Drivers of mammal communities across multi-dimensional landscape gradients in the semi-arid Baviaanskloof catchment, South Africa

Abstract

Private lands are vital for biodiversity conservation, particularly when protected areas are insufficient to maintain ecological processes and species persistence. In mixed land-use areas like the Baviaanskloof catchment in the Eastern Cape, South Africa, conserving biodiversity could protect up to half of the country's wild species, given that nearly 80% reside outside protected areas. This study focuses on the biodiversity of medium to large mammals within this mixed-use landscape, where natural habitats intermingle within protected areas and agricultural lands. From January 2020 to April 2022, 131 camera traps captured data on 34 species across 21,020 trap days in both agricultural zones and the Baviaanskloof Nature Reserve, spanning a range of habitats including forest, savanna, fynbos, agricultural lands, and thicket, from intact to severely degraded states. Because survey design can bias detection, camera-trap configurations were quantified for their influence on species detectability to optimise mammalian biodiversity assessments. Furthermore, the study assessed species occupancy in response to a combination of topographic variables, such as landscape units, aspect, slope, ruggedness, and solar gain, and anthropogenic factors including proximity to human settlements and the impacts of livestock and crop farming. This comprehensive approach allowed for a nuanced understanding of how various environmental and human factors influence wildlife populations in mixed-use landscapes. Bayesian multi-species occupancy models were employed to estimate species occupancy, richness, and community composition, influenced by these land use and environmental variables. Biodiversity indices showed significant species diversity variability across different camera trap deployments. Logistic growth models suggested that species discovery plateaued at around 153 sampling days, though detecting rarer species might require prolonged monitoring. The study also examined how camera placement, survey duration, and setup parameters like elevation, angle, and positioning affected species detectability and capture frequency. Road-based cameras more frequently detected species that regularly use these paths, whereas cameras set away from roads more effectively captured elusive species like cryptic carnivores that shy away from areas with high human activity. Cameras set between 40 to 70 cm above ground improved detections, whereas those over 100 cm had reduced effectiveness. Optimal detection angles ranged from 50° to 80°, with north-and south-facing setups having higher capture rates than west-facing ones. Vegetation transects provided insights into ground cover, height, and structural complexity, and horizontal visibility measurements at various distances helped evaluate line-of-sight impacts on species occupancy and habitat selection. Distinct preferences for vegetation types were evident among the species. Forest areas, with dense tree cover and abundant shrubs, supported species requiring concealment, such as carnivores and primates. Savannas, featuring medium to high visibility, attracted herbivores, which benefit from clear sightlines for spotting predators. Fynbos habitats, characterised by dense, low-lying vegetation, were favoured by smaller herbivores and carnivores. Agricultural settings attracted species adapted to anthropogenic food sources, utilising the altered landscape. Thicket habitats, depending on their level of disturbance, supported different levels of occupancy; intact thickets catered to species needing concealment, while degraded thickets still accommodated adaptable species. Disturbance gradients significantly influenced habitat suitability, with intact thickets showing the highest mammalian richness, whereas severely degraded areas, characterised by increased visibility and reduced vegetation complexity, adversely affected species occupancy. This study underscored the critical role of microhabitat vegetation structure in determining habitat utilisation across species. Topographical complexity identified floodplains, valleys, and areas with lower slopes and low ruggedness as biodiversity hotspots, vital for providing water and forage. These features supported a high species richness and facilitated coexistence. Conversely, steeper, more rugged terrains with higher solar gain served as crucial refuges for specialised taxa. The integration of detailed vegetation and topographic metrics into conservation strategies is particularly important in mountainous landscapes, where these factors significantly influence species occupancy and distribution. This study demonstrates how private lands can complement protected areas, offering crucial habitats for a diverse range of species and highlighting the potential of sustainable land-use practices to balance biodiversity conservation with agricultural productivity. This approach underscores the importance of community-based initiatives, habitat connectivity, and landscape-scale planning to ensure the sustainability of multidimensional semi-arid landscapes like the Baviaanskloof.