The network structure of cross-feeding impacts microbial community diversity under growth-inhibiting stresses

Perturbations such as dietary shifts and drug treatment can reduce gut microbiome diversity, with negative consequences for host health, yet predicting diversity responses remains challenging because microbial species interact through multiple mechanisms. While nutrient competition and cross-feeding both influence microbiota assembly, environmental stresses such as antibiotics are typically studied experimentally in monoculture, and most theoretical frameworks consider nutrient competition alone. To investigate how these processes jointly shape community structure, we develop a consumer-resource model that incorporates nutrient competition, growth-inhibiting stress, and metabolite cross-feeding with a unified framework spanning varied cross-feeding architectures. For three-species communities, coexistence during narrow-spectrum growth inhibition is maximized by cyclic cross-feeding networks, whereas fully connected cross-feeding networks maximize coexistence during broad-spectrum growth inhibition. However, the benefits of cyclic cross-feeding depend strongly on community size and stress targeting: in communities with more than six species and six resources, cyclic networks can destabilize coexistence. These results are robust to inefficient leakage, dead-end metabolites, and embedding in larger communities, and large communities generalize to random leakage architectures in which connectivity determines the response to stress. Together, this framework shows that cross-feeding network architecture can fundamentally reshape how microbial communities respond to growth-inhibiting stresses. Here, the authors develop a consumer-resource model to investigate how cross-feeding shapes microbial diversity and show that cyclic networks support coexistence under narrow-spectrum stress but destabilize larger communities, while fully connected networks favor coexistence under broad-spectrum stress. Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give... [797 chars]