Non-random patterns of vegetation clearing and potential biases in studies of habitat area effects

Context: Native vegetation extent is often a proxy for habitat area in studies of human-modified landscapes. However, the loss and retention of native vegetation is rarely random among landscapes. Instead, the extent of native vegetation in landscapes may be correlated with abiotic factors, thereby obscuring or distorting relationships between ecological phenomena and area. Objectives: We asked: (1) how has the potential for non-random vegetation loss to confound area effects been addressed in the landscape ecology literature? (2) Are consistent patterns of non-random vegetation loss and retention evident from modified regions of two countries? Methods: We reviewed 118 papers that related area to an ecological response, to determine whether potential biases associated with non-random vegetation loss and retention were considered. We then analysed ~18,000 100 km landscape units in Australia and South Africa to identify how different abiotic factors correlate with the extent of native vegetation retained in those landscapes. Results: Only 21% of the studies we reviewed explicitly or implicitly considered spatial biases in vegetation clearing. Yet, across modified regions of Australia and South Africa, landscape-scale native vegetation extent was consistently and often strongly related to abiotic factors, particularly soil properties and topographic variability. Conclusion: Patterns of vegetation clearing and retention commonly reflect underlying abiotic heterogeneity. These biases, which are infrequently highlighted in studies focussing on area effects, have implications for how we assess the importance of vegetation extent for species and assemblages. Failure to account for correlates of vegetation extent risks erroneous area-based conservation prescriptions in human-modified environments.