Invasive mutualists erode native pollination webs

Abstract

Plant-animal mutualisms are characterized by weak or asymmetric mutual dependences between interacting species, a feature that could increase community stability. If invasive species integrate into mutualistic webs, they may alter web structure, with consequences for species persistence. However, the effect of alien mutualists on the architecture of plant-pollinator webs remains largely unexplored. We analyzed the extent of mutual dependency between interacting species, as a measure of mutualism strength, and the connectivity of 10 paired plant-pollinator webs, eight from forests of the southern Andes and two from oceanic islands, with different incidences of alien species. Highly invaded webs exhibited weaker mutualism than less-invaded webs. This potential increase in network stability was the result of a disproportionate increase in the importance and participation of alien species in the most asymmetric interactions. The integration of alien mutualists did not alter overall network connectivity, but links were transferred from generalist native species to super-generalist alien species during invasion. Therefore, connectivity among native species declined in highly invaded webs. These modifications in the structure of pollination webs, due to dominance of alien mutualists, can leave many native species subject to novel ecological and evolutionary dynamics.

Figure 1. Relation of Mutualism Strength to Invasion Index for Ten Pollination Networks

Mutualism strength was estimated by averaging the product of the mutual dependences over (A) all pairs of interacting species sampled in each web and (B) only pairs of interacting native species. The invasion index was the average proportion of the total interaction frequency represented by alien plants and alien pollinators. For each web, mutualism strength was standardized by the mean of the distribution of values generated by random re-distribution of observed interaction frequencies (see Materials and Methods). In each panel, symbols depict observations for the five pairs of webs analyzed (, Challhuaco; ▴, Cerro Otto; ▾, Llao-llao; •, Puerto Blest; □, Oceanic islands). Solid lines indicate the best linear fits (y = −0.18 – 0.42x, r ²= 0.427, F = 5.95, df = 1, 8, p = 0.041 for (A), and y = −0.34 + 0.27x, r ²= 0.250, F = 2.67, df = 1, 8, p = 0.14 for (B)). N = 10 pollination webs. Dotted line segments join points representing webs from paired disturbed and undisturbed areas in the south-Andean forest region or the two webs from oceanic islands. The gray zone is the region delimited by the 2.5 and 97.5 percentiles from the standardized random distributions.

Figure 2. Means and Composite Frequency Distributions of Interaction Asymmetry for Pairs of Interacting Species

The graphs depict asymmetries for interactions recorded in lightly invaded and highly invaded webs. Gray dots (upper panels) and bars (lower panels) indicate asymmetries for pairs of interacting native species, and black dots and bars depict pairs of interacting species in which at least one is alien. Sample sizes for each mean and frequency distribution can be derived from the column “Number of links” in Table 1. Error bars represent one standard error.

Figure 3. Relation of a Species' Strength to Its Degree

A species' strength is the sum of the dependences of the species with which it interacts, whereas its degree is the number of species with which it interacts. The figure depicts separate relations for native (gray circles) and alien (black circles) plant and pollinators species present in lightly and highly invaded webs. The dashed and continuous lines represent the best-fit linear regressions for native and alien species, respectively (p < 0.001 in all cases). Sample sizes for natives and aliens are, respectively, 36 and 15 for plants in lightly invaded webs, 37 and 34 for plants in highly invaded webs, 113 and 15 for pollinators in lightly invaded webs, and 162 and 19 for pollinators in highly invaded webs.

Figure 4. Relation of Connectance to the Number of Species

Connectance represents the percentage of possible interactions that were actually recorded. The solid line indicates the best-fit negative exponential regression line (y = 40.9 × e ^–0.016x, F = 221.6, df = 2, 8, p < 0.0001). N = 10 pollination webs. Symbols as in Figure 1.

Figure 5. Relation of the Percent Difference in Connectance to Invasion Index

Percent difference represents the connectance of the native plant-pollinator sub-web (C_N) relative to the entire web (C_C) and was estimated as 100 · . Negative and positive connectance differences indicate that native sub-webs are less or more connected than their respective entire webs. We compared the connectances of the native sub-web and the sub-web formed by all other interactions by means of χ ² test with df = 1. Asterisks indicate significant (p < 0.05) deviations. The solid line represents the best linear fit (y = 21.2 – 61.6x, r ²= 0.300, F = 3.41, df = 1, 8, p = 0.10). N = 10 pollination webs. Symbols as in Figure 1.