Metacommunity Model Research Articles

Extinction and Ecosystem Function Debt Across Dispersal Rate and Behaviour in a Heterogeneous Metacommunity Model

ABSTRACTAimHabitat destruction causes “extinction debt” and is also thought to produce ecosystem function debt, but theory of their magnitude and nature is limited. Heterogeneous landscapes are fundamental to the maintenance of species richness and ecosystem function, while directed or undirected dispersal behaviour, such as dispersal of seeds by animals or by the wind, is also important, especially after habitat destruction. We therefore consider extinction and ecosystem function debt under different dispersal rates and behaviours in heterogeneous landscapes.MethodsWe use a classic heterogeneous metacommunity model to capture the dynamics of competing species in local patches linked by dispersal and varying in environmental conditions. We remove one patch at a time and measure extinction debt and ecosystem function debt by the number/proportion of delayed extinctions and the amount of biomass change, respectively.ResultsWe reveal three species extinction regimes as dispersal increases: (1). species most adapted to the removed habitat are most at risk; (2). similarly adapted species are also at risk; (3). patch removal shifts competitive balance among the few species coexisting at high dispersal, where competition is strong. We find surprisingly that destruction of habitat can hasten the extinction of those species best adapted to harsh environments and that the proportion of diversity at risk from extinction actually increases with dispersal because competition is intense there. Finally, there can be a small ecosystem credit but extinction debt when dispersers reroute to potentially more favourable remaining habitats (directed dispersal), especially when harsh environments are removed. However, ecosystem debt occurs and can be large under undirected dispersal.Main ConclusionsThe magnitude and nature of extinction and ecosystem function debts depend on species dispersal rates and behaviours, as well as the environmental conditions of the disturbed habitats. Conservation actions will be more successful if they consider these factors.

Habitat Fragmentation Affects Richness—A View Through a Metacommunity Lens

Habitat fragmentation is often assumed to negatively impact species diversity because smaller, more isolated populations on smaller habitat patches are at a higher extinction risk. However, some empirical and theoretical studies suggest that landscapes with numerous small habitat patches may support higher species richness, although the circumstances remain elusive. We used an agent-based metacommunity model to investigate this and simulate landscapes of the same total area but diverse patch sizes. Our model, as generic and unbiased by specific assumptions as possible, aimed to explore which circumstances may be more conducive to supporting higher biodiversity. To this end, most parameters and behaviors were random. The model included generalized species traits, dispersal, and interactions to explore species richness dynamics in fragmented landscapes of distinct patch sizes. Our results show that landscapes with many small patches maintain higher species richness than those with fewer large patches. Moreover, the relationship between patch connectivity and species richness is more pronounced in landscapes with smaller patches. High connectivity in these landscapes may support species diversity by preventing local extinctions and facilitating recolonization. In contrast, connectivity is less significant in large-patch landscapes, where generalist species dominate. The findings highlight the complex interplay between patch size quality, connectivity, species traits, and diverse interactions among species in determining species richness. We suggest the patterns produced by the model represent null predictions and may be useful as a reference for a diversity of more specialized questions and predictions. These insights may also have specific implications for conservation strategies, suggesting that maintaining a mosaic of small, well-connected patches could enhance biodiversity in fragmented landscapes.

Self‐organised pattern formation promotes consumer coexistence by fluctuation‐dependent mechanisms

Abstract In spatially extended ecosystems, habitat heterogeneity facilitates coexistence of species if each competitor experiences environmental conditions in a particular habitat that provide a growth advantage compared to other species, thereby making it locally competitively superior. If the competitive hierarchy among species is the same everywhere, heterogeneity‐facilitated coexistence is possible provided that superior competitors disperse maladaptively towards unfavourable habitats or if they hedge insufficiently against fluctuating environmental conditions. We use a generic two‐patch metacommunity model to show that the latter mechanisms also operate in metacommunities with homogeneous habitat quality when heterogeneous biomass distributions emerge from self‐organised pattern formation. The model consists of an abiotic resource, an autotroph producer and two competing heterotroph consumer species of which one is always competitively inferior to the other, irrespective of resource availability. If the induced biomass patterns are static in time, a lower dispersal rate can allow the inferior competitor to avoid competitive exclusion by retaining most of its biomass in the patch with the higher resource density. However, if the biomass patterns fluctuate spatio‐temporally, the inferior competitor must adopt a higher dispersal rate than the superior competitor to persist. This increased movement enables the inferior competitor to effectively distribute its biomass across space, thereby achieving a higher growth rate during periods of recovery from local population minima. Strikingly, we find a novel coexistence mechanism that emerges if the competitors differ in their abilities to induce pattern formation. Similar to relative nonlinearity in resource use (based e.g. on a gleaner–opportunist trade‐off), the dominant species modifies the spatial or spatio‐temporal variation in the distribution of the resource in a way that favours its competitor. This prevents competitive exclusion due to differently effective dispersal strategies. We conclude that while temporal instabilities that cause, for example, predator–prey oscillations are usually regarded as jeopardising species' persistence, spatial instabilities that give rise to self‐organised pattern formation should be interpreted more positively, as they provide a generic mechanisms for maintaining diversity in metacommunities without requiring a priori habitat heterogeneity. Read the free Plain Language Summary for this article on the Journal blog.

Understanding the transmission of bacterial agents of sapronotic diseases using an ecosystem-based approach: A first spatially realistic metacommunity model.

Pathogens such as bacteria, fungi and viruses are important components of soil and aquatic communities, where they can benefit from decaying and living organic matter, and may opportunistically infect human and animal hosts. One-third of human infectious diseases is constituted by sapronotic disease agents that are natural inhabitants of soil or aquatic ecosystems. They are capable of existing and reproducing in the environment outside of the host for extended periods of time. However, as ecological research on sapronosis is infrequent and epidemiological models are even rarer, very little information is currently available. Their importance is overlooked in medical and veterinary research, as well as the relationships between free environmental forms and those that are pathogenic. Here, using dynamical models in realistic aquatic metacommunity systems, we analyze sapronosis transmission, using the human pathogen Mycobacterium ulcerans that is responsible for Buruli ulcer. We show that the persistence of bacilli in aquatic ecosystems is driven by a seasonal upstream supply, and that the attachment and development of cells to aquatic living forms is essential for such pathogen persistence and population dynamics. Our work constitutes the first set of metacommunity models of sapronotic disease transmission, and is highly flexible for adaptation to other types of sapronosis. The importance of sapronotic agents on animal and human disease burden needs better understanding and new models of sapronosis disease ecology to guide the management and prevention of this important group of pathogens.

Interspecific dispersal constraints suppress pattern formation in metacommunities.

Decisions to disperse from a habitat stand out among organismal behaviours as pivotal drivers of ecosystem dynamics across scales. Encounters with other species are an important component of adaptive decision-making in dispersal, resulting in widespread behaviours like tracking resources or avoiding consumers in space. Despite this, metacommunity models often treat dispersal as a function of intraspecific density alone. We show, focusing initially on three-species network motifs, that interspecific dispersal rules generally drive a transition in metacommunities from homogeneous steady states to self-organized heterogeneous spatial patterns. However, when ecologically realistic constraints reflecting adaptive behaviours are imposed-prey tracking and predator avoidance-a pronounced homogenizing effect emerges where spatial pattern formation is suppressed. We demonstrate this effect for each motif by computing master stability functions that separate the contributions of local and spatial interactions to pattern formation. We extend this result to species-rich food webs using a random matrix approach, where we find that eventually, webs become large enough to override the homogenizing effect of adaptive dispersal behaviours, leading once again to predominately pattern-forming dynamics. Our results emphasize the critical role of interspecific dispersal rules in shaping spatial patterns across landscapes, highlighting the need to incorporate adaptive behavioural constraints in efforts to link local species interactions and metacommunity structure. This article is part of the theme issue 'Diversity-dependence of dispersal: interspecific interactions determine spatial dynamics'.

Habitat amount modulates biodiversity responses to fragmentation.

Anthropogenic habitat destruction leads to habitat loss and fragmentation, both of which interact to determine how biodiversity changes at the landscape level. While the detrimental effects of habitat loss are clear, there is a long-standing debate about the role of habitat fragmentation per se. We identify the influence of the total habitat amount lost as a modulator of the relationship between habitat fragmentation and biodiversity. Using a simple metacommunity model characterized by colonization-competition (C-C) trade-offs, we show that the magnitude of habitat loss can induce a unimodal response of biodiversity to habitat fragmentation. When habitat loss is low, habitat fragmentation promotes coexistence by suppressing competitively dominant species, while habitat fragmentation at high levels of habitat loss can shape many smaller isolated patches that drive extinctions of superior competitors. While the C-C trade-off is not the only mechanism for biodiversity maintenance, the modulation of habitat fragmentation effects by habitat loss is probably common. Reanalysis of a globally distributed dataset of fragmented animal and plant metacommunities shows an overall pattern that supports this hypothesis, suggesting a resolution to the debate regarding the relative importance of positive versus negative fragmentation effects.

Multi-scale effects of behavioral movement deterrents on invasive carp metapopulations

Behavioral deterrents of among-pool movement represent a promising tool for controlling invasive fish populations. To date, much of the research in this area has been focused on the direct effectiveness of different methods of deterrence. However, the effect of these structures on populations in spatially complex habitats is unknown. We combine a metacommunity model with movement data of two invasive species (bighead carp and silver carp) in a large river to assess local and river-wide scale population outcomes of deterrent locations. We calculated (1) which potential deterrent locations are most effective at reducing the growth at the invasion front (2) the river-scale population effects at each location, and (3) what, if any, are the risks imposed by altering the current spatial dynamics. We found that the effects on the population dynamics at the invasion front varied with the location of deterrents, ranging from near zero to effects equal to the reduction in an individual’s movement across the deterrent. The river-scale population growth rate was slightly increased by all potential deterrent placements because the deterrents tended to concentrate more of the river-scale population into pools with the highest recruitment rates. The short-term, transient dynamics followed a strictly decreasing pattern after deterrent placement suggesting no additional short-term risk. These results suggest that deterrents can be an effective and low-risk intervention for the control of invasive carp, although the population level effect will depend on the interaction of the traits and behavior of the species with the physical character and spatial structure of the habitat.

Tipping points emerge from weak mutualism in metacommunities.

The coexistence of obligate mutualists is often precariously close to tipping points where small environmental changes can drive catastrophic shifts in species composition. For example, microbial ecosystems can collapse by the decline of a strain that provides an essential resource on which other strains cross-feed. Here, we show that tipping points, ecosystem collapse, bistability and hysteresis arise even with very weak (non-obligate) mutualism provided the population is spatially structured. Based on numeric solutions of a metacommunity model and mean-field analyses, we demonstrate that weak mutualism lowers the minimal dispersal rate necessary to avoid stochastic extinction, while species need to overcome a mean threshold density to survive in this low dispersal rate regime. Our results allow us to make numerous predictions for mutualistic metacommunities regarding tipping points, hysteresis effects, and recovery from external perturbations, and let us draw general conclusions for ecosystems even with random, not necessarily mutualistic, interactions and systems with density-dependent dispersal rather than direct mutualistic interactions.

Early warning indicators capture catastrophic transitions driven by explicit rates of environmental change.

In response to external changes, ecosystems can undergo catastrophic transitions. Early warning indicators aim to predict such transitions based on the phenomenon of critical slowing down at bifurcation points found under a constant environment. When an explicit rate of environmental change is considered, catastrophic transitions can become distinct phenomena from bifurcations, and result from a delayed response to noncatastrophic bifurcations. We use a trophic metacommunity model where transitions in time series and bifurcations of the system are distinct phenomena. We calculate early warning indicators from the time series of the continually changing system and show that they predict not the bifurcation of the underlying system but the actual catastrophic transition driven by the explicit rate of change. Predictions based on the bifurcation structure could miss catastrophic transitions that can still be captured by early warning signals calculated from time series. Our results expand the repertoire of mechanistic models used to anticipate catastrophic transitions to nonequilibrium ecological systems exposed to a constant rate of environmental change.

Effects of dispersal and temperature variability on phytoplankton realized temperature niches.

Phytoplankton species exhibit fundamental temperature niches that drive observed species distributions linked to realized temperature niches. A recent analysis of field observations of Prochlorococcus showed that for all ecotypes, the realized niche was, on average, colder and wider than the fundamental niche. Using a simple trait-based metacommunity model that resolves fundamental temperature niches for a range of competing phytoplankton, we ask how dispersal and local temperature variability influence species distributions and diversity, and whether these processes help explain the observed discrepancies between fundamental and realized niches for Prochlorococcus. We find that, independently, both dispersal and temperature variability increase realized temperature niche widths and local diversity. The combined effects result in high diversity and realized temperature niches that are consistently wider than fundamental temperature niches. These results have broad implications for understanding the drivers of phytoplankton biogeography as well as for refining species distribution models used to project how climate change impacts phytoplankton distributions.

Assessing mutualistic metacommunity capacity by integrating spatial and interaction networks

We develop a spatially realistic model of mutualistic metacommunities that exploits the joint structure of spatial and interaction networks. Assuming that all species have the same colonisation and extinction parameters, this model exhibits a sharp transition between stable non-null equilibrium states and a global extinction state. This behaviour allows defining a threshold on colonisation/extinction parameters for the long-term metacommunity persistence. This threshold, the ‘metacommunity capacity’, extends the metapopulation capacity concept and can be calculated from the spatial and interaction networks without needing to simulate the whole dynamics. In several applications we illustrate how the joint structure of the spatial and the interaction networks affects metacommunity capacity. It results that a weakly modular spatial network and a power-law degree distribution of the interaction network provide the most favourable configuration for the long-term persistence of a mutualistic metacommunity. Our model that encodes several explicit ecological assumptions should pave the way for a larger exploration of spatially realistic metacommunity models involving multiple interaction types.

Effects of compositional heterogeneity and spatial autocorrelation on richness and diversity in simulated landscapes.

Landscape structure plays a key role in mediating a variety of ecological processes affecting biodiversity patterns; however, its precise effects and the mechanisms underpinning them remain unclear. While the effects of landscape structure have been extensively investigated both empirically and theoretically from a metapopulation perspective, the effects of spatial structure at the landscape scale remain poorly explored from a metacommunity perspective. Here, we attempt to address this gap using a spatially explicit, individual-based metacommunity model to explore the effects of landscape compositional heterogeneity and per se spatial configuration on diversity at the landscape and patch levels via their influence on long-term community assembly processes. Our model simulates communities composed of species of annual, asexual organisms living, reproducing, dispersing, and competing within grid-based, fractal landscapes that vary in their magnitude of spatial environmental heterogeneity and in their degree of spatial environmental autocorrelation. Communities are additionally subject to temporal environmental fluctuations and external immigration, allowing for turnover in community composition. We found that compositional heterogeneity and spatial autocorrelation had differing effects on richness, diversity, and the landscape and patch scales. Landscape-level diversity was driven by community dissimilarity at the patch level and increased with greater heterogeneity, while landscape richness was largely the result of the short-term accumulation of immigrants and decreased with greater compositional heterogeneity. Both richness and diversity decreased in variance with greater compositional heterogeneity, indicating a reduction in community turnover over time. Patch-level richness and diversity patterns appeared to be driven by overall landscape richness and local mass effects, resulting in maximum patch-level richness and diversity at moderate levels of compositional heterogeneity and high spatial autocorrelation.

Increased dispersal explains increasing local diversity with global biodiversity declines.

The narrative of biodiversity decline in response to human impacts is overly simplistic because different aspects of biodiversity show different trajectories at different spatial scales. It is also debated whether human-caused biodiversity changes lead to subsequent, accelerating change (cascades) in ecological communities, or alternatively build increasingly robust community networks with decreasing extinction rates and reduced invasibility. Mechanistic approaches are needed that simultaneously reconcile different aspects of biodiversity change, and explore the robustness of communities to further change. We develop a trophically structured, mainland-archipelago metacommunity model of community assembly. Varying the parameters across model simulations shows that local alpha diversity (the number of species per island) and regional gamma diversity (the total number of species in the archipelago) depend on both the rate of extirpation per island and on the rate of dispersal between islands within the archipelago. In particular, local diversity increases with increased dispersal and heterogeneity between islands, but regional diversity declines because the islands become biotically similar and local one-island and few-island species are excluded (homogenisation, or reduced beta diversity). This mirrors changes observed empirically: real islands have gained species (increased local and island-scale community diversity) with increased human-assisted transfers of species, but global diversity has declined with the loss of endemic species. However, biological invasions may be self-limiting. High-dispersal, high local-diversity model communities become resistant to subsequent invasions, generating robust species-community networks unless dispersal is extremely high. A mixed-up world is likely to lose many species, but the resulting ecological communities may nonetheless be relatively robust.

Connecting local and regional scales with stochastic metacommunity models: Competition, ecological drift, and dispersal

AbstractDespite the well known scale‐dependency of ecological interactions, relatively little attention has been paid to understanding the dynamic interplay between various spatial scales. This is especially notable in metacommunity theory, where births and deaths dominate dynamics within patches (the local scale), and dispersal and environmental stochasticity dominate dynamics between patches (the regional scale). By considering the interplay of local and regional scales in metacommunities, the fundamental processes of community ecology—selection, drift, and dispersal—can be unified into a single theoretical framework. Here, we analyze three related spatial models that build on the classic two‐species Lotka–Volterra competition model. Two open‐system models focus on a single patch coupled to a larger fixed landscape by dispersal. The first is deterministic, while the second adds demographic stochasticity to allow ecological drift. Finally, the third model is a true metacommunity model with dispersal between a large number of local patches, which allows feedback between local and regional scales and captures the well studied metacommunity paradigms as special cases. Unlike previous simulation models, our metacommunity model allows the numerical calculation of equilibria and invasion criteria to precisely determine the outcome of competition at the regional scale. We show that both dispersal and stochasticity can lead to regional outcomes that are different than predicted by the classic Lotka–Volterra competition model. Regional exclusion can occur when the nonspatial model predicts coexistence or founder control, due to ecological drift or asymmetric stochastic switching between basins of attraction, respectively. Regional coexistence can result from local coexistence mechanisms or through competition‐colonization or successional‐niche trade‐offs. Larger dispersal rates are typically competitively advantageous, except in the case of local founder control, which can favor intermediate dispersal rates. Broadly, our models demonstrate the importance of feedback between local and regional scales in competitive metacommunities and provide a unifying framework for understanding how selection, drift, and dispersal jointly shape ecological communities.

Preparing for and managing crown-of-thorns starfish outbreaks on reefs under threat from interacting anthropogenic stressors

Crown-of-thorns starfish (CoTS) outbreaks rank among the greatest threats to coral throughout the Indo-Pacific. In the future, reefs already stressed by CoTS will be further burdened by overfishing and nutrient loading. How much these two factors will exacerbate CoTS outbreak severity is still uncertain. Furthermore, the CoTS management literature has focused on the Great Barrier Reef, whereas outbreak damage is rising across the Indo-Pacific. Here, we use a metacommunity model to simulate CoTS outbreaks in areas with high and growing levels of fishing pressure and offshore nutrient input. We model outbreaks on reefs adjacent to two cities within the range of CoTS that have less prior literature coverage: Cebu City, Philippines, and Jeddah, Saudi Arabia. We observe that the combination of population increases and urbanization of previously rural areas can drive complex patterns of multi-stressor interaction. We find that CoTS removal on intermediate spatial scales significantly improves regional-scale coral health, and provide guidelines under which each of four CoTS management strategies is optimal for conservation. We find that coral decline due to overfishing can be sharper on reefs with CoTS, and that nutrification can induce a shift from discrete outbreak waves to continuous CoTS presence. Our work shows the importance of long-term planning for reef management, and highlights how reef stressors can interact in potentially unforeseen ways.

Priority Effects Determine How Dispersal Affects Biodiversity in Seasonal Metacommunities.

AbstractThe arrival order of species frequently determines the outcome of their interactions. This phenomenon, called the priority effect, is ubiquitous in nature and determines local community structure, but we know surprisingly little about how it influences biodiversity across different spatial scales. Here, we use a seasonal metacommunity model to show that biodiversity patterns and the homogenizing effect of high dispersal depend on the specific mechanisms underlying priority effects. When priority effects are driven only by positive frequency dependence, dispersal-diversity relationships are sensitive to initial conditions but generally show a hump-shaped relationship: biodiversity declines when dispersal rates become high and allow the dominant competitor to exclude other species across patches. When spatiotemporal variation in phenological differences alters species' interaction strengths (trait-dependent priority effects), local, regional, and temporal diversity are surprisingly insensitive to variation in dispersal, regardless of the initial numeric advantage. Thus, trait-dependent priority effects can strongly reduce the effect of dispersal on biodiversity, preventing the homogenization of metacommunities. Our results suggest an alternative mechanism that maintains local and regional diversity without environmental heterogeneity, highlighting that accounting for the mechanisms underlying priority effects is fundamental to understanding patterns of biodiversity.

Habitat restoration and the recovery of metacommunities

Abstract Ecosystem restoration is becoming a widely recognised solution to the biodiversity crisis. However, there is still a gap between restoration science and practice. Specifically, we lack a theoretical framework that would improve our understanding of ecosystems' recovery and allow us to optimise restoration design. Here, we narrow this gap by developing spatially explicit metacommunity models and studying the recovery dynamics of communities during restoration. We show that community response depends on how damaged the landscape is prior to restoration, with highly fragmented landscapes imposing greater challenges to community recovery. In such cases, recovery depends on the type of interaction and the structure of the interaction network. Furthermore, we demonstrate that community recovery can be maximised with careful spatial planning. Specifically, when recovering communities composed of antagonistic interactions, restoration should target areas adjacent to the most species‐rich sites. In the case of mutualistic communities, the same strategy should be adopted in the short term, whereas in the long term, restoration should be extended to sites that improve the overall connectivity of the landscape. Synthesis and applications: Our results highlight the importance of considering interactions between species and spatial planning in restoration projects. Moreover, they provide insights into improving the efficiency of restoration and, thus, can help guide the design of restoration projects.

Maintenance of biodiversity in multitrophic metacommunities: Dispersal mode matters.

Although metacommunity models generally formulate dispersal as a random, passive process, mounting evidence suggests that dispersal can be an active process depending on species fitness over the landscape, particularly in multitrophic communities. How different dispersal modes (i.e. from random to increasingly fitness-dependent dispersal) modulate the effect of dispersal on biodiversity remains unclear. Here, we used a metacommunity model of food webs to investigate the effects of dispersal and habitat heterogeneity on biodiversity and how these effects may be dependent on dispersal mode. Our results showed that compared to isolated systems, random dispersal increased local food web diversity ( diversity) but decreased across-community dissimilarity ( diversity) and regional food web diversity ( diversity), consistent with findings from competitive metacommunity models. However, fitness-dependency could alter the effects of dispersal on biodiversity. Both and diversity increased with the strength of fitness-dependency of dispersal, while diversity peaked at intermediate fitness-dependency. Notably, strong fitness-dependent dispersal maintained levels of and diversity similar to those observed in isolated systems. Thus, random dispersal and isolation (i.e. no dispersal) can be considered as two extremes along the continuum of fitness-dependent dispersal, in terms of their effects on biodiversity. Moreover, both biodiversity-habitat heterogeneity and biodiversity-habitat connectivity relationships depended on the dispersal mode. Strikingly, under random dispersal, diversity decreased with habitat heterogeneity and connectivity, but under strong fitness-dependent dispersal, it increased with habitat heterogeneity and remained unchanged as habitat connectivity increased. Our study highlights the context dependence of dispersal effects on biodiversity in heterogeneous landscapes. Our findings have useful implications for biodiversity conservation and landscape management, where management strategies should account for different modes of dispersal across taxa, thus different responses of biodiversity to habitat heterogeneity and connectivity.

Heterogeneity in the isolation of patches may be essential for the action of metacommunity mechanisms

The spatial isolation gradient of communities and the gradient in the species dispersal ability are recognized as determinants of biodiversity in metacommunities. In spite of this, mean field models, spatially explicit models, and experiments were mainly focused on idealized spatial arrangements of communities leaving aside the combining role of dispersal and isolation gradients in metacommunity processes. Consequently, we have an incipient understanding of the role of the real spatial arrangement of communities on biodiversity patterns. We focus on six metacommunities for which confident information about the spatial arrangement of water bodies is available. Using coalescent metacommunity models and null models that randomize the location of water bodies, we estimated the potential effect of the landscape on biodiversity and its dependence on species dispersal ability. At extremely low or high dispersal abilities, the location of ponds does not influence diversity because different communities are equally affected by the low or high incoming dispersal. At intermediate dispersal abilities, peripheral communities present a much lower richness and higher beta diversity than central communities. Moreover, metacommunities from real landscapes host more biodiversity than randomized landscapes, a result that is determined by the heterogeneity in the geographic isolation of communities. In a dispersal gradient, mass effects systematically increase the local richness and decrease beta diversity. However, the spatial arrangement of patches only has a large importance in metacommunity processes at intermediate dispersal abilities, which ensures access to central locations but limits dispersal in isolated communities. The ongoing reduction in spatial extent and simplification of the landscape may consequently undermine the metacommunity processes that support biodiversity, something that should be explicitly considered in preserving and restoring strategies.

Community composition exceeds area as a predictor of long-term conservation value.

Conserving biodiversity often requires deciding which sites to prioritise for protection. Predicting the impact of habitat loss is a major challenge, however, since impacts can be distant from the perturbation in both space and time. Here we study the long-term impacts of habitat loss in a mechanistic metacommunity model. We find that site area is a poor predictor of long-term, regional-scale extinctions following localised perturbation. Knowledge of the compositional distinctness (average between-site Bray-Curtis dissimilarity) of the removed community can markedly improve the prediction of impacts on regional assemblages, even when biotic responses play out at substantial spatial or temporal distance from the initial perturbation. Fitting the model to two empirical datasets, we show that this conclusions holds in the empirically relevant parameter range. Our results robustly demonstrate that site area alone is not sufficient to gauge conservation priorities; analysis of compositional distinctness permits improved prioritisation at low cost.