The study of experimental communities is fundamental to the development of ecology. Yet, for most ecological systems, the number of experiments required to build, model or analyse the community vastly exceeds what is feasible using current methods. Here, we address this challenge by presenting a statistical approach that uses the results of a limited number of experiments to predict the outcomes (coexistence and species abundances) of all possible assemblages that can be formed from a given pool of species. Using three well-studied experimental systems—encompassing plants, protists, and algae with grazers—we show that this method predicts the results of unobserved experiments with high accuracy, while making no assumptions about the dynamics of the systems. These results demonstrate a fundamentally different way of building and quantifying experimental systems, requiring far fewer experiments than traditional study designs. By developing a scalable method for navigating large systems, this work provides an efficient approach to studying highly diverse experimental communities.
Subscribe to Journal
Get full journal access for 1 year
only $8.25 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Tax calculation will be finalised during checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
The data and code needed to replicate the central findings of this work are available at https://git.io/fjvON
Gause, G. F. The Struggle for Existence (Williams & Wilkins Company, 1934).
Yoshida, T., Jones, L. E., Ellner, S. P., Fussmann, G. F. & Hairston, N. G. Jr. Rapid evolution drives ecological dynamics in a predator–prey system. Nature 424, 303–306 (2003).
Costantino, R. F., Desharnais, R., Cushing, J. M. & Dennis, B. Chaotic dynamics in an insect population. Science 275, 389–391 (1997).
Dai, L., Vorselen, D., Korolev, K. S. & Gore, J. Generic indicators for loss of resilience before a tipping point leading to population collapse. Science 336, 1175–1177 (2012).
Cadotte, M. W. Experimental evidence that evolutionarily diverse assemblages result in higher productivity. Proc. Natl Acad. Sci. USA 110, 8996–9000 (2013).
Heemsbergen, D. A. et al. Biodiversity effects on soil processes explained by interspecific functional dissimilarity. Science 306, 1019–1020 (2004).
Friedman, J., Higgins, L. M. & Gore, J. Community structure follows simple assembly rules in microbial microcosms. Nat. Ecol. Evol. 1, 0109 (2017).
Dormann, C. F. & Roxburgh, S. H. Experimental evidence rejects pairwise modelling approach to coexistence in plant communities. Proc. R. Soc. B 272, 1279–1285 (2005).
Bell, T., Newman, J. A., Silverman, B. W., Turner, S. L. & Lilley, A. K. The contribution of species richness and composition to bacterial services. Nature 436, 1157–1160 (2005).
Tilman, D. et al. Diversity and productivity in a long-term grassland experiment. Science 294, 843–845 (2001).
Kuebbing, S. E., Classen, A. T., Sanders, N. J. & Simberloff, D. Above- and below-ground effects of plant diversity depend on species origin: an experimental test with multiple invaders. New Phytol. 208, 727–735 (2015).
Rakowski, C. & Cardinale, B. J. Herbivores control effects of algal species richness on community biomass and stability in a laboratory microcosm experiment. Oikos 125, 1627–1635 (2016).
Pennekamp, F. et al. Biodiversity increases and decreases ecosystem stability. Nature 563, 109–112 (2018).
Pennekamp, F. et al. Dynamic species classification of microorganisms across time, abiotic and biotic environments—a sliding window approach. PLoS ONE 12, e0176682 (2017).
Sun, G. Q. Mathematical modeling of population dynamics with Allee effect. Nonlinear Dyn. 85, 1–12 (2016).
Holland, J. N., Okuyama, T. & DeAngelis, D. L. Comment on ‘Asymmetric coevolutionary networks facilitate biodiversity maintenance’. Science 313, 1887–1887 (2006).
Fussmann, G. F. & Heber, G. Food web complexity and chaotic population dynamics. Ecol. Lett. 5, 394–401 (2002).
Bairey, E., Kelsic, E. D. & Kishony, R. High-order species interactions shape ecosystem diversity. Nat. Commun. 7, 12285 (2016).
Mayfield, M. M. & Stouffer, D. B. Higher-order interactions capture unexplained complexity in diverse communities. Nat. Ecol. Evol. 1, 0062 (2017).
Levine, J. M., Bascompte, J., Adler, P. B. & Allesina, S. Beyond pairwise mechanisms of species coexistence in complex communities. Nature 546, 56–64 (2017).
Carrara, F., Giometto, A., Seymour, M., Rinaldo, A. & Altermatt, F. Inferring species interactions in ecological communities: a comparison of methods at different levels of complexity. Methods Ecol. Evol. 6, 895–906 (2015).
Maynard, D. S. et al. Diversity begets diversity in competition for space. Nat. Ecol. Evol. 1, 0156 (2017).
Kraft, N. J. B., Godoy, O. & Levine, J. M. Plant functional traits and the multidimensional nature of species coexistence. Proc. Natl Acad. Sci. USA 112, 797–802 (2015).
Fox, J. W. The dynamics of top-down and bottom-up effects in food webs of varying prey diversity, composition, and productivity. Oikos 116, 189–200 (2007).
Crawford, K. M. & Knight, T. M. Competition overwhelms the positive plant–soil feedback generated by an invasive plant. Oecologia 183, 211–220 (2017).
Beveridge, O. S., Petchey, O. L. & Humphries, S. Direct and indirect effects of temperature on the population dynamics and ecosystem functioning of aquatic microbial ecosystems. J. Anim. Ecol. 79, 1324–1331 (2010).
Golberg, D. E. Neighborhood competition in an old-field plant community. Ecology 68, 1211–1223 (1987).
Sarnelle, O. & Wilson, A. E. Type III functional response in Daphnia. Ecology 89, 1723–1732 (2008).
Xiao, Y. et al. Mapping the ecological networks of microbial communities. Nat. Commun. 8, 2042 (2017).
Beisner, B., Haydon, D. & Cuddington, K. Alternative stable states in ecology. Front. Ecol. Environ. 1, 376–382 (2003).
Scheffer, M. & Carpenter, S. R. Catastrophic regime shifts in ecosystems: linking theory to observation. Trends Ecol. Evol. 18, 648–656 (2003).
Seekell, D. A., Cline, T. J., Carpenter, S. R. & Pace, M. L. Evidence of alternate attractors from a whole-ecosystem regime shift experiment. Theor. Ecol. 6, 385–394 (2013).
Steffen, T. Control Reconfiguration of Dynamical Systems: Linear Approaches and Structural Tests (Springer Science & Business Media, 2005).
Hoffman, M. D. & Gelman, A. The No-U-Turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo. J. Mach. Learn. Res. 15, 1593–1623 (2014).
Carpenter, B. et al. Stan: a probabilistic programming language. J. Stat. Softw. 76, 1–32 (2017).
We thank S. Kuebbing, C. Rakowski, F. Pennekamp and O. Petchey for making their data available and for comments on earlier drafts of this manuscript. We thank C. Serván, M. Pascual, J. Bergelson, E. Baskerville, G. Barabás and E. Friedlander for assistance and suggestions throughout this study.
The authors declare no competing interests.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
For each combination of the four competitors (c1 to c4, colours), we ran 96 simulations starting from different initial conditions. Each panel shows the location of the simulations’ endpoints (solid circles), as well as the true location of the equilibria for the system (crosses, computed analytically). Experiments resulting in a lack of coexistence are represented as half-points at the bottom of the graph. The predictions obtained using our method are reported using open symbols. There are two cases: we use circles for predictions of (locally) stable endpoints, and triangles for unstable ones; the stability is calculated under the assumption of Lotka-Volterra dynamics and equal growth rates. For this system, as expected, we recover a perfect fit for the positive densities. We can also correctly predict the lack of coexistence among triplets. We predict perfectly the position of the four-species equilibrium, and correctly classify it as unstable, despite having used growth rates that differ substantially from each other.
Colours and symbols are as in Fig. 17. Notable features of this system are: competitors might (c1-c2, c2-c3) or might not (c1-c3, c1-c4) coexist in pairs. However, in one case a feasible but unstable equilibrium exists (c1-c3, correctly predicted by our method), while in the other there is no feasible equilibrium (c1-c4, also correctly predicted). The system including c1-c3-c4 shows dependence on initial conditions (some trajectories collapse to another system, while others converge to equilibrium), signaling a locally (but not globally) stable equilibrium. The method correctly identifies the position of the 4-species equilibrium surrounded by the limit cycle. However, it suggests stability for the equilibrium, while it must be unstable to give rise to the stable limit cycle. The misclassification stems from the fact that in the calculation of stability, we consider growth rates to be equal (because we cannot infer growth rates from endpoints), while this is not the case here.
Extended Data Fig. 3 Simulation results for competition with Allee effects, in which competitors cannot grow when rare.
This system exhibits multistability in all cases (half-points at the bottom of each graph signal trajectories that resulted in extinctions). Despite the fact that the model contains cubic terms (while our method can deal only with quadratic terms), the in-fit is excellent, in all cases fitting the location of the endpoints perfectly. Experiments in which the species do not coexist are however misclassified—while there exist equilibria close to the prediction, they are unstable, rather than stable as predicted by our method.
Extended Data Fig. 4 Simulation results for a system of facultative mutualism between two classes of competitors.
Plants are denoted by p, and animals are denoted by a. Despite the non-linear functional response, the method predicts the location of the endpoints (all characterized by equilibrium dynamics) almost perfectly. The method also correctly predicts the lack of coexistence between the two plants (panel p1-p2).
Two resources and two consumers are simulated in all possible combinations, giving rise to cases of coexistence at equilibrium, stable limit cycles, or extinctions. In all cases, the proposed method predicts the location of the equilibria of the nonlinear system quite perfectly, making the correct inference for all cases in which species cannot coexist.
Despite the strong effect of HOIs, the recovered solution is close to all endpoints. Moreover, the method correctly predicts that coexistence between c1 and c2, or c1, c2 and c3 is precluded. The method however predicts coexistence between two triplets, despite simulations showing that either no feasible equilibrium exists, or it is unstable.
We simulated a 6-species GLV model, in which all 63 possible assemblages lead to coexistence. We measured abundances at these endpoints by adding noise, and producing five ‘replicates’. For each design, we use the specified number of assemblages to fit the model, and predict out-of-fit the abundance of all species at all other endpoints. Designs that produce qualitatively wrong predictions (that is, predicting a lack of coexistence for assemblages that do in fact coexist) are represented as vertical bars at the bottom of each boxplot. The horizontal dashed line marks the performance of the monoculture + leave-one-outs design, which fares among the best despite using only 12 assemblages to predict the remaining 51.
Extended Data Fig. 8 The quality-of-fit for the protist system at 15 °C as a function of the sampling day.
Rather than sample the community at days 15-17, as in the main text, we ‘ended’ the experiment at the indicated day, ±2 days, and fit the model with the corresponding endpoints. These results demonstrate that there is a clear initial period where the model fits poorly due to transient dynamics; followed by a stable period between days 10 and 20 where the approach performs well; followed by a period where the quality of fit starts to deteriorate as the species decline in abundances. The point in red denotes the point used in the main analysis, independently identified by quantifying when total biomass of the community stabilized.
About this article
Cite this article
Maynard, D.S., Miller, Z.R. & Allesina, S. Predicting coexistence in experimental ecological communities. Nat Ecol Evol 4, 91–100 (2020). https://doi.org/10.1038/s41559-019-1059-z
Observed Ecological Communities Are Formed by Species Combinations That Are among the Most Likely to Persist under Changing Environments
The American Naturalist (2021)
Nature Communications (2021)
Ecology Letters (2021)
Physical Review X (2021)