Co-occurrence patterns and the large-scale spatial structure of benthic communities in seagrass meadows and bare sand

Casper Kraan,Carsten F Dormann,Simon F Thrush

doi:10.1186/s12898-020-00308-4

Abstract

BackgroundSpecies distribution models are commonly used tools to describe diversity patterns and support conservation measures. There is a wide range of approaches to developing SDMs, each highlighting different characteristics of both the data and the ecology of the species or assemblages represented by the data. Yet, signals of species co-occurrences in community data are usually ignored, due to the assumption that such structuring roles of species co-occurrences are limited to small spatial scales and require experimental studies to be detected. Here, our aim is to explore associations among marine sandy-bottom sediment inhabitants and test for the structuring effect of seagrass on co-occurrences among these species across a New Zealand intertidal sandflat, using a joint species distribution model (JSDM).ResultsWe ran a JSDM on a total of 27 macrobenthic species co-occurring in 300,000 m2 of sandflat. These species represented all major taxonomic groups, i.e. polychaetes, bivalves and crustaceans, collected in 400 sampling locations. A number of significant co-occurrences due to shared habitat preferences were present in vegetated areas, where negative and positive correlations were approximately equally common. A few species, among them the gastropods Cominella glandiformis and Notoacmea scapha, co-occurred randomly with other seagrass benthic inhabitants. Residual correlations were less apparent and mostly positive. In bare sand flats shared habitat preferences resulted in many significant co-occurrences of benthic species. Moreover, many negative and positive residual patterns between benthic species remained after accounting for habitat preferences. Some species occurring in both habitats showed similarities in their correlations, such as the polychaete Aglaophamus macroura, which shared habitat preferences with many other benthic species in both habitats, yet no residual correlations remained in either habitat.ConclusionsFirstly, analyses based on a latent variable approach to joint distributions stressed the structuring role of species co-occurrences beyond experimental scales. Secondly, results showed context dependent interactions, highlighted by species having more interconnected networks in New Zealand bare sediment sandflats than in seagrass meadows. These findings stress the critical importance of natural history to modelling, as well as incorporating ecological reality in SDMs.

Highlights

Species distribution models are commonly used tools to describe diversity patterns and support conservation measures
We ran a joint species distribution analysis on a total of 27 species co-occurring in 300,000 m2 of New Zealand intertidal sandflat, representing all major taxonomic groups, i.e. polychaetes, bivalves and crustaceans
Eleven species only occurred in seagrass meadows, 6 species were restricted to bare sand flats, whereas 10 species were common (> 25% of sampling locations) to both habitats (Table 1)

Summary

Introduction

Species distribution models are commonly used tools to describe diversity patterns and support conservation measures. Co-occurrence patterns of species across a landscape may arise due to shared habitat preferences, dispersal patterns, community interactions (e.g. facilitation, competition) or the interaction of these processes [1, 2] This interest in joint distributions of species relates. Manipulative experiments are unfeasible, and detection of interactions relies on correlative evidence, recently in the form of joint analyses of species abundances or occurrences (joint species distribution models: JSDMs). With this method, environmental drivers of species distributions are accounted for, and remaining correlation in the model residuals indicate association between species, often interpreted as biotic interactions Environmental drivers of species distributions are accounted for, and remaining correlation in the model residuals indicate association between species, often interpreted as biotic interactions (e.g. [6, 7])

Objectives

Methods

Results

Discussion

Conclusion