1. Phenological overlap between crop flowering and pollinators is a crucial trait for the pollination of more than 75% of the world’s crops. However, crop management rarely considers the seasonal aspect of plant-pollinator mutualism. Here, we explore the phenological overlap between crops and pollinators and how it affects pollination and fruit production.
2. We measured the abundance and richness of native and non-native pollinators visiting raspberry flowers at two different times during the flowering season (i.e., early and late flowering periods) and examined their effects on crop yield in 16 fields.
3. The community of pollinators foraging on raspberry flowers was more diverse and dominated by native pollinators during the early flowering period when most native plants were in flower. Later in the season, when native flower resources in the environment declined, raspberry flowers were visited mainly by two non-native bees: managed honeybees and the invasive bumblebee Bombus terrestris.
4. Pollinator contribution to raspberry yield was twice as high in the early flowering period compared to the late period (61% vs. 31% increase in drupelet set, respectively). Flower damage caused by extremely high visitation frequency by non-native bees was also six times lower in the early than in the late flowering period (5% vs. 30% of damaged flowers, respectively).
5. Synthesis and applications: Providing sufficient pollen and nectar resources to support wild pollinators over extended periods in agricultural landscapes can contribute to crop pollination and ensure high fruit weight and quality. This can be achieved by restoring natural and semi-natural areas next to crop fields with native, long-flowering plant species. Additionally, growers and crop breeding programmes should consider selection of flowering time to coincide with the period of high diversity or abundance of native pollinators in order to reduce dependence on managed pollinators.

In each field, we studied raspberry flower visitors along 20, ~25m transects: 8 to estimate pollinator abundance (i.e., number of pollinators in 200 flowers) and 12 to estimate pollinator richness (i.e., number of pollinator species in 200 flowers) per flowering period (i.e., 20 transects x 2 periods = 40 transects per field in total) (Vaissière et al. 2011). Along each transect, we observed a total of 200 focal randomly-selected flowers while walking at a regular pace, and recorded the presence and identity of a flower visitor following the methodology proposed by Vaissière et al. (2011). Although these authors do not consider the time spent in each census as a variable to account for, we were careful not to spend more time surveying the 200 flowers in the transects in which we observed higher pollinator activity. Pollinators were censused under sunny and moderate or no windy conditions. Half of the transects were monitored in the morning, between 10:00 and13:00, and the second half in the afternoon, between 14:00 and 17:00. In each field, we surveyed flower visitors over two non-consecutive days within each flowering period (early and late).

We quantified reproductive success in terms of “drupelet set” (i.e., the proportion of ovules that set a drupelet) in naturally (open) pollinated and isolated plants (i.e., plants surrounded with fine mesh precluding pollinator visitation). Drupelet set is a close predictor of fruit weight. In each field, six to eight flowering stems were tagged and randomly assigned to one of two treatments: (a) open (i.e., 3 to 4 stems per field/period), or (b) isolated (i.e., 3 to 4 stems per field/period). Later, approximately four weeks after the survey of flower visitors, we randomly collected fruits from each tagged raspberry stem, totalling 2,668 fruits. Fruits were harvested close to maturity (a few days before they easily detached from the receptacle), transported to the laboratory in an electric cooler, and stored in a freezer until processing. We used a magnifying glass (20x) to count (1) the number of developed drupelets; (2) the number of undeveloped drupelets (i.e., presence of a dried stigma without a drupelet); and, in the case of fruits from the open-pollination treatment, (3) the presence/absence of flower damage, measured as the presence of holes in the vestigial sepals. Finally, each fruit was weighed using an electronic scale. The proportion of ovules that developed into a drupelet (i.e., drupelet set) is strongly correlated with fresh fruit weight and can therefore be used as a proxy of crop yield.