Remotely sensed vegetation productivity predicts breeding activity and drought refuges for a threatened bird in semi‐arid Australia

Abstract

AbstractRefuges play a critical role protecting species against the effects of climate change. Managing high priority refuges could improve species resilience and facilitate dispersal during periods of environmental change. In this study, we identified drought refuges in semi‐arid Australia for a threatened bird, the malleefowl Leipoa ocellata. Using a Poisson regression model, we quantified the effect of remotely sensed vegetation indices, rainfall, soil moisture and site characteristics on malleefowl breeding activity at 144 sites surveyed from 2000 to 2017 during and after drought. We tested the effect of two vegetation productivity indices on malleefowl breeding activity – the Normalized Difference Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI) – averaged across three temporal and three spatial scales during the mound building and incubation stage of the breeding cycle. We found that NDVI and EVI were better predictors of malleefowl breeding activity than soil moisture and winter rainfall. The model with the lowest Akaike information criterion value contained NDVI averaged over 3 months (June–August) and a 1‐km radius. Malleefowl breeding activity had a strong positive association with NDVI (0.42 ± se 0.03) and a negative association with slope (−0.34 ± se 0.03) and vegetation patch size (−0.23 ± se 0.02). We found the proportion of refugia (top 20% of predicted breeding activity) in protected areas was highly variable, decreasing from 42% in an extreme wet year (2011) to 14% in an extreme drought year (2007). Expanding the reserve network to include refugia predicted to occur in the south of semi‐arid Victoria could improve resilience of malleefowl to climate change. We demonstrate how remotely sensed vegetation indices combined with citizen science data can identify where to protect native vegetation with high, stable productivity. Our approach could be applied to a broad range of species in semi‐arid regions vulnerable to climate change.