Spatial heterogeneity regulates plant-pollinator networks across multiple landscape scales

Abstract

Mutualistic plant-pollinator interactions play a key role in biodiversity conservation and ecosystem functioning. In a community, the combination of these interactions can generate emergent properties, e.g., robustness and resilience to disturbances such as fluctuations in populations and extinctions. Given that these systems are hierarchical and complex, environmental changes must have multiple levels of influence. In addition, changes in habitat quality and in the landscape structure are important threats to plants, pollinators and their interactions. However, despite the importance of these phenomena for the understanding of biological systems, as well as for conservation and management strategies, few studies have empirically evaluated these effects at the network level. Therefore, the objective of this study was to investigate the influence of local conditions and landscape structure at multiple scales on the characteristics of plant-pollinator networks. This study was conducted in agri-natural lands in Chapada Diamantina, Bahia, Brazil. Pollinators were collected in 27 sampling units distributed orthogonally along a gradient of proportion of agriculture and landscape diversity. The Akaike information criterion was used to select models that best fit the metrics for network characteristics, comparing four hypotheses represented by a set of a priori candidate models with specific combinations of the proportion of agriculture, the average shape of the landscape elements, the diversity of the landscape and the structure of local vegetation. The results indicate that a reduction of habitat quality and landscape heterogeneity can cause species loss and decrease of networks nestedness. These structural changes can reduce robustness and resilience of plant-pollinator networks what compromises the reproductive success of plants, the maintenance of biodiversity and the pollination service stability. We also discuss the possible explanations for these relationships and the implications for landscape planning in agricultural areas.

Fig 1

Left side: A—highlighted in green, dark gray and light gray, the state of Bahia, Brazil and South America, respectively; B—At the center of the state of Bahia (green), the studied region, with the lands of the agricultural partnership in red and Chapada Diamantina National Park in purple; C—example of the 3 km buffer used for the selection of sampling units; D—arrangement of the 27 selected sampling units (red dots) in the study region and the land cover classification used for the calculation of landscape metrics.

Fig 2

Relationship between the characteristics of networks (Y axis) and selected models (X axis): A—Number of links of networks; B—Network interaction strength asymmetry; C—network nestedness; LV—Local vegetation; PLD—Proximal landscape diversity; PLC—Proximal Landscape configuration; BLD—Broad landscape diversity

Fig 3. Scheme showing the qualitative relationship between landscape heterogeneity, vegetation heterogeneity and the networks structure.