Habitat‐ and rainfall‐dependent biodiversity responses to cattle removal in an arid woodland–grassland environment.

Biodiversity conservation in rangeland environments is often addressed by removing livestock, but inconsistent responses by biota mean that the efficacy of this form of management is hotly debated. Reasons for this inconsistency include the usually short duration and small spatial scale of manipulations compared to the area of grazing properties, as well as divergent responses amongst biota. In low-productivity arid environments, the pulse-reserve dynamic also complicates the outcome of manipulations. Here, we tested and extended these ideas in a heterogeneous desert environment in central Australia that consists of small patches of open woodland (gidgee) in a grassland (spinifex) matrix. Taking advantage of a controlled property-scale removal of cattle, and a rain event that stimulated productivity, we first quantified differences in the vegetation and small vertebrates of these two habitats, and then tracked the diversity, composition, and abundance of these biota for 6–19 months post-rain. We predicted that the two habitats would differ in the structure, composition, and reproductive output of their constituent plant species. We predicted also that the effects of cattle removal would interact with these habitat differences, with the abundance, richness, and diversity of small mammals and reptiles differing across habitats and grazing treatments. As anticipated, plant species composition in woodland was distinct from that in grassland and varied over time. The effects of cattle removal were habitat specific: Plant composition responded to de-stocking in woodland, but not in grassland; flowers were more abundant, and palatable plant cover also was greater following cessation of grazing pressure. The responses of small mammals but not reptiles showed some accord with our predictions, varying over time but inconsistently with treatment, and perhaps reflected high variability in capture success. We conclude that the timing and length of sampling are important when evaluating the responses of biota to livestock removal, as is the inclusion of all key habitats in the sampling regime.