(A) Means and standard errors of the ovary sizes in the subsets of bees selected from the extreme tails of the ovariole number and GRS and distributions; LL = small ovaries (3–9 ovarioles) and low GRS (0–2), HH = large ovaries (17–29 ovarioles) and high GRS (4–7), n = 44 and 43, respectively. L and H were laboratory handling controls that were selected only on the basis of ovary size, GRS was not determined (n = 10; ovariole number was 3–9 and 12–23, respectively, H spanned lower ovary sizes than HH as there were not a sufficient number of bees to obtain in the 17–29 range). HH and LL differ significantly for ovariole number (P<0.0001, one-way ANOVA). (B) Means and standard errors of the log-transformed vitellogenin mRNA expression level given as a relative quantity (RQ). Bees with large ovaries and high GRS are characterized by significantly higher vitellogenin levels on average (P<0.005, one-way ANOVA). The controls show no significant effect of handling.