Date Published: April 19, 2019
Publisher: Public Library of Science
Author(s): Nathalie Cavada, Rasmus Worsøe Havmøller, Nikolaj Scharff, Francesco Rovero, Angela K. Fuller.
With biodiversity facing unparalleled threats from anthropogenic disturbance, knowledge on the occurrences of species and communities provides for an effective and fast approach to assess their status and vulnerability. Disturbance is most prominent at the landscape-level, for example through habitat loss from large-scale resource extraction or agriculture. However, addressing species responses to habitat changes at the landscape-scale can be difficult and cost-ineffective, hence studies are mostly conducted at single areas or habitat patches. Moreover, there is a relative lack of studies on communities, as opposed to focal species, despite the former may carry more comprehensive information. Here, we used a multi-region, multi-species hierarchical occupancy model to study a meta-community of mammals detected by camera traps across five distinct areas within a heterogeneous landscape in Tanzania, and aimed to assess responses to human disturbance and environmental variables. Estimated species richness did not vary significantly across different areas, even though these held broadly different habitats. Moreover, we found remarkable consistency in the positive effect of distance to human settlements, a proxy for anthropogenic disturbance, on community occupancy. The positive effect of body size and the positive effect of proximity to rivers on community occupancy were also shared by communities. Results yield conservation relevance because: (1) the among-communities consistency in responses to anthropogenic disturbance, despite the heterogeneity in sampled habitats, indicates that conservation plans designed at the landscape-scale may represent a comprehensive and cost-efficient approach; (2) the consistency in responses to environmental factors suggests that multi-species models are a powerful method to study ecological patterns at the landscape-level.
Addressing the current unparalleled loss of biodiversity requires an understanding of what shapes ecological communities, especially those exposed to anthropogenic threats, how they vary in space and time, and what are the implications of such variation [1–3]. This is because determining vulnerability of communities likely yields greater potential for informing conservation than single species or guilds . Community level approaches not only provide inference on individual species, but they also address patterns at the ecosystem level, which is an ecologically- and cost-effective approach for management actions . Moreover, even if management actions might be designed for a limited set of target species, management likely exerts an influence that extends to other species within a landscape . In this context, tropical forest mammals are of central relevance: they are among the most threatened taxa on earth and yet they carry out fundamental ecosystem functions as seed consumers and dispersers, predators, and prey . Changes in mammalian communities are therefore likely to have consequences for ecosystem stability . Moreover, the larger-bodied species tend to require large undisturbed areas and are particularly impacted by uncontrolled threats such as logging, hunting and environmental degradation [9–11], to which they are variably sensitive [12,13]. Previous research on anthropogenic effects on mammal communities has had a marked focus on variation of species richness [1,14] and abundance of individual species [3,15]. Results from these studies generally point to a decline in both metrics across gradients from intact versus degraded, modified and/or hunted habitats. Relative less attention, however, has been devoted to metrics that describe the overall abundance and distribution of communities, such as occupancy . This is important because average community occupancy may provide for a finer resolution assessment of community responses to anthropogenic disturbance and environmental change than species richness . This, in turn, may provide for an informative tool in conservation management .
Camera traps yielded 3725 events per day of 48 species of medium-to-large, ground-dwelling mammals: 29 species in Matundu, 33 in Mbatwa, 24 in Lumemo, 25 in Ndundulu and 24 in Mwanihana (see S1 Table for the checklist of species by taxonomic order and body mass). Only 13 of 48 species detected were shared among all study areas. These included large and wide-ranging, or common herbivores such as elephant (Loxodonta africana), African buffalo (Syncerus caffer), and bushbuck (Tragelaphus scriptus), small-bodied antelopes such the Harvey’s duiker (Cephalophus harveyi) and suni (Neotragus moschatus), the aardvark (Orycteropus afer), as well as both large (such as the leopard, Panthera pardus), and small (bushy-tailed mongoose, Bdeogale crassicauda) carnivores. Estimates of community size (median) at the region level ranged from 28 species in Lumemo to 37 species in Mbatwa, with 95% BCIs that overlapped among regions (Table 3).
We studied species richness and community occupancy across five sampled areas in a highly heterogeneous landscape in Tanzania. We investigated how habitat variables and anthropogenic disturbance affect meta-community occupancy while also analysing species-specific effects, thereby conducting a comprehensive evaluation of mammalian occurrence patterns at the landscape scale. Despite the pronounced differences in community composition (only 27% of species occurred in all study areas), and the heterogeneity of habitats where they occur, we found shared patterns of responses by the whole meta-community to the drivers considered. In particular, we found that occupancy of the meta-community was positively related to the distance from human settlements with a significant relationship. We consider this covariate a proxy for decreasing human disturbance. The shared pattern of responses we found is in line with previous findings on generalized mammal responses to extent of protected habitat and disturbance across the Eastern Arc Mountains , and, more specifically, with similar findings from studies on the abundance of leopard  and of arboreal primates in the Udzungwa Mountains . The latter study, in particular, found similar responses for the density of three species of arboreal primates to signs of human disturbance, with these signs being generally more abundant in forests more heavily surrounded by human settlements. These evidences altogether point to a marked impact of anthropogenic disturbance on species and communities at the landscape-level, and suggest that the mammal meta-community in the area is facing threats from anthropic settlements and activities regardless of the high variability in species composition and habitat type. While the increasing impact on mammals near anthropic areas and the positive effect of protected habitat are documented patterns [10,53,54], our finding of consistent responses at the meta-community level is noteworthy and of conservation relevance, as it suggests that effective conservation plans could be designed and coordinated across the landscape. On the contrary, current conservation efforts tend not to be coordinated, with different protected areas and authorities involved, and management strategies that are not harmonized across the landscape [37,55]. As a result, there are areas that are relatively well protected and areas that are not, regardless of their biodiversity importance. Therefore, coordinating protection efforts appears a strategy to increase the cost-effectiveness of law enforcement, as well as guide the design of comprehensive conservation measures such as buffer zones, wildlife management areas, and community-conservation schemes. Moreover, the among-species consistency in responses to covariates within communities supports the notion that species-specific conservation action plans (typically designed for endemic and/or highly threatened and charismatic species) will likely provide for benefits to the whole guild or community of mammals .