Quaternary History

Despite recent progress and new fossil pollen sequences obtained by coring swamps and lakes in Africa, it is not possible to reconstruct the Quaternary history of all the different forest units described in Section 5.2. Not all of them being documented yet, the presentation of fossil pollen records follows a chronological order, starting with the last glacial period, and then discussing the last 10,000 years of the Holocene. The list of sites where fossil pollen are available is given in Table 5.1 and Figure 5.1. Evidence for climate changes based upon other sources of information—such as lake sediments, paleosols, stable isotopic, phytoliths, diatoms—has been summarized in Battarbee et al. (2004).

5.4.1 Ice-age record

Marine sediments have provided pollen data related to vegetation change on the continent. These sequences have the great advantage of providing land-sea linkage on a straightforward isotopic chronology (Bengo and Maley, 1991; Dupont and Wienelt,

Table 5.1. List of fossil pollen sites located within the African lowland rainforest.



Elevation Rainfall (m) (mm/yr)


Barombi Mbo

4°67' N

, 09°40'E



Maley and Brenac (1998),


4°30' N

, 09°20'E



Richards (1986)

Ossa (OW4)

3°40' N

, 10°05'E



Reynaud-Farrera et a/. (1996)


0°43' S,




Ngomanda et a/. (2005)


3°31' S,




Elenga et a/. (1991)


3°50' S,




Vincens et a/. (1994, 1998)


4°00' S,




Elenga et a/. (1992)


4° 15' S,




Elenga et a/. (1996)


4°04' S,




Elenga et a/. (1994)






Elenga et a/. (2001)

1996; Dupont and Behling, 2006). Offshore transported pollen include taxa from different vegetation zones that are mixed together and can hardly be interpreted in terms of past expansion of each of the different forest types. Indeed, marine records are the only source of information for older geological time periods (see Chapter 1). Here, we will discuss the oldest continental evidence provided by lacustrine pollen sequences. Lake records span the last 30,000 years including part of the last glacial period (ice age) and its maximum in the Last Glacial Maximum (LGM).

The ice-age record for African tropical lowland forests is known from two sites: Barombi Mbo within a forested region of Cameroon (Maley and Brenac, 1998) and Ngamakala within secondary grassland nearby the Congo River (Elenga et a/.,1994). A third record, from lake Bosumtwi, West Africa, remains poorly documented through a preliminary pollen diagram (Maley, 1991). Except for Ce/tis, O/ea it does not contain any detail about the forest composition. Although the Bosumtwi record is informative about the lowland rainforest along the Guinean gulf, this review does not include it. A long core spanning the last 1.1 Myr was raised in 2003 and is still being analyzed. This record will be of immense significance to the reconstruction of West African paleoclimates. Barombi Mbo, evergreen and semi-evergreen forests (Cameroon)

North of the equator, the small crater lake Barombi Mbo ("Mbo" means lake in the local language) (4°40'N, 9°24'E) is located 15 km north of Mount Cameroon and 50 km inland from the Atlantic coast. Core MB-6 recovered from the deepest part of the lake (Maley et a/., 1990) yields a remarkably complete record for the last 32,000 years (2714Ckyr bp), including the LGM. The lake is situated at low elevation (300 m a.s.l.) and presently surrounded by forest (Figure 5.6, see color section). The crater lies within the wide belt of lowland evergreen Biafran forest dominated by Caesalpinia-ceae bordered by two large bands of semi-evergreen forest. Patches of semi-evergreen forest occupy areas under the rain shadow of Mount Cameroon, which causes a decrease in precipitation and a reduction in the length of the rainy season. Lying

Causes Rainforest Decrease

I I Trees I I Aquatics I I Poaceae

Figure 5.7. Synthetic pollen diagram from core BM-6, Lake Barombi Mbo, Cameroon, presented according to interpolated 14C ages (after Maley and Brenac, 1998) (% calculated versus pollen sum including all identified taxa, excluding spores).

I I Trees I I Aquatics I I Poaceae

Figure 5.7. Synthetic pollen diagram from core BM-6, Lake Barombi Mbo, Cameroon, presented according to interpolated 14C ages (after Maley and Brenac, 1998) (% calculated versus pollen sum including all identified taxa, excluding spores).

within Mount Cameroon's rainshadow, Barombi Mbo receives 2,350 mm/yr with a 3-month dry season from December to February. This relatively low rainfall contrasts with 9,000 mm/yr of windward (coastal) precipitation on the side of Mount Cameroon. The laminated sediments of the 23.5-m core were regularly deposited and present no hiatus, an exceptional situation for African lakes. Twelve AMS radiocarbon dates provide a reliable depth/age curve (Giresse et al., 1991, 1994). In the pollen diagram each sample corresponds to a 1-cm thickness of sediment averaging c. 10 to 15 years of deposition. Pollen analyses were made at c. 200-yr intervals in the Holocene, and c. 300yr in the glacial period (Maley and Brenac, 1998; Elenga et al., 2004). All the results discussed here follow the 14C chronology provided by the authors.

At Barombi Mbo, the curve of total arboreal pollen (Figure 5.7) provides a good average estimate of the forest cover surrounding the lake, although pollen deposition into the lake integrates a much larger basin. It clearly shows that, during the last glacial period—from 27 to 1014Ckyrbp (c. 32-11.5kcal. yr bp)—the area around the lake remained forested. However, between 20 to 1014Ckyr bp (c. 24-11.5kcal. yr bp) the tree cover was significantly reduced. Some fluctuations are depicted by the curve of total arboreal pollen, which would have been less marked if aquatics (sedges) had been eliminated from the pollen sum on which relative frequencies are calculated. High abundances of sedges (aquatics illustrated in Figure 5.7) coincide with the Last Glacial Maximum (LGM). Such peaks attest to enlarged herbaceous wetlands (including grass) that occupied emerged land on the shore line. A probable explanation for wetland expansion is falling lake levels in response to drier climatic conditions.

Glacial period and refuge hypothesis

Regarding the detailed pollen composition of the tree component, the last glacial period can be subdivided into two distinct phases (Figure 5.8). During the first phase from c. 32-24 kcal. yr bp, the total AP pollen (c. 80%) indicates a dense canopy cover (Figure 5.7) which remained fairly stable throughout and includes the highest frequencies (7 to 10%) of the Caesalpiniaceae evergreen component. Out of the 150 identified pollen taxa, 20 different genera are included in the Caesalpiniaceae curve (Maley and Brenac, 1998). As their pollen is normally under-represented in modern pollen rain (Reynaud-Farrera, 1995, and Section 5.3), the Caesalpiniaceae may have been a very important and diverse component within the forest at that time. Sapo-taceae, a component of mature semi-evergreen forest (Elenga et al., 2004), were also present (1 to 3%), together with other components of this subflora. Among markers of the mountain forest, Podocarpus pollen was found at such low frequencies (<1.5%) that it is unlikely that the trees occurred close to the lake. Today, Podocarpus is present on Mount Koupe (2,050 m) and its fossil occurrence could well be attributed to longdistance transport from this mountain. In marine cores Podocarpus pollen is quite abundant in sediment dating from the glacial period (Marret et al., 1999). In contrast, Olea was recorded at higher pollen percentages (>10%) indicating that their trees may have been present near the lake. The Olea pollen curve shows a remarkable pattern through time. An increasing trend started at 2414Ckyr bp (28 kcal. yr bp), reaching a maximum (30%) at 2014Ckyr bp (24 kcal. yr bp), and then decreasing again to 5% at 1714Ckyr bp (c. 20 kcal. yr bp). From 24 kcal. yr bp to 20 kcal. yr bp, the decreasing trend of Olea pollen is in good correspondence with the 4,000-yr duration of the LGM chronozone placed between 23 kcal. yr bp and 19 kcal. yr bp on marine records (Mix et al., 2001). Today, Olea capensis grows on Mount Cameroon at an elevation of 1,600 m and much higher, such as in cloud forest. Its abundance in the fossil pollen record has been explained by the impact of stratiform clouds and associated fogs produced by sea surface temperature cooling of the Atlantic (Maley, 1989; Maley and Elenga, 1993). That such processes may have played a role cannot be ruled out. Interestingly, the maximum of Olea percentages occurred slightly before the LGM and corresponds to the timing of the Dansgaard-Oeschger event 2 (DO2) and the last S18O maximum of the Antartica Byrd ice core (Mix et al., 2001). The pollen/climate transfer function in East Africa indicates a glacial continental cooling of 3 ± 2° C in the tropical region (Bonnefille et al., 1990,1992; Vincens et al., 1993)—a maximum value—since the effect of lower carbon dioxide content of the atmosphere could not be taken into account. Using the present day lapse rate, such an estimate corresponds to a 600-m shift in elevation, much less than the 1,300-m necessary lowering for Olea to reach the Barombi Mbo lowlands. Originally, tropical cooling was inferred from a significant shift in altitudinal distribution of vegetation zones on East African mountains (Flenley, 1979). The descent of vegetation on tropical mountains results from the

Lake Barombi
Figure 5.8(a). Detailed pollen diagram from core BM-6, Lake Barombi Mbo, Cameroon, presented according to interpolated 14C ages: trees and shrubs (after Maley and Brenac, 1998) (% calculated versus pollen sum including all identified taxa, excluding spores).

Semi-evergreen forest

Mountain Forest

Evergreen Forest


Semi-evergreen forest

Mountain Forest

Evergreen Forest

Photos Quaternary Lake Cores
Figure 5.8(b). Detailed pollen diagram from core BM-6, Lake Barombi Mbo, Cameroon, presented according to interpolated C ages: herbs (after Maley and Brenac, 1998) (% calculated versus pollen sum including all identified taxa, excluding spores).

associated effects of both decreasing temperature and rainfall. The Barombi Mbo record clearly demonstrates an individualistic movement of O/ea into the lowland vegetation during glacial time. This cannot be forced by lower CO2 concentration as it affected a tree which is a C3 plant. Applying 3°C-cooling at the Barombi Mbo will lead to a value of 21°C (24 — 3 = 21°C) for mean annual temperature, a value above the 18°C threshold for tropical highland forests in India (Bonnefille et a/., 1999; Barboni and Bonnefille, 2001), and above the 15°C threshold used to define the tropical biome (Prentice et a/., 1992). Under such conditions, O/ea could reach the lowland rainforest where other tropical trees remained. The pattern shown by the O/ea curve in the Barombi Mbo fossil record provides a good example of how plants individualistically responded to climatic changes. Significant rainfall decrease, during glacial time at the equatorial latitude, was estimated around 20 to 30% of the present value (Bonnefille et a/., 1990; Bonnefille and Chalie, 2000). Applying this estimate at Barombi Mbo leaves enough precipitation (1,500 mm/yr) to maintain a forest cover, during the glacial period, prior to the LGM.

During the second phase of the glacial period (24-11.5 kcal. yr bp), total AP pollen dropped with decreasing abundance of typical Biafran evergreen forest taxa, whereas semi-evergreen components—such as Ce/tis and Antiaris (Moraceae)— become more abundant, although their pollen frequencies show large fluctuations. Lowland species of Strombosia, Flacourtiaceae, Sapotaceae, Antiaris, Hymenostegia (Caesalpiniaceae), Ber/inia, and other Caesalpiniaceae are still present, but decreased significantly (Maley and Brenac, 1998). Isotopic studies from the same core point to an increased proportion of C4 grasses, likely favored by low CO2 concentration of the global atmosphere at that time (Giresse et a/., 1994). The increase in grass pollen does not overlap the O/ea phase (29-22 kcal. yr bp), but follows it, becoming more abundant between 24 kcal. yr bp and 11.5 kcal. yr bp, synchronously with the increase in Cyperaceae (Figure 5.8). The different patterns of the Poaceae and the Cyperaceae curves may indicate that the peak of Poaceae is not related to subaquatic grasses, but rather come from open grassland inside the forest. Pollen/biome reconstruction at 22 kcal. yr bp emphasized the replacement of rainforest by a tropical seasonal forest (Elenga et a/., 2000c). However, during the minimum extent of forest which lasted 5,000 years (24-19 kcal. yr bp), two sharp increases in tree cover are observed. They attest that forest expanded significantly during glacial time, although fluctuations in tree pollen percentages would have been minimized by excluding Cyperaceae from the pollen sum in the calculation of relative percentages. The maximum of grass pollen associated with the greatest opening of the forest is dated at 18 kcal. yr bp, a radiocarbon date that fits Heinrich Event H1 (Mix et a/., 2001), and therefore occurred a long time after the LGM. If the peak of O/ea registers the maximum cooling and the peak of grasses the maximum aridity, these were delayed by at least 5,000 years. Aridity and cooling were decoupled and a complex pattern of forest dynamics is evidenced during the glacial period when the climatic impact of the two Heinrich Events H1 and H2 affected the lowland rainforest at Barombi Mbo. Nevertheless, rainforest appears very sensitive to global climatic changes. While considering the high topography of Mount Cameroon and the high precipitation gradient, a great variety of climatic conditions must have prevailed in the region in the past, just as it does today. During glacial time, enough precipitation could have existed on the western slopes, allowing the persistence of evergreen forests there at the same time as semi-evergreen forests at Barombi Mbo, at the eastern base of Mount Cameroon. During glacial time, the coastal area expanded as sea level fell and offered new opportunities for new land occupation. Various forest refuges could have existed during glacial time and could be located on direct evidence by means of new palynological studies, rather than postulated on various hypotheses (Maley, 1996). Ngamakala, savanna contact with semi-evergreen rainforest (Congo)

On the right bank of the Congo River, the small Ngamakala lake (4°04'S, 15°23'E, 400 m), is located at the southern end of the Bateke plateau (Figure 5.1) where mesophilous, hygrophytic forests are related to humid edaphic conditions (Descoing, 1960). The lake—1 km wide—is now covered by Sphagnum (Sphagnaceae) and clumps of trees of Alstonia boonei (Apocynaceae). It is surrounded by a wooded Loudetia demeusei (Poaceae) savanna, with Pentaclethra (Mimosoideae) new growth (Makany, 1976). The results of pollen analysis of a 160-cm core show that, from c. 30-17kcal. yr bp, the fossil pollen sequence was dominated by Sapotaceae and Syzygium (Figure 5.9). With Fabaceae (Leguminosae) and Canthium, later identified among the unknown (Elenga et al., 2004), percentages of tree pollen exceed 80%. This record clearly indicates a forested environment during the glacial period. This forest was developed on a swamp attested by the occurrence of aquatic plants—such as Xyris, Laurembergia and the floating Nymphea. The forest existing there during the last glacial period included significant Fabaceae with Combretaceae, Alchornea, Campylospermum (Ochnaceae), Cleistanthus (Euphorbiaceae), Canthium, and Celtis. Rare pollen of other trees—such as Crudia gabonensis (Caesalpinioideae), Guibourtia, and Tetracera (Dilleniaceae)—provide a link with modern surface samples of the central Congo Basin (Elenga et al., 1994). But, except for Celtis and Sapotaceae, the glacial forest in southern Congo had no floristic resemblance to that documented at Barombi Mbo in Cameroon at the same time. More specifically, the Ngamakala record does not show any of the highland taxa pointing to cooler temperatures, such as observed in the Cameroon record. The swamp environment may not have been favorable to the growth of Olea, but the lack of Podocarpus is more surprising. If Podocarpus had been present at mid-elevation on the Bateke plateau, its abundant pollen would have been blown away, and at least a few grains found in the Ngamakala sediment. However, no valid conclusion can be drawn until another glacial sequence from southern Congo confirms it. As in the Cameroon record, variations within the relative abundance of the different trees are observed during the glacial period, although masked by abundant Syzygium and Sapotaceae. The Ngamakala core had a very low sedimentation rate with only an 80-cm thickness of sediment deposited during the glacial interval (from 29kcal. yr bp to 17kcal. yr bp), and conventional dating is not accurate enough. The time resolution interval between adjacent samples (350 years) is greater than that of Barombi Mbo and insufficient to address short-term climatic variability during the glacial period. At Ngamakala, no

36501180 |

39401130 |

108801160 |

132601220 |

140901230 |

36501180 |

39401130 |

108801160 |

132601220 |

140901230 |

Quaternary Period Tropical Rain Forest

108801160 I

132601220 |2 14090±230 |

Was this article helpful?

0 0
Waste Management And Control

Waste Management And Control

Get All The Support And Guidance You Need To Be A Success At Understanding Waste Management. This Book Is One Of The Most Valuable Resources In The World When It Comes To The Truth about Environment, Waste and Landfills.

Get My Free Ebook


  • ivan
    What is the history of the tropical rainforest?
    8 years ago

Post a comment