Integrating large mammalian herbivores into Dynamic Global Vegetation Models: Status and next steps

Abstract Zoogeochemistry, the study of how animals influence elemental cycling within ecosystems, bridges the fields of zoology, ecology and biogeochemistry. Yet, most terrestrial ecosystem models, especially Dynamic Global Vegetation Models (DGVMs), remain largely plant‐centric and rarely capture animal‐driven feedbacks. This omission could lead to misrepresentation of carbon and nutrient cycling, which is particularly pertinent at a time of continued widespread defaunation but also animal introductions (trophic rewilding) gaining traction as a potential mitigation measure. Here, we review global evidence from field observations, experimental manipulations, remote sensing and modelling to understand how animals, influence carbon, nitrogen and phosphorus cycling through consumption (e.g. herbivory), transformation and redistribution (via faeces, urine and carcasses) and physical disturbances (e.g. trampling). While zoogeochemistry encompasses all animals, this review focuses primarily on terrestrial vertebrates, particularly large mammals. We then trace the historical development of how animals have been represented in contemporary DGVMs, beginning with implicit representations and progressing towards explicit representations. Persistent gaps remain in capturing key animal‐driven processes, and relating these to functional traits (body size, foraging strategy, etc.), mammal‐mediated biogeochemical cycling, animal–microbial interactions and in development of adequate model validation approaches. To address these challenges, we propose a modular framework guided by the principle of ‘complexity by necessity,’ which links trait‐based animal cohorts to above‐ and below‐ground processes through consumption, defecation and other animal‐driven inputs, capturing spatial behaviours, plant–microbe–soil interactions and predator–prey dynamics across scales. Integrating zoogeochemical mechanisms into DGVMs would replace the traditional black‐box approach to animal processes with explicit mechanistic modelling. This offers a process‐based tool to quantify local animal impacts, scale across large regions and rigorously evaluate the expected outcomes for ecosystem structure and biogeochemistry. In the future, we call for better implementation of animal interactions and habitat‐mediated feedbacks, such as behavioural effects through predation or fear (predator–prey interactions). To more accurately simulate the nutrient cycling in soil, we suggest expanding models to include interactions with soil‐engineering invertebrates and larger animals. Collectively, these improvements would advance our mechanistic understanding of animal‐driven biogeochemistry and their importance for predicting nutrient cycles and animal–plant–soil feedbacks within Earth system models. Read the free Plain Language Summary for this article on the Journal blog.