Habitat use of and threats to African large carnivores in a mixed‐use landscape

Abstract

Large carnivores increasingly inhabit human-affected landscapes, which exhibit heterogeneity in biotic resources, anthropogenic pressures, and management strategies. Understanding large carnivore habitat use in these systems is critical for their conservation, as is the evaluation of competing management approaches and the impacts of significant land-use changes. We used occupancy modeling to investigate habitat use of an intact eastern African large carnivore guild across the 45,000 km 2 Ruaha-Rungwa landscape in south-central Tanzania. We determined the relative impact on five large carnivore species of biotic, anthropogenic, and management factors at the scales of home range selection and short-term use within home ranges. We also specifically tested the effect of abandonment of trophy hunting areas on large carnivore occurrence. Patterns of habitat use differed among species. Lions (Panthera leo) appeared affected by top-down limitation, as their occurrence was significantly negatively associated with illegal human activity (β = -0.63 [SE 0.28]). African wild dogs (Lycaon pictus), instead, were limited by biotic features; the species was negatively associated with riverine areas of high sympatric predator density (β = -1.00 [SE 0.43]) and used less-productive habitats. Spotted hyaena (Crocuta crocuta) and leopard (Panthera pardus) persisted in more disturbed areas and across habitat types. Large carnivore occurrence was not affected by whether an area was used for photographic or trophy hunting tourism; regular law enforcement was instead a better predictor of occurrence. All species fared better in actively managed hunting areas than those that had been abandoned by operators. Overall, our findings highlight the divergent habitat requirements within large carnivore guilds and the importance of adopting an integrated approach to large carnivore conservation planning in modern systems. We also identified a novel threat to African conservation areas in the form of decreased management investments associated with the abandonment of trophy hunting areas.

Trends and biases in African large carnivore population assessments: identifying priorities and opportunities from a systematic review of two decades of research

“…In the context of large carnivore population status assessments and monitoring, population density is a commonly-employed and meaningful metric ( Boitani & Powell, 2012 ). Although other metrics of population status exist ( e.g ., occupancy— Strampelli et al, 2022a ), population density is increasingly recommended as the metric of choice to assess and monitor large carnivore populations ( Elliot & Gopalaswamy, 2016 ). In addition to estimating current status, estimating densities also allows comparison over time and between sites, enabling researchers and managers to understand how population status varies with biotic factors ( Searle et al, 2021 ), anthropogenic disturbances ( Balme, Slotow & Hunter, 2010 ; Henschel et al, 2011 ), or land management strategies ( Swanepoel, Somers & Dalerum, 2015 ).…”

Spatial co-occurrence patterns of sympatric large carnivores in a multi-use African system

“…The complex is unfenced, and is surrounded by unprotected village lands to the south and east, and by additional OAs and GCAs to the west and north. Law enforcement is generally higher in the NP and the GRs, due to greater availability and investments of resources, and lower in the GCA, OA, and WMAs [47]. Nevertheless, due to boundary disputes, human settlements and intensive agriculture are present in an area of ~2,100 km 2 in south-west Ruaha NP [48].…”

“…We employed a spatially replicated occupancy sampling approach [47,[55][56][57]. Each transect was divided into 500 m segments, and we recorded whether sign of each species was detected (1) or non-detected (0) within each segment.…”

“…Occupancy models that explicitly address the issue of spatial autocorrelation [58] cannot be implemented in two-species analyses. As a result, at the home range scale, segment length was set to account for spatial autocorrelation (Markovian dependence) in detection; this was achieved by testing for this in Program PRESENCE 2.12.37, increasing sampling occasion length until there was no more evidence of spatial autocorrelation [47,59]. As this was achieved at a length equal to or shorter than 5 km for all species [47], sampling occasions consisted of 5 km transect segments in all analyses at the home range scale.…”

“…As a result, at the home range scale, segment length was set to account for spatial autocorrelation (Markovian dependence) in detection; this was achieved by testing for this in Program PRESENCE 2.12.37, increasing sampling occasion length until there was no more evidence of spatial autocorrelation [47,59]. As this was achieved at a length equal to or shorter than 5 km for all species [47], sampling occasions consisted of 5 km transect segments in all analyses at the home range scale. At the short-term use scale, sampling occasions comprised 2-4 continuous 500 m segments, and it was therefore not possible to account for spatial autocorrelation following the method employed at the home range scale.…”