A bi-variate framework to model microbiome resilience in healthy dogs

Abstract

IntroductionEcological resilience is the capacity of an ecosystem to maintain its state and recover from disturbances. This concept can be applied to the gut microbiome as a marker of health.MethodsSeveral metrics have been proposed to quantify microbiome resilience, based on the prior choice of some salient feature of the trajectories of microbiome change. We propose a data-driven approach based on compositional and functional data analysis to quantify microbiome resilience. We demonstrate the validity of our approach through applications to sled dogs undergoing three types of exercise: running on an exercise wheel, pulling an all-terrain vehicle, and pulling a sled.ResultsMicrobiota composition was clearly impacted by each exercise type. Log-ratio analysis was utilized for dimensionality reduction and identified 33 variables (taxa) explaining 90% of the variance. Functional principal component analysis identified two scores (FPCA 1 and FPCA2) which explained 76% and 19% of the variability of the trajectories, respectively. More resilient trajectories corresponded to low values of FPCA1 and FPCA2 values close to zero. Levels of chemokines MCP-1 and KC-like, which increased significantly after exercise and returned to pre-exercise levels within 24 h, were significantly associated with FPCA scores as well.DiscussionTo our knowledge, this is the first study proposing a principled approach to quantify microbiome resilience in healthy dogs and associate it with immune response to exercise-related stress.