Andrew, ME; Fox, E (2020). Modelling species distributions in dynamic landscapes: The importance of the temporal dimension. JOURNAL OF BIOGEOGRAPHY, 47(7), 1510-1529.

Aim Temporal variation in environmental conditions merits greater consideration in species distribution models (SDMs), and is especially relevant in dynamic systems, such as arid and semi-arid environments, where precipitation and resource pulses are unpredictable in space and time. Species responses to this spatio-temporal variation include tracking resource availability with nomadic life histories or reliance on refugia that provide stable, high-quality habitat. This study evaluated whether inclusion of temporal variables representing short- and long-term environmental variation improves SDMs. Location Great Western Woodlands, Western Australia. Taxon Birds. Methods Temporal information was added to SDMs of 36 bird species with two types of environmental predictor variables: (a) time-specific estimates of gross primary productivity (GPP) and precipitation paired to survey dates; and (b) estimates of the temporal stability of GPP over the MODIS archive, providing an index of the degree to which sites might function as habitat refugia from disturbance and drought. Predictions were compared with those from SDMs using static variables from single time points. Results Model predictions revealed considerable temporal variation in species distributions and various spatio-temporal responses to environmental variation. Temporal variables modestly improved model performance overall, but results were species-specific. Time-specific environmental variables were most important to models of nomadic and nectivorous species, which experienced the greatest improvements in model performance when including temporal variables, and showed the greatest disagreement between temporal and static model predictions, and the predictions of temporal models applied to different time points. Long-term environmental variation was relevant across species groups, though the chosen measure of variability differed between species. Main Conclusions Temporal variation influences species distributions and may need to be explicitly included in SDMs. However, the nature and degree of species' responses to environmental dynamics depends on species movement behaviours; appropriate selection of temporal variables for SDMs requires ecological understanding of the species of interest. Results of spatio-temporal SDMs can identify the extent to which landscape management must account for temporal variability to achieve conservation goals.