Spatial variation in ant–tree bipartite networks is driven by a bottom‐up process

Abstract

Spatial shifts in the organisation and pattern of ecological networks may be driven by species composition turnover, changes in species abundance and/or abiotic effects on species interactions. To test the above assumptions, we collected arboreal ants and recorded host tree species from five latitudinal sites in Xishuangbanna, China. We constructed bipartite networks and compared spatial variation in species composition, interactions and stability change. We also analysed the relationship between network structure metrics and network stability. Results showed that ant composition differed significantly among the five sites, primarily determined by weaver ants (Oecophylla smaragdina) and Crematogaster ant species. Dominant tree species included Ficus (32 individuals), Hevea (14 individuals), Castanopsis (11 individuals) and Pittosporopsis (8 individuals). Ant richness was positively correlated with tree richness, suggesting that host trees support ant diversity as a fundamental environment. The ant–tree bipartite networks showed low connectance, high specificity and low niche overlap. Links per species and network node were positively correlated with network stability, whereas connectance, nestedness, specificity and niche overlap were not. Ants and trees formed a nonspecific and sparse bipartite network, and the number of nodes and links per species promoted network stability. Because plant richness was positively related to ant richness, and because arboreal ants feed predominantly on extrafloral nectaries, seeds, insects and honeydew secreted from hemipteran insects, we believe the bottom-up process plays a central role in determining ant–tree interaction networks.