Insect pollinator visitations to soybean crops, and soybean crop yield, along a latitudinal gradient in South America, 2020-2022

Fieldwork instrumentation used. No fieldwork instrumentation was used outside of GPS to position the sites’ locations.

Methods of collection. Five study areas were located along a N-S latitudinal gradient from southern Brazil to central Argentina. Within each area, five soybean sites were selected. At each site sampling areas were established at the field edge (50 m from field boundary) and the field interior (150 m from the field boundary).

Pollinator sampling: During flowering, we conducted repeated measurements of pollinator abundance and diversity in each soybean field by sampling along 100-m-long transects (1 m wide) placed on the field margin (0 m from field edge) and within the crop at 50 and 150 m from the field edge. Pollinator sampling along transects was undertaken by slowly walking the transect for 20 minutes and recording all insects seen visiting flowers. This was done on three occasions each day for an early (8:30-10:00h), mid (12-14:00h) and late (15:30-17:30) sample round. Pollinators were identified where possible to species or species aggregates in the field. Where this was not possible individuals were collected for subsequent lab-based identification ideally to species, or where this was not possible to generic or some other level of identification. If necessary, a morphospecies concept was applied allowing the identification of recognizable taxonomic units even where taxonomic resolution was poor. Where possible the name of flowering plants in the field margin that received visits from the pollinators were recorded. Weather data was recorded in the form of temperature, humidity and wind speed.

Pollination dependence of oilseed rape: At each sampling location in a field two (50 m and 150 m from the edge) sampling transects from the field edge were established (A and B). We assessed pollinator dependence using a caging experiment consisting of two treatments: (i) closed to pollinators by covering crops within 1m2 enclosed by cages of fine mesh during the flowering period, (ii) open pollination (i.e., plot of 1m2 without manipulation). We harvested 5 plants per treatment, each taken from in the middle of the cage. Following harvest, we measured: (i) Total number of pods on the plant; (ii) Average number of seeds per pod; (ii) Weight of the seeds (grams) before drying; iii) Total dried weight of seeds from the plant (grams). Additional information on crop variety, date of sowing and data of first observed flowers was recorded. Sampling was undertaken according to site in the harvest years of 2020, 2021 and 2022.

Nature and units of recorded variables. All data on abundance of pollinators represents counts of individuals. Data on fruit set represents counts of the number of fruits, buds and flowers.

Processing steps performed on the data. No data has been transformed with the exception of mean seeds per pod values.

Quality control/assessment applied to the data. All field workers were either experts with a long track record of pollinator identification or underwent extensive training. Questionable specimens were collected for subsequent identification under laboratory conditions. Data was collected in the field using pre-prepared data sheets. Data sheets were checked both visually and following data entry into digital format by confirming where outlier abundances (>2SD from mean) were found that these were consistent with raw field data sheets. Balances used to weigh seeds are regularly calibrated.

Limitations on the data reliability. For the seed set data there is no shaded control that allows access to pollinators, although a preliminary study suggested that this treatment did not differ from a non-shaded open to pollinators control.