Plants need pollinators or seed dispersers to thrive. Similarly, pollinators and seed dispersers depend on plants for their survival. Recent work suggested that the particular way plants and animals build such mutualistic interactions safeguards them against environmental disturbances and increases the stability of their communities. At the same time, however, the unprecedented rapid environmental change that these communities are experiencing challenges their long-term survival. Therefore, it is necessary to develop ways of measuring the resilience of mutualistic communities that may help us to prevent them from potentially collapsing. Here, we show how we can detect tipping points in mutualistic communities reconstructed from empirical plant-pollinator networks. We simulated scenarios of gradual environmental change that led to species loss and community collapse. Prior to collapse, we monitored abundances of all species and estimated variance, autocorrelation, and skewness as indicators of community resilience. We tested how the structure of the community affected our ability to detect transitions, as well as whether we could identify flagship indicator species that best signal community-wide transitions.
Results/Conclusions
We found that variance and autocorrelation in species dynamics increased prior to community collapse both when measured at species and community levels. In contrast, skewness showed no particular trend. The variation in nestedness measured across our communities did not have a significant infuence on the performance of the indicators, but it had an effect on the identity of species lost at the progressive deterioration of the environment. Estimating indicators of resilience based on low abundant species was more informative than dominant species, while the characteristics of the species identified as flagship indicator species depended on overall community properties. Our results show, for the first time, that generic indicators of resilience can be used to detect tipping points in complex ecological networks. This may offer novel insight for the design of monitoring and conservation policies for anticipating unexpected events in a variety of ecological networks, ranging from metapopulations and host-parasite communities to food webs.