Cherry flower visitors and cherry pollination

Fieldwork was carried out in spring 2020 in orchards within the federal states Sachsen-Anhalt and Thuringia, Germany (Figure 1). Both federal states are dominated by agriculture (>60% of land cover). We selected 17 sites, of which two were experimental orchards and 15 were commercial mixed fruit orchards (Table S.1). The experimental orchards are owned by research institutions for conducting research experiments but are managed as commercial orchards in order to simulate agriculturally field-realistic conditions. The size of the orchards devoted to sweet cherry (Prunus avium) cultivation ranged between 0.2 ha and 36 ha (6.6 8.6; mean SD). Distances between field sites were >2 km (47.9 28.9 km; mean SD) to ensure spatial independence.
Pollination management consisted of providing honey bee hives and nesting material for mason bees (Osmia cornuta). Though Osmia bicornis is common in the study region, its flight period does commence until after cherry bloom has ceased and therefore, we did not consider it further in this study. Commercial bumble bee nests were not used by cherry farmers in this study. To investigate the effect of bee management on bee abundance and subsequently on sweet cherry fruit set, we selected sites with varying pollinator management. Mason bee management ranged from zero (i.e., no nesting material provided) to 8.6 nesting boxes per ha. On-farm bee management was decided prior to the study by the farmer and therefore the size of the nesting boxes varied across sites. To make them comparable, we set the standard size of a nesting material box to be 100 x 54 cm containing approximately 500 tubes made of paper, bamboo or drilled wood, each of a length of 13-16 cm and a diameter of 4-10 mm. Farmers initiated mason bee management in different years, ranging from 2002 to 2020 (General_info_orchards). In the two sites at which farmers started providing nesting material in the same year as the study (2020), Osmia cornuta cocoons were provided in addition to the nesting material to ensure similar conditions (General_info_orchards). The number of honey bee colonies per site (General_info_orchards) varied from zero to 20 hives per ha.

We quantified the abundance of insect flower-visitors during the peak bloom of sweet cherry (16.04.2020 23.04.2020) at each site for one day. For each site, two transect walks of 90 minutes each (in total 180 minutes) were performed alongside cherry trees on a sunny day, one in the morning and one in the afternoon on the same day. Transect walks were performed >50 m from the edge of the orchard and included trees of the cultivar Regina used for a pollination experiment (described below). One observer continuously walked alongside the Regina cultivar, while noting all visible flower visitors. Flower visitors that touched the reproductive parts of a cherry flower were counted and identified to morphological groups: honey bees, bumble bees, mason bees (Osmia cornuta), other wild bees, butterflies, flies, beetles, and ants. Though flower-visitor observations performed only for one day might be a limiting factor, assessments in cherries with a very limited flowering period of around 5-8 days restrict observations over several days in multiple orchards. Ambient (shade) temperatures during the transect walks were recorded with a digital thermometer; ranging from 9C to 22C. Data can be found in Abundance_Transectwalk_Cherry.

In order to quantify provision of pollination services, we studied one of the most common sweet cherry cultivars, Regina, which was present in all 17 study sites. This cultivar is self-sterile (S-alleles: S1S3) and requires cross-pollination for successful fruit development (Lech et al., 2008; Holzschuh et al., 2012). We chose at least one row of Regina trees in each orchard, which was planted either next to a cross-compatible pollinizer cultivar or which was interspersed with a pollinizer cultivar in the same row. Pollinizer cultivars varied across sites (General_info_orchards). We selected 20 Regina trees in each orchard that were at least 50 m from the orchard edge. On each tree, we applied three flower treatments: insect exclusion (bagged: B), hand pollination (hand: H), and open insect pollination (open: O). For each treatment, we chose a flower bundle, which we marked with coloured ribbons, cord, and barrier tape to later locate the treatments. In some cases, several flower bundles were used for one treatment if one flower bundle contained less than three flowers. During full bloom and on the same day as observations of flower visitors, we counted all open and receptive flowers per bundle and removed old (over-flowered) or young (still closed) flowers. The insect exclusion treatment (B) was bagged in fine netting (1 mm PVC mesh) prior to cherry bloom (6.04.2020 12.04.2020) to prevent insect pollination and remained bagged throughout cherry bloom; it enabled us to estimate the contribution of wind pollination and autonomous self-pollination to fruit set. Flower bundles of treatment H were manually pollinated with pollen from at least two flowers of an adjacent pollinizer (see Table S.1 for pollinizer cultivars per site) as a measure of maximal fruit set. Treatment O remained unmanipulated as a measure of current pollination service provision at each fruit orchard accessible to all flower-visiting insect including honey bees and mason bees.

Sweet cherry fruit set was counted three times, once in May to assess initial fruit set (ca. four weeks after flowering), once in June after the so-called June fall (ca. eight weeks after flowering), and once prior to harvest to assess final fruit set, approximately in the beginning of July (ca. 12 weeks after the experimental manipulations). We divided the number of developed fruits per bundle by the number of flowers per bundle for each fruit count period (i.e., May, June, July) to give the percentage fruit set per treatment. In addition, we recorded the weight of each cherry fruit during the final fruit set count in July. Data can be found in Pollination_experiment_cherry.

For all data entries entered as NA, data was not possible to collect.