Comparative gut microbiome research through the lens of ecology: theoretical considerations and best practices

Abstract

Comparative approaches in animal gut microbiome research have revealed patterns of phylosymbiosis, dietary and physiological convergences, and environment-host interactions. However, most large-scale comparative studies, especially those that are highly cited, have focused on mammals, and efforts to integrate comparative approaches with existing ecological frameworks are lacking. While mammals serve as useful model organisms, developing generalised principles of how animal gut microbiomes are shaped and how these microbiomes interact bidirectionally with host ecology and evolution requires a more complete sampling of the animal kingdom. Here, we provide an overview of what past comparative studies have taught us about the gut microbiome, and how community ecology theory may help resolve certain contradictions in comparative gut microbiome research. We explore whether certain hypotheses are supported across clades, and how the disproportionate focus on mammals has introduced potential bias into gut microbiome theory. We then introduce a methodological solution by which public gut microbiome data of understudied hosts can be compiled and analysed in a comparative context. Our aggregation and analysis of 179 studies shows that generating data sets with rich host diversity is possible with public data and that key gut microbes associated with mammals are widespread across the animal kingdom. We also show the effects that sample size and taxonomic rank have on comparative gut microbiome studies and that results of multivariate analyses can vary significantly with these two parameters. While challenges remain in developing a universal model of the animal gut microbiome, we show that existing ecological frameworks can help bring us one step closer to integrating the gut microbiome into animal ecology and evolution.

Keywords: comparative; ecology; evolution; gut microbiome; host–microbe interactions.

Fig. 1

Number of gut microbiome studies across different host clades, and whether a study was comparative or not. Studies were collected in September 2023 using the Scopus document search tool. We counted only published journal articles and excluded all other article types (including reviews). The top row includes studies with human subjects and laboratory animals (w/ humans & lab animals) while the bottom row does not include these studies (w/o humans or lab animals). Both 16S and metagenomic studies are included.

Fig. 2

An ecosystem selection ‘ball and cup’ model of the animal gut microbiome that accounts for the multivariate shaping of gut microbiome composition across different host clades. A narrow resilience curve (black line on top left) results in a selective ecosystem due to strong host selective pressures such as immune filtering (blue arrow), whereas a wider curve (top right) results in a less‐selective ecosystem due to weaker host selective pressures and stronger microbial forces. The red arrow denotes stochastic microbial forces such as dispersal, genetic drift, and immigration. Each shaded colour under the resilience curve denotes a host factor that contributes to the width of the curve. We provide four example curves with varying widths across four groups of hosts. The width of the resilience curves is a proxy for the strength of phylosymbiosis within a given clade of hosts. The model predicts that tighter curves limit microbial forces due to highly selective host forces while wider curves allow the opposite to occur due to less‐selective host forces.

Fig. 3

Results of Adonis analysis of the effects of sample size per host species on the calculated influence of taxonomic rank on animal gut microbiomes. Samples from each host were randomly subsampled at 999 permutations; the R2 value plotted is the mean from all permutations. Error bars are too small to show on the graph but were all <0.001. Colours denote host taxonomic ranks which were treated as separate factors. Because Study and Species are nearly identical (most studies had a unique host species), study, as a factor, is included to explore the batch effect after taking taxonomic rank into account. We rarefied the data to 1000 reads and used unweighted UniFrac distance matrices as inputs for the Adonis test.

Fig. 4

An annotated phylogeny of the presence of Akkermansia muciniphila and Faecalibacterium prausnitzii in the gut microbiomes of hosts from 179 compiled studies. Orange and blue colours at the tips denote the presence of these gut microbes. Each tree tip represents a host species included in our database.