Spiders are dominant predators in terrestrial ecosystems and feed on prey from the herbivore and detritivore subsystem (dual subsystem omnivory) as well as on other predators (intraguild predation). Little is known about how global change potentially affects the importance of different prey groups in predator diets. In this meta-analysis we identify the impact of climatic conditions, land-use types and functional traits of spider species on the relative importance of Hemiptera, Araneae and Collembola prey in spider diets. We use a dataset including 78 publications with 149 observational records of the diet composition of 96 spider species in agricultural and non-agricultural habitats in 24 countries worldwide. The importance of Hemiptera prey was not affected by climatic conditions and was particularily high in smaller spider species in agricultural habitats. Araneae prey was most important for actively hunting, larger spider species in non-agricultural habitats. Collembola prey was most important for small, actively hunting spider species in regions with higher temperature seasonality. Spider species with a higher importance of Araneae prey for their diet also had higher importances of Collembola and lower importances of Hemiptera prey. Future increases of temperature seasonality predicted for several regions worldwide may go along with an increasing importance of Collembola prey which also related to a higher importance of intraguild prey here. Two global change drivers predicted for many regions of the world (increasing climatic seasonality and ongoing conversion of non-agricultural to agricultural land) both hold the potential to increase the importance of Collembola prey in spider diets. The importance of Hemiptera and Araneae prey may however show contrasting responses to these two drivers. These complex potential effects of global change components and their impact on functional traits in spider communities highlight the importance to simultaneously consider multiple drivers of global change to better understand future predator-prey interactions.

This study is based on a global database about the diet composition of hunting and web-building spider species in natural ecosystems used in Birkhofer and Wolters (2012) with the addition of data from agricultural ecsoystems and updates from Diehl et al. (2013b), Michalko & Pekár (2015a), Arvidsson et al. (2020) and Mezőfi et al. (2020). All data in the original publications are derived by direct visual records of prey or prey remains in spider species in field studies, not including data from molecular or experimental studies. Note that only subsets of data from the original database from non-agricultural (82 cursorial and web-building spider species in natural habitats: Birkhofer & Wolters 2012) or only for web-building spiders (63 spider species in agricultural, natural and forest habitats: Birkhofer et al. 2018) were previously published. The database includes 118 unique publications that reported 310 datasets about diet compositions in individual spider species worldwide. All datasets that were based on fewer than 20 recorded prey items per spider species in individual studies or did address spider species in forest habitats were excluded for this study. The selection of a minimum of 20 records was based on the fact that spiders in each study could theoretically reach the maximum diet breadth, including prey from all 20 prey orders that were originally recorded across datasets (see Birkhofer & Wolters 2012). The selection of non-forest habitats was based on the aim to compare habitats that share a major structural characteristic by not being dominated by dense, natural tree cover. Forests further have a very different invertebrate community compared to grasslands and arable fields (e.g. Birkhofer et al. 2015), which would limit a comparison of diets between major habitat types. The remaining 78 publications provided 149 datasets on the diet composition of 96 spider species worldwide (Figure 1). This database was used to calculate the relative contribution of each prey order to the overall diet in each dataset as percentage value. The percentages of Hemiptera, Collembola and Araneae prey were then extracted to reflect the relative contribution of these prey orders to the diet of individual spider species.

No missing values