Widespread analytical pitfalls in empirical coexistence studies and a checklist for improving their statistical robustness

Abstract

1. Modern Coexistence Theory (MCT) offers a conceptually straightforward approach for connecting empirical observations with an elegant theoretical framework, gaining popularity rapidly over the past decade. However, beneath this surface-level simplicity lie various assumptions and subjective choices made during data analysis. These can lead researchers to draw qualitatively different conclusions from the same set of experiments. As the predictions of MCT studies are often treated as outcomes, and many readers and reviewers may not be familiar with the framework's assumptions, there is a particular risk of 'researcher degrees of freedom' inflating the confidence in results, thereby affecting reproducibility and predictive power. 2. To tackle these concerns, we introduce a checklist consisting of statistical best-practices to promote more robust empirical applications of MCT. Our recommendations are organised into four categories: presentation and sharing of raw data, testing model assumptions and fits, managing uncertainty associated with model coefficients, and incorporating this uncertainty into coexistence predictions. 3. We surveyed empirical MCT studies published over the past 15 years and discovered a high degree of variation in the level of statistical rigour and adherence to best practices. We present case studies to illustrate the dependence of results on seemingly innocuous choices among competition model structure and error distributions, which in some cases reversed the predicted coexistence outcomes. These results demonstrate how different analytical approaches can profoundly alter the interpretation of experimental results, underscoring the importance of carefully considering and thoroughly justifying each step taken in the analysis pathway. 4. Our checklist serves as a resource for authors and reviewers alike, providing guidance to strengthen the empirical foundation of empirical coexistence analyses. As the field of empirical MCT shifts from a descriptive, trailblazing phase to a stage of consolidation, we emphasise the need for caution when building upon the findings of earlier studies. To ensure that progress made in the field of ecological coexistence is based on robust and reliable evidence, it is crucial to subject our predictions, conclusions, and generalizability to a more rigorous assessment than is currently the trend.

Quantifying soil microbial effects on plant species coexistence: A conceptual synthesis

“…The rapidly developing toolkit for evaluating multispecies coexistence dynamics (e.g., Song et al, 2018) also provides a promising avenue for building toward a more comprehensive understanding of how microbes shape the dynamics of diverse plant communities. As studies scale up to include the experimental treatments necessary for parameterizing such models, ensuring that the observed patterns are consistent with model assumptions and that uncertainty in parameter estimates is properly propagated to long‐term conclusions should remain a priority (Terry and Armitage, 2023).…”

An experimental validation test of ecological coexistence theory to forecast extinction under rising temperatures

“…The selected population model for both species used a simplified Brière thermal performance function 36 and a Beverton-Holt competition model with temperature-dependent competition exerted by D. pallidifrons : where N is the number of individual females of species i in generation g, T g is the temperature in generation g, a, T min and T max are parameters of the thermal performance function defining the low-density growth rate, and α terms are competition coefficients. Following standard approaches 23,34,37 , coefficients from these fitted models were used to infer the population fitness differences and niche overlap at each temperature and categorise the invasion growth rate of each species. With increased temperature, the models predicted a shift from D. pallidifrons (the resident, highland species) excluding D. pandora (the lowland, invader species), through the zone of likely coexistence, to the exclusion of D. pallidifrons by D. pandora (Figure 2a).…”

“…Individual vials in the same treatment could be highly divergent, with some vials containing apparently stressed communities with late-developing pupae. Capturing this variation is a major, but underacknowledged 34 , challenge for applying coexistence theory. The best-fit models incorporated zero-inflated negative-binomial error distributions with temperature dependent parameters, a considerably more complex expression than is frequently assumed 34 .…”

“…Capturing this variation is a major, but underacknowledged 34 , challenge for applying coexistence theory. The best-fit models incorporated zero-inflated negative-binomial error distributions with temperature dependent parameters, a considerably more complex expression than is frequently assumed 34 . Modelling the system with approaches that did not have the flexibility to represent this variability gave different predictions.…”

Building modern coexistence theory from the ground up: The role of community assembly

“…permanence) to finding the equilibria of all subcommunities, which requires a simple way to find these equilibria. Additionally, the computed invasion graph only approximates the real invasion graph, as the model may misrepresent the natural community, measurement errors (Bowler et al, 2022) or wrong community model (Terry & Armitage, 2023). Empirically measuring the actual invasion graph maybe even more prohibitive, but may be possible for short-lived communities (Chang et al, 2023).…”