The rise of a marine generalist predator and the fall of beta diversity.

Kari E Ellingsen,Marti J Anderson,Andrey V Dolgov,Nigel G Yoccoz,Edda Johannesen,Torkild Tveraa,Kenneth T Frank,Nancy L Shackell

doi:10.1111/gcb.15027

Abstract

Determining the importance of physical and biological drivers in shaping biodiversity in diverse ecosystems remains a global challenge. Advancements have been made towards this end in large marine ecosystems with several studies suggesting environmental forcing as the primary driver. However, both empirical and theoretical studies point to additional drivers of changes in diversity involving trophic interactions and, in particular, predation. Moreover, a more integrated but less common approach to the assessment of biodiversity changes involves analyses of spatial β diversity, whereas most studies to date assess only changes in species richness (α diversity). Recent research has established that when cod, a dominant generalist predator, was overfished and collapsed in a northwest Atlantic food web, spatial β diversity increased; that is, the spatial structure of the fish assemblage became increasingly heterogeneous. If cod were to recover, would this situation be reversible, given the inherent complexity and non-linear dynamics that typify such systems? A dramatic increase of cod in an ecologically similar large marine ecosystem may provide an answer. Here we show that spatial β diversity of fish assemblages in the Barents Sea decreased with increasing cod abundance, while decadal scale changes in temperature did not play a significant role. These findings indicate a reversibility of the fish assemblage structure in response to changing levels of an apex predator and highlight the frequently overlooked importance of trophic interactions in determining large-scale biodiversity patterns. As increased cod abundance was largely driven by changes in fisheries management, our study also shows that management policies and practices, particularly those involving apex predators, can have a strong effect in shaping spatial diversity patterns, and one should not restrict the focus to effects of climate change alone.

Highlights

Ecologists strive to understand the processes responsible for generating and modifying diversity in natural ecosystems (Chase et al, 2018; Ricklefs, 1987)
We explored the performance of competing models using both widely applicable information criterion (WAIC) and Leave-One-Out cross-validation Information Criterion (LOOIC) as the latter can be more robust in the case of weak priors or influential observations (Vehtari et al, 2017)
We document a decadal-scale pattern of spatial homogenization of fish assemblages in the Barents Sea. This decline in β diversity can be explained by an increase in abundance of cod, a generalist apex predator

Summary

| INTRODUCTION

Ecologists strive to understand the processes responsible for generating and modifying diversity in natural ecosystems (Chase et al, 2018; Ricklefs, 1987). Many highly perturbed ecosystems shift from one state to another, but a reversal back to a pre-disturbance state rarely occurs following removal of the perturbation (Scheffer, Carpenter, & Young, 2005) If this process were reversible, we would expect that the recent increase in Barents Sea cod would result in a decline of β diversity, that is, a broad-scale spatial homogenization of resident fish assemblages. We delineated large subregions (of approximately equal spatial extent) having contrasting patterns in cod abundance and temperature through time, and the patterns of β diversity within each of these subregions were assessed This design-based approach (Bråthen et al, 2007; Butsic, Lewis, Radeloff, Baumann, & Kuemmerle, 2017; Shadish, Cook, & Campbell, 2002) to the analysis of observational data allowed us to disentangle the relative importance of an apex predator (cod) and a physical climate variable (bottom temperature) on the β diversity of the resident fish community. Ten years of monitoring data derived from a systematic, large-scale scientific sampling program conducted annually from 2004 to 2013 were used for the analysis

| Study design

Findings

| DISCUSSION