Biodiversity and Conservation

, Volume 17, Issue 5, pp 1241–1249

Mammal distribution in a Central African logging concession area

Authors

    • Centre International de Recherche Agronomique pour le Développement (CIRAD)
  • Robert Nasi
    • Center for International Forestry Research (CIFOR)Centre International de Recherche Agronomique pour le Développement (CIRAD)
Original Paper

DOI: 10.1007/s10531-007-9300-5

Cite this article as:
Van Vliet, N. & Nasi, R. Biodivers Conserv (2008) 17: 1241. doi:10.1007/s10531-007-9300-5

Abstract

We used data collected during a routine forest inventory to prepare a management plan for a logging concession in Gabon to identify the biophysical and human factors that better explain the distribution of mammal species within the logged-over landscape. Results of a Multiple Correspondence Analysis show that the distribution of mammals within the forest concession is more influenced by roads and hunting than by the direct effects of logging. The structure of the canopy and that of the understorey are also important factors explaining the distribution of mammals within the concession. Linear regressions and Spearman correlation tests were computed to assess the significance of the correlation between the probability of encounter for a particular species and the distance from main roads. Small diurnal monkeys were found far from the villages and between 3 and 10 km from the main roads. Elephants were equally found close or far from roads and do not seem to be affected by hunting activities. Red duikers and the yellow back duikers avoided hunted zones and were significantly more abundant far from roads. Other species like gorillas, chimpanzees or forest buffaloes show no negative relationship with distance to roads and were observed close to villages. We show also how using routinely collected data without research purposes, can be used to propose practical recommendations to managers to limit the negative impacts of logging activities within the concession.

Keywords

Mammal distributionLogging concessionEco-geographical factorsHuntingRoadsGabon

Introduction

In the last decade, large blocks of Central Africa forests have been surveyed by logging companies to implement the new forestry laws that require a detailed and comprehensive forest management plan. These surveys aimed, as primary goal, at estimating the timber potential from commercial species but did also consider other ecological parameters including fauna (van Vliet et al. 2004).

Between 2001 and 2003, the company CBG (“Compagnie des Bois du Gabon”) carried out a forest management survey within its Mandji logging concession to develop the forest management plan required by Law 16/01 (forest code of the Gabonese Republic). Data was collected systematically along transects covering the entire CFAD and considered, aside of timber resources, fauna, flora, biophysical parameters and indices of human activities.

We analysed this invaluable data, collected during a routine management planning process, using various statistical methods to assess the scientific usefulness of such data and see if we could understand which factors explain mammal distribution in this logging concession.

Material and methods

The Mandji concession (Fig. 1), covers 352,100 ha over two provinces of southwest Gabon (Ogooué Maritime and Ngounié) between 1°29′–2°22′ latitude South and 9°37′–10°56′ longitude East (Pélissier 2004). The concession borders the Loango and the Moukalaba-Doudou National Parks. This forest block has been logged numerous times since the 1950s (WRI 2000) leaving a mosaic of harvested and untouched forests.
https://static-content.springer.com/image/art%3A10.1007%2Fs10531-007-9300-5/MediaObjects/10531_2007_9300_Fig1_HTML.gif
Fig. 1

The Mandji forest concession under sustainable management plan and the system of transects used for “biodiversity surveys”

According to Caballé (1978), the most common vegetation type found within the CFAD is a coastal dense evergreen tropical forest dominated by Aukoumea klaineana and Sacoglottis gabonensis. Based on a vegetation map developed using satellite images (Ouar 2003) and on field observations, seven vegetation types were distinguished in this heterogeneous landscape: (1) a mosaic of forests and savannas, (2) Aukoumea klaineana mono-dominant forests, (3) hill dense forests at the edge of the Doudou mountains, (4) lowland dense forests, (5) riverine forests, (6) swamps and (7) secondary forests. From a floristic point of view, the hill and lowland forests are very similar and characterized by a high floristic richness (290 tree species) without any dominance of a particular species. Mono-dominant forests are still diverse (248 tree species) but the canopy is largely dominated by one species, Aucoumea klaineana. Riverine forests are also less diversified and characterized by the abundance of Diospyros spp., Diogoa zenkeri and Anthostema aubryanum. Secondary forests have the lowest floristic diversity (169 tree species) and are locally dominated by pure stands of Musanga cecropioides.

The forest inventory was carried out along 159 parallel and equidistant line transects covering 1% of the overall concession area. The survey units were 20 m by 200 m adjacent plots (5,711 plots) centred on the transect line.

The mammal survey was carried out using line transect techniques and the information on seen animals and pellet groups was referred to the corresponding plot. More than 20 mammal species were surveyed. For analyses we selected species on the basis of (1) importance for local people (Cephalophus spp., Atherurus africanus and small diurnal monkeys), (2) charismatic and international conservation value (Gorilla gorilla, Pan troglodytes, Loxodonta africana, (3) total protection status in Gabon (Hyemoniscus africanus and Cephalophus sylvicultor) (Table 1). As in most surveys, diurnal monkey species (Cercopithecus cephus, C. nictitans, C. pogonias, Lophocebus albigena) and red duiker species (Cephalophus dorsalis, C. callipygus, C. leucogaster, C. nigrifrons and C. ogylbi) were lumped together respectively under the name “small diurnal monkeys” and “red duikers”. Small diurnal monkeys are often found in multi-species groups, which might indicate similar distributions for the different species grouped under the same name. For red duikers, knowing that C. callipygus and C. dorsalis represent respectively 49% and 33% of the individuals in a sample of red duikers (Feer 1996), it is likely that the results for “red duikers” actually reflect the distributions of these two most common and sympatric species.
Table 1

Common and scientific names of the mammal species selected for our analysis

Scientific names

Common names

Atherurus africanus

Brush-tailed porcupine

C. cephus, C. nictitans, C. pogonias, Lophocebus albigena

Small diurnal monkeys

C. dorsalis, C. callipygus, C. leucogaster, C. nigrifrons et C. ogylbi

Red duikers

C. sylvicultor

Yellow back duiker

Cephalophus monticola

Blue duiker

Gorilla gorilla

Gorilla

Hyemoniscus aquaticus

Water chevrotain

Loxodontha africana

Elephant

Pan troglodytes

Chimpanzee

Potamochoerus porcus

Red river hog

Syncerus caffer

Buffalo

All tree species were identified and their diameter at breast high (DBH) was measured. A total of 315 species were recorded but we confined our analyses to 60 species identified by Gauthier-Hion et al. (1985) as potential fruit producers for the large mammals studied here. We also neglected individuals with a DBH lower than 40 cm, as most trees fruit above this diameter. The abundance and species richness of these 60 tree species were calculated for each plot.

Plots were also characterized by a suite of biophysical parameters—Topographic positions: (1) valley; (2) plain; (3) plateau; (4) hilltop; Canopy cover: 1: open canopy; 2: semi-open canopy; 3: closed canopy;—Abundance of understorey vegetation: (1: clear; 2: semi-dense; 3: dense);—Abundance of herbaceous species (Maranthaceae, Zingiberaceae or ferns): (1: absent; 2: scarce; 3: understorey entirely covered);—Abundance of lianas (1: absence; 2: few; 3: abundant) and—Soil type: 1 clayey, 2 sandy, 3 rocky.

Signs of human activities, such as logging damages or infrastructures (gaps, roads, skidding trails, log yards) and hunting indices (snares, cartridges, hunting camps) were also recorded for each plot.

All plots where also characterized by their distance to waterways (less than 100 m; from 100 m to 1 km; more than 1 km), to main roads (less than 3 km, from 3 to 7 km and more than 10 km) and to villages (less than 5 km, from 5 to 15 km, and more than 15 km)

Small rivers, villages and main roads where digitized from a 1/50,000 map and integrated with the plots in a GIS (Mapinfo®) where all ecological information on fauna, flora and biophysical and human variables were also added. We used this GIS to build the distribution maps for each species and make a visual description of their distribution patterns.

We used Xlstat2006® to conduct a multiple correspondence analysis (MCA) to identify the biophysical and human factors that better explain the distribution of mammal species within the logging concession. We used linear regression and Spearman correlation test to examine if the probability of encounter of a species co-varied with the main discriminant variables obtained with the MCA. For species that showed no linear correlation, we used the t-test to detect significant differences between means.

Results

The habitats that sustain the highest mammal richness are the lowland dense forests and the savannas-forests mosaics. Dung counts and seen animals for each species highlight three main distribution patterns: (1) abundant species evenly distributed all over the concession (elephant and red duikers); (2) abundant species present all over the concession but with a high concentration in particular patches (Cephalophus monticola; small monkeys); rare species with a clear patchy distribution pattern (Hyemoschus aquaticus, Gorilla gorilla, Pan troglodytes and Syncerus caffer).

The GIS maps show that 20% of the concession is located less than 3 km from a main roads and less than 5 km from a village. Most hunting traces are located less than 3 km from the main roads or at less than 5 km from the bigger cities (Rabi, Mandji, Guietsou, Mbongou1). We found a strong significant correlation (Spearman Coef. = 0.676; P < 0.000) between hunting traces and distance from roads.

The multiple correspondence analyses between mammals, biophysical and human variables, highlight the factors explaining the distribution of fauna within the landscape (Fig. 1). Axis 1, which best explains the sample variance (50.3%), opposes plots with a dense canopy located far from roads on clay soils to open canopy plots located on sandy soils. Axis 2 (9.2% of the variance) opposes hunted plots with few lianas and herbaceous plants to un-hunted plots with abundant lianas and herbaceous plants. The results of the MCA show that the distribution of mammals within the forest concession is much more influenced by roads and hunting than by direct effect of logging. The structure of the canopy and that of the understorey are also important factors explaining the distribution of mammals within the concession. The first axis has the highest correlation with most species; small diurnal monkeys and C. sylvicultor are particularly well represented along this axis. Small monkeys show a strong link to sandy plots with an open canopy located 3–10 km from main roads. Cephalophus sylvicultor is linked to clayey plots in lowland dense forest without herbaceous plants in the understorey, located farther than 1 km from waterways and 10 km from main roads. Red duikers and elephants are well represented by axis 2. The red duikers are present in un-hunted plots with abundant Marantaceae/Zingiberaceae and lianas. Elephants appear to prefer dense forest plots, hunted or not, located uphill, less than 3 km from main roads, with some Marantaceae/Zingiberaceae and lianas and rocky soil.

Since distance to main roads (significantly correlated to hunting activities), appears to be the main human factor influencing mammal distribution, we assessed the nature and significance of the correlation between the probability of encounter (P = # of observations for species i divided by the number of plots at distance j) and the distance from main roads, as shown in Fig. 2. Linear regressions and Spearman correlation tests were only computed for those species that presented a clear distribution pattern. Syncerus caffer, Pan troglodytes, Atherurus africanus, Hyemoschus aquaticus, Gorilla gorilla and Potamochoerus porcus do not exhibit a clear distribution pattern in relation with distance from main roads. Small monkeys have a distribution that follows a humped back pattern: the probability to see small monkeys increases with distance to roads until a distance of 7 km, and then diminishes (Figs. 3, 4). Duikers’ abundance increases with distance to roads but the correlation between the encounter probability and distance from roads is only significant for red duikers (P < 0.001). Loxodonta africana is equally found at all distances from roads (Fig. 5). The weak correlation (R² = 0.0032), suggests no real links between elephant abundance and distance from roads. Cephalophus sylvicultor showed no linear correlation between abundance and distance to roads (Fig. 6). However, the result of the t-test showed significant differences between the abundance at less than 10 km and at more than 10 km from the main roads (P = 0.007).
https://static-content.springer.com/image/art%3A10.1007%2Fs10531-007-9300-5/MediaObjects/10531_2007_9300_Fig2_HTML.gif
Fig. 2

Relationship between mammal distribution and ecological and human factors as shown by axis F1 and F2 of the Multiple Correspondence Analyses

https://static-content.springer.com/image/art%3A10.1007%2Fs10531-007-9300-5/MediaObjects/10531_2007_9300_Fig3_HTML.gif
Fig. 3

Small monkeys’ distribution in relation to distance from roads show a humped back pattern (P = # of observations for species i divided by the number of plots at distance j)

https://static-content.springer.com/image/art%3A10.1007%2Fs10531-007-9300-5/MediaObjects/10531_2007_9300_Fig4_HTML.gif
Fig. 4

Duikers are positively correlated to distance from roads (P = # of observations for species i divided by the number of plots at distance j)

https://static-content.springer.com/image/art%3A10.1007%2Fs10531-007-9300-5/MediaObjects/10531_2007_9300_Fig5_HTML.gif
Fig. 5

Loxodonta africana is equally found at all distances from roads (P = # of observations for species i divided by the number of plots at distance j

https://static-content.springer.com/image/art%3A10.1007%2Fs10531-007-9300-5/MediaObjects/10531_2007_9300_Fig6_HTML.gif
Fig. 6

Cephalophus sylvicultor shows no linear relation between abundance and distance to roads (P = # of observations for species i at distance j) but is more abundant far from roads as shown by the comparison of abundance at less than 10 km and more than 10 km from roads

Discussion

Some of the most hunted species for the consumption by local people occur mainly far from areas with a significant human activity. The red duikers and C. sylvicultor avoid hunted zones and are significantly more abundant far from roads. Laurance et al. (2006) have shown similar results in South-east Gabon, concerning the impact of roads on duiker distribution. Small diurnal monkeys are found far from the villages and between 3 and 10 km from the main roads. Similarly, Blom et al. (2004) found that all monkeys in the Dzanga-Sangha National Park in Central African Republic were positively associated with increasing distance from the main village of Bayanga. Decreasing hunting pressure linked to distance to roads could explain the increase of small monkeys’ abundance until 7 km distance from roads. Further from this limit, very few people seem to hunt anymore and their distribution becomes more determined by the structure of the forest (with a preference for open canopy forests). Likewise Laurance et al. (2006) found no linear relationship between small monkeys’ abundance and distance from roads.

Other important species, on the contrary, do not seem to be affected by human presence. The distribution of Hyemoschus aquaticus, Pan troglodytes, Gorilla Gorilla, Atherurus africanus, Syncerus caffer and Potamochoerus porcus, does not seem no be influenced by human presence as these species show no negative relationship with distance to roads and were regularly observed close to villages. In our study site, elephants were found equally near to or far from roads and do not seem to be affected by hunting activities. This is contrary to what Blom et al. (2004) demonstrated in a Dzanga-Sangha where elephants avoided the proximity of roads. According to Barnes et al. (1991), elephants are attracted by secondary forests given the diversity of available food resources and, in our site, elephants were often found in plots with dense herbaceous understorey usually occurring in old secondary forests.

Past and present logging activity signs can be found all over the concession, apparently without significantly affecting the distribution of mammals. From results obtained in Kibale Forest (Uganda), we would have expected duiker’s abundance to be greater in unlogged than in logged plots. Cephalophus monticola seemed particularly affected by forest logging activities (Struhsaker 1998) this is not the case in our site. At Lopé (Gabon), densities of Pan troglodytes dropped about 20% after logging (White 1998). In our study site, where logging has been more or less continuous since the 1950s, Pan troglodytes is still present and does not seem avoiding logged over areas. However, contacts with Pan troglodytes were scarce and their populations might already been drastically reduced. This species should be monitored to ensure that it is not being driven to local extinction.

The type of understorey seems another determinant factor for the distribution of duikers. The MCA shows that the red duikers prefer the forests where the understorey is covered by Marantaceae and lianas. The use of low dense vegetation as a visual protection during the day and the use of more open understorey during the darkest hours of the day or at night, can explain the importance for duikers of forest habitats’ structure (Dubost 1980).

Conclusions

Our study shows that usual multivariate analyses allow the extraction of ecological patterns out of data collected mainly for production objectives during forest management inventories. The use of forest inventories’ data at regional scales, and the choice of appropriate analysis methods provide the possibility to highlight certain properties of ecosystems, such as the relationships between fauna, habitat, floristic composition and human activities. An important consideration to keep in mind is that more than 30 millions of hectares have been recently inventoried in the region (Nasi et al. 2006)

Results of our analyses can be used as practical recommendations for managers in order to limit the negative impacts of logging activities on fauna within forest concessions under sustainable management. The road network seems to be at the hart of the problem since hunting intensity is strongly correlated with distance to roads. An optimal planning of the road network should help limiting the direct impacts (avoiding passing close to the National Park boundaries, limiting secondary roads...) while a better control of access (blocking used skidtrails or secondary roads, manned gates on main roads...) should help limiting commercial hunting activities. The results of this study show also that some common game species (mainly A. africanus but also C. monticola ) are resistant to human pressure such as habitat degradation or hunting. A sustainable hunting management plan could be considered for such species so as to satisfy local people’s needs. On the other hand, for vulnerable species, such as Pan troglodytes, a monitoring programme should ensure the maintenance of biodiversity within the logging concessions.

Acknowledgements

We would like to thank the Compagnie des Bois du Gabon (CBG), which kindly accepted to share the database used in this study. We also thank Benoit Demarquez and Cyril Pelissier from TEREA (for the elaboration of the Geographical Information System that was used to build mammal distribution maps and for their relevant comments on this paper.

Copyright information

© Springer Science+Business Media B.V. 2007