Search

Models of natural processes necessarily sacrifice some realism for the sake of tractability. Detailed, parameter‐rich models often provide accurate estimates of system behaviour but can be data‐hungry and difficult to operationalize. Moreover, complexity increases the danger of 'over‐fitting', which leads to poor performance when models are applied to novel conditions. This challenge is typically described in terms of a trade‐off between bias and variance (i.e. low accuracy vs. low precision).In studies of ecological communities, this trade‐off often leads to an argument about the level of detail needed to describe interactions among species. Here, we used data from a grassland biodiversity experiment containing nine locally abundant plant species (the Jena 'dominance experiment') to parameterize models representing six increasingly complex hypotheses about interactions. For each model, we calculated goodness‐of‐fit across different subsets of the data based on sown species richness levels, and tested how performance changed depending on whether or not the same data were used to parameterize and test the model (i.e. within vs. out‐of‐sample), and whether the range of diversity treatments being predicted fell inside or outside of the range used for parameterization.As expected, goodness‐of‐fit improved as a function of model complexity for all within‐sample tests. In contrast, the best out‐of‐sample performance generally resulted from models of intermediate complexity (i.e. with only two interaction coefficients per species—an intraspecific effect and a single pooled interspecific effect), especially for predictions that fell outside the range of diversity treatments used for parameterization. In accordance with other studies, our results also demonstrate that commonly used selection methods based on AIC of models fitted to the full dataset correspond more closely to within‐sample than out‐of‐sample performance.Synthesis. Our results demonstrate that models which include only general intra and interspecific interaction coefficients can be sufficient for estimating species‐level abundances across a wide range of contexts and may provide better out‐of‐sample performance than do more complex models. These findings serve as a reminder that simpler models may often provide a better trade‐off between bias and variance in ecological systems, particularly when applying models beyond the conditions used to parameterize them. [ABSTRACT FROM AUTHOR]

"Old fields" are ecosystems that have been previously managed and subsequently abandoned, usually from agricultural use. These systems are classic testing grounds for hypotheses about community assembly. However, old field succession can be difficult to predict: seemingly similar fields often diverge in terms of species composition and environmental conditions.Here, we test the relative roles of contingency and stochasticity in driving vegetative successional dynamics. We draw on three decades of surveys in 24 old fields at the Cedar Creek Ecosystem Science Reserve (Minnesota, USA), and focus on five drivers that are known to shape local plant communities: soil fertility, fire, climate, competition, and demography. These drivers can contribute to contingency when they act consistently across fields and years (e.g., soil nitrogen accumulation, experimental fire regimes, or average climate), or to stochasticity when their effects are variable (e.g., annual variations in weather, or colonization and mortality events).We proceed in two steps. First, we fit regressions estimating abundance, colonization, and mortality for eight major functional groups in relation to these five drivers. We then use these regressions to parameterize a series of metacommunity simulation models, and test whether observed levels of stochasticity and variation in the drivers are sufficient to explain successional divergence.All drivers were significantly associated with plant species abundances, colonization, and mortality. Contingent factors strongly altered predicted successional trajectories. However, replicate simulations with similar conditions followed similar successional trajectories, suggesting that stochastic processes did not lead to divergence. This robustness of successional dynamics may be explained by compensatory trade‐offs. For example, species that were abundant late in succession typically suffered from low colonization rates and high mortality rates early in succession.Synthesis. Average successional dynamics among old fields at Cedar Creek follow largely consistent trends. Though dynamics of individual fields vary, much of this variation can be explained by contingent factors. Stochastic processes appear not to be sufficiently strong to create divergent successional trajectories among fields with similar sets of drivers. Our results therefore suggest that divergence among successional trajectories in chronosequences may be the result of predictable contingent factors, rather than unpredictable stochastic fluctuations. Average successional dynamics among old fields at Cedar Creek follow largely consistent trends. Though dynamics of individual fields vary, much of this variation can be explained by contingent factors. For example, regular burning (left half of picture) can lead to reduced leaf litter and changes in plant species abundances relative to unburned fields (right half of picture). In contrast, stochastic processes appear not to be sufficiently strong to create divergent successional trajectories among fields with similar sets of drivers. Our results suggest that divergence among successional trajectories in chronosequences may be the result of predictable contingent factors, rather than unpredictable stochastic fluctuations. Photo credit: Adam Clark, April 2017. Field 5 at Cedar Creek. [ABSTRACT FROM AUTHOR]

Abstract Because the Lotka–Volterra competitive equations posit no specific competitive mechanisms, they are exceedingly general, and can theoretically approximate any underlying mechanism of competition near equilibrium. In practice, however, these models rarely generate accurate predictions in diverse communities. We propose that this difference between theory and practice may be caused by how uncertainty propagates through Lotka–Volterra systems. In approximating mechanistic relationships with Lotka–Volterra models, associations among parameters are lost, and small variation can correspond to large and unrealistic changes in predictions. We demonstrate that constraining Lotka–Volterra models using correlations among parameters expected from hypothesized underlying mechanisms can reintroduce some of the underlying structure imposed by those mechanisms, thereby improving model predictions by both reducing bias and increasing precision. Our results suggest that this hybrid approach may combine some of the generality of phenomenological models with the broader applicability and meaningful interpretability of mechanistic approaches. These methods could be useful in poorly understood systems for identifying important coexistence mechanisms, or for making more accurate predictions. Highlights • Lotka–Volterra models are commonly used to study interspecific interactions. • However, these models can fail when interaction strengths among species are uncertain. • We show that underlying mechanisms constrain relationships among model parameters. • We approximate these constraints by deriving expected covariance among parameters. • We find that covariance significantly improves predictions in diverse communities. [ABSTRACT FROM AUTHOR]