Landscape-specific thresholds in the relationship between species richness and natural land cover

Thresholds in the relationship between species richness and natural land cover can inform landscape-level vegetation protection and restoration targets. However, landscapes differ considerably in composition and other environmental attributes. If the effect of natural land cover on species richness depends on (i.e., interacts with) these attributes, and this affects the value of thresholds in this relationship, such dependencies must be considered when using thresholds to guide landscape management. We hypothesized that the amount of natural land cover at which a threshold occurs would differ in predictable ways with particular anthropogenic, abiotic, and biotic attributes of landscapes. To test this, we related woodland bird species richness in 251 landscapes, each 100\ua0km, to natural land cover in south-east Australia. We compared the fit of exponential and threshold models of the richness–natural land cover relationship, focussing on the extent of natural land cover at which thresholds presented among landscapes that differed in matrix land use intensity, heterogeneity, productivity, and the prevalence of strong biotic interactors. We used linear mixed modelling to examine how interactions between natural land cover and the various landscape attributes affected the fit of models of species richness. Threshold models of the richness–natural land cover relationship were always a better fit than exponential models. Threshold values did not vary consistently with specific landscape attributes, with the exception of landscapes that were classified by the prevalence of strong biotic interactors (hypercompetitive native birds of the genus Manorina). Natural land cover had a more positive effect on species richness in landscapes when Manorina prevalence was higher. This positive interaction provided the biggest improvement in explanatory power of models of species richness. Synthesis and applications. While we detected an interaction between Manorina prevalence and the area of natural land cover, generalities relating to the underlying nature of thresholds in the richness–natural land cover relationship remain elusive. Complex interactions, relating to various landscape attributes and associated ecological processes, likely underpin variation in threshold values. Until these complexities are better understood, the use of thresholds for informing landscape management and conservation target setting should be approached with caution