Macroecological distributions of gene variants highlight the functional organization of soil microbial systems

The recent application of macroecological tools and concepts has made it possible to identify consistent patterns in the distribution of microbial biodiversity, which greatly improved our understanding of the microbial world at large scales. However, the distribution of microbial functions remains largely uncharted from the macroecological point of view. Here, we used macroecological models to examine how the genes encoding the functional capabilities of microorganisms are distributed within and across soil systems. Models built using functional gene array data from 818 soil microbial communities showed that the occupancy-frequency distributions of genes were bimodal in every studied site, and that their rank-abundance distributions were best described by a lognormal model. In addition, the relationships between gene occupancy and abundance were positive in all sites. This allowed us to identify genes with high abundance and ubiquitous distribution (core) and genes with low abundance and limited spatial distribution (satellites), and to show that they encode different sets of microbial traits. Common genes encode microbial traits related to the main biogeochemical cycles (C, N, P and S) while rare genes encode traits related to adaptation to environmental stresses, such as nutrient limitation, resistance to heavy metals and degradation of xenobiotics. Overall, this study characterized for the first time the distribution of microbial functional genes within soil systems, and highlight the interest of macroecological models for understanding the functional organization of microbial systems across spatial scales.

Local and Geographic Factors Shape the Occupancy-Frequency Distribution of Freshwater Bacteria

Species prevalence across the landscape is related to their local abundance, which is a result of deterministic and stochastic processes that select organisms capable of recolonizing sites where they were once extinct, a process known as the rescue effect. The occupancy-frequency distribution (OFD) describes these patterns and has been extensively used to understand organism's distribution but has been poorly tested on microorganisms. In order to test OFD on freshwater bacteria, we collected data from 60 shallow lakes distributed across a wide area in southeastern Brazil, to determine the bacterial operational taxonomic units (OTUs) that were present in all sites (core) and at only one site (satellite). Then, we analyzed the spatial abundance distributions of individual OTUs to understand the influence of local abundances on regional occupancy patterns. Finally, we tested the environmental factors that influenced occupancy and abundance. We found a significant bimodal OFD for freshwater bacteria using both OTUs (97% clustering) and amplicon sequence variants (ASVs, unique sequences), with 13 core OTUs and 1169 satellite OTUs, but only three core ASVs. Core organisms had a bimodal or gamma abundance distribution. The main driver of the core community was pH, while nutrients were key when the core community was excluded and the rest of the community (mild and satellite taxa) was considered. This study demonstrates the close relationship between local environmental conditions and the abundance and dispersion of microorganisms, which shapes their distribution across the landscape.

A self-similarity model for the occupancy frequency distribution

The shapes of interspecific range-size distributions at scales finer than the geographic range are highly variable. However, no numerical model has been developed as a basis for understanding this variation. Using self-similarity conditions, we present an occupancy probability transition (OPT) model to investigate the effect of sampling scale (i.e. sample grain) and species saturation (strongly positively correlated with the fractal dimension) on the shape of occupancy frequency distributions (fine scale expression of range-size distributions). In accordance with empirical observations, the model showed that core-modes are likely to be rare in occupancy frequency distributions. The modal occupancy shifted from core to satellite with an increase in sample grain (from coarse scale to fine scale) at a linear rate after log-transformation of occupancy. Saturation coefficients above a particular threshold generated multimodality. Bimodal distributions arose from a combination of different occupancy probability distributions (OPDs), with species-specific saturation coefficients generating occupancy frequency distributions of the shape commonly observed empirically, i.e. bimodal with a dominant satellite mode. This is a consequence of the statistical properties of the OPD, and is also largely insensitive to species richness. The OPT model thus provides a null model for the shape of occupancy frequency distributions. Furthermore, it demonstrates that the sample grain of a study, sampling adequacy (based on a linear sampling assumption) and the distribution of species saturation coefficients in a community are together largely able to explain the patterns observed in empirical occupancy frequency distributions.

Occupancy frequency distributions: patterns, artefacts and mechanisms

Numerous hypotheses have been proposed to explain the shape of occupancy frequency distributions (distributions of the numbers of species occupying different numbers of areas). Artefactual effects include sampling characteristics, whereas biological mechanisms include organismal, niche-based and meta-population models. To date, there has been little testing of these models. In addition, although empirically derived occupancy distributions encompass an array of taxa and spatial scales, comparisons between them are often not possible because of differences in sampling protocol and method of construction. In this paper, the effects of sampling protocol (grain, sample number, extent, sampling coverage and intensity) on the shape of occupancy distributions are examined, and approaches for minimising artefactual effects recommended. Evidence for proposed biological determinants of the shape of occupancy distributions is then examined. Good support exists for some mechanisms (habitat and environmental heterogeneity), little for others (dispersal ability), while some hypotheses remain untested (landscape productivity, position in geographic range, range size frequency distributions), or are unlikely to be useful explanations for the shape of occupancy distributions 'species specificity and adaptation to habitat, extinction-colonization dynamics). The presence of a core (class containing species with the highest occupancy) mode in occupancy distributions is most likely to be associated with larger sample units, and small homogenous sampling areas positioned well within and towards the range centers of a sufficient proportion of the species in the assemblage. Satellite (class with species with the lowest occupancy) modes are associated with sampling large, heterogeneous areas that incorporate a large proportion of the assemblage range. However, satellite modes commonly also occur in the presence of a core mode, and rare species effects are likely to contribute to the presence of a satellite mode at most sampling scales. In most proposed hypotheses, spatial scale is an important determinant of the shape of the observed occupancy distribution. Because the attributes of the mechanisms associated with these hypotheses change with spatial scale, their predictions for the shape of occupancy distributions also change. To understand occupancy distributions and the mechanisms underlying them, a synthesis of pattern documentation and model testing across scales is thus needed. The development of null models, comparisons of occupancy distributions across spatial scales and taxa, documentation of the movement of individual species between occupancy classes with changes in spatial scale, as well as further testing of biological mechanisms are all necessary for an improved understanding of the distribution of species and assemblages within their geographic ranges.