Intrinsic ecological dynamics drive biodiversity turnover in model metacommunities

ARTICLE https://doi.org/10.1038/s41467-021-23769-7 OPEN Intrinsic ecological dynamics drive biodiversity turnover in model metacommunities 1234567890():,; Jacob D. O’Sullivan 1 ✉, J. Christopher D. Terry 1 & Axel G. Rossberg1 Turnover of species composition through time is frequently observed in ecosystems. It is often interpreted as indicating the impact of changes in the environment. Continuous turnover due solely to ecological dynamics—species interactions and dispersal—is also known to be theoretically possible; however the prevalence of such autonomous turnover in natural communities remains unclear. Here we demonstrate that observed patterns of compositional turnover and other important macroecological phenomena can be reproduced in large spatially explicit model ecosystems, without external forcing such as environmental change or the invasion of new species into the model. We find that autonomous turnover is triggered by the onset of ecological structural instability—the mechanism that also limits local biodiversity. These results imply that the potential role of autonomous turnover as a widespread and important natural process is underappreciated, challenging assumptions implicit in many observation and management tools. Quantifying the baseline level of compositional change would greatly improve ecological status assessments. 1 School of Biological and Chemical Sciences, Queen Mary University of London, London, UK. ✉email: NATURE COMMUNICATIONS | (2021)12:3627 | https://doi.org/10.1038/s41467-021-23769-7 | www.nature.com/naturecommunications 1 ARTICLE C NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23769-7 hange in species composition observed in a single location through time, called community turnover, is observed in all natural ecosystems. Potential drivers of such biotic change include changes in the abiotic environment, random demographic fluctuations (referred to as community drift) and population dynamics driven by ecological interactions and dispersal. Analysis of community time series suggests communities turn over at a faster rate than can be explained by random drift1,2. Climate change and other anthropogenic pressures are known to contribute to community turnover3–7 and there is evidence to suggest that turnover is accelerating in some biomes8. Ecological assessments, projections and mitigation strategies are therefore commonly designed around the assumption that communities turn over predominantly in response environmental change and direct anthropogenic pressures9. The extent to which processes intrinsic to ecosystems contribute to turnover, however, remains poorly understood10. Understanding the expected amount of temporal turnover due to such intrinsic processes is of vital importance if ecological change is to be accurately interpreted7. If strong temporal community turnover were a natural phenomenon that can arise independently of changes in the abiotic environment, then observed shifts in the composition of ecological communities would not on their own demonstrate external pressures. In theoretical models of ecological communities population abundances do not necessarily arrive at fixed points. Instead, such systems can manifest persistent dynamics which we refer to here as ‘autonomous’ since they do not depend on variation in the external environment or other extrinsic drivers. When these population fluctuations are strong, changes in the abundances of species can be dramatic and even drive species locally extinct; if an excluded species retains occupancy in adjacent patches11, it may re-colonise at some future time. We refer to as ‘autonomous turnover’ local compositional changes involving colonisationextinction processes or significant restructuring of relative abundances, driven by such autonomous population dynamics. Limited availability of historical turnover data before the onset of widespread anthropogenic impacts poses considerable challenges when trying to establish the natural baseline of turnover. Nevertheless, broad consistency amongst the species-time-area relationships observed in extant assemblages12,13 points to a consistency in the dominant underlying biological process. It is reasonable to expect, therefore, that the drivers of such spatiotemporal turnover can be probed using theoretical models. Previous theoretical11,14–17 and experimental studies18 have shown how competitive ecological communities (for specific network structures or parameter combinations) can generate any type of dynamical behaviour, including persistent chaotic cycles. Likewise, antagonistic interactions between predator and prey species have been shown in both theory and experiment to lead to persistent population oscillations in the absence of external variation19,20. However, these cyclic processes are different from, and have not usually been associated with, empirical observations of acyclic, directional compositional turnover1,2. An important distinction between these processes lies in the role of space. While cyclic forms of community dynamics can lead to characteristic spatial structures17,18,21, cyclic dynamics do not in principle require space19,20. Acyclic turnover, on the other hand, manifestly involves colonisation by species from surrounding patches and therefore explicitly requires that a community is embedded in a spatially structured ecological neighbourhood. Here we ask: can community dynamics enabled by spatial structure account for the observed macroecological patterns in compositional turnover? We address this question drawing on recent advances in the theory of spatially extended ecological communities, so called “metacommunities”22, using a 2 population-dynamical simulation model with explicit spatial and environmental structure23 that has been shown to reproduce fundamental spatial biodiversity patterns. As previously shown, such metacommunity models can exhibit a phenomenon called “ecological structural instability”24, as a result of which species richness at both local and regional scales is intrinsically regulated23. The structural stability of a system refers to its capacity to sustain changes in parameters without undergoing qualitative changes in dynamical behaviour25. As such, ecological structural stability is taken to describe in particular the capacity of a community to persist in the face of small biotic or abiotic perturbations24,26–30. Empirical observation of many of the emergent phenomena associated with ecological structural instability provides compelling indirect evidence for the prevalence of structural instability in nature23,31. Our understanding of the impact of structurally unstable diversity regulation on temporal community-level properties, however, remains incomplete32. Here we build upon earlier work by exploring the spatio-temporal patterns that emerge in metacommunity models. We find that, when expanding the spatial and taxonomic scale of simulations beyond those studied previ (...truncated)