Unity makes strength: clustered extinctions have stronger, longer-lasting effects on metacommunities dynamics
Spatial distribution of local patch extinctions drives recovery dynamics in metacommunities
In this article, Saade et al. (2021) investigate how the rate of local extinctions and their spatial distribution affect recolonization dynamics in metacommunities. They use an elegant combination of microcosm experiments with metacommunities of freshwater ciliates and mathematical modelling mirroring their experimental system. Their main findings are (i) that local patch extinctions increase both local (α-) and inter-patch (β-) diversity in a transient way during the recolonization process, (ii) that these effects depend more on the spatial distribution of extinctions (dispersed or clustered) than on their amount, and (iii) that they may spread regionally.
Microcosm experiments are already quite cool just by themselves and have contributed largely to conceptual advances in community ecology (see Fraser and Keddy 1997, or Jessup et al. 2004 for reviews on this topic), but they are here exploited to a whole further level by the fitting of a metapopulation dynamics model. The model allows both to identify the underlying mechanisms most likely to generate the patterns observed (here, competitive interactions) and to assess the robustness of these patterns when considering larger spatial or temporal scales. This release of experimental limitations allows here for the analysis of quantitative metrics of spatial structure, like the distance to the closest patch, which gives an interesting insight into the functional basis of the effect of the spatial distribution of extinctions.
A major strength of this study is that it highlights the importance of considering the spatial structure explicitly. Recent work on ecological networks has shown repeatedly that network structure affects the propagation of pathogens (Badham and Stocker 2010), invaders (Morel-Journel et al. 2019), or perturbation events (Gilarranz et al. 2017). Here, the spatial structure of the metacommunity is a regular grid of patches, but the distribution of extinction events may be either regularly dispersed (i.e., extinct patches are distributed evenly over the grid and are all surrounded by non-extinct patches only) or clustered (all extinct patches are neighbours). This has a direct effect on the neighbourhood of perturbed patches, and because perturbations have mostly local effects, their recovery dynamics are dominated by the composition of this immediate neighbourhood. In landscapes with dispersed extinctions, the neighbourhood of a perturbed patch is not affected by the amount of extinctions, and neither is its recovery time. In contrast, in landscapes with clustered extinctions, the amount of extinctions affects the depth of the perturbed area, which takes longer to recover when it is larger. Interestingly, the spatial distribution of extinctions here is functionally equivalent to differences in connectivity between perturbed and unperturbed patches, which results in contrasted “rescue recovery” and “mixing recovery” regimes as described by Zelnick et al. (2019).
Furthermore, this study focuses on local dynamics of competition and short-term, transient patterns that may have been overlooked by more classical, equilibrium-based approaches of dynamical systems of metacommunities. Indeed, in a metacommunity composed of several competitors, early theoretical work demonstrated that species coexistence is possible at the regional scale only, provided that spatial heterogeneity creates spatial variance in fitness or precludes the superior competitor from accessing certain habitat patches (Skellam 1951, Levins 1969). In the spatially homogeneous experimental system of Saade et al., one of the three ciliate species ends up dominating the community at equilibrium. However, following local, one-time extinction events, the community endures a recolonization process in which differences in dispersal may provide temporary spatial niches for inferior competitors. These transient patterns might prove essential to understand and anticipate the resilience of natural systems that are under increasing pressure, and enduring ever more frequent and intense perturbations (IPBES 2019). Spatial autocorrelation in extinction events was previously identified as a risk for stability and persistence of metacommunities (Ruokolainen 2013, Kahilainen et al. 2018). These new results show that autocorrelated perturbations also have longer-lasting effects, which is likely to increase their overall impact on metacommunity dynamics. As spatial and temporal autocorrelation of temperature and extreme climatic events are expected to increase (Di Cecco and Gouthier 2018), studies that investigate how metacommunities respond to the structure of the distribution of perturbations are more necessary than ever.
Badham J, Stocker R (2010) The impact of network clustering and assortativity on epidemic behaviour. Theoretical Population Biology, 77, 71–75. https://doi.org/10.1016/j.tpb.2009.11.003
Di Cecco GJ, Gouhier TC (2018) Increased spatial and temporal autocorrelation of temperature under climate change. Scientific Reports, 8, 14850. https://doi.org/10.1038/s41598-018-33217-0
Fraser LH, Keddy P (1997) The role of experimental microcosms in ecological research. Trends in Ecology & Evolution, 12, 478–481. https://doi.org/10.1016/S0169-5347(97)01220-2
Gilarranz LJ, Rayfield B, Liñán-Cembrano G, Bascompte J, Gonzalez A (2017) Effects of network modularity on the spread of perturbation impact in experimental metapopulations. Science, 357, 199–201. https://doi.org/10.1126/science.aal4122
IPBES (2019) Summary for policymakers of the global assessment report on biodiversity and ecosystem services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. S. Díaz, J. Settele, E. S. Brondízio E.S., H. T. Ngo, M. Guèze, J. Agard, A. Arneth, P. Balvanera, K. A. Brauman, S. H. M. Butchart, K. M. A. Chan, L. A. Garibaldi, K. Ichii, J. Liu, S. M. Subramanian, G. F. Midgley, P. Miloslavich, Z. Molnár, D. Obura, A. Pfaff, S. Polasky, A. Purvis, J. Razzaque, B. Reyers, R. Roy Chowdhury, Y. J. Shin, I. J. Visseren-Hamakers, K. J. Willis, and C. N. Zayas (eds.). IPBES secretariat, Bonn, Germany. 56 pages. https://doi.org/10.5281/zenodo.3553579
Jessup CM, Kassen R, Forde SE, Kerr B, Buckling A, Rainey PB, Bohannan BJM (2004) Big questions, small worlds: microbial model systems in ecology. Trends in Ecology & Evolution, 19, 189–197. https://doi.org/10.1016/j.tree.2004.01.008
Kahilainen A, van Nouhuys S, Schulz T, Saastamoinen M (2018) Metapopulation dynamics in a changing climate: Increasing spatial synchrony in weather conditions drives metapopulation synchrony of a butterfly inhabiting a fragmented landscape. Global Change Biology, 24, 4316–4329. https://doi.org/10.1111/gcb.14280
Levins R (1969) Some Demographic and Genetic Consequences of Environmental Heterogeneity for Biological Control1. Bulletin of the Entomological Society of America, 15, 237–240. https://doi.org/10.1093/besa/15.3.237
Morel-Journel T, Assa CR, Mailleret L, Vercken E (2019) Its all about connections: hubs and invasion in habitat networks. Ecology Letters, 22, 313–321. https://doi.org/10.1111/ele.13192
Ruokolainen L (2013) Spatio-Temporal Environmental Correlation and Population Variability in Simple Metacommunities. PLOS ONE, 8, e72325. https://doi.org/10.1371/journal.pone.0072325
Saade C, Kefi S, Gougat-Barbera C, Rosenbaum B, Fronhofer EA (2021) Spatial distribution of local patch extinctions drives recovery dynamics in metacommunities. bioRxiv, 2020.12.03.409524, ver. 4 peer-reviewed and recommended by Peer Community in Ecology. https://doi.org/10.1101/2020.12.03.409524
Skellam JG (1951) Random Dispersal in Theoretical Populations. Biometrika, 38, 196–218. https://doi.org/10.2307/2332328
Zelnik YR, Arnoldi J-F, Loreau M (2019) The three regimes of spatial recovery. Ecology, 100, e02586. https://doi.org/10.1002/ecy.2586
Elodie Vercken (2021) Unity makes strength: clustered extinctions have stronger, longer-lasting effects on metacommunities dynamics. Peer Community in Ecology, 100084. 10.24072/pci.ecology.100084
Revision round #214 Apr 2021
Decision round #2
Thank you for your revision and for addressing carefully the comments of the reviewers and myself. You have been doing a great job, and I reckon that most
I still have one remaining concern that needs to be addressed before I can recommend your manuscript. Although the core topic of your manuscript is related to the effect of spatial structure in metacommunities, I feel that the issue of space is not fully appreciated in its present version yet.
When analyzing a larger landscape, you find (i) no difference for metrics measured in unperturbed patches in comparison with smaller landscapes; (ii) for other metrics, a larger effect of the spatial distribution in larger landscapes, which is driven by a quantitative modification of metrics measured in clustered extinctions.
These two points highlight the fact that the position of a given patch relative to perturbed or unperturbed patches is crucial here, and should be acknowledged and discussed more explicitly.
Regarding the first point (i): you analyze only unperturbed patches that are adjacent to a perturbed patch, so the local environment of these patches is more or less the same regardless of landscape size. This actually looks like the equivalent of an edge effect. It is quite interesting, and might be worth commenting.
For the second point (ii): it does seem that the size of the cluster has a strong effect (no modification for dispersed extinctions, that do not change in larger landscapes), which is further supported by the analysis of recovery dynamics on Fig S18 and S19. The distance to an unperturbed patch (which is correlated to the size/depth of the perturbation cluster) has a clear effect on the recovery dynamics. I do believe that the use of quantitative indicators in the analysis of the simulations in large landscapes (not in addition to the clustered/dispersed factor, but in replacement of it) could give more insight on the general role of space here, outside of the specific clustered/dispersed modalities. Patches from all modalities could be pooled and analyzed with regard to their actual local environment. Multi-model comparison methods could be used to determine which indicator is most informative as different indicators are likely to be correlated (e.g., the distance to the closest perturbed/unperturbed patch or the number of perturbed/unperturbed patches in the neighbourhood).
I reckon that the discussion could be improved with a part addressing explicitly these issues related to a more quantitative interpretation of spatial structure, and would give the opportunity to make better use of Fig. S7-S8 and S18-S19, the latter not being called in the main text yet. Some parts of the discussion already address the issue of space (e.g. L. 435-439, 472-476, 548-551, 561-567, 591-586), but they are scattered over different paragraphs and could be brought together for more clarity.
Revision round #103 Feb 2021
Decision round #1
Dear Authors, I would like to apologize for the delay in getting back to you. I have been waiting for a 3rd review for more than 3 weeks now, so I decided to proceed with the first two reviews only, to avoid keeping you waiting any longer. Both reviewers were very positive about your work, which they found thorough and clearly written. The elegant link between experimental results and model simulations was particularly appreciated. I reckon that most of the comments we had should be easily addressed, but please pay particular attention to the suggestions concerning results presentation, the analysis of dynamical response variables, and the analysis (or at least discussion) of effects of temporal synchrony in perturbations. I will be most happy to recommend your manuscript once you have addressed these comments. Thank you for sharing your work with us. Best regards, Elodie Vercken. My own comments: - I wondered why you used the terms “spatial clumping” rather than “spatial autocorrelation” - The size of the landscape sets a strict limit to the range of spatial clumping that can be explored. I reckon you could use the model to investigate larger landscapes, to check whether the dominant influence of the spatial distribution of extinctions over extinction rates holds in less constraint configurations. - Also, I think it would be informative to compare the distributions of some functional indicators between experimental modalities (e.g. distance to the closest extinct patch, proportion of extinct patches in a 1-patch neighbourhood), as it would help to interpret the results (see for instance l. 464-465). Would it be possible to analyze the different response variables with such quantitative covariates, rather than “clumped vs dispersed”? - It seems that the results are sensitive to dispersal (l. 287-290); Maybe it would be interesting to run a sensitivity analysis relative to dispersal on the model outcomes. It might also interact with landscape size, as the influence of higher dispersal rates should be stronger in larger landscapes. - Figures : why is the modality “no extinction” represented on Fig 2 and not on Fig 1? (also, legend says that these represent diversity in extinct patches or extinct landscapes, so something is not right here). - Fig 3: why are only non-extinct patches adjacent to an extinct one included in the analysis? It does raise again the issue of landscape size, and the potential large-scale influence of extinction events. - I did not understand how you estimate the consistency between experimental results and the different modelling scenarios. For instance, L. 340-341, you state that the “competition-colonization trade-off” scenario is more consistent with experimental results, while based on the effect sizes on Fig 3a and Fig 4a, I would tend to say that the “empirical interactions” scenario is a better fit? **Additional requirements of the managing board**: We would like to receive your revision within 2 months. If you need more time, just tell us. As indicated in the 'How does it work?’ section and in the code of conduct, please make sure that: -Data are available to readers, either in the text or through an open data repository such as Zenodo (free), Dryad or some other institutional repository. Data must be reusable, thus metadata or accompanying text must carefully describe the data. -Details on quantitative analyses (e.g., data treatment and statistical scripts in R, bioinformatic pipeline scripts, etc.) and details concerning simulations (scripts, codes) are available to readers in the text, as appendices, or through an open data repository, such as Zenodo, Dryad or some other institutional repository. The scripts or codes must be carefully described so that they can be reused. -Details on experimental procedures are available to readers in the text or as appendices. -Authors have no financial conflict of interest relating to the article. The article must contain a "Conflict of interest disclosure" paragraph before the reference section containing this sentence: "The authors of this preprint declare that they have no financial conflict of interest with the content of this article." If appropriate, this disclosure may be completed by a sentence indicating that some of the authors are PCI recommenders: “XXX is one of the PCI XXX recommenders.”