Predicting oxygen thresholds of marine taxa to improve ecological forecasts

Species' ranges are shifting in response to increasing temperature and decreasing oxygen in coastal oceans. Predicting these shifts is limited by information on physiological oxygen thresholds and how they depend on temperature. Here we collate laboratory-derived measurements of a common oxygen threshold, p_crit, for 148 animal species that span six phyla, and fit a hierarchical model based on taxonomy to impute threshold values for all based on taxonomy, body size, and environmental temperature. As expected, p_crit increased with increasing temperature and body size, and these temperature- and body size effects were broadly similar among species. Generally, variation in p_crit measurements was estimated to be most pronounced at the taxonomic family and species level, although the partitioning of variance was relatively imprecise. We demonstrate application of these estimates for species distribution modeling of six groundfish species that reside in the coastal waters of the US and Canadian Pacific coast, finding that models that used imputed values sometimes – but not always – improved the performance of species distribution models compared to models that use environmental pO₂ alone. This modeling framework and data can support species distribution modeling for marine species by providing an alternative way to consider the role of shifting oxygen levels and temperatures on species ranges.