1 | Introduction
Research on the microbiome ecology of wild animals, where ecological and evolutionary processes free from human intervention shape diversity, is key to understanding the form and causes of observed microbiome variation in nature. Exploration of new systems necessarily begins with characterizing the form and putative causes of microbiome variation observed among hosts, host populations, or host species. Following methods of clinical research, aspects of host physiology (Amato et al., 2014; Stothart, Palme, & Newman, 2019; Suzuki et al., 2019) and diet (Kartzinel, Hsing, Musili, Brown, & Pringle, 2019; Teyssier et al., 2020) remain heavily emphasized to the neglect of the fuller breadth of ecological processes which shape community variation (selection, ecological drift, and dispersal; Vellend et al., 2014). Niche-based processes (i.e., selection) are sometimes assumed by default, with stochastic processes such as ecological drift (chance demographic change in a population) more rarely taken into account (Adair & Douglas, 2017; Zhou & Ning, 2017). Similarly, microbiota dispersal (e.g. spatial, temporal, or social isolation)—while generally ignored beyond co-housing considerations in clinical research—is a natural process integral to microbial ecology within free-living systems (Greyson-Gaito et al., 2020). Yet, studies of wildlife microbiomes also overlook dispersal in favour of putative mechanisms related to host diet and physiology.
While host physiology and diet clearly shape wildlife-microbiome variation, the ability of microbiota to disperse between host intestinal tracts arguably supersede the importance of either in governing microbiome diversity (Miller, Svanbäck, & Bohannan, 2018). Correlations between microbiome composition and social networks in gregarious hosts illustrate the importance of microbial community connectivity and bacterial dispersal on microbiome diversity (savannah baboons [Papio cynocephalus] , Tung et al., 2015; Sarkar et al., 2020). Bacterial dispersal between hosts can occur through grooming (rhesus macaques [Macaca mulatta ], Balasubramaniam et al., 2018), coprophagy (Brandt’s vole [Lasiopodomys brandtii ], Bo et al., 2020), shared environment (humans [Homo sapiens ], Rothschild et al., 2018), or copulation (black legged kittiwakes [Rissa tridactyla ], White et al., 2010). Therefore, we would expect rates of bacterial dispersal to decrease as a function of the time and space separating hosts. An effect of spatial separation has been demonstrated at large spatial scales between (sub)populations of red squirrels (Tamiasciurus hudsonicus ; ~7km; Ren et al., 2017), bighorn sheep (~150 km; Couch et al., 2020), house mice (Mus musculus ; ~1100 km; Linnenbrink et al., 2013), American pika (Ochotona princeps ; ~1400 km; Kohl, Varner, Wilkening, & Dearing, 2018), red colobus (Procolobus rufomitratus ; ~1100km; Mccord et al., 2014), and between pairs of predator and prey species (~12100 km; Moeller et al., 2017).
The affects of spatial separation on microbial dispersal between social groups of host individuals within populations are more rarely considered. One study of a single focal population of house mice found a greater importance of fine-scale habitat heterogeneity than spatial separation (Goertz et al., 2019). Conversely, spatial structuring of the microbiome has been reported among a contiguous moose population spanning 150 km (Fountain-Jones et al., 2020). Similar effects of spatial proximity have been observed among semi-feral ponies (40 km2; Antwis, Lea, Unwin, & Shultz, 2018), but were limited to comparisons between three large social groups (bands). Regardless of the spatial scale considered, many studies do not control for local environmental variation, which can be problematic given an expectation of spatial autocorrelation in environmental conditions. Conversely, studies which consider environmental terms often do not consider spatial processes.
Although spatial proximity and social networks can facilitate dispersal between hosts and can thereby drive microbiome similarity, strong dispersal limitation can cause greater than expected divergence between communities and unpredictable β-diversity patterns. In a meta-population context, dispersal between communities are thought to stabilize populations (Crowley, 1981), so long as dispersal is not so high as to drive spatial synchrony (Yaari, Ben-Zion, Shnerb, & Vasseur, 2012). Conversely, dispersal limitation among isolated biological communities increases the strength of ecological drift and heightens the risk of local extinctions (Vellend, 2010). Hosts disconnected from the broader meta-community of conspecific microbiomes (Miller et al., 2018)—those in low density populations, at the fringes of populations, or experiencing social isolation—may be at greater risk of stochastic microbiome dysregulation. Evidence for dispersal limitation induced dysregulation of the microbiome has been demonstrated through experimental prevention of coprophagy in Brandt’s voles (Bo et al., 2020). While similar concerns have been raised with respect to wildlife in captivity (McKenzie et al., 2017; Trevelline, Fontaine, Hartup, & Kohl, 2019), this effect remains to be explicitly tested in free-living settings.
Dispersal limitation can feed ecological drift but so too can dietary and physiological factors which are often assumed to be deterministic. For example, different diets can exert divergent selective pressures, but can also differ in the energy made accessible to the microbiome and the diversity of metabolic niche space supported. Labile high energy diets may fail to support fibrolytic and cellulolytic niche-space in the microbiome (Oliphant & Allen-Vercoe, 2019) and can destabilize microbial communities in a process similar to the paradox of enrichment (Coyte, Schluter, & Foster, 2015; Rosenzweig, 1971). Similarly, while different host physiological states might select for different microbial functions (Foster, Schluter, Coyte, & Rakoff-Nahoum, 2017), a loss of host homeostatic control among physiologically stressed hosts might result in community instability and greater stochastic variation (Zaneveld, McMinds, & Vega Thurber, 2017). Microbiome β-dispersion, a measure of microbiome variance, is one indication of the relative strength of stochasticity. A second indication of stochasticity is the failure of communities to deviate from the stochastic expectations as predicted by null modelling approaches. Despite past misuse (for overview see: Narwani, Matthews, Fox, & Venail, 2015), phylogenetic null modelling methods are valuable to consider alongside conventional β-diversity metrics, as traditional diversity metrics can be influenced by system gamma diversity and imbalances in alpha diversity between communities (Chase, Kraft, Smith, Vellend, & Inouye, 2011; Gering & Crist, 2002; Zhou & Ning, 2017).
Here we directly contrast the ability of host physiology, habitat heterogeneity, and spatial measures to explain variation in the faecal bacterial microbiome of feral horses using 86 adult females from the closed population of Sable Island (Nova Scotia, Canada). Building on a comprehensive, long-term, detailed individual-based study of ecology and evolution for this population (Richard, Simpson, Medill, & Mcloughlin, 2014), we apply a combination of conventional diversity analyses and null modeling approaches to evaluate the evidence for drift, dispersal, and niche-based processes. If environmental conditions and host physiology—sources of selective pressures which shape the microbiome—are more similar among populations than between populations or between species, then we would predict microbial dispersal patterns and ecological drift to play a comparably large role in shaping inter-individual microbiome variation within populations. Specifically, we predict that phylogeny-independent diversity measures will be most strongly influenced by spatial and social variables, reflecting microbial dispersal patterns. Conversely, we predicted that host physiology and local habitat heterogeneity would better explain variation in phylogeny-weighted metrics of microbiome diversity, reflecting different selective pressures imposed on host-associated microbiota between host physiological states or diets. Our study represents one of the first direct comparisons between environmental and spatial effects on host-associated microbiomes in the wild at a within population scale, with consideration offered to alternative ecological processes.