The network structure of microbial cross-feeding impacts community diversity during antibiotic treatment

Antibiotics can reduce gut microbiome diversity, with negative health effects on the host, but predicting community diversity during antibiotic treatment is challenging due to the many mechanisms by which species interact. While both nutrient competition and cross-feeding can play major roles in microbiota assembly, the effect of antibiotics is often studied experimentally in monoculture, and theoretical studies have focused on communities that interact via nutrient competition alone. To investigate the interplay between nutrient competition, antibiotics, and cross-feeding, we introduce a consumer-resource model that includes all three interactions, and captures a wide range of cross-feeding network architectures with a single parameter. For three-species communities, we found that coexistence during narrow-spectrum antibiotic treatment was maximal with a cyclic cross-feeding network, while fully connected cross-feeding networks maximized coexistence during broad-spectrum antibiotic treatment. However, the effects of cyclic cross-feeding were particularly sensitive to community size and the number of targeted species (antibiotic spectrum); for communities with more than six species and resources, cyclic cross-feeding can be detrimental to coexistence due to instability. Our findings highlight the complex effects of cross-feeding network architecture on coexistence when growth inhibition reshapes the nutrient competition landscape, with potential applications to microbial communities in all natural environments, where they are often exposed to growth-inhibiting agents such as drugs, temperature, and pH modulation.