Climate change has strong effects on traits such as phenology and physiology. Studies typically assume that climate-induced trait changes will have consequences for population dynamics, but explicit tests are rare. Body condition reflects energy storage and may directly affect how much can be invested in reproduction and survival. However, the causal pathway by which decreased body condition impacts population dynamics has never been quantified across multiple populations and species. Therefore, we lack a general understanding of the consequences of changes in condition for variables more relevant for conservation, such as population size. Using structural equation modeling, we investigate how temperature-induced changes in body condition affect reproduction, and the subsequent impact on population growth rates of 19 bird species across 80 Dutch sites over a 21-year period. Warmer temperatures were associated with decreased body condition, which led to both decreased and increased reproduction at different sites, cancelling out any overall effect. The indirect effect of temperature on population growth (via body condition and reproduction) only explained within-species variation in the total effects of temperature on population growth. Instead, the direct effect of temperature on population growth (unrelated to condition and reproduction) was the most important pathway underlying the total effects of temperature on population growth, suggesting that unknown variables are mediating this effect. About half of the species are expected to increase under global warming, but this variation was not associated with any species characteristic. Overall, body condition responses to global warming are common, but their consequences on reproduction and subsequently population growth contribute relatively little to the total temperature impacts on population dynamics. Given that warming temperatures have strong effects on population dynamics, understanding the pathways via which temperature impacts population dynamics will be crucial for our ability to predict climate change effects in the future and improve conservation efforts.

Body mass, body size and reproduction data were collected as part of the Dutch Constant Effort Site (CES) program, which covers 80 sites across the Netherlands, spanning 21 years (1994-2014; see Appendix 1 Fig S1 for map). The CES-project follows a standardised protocol where birds are captured using mist nets from the 13^th April until the 13^th August every year (Robinson et al. 2009). Most sites were sampled 12 times per year (range 9-12). Captured birds are ringed and morphometric measurements taken, including body mass (grams) and wing length (maximum chord measurement; Svensson 1992). The sex and age-class (juvenile or adult) are classified based on plumage (wherever possible). For all species we consider juveniles to be less than one year old and born in the year of capture.

Independent annual population growth rates were calculated using abundance count data from the Dutch Breeding Bird Monitoring Program (BMP) which has been running since 1984. This bird counting methodology uses fixed study plot territory mapping, a bird counting methodology that produces reliable estimates of total bird abundance compared to point/transect counts, and makes year-to-year comparisons possible at the site level (Van Turnhout et al. 2010).  Study plots are visited 5-10 times per year, with every visit covering the whole plot.

We calculated the weighted mean abundance for each of the CES sites by averaging the 5 closest BMP sites that had the same habitat type (typically within a 15km radius of the CES sites).

Estimates in this data set are calculated per the methodology in the article. Please see the metadata tab for specific information.