Future of Food Webs: The Role of Biotic Interactions in Predicting the Impact of Climate and Land Use Change

The world we share with millions of species is changing rapidly and dramatically. Climate change, pollution, overexploitation, and the ongoing modification of natural habitats pose serious challenges for all species. Ecologists have always desired to predict the future of ecosystems and species. However, only recently have advanced modeling techniques and increased ecological data made it possible to use complex models to predict what will happen to communities over the next century.

Species distribution models (SDMs) attempt to predict where a species is likely to occur based on its relationship with environmental conditions. They can be used to predict how species' ranges will shift by considering different climate and land use change scenarios. However, species do not live independently of each other. A complex web of interactions links them, with direct and indirect effects spreading from one to another. The changing environment affects not only the species themselves but also the relationships between them (Van der Putten et al. 2010).

Food web models are designed to investigate the structure and functioning of communities by focusing on one of the most important relationships: trophic interactions. The strength of these models is that they consider not only direct interactions between species but also indirect effects crossing multiple species. They are suitable for quantifying ecological processes such as productivity (Wang and Brose 2018) or the impacts of invasion and disturbances (e.g., Woodward and Hildrew 2001). Furthermore, they can be used to investigate important conservation issues, such as finding keystone species (Jordán et al. 2006).

Previous studies have shown that climate change can strongly affect food webs. Differences in the species' thermal vulnerability and dispersal ability may lead to trophic mismatches (Thakur 2020) and the simplification and rewiring of food webs (Bartley et al. 2019). Land-use intensification may have similar effects, favouring particular trophic groups and inducing structural changes in food webs (Botella et al. 2024). An important question is, therefore, how these rearrangements will affect the structural properties of food webs and whether there are spatial patterns in which networks will be more sensitive to future environmental changes. Hao et al. (2025) sought to answer this question in their ambitious research.

Large amounts of reliable ecological data are needed to compare food webs over time and space. However, collecting all the information about who eats whom in a community is difficult. Monitoring or measuring dietary preferences is time-consuming and sometimes expensive. One possible solution is to infer interactions based on the functional traits of the possibly interacting species. Eklöf et al. (2013) showed that knowledge of a few relevant functional traits may be sufficient to predict all trophic interactions of a community. This approach can help not only to fill knowledge gaps when building food web models but also to predict the likelihood of interactions between species that have not yet met but may have overlapping ranges in the future.

Hao et al. (2025) combined this trait-based interaction inference with species distribution modelling to compare the current and potential future structures of terrestrial vertebrate food webs worldwide. To do this, they collected all available data from databases and articles on tetrapods' dietary preferences and other traits, and created an empirical meta-food web. Then, based on this network and the trait database of the species, machine learning was used to predict a potential meta-web. This latter network included all possible trophic interactions that could be realised between species when their range and habitat preferences overlap. In addition, species distribution models were used to predict the current and future ranges of the species, accounting for different climate change scenarios, land-use change, habitat and elevation preferences of the species, and group-level dispersal abilities. The distribution data were then integrated with the potential metaweb to predict and compare the structure of local food webs.

Although the models focused solely on terrestrial vertebrates, representing a small subset of the actual food web, they successfully reproduced known spatial patterns in the structural indicators of the food webs. However, the future they projected is not very promising: without considering dispersion, the model simulated a significant reduction in species number and network connections leading to a decrease in complexity and an increase in connectivity. These changes are most pronounced in tropical (and Arctic) areas, particularly affecting the basal (mainly amphibian and reptile) species. While maximum species dispersal mitigated or even reversed these changes in some areas the networks still exhibited considerable variance in their response to environmental changes, part of which can be well explained by spatial (latitudinal or topographic) arrangement. This result highlights why these large-scale comparative studies are important. The results are in line with those of previous, smaller-scale studies that highlight the importance of dispersal in climate change adaptation and the necessity of considering land-use change alongside climate change in similar research.

Each additional parameter in the models can help to make the results more nuanced, but also increases the amount of data required and the uncertainty involved. Consequently, this work, like other global-scale studies, is simplistic in many respects and has limitations. A major limitation of such a complex model can be the quantity and quality of data. Although the quantity of ecological and other functional trait data is increasing rapidly, there are still many knowledge gaps. The traits of invertebrates or certain groups' feeding strategy and dispersal ability are still poorly documented. In particular, data from areas outside Europe and North America are under-represented (Cameron et al. 2019). Hao et al. have skilfully navigated data-poor fields, highlighting important correlations, yet there remains ample scope for further research. One important future direction is extending their food web models by including invertebrates, plants, and the below-ground (brown) food web. Although this would require a lot of targeted data collection, it would be interesting to see if these more complex and realistic food webs are less sensitive to expected climate and land use changes. Similar modelling of freshwater food webs could be equally insightful. Furthermore, deeper analysis of the local food webs from the current research and comparison with known food webs could validate the results and enhance the understanding of how these systems work.

In addition to abiotic factors, consideration of biotic interactions is essential for more accurate predictions of future distributions and community structures. Thus, research such as the work of Hao et al. significantly contributes to understanding how vulnerable communities are and how they will respond to the grand challenges of future environmental change. This knowledge is essential to developing appropriate conservation strategies to mitigate the expected negative impacts.

Anett Endrédi: conceptualization, writing – original draft, writing – review and editing.

The author declares no conflicts of interest.

This article is a Invited Commentary on Hao et al., https://doi.org/10.1111/gcb.70061.