Does the disruption of the keystone interaction by non-native ungulates produce cascading effects on the community?
Network complexity - We found differences on network complexity in pollination networks (F 1,7 = 5.032, P = 0.029) but only marginal in seed dispersal networks (F 1,7 = 3.849, P = 0.087) between intact forest sites with the keystone interaction and sites invaded by non-native ungulates. The complexity on both pollination and seed dispersal networks was lower in invaded sites than in sites with the keystone interaction, exhibiting 23% and 59% fewer motifs respectively, and hence, fewer direct and indirect ecological interactions supporting the community (Fig. 4a and 4c). In addition, both pollination and seed dispersal networks exhibit 12% and 11% fewer frequency of largest motifs integrated by six species in invaded sites than in sites with the keystone interaction (Fig 4b and 4d).
Functional redundancy among generalist species - The number of pollinator and plant species that occupied generalist positions was twice as high in intact forest sites with the keystone interaction than in invaded sites (Fig. 5a; Table S3). Moreover, in presence of the keystone interaction a great number of pollinator species occupied the most generalist position providing more network stability. On the other hand, the low functional redundancy among generalist species found in invaded sites suggests low stability in pollinator networks. This analysis was not possible in seed dispersal networks because of the reduced number of frugivorous species.
Ecological importance of the keystone interaction members - We found differences in the ecological importance of the members of the keystone interaction between intact forest sites with the keystone interaction and invaded sites (Fig. 5b). In sites with the keystone interaction, the number of positions occupied by the mistletoe was at least three times higher than in invaded sites (Z 1,7 = -2.998, P = 0.002), suggesting a greater ecological importance of the mistletoe in the pollination network. In addition, we caught no individuals of D. gliroides in any field season in the invaded sites, which indicates that the marsupial may be ecologically extinct in these sites (Rodriguez-Cabal & Branch 2011). Consequently, we were unable to assess statistically the ecological importance of the marsupial between treatments. Furthermore, we found no differences in the ecological importance of the hummingbird between treatments (Z 1,7 = -0.715, P = 0.475).