How useful are thermal vulnerability indices?

To forecast climate change impacts across habitats or taxa, thermal vulnerability indices (e.g., safety margins and warming tolerances) are growing in popularity. Here, we present their history, context, formulation, and current applications. We highlight discrepancies in terminology and usage, and we draw attention to key assumptions underpinning the main indices and to their ecological and evolutionary relevance. In the process, we flag biases influencing these indices that are not always evaluated. These biases affect both components of index formulations, namely: (i) the characterisation of the thermal environment; and (ii) an organism’s physiological and behavioural responses to more frequent and severe warming. Presently, many outstanding questions weaken a thermal vulnerability index approach. We describe ways to validate vulnerability index applications and outline issues to be considered in further developing these indices. Accurate forecasts of the impacts of climate change are essential for managing biodiversity and minimising further species losses. Indices reflecting forecasts are useful when they are simple enough to be implemented at large scales, but this needs to be balanced against their ability to predict actual impacts.Thermal vulnerability indices such as safety margins and warming tolerance have increased in popularity for making both regional and global predictions of species vulnerability to climate change, but concerns have been raised about their forecasting potential.By estimating the deviation between species’ thermal sensitivity and thermal exposure, these indices often do not reflect variation relevant to the organism. They also fall short of considering many intrinsic and extrinsic factors that affect species vulnerability, weakening their usefulness.Under specific circumstances, safety margins can however answer particular questions relating to species’ relative sensitivity and exposure, and can be complemented by other approaches, all of which require validation.