Temperature rise may alter consumer diets through increased metabolic demand and altered resource availability. However, current theories assessing dietary shifts with warming do not account for change in resource availability. It is unknown whether consumers will increase consumption rates or consume different resources to meet increased energy requirements and whether dietary change will lead to associated variation in morphology and nutrient utilisation. Here, we used populations of Gambusia affinis across parallel thermal gradients in New Zealand and California to understand the influence of temperature on diets, morphology, and stoichiometric phenotypes. Our results show that with increasing temperature in New Zealand, mosquitofish consumed more plant material, whereas, in California, mosquitofish shifted towards increased consumption of invertebrate prey. In both regions, populations with plant-based diets had fuller guts, longer gut length to body length ratios, more dorsally orientated mouths, and reduced body elemental %C and N/P. Together, our results show multiple pathways by which consumers may alter their feeding patterns with rising temperatures, and they suggest that warming-induced changes to resource availability may be the principal determinant of which pathway is taken.

We collected Gambusia affinis from sites along parallel temperature gradients in NZ and CA using a seine net. In CA, we sampled between the 30^th of May and the 1^st of June 2016, and in NZ, we sampled from the 8^th of February to the 21^st of February 2017. All fish were euthanised onsite and frozen immediately. From each population 40 fish were photographed for morphological triats, dissected for gut traits, and 20 fish from each population were dried and ground for elemental composition. At each site we took samples of macroinverebrates and zooplankton. Gut contents were summarized as relative importance for analyses, and are presented as proportional occurrence of the major dietary categories in the fishes diet in the data file. 

• Where gut length data are not available (due to the absence of a photograph of the dissected gut) these cells have been intentionally left blank.
• Gut content and fullness values of '0' indicate an empty gut.