A bee's-eye view of landscape change: differences in diet of 2 Andrena species (Hymenoptera: Andrenidae) between 1943 and 2021

Abstract

Declines in pollinating insects have been linked to changes in land cover, affecting the availability of nesting sites and floral resources. Our study is the first analysis of changes in pollen load composition of 2 mining bees, Andrena barbilabris (Kirby) and Andrena flavipes (Panzer) (Hymenoptera: Andrenidae), at the same sites in central England, over 75 years. This provides a unique opportunity to remove spatial variation and review temporal changes in pollen diet within the context of landscape change. We analyzed modern-day pollen load composition for these species and compared it with historical data from the same sites. We then examined potential links between land-use change and the bees' diets. Both bees showed dietary flexibility and lower diet breadth for A. barbilabris, and the bees' foraging strategies appear to have changed. Andrena flavipes collected more pollen taxa in a single load, while A. barbilabris appeared to source pollen from greater distances. Landscape changes at the studied sites have affected the nutritional environment for these bees. Our findings are supported by an existing assessment of floral resources, which found floral diversity has decreased overall in both the habitats used by these bees. However, more research is needed on the nutritional content of pollens used by these bees, both now and historically, to estimate how pollen diversity has changed. The bee's-eye view underlines the importance of understanding how species respond to local changes so that effective conservation strategies can be developed.

Keywords: agriculture; biodiversity; pollen.

Landscape changes over 80 years at 2 UK sites, appear to be reflected in the pollen diet of 2 Andrena bees.

Fig. 1.

Changes in the use of different pollens for Andrena flavipes between 1945 (n = 130) and 2021 (n = 30). Counts are the number of pollen grains + 1 and log-transformed to base 10. The horizontal lines signify the median; the boxes show the interquartile range; the whiskers extend from the box to the minimum and maximum data point if these are within 1.5 times the interquartile range; other values are outliers and are plotted as dots. Significant differences in pollen load composition between time periods were identified using the Wilcoxon rank-sum test with Bonferroni’s correction for multiple comparisons.

Fig. 2.

Changes in the use of different pollens for Andrena barbilabris between 1943 and 1944 (n = 83) and 2021 (n = 30). Counts are the number of pollen grains + 1 and log-transformed to base 10. The horizontal lines signify the median; the boxes show the interquartile range; the whiskers extend from the box to the minimum and maximum data point if these are within 1.5 times the interquartile range; other values are outliers and are plotted as dots. Significant differences in pollen load composition between time periods were identified using the Wilcoxon rank-sum test with Bonferroni’s correction for multiple comparisons.

Fig. 3.

A) Satellite imagery (Google Maps) and B) Land cover classes (based on CEH 2021 land cover maps) of Andrena barbilabris nest site on Site 2, with buffers shown at 500 m, 600 m, 700 m, and 800 m. Areas of recent clear-fell and Castanea sativa trees were marked. Maps produced using QGIS 3.20.3 (2022).

Fig. 4.

A) Species accumulation curves of the pollen genera collected by Andrena flavipes between 1945 (n = 130) and 2021 (n = 30). B) Species accumulation curves of the pollen genera collected by Andrena barbilabris between 1943 and 1944 (n = 83) and 2021 (n = 30).

Fig. 5.

A) Differences in pollen load composition of Andrena flavipes between 1945 (n = 130) and 2021 (n = 30). Analysis of Permutational multivariate analysis of variance (PERMANOVA) results: F₁ = 11.29, P = 0.001. B) Differences in pollen load composition of Andrena barbilabris between 1943/1944 (n = 83) and 2021 (n = 30). Permutational multivariate analysis of variance (PERMANOVA) results: F₁ = 10.14, P = 0.001. Each point represents a single-pollen load from one bee, and these are color coded according to the sampling period. Ellipses show 95% CI.