Station d'Etudes des Gorilles et Chimpanzés

Réserve de Faune de la Lopé, Gabon

August 1997

The Station d'Etudes des Gorilles et Chimpanzés (SEGC) was established 12 km into the Lopé Reserve, central Gabon, in 1983 by Caroline Tutin and Michel Fernandez, following the completion of a nationwide survey of gorillas and chimpanzees (Tutin & Fernandez, 1984). The census had revealed numbers far higher than anybody had expected and in the light of this it was decided to undertake a study of sympatric gorillas and chimpanzees. The Lopé was chosen because, at the time, it was the only one of Gabon's five faunal reserves actively protected by staff from the Ministry of 'Eaux et Forêts'. The construction and running costs of SEGC were, and continue to be financed by the Centre International de Recherches Medicales de Franceville (CIRMF), an institution co-funded the Gabonese government, ELF Gabon and the 'Cooperation Française'. SEGC is also affiliated to two other institutions: the University of Stirling, Scotland and the Ministry of 'Eaux et Forêts', Gabon.

What was initially planned as a two year field study of the ecology and behaviour of gorillas and chimpanzees has evolved into a long term study of many aspects of forest ecology. SEGC continues to receive core funding from CIRMF but additional support for the station in the form of grants has come from L.S.B. Leakey Foundation, the World Wide Fund for Nature, the World Society for the Protection of Animals; whilst The Wildlife Conservation Society (formerly the New York Zoological Society) has supported the presence of one researcher (LW) at SEGC since January 1989.

In the early days the emphasis at SEGC was on gorilla and chimpanzee ecology. Most field days involved many hours trying to locate apes by following their feeding trail. In addition to Caroline Tutin and Michel Fernandez, who were based permanently at SEGC, a number of students joined the team to work on either gorillas or chimpanzees.

Since 1989 the research goals at SEGC have become more diverse and the emphasis today has switched to the rain forest ecosystem as a whole, rather than concentrating on just one or two animal species.

The aim of this document is to give a synthesis of what has been achieved over the years and to give some idea of the future direction of research at SEGC. After a review of the history of scientific research at Lopé, major findings on the flora and fauna are presented. The importance of scientific research for management of the Reserve and role that the SEGC has played in developing methodologies and in training are described. Finally, future directions of research at SEGC are briefly outlined and the publication resulting from research at the station are listed in an Appendix.

A BRIEF HISTORY OF SCIENTIFIC RESEARCH IN THE LOPE RESERVE

Early research in the Lopé Reserve was botanical. It was the area's savannas, enclosed within tropical rain forest, which caught the interest. Aubreville published a number of papers speculating on the origins of the savanna forest mosaic in the Lopé Region (e.g., Aubreville, 1967). Later Descoings (1974) spent time in the region identifying and mapping different species assemblages in the savanna in the north of the Reserve and those further east towards Booué (Fig. 1). However, up until the early 1980's there was only limited botanical work in the forests of Lopé, and no research on animals had been done.

In 1982 Richard Oslisly and his colleagues began a study of the archaeology of the region. Our knowledge of the history of human occupation in the Lopé area, which is thought to date back at least 350,000 years, is due almost wholly to their detailed and on going study which has documented several stone age cultures, the arrival of metallurgy about 2500 years ago and the appearence of the current populations about 700 years ago (e.g., Oslisly 1993, 1995; Oslisly & Fontugne, 1993; Oslisly & Peyrot, 1992 a, b).

In 1982, Patrice Christy, an ornithologist, made the first of many visits to the Lopé which have resulted in a comprehensive inventory of bird species occurring in northern parts of the Reserve (Christy & Clarke, 1994).

In 1983 Michael Harrison, undertook a 9-month study of the ecology of black colobus, Colobus satanas, in forest bordering the savanna in the north of the Reserve (Harrison, 1986; Harrison & Hladik, 1986; see Fig. 1). During the course of his study he undertook monthly follows of a focal group of black colobus, collected specimens of all colobus foods for identification, and established three botanical transects along which he measured and identified all trees with circumference ³50 cm at breast height (1.3 m) in an area of 2 hectares. In addition he measured food availability for colobus by following phenology patterns of leaf, flower and fruit production.

Also in 1983, Michel Fernandez and Caroline Tutin began the long-term study on gorillas and chimpanzees at the Station d'Etudes des Gorilles et Chimpanzés. The small research station, which houses 4-6 people, is located on the forest edge in the north of the Reserve (Fig. 1). The study of the ecology of apes (Gorilla g. gorilla & Pan t. troglodytes ) has monitored diet and seasonal movements of sympatric gorillas and chimpanzees over a 12 year period.

Figure 1: Location of the Lopé Reserve and various sites mentioned in the text.

It is impossible to study the biology of any animal without an understanding of its habitat. So the ape study has involved identification of the plant species consumed or otherwise used (e.g., for nest construction [both species] or during tool use [chimpanzees]). In addition we have undertaken habitat analyses in order to try to understand seasonality of food choice (determined in part by availability) and ranging. Liz Williamson, the first doctoral student to complete field work at SEGC, identified and measured all trees with diameter at breast height (dbh, - measured at 1.3 m above the ground, or above buttresses) ³ 10 cm on five botanical transects (one 1-km line transect and four elephant paths) covering 4 hectares (Williamson, 1988). Williamson found that the most common tree species in the SEGC study area was new to science, and it was described in her honour as Cola lizae (Sterculiaceae). This was the first of a number of new species to be described from collections in Lopé, the most recent of which are Conceveiba macrostachys (Euphorbiaceae), a new genus for Africa and the most common tree in the forest in the centre of the Reserve; Dialium lopense (Caesalpiniaceae), which is one of the dominant trees in the forest galleries in the savannas in the north of the Reserve, and Engomegoma gordonii (Olacaceae), a new genus described in 1995 which is amongst the 10 dominant large trees in the south of the Reserve.

The SEGC research team began to collect plant phenology data in 1984 and this data set continues to be collected regularly. It has been of great help in the quest to understand the lives of gorillas and chimpanzees, and indeed for other rain forest mammals and birds, to have a record of fruit production in particular (e.g., Tutin et al., 1991a). Rogers and Williamson (1987) undertook stem counts of Marantaceae and Zingiberaceae along Williamson's 1-km line transect and demonstrated the abundance of plants of these two families in the understory (cf. Wrangham et al., 1993). The work at SEGC continues to track the ecology of gorillas and chimpanzees and is gradually diversifying into a detailed study of pattern and process in the Lopé ecosystem. Upwards of 60 publications have resulted from this work (see appendix).

Between 1985-1987 Jan Reitsma, established four 1-hectare plots in different parts of Gabon. One was in the Lopé Reserve, about 30 km to the south-west of SEGC. He identified and measured all trees and lianes ³ 10 cm dbh and identified all plant species (of any size) in two 10 m x 10 m sub-plots within each of the four sites (Reitsma 1988). Lopé was the least species diverse of Reitsma's four plots and he suggested that this was related to poor soils and low rainfall. One surprising feature of the plot in Lopé was that it contained several large Sacoglottis gabonensis (Humiriaceae) trees. This species was thought to be restricted principally to the coastal sedimentary basin and had not previously been recorded so far inland (Williamson 1988 also recorded the presence of scattered individuals in the SEGC study area).

In 1986 Marcel Alers and Allard Blom, began a study of the ecology and behaviour of forest buffalo, Syncerus caffer nanus, in the northern part of the Lopé Reserve. As part of this study they undertook classification and mapping of savanna areas between Lopé village and SEGC. They divided the savannas into five types based on plant species composition. Erosion and moisture content of the soil were two of the main physical features which determined savanna vegetation type. They produced a simplified vegetation map.

In 1989 Lee White began a doctoral study of the effects of commercial logging on rain forest vegetation and wildlife. This marked the first time that activities of SEGC were extended out of the main ape study area to the Reserve as a whole. At the same time, Gordon McPherson of the Missouri Botanical Garden made a first collecting trip to the region and he has continued to collaborate closely on the botanical inventory of the Reserve ever since.

For the logging study five sites with different logging histories were identified. Botanical inventories and mammal censuses were undertaken using methods adopted from similar studies elsewhere, with some adaptation as appropriate to ecological conditions in Lopé and the aims of the study - see Figure 1. When it became evident that there were important vegetation differences between sites that could not be explained by recent disturbance, the logging study evolved into an investigation of vegetation history and the effects of differences in vegetation composition on mammalian biomass (White, 1992).

From 1989 we began to diversify the research program to look at other species which shared the apes' habitat. We intensified long-term data collection on the biology of forest elephants and and three more PhD theses were completed, one looking at habitat preferences and nest building of chimpanzees (Wrogemann, 1992), one on the biology of grey-cheeked mangabeys (Cercocebus albigena ) (Ham, 1994) and one on the role of gorillas as seed dispersers (Voysey, 1995 ). In 1993, Alphonse Mackanga-Missandzou, the first Gabonese national candidate for a PhD in wildlife ecology, began a study of the wildlife and vegetation of the forest-savanna contact zone.

In 1992 a regional conservation program funded by the European community, ECOFAC, selected Lopé as the protected area in Gabon in which it would operate. The goal of ECOFAC is to promote the conservation of tropical forests through the development of sustanable economic activities within local comunities. At Lopé the main development project is the introduction of eco-tourism which creates jobs and opens a viable source of income to local people. In addition, ECOFAC has funded both ecological and sociological research in the Reserve in order to create a data-base that will allow informed management decisions to be made. Michel Fernandez directs ECOFAC Gabon and close links exist with SEGC.

Research on the Lopé gorillas suffered a set-back in 1993 when Porthos, the adult male 'silverback' of SEGC's focal gorilla group died as a result of an attack by another gorilla. His group broke up and his previous home range in the centre of our study area has remained practically devoid of gorillas. We continue to monitor the study area routinely - since the length of time it takes before the area is colonised by a new group may tell us a great deal about the dynamics of gorilla populations. However, the absence of gorillas and the start of ECOFAC have resulted in a shift in emphasis of research at SEGC.

Our new themes are most concerned with vegetation changes which have occurred since the last ice age (that is, the last 20,000 years or so) and their implications for rain forest mammals. Lopé is ideal for such studies, because the mosaic of forest and savanna, which occurs in the north of the Reserve, resembles the vegetation of much of tropical Africa as it was after the last glacial peak, when forest was expanding out of a few rain forest refugia into open grasslands. The juxtaposition of forest and savanna opens up a window into the past. A new emphasis on archaeology, in collaboration with Richard Oslisly at the Natural History Museum in Paris, has begun to highlight the important role humans have played in forest history, particularly over the last 2500 years. At the same time, the long term ecological data sets are gradually coming into their own, highlighting possible implications of the rapid climate change the world is experiencing today. Our research may eventually allow us to predict future changes in rain forest vegetation in Africa.

SUMMARY OF RESEARCH FINDINGS AT SEGC

Many researchers have passed through the Lopé Reserve and through the continued presence of the field station at SEGC the published and unpublished results of the work of these assorted scientists have been compiled into an extensive data base. This can be divided into four broad categories: Inventories; Vegetation Description, Dynamics and History; Ecology and Behaviour of large mammals; and Long-term Data Sets. Major findings are described below emphasising links between fields of research and the application of these findings to management of the Lopé Reserve and of tropical rain forest ecosystems.

1) Inventories

There has been no concerted attempt to compile a comprensive biological inventory in the Reserve. A list of large and medium sized mammals based on field observations and supplemented by examination of dead animals, includes 45 forest dwelling species (Tutin et al., in preparation). Systematic inventories have not been made of small mammals.

The majority of researchers have taken an interest in the botany of the Reserve and more recently ECOFAC has supported systematic botanical collection aimed at producing a plant species list for the Reserve. It can be argued that a detailed botanical inventory is the building block for all ecological research. In Lopé, early research focussed on primate ecology but in order to understand the ecology of primates it was vital to identify and describe the distribution of food species and to document patterns of plant food production in their habitat (e.g., Tutin et al., 1991 a, 1994). If one is to make meaningful comparisons across different sites detailed botanical inventories are essential (Mitani et al., 1994).

In addition, compilation of a detailed botanical inventory is a form of research in itself. Species composition will reflect both the diversity of habitats and the vegetation history of the site. The process of compiling a species list contributes to our understanding of vegetation types and is a useful means of training research assistants. The process also encourages collaboration between field sites and taxonomists working in herbaria worldwide, and in this way stimulates the ongoing process of describing and cataloguing the numerous species as yet unknown to science.

Tutin et al. (1994) recorded 676 species in the SEGC study area which covers about 50 km2 in the north of the Lopé Reserve (see Figure 1) and predicted that the total count would rise to about 1400. We have been able to compile a record of all botanical collections made in the Reserve by contacting botanists known to have worked there. During the course of daily field work we systematically collect all fertile plants with which we are unfamiliar. In addition, we undertake specific collecting trips to particular habitats or more remote areas. Plants are dried in the field. One specimen is kept for the field herbarium at SEGC. A second replicate is kept in Gabon for the National Herbarium in Libreville (LBV). A third is sent to the ECOFAC collection currently being established by Jean Lejoly at Université de Bruxelles Libre and all other replicates go to Missouri Botanical Garden (MO), in view of our profitable, long-term collaboration with Gordon McPherson at that institution.

To date we have a list of about 1250 collections to add to our own, numbering just over 1500, making a total of over 2750 for the Reserve. These have been entered into a database at SEGC. A total of at least 1286 species of plant from 116 taxonomic families have been collected in the Reserve. A number of collections have been made close to, but not within the Reserve, mostly in the Forêt des Abeilles to the east. This forest is dominated by plants in the Caesalpiniaceae - a vegetation type which dominates eastern and southern parts of the Lopé Reserve, but in which we have done limited collecting to date. If these collections from close to the Reserve, but not within its physical limits, are added (Gordon McPherson, personal communication; Chris Wilks, personal communication; and Gesnot, 1994) a further 67 species and one additional family are added to the list. The list includes 8 new species and one new genus (and a second which is new for Africa) and 26 new records for Gabon. The list of plant species will continue to rise for many years to come.

In addition to the plant list there is a comparable list of 345 bird species known to occur within the Reserve. Many of these are restricted to specific habitats such as savanna, mature forest or the galleries along the River Ogooué. Many of the species found in savanna are specialists which would disappear if this habitat is not maintained (see below).

Further inventories are planned as part of the ECOFAC program. An inventory of reptiles and amphibia is underway, and work on insects will be undertaken in the future.

2) Vegetation Description, Dynamics and History

Vegetation studies have been undertaken in several parts of the Lopé Reserve on five 5-km botanical transects, using the methods subsequently adopted by ECOFAC. There were between 81-160 species of tree and liane ³ 10 cm dbh per 2.5 ha transect sample (Figure 3). In addition there were marked differences in species composition between sites. For example, Table 1 compares the first ten species ranked by basal area for the five transects.

Figure 3: Species Area Curves for plants ³ 10 cm dbh on the five transects.

By a combination of field observation and data analysis using two types of multivariate analysis (Two-way Indicator Species Analysis [TWINSPAN] [Hill, 1979a] and Canonical Community Ordination [CANOCO] [Ter Braak, 1988 - an extension of DECORANA - Hill, 1979b]) 21 vegetation types found in different proportions on one or more of the transects were defined. Of these, six are major habitat types while others (described in full in White, 1992) cover smaller areas.

1) Savanna - vegetation maintained by annual fires, dominated by grasses, with shrubs such as Crossopteryx ferruginea, Nauclea latifolia and Bridelia febrifuga patchily distributed in some areas. Large continuous areas of savanna are restricted to low altitude areas, whilst small isolated patches occur either around these zones, or on hill tops with altitude of about 250-450m.

2) Colonising Forest - occurs adjacent to savannas in areas protected from fire. Shrubs such as Psidium guineensis, Psychotria vogeliana and Antidesma vogelianum become common as well as the trees Aucoumea klaineana, Lophira alata and Sacoglottis gabonensis. Ground vegetation is dominated by grasses.

3) Monodominant Forest - dominated by Aucoumea klaineana and Lophira alata, but with other characteristic species (e.g., Klainedoxa gabonensis ). Many individuals are crooked or branch low and most are smaller than is usual for their species. Ground vegetation is sparse, but some herbs may become established, notably Aframomum longipetiolatum and Megaphrynium spp.

4) Marantaceae Forest - where trees are better formed, canopy cover is increased but dominated by Aucoumea klaineana and Lophira alata . Ground vegetation is more diverse and herbs abundant including Haumania liebrechtsiana, Aframomum sp. ?nov and Megaphrynium spp.

5) Mixed Marantaceae Forest - in which greater numbers of other tree species are present adding to the structural complexity and species diversity of the forest in which Aucoumea klaineana and Lophira alata are no longer dominant. Herbaceous plants of the Zingiberaceae and Marantaceae are abundant.

6) Mature Forest - increased tree diversity including species associated with more ancient forest (e.g., Coula edulis, Sindoropsis le-testui, Desbordesia glaucescens ) (cf. de Saint Aubin, 1963). Densities of Marantaceae and Zingiberaceae herbs are low.

b) Effect of vegetation on animal abundance.

Differences in vegetation type have a profound influence on densities of many animal species, particularly large mammals such as forest elephant, Loxodonta africana cyclotis and western lowland gorillas, Gorilla g. gorilla (White, 1994 a). Table 2 gives the large mammal biomass for transects 1-5. There are great differences between sites and these are mainly due to differences in vegetation composition of the five sites (White, 1994 a).

Site 1, the SEGC main study area, which is dominated by Marantaceae Forest, had the highest mammalian biomass yet recorded for a tropical rain forest. This was due mostly to the high forest elephant density, which averaged 3 km-2 over a period of three years, with seasonal peaks almost double this. This illustrates the importance of Marantaceae Forest for elephants and there is evidence that it is equally important for several other large mammal species, particularly gorillas and mandrills (see White et al., 1995).

Table 2: Estimated species biomass for the study sites.

P = present but not recorded on censuses.

New research in gallery forests and forest islands isolated in the savanna has shown that mammalian biomass in the forest-savanna contact zone is even higher than in Marantaceae Forest, exceeding 5970 kg km-2. However, in this habitat type the proportion made up by each species is different, with elephants becoming less dominant and buffalo, pigs, duikers and mandrills achieving much higher densities.

c) Mapping vegetation types.

Considering the marked differences in animal densities in different vegetation types, if we are to estimate the size of populations of different animal species within the Lopé Reserve we need to map the distribution of vegetation types. Mapping has been undertaken on two scales. First, using aerial photographs which already existed and radar images commissioned by ECOFAC, a simplified vegetation map of the whole reserve has been produced (Figure 4). Second, a more detailed map of the main SEGC study area in the north of the Lopé Reserve has been produced in order to study the relationship between different vegetation types (Figures 5 & 6).

Figure 4: Simplified vegetation map of the Lopé Reserve

Figure 5: Part of the Lopé Reserve showing the area covered by the detailed vegetation map.

Aerial photographs from 1982 and radar images taken in 1992 were available for the whole of the Reserve, and there were also some aerial photographs from 1957 (all three at 1:50,000 scale). On aerial photographs and radar images it was possible to distinguish Savanna and the major forest types. A 1:10,000 scale map was produced by tracing visible vegetation boundaries from 1:50,000 scale aerial photographs and radar images and enlarging. Locations of water courses, old logging loads and major elephant trails were marked onto these maps using unpublished records from SEGC, aerial photographs and an existing topographic map and by direct mapping using a sighting compass and a 'topofil' measure. The entire SEGC study area was then walked and each vegetation type was mapped with the aid of a compass, topofil and GPS unit, by walking its perimeter.

The vegetation map is presented in Figure 6 at a scale of about 1:150000.

Figure 6: Detailed vegetation map of part of the Lopé Reserve.

d) Vegetation dynamics.

This map clearly demonstrates the relationship between Colonising, Monodominant and the two types of Marantaceae Forest. In almost all locations where there is Colonising Forest isolated within the main forest block, there are successive rings of Monodominant, Marantaceae and Mixed Marantaceae Forest. This is in line with the theory that these vegetation types are linked on a succession.

In order to try to explain the large variation in botanical composition between sites the following model was developed:

It is generally accepted that global climatic fluctuations associated with ice age maxima and minima have caused changes in the distribution of forest vegetation in Africa (e.g., Hamilton, 1976, 1982; Maley, 1991): during maxima the African climate became cooler and drier and forest vegetation retreated into a series of isolated refugia; and during minima the climate became warmer and wetter and forests expanded. In some places forest composition today reflects these pre-historic changes. De Foresta (1990) described forest vegetation close to savanna isolates in the Makaba region of the Oriental Mayombe, Congo, and concluded that the characteristic formation "forêt clairsemée à Marantaceae", which had previously been described close to savannas in Cameroun (Letouzey, 1968), was evidence for recent colonisation of savanna by forest vegetation. This forest type occurs at Lopé, where it is generally close to savannas, extending up to 10-15 km from the modern savanna edge, although isolated patches exist further away. Aubreville (1967) suggested that savanna areas in Lopé had previously been more extensive, and that they had been recolonised by forest spreading from refuges to the north and south. Data from our research, and information on Lopé savannas (Oslisly, 1993; J. Maley, personal communication), suggest the following scenario, very much in line with that put forwards by Aubreville:

After the last major dry climatic phase (ice age maxima) around 18,000 years BP (e.g., Hamilton, 1982; Maley, 1991), when much or all of the area might have been savanna, forest vegetation started to re-colonise around 12,000 years BP. Maximum extension of the forest occurred in the middle Holocene (7-4,000 BP), at which point the Lopé savannas would probably have been much reduced. Savannas probably then re-opened between 3-4,000 BP, expanding during a marked arid phase between 2500 - 2000 BP (Maley, 1992), when they were probably more extensive than today. The return of more humid conditions between 1400 - 1500 BP (Maley, 1992) would have initiated a new phase of forest re-colonisation, which may be what we see underway today in Lopé, in places where fire does not interfere with the succession.

During arid periods it is generally accepted that the climate also became cooler and that montane vegetation descended to about 450-500m altitude (Hamilton 1976, 1982; Maley et al, 1990). Hence, forest cover was retained on mountains of about this height, whilst lower hills became covered by savanna. In addition, species composition of some of the gallery forests currently isolated in savanna in the north of the Reserve suggests that some of these are ancient and probably also survived through at least the most recent arid phase. Once climatic conditions became favourable once again, forest species spread out from major forest refugia to the north and south of Lopé, and from forested mountain tops and any galleries that remained. As the forest advanced the last areas to be recolonised were hill-tops, and many of these retain isolated patches of savanna on their crowns, which will presumably slowly disappear if protected from fire.

Tree species such as Lophira alata, Aucoumea klaineana and Sacoglottis gabonensis are able to establish in savanna edge conditions. They form a Colonising Forest extension that provides shade, moderating the extremes of temperature and humidity experienced in open savanna. Early colonising individuals have low, round canopies and branch low down, but successive generations of seedling grow taller and branch higher to escape from the shadow of their predecessors. As more individuals become established canopy cover increases, and there is a corresponding increase in relative humidity and rates of organic deposition, whilst soil temperatures come to resemble those found within the forest. Soil quality improves, and new species are able to become established forming a post-colonising formation. Increased shade reduces the competitive advantage of grasses and new species of herbs and shrubs appear.

As the process continues and more species establish, conditions become favourable for the growth of herbs such as Haumania liebrechtsiana, Megaphrynium spp. and Aframomum spp., i.e. Marantaceae Forest, characteristic of many savanna/forest areas (de Foresta, 1990; Letouzey, 1968). Haumania liebrechtsiana densities increase until it forms a tangled carpet smothering the ground and climbing up to 10m or more in dense towers that can engulf small trees. This formation may persist for long periods, as the ground cover interferes with establishment of the next generations of trees. Marantaceae forests have characteristic low stocking densities of trees, especially medium sized trees whose crowns form the middle canopy (de Foresta, 1990; Letouzey, 1968), giving them an open appearance (see photographs in: de Foresta 1990, p.335; Letouzey 1968, p.225) and there are sometimes extensive areas (up to about 1ha) with few or no trees at all (de Foresta, 1990).

There is a gradual build-up of canopy cover in Marantaceae forest, which results in decreased light levels and a decrease in herb densities, and new tree species invade which are characteristic of Mixed Forest. As these become established the succession proceeds towards Mature Closed Canopy Forest, characterised by increased dominance of the Caesalpiniaceae, Olacaceae and Myristicaceae, and further decreased herb densities.

Following this model, the vegetation in the Lopé is a dynamic formation which has been in constant flux for thousands of years. The situation today is a snapshot in time and the vegetation will continue to evolve. As a consequence, there is a complex mosaic of vegetation types, particularly in northern parts of the Reserve close to the savanna, which reflects the recent expansion of forest into savanna areas. Each vegetation type is a stage of the succession from savanna to mature forest, and as such they are all inter-related.

If one assumes that Monodominant Forest results from savanna colonisation and evolves into Marantaceae Forest, one can see clearly from the detailed vegetation map the evolution of the landscape: areas which are now occupied by Monodominant Forest were once Savanna; areas which are now occupied by Marantaceae Forest would then have been Monodominant or Colonising Forest. The rock outcrops are probably a result of erosion during savanna conditions (see Schwartz et al., 1990).

The research of Richard Oslisly and his colleagues on the history of Iron Age populations, who arrived in the area about 2500 years ago, has particular bearing on vegetation dynamics. At Lopé, Marantaceae Forests extend up to 20km from the savanna edge in places (Fig. 4) and are then replaced by 'Closed Canopy Forests' with increased species diversity and structural complexity and decreased densities of Marantaceae and Zingiberaceae herbs. Oslisly & Dechamps (1994) found evidence of fires about 20 km south of the current savanna-forest interface in Lopé, which dated from 1,400 - 1,500 BP. These were closely associated with remains of iron smelting furnaces which dated from the same period and it is likely that the fires were caused by humans. Two tree species identified from the charcoal, Sapium ellipticum and Erythroxylum sp., are currently restricted to the forest-savanna mosaic. This suggests that the savannas extended further south at this time. Recent discoveries of iron age villages far from the present day savannas, close to the transition between Marantaceae Forest and Closed Canopy Forest, and dated back at about 1800 years BP (Oslisly & White, unpublished data) provide further evidence that savannas extended far to the south of their current distribution at Lopé. All of these sites are located in mature forest types with open understorey. The age of these recent archaeological finds suggests that the current distribution of Marantaceae Forest at Lopé is most likely to be a result of changes in the vegetation during the last dry phase of the recent Holocene, which started about 3000 BP and lasted to 2500/2000 BP (Maley, 1992).

If this is so, there would have to have been a period between then and now when savannas fires were absent, or rare, allowing the forest to advance up to 20 km into the savanna. This would be unlikely if human populations continued to live in the area, since they would burn savannas regularly. However, Oslisly (1993, 1995) reports evidence of a population crash at about 1400 BP which lasted until about 700 BP: there are numerous archaeological remains from the Lopé area both before and after this period, but no evidence of human activity anywhere in central Gabon between these dates. Such a population crash, coinciding as it did with a relatively humid period in Africa's climatic history (see Maley, 1992), would have allowed the forest to advance rapidly into the savanna from its southern limit at the time. This scenario suggests that much of the vegetation in the Lopé Reserve was savanna up to 1500 years ago. This area is delimited today by the distribution of Marantaceae Forests (Fig. 4). Today these forests support the highest known mammalian biomass of any rainforest (White, 1994 a).

The age of some of the younger vegetation types can be estimated using evidence from old aerial photographs and data on the growth rates of indicator tree species. The Colonising phase lasts for about 50 - 100 years before the transition to Monodominant Forest. Monodominant Forests probably last for another 100 years or so, gradually evolving into Marantaceae Forest. The transition from Marantaceae Forest to Mixed Marantaceae Forest probably takes about 200 years and the evolution of Mixed Marantaceae Forest into older forest types with more open understory probably takes 300 - 500 years. Succession in rocky areas is slower. Individuals of Diospyros zenkeri up to 30 cm dbh occur in Mixed Rocky Forest - according to growth data (see above) these are about 597 years old. Presence of rock outcrops indicates that the surrounding area was once savanna, since the erosion necessary to reveal outcrops does not occur under forest conditions (Schwartz et al., 1990).

These dates are best guesses for the moment. We are currently awaiting the results of a project using 13C analyses (which use measures of 13C in the soil to distinguish between landscapes dominated by Savanna or Forest species) combined with 14C dating of charcoal in soil profiles, which should allow us to date these transitions with some precision.

Using the preliminary estimates of the age of different vegetation types vegetation changes in the SEGC / ECOFAC area from the present to 1500 years BP have been reconstructed (Figures 7 - 11).

Figure 7: Vegetation map of SEGC and ECOFAC areas; present
(Legend applies to Figures 7 - 11).

Figure 8: Vegetation map of SEGC and ECOFAC; situation 75-100 years ago.

Figure 9: Vegetation map of SEGC and ECOFAC; situation about 250 years ago.

Figure 10: Vegetation map of SEGC and ECOFAC; situation about 700 years ago.

Figure 11: Vegetation map of SEGC and ECOFAC; situation about 1400 years ago.

This reconstruction shows that about 75 -100 years ago there was a lot more Colonising Forest than today, suggesting a corresponding reduction in the number of savanna fires. This probably corresponds to a trend for villages to move out of the Reserve towards the Ogooué River under the colonial 'regroupment' policy, begun in the late ninteenth century (Pourtier, 1989). About 250 years ago there was a high proportion of Monodominant Forest, corresponding to a period, probably about a century before, when there was another period without fires. Alternatively, this may represent a gradual movement of the forest-savanna border during a period of irregular fires due to low population density (Richard Oslisly, personal communication). The situation about 700 years ago reflects the situation at the end of the 700 year absence of human populations when savanna fires lit by the new arrivals (Oslisly 1995) would have stopped colonisation and set much of the young Colonising Forest back to Savanna. The situation about 1400 years ago corresponds to the beginning of the hiatus. Some gallery forests which were thin or non-existant are indicated by the presence of rock outcrops. In others, densities of Caesalpiniaceae species today indicate that they survived through this period of maximum savanna extension.

It is probable that the situation just before the hiatus was little changed from the arrival of metalurgists, about 2500 years ago (Oslisly & Fontugne, 1992). This arrival coincided with a dry climatic phase which lasted from about 3000 to 2500-2000 years ago (Maley, 1992), during which fires of human origin may well have penetrated the forest, driving it south away from the population centre along the River Ogooué. Hence, vegetation distribution in the Reserve at this time probably corresponded to that portrayed in Figure 12.

Figure 12: Simplified vegetation map for the Reserve, as it is thought to have been about 2000 years ago.

Figure 13 shows profile diagrams which document the process of savanna colonisation. Note that all but three of the species colonising the savanna in these figures are dispersed by animals (see below). Table 3 shows the densities of the commoner shrubs and seedlings in Savanna and Colonising Forest - all the species with animal dispersed fruits represent foods which will attract animals out of the forest into open areas where they are more easily viewed by tourists.

Figure 13: profile diagrams in Savanna and Colonising Forest.
(horizontal and vertical scales are the same. All individuals ³ 5 cm dbh in a 5m strip along a line-transect are drawn)

a) Savanna

b) Colonising Forest (about 15 years since last fire).

c) Colonising Forest ( about 50 years since last fire)

Table 3: abundance of species which occurred at least five times in Savanna or Colonising Forest plots. (Sav = number of individuals in 5 savanna plots; Col = number of individuals in 5 Colonising Forest plots; Disp = Dispersal mode (A - animal, W - wind, ? - unknown).

Moving further along the succession we pass through Monodominant Forest, which is essentially a natural plantation of Okoumé, into Marantaceae Forest. Plants belonging to the families Marantaceae (arrowroot) and Zingiberaceae (gingers) are a major component of the understorey in certain tropical African forests, to the extent that 'Marantaceae Forest' is a recognised forest type in some areas (e.g. de Foresta, 1990; Hawthorne, in press; Koechlin, 1964; Letouzey, 1968; Maley, 1990; Rogers & Williamson, 1987; Swaine, 1992; White, 1992).

Several species of Marantaceae and Zingiberaceae provide important food items for apes (Gorilla gorilla & Pan troglodytes ), mandrills (Mandrillus sphinx ) and elephants (Loxodonta africana ) in the forests of central Africa (Badrian & Malenky, 1984; Carroll, 1988; Hoshino, 1986; Jones & Sabater-Pi, 1971; Kano, 1983; Kano & Mulavwa, 1984; Malenky & Stiles, 1991; Rogers & Williamson, 1987; White et al., 1995; Wrangham et al., 1991, 1993). In the Lopé, young leaves and pith of five species, Aframomum sp.? nov. (Zingiberaceae), Haumania liebrechtsiana, Megaphrynium velutinum, M. macrostachyum and Hypselodelphys violacea (Marantaceae) are eaten throughout the year by lowland gorillas (Gorilla g. gorilla) and chimpanzees (Pan t. troglodytes) (Rogers et al., 1990; Tutin & Fernandez, 1993 b; Williamson et al., 1990). When fruit is scarce, both species of ape at Lopé increase their consumption of these foods. In addition, gorillas eat large quantities of the aquatic Marantaceae, Marantochloa cordifolia, but only during the annual dry season when fruit was scarce (Rogers et al., 1988; Tutin et al., 1991a). Parts of 14 other species of Marantaceae and Zingiberaceae are eaten less frequently (Tutin & Fernandez, 1993 b; Williamson et al., 1990), but the five species named above are amongst the 'keystone' foods of the apes at Lopé; that is, they play prominent roles in sustaining apes through periods of food (i.e.fruit) scarcity, and are reliably available all year (Terborgh, 1986). As well as providing crucial foods, 62% of gorilla night nests (N=2435) are built using one or more of these species as nesting materials (Tutin et al., 1995).

Occurrence of dense stands of plants in the families Marantaceae and Zingiberaceae in some African rain forests has been attributed to factors other than savanna colonisation: association with permanent water (e.g. Rogers & Williamson, 1987; Wrangham et al., 1993); forest disturbance (Calvert, 1985; Carroll, 1988); fire (Hawthorne, in press; Swaine, 1992); and elephants (Calvert, 1985; Guillaumet, 1967); although in some central African forests there is no obvious explanation for their occurrence (D. Harris, Central African Republic, pers.comm.).

Figure 14 illustrates these next stages in the succession:

Figure 14: Profile diagrams for Monodominant and Marantaceae Forest.
(horizontal and vertical scales are the same. All individuals ³ 5 cm dbh in a 5m strip along a line-transect are drawn)

a) Monodominant Forest

b) Marantaceae Forest

e) Patterns of fruit production

Patterns of fruit production in the forest in the Lopé Reserve are highly seasonal (see Fig. 15). The short dry season (January / February) is the period of maximum fruit availability whilst in the long dry season (mid-June to mid-September) is a time of scarcity for frugivorous mammals. This has a serious impact on behaviour of some species. For example, gorillas shift their diet to certain Marantaceae and Zingiberaceae species and range less widely, whilst chimpanzees split into small silent groups in an attempt to maintain the fruit content of their diet (Tutin et al., 1991a).

Some plant species fruit over prolonged periods and hence their fruits are available during much of the year (Tutin et al., 1991a; White, 1994b; White et al., 1993). Others however have distinct fruiting periods, which can cause predictable movements of some large mammals (see below). Plants which provide a reliable source of food in the period of fruit scarcity (known as 'keystone' foods) are of particular importance. For example, at SEGC the bark of Iroko (Milicia excelsa ) is an important food for gorillas during the long dry season when fruit availability is at a minimum. However, this species is a valuable commercial wood much sought after by loggers. Even a light exploitation which involved removal of many of the Iroko trees could have a severe effect on gorilla densities.

Another lesson that comes out of these phenological studies is the important role played by animals as seed dispersers. Almost 75% of plant species in Lopé rely on animals to disperse their seeds - what will be the result if over-hunting kills most of the large mammals and birds ?

Figure 15: Number of plant species fruiting per month in the Lopé Reserve.

(Data recorded on monthly fruit fall counts along five 5-km line transects - see Fig. 1. Taken from White 1994 b)

3) Ecology and Behaviour of Large Mammals.

a) Gorillas and Chimpanzees.

I. Diet

Data on ape diet have been collected by observation of feeding, examination of feeding trail left by apes and analysis of faeces. To date we have identified 220 foods of gorillas and 182 of chimpanzees (Tutin & Fernndez, 1993b; Williamson et al., 1990). For both, fruit is the most important food class although seeds, leaves, flowers and insects are also eaten regularly. Figure 16 compares the diets of Lopé gorillas and chimpanzees in terms of the relative proportions of different food classes.

Figure 16. Comparison of gorilla and chimpanzee diet at Lopé

Both gorillas and chimpanzees eat fruit regularly and remains of at least one species of fruit were found in 96% of 5200 gorilla faeces and in 98% of 3150 chimpanzee faeces examined. Dietary overlap is high with 139 foods eaten by both apes. Overlap is highest for fruit foods and only 8% of gorilla fruits and 10% of chimpanzee fruits are not shared. The degree of overlap is lowest for animal foods although insect remains were found in 30% of faeces of both species. Chimpanzees use tools to obtain honey from bees nests and to catch ants in arboreal and subterranean nests while gorillas, who have not been seen to use tools in the wild, eat a variety of ant species that life in accessible nests (Tutin & Fernandez, 1992; Tutin et al., 1995). Chimpanzees at Lopé, but not gorillas, also kill and eat vertebrates and to date we have identified six prey species including three arboreal monkeys.

Both gorillas and chimpanzees harvest the majority of their plant foods from trees, climbing to heights of 40m many times each day to obtain fruit and leaves.

Mountain gorilla ecology is well documented from long-term studies in Rwanda and Zaire and their diet, in contrast to that of the Lopé gorillas, is almost entirely folivorous. Few succulent fruit are available in their montane habitat. It is now clear from research at Lopé and from studies of gorillas in tropical forests in CAR and Congo, that fruit are a prefered food of gorillas and are eaten whenever available. The Lopé gorillas are the most frugivorous population studied to date, due probably to the botanical diversity of their habitat. Their diet resembles that of the Lopé chimpanzees much more than that of mountain gorillas.

The importance of fruit for apes at Lopé means that their diets vary greatly from month to month as most fruit are produced seasonally (Tutin et al., 1991a). We have also documented large inter-annual variations in the amount of ripe fruit available due to failures of crops in some years. Figure 17 illustrates both seasonal and inter-annual variation in the availability of fruit foods.

Figure 17. Means and Standard Deviations of the Number of Tree Species bearing ripe fruit, 1986 - 1991.

Figure 18. Seasonal variation in gorilla and chimpanzee diet at Lopé.

a) Mean Number of Fruit Species / Sample

b) Mean Monthly Foliage Score

Seasonal variation in ape diet is illustrated by the pattern of diversity of fruit eaten each month (mean number of fruit species per faecal sample) and by the mean monthly foliage scores (the proportion of the faecal sample composed of non-fruit foods measured on an 8-point scale), as shown in Figure 18. When fruit is scarce during the annual dry season both gorillas and chimpanzees eat larger amounts of pith and young leaves of herbaceous plants. Chimpanzees maintain a higher intake of fruit than do gorillas and concentrate on the oily flesh of the palm nut (Elaeis guineensis) which fruits asynchronously. Gorillas, but not chimpanzees, eat large amounts of the inner bark of Iroko during the dry season, as mentioned in the previous section, and also much pith from two acquatic herbs. Chimpanzees eat some palm nuts throughout the year but gorillas only eat Iroko bast and pith of acquatic herbs during the dry season suggesting that their 'keystone' foods are not preferred.

Thus, although gorilla and chimpanzee diets at Lopé are similar, they diverge when the preferred food of both species (succulent fruit) is scarce with gorillas being more folivorous than chimpanzees.

Fruit is a high quality food being rich in easily digested sugar. The preference for fruit is not surprising as, in contrast to other food classes such as leaves and pith, fruit are 'designed' to be eaten: the plant's reward to a potential seed dispersers. Apes are seed dispersers 'par excellence' as their large body size means that they swallow many seeds which are excreted intact. Ben Voysey's study showed that in addition to transporting seeds away from the parent tree, gorillas deposit seeds of some tree species in ideal conditions for germination and seedling growth. Gorillas defecate beside their night nests and nests are usually built in light gaps where the herbs preferred for nest construction are common. Germination rate and seedling survival after 2 years was significantly greater at nest-sites than for seeds that fell below parent trees or that were left in dung deposited beside paths in the forest (Voysey, 1995).

II. Social Organisation