Wild bee responses to cropland landscape complexity are temporally-variable and taxon-specific: Evidence from a highly replicated pseudo-experiment

Abstract

Wild bees may benefit from the restoration of natural areas in agricultural regions. The abundance and diversity of wild bee species responds to the amount of nesting and foraging habitat, but it is less clear how the distribution of these resources (e.g., the landscape complexity) may affect bees. We implemented a pseudo-experiment to disentangle the effects of three components of landscape complexity for wild bees in a Canadian Prairie cropland region. We used an algorithm to identify 146 sites that minimized correlations in indices of patch richness (i.e., the diversity in land cover types) and contagion (i.e., their degree of interspersion), and that collectively captured a cross-section of landscape contexts that differed in the relative proportion of cropland to other non-crop land covers. We trapped bees at these locations repeatedly over time (1119 unique collection events; equivalent to 10,471 trap-days over two consecutive years), identifying 22,493 bees of 213 taxa, in order to model trends for bees at different times of the season. We found that increasing patch richness may support a greater number of bee taxa, but individual bee taxa varied considerably in their response to components of landscape complexity. The effect on the total abundance of wild bees was temporally-variable, with the amount of cropland positively associated with abundance earlier in the season when mass-flowering crops are in bloom, and negatively later in the season when semi-natural areas are likely to provide the most forage. The response of bee abundance to contagion also varied temporally, and demonstrated a “humped” effect later in the growing season, suggesting there is an optimum in the complementary resources provided by adjacent habitat types. Our study shows that increasing the amount or diversity of non-crop land covers in this region is not likely to have a consistent effect for the majority of species across the season. We argue that modifying croplands to support wild bees is likely to be a complex task, requiring study of the functional responses to landscape of bee species present in the region, and their interactions with the phenological variability in resources.