Piosphere contribution to landscape heterogeneity: a case study of remote‐sensed woody cover in a high elephant density landscape

Abstract

Vegetation gradients developing around water sources (i.e. piospheres) are important features of arid and semi‐arid ecosystems. Studied extensively in pastoral areas, piospheres have rarely been investigated in areas hosting rich herbivore diversity. We studied piospheres in woody cover assessed through remote‐sensing in Hwange National Park, Zimbabwe, which has one of the world's largest elephant populations. As no preferred statistical model for piosphere studies has emerged, we first contrasted results from ordinary least‐square (OLS) regressions on raw data with those from other statistical models (OLS on aggregated data and simultaneous autoregressive models on raw data), and selected the most parsimonious, unbiased, model to study the influence of artificial and natural waterholes, and the abundance of elephants, others browsers, and grazers on piospheres. OLS models provided unbiased parameter estimates, despite the strong spatial autocorrelation present in woody cover data, whereas other statistical models had important drawbacks. Using an OLS framework we showed that despite an important negative non‐linear mean effect, distance‐to‐water was a poor predictor of woody cover at any location. Woody cover was on average more reduced in the vicinity of water at artificial waterholes than at natural waterholes. Elephant abundance was not consistently associated with lower woody cover, and poorly explained woody cover heterogeneity, as did all other herbivore‐related variables. Our study indicates that piospheres may develop differently in pastoral and protected areas, suggesting the importance of herbivore diversity in ecosystem functioning. Our results also show that heterogeneity in woody cover persists within piospheres, calling for further investigation on the origin and role of this heterogeneity in the maintenance of ecological processes and biodiversity within these key‐areas of the landscape.