Can spatial distribution of ungulates be predicted by modeling camera trap data related to landscape indices? A case study in a fragmented mediterranean landscape

Abstract

Camera trap applications range from studying wildlife habits to detecting rare species, which are difficult to capture by more traditional techniques. In this work, we aimed at finding the best model to predict the distribution pattern of wildlife and to explain the relationship between environmental conditions with the species detected by camera traps. We applied two types of statistical models in a specific Mediterranean landscape case. The results of both models shown adjustments over 80 %. First, we ran a Principal Components Analysis (PCA). Discriminant, and logistic analyses were per-formed for ungulates in general, and three species in particular: Barbary sheep, mouflon, and wild boar. The same environmental conditions explained the presence of these species in all the proposed models. Hence, we proved the generally positive influence of patch size on the presence of ungulates and negative influence of the fractal dimension and density edge. We quantified the relationships between a suite of landscape metrics measured in different grids to test whether spatial heterogeneity plays a major role in determining the distribution of ungulates. We explained much of the variation in distribution with metrics, specifically related to habitat heterogeneity. That outcome highlighted the potential importance of spatial heterogeneity in determining the distribution of large herbivores. We discussed our results in the forestry conservation practices context and discuss potential ways to integrate ungulate management and forestry practices better.

Keywords

• camera trap
• discriminant analysis
• landscape metrics
• logistic analysis
• multivariant analysis