It is predicted that warmer conditions should lead to a loss of trophic levels, as larger bodied consumers, which occupy higher trophic levels, experience higher metabolic costs at high temperature. Yet, it is unclear whether this prediction is consistent with the effect of warming on the trophic structure of natural systems. Furthermore, effects of temperature at the species level, which arise through a change in species composition, may differ from those at the population level, which arise through a change in population structure. We investigate this by building species-level trophic networks, and size-structured trophic networks, as a proxy for population structure, for 18,648 stream fish communities, from 4,145,234 individual fish samples, across 7,024 stream locations in France from 1980 to 2008. We estimated effects of temperature on total trophic diversity (total number of nodes), vertical trophic diversity (mean and maximum trophic level), and distribution of biomass across trophic level (correlation between trophic level and biomass) in these networks. We found a positive effect of temperature on total trophic diversity in both species- and size-structured trophic networks. We found that maximum trophic level and biomass distribution decreased in species-level and size-structured trophic networks, but the mean trophic level decreased only in size-structured trophic networks. These results show that warmer temperatures associate with a lower vertical trophic diversity in size-structured networks, and a higher one in species-level networks. This suggests that vertical trophic diversity is shaped by antagonistic effects of temperature on population structure and on species composition. Our results hence demonstrate that effects of temperature do not only differ across trophic levels, but also across levels of biological organisation, from population to species level, implying complex changes in network structure and functioning with warming.

This dataset contains 4,145,234 individual fish samples (e.g. body length), from 50 species of fish, across 7,024 stream locations in France from 1980 to 2008. This data was collected by ONEMA via electric fishing.  In addition, we built species-level trophic networks, and size-structured trophic networks, as a proxy for population structure, for 18,648 stream fish communities, based on body-size dependent prey-predator interactions and specific information regarding the diet of fish extracted from the literature. All scripts and data set necessary to build the networks are provided.

It should be noted that the scripts were not intended initially for use by a third party and that adjustments may be necessary, such as changing the root directory at the beginning of the scripts.

More information in the README.txt file.