Significance

Species distribution models (SDMs) which identify statistical correlations between species presence/absence and various land cover and climate explanatory variables are increasingly used to generate predictions of species’ current and future distributions. A link between the probability of occurrence scores generated by SDMs and population-level parameters such as abundance is often inferred but it remains unclear whether SDM predictions are truly indicative of population processes. We find that, while SDMs built using species detection/non‐detection data can produce probability of occurrence estimates that are reasonable predictors of population density in butterflies and birds, they are generally poor at predicting population stability. For additional insights into the factors affecting species’ extinction risks which cannot easily be inferred from species’ occupancy data, count data from standardised monitoring schemes remain key.