Optimizing sampling for bee communities using bee bowls

Native bees are important contributors to pollination in natural and agricultural landscapes. Understanding the forces driving the spatial distribution of native bee (Hymenoptera: Anthophila) communities may be beneficial for developing bee monitoring programs and for advancing agricultural practices. Geostatistical analysis is one method used to characterize spatial patterns. We sought to create and optimize a geostatistical sampling protocol for native bees by determining the distance required for spatial independence between bee bowls.  A spatial study was conducted in an open field by placing pairs of fluorescent blue and yellow bee bowls in an 18m × 18m grid with a spacing of two meters for a total of 100 sampling points. Bowls were sampled every 24h for four days. Data was analyzed in ArcGIS 9.3 to characterize spatial patterns and to determine spatial trends. Mean trap catch over the four day period was 2.5 ± 1.89 (SD) bees per trap. The correlogram in ArcGIS best fit with a nugget model indicating spatial randomness at a distance of two meters, that is, at a distance of two meters bee bowls are spatially independent. This indicates that a distance of ≥ 2m is suitable for use in bee sampling protocols to avoid competitive effects between bee bowls. Further geostatistical analysis using distances less than two meters and anisotropic semivariograms along with determining seasonal and site-specific effects will also be of value for optimizing bee sampling plans.