摘要：Plant-pollinator interactions are susceptible to different types of anthropogenic disturbances. Although it is assumed that different human activities can disrupt these interactions, the direction and magnitude of their consequences are mediated by a series of environmental and biological variables that may have opposite effects on different aspects of these mutualisms. Therefore, the impact of human-driven habitat alteration on pollination and plant reproductive success can not be predicted a priori. The sign and magnitude of this impact will also depend on disturbance intensity, spatial scale and frequency. The current approach to study disturbance effects on plant reproduction is mostly comparative and highly reductionist (e.g., seed production is compared between disturbed vs. undisturbed sites). This approach leads to a lack of understanding on the proximal factors involved in the reproductive response of plants to habitat alteration, or even to incomplete or erroneous interpretations when responses are curvilinear or there is a low level of true replication. Here I review and comment on the different types of limitations arising from this comparative approach, and propose an alternative one. The proposed approach is exemplified in a simple conceptual model that can accommodate different aspects of the disturbance itself, including scale considerations, as well as aspects of plants, pollinators, and of the pollination process. This approach, even though of a correlative nature, allows a more mechanistic understanding of the effects of habitat disturbance on plant pollination and reproductive success by testing different proximal variables and causal models using path analysis and structural equation modeling. Also, the proposed approach can incorporate easily the disturbance patterns found in the field, better reflected in gradients than discrete classes. Finally, I advocate a greater integration between the mostly disconnected areas of plant reproductive ecology and landscape ecology
