical West African lowlands of cold elements were recorded in other sites (Van Zinderen Bakker and Clark, 1962; Maley and Livingstone, 1983; Maley, 1987). Pollen data from the littoral suggest that a dry episode occurred ca. 3000 yr B.P., approximately synchronous with human impact (Elenga et al., 1992). This paper presents new palynological results from two cores collected in the Ngamakala Pond on the Batrk6 plateaux. The records give informations on vegetation and hydrological changes at local and regional scale, since 24,000 yr B.P. 2. The Bat~k~ Plateaux The Batrk6 Plateaux (1-4°S, 14-16°E) cover an area of 12,000 km 2 at an altitude between 600 and 886 m and are surrounded by a hilly area of 70,000 km 2 which ranges in altitude between 350 and 800 m (Fig. 1). The modern climatic conditions prevailing in this region are of the equatorial type. They are characterized, according to latitudinal and altitudinal location, by a mean annual rainfall comprised between 1300 mm (Brazzaville meteorological station) and 1900 mm (Djambala meteorological station), with a dry season reaching 3-4 months. The mean annual temperature ranges between 22 and 25°C with a mean annual thermal amplitude of 6°C (A. Sec. N.A., 1964). The Bat6k6 region is located within the mosaic of Guineo-Congolian lowland rain forest and secondary grassland (White, 1983). It is covered for 90% of its area by more or less wooded savannas. Locally, the occurrence of hygrophytic associations is related to local humid edaphic conditions (Duvigneaud, 1949; Descoings, 1969; Makany, 1976). The clumps of forest, sometimes well developed, are either of anthropogenic origin and characterized by the abundance of cultivated plants such as Elaeis guineensis (oil palm) and Mangifera indica (Mango), or forested galleries with Pentaclethra eetveldeana and Parinari excelsa. The occurrence of typical rainforest species such as Pycnanthus angolensis, Dacryodes edulis, etc., in the less damaged zones could indicate that some of these galleries might be the relics of a formerly more expanded forest environment. The hygrophytic associations are either grasslands with Sphagnum, Xyris, Stipularia africana and other aquatic herbs, or swamp forests with Syzygium guineensis, Alstonia boonei, Xylopia rubescens, etc. 3. Material and methods The Ngamakala Pond (4°4'30"S, 15°23'E, 400 m) is located within the hill area defined above. It is a swampy depression 750 m long and 200 m wide, largely covered by Sphagnum. The canopy cover is dominated by Alstonia boonei (Fig. 2). 3.1. Core sampling and laboratory analysis The cores Gama 4 and Gama 1 have been collected, using a Russian corer (Fig. 2; Belokupitov and Beresnevich, 1955). On each core, samples for pollen study were taken every 3 cm. Therefore, a total of 70 distinct levels have been counted in the two cores. The sediments were processed using the standard method described by Faegri and Iversen (1975), working with successive chemical attacks of HF, HC1 and KOH. Pollen and spores were identified by comparison with the reference pollen collection of the Laboratoire de Grologie du Quaternaire (Marseille, France: 7500 African tropical specimen). Each pollen spectrum includes at least twenty taxa and a total pollen sum averaging 400 grains. Such total counts are significant to characterize the vegetational associations found in the study of modem pollen rain of equatorial forests from Gabon (Jolly, 1987). The relative frequencies of pollen types or pollen groups are calculated on the total pollen sum excluding the unidentifiable grains. 3.2. Lithology and chronology of the deposits The sediments of both cores Gama 1 and Gama 4 mainly consist of fibrous peat with silty or clayey layers. The bottom of the core Gama 4 is dominated by compact silty clay abruptly replaced by H. Elenga et al./Palaeogeography, Palaeoclimatology, Palaeoecology 109 (1994) 345-356 347 \ @ ~EE~ Mesophilous 1 Swamp forest ~ Savanna with Savanna with o t 20 km I © roc GAMBOMA (377m) Prom DJAMBALA (803m) BRAZZAVILLE (314m 1131 2O I0 26°C 1770 mm JASONDJFMAM Fig. 1. (a) Location of the study area in Congo. (b) Major vegetation types of the Bat6k6 region and location of the Ngamakala Pond. (c) Climatic diagrams of the three meteorological stations of the studied area (Atlas du Congo, 1969). a humid silty peat (Fig. 3). Around 250cm, a more clayey peat occurs which is contemporaneous with the bottom deposits o f the core Gama 1 (Fig. 4). The upper part of the two cores consist of a fibrous peat containing abundant roots in the process of decomposition. In the cores Gama 1 (between 40 and 0 cm) and Gama 4 (between 140 and 0 cm) the deposits composed of plant remains were saturated of water. Therefore, it has been impossible to collect these levels. Eight standard radiocarbon datings made on total organic matter content have been performed on the cores Gama 4 and Gama 1 (Laboratoire de G6ologie du Quaternaire, Marseille). The dates obtained are presented in Table 1. These indicate that the sequence found in Gama 4 covers the period ca. 24,000 to ca. 3500 yr B.P., but the ~4C dates suggest a possible gap between 10,880 + 160 and 3940_+ 130 yr B.P. The Gama 1 core registers the period ca. 3300 to ca. 900 yr B.P. 348 H. Elenga et al./Palaeogeography, Palaeoclimatology, Palaeoecology 109 (1994) 345-356 Gama 4 Gama 1 (A) N .... .~ ,.~ ,,,!L.~~ .~ ............... .. ~ ii~i~:~!i ,: ~>.:i~~!.i .~..~ ~.....!~.!i~.;,i :.i!!~ii.!: ,- ii~i~!!!!!:~,!!)i,, i,!~.... W E S 2rnN ~ ' ~ S ~ ~ ~ 12,,, 1 2 3 4 S 6 7 e 9 lo 11 Fig. 2. The Ngamakala Pond (Makany, 1976) and location of the core sites Gama 1 and Gama 4. 1 =fluctuating limit of water level; 2=Loudetia demeusei savanna; 3=new growth with Pentaclethra eetveldeana; 4=clumps of Alstonia boonei; 5=Sphagnum peat; 6=pond; 7=Stipularia africana zone; 8=peatbog extension; 9=pond extension; 10=floating upper part of the peat; 11= lower part of the peat largely saturated in water. Table 1 List of 14C ages performed on total organic matter in cores Gama 4 and Gama 1, Ngamakala Pond (LGQ, Marseille) Core Gama 4 Gama 1 Material Depth (cm) ~4C ages (yr B.P.) Laboratory (no.) matter matter matter matter matter matter matter 140-150 165-175 180-190 190-200 210-221 258-270 290-300 3650 +__180 3950 +_130 10,880_ 160 13,260 + 220 14,090___230 24,000 + 580 24,200 +480 LGQ 244 LGQ 446 LGQ 553 LGQ 447 LGQ 405 LGQ 501 LGQ 242 Organic matter Organic matter 10-20 100-110 930 +_140 3300 + 130 LGQ 564 LGQ 495 Organic Organic Organic Organic Organic Organic Organic 4. Palynological results T h e p o l l e n d i a g r a m s p r e s e n t e d in Figs. 3 a n d 4 include o n l y the m a i n t a x a which are a b u n d a n t a n d consistently occurring, a n d therefore are considered as the m o s t i m p o r t a n t for the p a l a e o e n v i r o n m e n t a l i n t e r p r e t a t i o n . F o u r m a i n pollen zones ( G 4 to G 1 ) have been differentiated on the frequency fluctuations. T h e y are d e s c r i b e d f r o m the b o t t o m to the t o p o f the cores. 4.1. Pollen zone G4 This zone is represented only at the b o t t o m o f the core G a m a 4, between 300 a n d 285 cm. D a t e d a r o u n d 24,000 y r B.P., it is c h a r a c t e r i z e d b y the H. Elenga et al./Palaeogeography, Palaeoclimatology, Palaeoecology 109 (1994) 345-356 349 ,e- S~ 36~o ± 18o l;::~ ] G2 :.'7.. 3940 ± 130 J 10880± 160l ) 13260 ± 220 J == ?:?.-! 14090 ± 230 J i G3 ), ~o- 24000 ± 580 I i. ;...-. L. I= IG4 24200 + 480 J +<1% C (~) -~ ~ 6' .~;~f." ~,~ ,.~bo # .. ~..~+,~,,÷L..;yo. © ,~ .~ ±o~ov..- I~1 SYZYGIUM ~ II Or.E.S AP ~ [] UNDETERMINED NAP / FER.S i~o - 3650 + 180 I .,o. G2 3940 + 130 J 10880 + 160 I ;~T¢. ?!;' ri~"! !, ~3260 ± ~201 ~.:~;: iI i: ! i. i '/ ! . ..... 24200±48o I.o G3 , G4 I *<1% o lO% o 50 lOO% Fig. 3. Pollen diagram from core Gama 4, Ngamaka]a Pond (values are relative percentages of total pollen and spores, unidentifiables grains excluded). (A) Main arboreal and climber taxa. (B) Main herbaceous taxa and spores. (C) Synthetic diagram. 1t. Elenga et al./Palaeogeography, Palaeoclimatology, Palaeoecology 109 (1994) 345-356 350 xx~ ~4,v..& *~ _~. ® ~4: .~., . o ~ L ~ ¢ ; . ~ - ~ o~.4. ,v~ _~ ~,-,',~.¢: ~ ~_÷ v ~,~ / / | 40. 930 ± 140 I I ;./,- -- '~t".,: I l i d3 ' I ' ] , 33Q0:1:140 II~m I ' o i, I +o% *<1% ,r~ (, ~ ~,',# v+ _d ,., ,~ 4e" q..~,q" •l SYZYOIUM • ~] OTHIE. ~ [] UNC~TEI~AINED ~ NAP FEP4~I$ © •,r~; 1oo ) ~. , .re" 3300 ± 140 I,~a ' o lo% 0 • < 1% bad-decomposed peat ( f l ~ s t ) silty peat 50 100% clayey peat Fig. 4. Pollen diagram from core G a m a 1, N g a m a k a l a Pond (values are relative percentages of total pollen and spores, unidentifliables grains excluded). (A) Main arboreal and climber taxa. (B) Main herbaceous taxa and spores• (C) Synthetic diagram• H. Elenga et al./Palaeogeography, Palaeoclimatology, Palaeoecology 109 (1994) 345-356 dominance of arboreal taxa which represent more than 70% of the total pollen count. The most abundant ones are the Sapotaceae (35%), Syzygium (20%) associated with some Alchornea (5%0), Macaranga, Crudia type and Celtis (1-2%). The Combretaceae (10%), probably belonging to the genus Combretum as shown by the numerous botanical studies undertaken on this site (Makany, 1976), and Tetracera (less than 1%) represent the main climber components in the pollen spectra. Herbaceous taxa are very scarce. Pteridophyta represent mean percentages of about 3%. 4.2. Pollen zone G3 This zone, such as the previously described, is present only in the core Gama 4, between 285 and 190 cm. It is dated between ca. 24,000 and ca. 13,000 yr B.P. The vegetation shows a significant change. The Sapotaceae (20%) and Syzygium (10%) decrease. During the same time, arboreal taxa such as Cleistanthus (5%) and Celtis (5%) develop. The climbers are more abundant with always Combretaceae (15%) and Tetracera (1-2%) dominant. The herbaceous taxa increase, particularly the aquatic and swampy components such as Xyris (10%), Nymphaea lotus type (10%) and Laurembergia tetrandra. Gramineae and Spermacoce are present, but in low percentages (2%). 4.3. Pollen zone G2 This zone is present in the core Gama 4 between 190 and 140 cm, and only its upper part in the core Gama 1, between 150 and 140 cm. It corresponds to the period from ca. 11,000 to ca. 3000 yr B.P. and is characterized by a large representation of arboreal taxa, particularly of the Sapotaceae (45%) and Syzygium (35%) associated with some Alchornea and Celtis. The climbers (Combretaceae (5%) and Tetraeera (less than 1%)) greatly decrease such as the herbaceous taxa (less than 5%). 4.4. Pollen zone G1 It is represented only in the core Gama 1, between 140 and 40 cm and covers the episode 351 from ca. 3000 yr B.P. to the sub-actual period. It has been divided into two sub-zones: Sub-zone Gla It is characterized by an abrupt increase in the frequencies of herbaceous taxa with a mean value of about 60% of the total count. The most abundant components are the Gramineae (30%), the Cyperaceae (10%), Stipularia africana (10%), Laurembergia tetrandra and Xyris (2-5%). Pteridophyta are also well represented. The arboreal taxa significantly decrease with only 35%. Elaeis guineensis (oil palm) appears at the bottom of this sub-zone. Sub-zone Glb In this sub-zone, an increase of arboreal taxa such as Syzygium associated with Allophylus and Campylospermum is registered. 5. Major environmental changes and paleoclimatic implications Pollen analysis of the deposits recovered in the Ngamakala swamp, Bat6k6 region, clearly show that the dynamics of the vegetation occurring on this site during the late Quaternary was characterized by extension and retreat of the local forested environment. At approximately 24,000 yr B.P. an hydromorphous forest occupies a large area in the Ngamakala depression. The low representation of herbaceous elements, particularly of Gramineae which have a great local pollen production as demonstrated in modern pollen rain studies in Congo (Elenga, 1992), suggests that grassland was missing at the site. Such a palaeobotanical environment would indicate a humid climatic episode with precipitation certainly higher than today. Although these data have been obtained on a single site, it can be supposed that ca. 24,000 yr B.P. most of the Batik6 depressions were occupied by hydromorphous forests. Between ca. 24,000 and 13,000 yr B.P., pollen data show a retreat of the arboreal strata on the Ngamakala site giving way to hygrophytic grasslands rich in Xyris, Nymphaea lotus and some 352 H. Elenga et al./Palaeogeography, Palaeoclimatology, Palaeoecology 109 (1994) 345-356 other herbaceous components which were particularly well developed before ca. 14,000 yr B.P. During this period, the vegetation could be interpreted as a mosaic of swampy grasslands with free and open water indicated by local patches of forests, suggesting drier climatic conditions than before. Such a reconstruction is consistent with the modern distribution of the aquatic plants identified in the fossil spectra. Indeed, Nymphaea lotus can develop in temporary pools flooded only during four months a year. So, it is possible that between 24,000 and 13,000 yr B.P., a rainfall decrease coincided with a more contrasting seasonal distribution of precipitation. During the long dry season, the drop of the surface water level could have permitted the development of Xyris on the emerged sandy banks of the site, whereas Laurembergia tetrandra and Stipularia africana occupied more boggy soils. At the same time, the forest, rather open as confirmed by the presence of numerous climbers, could have remained in the lowest zones of the depression where water would have been present. Such occurrence of water in the Batrk6 depressions during a period interpreted as dry can be explained by the fact that these low zones would have taken advantage of the flow of underground water of which the circulation was favoured by the sandy nature of the substratum. After 13,000 yr B.P., pollen data give evidence of an expansion of the hydromorphous forests in the depression. The floristic composition is close to the one registered ca. 24,000 yr B.P. This new extension of forests corresponds to the onset of the Holocene climatic optimum now well known all over Africa and characterized by the establishment of higher precipitations than before. The succession of the dates 10,880___ 160 and 3940 + 130 yr B.P. indicates that early and middle Holocene deposits are not represented in core Gama 4, such as in the first record of the Kashiru swamp (Bonnefille and Riollet, 1988). From ca. 3000 yr B.P. onwards, a new retreat of the hydromorphous forests is registered but it is less important than during the period ca. 24,000 to ca. 13,000 yr B.P. The pollen spectra indicate at the same time a re-expansion of the hygrophytic grasslands. This period is marked by an increase of the Gramineae which has no equivalent in older deposits. This phenomenom, which could be related to human activities as shown also by the occurrence of Elaeis guineensis at the same time, will be discussed later. A new expansion of the forested area seems to take place ca. 900 yr B.P. This trend is today still observed in many Congolese regions, but it is largely slowed down by human pressure. 6. Discussion The humid episode dated at ca. 24,000 yr B.P. and characterized by an extension of forests in the swampy depressions of the Batrk6 region, is for the first time demonstrated in Congo. It would correspond to a humid phase within the Leopoldvillian arid period dated between ca. 30,000 and ca. 12,000 yr B.P. as recognized in West Central Africa (Lanfranchi and Schwartz, 1990). This wet phase was recently described in a pollen sequence from lake Barombi Mbo, Cameroun (Maley and Brenac, 1987), where ca. 24,000 yr B.P. high frequencies of arboreal taxa (more than 60%) indicate the forested nature of the vegetation. In East Tropical Africa, palynological data supply evidence of a similar humid climatic phase between 25,000 and 20,000 yr B.P. (Bonnefille and Riollet, 1988; Vincens, 1986, 1991 a,b). Also other approaches led to the recognition of this event. Limnological studies indicate that in this period many tropical African lakes have registered a high or intermediate water level (Street and Grove, 1979; Gasse et al., 1989). In the interior Zaire basin, Preuss (1990) describes phenomena of podsolization on alluvial deposits between 26,000 and 23,000 yr B.P. related to higher precipitation. Many data obtained in Tropical Africa seem to confirm the occurrence ca. 24,000 yr B.P. of a humid climatic fluctuation within the dry period described between 30,000 and 12,000 yr B.P. and contemporaneous to the last glacial period in the northern hemisphere. Ca. 11,000 B.P., contrary to pollen results previously obtained in the Bois de Bilanko depression located 30 km northward (Elenga et al., 1991 ), no montane element such as Podocarpus latifolius, Olea capensis or Ilex mitis is present in the vicinity H. Elenga et al./Palaeogeography, Palaeoclimatology, Palaeoecology109 (1994)345-356 of the Ngamakala Pond. Therefore, it can be supposed that the lowest limit for their development was situated between the altitude of Bois de Bilanko (700 m) and of Ngamakala (400 m). Such a downward migration to their modem altitudinal range has been interpreted as a drop of the mean annual temperature of about 4-5°C compared to modem values when applying a mean temperature gradient of 0.6°C/100 m displacement of the vegetation (Elenga et al., 1991). Considering the very short distance between the two sites and their morphological location, it seems possible that any other factor such as an important cloudiness has influenced this distribution, although this factor has been used to explain the modem occurrence of Podocarpus latifolius in one residual settlement in the Chaillu Massif (Congo) at 700 m in altitude (Maley et al., 1990b). A climatic optimum is reached ca. 10,000-9000 yr B.P. It is marked by the expansion of the humid rain forests on the Bosumtwi (Ghana) and Barombi Mbo (Cameroon) sites (Maley and Livingstone, 1983; Maley and Brenac, 1987; Maley, 1991 ) and, on the East African mountains, by the spread to high elevation up to 2500 m of the forest belt (Hamilton, 1982; Bonnefille et al., 1991; Jolly and Bonnefille, 1991). In marine sediments, the ~180 curve obtained on a core off the Congolese coast clearly registers an increase in precipitation (Giresse et al., 1982). In the Bat6k6 region, this warm and humid period is though not well documented. At appoximately 3000 yr B.P. a new establishment of dry climatic conditions is registered. This change has been previously assessed in Congo (Caratini and Giresse, 1979; Dechamps et al., 1988a,b; Schwartz et al., 1990a,b,c; Elenga et al., 1992; Schwartz, 1992) and in Cameroon (Richards, 1986; Brenac, 1988; Maley, 1991, 1992) at the same time. It was also detected in other African regions between ca. 4000 and ca. 2500 yr B.P. (Kendall, 1969; Vincens, 1987, 1989, 1993; Bonnefille and Riollet, 1988; Taylor, 1990; Jolly and Bonnefille, 1991; Mworia, 1991; Ssemmanda and Vincens, 1993; Jolly, 1993). In West Equatorial Africa, this dry phase ca. 3000 yr B.P., more abrupt and more important than the one previously recognized, may have led to local fragmen- 353 tation of the African forest. The modem evidences of such a fragmentation would be found in the isolated enclosed savannas in some forested areas of Congo (De Foresta, 1990; Schwartz et al., 1990c). This opening of the forest would have contributed to migration of the Bantu populations originating from the Cameroon Grassfields towards the South of the African continent (Schwartz, 1992). Moreover, the results obtained from the pollen sequences of the Ngamakala site indicate that the environment at ca. 3000 yr B.P. had no equivalent in the past 24,000 yr. Even during the glacial period interpreted as been the driest one, Gramineae were never as abundant as during the period 3000-950 yr B.P. It seems that the savannas found in the Batrk6 Plateaux today are not to be compared to the vegetation occurring there between 24,000 and 13,000 yr B.P., which was probably of a more wooded type. Such a conclusion has been previously proposed on the Congolese littoral (Elenga et al., 1992). 7. The environment from 3000 yr B.P. onwards: human interference on the landscape The fact that the modem savannas have no equivalent during the former late Quaternary dry climatic phases and the early occurrence of Elaeis guineensis (oil palm), an anthropic species probably originating from the Gulf of Guinea (Ghana-Nigeria) where its pollen grains have been identified in older sediments than those presented here (Sowunmi, 1981) raise the problem of the effect of human impact on the physionomy and the distribution of modern vegetation. Firstly, it is noticeable that in the Ngamakala pollen diagrams the appearance of oil palm pollen is not exactly synchronous with the increase of grains from grassland savanna, but closely follows it. Similar results have previously been obtained from the Congolese littoral (Elenga, 1992). The comparison between investigations on macroflora remains in situ (Dechamps et al., 1988a) and on pollen from peat deposits (Elenga et al., 1992) allows to place the beginning of the dry late Holocene phase at ca. 3100-3000 yr B.P. In return, 354 H. Elenga et al./Palaeogeography, Palaeoclimatology, Palaeoecology 109 (1994) 345-356 the oil palm only appears in this region between 2850 and 2700 yr B.P., maybe at the same time as ceramics, though for the m o m e n t there are some divergences on their dating (Denbow, 1990; Schwartz, 1992). It seems that on the Congolese littoral, as in the Bat6k6 region, the establishment of proto-agricultural populations, assimilated to Bantu-speaking groups, follows the start of the dry climatic event. Similar conclusions have been found in Rwanda by Van Grunderbeek et al. (1984). Two other arguments are in favor of a, climatic origin of the detected environmental change at ca. 3000 yr B.P. The first one is the synchronism of this event in as different regions as the Congolese littoral, the Bat6k6 plateaux and the Lake Barombi M b o in Cameroon (Schwartz, 1992; Maley, 1992). The second one is that this change has also been observed on hydromorphous podzols, chemically corresponding to very p o o r soils influenced by a fluctuating water table, conditions which are not favourable for agricultural activities (Schwartz, 1988; Schwartz et al., 1989). So, the hypothesis of an abrupt climatic change towards drier conditions ca. 3000 yr B.P., leading to a partial fragmentation of the African rain forest of which the Iron Age populations would have taken advantage by their migrations, must be taken in serious consideration (Schwartz, 1992). 8. Conclusion The pollen sequences presented in this paper provide new detailed information mainly concerning the evolution of the vegetation in the Bat6k6 swampy depressions. They are in agreement with the patterns previously suggested by studies on macroflora remains (Delibrias et al., 1983; Schwartz, 1988; Dechamps et al., 1988b) and with pollen record from marine cores of the Congolese coast (Caratini and Giresse, 1979; Bengo and Maley, 1991). As new evidence, we must emphasize the occurrence of a humid phase ca. 24,000 yr B.P., the decrease of forest cover and the development of a swampy herbaceous vegetation between 24,000 and 13,000 yr B.P. indicating slightly drier conditions but not arid as suggested by geomorphological and pedological data. It is clear that during the Late Glacial Maximum, forested areas were present in the Bat6k6 region at the southern limit of the Bat6k6 Plateaux, near the Congo river. The pollen data confirm that the last 3000 yr B.P. were the driest ones of the Holocene. Acknowledgements The work presented here is part of Elenga's thesis defended at Marseille (1992), in the Laboratoire de Gkologie du Quaternaire CNRS. The authors are i n d e b t e d to R. Bonnefille for laboratory support, advice and encouragement during this research and revision of the manuscript; we acknowledge M. Servant, G. Caball6 and C. de N a m u r for helpful comments and discussions. Radiocarbon dating was provided by R. Lafont (LGQ-Marseille). We thank G. Buchet, G. Riollet and M. Decobert for technical assistance; N. Buchet and R. Smadja for computer storage; J. Adams for the English translation. Financial support comes from C N R S and O R S T O M . References A.SEC.N.A. Aper~u sur le climat du Congo Brazzaville. Serv. M6t6orol. Brazzaville, 23 pp, Atlas du Congo, 1969. Le climat du Congo. ORSTOM/Min. Coop., Paris. Belokupitov, I.E. and Beresnevich, V.V., 1955. Giktorf's peat borers. Turf Prom., 8: 9-10. Bengo, M.D. and Maley, J., 1991. Analysesdes flux polliniques sur la marge sud du Golfe de Guin6e depuis 135 000 ans. C. R. Acad. Sci. Paris, Ser. 2, 313: 843-849. Bonnefille, R. and Riollet, G., 1988. The Kashiru pollen sequence (Burundi). Palaeoclimatic implications for the last 40,000 yrs B.P. in tropical Africa. Quat. Res., 30: 19-35. Bonnefille, R., Riollet, G. and Buchet, G., 1991. Nouvelle s6quence pollinique d'une tourbi~re de la crete Za'ire-Nil (Burundi). Rev. Palaeobot. Palynol., 67: 315-330. Brenac, P., 1988. Evolution de la v6g6tation et du climat dans l'Ouest Cameroun entre 25 000 et 11 000 ans B.P. Act. 10th Symp. APLF, Bordeaux, Trav. Sec. Sci. Tech. Inst. Fr. Pondichery, 25: 91-103. Caratini, C. and Giresse, P., 1979. Contribution palynologique la connaissance des environnementscontinentaux et marins du Congo h la fin du Quaternaire. C. R. Acad. Sci. Paris, Ser. D, 288: 379-382. Dechamps, R., Guillet, B. and Schwartz, D., 1988a. D6couverte d'une flore foresti~re mi-Holoc~ne (5800-3100 B.P.) con- 1-1. Elenga et al./Palaeogeography, Palaeoclimatology, Palaeoecology 109 (1994)345-356 servre in situ sur le littoral pontrnrgrin (R.P. du Congo). C. R. Acad. Sci. Paris, Ser. 2, 306: 615-618. Dechamps, R., Lanfranchi, R., Le Cocq, A. and Schwartz, D., 1988b. Reconstitution d'environnements quaternaires par l'rtude de macrorestes v~grtaux (Pays Bateke, R.P. du Congo). Palaeogeogr., Palaeoclimatol., Palaeoecol., 66: 33-44. De Foresta, H., 1990. Origine et 6volution des savanes intramayombiennes (R.P. du Congo). 1I. Apports de la botanique foresti~re. In: R. Lanfranchi and D. Schwartz (Editors), Paysages Quaternaires de l'Afrique Centrale Atlantique. ORSTOM, Paris, pp. 236-355. Delibrias, G., Giresse, P., Lanfranchi, R. and Le Cocq, A., 1983. Datations des drprts holorganiques quaternaires sur la bordure occidentale de la Cuvette Congolaise (R.P. du Congo); corrrlations avec les srdiments marins voisins. C. R. Acad. Sci. Paris, Ser. 2, 296: 463-466. Denbow, J., 1990. Congo to Kalahari: data and hypotheses about the political economy of the western stream of the Early Iron Age. Afr. Archaeol. Rev., 8: 139-175. De Ploey, J., 1963. Quelques indices sur l'rvolution morphologique et palroclimatique des environs du Stanley-Pool (Congo). Stud. Univ. Lovanium, Univ. Kinshasa 17, 16 pp. De Ploey, J., 1965. Position gromorphologique, gen~se et chronologie de certains drpbts superficiels du Congo occidental. Quaternaria, 7: 131-154. De Ploey, J. and Van Morsel, H., 1963. Contribution /t la connaissance chronologique et palrogrographique des gisements prrhistoriques des environs de Lropoldville (Congo). Stud. Univ. Lovanium, Mus. Prrhist., 19 pp. Descoings, B., 1969. Phytogrographie. Esquisse phytogdographique du Congo. In: Atlas du Congo, ORSTOM, Paris, 2 pp. Duvigneaud, P., 1949. Les savanes du Bas-Congo. Essai de phytosociologie topographique. Lejeunia 10, 192 pp. Elenga, H., 1992. Vrgrtation et climat du Congo depuis 24 000 ans B.P. Analyse palynologique de srquences srdimentaires du Pays Bateke et du littoral. Th~se Doct. Univ. AixMarseille 1II, 237 pp. Elenga, H., Vincens, A. and Schwartz, D., 1991. Prrsence d'616ments forestiers montagnards sur les Plateaux Batrk6 (Congo) au Plristoc~ne sup~rieur: nouvelles donnres palynologiques. Palaeoecol. Afr., 22: 239-252. Elenga, H., Schwartz, D. and Vincens, A., 1992. Changements climatiques et action anthropique sur le littoral congolais au cours de l'Holoc~ne. Bull. Soc. Grol. Fr., 163(1): 83-90. Faegri, K. and Iversen, J., 1975. Textbook of Pollen Analysis. Blackwell, Oxford, 295 pp. Gasse, F., Ledre, V., Massault, M. and Fontes, J.C., 1989. Lake Tanganyika in phase with the ocean during the last glaciation and deglaciation. Nature, 342: 57-59. Giresse, P., Lanfranchi, R. and Peyrot, B., 1981. Les terrasses alluviales en Rrpublique Populaire du Congo. Bull. ASSEQUA, 62/63 pp. 43-66. Giresse, P., Bongo-Passi, G., Delibrias, G. and Duplessy, J.C., 1982. La lithostratigraphie des srdiments hdmiprlagiques du delta profond du fleuve Congo et ses indications sur les 355 palroclimats de la fin du Quaternaire. Bull. Soc. Grol. Fr., 24(4): 803-815. Hamilton, A.C., 1982. Environmental History of East Africa. A Study of the Quaternary. Academic Press, London, 328 pp. Jolly, D., 1987. Reprrsentation pollinique des forrts sempervirentes du N.E. du Gabon. Mrm. DEA, Univ. Montpellier, 85 pp. (unpubl.). Jolly, D., 1993. Evolution et dynamique des 6cosyst~mes du Burundi. Pollen et statistique. Th~se Doct. Univ. AixMarseille II, 143 pp. Jolly, D. and Bonnefille, R., 1991. Diagramme pollinique d'un sondage holo~ne de la Kuruyange (Burundi, Afrique Centrale). Palaeoecol. Afr., 22: 265-274. Kendall, R.L., 1969. An ecological history of the lake Victoria Basin. Ecol. Monogr., 39: 121-176. Koechlin, J., 1961. La Vrgrtation des Savanes dans le Sud de la Rrpublique du Congo. ORSTOM, Paris, 310 pp. Lanfranchi, R., 1979. Recherches prrhistoriques dans la moyenne vallre du Niari (R.P. du Congo). Thrse 3i~me cycle, Univ. Paris I, 2 Vol., 675 pp. Lanfranchi, R. and Schwartz, D. (Editors), 1990. Paysages Quaternaires de l'Afrique Centrale Atlantique. In: Collect. Didactiques. ORSTOM, Paris, 535 pp. Makany, L., 1976. Vrgrtation des plateaux Trkr. Trav. Univ. Brazzaville 1,301 pp. Maley, J., 1987. Fragmentation de la for6t dense humide africaine et extension de biotopes montagnards au Quaternaire rrcent: nouvelles donnres polliniques et chronologiques. Implications palroclimatiques et biogrographiques. Palaeocol. Afr., 18: 307-334. Maley, J., 1991. The African rainforest vegetation and paleoenvironments during Late Quaternary. Climatic Changes, 19: 79-98. Maley, J., 1992. Commentaires sur la note de D. Schwartz. Mise en 6vidence d'une prjoration climatique entre ca. 2500 et 2000 B.P. en Afrique tropicale humide. Bull. Soc. Grol. Ft., 16(3): 363-365. Maley, J. and Brenac, P., 1987. Analyses polliniques prrliminaires du Quaternaire rrcent de l'Ouest Cameroun: raise en 6vidence de refuges forestiers et discussion des problrmes palroclimatiques. Mrm. Trav. E.P.H.E., Inst. MontpeUier, 17: 129-142. Maley, J. and Livingstone, D.A., 1983. Extension d'un 616ment montagnard dans le Sud du Ghana (Afrique de l'Ouest) au Plristoc~ne sup~rieur et /~ l'Holoc~ne infrrieur: premieres donnres polliniques. C. R. Acad. Sci. Paris, Ser. 2, 296: 1287 1292. Maley, J., Livingstone, D.A., Giresse, P., Thouveny, N., Brenac, P., Kelts, K., Kling, G., Stager, C., Haag, M., Fournier, M., Bandet, Y., Williamson, D. and Zogning, A., 1990a. Lithostratigraphy, volcanism, paleomagnetism and palynology of Quaternary lacustrine deposits from Barombi Mbo (West Cameroon): preliminary results. J. Volcanol. Geotherm. Res., 42:319 335. Maley, J., Caball6, G. and Sita, P., 1990b. Etude d'un peuplement rrsiduel/t basse altitude de Podocarpus latifolius sur le flanc congolais du massif du Chaillu. Implications 356 H. Elenga et aL/Palaeogeography, Palaeoclimatology, Palaeoecology 109 (1994) 345-356 palroclimatiques et biogrographiques. Etude de la pluie pollinique actuelle. In: R. Lanfranchi and D. Schwhartz (Editors), Paysages Quaternaires de l'Afrique Centrale Atlantique. ORSTOM, Paris, pp. 336-352. Mworia, M., 1991. Vegetation response to climatic change in Central Rift Valley, Kenya. Quat. Res., 35: 234-245. Preuss, J., 1990. L'rvolution des paysages du bassin intrrieur du Za'fre pendant les quarante derniers millrnaires. In: R. Lanfranchi and D. Schwartz (Editors), Paysages Quaternaires de l'Afrique Centrale Atlantique. ORSTOM, Paris, pp. 260-270. Richards, K., 1986. Preliminary results of pollen analysis of a 6000 year core from Mboandong, a crater lake in Cameroon. Hull Univ. Geogr. Dep., Misc. Set., 32: 14-28. Schwartz, D., 1988. Some podzols on Bateke sands and their origins, People's Republic of Congo. Geoderma, 43(2/3): 229-247. Schwartz, D., 1992. Ass~chement climatique vers 3000 B.P. et expansion Bantu en Afrique centrale atlantique: quelques reflexions. Bull. Soc. Grol. Ft., 163(3): 353-361. Schwartz, D., Dechamps, R. and Guillet, B., 1989. Une flore holoc~ne (8000 B.P.) drcouverte h Ngidi (Congo). Nsi, 5: 9-14. Schwartz, D., De Eoresta, H., Dechamps, R. and Lanfranchi, R., 1990a. Drcouverte d'un premier site de l'glge du fer ancien (2110 B.P.) dans le Mayombe congolais. Implications palrobotaniques et prdologiques. C. R. Acad. Sci. Paris, Ser. 2, 310: 1293-1298. Schwartz, D., Guillet, B. and Dechamps, R., 1990b. Etude de deux flores foresti&es mi-Holoc~ne (6000-3000 B.P.) et subactuelle (500 B.P.) conservres in situ sur le littoral pontrnrgrin (Congo). In: R. Lanfranchi and D. Schwartz (Editors), Paysages Quaternaires de l'Afrique Centrale Atlantique. ORSTOM, Paris, pp. 283-297. Schwartz, D., Lanfranchi, R. and Mariotti, A., 1990c. Origine et 6volution des savanes intramayombiennes (R.P. du Congo). I. Apports de la prdologie et de la biogrochimie isotopique (14C et 13C). In: R. Lanfranchi and D. Schwartz (Editors), Paysages Quaternaires de l'Afrique Centrale Atlantique. ORSTOM, Paris, pp. 314-325. Sowunmi, M.A., 1981. The late quaternary environmental changes in Nigeria. Pollen Spores, 23(1): 125-148. Ssemmanda, I. and Vincens, A., 1993. Vrgrtation et climat dans le bassin du lac Albert (Ouganda, Za'ire) depuis 13 000 ans B.P. Apport de la palynologie. C. R. Acad. Sci., Paris, Ser. 2, 316: 561-567. Street, F.A. and Grove, A.T., 1979. Global maps of lake-level fluctuations since 30 000 yr B.P. Quat. Res., 12:83-118. Taylor, D.M., 1990. Late Quaternary pollen records from two Uganda mires: evidence for environmental change in the Rukiga Highlands of Southwest Uganda. Palaeogeogr., Palaeoclimatol., Palaeoecol., 80: 283-300. Van Grunderbeek, M.C., Doutrelepont, H. and Roche, E., 1984. Influence humaine sur le milieu au Rwanda et au Burundi h l'$ge du fer ancien (220-665 A.D.). Apports de la palynologie et de l'rtude de charbons de bois, Rev. Palrobiol., Vol. Sprc., pp. 221-229. Van Zinderen Bakker, E.M. and Clark, J.D., 1962. Pleistocene climates and cultures in North-Eastern Angola. Nature, 196: 639-642. Vincens, A., 1986. Diagramme pollinique d'un sondage plristoc~ne suprrieur-holockne du lac Bogoria (Kenya). Rev. Palaeobot. Palynol., 47: 169-192. Vincens, A., 1987. La srdimentation plristoc~ne suprrieur-holoc~ne. Pollens et spores: indices de l'rvolution climatique. In: J.J. Tiercelin and A. Vincens (Coordinators), Le Demigraben de Baringo-Bogoria, Rift Gregory, Kenya. 30 000 Ans d'Histoire Hydrologique et Srdimentaire. Bull. Cent. Rech. Explor. Prod. Elf-Aquitaine, 11:437 451. Vincens, A., 1989. Palroenvironnements du bassin NordTanganyika (Zaire, Burundi, Tanzanie) au cours des 13 derniers mille ans: apport de la palynologie. Rev. Palaeobot. Palynol., 61: 69-88. Vincens, A., 1991a. Late Quaternary vegetation history of the South-Tanganyika basin. Climatic implications in South Central Africa. Palaeogeogr., Palaeoclimatol., Palaeoecol., 86: 207-226. Vincens, A., 199lb. Vrgrtation et climat dans le bassin SudTanganyika entre 25 000 et 9000 ans B.P. : nouvelles donnres palynologiques. Palaeoecol. Afr., 22: 253-263. Vincens, A., 1993. Nouvelle srquence pollinique du lac Tanganyika. 30 000 ans d'histoire botanique et climatique du Bassin Nord. Rev. Palaeobot. Palynol., 78: 381-394. White, F., 1983. The Vegetation Map of Africa. UNESCO, Paris, 356 pp.