Can plant-pollinator network metrics indicate environmental quality?

Abstract

Plant-pollinator interaction networks may be more informative than the diversity of species in the evaluation of the effects of environmental change. Considering that networks vary with the integrity of ecosystems, their changes may help to predict the consequences of anthropogenic impacts on biodiversity and ecological processes. This characteristic highlights its use as environmental quality indicator. However, to employ interaction networks as ecological indicators it is necessary to identify the most sensitive metrics and understand how and why they vary with environmental changes. This review aimed to identify, in empirical studies, which network metrics have been evidenced as being more sensitive to changes in environmental quality. We analyzed published empirical studies, that applied the network approach on environmental quality gradients. In addition to the network metric behavior, we studied the interactions between them and possible causes of their variation. The available empirical data indicated that degree, nestedness and connectance did not have a simple, linear or unidirectional response to habitat degradation. Conversely, the metrics interaction asymmetry, d' (reciprocal specialization index of the species) showed the most consistent responses to environmental change. The role of the species changed, ranging between generalists and specialists under different conditions. In addition, specialist species with morphological and behavioral constraints were lost in worse environmental quality situations. The identity of interacting species and their role in the network, with a further specification of groups and interactions most affected, are the properties with greater potential to indicate changes in environmental quality. Most of the available studies focused on metrics at the network level, but several studies and this review indicate that the patterns at the network level can be better understood in the light of metrics analyzed at the species level. Our results provide information that enrich the network analysis, highlighting the need to consider important features that are often neglected. Discussions and information compiled here are important for deciding how to look at empirical data and what to look for, as well as to indicate some caveats when interpreting data on plant-pollinator interactions with a complex network approach. Network metrics can be good indicators of environmental quality if the underlying ecological causes of the numerical changes are carefully analyzed.