Biotic interactions drive multitrophic species community assembly. Yet, explicitly incorporating this process in species distribution models (SDMs) is particularly challenging, even when biotic interactions are known. Here, we propose a framework that combines knowledge of trophic interactions with Bayesian structural equation models that model each species as a function of its prey or predators and environmental conditions. We tested and validated our framework on realistic simulated communities from different theoretical models. We showed that our framework improves the inference of both species’ fundamental and realized niches compared to classical SDMs (mean performances increased by 11% and 4% respectively), especially for species with strong biotic control, thus increasing model predictive performance (up to 99% and 70% improvement, respectively). Our framework can easily integrate various SDM extensions (e.g., occupancy models) and algorithms, and stands out as a novel solution for modeling multitrophic community distributions when trophic interactions are known or assumed.