No abstract

No abstract

The landscape heterogeneity hypothesis posits that increasing variance of land cover types in agricultural landscapes will increase landscape level biodiversity. This hypothesis, however, does not detail which component of landscape structural heterogeneity, compositional (the type and amount) or configurational (the shape and arrangement) has the greatest influence on biodiversity and at what spatial scale(s). We investigated how dung beetle (Scarabaeidae) alpha‐diversity responded to landscape structural heterogeneity at a variety of spatial scales in an agricultural mosaic‐landscape in north‐eastern Swaziland. We also compared the effect of these components to the effect of variation in the amounts of major land‐cover types in the landscape and plot level vegetation structure. We used pitfall traps to sample the dung beetle community along a gradient of heterogeneity and used linear mixed effects models to compare the effect of each component on dung beetle richness and Shannon diversity at four separate spatial scales (i.e. 1‐,1.5‐, 2‐ & 3‐km). Land‐cover diversity and number of patches represented compositional and configurational landscape heterogeneity respectively. Landscape compositional heterogeneity was negatively correlated with dung beetle richness at the 1.5‐km and 2‐km spatial scales. The percentage savanna in the landscape was positively correlated with dung beetle richness at the 3‐km and 2‐km scales. Landscape level heterogeneity may enhance diversity of some taxonomic groups, but this was not the case for dung beetles in a southern African savanna. The best way to maintain their diversity is to create or maintain large continuous blocks of savanna while limiting intensive agriculture.

The analysis of spatial point patterns has greatly advanced our understanding of ecological processes. However, the methods currently available for analyzing replicated spatial point patterns (RSPPs) are rarely used by ecologists. One barrier to the use of RSPP analyses is a lack of software to implement the approaches that have been developed in the statistical literature. Here, we provide a practical guide to RSPP analysis and introduce the RSPPlme4 R package that implements the approaches we discuss. The methods we outline use a linear modeling framework to link variation in the spatial structure of point patterns to discrete and continuous explanatory covariates. We describe methods for linear models and mixed-effects models of RSPPs, including approaches to estimating confidence intervals via semi-parametric bootstrapping. The syntax for model fitting is similar to that used in linear and linear mixed-effects modeling packages in R. The RSPPlme4 package also allows users to easily plot the results of model fits. We hope that this tutorial will make methods for RSPP analysis accessible to a wide range of ecologists and open new avenues for gaining insight into ecological processes from spatial data.