Predicting the responses to perturbation in ecosystems has been an important and challenging task in ecology. In this aspect, the studies on ecological networks have largely advanced our understanding of responses to pulse perturbation. By contrast, the effects of network properties on the responses to press perturbation have received much less attention. The difficulties are two-fold: (1) the complex indirect interactions embedded in the responses to press perturbation, and (2) the lack of comparable food webs with interaction strength. The authors compiled a global dataset of food webs with consistent and reliable estimation of interaction strength. Based on this dataset, they applied matrix theories to estimate net interactions that take both direct and indirect interactions into account, which can be used to predict the network-level and species-level responses to press perturbation. This is the first work that has examined the net interactions in food webs on a global scale. The re-definition of species interactions in terms of net interactions is a vital complement to the traditional feeding relationships present in food webs. The associations of food web topological structures with net interaction strength and the importance of indirect effects can help target ecosystems and species that are more vulnerable to sustained perturbation in the application of conservation and biodiversity management.

Fig. 1. The net effects, net strengths, and proportion of sign changes for per biomass growth in food webs (i.e., net interaction matrix, T_I). (A) The boxplots show the variation of net effects across 169 food webs. The red line indicates values of zero. (B) Density distribution of absolute net interaction strength in each food web. The net strength was scaled by dividing the sum of net strengths in each web, and the density estimate was scaled to a maximum of one. (C)Density distribution of the proportions of sign changes from direct to net effects for positive, negative, and total direct interactions in all food webs. The three dashed lines that are highly overlapped indicate mean values.

Fig. 2.  The different proportions of sign changes from direct to net interaction effects for per capita growth in food webs (i.e., net interaction matrix, TI). (A) The sign changes from direct predator-prey (-+) interactions to net prey-predator (+-), mutualism (++), and competition (-) interaction effects. Note that the “-+ → -+” means no change of interaction sign, and “-+ →+-” means the predator and prey are reversed, i.e., from predator-prey to prey-predator. (B) The sign changes from no direct interactions (00) to net predator-prey (-+/+-), mutualism (++), and competition (-) interaction effects. Different lowercase letters indicate significant differences from pairwise comparisons (P < 0.05).