Night and day: evaluating transect methodologies to monitor duikers in the Dzanga‐Sangha Protected Areas,Central African Republic

Across the Congo Basin, human hunting pressure poses the biggest threat to small‐ and medium‐bodied forest ungulates (genus Philantomba and Cephalophus, commonly known as duikers). The exploitation of these species has cascading effects on larger ecosystem processes, as well as on human subsistence practices. This study compares encounter rates and estimated population densities of duiker species, specifically Philantomba monticola (blue duiker) in the Dzanga‐Sangha Protected Areas (APDS), Central African Republic (CAR). Data were collected using direct observations of individual animals during diurnal (135 km) and nocturnal (150 km) transects in the APDS, with abundance estimates produced using DISTANCE software. Transect data demonstrate that within hunted forests similar to APDS, nocturnal rather than diurnal transects yield more individual observations of ungulates. Despite hunting pressure in the region, estimates presented for APDS suggest some of the highest density estimates reported for blue duikers in western and central Africa (58.6 blue duikers per km²). This study directly contributes current regional data on the status of duiker populations at APDS and in the larger Sangha Trinational Landscape (TNS, UNESCO). More broadly, we highlight the potential importance of nocturnal transect data to the development of adaptive management regimes in hunted forests.     

Primate and ungulate abundance in response to multi‐use zoning and human extractive activities in a Central African Reserve

This study examines the abundance of key mammal species at the Dzanga‐Sangha Reserve (RDS) in the Central African Republic with respect to conservation zoning and human activities in the reserve. RDS has been funded as an integrated conservation and development project since the mid‐1980s. This study illustrates distinct wildlife responses to logging and hunting in RDS sectors that vary in protection and enforcement levels and the erosion of some critical animal communities across the RDS in the face of challenges of increasing human populations and flows of arms and ammunitions there. Our results show elephants (Loxodonta africana cyclotis) to be appreciably absent close to human settlements, and increasingly vulnerable to hunting in the more integrally protected sectors far from town. We have found that duikers (Cephalophus sp.) and western lowland gorillas (Gorilla g. gorilla) make use of recently logged areas but are vulnerable to hunting there. These species are now most abundant farthest from human settlements. Our results have implications for the formulation of adaptive management plans that would benefit from the inclusion of nuanced understandings of site‐specific and species‐specific responses to microhabitats and the particular kinds of human extractive activities and challenges in the region.     

A survey of the apes in the Dzanga-Ndoki National Park, Central African Republic: a comparison between the census and survey methods of estimating the gorilla (Gorilla gorilla gorilla) and chimpanzee (Pan troglodytes) nest group density

A survey of apes was carried out between October 1996 and May 1997 in the Dzanga sector of the Dzanga‐Ndoki National Park, Central African Republic (CAR), to estimate gorilla (Gorilla gorilla gorilla) and chimpanzee (Pan troglodytes) densities. The density estimates were based on nest counts. The strip transect census and the line transect survey method (Standing Crop Nest Count) were used to estimate the gorilla nest group density. The strip transect has been most commonly used to date. It assumes that all nest groups within the width of the strip are detected, but as this assumption is easily violated in the dense tropical rain forest, the line transect survey was also used. In this method, only the nest groups on the transect line itself should be detected. This method proved to be an adequate and easy technique for estimating animal densities in dense vegetation. The gorilla density of 1.6 individuals km^?2 (line transect survey method) found for the Dzanga sector is one of the highest densities ever reported in the literature for the Western lowland gorilla. The density estimate for chimpanzees was 0.16 individuals km^?2 (census method). The results of this study confirm the importance of the Dzanga‐Ndoki National Park for primate conservation.     

GPS telemetry of forest elephants in Central Africa: results of a preliminary study

Few data exist on the ranging behaviour of forest elephants. A feasibility study on the use of GPS telemetry as a tool to study ranging, seasonal movements and distribution was implemented in the Dzanga‐Sangha and Nouabalé‐Ndoki National Parks Complex of Central African Republic and Congo. The study consisted of two parts – a thorough hand‐held testing of an elephant GPS telemetry collar under tropical forest conditions and the deployment of collars on two elephants. During the feasibility study the system performance was satisfactory; GPS fix acquisition success rate, VHF and UHF collar–researcher communications were adequate. Two elephants, a mature bull and an adult female, were immobilized and fitted with GPS collars in October 1998. After deployment, the female's GPS collar performed well initially, but in less than a month the GPS within the collar stopped acquiring fixes. She was subsequently located using VHF tracking. The male was never relocated strongly suggesting complete failure of the collar. Despite these setbacks, the small amount of data retrieved provide an important first insight into forest elephant ranging and daily activity patterns, with significant conservation implications. When technical difficulties of reliability are overcome, GPS telemetry will provide an exceptionally useful tool in forest elephant research and management.     

Roadside conditions as predictor for wildlife crossing probability in a Central African rainforest

The negative effects of roads on wildlife in tropical rainforests are poorly understood. Road construction has high priority in Africa, while negative impacts of roads on wildlife movement often are neglected. This study aims at providing information on the effects of roads on crossing behaviour of rainforest wildlife. The probability that wildlife would cross forest roads was analysed for association with ten different factors that were linked to road presence or construction. Factors were divided into three classes: vegetation cover, topography and human influence. A trackplot survey was done in southern Cameroon, Africa. Trackplots were laid along a 32 km unpaved logging road that intersects Campo‐Ma’an National Park. Tracks of several species were found frequently (e.g. genets and porcupines); while others were found only sporadically (e.g. forest duikers and apes). The actual physical obstacles found along the road (e.g. logs, banks, etc.) were highly negatively correlated with crossing probabilities. For all wildlife species high vegetation cover was positively correlated to crossing probability. This study indicates that roads have a large impact on wildlife, and suggests which factors could be altered during road construction and maintenance in order to mitigate these impacts.     

Censusing rainforest game species with communal net hunts

This paper presents a method to address two wildlife management problems in central African rainforests: the need for local communities to take responsibility for wildlife management, and the lack of simple and appropriate wildlife monitoring techniques. The method uses encounters of game species during net hunts to calculate abundance indices as well as to estimate population densities for the four principal game species in the Dzanga–Sangha region: the duikers Cephalophus monticola (10.7–20.4 km^?2), C. dorsalis (1.2–2.0 km^?2), and C. callipygus (0.9–1.2 km^?2), and the brush-tailed porcupine Atherurus africanus (2.7–5.3 km^?2). Game species behaviour, the hunting practice, and comparisons with results from other research across central Africa suggest that the method can provide valid density estimates for C. monticola and C. dorsalis, but only abundance indices for C. callipygus and A. africanus. Nevertheless, the method can be applied by hunters in the course of their normal activities, and is adapted to the local habitat types and game species. As such, it can be an important tool for local communities in developing sustainable wildlife management programmes.     

Extent and ecological consequences of hunting in Central African rainforests in the twenty-first century

Humans have hunted wildlife in Central Africa for millennia. Today, however, many species are being rapidly extirpated and sanctuaries for wildlife are dwindling. Almost all Central Africa''s forests are now accessible to hunters. Drastic declines of large mammals have been caused in the past 20 years by the commercial trade for meat or ivory. We review a growing body of empirical data which shows that trophic webs are significantly disrupted in the region, with knock-on effects for other ecological functions, including seed dispersal and forest regeneration. Plausible scenarios for land-use change indicate that increasing extraction pressure on Central African forests is likely to usher in new worker populations and to intensify the hunting impacts and trophic cascade disruption already in progress, unless serious efforts are made for hunting regulation. The profound ecological changes initiated by hunting will not mitigate and may even exacerbate the predicted effects of climate change for the region. We hypothesize that, in the near future, the trophic changes brought about by hunting will have a larger and more rapid impact on Central African rainforest structure and function than the direct impacts of climate change on the vegetation. Immediate hunting regulation is vital for the survival of the Central African rainforest ecosystem.     

Phylogeography of Mus (Nannomys) minutoides (Rodentia,Muridae) in West Central African savannahs: singular vicariance in neighbouring populations

We studied the phylogeography of the strict savannah pygmy mice Mus (Nannomys) minutoides in West Central Africa. A total of 846 base pairs of the cytochrome b sequence were obtained for 66 individuals collected in Gabon, Cameroon, Republic of Congo and Central African Republic. These sequences were compared to those of M. minutoides from other African countries and to eight other species of the genus Mus. We performed maximum likelihood, Bayesian and nested clade analyses, as well as neutrality tests and time estimates. We show that M. minutoides is a well‐differentiated monophyletic species that separated from other pygmy mice 1.17 Myr ago. A distinct West Central African M. minutoides clade diverged early from the other African populations of the species, with a more recent common ancestor dating 0.14 Myr. West Central African populations are globally homogeneous, despite the present fragmentation of savannahs by the rain forest. However, our analyses show an unexpected vicariance between geographically close savannahs, embedded in the rain forest in Central Gabon. One of these populations is genetically more similar to very distant peripheral populations than to three closely neighbouring populations situated on both sides of the Ogooué River. A non‐river geographical barrier probably persisted in this area, durably isolating these local populations. This hypothesis about the history of the savannah landscape should be testable through the biogeographical analysis of other strict savannah small mammal species.     

Group size, density and biomass of large mammals in the Réserve de Faune du Petit Loango, Gabon

Group size, density and biomass of large‐bodied diurnal mammal species in the Réserve de Faune du Petit Loango, Gabon (now Parc National de Loango) was determined over a 12‐month period using standard line‐transect methods. Petit Loango encompasses a range of distinct habitat types, including coastal scrub, savanna, swamps and disturbed and mature forest. Such intact coastal habitats are increasingly rare on the Central‐West African coastline. Faecal censusing indicated highest forest elephant (Loxodonta africana cyclotis) and buffalo (Syncerus caffer nanus) ecological densities at the extreme coast (2.48 and 1.29 km⁻² respectively), probably reflecting high intensity of use of this habitat. Ape density was comparable with that at other Central African study sites at 1.01 individuals km⁻². Mean total biomass of diurnal primates, elephants and other ungulates over the 20 km² site was 3290 kg km⁻². Forest elephants and red river hogs (Potamochoerus porcus) constituted the bulk of the biomass, at 67% and 14% respectively. Primates made up 5% of the biomass. This is the first estimation of mammal density and biomass over an annual cycle at a Central African coastal site, and provides baseline data for long‐term studies in such habitats and to aid habitat and wildlife management decisions.     

The pristine rain forest? Remnants of historical human impacts on current tree species composition and diversity

Abstract Aim Tropical rain forests are often regarded as pristine and undisturbed by humans. In Central Africa, community‐wide disturbances by natural causes are rare and therefore current theory predicts that natural gap phase dynamics structure tree species composition and diversity. However, the dominant tree species in many African forests recruit poorly, despite the presence of gaps. To explain this, we studied the disturbance history of a species‐rich and structurally complex rain forest. Location Lowland rain forest in Southern Cameroon. Methods We identified the recruitment conditions of trees in different diameter classes in 16 ha of species‐rich and structurally complex ‘old growth’ rain forest. For the identification of recruitment preference we used independent data on the species composition along a disturbance gradient, ranging from shifting cultivation fields (representing large‐scale disturbance), to canopy gaps and old growth forest. Results In nine of sixteen 1‐ha forest plots the older trees preferred shifting cultivation fields for recruitment while younger trees preferred gaps and closed forest conditions. This indicates that these nine sites once experienced large‐scale disturbances. Three lines of evidence suggest that historical agricultural use is the most likely disturbance factor: (1) size of disturbed and undisturbed patches, (2) distribution of charcoal and (3) historical accounts of human population densities. Main conclusions Present‐day tree species composition of a structurally complex and species‐rich Central African rain forest still echoes historical disturbances, most probably caused by human land use between three to four centuries ago. Human impact on African rain forest is therefore, contrary to common belief, an issue not of the last decades only. Insights in historical use will help to get a more balanced view of the ‘pristine rain forest’, acknowledging that the dualism between ‘old growth’ and ‘secondary’ forest may be less clear than previously thought.     

Gut microbiome composition and metabolomic profiles of wild western lowland gorillas (Gorilla gorilla gorilla) reflect host ecology

The metabolic activities of gut microbes significantly influence host physiology; thus, characterizing the forces that modulate this micro‐ecosystem is key to understanding mammalian biology and fitness. To investigate the gut microbiome of wild primates and determine how these microbial communities respond to the host's external environment, we characterized faecal bacterial communities and, for the first time, gut metabolomes of four wild lowland gorilla groups in the Dzanga‐Sangha Protected Areas, Central African Republic. Results show that geographical range may be an important modulator of the gut microbiomes and metabolomes of these gorilla groups. Distinctions seemed to relate to feeding behaviour, implying energy harvest through increased fruit consumption or fermentation of highly fibrous foods. These observations were supported by differential abundance of metabolites and bacterial taxa associated with the metabolism of cellulose, phenolics, organic acids, simple sugars, lipids and sterols between gorillas occupying different geographical ranges. Additionally, the gut microbiomes of a gorilla group under increased anthropogenic pressure could always be distinguished from that of all other groups. By characterizing the interplay between environment, behaviour, diet and symbiotic gut microbes, we present an alternative perspective on primate ecology and on the forces that shape the gut microbiomes of wild primates from an evolutionary context.     

Caught in the mesh: roads and their network‐scale impediment to animal movement

Roads have a pervasive multi‐faceted influence on ecosystems, including pronounced impacts on wildlife movements. In recognition of the scale‐transcending impacts of transportation infrastructure, ecologists have been encouraged to extend the study of barrier impacts from individual roads and animals to networks and populations. In this study, we adopt an analytical representation of road networks as mosaics of landscape tiles, separated by roads. We then adapt spatial capture–recapture analysis to estimate the propensity of wildlife to stay within the boundaries of the road network tiles (RNTs) that hold their activity centres. We fit the model to national non‐invasive genetic monitoring data for brown bears Ursus arctos in Sweden and show that bears had up to 73% lower odds of using areas outside the network tile of their home range centre, even after accounting for the effect of natural barriers (major rivers) and the decrease in utilization with increasing distance from a bear's activity centre. Our study highlights the pronounced landscape‐level barrier effect on wildlife mobility and, in doing so, introduces a novel and flexible approach for quantifying contemporary fragmentation from the scale of RNTs and individual animals to transportation networks and populations.     

Community‐based wildlife management area supports similar mammal species richness and densities compared to a national park

Community‐based conservation models have been widely implemented across Africa to improve wildlife conservation and livelihoods of rural communities. In Tanzania, communities can set aside land and formally register it as Wildlife Management Area (WMA), which allows them to generate revenue via consumptive or nonconsumptive utilization of wildlife. The key, yet often untested, assumption of this model is that economic benefits accrued from wildlife motivate sustainable management of wildlife. To test the ecological effectiveness (here defined as persistence of wildlife populations) of Burunge Wildlife Management Area (BWMA), we employed a participatory monitoring approach involving WMA personnel. At intermittent intervals between 2011 and 2018, we estimated mammal species richness and population densities of ten mammal species (African elephant, giraffe, buffalo, zebra, wildebeest, waterbuck, warthog, impala, Kirk's dik‐dik, and vervet monkey) along line transects. We compared mammal species accumulation curves and density estimates with those of time‐matched road transect surveys conducted in adjacent Tarangire National Park (TNP). Mammal species richness estimates were similar in both areas, yet observed species richness per transect was greater in TNP compared to BWMA. Species‐specific density estimates of time‐matched surveys were mostly not significantly different between BWMA and TNP, but elephants occasionally reached greater densities in TNP compared to BWMA. In BWMA, elephant, wildebeest, and impala populations showed significant increases from 2011 to 2018. These results suggest that community‐based conservation models can support mammal communities and densities that are similar to national park baselines. In light of the ecological success of this case study, we emphasize the need for continued efforts to ensure that the BWMA is effective. This will require adaptive management to counteract potential negative repercussions of wildlife populations on peoples' livelihoods. This study can be used as a model to evaluate the effectiveness of wildlife management areas across Tanzania.     

Indirect evidence for the genetic determination of short stature in African Pygmies

Central African Pygmy populations are known to be the shortest human populations worldwide. Many evolutionary hypotheses have been proposed to explain this short stature: adaptation to food limitations, climate, forest density, or high mortality rates. However, such hypotheses are difficult to test given the lack of long-term surveys and demographic data. Whether the short stature observed nowadays in African Pygmy populations as compared to their Non-Pygmy neighbors is determined by genetic factors remains widely unknown. Here, we study a uniquely large new anthropometrical dataset comprising more than 1,000 individuals from 10 Central African Pygmy and neighboring Non-Pygmy populations, categorized as such based on cultural criteria rather than height. We show that climate, or forest density may not play a major role in the difference in adult stature between existing Pygmies and Non-Pygmies, without ruling out the hypothesis that such factors played an important evolutionary role in the past. Furthermore, we analyzed the relationship between stature and neutral genetic variation in a subset of 213 individuals and found that the Pygmy individuals' stature was significantly positively correlated with levels of genetic similarity with the Non-Pygmy gene-pool for both men and women. Overall, we show that a Pygmy individual exhibiting a high level of genetic admixture with the neighboring Non-Pygmies is likely to be taller. These results show for the first time that the major morphological difference in stature found between Central African Pygmy and Non-Pygmy populations is likely determined by genetic factors.     

Centres of dispersal and evolution in the neotropical region

For the first time, centres of dispersal in the sense of De Lattin (1957) have been analysed for South and Central America. According to the results of this analysis, at least 40 centres have to be considered, the situation of which has been influenced by quaternary climatic fluctuations and vegetation fluctuations. The latest rate of differentiation of the faunal elements, which have to be associated with the centres (subspecific level) can be understood for a temporary indicator of the last function of the centres in the sense of centres of preservation of faunas and floras during regressive phases. The isolation of the montane forest centres occured before that of most of the lowland forest centres of the central and northern part of South America. During a postglacial arid phase, there has been a gene flow, via open migration roads, between numerous Amazonian Hylea‐Campo populations and populations of the costal savanna of Guiana, Venezuela and the Campo Cerrados. At the same time, these migration roads played the role of barriers of dispersal to the forest fauna and were reconquered by the forest at the end of the aridity phase. At the present time, transition zones of subspecificly differentiated forest populations can be found within the range of these old migration roads. Although the Campo Cerrado has been acting as a barrier of dispersal to forest species since the beginning of the postglacial arid phase, it had obviously been passed through by Amazonian and Serra‐do‐Mar faunal elements prior to that time. The location of the centres of dispersal and the differentiation of their faunal elements support the assumption that the great richness in species of neotropic biomes may be attributed to a large extent to quaternary biochore shifts and the resulting constant isolation of populations. Independent of this, however, numerous faunal elements indicate that this latest phase of differentiation is, in many cases, based on much older ones.     

Reductions in primate abundance and diversity in a multiuse protected area: synergistic impacts of hunting and logging in a congo basin forest

This article explores spatial and temporal changes in diurnal primate abundance and behavior in response to hunting, logging, and conservation at the Dzanga Sangha Dense Forest Reserve (RDS), Central African Republic over time. We use a combination of line-transect surveys in 2002 and 2009 (N = 540 km) and ethnographic interviews (N = 210) to investigate changes in the status of cercopithecines and colobines at RDS, with additional comparisons to earlier work. This protected area was lightly logged in the 1970s and the park was gazetted in 1990, with multiple-use reserve sectors allocated. Since the park's inception, hunting and the trade of primates have increased, along with human migration, greater accessibility of arms, and reduction of preferred ungulate prey. Primates have declined in both the park and reserve sectors. Our data further suggest that at RDS hunting has had a greater impact on primate diversity and abundance than logging. We have identified changes in species-specific vulnerability to hunting over time, with Cercopithecus nictitans and Lophocebus albigena initially having appeared to be relatively resistant to hunting pressure in 2002. However, subsequently as gun hunting has increased at RDS, these species have become vulnerable. Although monkeys at RDS have been responding behaviorally to increased gun hunting, they are not able to keep pace with changing hunting practices. This study allows us to begin to understand synergistic impacts of hunting and logging, necessary if we are to recommend strategies to better secure the future of primates in multiuse protected areas.     

Wildlife roadkill patterns in a fragmented landscape of the Western Amazon

One of the most evident and direct effects of roads on wildlife is the death of animals by vehicle collision. Understanding the spatial patterns behind roadkill helps to plan mitigation measures to reduce the impacts of roads on animal populations. However, although roadkill patterns have been extensively studied in temperate zones, the potential impacts of roads on wildlife in the Neotropics have received less attention and are particularly poorly understood in the Western Amazon. Here, we present the results of a study on roadkill in the Amazon region of Ecuador; a region that is affected by a rapidly increasing development of road infrastructure. Over the course of 50 days, in the wet season between September and November 2017, we searched for road‐killed vertebrates on 15.9 km of roads near the city of Tena, Napo province, for a total of 1,590 surveyed kilometers. We recorded 593 dead specimens, predominantly reptiles (237 specimens, 40%) and amphibians (190, 32%), with birds (102, 17%) and mammals (64, 11%) being less common. Recorded species were assigned to three functional groups, based on their movement behavior and habitat use (“slow,” “intermediate,” and “fast”). Using Ripley's K statistical analyses and 2D HotSpot Identification Analysis, we found multiple distinct spatial clusters or hotspots, where roadkill was particularly frequent. Factors that potentially determined these clusters, and the prevalence of roadkill along road segments in general, differed between functional groups, but often included land cover variables such as native forest and waterbodies, and road characteristics such as speed limit (i.e., positive effect on roadkill frequency). Our study, which provides a first summary of species that are commonly found as roadkill in this part of the Amazon region, contributes to a better understanding of the negative impacts of roads on wildlife and is an important first step toward conservation efforts to mitigate these impacts.     

Acoustic estimation of wildlife abundance: methodology for vocal mammals in forested habitats

Habitat loss and hunting pressure threaten mammal populations worldwide, generating critical time constraints on trend assessment. This study introduces a new survey method that samples continuously and non‐invasively over long time periods, obtaining estimates of abundance from vocalization rates. We present feasibility assessment methods for acoustic surveys and develop equations for estimating population size. As an illustration, we demonstrate the feasibility of acoustic surveys for African forest elephants (Loxodonta africana cyclotis). Visual surveys and vocalizations from a forest clearing in the Central African Republic were used to establish that low‐frequency elephant calling rate is a useful index of elephant numbers (linear regression P < 0.001, r_adj.² = 0.58). The effective sampling area was 3.22 km² per acoustic sensor, a dramatic increase in coverage over dung survey transects. These results support the use of acoustic surveys for estimating elephant abundance over large remote areas and in diverse habitats, using a distributed network of acoustic sensors. The abundance estimation methods presented can be applied in surveys of any species for which an acoustic abundance index and detection function have been established. This acoustic survey technique provides an opportunity to improve management and conservation of many acoustically‐active taxa whose populations are currently under‐monitored.     

Conserving large carnivores: dollars and fence

Conservationists often advocate for landscape approaches to wildlife management while others argue for physical separation between protected species and human communities, but direct empirical comparisons of these alternatives are scarce. We relate African lion population densities and population trends to contrasting management practices across 42 sites in 11 countries. Lion populations in fenced reserves are significantly closer to their estimated carrying capacities than unfenced populations. Whereas fenced reserves can maintain lions at 80% of their potential densities on annual management budgets of $500 km⁻², unfenced populations require budgets in excess of $2000 km⁻² to attain half their potential densities. Lions in fenced reserves are primarily limited by density dependence, but lions in unfenced reserves are highly sensitive to human population densities in surrounding communities, and unfenced populations are frequently subjected to density‐independent factors. Nearly half the unfenced lion populations may decline to near extinction over the next 20–40 years.     

Climatic and cultural changes in the west Congo Basin forests over the past 5000 years

Central Africa includes the world''s second largest rainforest block. The ecology of the region remains poorly understood, as does its vegetation and archaeological history. However, over the past 20 years, multidisciplinary scientific programmes have enhanced knowledge of old human presence and palaeoenvironments in the forestry block of Central Africa. This first regional synthesis documents significant cultural changes over the past five millennia and describes how they are linked to climate. It is now well documented that climatic conditions in the African tropics underwent significant changes throughout this period and here we demonstrate that corresponding shifts in human demography have had a strong influence on the forests. The most influential event was the decline of the strong African monsoon in the Late Holocene, resulting in serious disturbance of the forest block around 3500 BP. During the same period, populations from the north settled in the forest zone; they mastered new technologies such as pottery and fabrication of polished stone tools, and seem to have practised agriculture. The opening up of forests from 2500 BP favoured the arrival of metallurgist populations that impacted the forest. During this long period (2500–1400 BP), a remarkable increase of archaeological sites is an indication of a demographic explosion of metallurgist populations. Paradoxically, we have found evidence of pearl millet (Pennisetum glaucum) cultivation in the forest around 2200 BP, implying a more arid context. While Early Iron Age sites (prior to 1400 BP) and recent pre-colonial sites (two to eight centuries BP) are abundant, the period between 1600 and 1000 BP is characterized by a sharp decrease in human settlements, with a population crash between 1300 and 1000 BP over a large part of Central Africa. It is only in the eleventh century that new populations of metallurgists settled into the forest block. In this paper, we analyse the spatial and temporal distribution of 328 archaeological sites that have been reliably radiocarbon dated. The results allow us to piece together changes in the relationships between human populations and the environments in which they lived. On this basis, we discuss interactions between humans, climate and vegetation during the past five millennia and the implications of the absence of people from the landscape over three centuries. We go on to discuss modern vegetation patterns and African forest conservation in the light of these events.