Maritz, BryanClegg, Bruce WVan Wyk, Kurt M2026-10-142026-10-142026https://hdl.handle.net/10566/25603Reptiles are the most diverse class of terrestrial vertebrates in the world, yet they remain poorly studied and poorly protected. Due to taxonomic bias in the allocation of funds for conservation and research, a general lack of public interest and awareness, and the low detectability of most species, we know relatively little about how reptiles are responding to global anthropogenic change. However, a growing body of evidence suggests that their response to habitat transformation is widely negative. While advances in the field of ecological modelling have greatly improved our ability to account for imperfect detection when estimating abundance, it is unclear whether these tools can provide reliable estimates for African savanna reptiles. In sub-Saharan Africa, reptile abundances are thought to have declined to 56–77% of their estimated pre-industrial values, primarily as a consequence of agricultural land-use change. Zimbabwe’s land-use history and current protected area network form an opportune context in which to study the impact of historical land-use change on reptiles while testing different approaches to estimating the relative abundance of different species at different spatial scales. South-eastern Zimbabwe is a region of national and international conservation significance. As the current status of the country’s reptiles is largely unknown, I provide a systematic account of all known reptile occurrence records in the region to inform national conservation objectives. The examined records, derived from museum data, field research, and citizen science, span the years 1818–2024 and comprise 121 taxa, of which four are of global conservation concern. This account increases the number of published reptile occurrence records from the region by an estimated 285%. While the diversity of species present in an area may be known, failure to account for imperfect detection can produce erroneous estimates of abundance, which can have severe consequences for species conservation. Using a dataset of reptile observations from Malilangwe Wildlife Reserve in south-eastern Zimbabwe, I compare the abundance estimates of five N-mixture model variants, which explicitly account for variation in detectability. Acceptable model fit was only achieved for the most frequently detected species and varied significantly both between model variants and according to species occupancy and detection probability. Pairs of model variants were rarely congruent in their abundance estimates, such that fitting a single model variant to data sampled from an ecologically diverse community has the potential to yield artefactual variation in abundance estimates. With more data, the a priori matching of species and method of abundance estimation may be possible, but rigorous and adaptive survey design may remain a more reliable means of avoiding bias overall. Spatially replicated trapping surveys are a robust method of sampling reptile communities. Using three seasons of trap data, I compared reptile counts in abandoned cotton fields and adjacent, untransformed woodland on Malilangwe Wildlife Reserve and modelled the effect of individual habitat structure variables on abundance. Diversity was 14–33% lower in historically transformed habitat, while abundance was nearly 75% lower, more than six decades after transformation took place. The results suggest that the decline is best explained by the structural homogenisation of habitat, although differential species responses point to multiple, overlapping mechanisms. The results caution against the maintenance of land-use legacies in transformed savanna landscapes, in terms of both habitat structure and community composition. The relationship between species richness and area is widely used to describe ecological patterns and to inform conservation planning, but is commonly non-linear, limiting the scalability of primary ecological data. I demonstrate that reptiles are particularly prone to exhibiting some of the shortfalls of the species-area relationship. Flexible models that account for sources of non-linearity substantially improve estimates of species richness, informing biodiversity surveys and protected area management. This also improves the interpretability of survey results and species inventories across spatial scales, linking localised surveys with landscape conservation. Through analyses of diversity and abundance at multiple spatial scales, this thesis emphasises that reptile ecology, and perhaps community ecology as a whole, is limited by both the availability and quality of data. Nevertheless, I argue that explicitly acknowledging and accounting for these limitations will ultimately improve conservation outcomes for reptiles – both by motivating for greater support for reptile conservation and research, and by improving confidence in the accuracy and scalability of results. There is ample scope to expand this work in African savannas, where land use pressure and reptile diversity remain high.enCommunity ecologyHerpetologyImperfect detectionLand-use changeN- mixture modelThesis