Resilience of dynamic ecological networks with multiple interaction types

Background/Question/Methods

The relationship between complexity and stability is not only a fundamental ecological debate but also an important and current multidisciplinary issue. The network paradigm is ideally suited to inquiry in this area and continues to be widely used to probe the dynamics of complex systems in nature and society. Most ecological network research, however, has so far focused on particular interaction types, dynamics of highly simplified systems or empirical energy flows applicable only to specific systems. Few have tried to overcome these limitations concurrently by constructing realistically complex systems using simple rules and examining their emergent properties under a general theoretical framework. In this study, we first designed a simulation model combining ecological network topology, bioenergetics and population dynamics using fundamental ecological principles and mechanistic equations. We then used the model to investigate how the addition of different types of non-predator-prey interactions (competition, mutualism, parasitism) to food webs affect their resilience. Resilience was quantified in terms of the mean time taken to recover half the energy lost in perturbations. We accounted for parameter uncertainty via Monte Carlo simulations.

Results/Conclusions

Linear mixed-effects regression, with presence or absence of interaction types as dummy variables and model parameterization as a random effect, associated mutualism with shorter recovery time and parasitism with longer recovery time (n = 212, p << 0.0001). Direct competition among basal species was also associated with longer recovery time but the effect was less significant (p = 0.004). Mutualism had the aforementioned effect regardless of the presence or absence of parasitism. A separate simulation experiment, keeping connectance constant across model configurations with and without mutualism and the number of species constant across configurations with and without parasitism, produced qualitatively identical results for mutualism and parasitism (n = 2102, p << 0.0001). The effects are present despite the large variation across simulated ecosystems that reflects the variety of real ecosystems. We discuss possible ecological mechanisms by which non-predator-prey interactions can affect ecosystem resilience as evidenced by the simulations. Our findings illustrate the potential effects of non-predator-prey interactions and support the cause for accounting for these interactions in investigations of ecological communities. We advocate systems approaches as part of our arsenal of tools for better understanding and anticipation of the trajectories of complex ecosystems under rapid anthropogenic change.