Effect of nitrogen enrichment
Contrary to our expectations, we did not observe effects of nitrogen enrichment sources on crop pollinators and pollination. It is possible that the proxies used in our study do not adequately represent the real nitrogen exposure levels in our study fields. Indeed, while pollinators can be affected by local (i.e. within field) changes in nutrient availability (David et al. 2019), our proxies for nitrogen levels are taken at much broader scales. The amount of nitrogen that is in reality deposited in a specific location of the biosphere may not be well represented by the NO2 levels measured in the troposphere at much larger spatial resolution of the available data from the Sentinel-5 satellite (i.e., 7×3.5 km). Similarly, the estimated mean application rate of fertilizers at each study region (which is based on average application levels for each crop at country level, and do not consider personal decisions of landowners) may not be of a sufficient resolution to detect changes in fertilization rate and its effects at the local scale. For example, the presence of (semi-)natural habitats in the landscape will inevitably decrease the estimated average fertilizer application rate at 1km2 resolution, while a high proportion of highly enriched crops, such as cereals, maize or fertilised grasslands (e.g., ray grass) will tend to increase the estimated average application rate. Future works involving farmer interviews asking for the actual amount of fertilizer applied to better characterize nutrient availability would be important. Moreover, although we had a clear gradient of N fertilizer input across sites, all study sites were located in landscapes with a critical positive surplus of nitrogen inputs (that goes up to 20 t.km-2 for the year 2010) (European Environment Agency, 2020). Consequently, it is possible that throughout the study region pollinator communities are dominated by nitrophilous species (Carvalheiro et al. 2020) well adapted to high nitrogen conditions and the negative effects of nitrogen on pollinators and their contribution to crop production are no longer detectable in our specific study sites.
Finally, it is possible that functional composition of pollinators has changed along the nitrogen availability gradient but with no net change in pollinator abundance, or their contribution to crop production. Indeed, N enrichment can have contrasting effects on pollinator species. Pollinators with more diversified diets might be less affected by landscape eutrophication potentially due to their ability to forage on a higher diversity of flowers in a diverse set of habitats (Pöyry et al. 2017; Carvalheiro et al. 2020). N deposition that changes soil nutrient availability is an important driver of plant species composition change and result in the decline of oligotrophic plant species, such as nitrogen fixing Fabaceae species (Roth et al.2013, 2019). Fabaceae are the main food resource of most bumble bee species and many other solitary bees (Goulson et al. 2008; Kleijn & Raemakers 2008; Connop et al. 2010). Thus, species specialised on Fabaceae (and other N sensitive plants), can have more difficulty in finding adequate resources be more susceptible to the effects of N enrichment than other pollinator species (Stevens et al. 2018). But if, for the crops studied here, species that prefer nitrophilous environments (see Carvalheiro et al. 2020) are equally efficient for crop pollination than species which are negatively affected, pollinator community compositional changes would not affect the net crop pollination outcome.