COS 141-7
Reduced specialization and modularity in an intimate mutualism diversifying on young oceanic islands
Recent work in ecological network structure has shown that disparate interactions can show similar topological properties. This raises two important problems in species interactions ecology. First, to what extent does interaction intimacy, the degree of biological integration between species, predict patterns of specialization in ecological networks? Second, how does network structure vary spatiotemporally in taxonomically related interactions? Prior work has suggested that high-intimacy mutualisms have high modularity and high reciprocal specialization compared to low-intimacy mutualisms. However, such high-intimacy assemblages often are characterized by low species richnesses (Nplants, Nanimals < 10), for which many network analysis methods are unsuitable. Here we use novel methods to examine network structure in the brood-pollination mutualism between leafflower trees (Phyllanthaceae: Glochidion) and leafflower moths (Lepidoptera: Epicephala) in continental (Japan) and oceanic island (Society Islands) assemblages. We sampled moths from 10 leafflower species on three of the Society Islands, determined species diversity and network structure using molecular and morphological approaches, and compared these to equivalent data from four sites in Japan from a previous study. We measured modularity using MODULAR, measured indirect effects using pathway proliferation, and measured reciprocal specialization using information entropy and multiplication of dependences of plant and animal taxa in the interaction matrix.
Results/Conclusions
We find more asymmetric regional assemblage richness in the Societies (10 leafflower spp. and 3 moth spp.) than in Japan (5 leafflower spp. and 5 moth spp.). One Societies Epicephala taxon appears to be a “super-generalist” which interacts with 9 out of 10 available hosts. All Japanese assemblages show high modularity and reciprocal specialization. Two Societies assemblages (Huahine and Tahiti) lack modularity while the third (Moorea) is modular. Pathway proliferation results show that structurally similar networks may have important differences in their potential for indirect effect propagation. For example, Tahiti and Huahine have a one-order-of-magnitude difference in the number of higher pathway lengths connecting pairs of species. These indicate that a) in continental regions, brood-pollination assemblages show similarly high modularity and reciprocal specialization as do ant-myrmecophyte assemblages, supporting the hypothesis that biological intimacy promotes these network properties; b) network properties and specialization can vary spatiotemporally within the same mutualism; c) “super-generalist” taxa may also exist in high-intimacy mutualisms; d) highly connected, non-modular small networks vary in their potential for the propagation of indirect effects, with potentially important consequences for convergent trait evolution. Finally, the novel methods presented here can be profitably applied to the topological analysis of other small networks.