Complex Variation in Afrotropical Mammal Communities With Human Impact

Abstract

The diversity and composition of mammal communities are strongly influenced by human activities, though these relationships may vary across broad scales. Understanding this variation is key to conservation, as it provides a baseline for planning and evaluating management interventions. We assessed variation in the structure and composition of Afrotropical medium and large mammal communities within and outside protected areas, and under varying human impact. We collected data at 512 locations from 22 study sites in 12 Afrotropical countries over 7 years and 3 months (2011-2018) with 164,474 camera trap days in total. Half of these sites are located inside protected areas and half in unprotected areas. The sites are comparable in that they all harbor at least one great ape species, indicating a minimum level of habitat similarity, though they experience varying degrees of human impact. We applied Bayesian Regression models to relate site protection status and the degree of human impact to mammal communities. Protected area status was positively associated with the proportion of all threatened species, independent of the degree of human impact. Similarly, species richness was associated with area protection but was more sensitive to human impact. For all other attributes of the mammal communities, the pattern was more complex. The influence of human impact partially overrides the positive effects of protected area status, resulting in comparable mammal communities being observed both within protected areas and in similarly remote locations outside these areas. We observed a common pattern for large carnivores, whose probability of occurrence declined significantly with increasing human impact, independent of site protection status. Mammal communities benefit from sustainability measures of socio-economic context that minimize human impact. Our results support the notion that conservation of mammalian species can be achieved by reducing human impact through targeted conservation measures, adopting landscape-level management strategies, fostering community engagement, and safeguarding remote habitats with high mammal diversity.

Keywords: camera traps; mammal community composition; protected areas; threatened species; trophic guilds.

FIGURE 1

Location of PanAf research sites included in this study. The background represents estimated aboveground carbon density derived from Baccini et al. (2012). Greener colors indicate wetter forest‐dominated habitats and redder colors indicate drier, more open savanna or woodland habitats. Countries included in this study are labeled on the map following standard ISO alpha‐3 codes.

FIGURE 2

Posterior distribution and density of human footprint, site protection and the interaction between them on biodiversity variables. The plots show the estimates (dots; mean of the posterior distribution) and the 67%, 87%, and 97% credible intervals (blue bars). Additionally, the density of the posterior distribution is shown as a curved line above the horizontal credibility intervals.

FIGURE 3

Posterior predictions of protection status on biodiversity variables represented as violin plots. The colored rectangles depict the 50% credibility intervals and the whiskers the 97% credibility intervals of the predicted posterior distribution (the horizontal black line depicts the mean). The lighter lines represent 150 draws from the posterior. Each dot represents a site and darker colors represent overlapping sites. Turquois colors represent protection, and red color represents unprotected areas.

FIGURE 4

Posterior predictions for human footprint on biodiversity variables. The darker line represents the mean of the posterior distribution, and the lighter lines represent 150 draws from the posterior. The size of the circles indicates the sample size per value combination.

FIGURE 5

Posterior predictions for the interaction between protection status and the human footprint on biodiversity variables are represented by the continuous lines, where the darker one is the mean of the distribution, and the lighter lines represent 150 draws from the posterior. The size of the circles indicates the sample size per value combination.