New indices for rapid assessment of pollination services based on crop yield data: France as a case study

Local studies indicate that animal pollination is essential for crop productivity, but its effectiveness and temporal stability vary broadly across regions. However, there is no simple and rapid method to assess the pollination services over large areas. Here, we introduce two new indices to measure pollination services and its temporal stability in farmland, the Pollination Services Index and the index of Temporal Variation in Pollination Services, that only require readily available data, namely crop yield and crop pollinator dependence. The philosophy of these indices is to compare at a given site the standardized yield, or the temporal variation in yield, among crops that have different levels of dependence on animal pollinators for their production. We expect that where there is a shortage of pollinators, standardized crop yields should decrease, and temporal variation in crop yields should increase, with increasing dependence on pollinators. The Pollination Services Index in a given area is thus defined as the slope of the linear regression between standardized crop yield and crop pollinator dependence; the Temporal Variation in Pollination Services is defined as the slope of the linear regression between the inter-annual coefficient of variation of crop yield and crop pollinator dependence. We calculated these two indices in France, where we show extensive spatial variation in the estimated pollination services. We further show that this variation in the Pollination Services Index is negatively correlated with the index of Temporal Variation in Pollination Services, and positively correlated with habitat quality for pollinators. At a spatial resolution of administrative departments (ca. 5,800 km2), the results show that intensive farming is positively correlated with higher Temporal Variation in Pollination Services but is not significantly associated with the Pollination Services Index, probably due to multicollinearity issues. Despite some limitations that deserve further attention, these indices may constitute promising, cost-effective tools to highlight regions of pollination deficit over large areas.