Modelling the distribution of a key habitat feature to guide future on-ground habitat assessment for an endangered specialist songbird

Abstract

Species distribution models (SDMs) are often used during the planning stage of reintroduction programmes to model species' occurrence with the aim of selecting potential release sites. However, for many endangered species, only a low number of records are available, leading to models with low accuracy. When planning reintroductions for rare species, an alternative approach may be to model surrogate species that are more abundant or easier to locate. Here, we modelled the distribution of white gum (Eucalyptus viminalis), the preferred food tree of the forty-spotted pardalote (Pardalotus quadragintus), a rare songbird for which reintroduction has been proposed. Using boosted regression trees, we modelled white gum distribution under current and future climate conditions with the aim of identifying areas of high probability of occurrence that later can be used to plan on ground habitat assessments for reintroductions. Our model show areas with high probability of white gum occurrence outside its currently mapped distribution, indicating that there may be opportunities for reintroduction of pardalotes beyond their current range. Predictions of future climate scenarios showed climate space shifts, not only with some decrease but also with substantial increase in the probability of suitability for occurrence under some scenarios. Our spatial predictions for white gum may be used to design a survey to ground-truth our model and undertake a comprehensive habitat assessment for other habitat features forty-spotted pardalotes need to persist. The approach used in our study may be used for other highly specialized species, not only in the context of reintroduction planning but also in the general management of data-poor specialist species that depend on a more common resource.