Date Published: June 11, 2019
Publisher: Public Library of Science
Author(s): Clare E. Aslan, Tina Heger.
Resilience theory aims to understand and predict ecosystem state changes resulting from disturbances. Non-native species are ubiquitous in ecological communities and integrated into many described ecological interaction networks, including mutualisms. By altering the fitness landscape and rewiring species interactions, such network invasion may carry important implications for ecosystem resistance and resilience under continued environmental change. Here, I hypothesize that the tendency of established non-native species to be generalists may make them more likely than natives to occupy central network roles and may link them to the resistance and resilience of the overall network. I use a quantitative research synthesis of 58 empirical pollination and seed dispersal networks, along with extinction simulations, to examine the roles of known non-natives in networks. I show that non-native species in networks enhance network redundancy and may thereby bolster the ecological resistance or functional persistence of ecosystems in the face of disturbance. At the same time, non-natives are unlikely to partner with specialist natives, thus failing to support the resilience of native species assemblages. Non-natives significantly exceed natives in network centrality, normalized degree, and Pollination Service Index. Networks containing non-natives exhibit lower connectance, more links on average, and higher generality and vulnerability than networks lacking non-natives. As environmental change progresses, specialists are particularly likely to be impacted, reducing species diversity in many communities and network types. This work implies that functional diversity may be retained but taxonomic diversity decline as non-native species become established in networks worldwide.
Global environmental change alters both the composition and dynamics of ecological communities, with the potential to disrupt or erode critical ecological functions [1,2]. According to resilience theory, ecosystems undergoing significant changes in species composition may enter alternative stable states, wherein they exhibit fundamental shifts in character with potential associated losses in ecosystem functions and services . Resistant systems can absorb substantial change without transitioning in state, and resilient systems can return to their original state after disruption .
I used standard network analysis and quantitative research synthesis, coupled with extinction simulations, to compare the network roles of native and non-native species in published, empirical networks. I restricted my analyses to pollination and seed dispersal networks because they have been described sufficiently to permit well-replicated, species-level analyses. I used only mutualistic networks that were published in their entirety or provided directly by authors and contained at least one known non-native species.
Species level analyses found that natives and non-natives differed significantly in measured species-level metrics, when network was employed as a blocking factor. Non-native species exceeded native species in normalized degree (t = 2.0053; p = 0.0450) (Fig 2a), consistent with the hypothesis that non-natives would interact with a higher proportion of possible partners than would natives. Non-natives also exhibited significantly higher betweenness (t = 3.6408; p = 0.0003) and closeness (t = 3.0762; p = 0.0021) (Fig 2a). Finally, PSI was significantly higher for non-native animals (0.24 ± 0.016 SE) than for native animals (0.19 ± 0.0048 SE) (t = 4.1368; p < 0.0001) (Fig 2a). The non-native species in the analyzed networks appear to strengthen redundancy by interacting with species already well-linked within the networks. This is evidenced by significant differences between native and non-native species in normalized degree, betweenness, closeness, and PSI, as well as higher mean links per species, generality, and vulnerability for full networks than native-only or control networks. Results also suggest that non-native species in these networks rarely partner with specialist species that are poorly-linked within the network, since random extinctions of natives more often remove the sole partner of remaining species than do extinctions of non-native species. In other words, loss of non-natives will most often remove redundant links from networks. Empirical networks in this study were less connected and more generalized than null networks, indicating that a small proportion of species in empirical networks are particularly well-linked and accompany a large number or tail of poorly linked species, relative to taxa in simulated null networks. Source: http://doi.org/10.1371/journal.pone.0217498