Communities in high definition: Spatial and environmental factors shape the micro‐distribution of aquatic invertebrates

Abstract

According to metacommunity theories, the structure of natural communities is the result of both environmental filtering and spatial processes, with their relative importance depending on factors including local habitat characteristics, functional features of organisms, and the spatial scale considered. However, few studies have explored environmental and spatial processes in riverine systems at local scales, explicitly incorporating spatial coordinates into multi‐taxa distribution models. To address this gap, we conducted a small‐scale study to discriminate between abiotic and biotic factors affecting the distribution of aquatic macroinvertebrates, applying metacommunity concepts. We studied a mountain section in each of three perennial streams within the Po River Basin (northern Italy). We sampled macroinvertebrates both in summer and winter, using specific in situ 50‐point random sampling grids. Environmental factors, including benthic organic matter (BOM), flow velocity, water depth, and substrate were recorded together with spatial coordinates for each sampling point. The relationships between community metrics (taxon richness, abundance, biomass, biomass–abundance ratio, and functional feeding groups) and explanatory variables (environmental and spatial) were assessed using generalised additive models. The influence of the explanatory variables on community structure was analysed with joint species distribution models. Environmental variables—primarily BOM—were the main drivers affecting community metrics, whereas the effects of spatial variables varied among metrics, streams, and seasons. During summer, community structure was strongly affected by BOM and spatial position within the riverbed, the latter probably being a proxy for mass effects mediated by biotic and stochastic processes. In contrast, community structure was mainly shaped by hydraulic variables in winter. Using macroinvertebrate communities as a model group, our results demonstrate that metacommunity concepts can explain small‐scale variability in community structure. We found that both environmental filtering and biotic processes shape local communities, with the strength of these drivers depending on the season. These insights provide baseline knowledge that informs our understanding of ecological responses to environmental variability in contexts including restoration ecology, habitat suitability modelling, and biomonitoring.

Alpha and beta diversity and species co‐occurrence patterns in headwaters supporting rare intermittent‐stream specialists

“…(2020) and Burgazzi et al. (2020), we sampled the posterior distributions of model parameters using four MCMC chains each run for 150,000 iterations. The first 50,000 iterations of each chain were used as a conservative burn‐in period and discarded, and the remaining 100,000 were thinned by 100, producing 4,000 posterior samples (1,000 samples per chain).…”

“…We fitted an HMSC model to the most common taxa in our dataset, with AFD and DPS included as environmental covariates and sample site as a spatially structured random effect, implemented by means of spatial latent factors (Ovaskainen et al, 2016). To avoid convergence problems associated with zero-inflation, and to ensure that the HMSC procedure was computationally feasible, we only included a subset of taxa in the analysis (Burgazzi et al, 2020;Elo et al, 2021), retaining those present in at least 10% of samples and accounting for at least 0.1% of total abundance (n = 57). We incorporated two dispersal traits (Sarremejane, Cid, et al, 2020), niches was analysed using Trait r 2 values (Tikhonov et al, 2020).…”

“…We fitted a probit model to presence-absence data as this gave significantly better Markov chain Monte Carlo (MCMC) convergence than a Poisson model fitted to the full count data. The number of spatial latent variables used to prescribe the structure of the random effect (sample site) was constrained to between five and 10 to avoid overfitting(Tikhonov et al, 2020).FollowingTikhonov et al (2020) andBurgazzi et al (2020), we sampled the posterior distributions of model parameters using four MCMC chains each run for 150,000 iterations. The first 50,000 iterations of each chain were used as a conservative burn-in period and discarded, and the remaining 100,000 were thinned by 100, producing 4,000 posterior samples (1,000 samples per chain).…”

Replicate DNA metabarcoding can discriminate seasonal and spatial abundance shifts in river macroinvertebrate assemblages

“…As samples were all taken from the same 100‐m stretch of river, they were considered as one interacting community. Conversely, the distribution of taxa within the stream bed is known to be highly uneven at a microhabitat scale, with many taxa clustered according to changes in flow, substrate and availability of coarse organic matter (e.g., Burgazzi et al, 2020; Mathers et al, 2017). As a result, not all taxa were expected to be present for detection in every 0.8‐m 2 sample.…”

Joint species distribution models unveil co‐occurrences between freshwater mussels and their fish hosts

“…However, how to disentangle the effects of biotic interactions from unmeasured environmental variables remains unclear and requires empirical tests using species with known interactions (Zurell et al, 2018). Furthermore, JSDMs have rarely been applied to freshwater communities (e.g., Burgazzi et al, 2020; Perrin et al, 2022; Wagner et al, 2019), and, to the authors' best knowledge, only once on freshwater mussels (Inoue et al, 2017).…”