Factors shaping community assemblages and species co-occurrence of different trophic levels.

Valeria Trivellone,Diego Balseiro,Stephane Dray,Stephanie Bougeard,Simone Giavi,Patrik Krebs,Marco Moretti

doi:10.1002/ece3.3061

Abstract

Species assemblages are the results of various processes, including dispersion and habitat filtering. Disentangling the effects of these different processes is challenging for statistical analysis, especially when biotic interactions should be considered. In this study, we used plants (producers) and leafhoppers (phytophagous) as model organisms, and we investigated the relative importance of abiotic versus biotic factors that shape community assemblages, and we infer on their biotic interactions by applying three‐step statistical analysis. We applied a novel statistical analysis, that is, multiblock Redundancy Analysis (mbRA, step 1) and showed that 51.8% and 54.1% of the overall variation in plant and leafhopper assemblages are, respectively, explained by the two multiblock models. The most important blocks of variables to explain the variations in plant and leafhopper assemblages were local topography and biotic factors. Variation partitioning analysis (step 2) showed that pure abiotic filtering and pure biotic processes were relatively less important than their combinations, suggesting that biotic relationships are strongly structured by abiotic conditions. Pairwise co‐occurrence analysis (step 3) on generalist leafhoppers and the most common plants identified 40 segregated species pairs (mainly between plant species) and 16 aggregated pairs (mainly between leafhopper species). Pairwise analysis on specialist leafhoppers and potential host plants clearly revealed aggregated patterns. Plant segregation suggests heterogeneous resource availability and competitive interactions, while leafhopper aggregation suggests host feeding differentiation at the local level, different feeding microhabitats on host plants, and similar environmental requirements of the species. Using the novel mbRA, we disentangle for the first time the relative importance of more than five distinct groups of variables shaping local species communities. We highlighted the important role of abiotic processes mediated by bottom‐up effects of plants on leafhopper communities. Our results revealed that in‐field structure diversification and trophic interactions are the main factors causing the co‐occurrence patterns observed.

Highlights

At a given point in space and time, the composition of species assemblage is the result of at least two processes that have been concurrently brought to completion: dispersion and habitat filtering
Our data show that more than half of the variation across plant (51.8%) and leafhopper (54.1%) assemblages is mainly driven by habitat filtering processes, in particular topography and, to a lesser degree, by biotic factors mainly structured by habitat conditions
The contribution of pure abiotic factors was higher compared to pure biotic factors in both plants (9.6% vs. 4.9%) and leafhoppers (14.8% vs. 3.8%), respectively, indicating that abiotic filtering processes are relatively more important than biotic factors in shaping community assemblages at both trophic levels

Summary

Introduction

At a given point in space and time, the composition of species assemblage is the result of at least two processes that have been concurrently brought to completion: dispersion and habitat filtering. Asymmetrical canonical ordination methods are applied for modeling a response variable Y (i.e., species communities observed) using a set of explanatory variables assembled in a data matrix X (i.e., usually abiotic factors), providing the proportion of the variation of the response data matrix Y that is accounted for by the explanatory matrix X. One of the pivotal ecological challenges at the moment is to determine how to statistically include in the analyses the contribution of biotic interactions, especially given the fact that observed species interaction matrices (who interacts with whom) are lacking (Ovaskainen, Abrego, Halme, & Dunson, 2016; Wisz et al, 2013). Even though surrogates for biotic relationships or reduced matrices for species interactions could be included as predictors of community composition, the issue on how properly depict the different kind of asymmetrical and symmetrical interactions and coping with large biotic interaction matrices in statistical models is still poorly explored

Objectives

Findings

Discussion

Conclusion