Costs of reproduction are present but latent in eusocial bumblebee queens

Abstract

Background: The standard evolutionary theory of ageing proposes that ageing occurs because of a trade-off between reproduction and longevity. Eusocial insect queens exhibit positive fecundity-longevity associations and so have been suggested to be counter-examples through not expressing costs of reproduction and through remodelling conserved genetic and endocrine networks regulating ageing and reproduction. If so, eusocial evolution from solitary ancestors with negative fecundity-longevity associations must have involved a stage at which costs of reproduction were suppressed and fecundity and longevity became positively associated. Using the bumblebee (Bombus terrestris), we experimentally tested whether queens in annual eusocial insects at an intermediate level of eusocial complexity experience costs of reproduction, and, using mRNA-seq, the extent to which they exhibit a remodelling of relevant genetic and endocrine networks. Specifically, we tested whether costs of reproduction are present but latent, or whether a remodelling of relevant genetic and endocrine networks has already occurred allowing queens to reproduce without costs.

Results: We experimentally increased queens' costs of reproduction by removing their eggs, which caused queens to increase their egg-laying rate. Treatment queens had significantly reduced longevity relative to control queens whose egg-laying rate was not increased. Reduced longevity in treatment queens was not caused by increased worker-to-queen aggression or by increased overall activity in queens. In addition, treatment and control queens differed in age-related gene expression based on mRNA-seq in both their overall expression profiles and the expression of ageing-related genes. Remarkably, these differences appeared to occur principally with respect to relative age, not chronological age.

Conclusions: This study represents the first simultaneously phenotypic and transcriptomic experimental test for a longevity cost of reproduction in eusocial insect queens. The results support the occurrence of costs of reproduction in annual eusocial insects of intermediate social complexity and suggest that reproductive costs are present but latent in queens of such species, i.e. that these queens exhibit condition-dependent positive fecundity-longevity associations. They also raise the possibility that a partial remodelling of genetic and endocrine networks underpinning ageing may have occurred in intermediately eusocial species such that, in unmanipulated conditions, age-related gene expression depends more on chronological than relative age.

Keywords: Age-related gene expression; Bombus terrestris; Costs of reproduction; Eusociality; Evolutionary theory of ageing; mRNA-seq.

Fig. 1

Outline of experimental design to test for costs of reproduction in Bombus terrestris queens. a Experimental treatments: R: Removal queens: eggs removed and counted; C: Control queens: eggs removed, counted, and replaced. Day 1 was the first day that new egg cells were observed in all colonies. b RNA sample collection strategy. G1, G2: random subgroups of queens within treatments. TP1, TP2: time-points for sampling of queens for RNA extraction (R queens: TP1 on day 37, TP2 on day 85; C queens: TP1 on day 89, TP2 on day 134). TP1G, TP2G: subsets of queens sampled for RNA extraction at TP1 and TP2, respectively. One R queen (Q1) was not assigned to G1 or G2, and was used to provide life-history data only. One C queen (Q57 in the G1 subgroup) was censored and therefore not included in longevity analyses or sampled for RNA. Final sample sizes were: R:TP1G (N = 6), R:TP2G (N = 6), R:life-history (N = 2 + 9 + 12 + 1 = 24); C:TP1G (N = 6), C:TP2G (N = 6), C:life-history (N = 2 + 9 + 12 − 1 = 22). See ‘Methods’ for full details

Fig. 2

Fertility measures for R (eggs removed) and C (eggs removed and replaced) Bombus terrestris queens/colonies. a Queen fertility (mean number of eggs produced per 48-h period) for R (N = 36) and C (N = 35) queens before treatment started (baseline mean queen fertility: days 1 and 3) and during treatment until the first worker egg-laying was observed (during-treatment mean queen fertility: days 5–25 inclusive). Queen fertility over days 5–25 represented the fertility response of queens alone to the treatment. NS, not significant; *** p < 0.001. Black circles represent means for individual queens. b Colony fertility over time for R (N = 36 on day 1 declining to N = 1 on day 134) and C (N = 35 on day 1 declining to N = 1 on day 148) colonies until day 147. Red dashed line: day 5, manipulations started; black dashed line: day 26, when worker egg-laying first observed in any colony. Daily sample sizes are in Additional file 2: Table S23. Outliers not shown. For a and b, black horizontal bars: medians; boxes: interquartile ranges; whiskers: 1.5 × interquartile range. R colonies had significantly higher colony fertility than C colonies after treatment had started (see ‘Results’). c Observed worker egg-laying over time, recorded every 4 days during 10-min observations in R (N = 36 on day 1 declining to N = 1 on day 134) and C (N = 35 on day 1 declining to N = 1 on day 148) colonies. Black dashed line: day 26, when worker egg-laying first observed in any colony. Points are offset around each integral value on the Y axis. From day 26, R colonies had significantly more worker egg laying events than C colonies (see ‘Results’)

Fig. 3

Queen longevity and worker aggression for experimental Bombus terrestris queens/colonies. a Survivorship of R (eggs removed; N = 24) and C (eggs removed and replaced; N = 22) queens; b Longevity (days from start of experiment) of R (eggs removed; N = 24) and C (eggs removed and replaced; N = 22) queens. Black horizontal bars: medians; boxes: interquartile ranges; whiskers: 1.5 × interquartile range. ** p < 0.01. In a and b, R queens had significantly reduced survivorship/longevity relative to C queens (see ‘Results’); c Observed worker-to-queen aggression over time (recorded every 4 days during 10-min observations) in R (eggs removed; N = 36 on day 1 and N = 1 on day 134) and C (eggs removed and replaced; N = 35 on day 1 and N = 1 on day 148) colonies. Daily sample sizes are in Additional file 2: Table S23. Points are offset around each integral value on the Y axis. There was no significant difference in observed worker-to-queen aggression between R and C colonies (see ‘Results’)

Fig. 4

Comparisons of mRNA-seq gene expression profiles by chronological age in experimental Bombus terrestris queens. Arrows and associated numbers denote differentially expressed genes (DEGs) between two conditions (increasing in chronological age from left to right) for a brain; b fat body; and c ovaries. R conditions, eggs removed; C conditions, eggs removed and replaced. Red numbers, above arrows: numbers of genes upregulated (more expressed with chronological age) between two conditions linked by an arrow; blue numbers, below arrows: numbers of genes downregulated (less expressed with chronological age) between two conditions linked by an arrow. d Scale displaying the day of the experiment on which each condition was sampled. TP1: time-point 1, TP2: time-point 2. Sample sizes: brain: R:TP1G (N = 6), R:TP2G (N = 5); C:TP1G (N = 6), C:TP2G (N = 5); fat body: R:TP1G (N = 6), R:TP2G (N = 6); C:TP1G (N = 6), C:TP2G (N = 4); ovaries: R:TP1G (N = 6), R:TP2G (N = 6); C:TP1G (N = 6), C:TP2G (N = 6)

Fig. 5

Comparison of changes in mRNA-seq gene expression profiles with relative age in Bombus terrestris queens. Euler diagrams of overlaps between differentially expressed genes (DEGs), i.e. genes differentially expressed between the two time-points TP1 and TP2 within treatments and shared between R queens (eggs removed) and C queens (eggs removed and replaced) for: a, b brain; c, d fat body; and e, f ovaries. (In panel e, because their low values mean there is a lack of space, numbers of DEGs for R queens and shared between R and C queens are shown adjacent to the relevant area.) Asterisks (*), significant overlap in DEGs (Fisher’s exact test, p < 0.05 after Bonferroni correction). Upregulated DEGs: DEGs significantly more expressed in TP2G than TP1G, i.e. that increase expression with queen relative age; downregulated DEGs: DEGs significantly more expressed in TP1G than TP2G, i.e. that decrease expression with queen relative age. Results of statistical tests are in Additional file 2: Table S7 and identities of overlapping genes are in Additional file 2: Table S8. Sample sizes: brain: R:TP1G (N = 6), R:TP2G (N = 5); C:TP1G (N = 6), C:TP2G (N = 5); fat body: R:TP1G (N = 6), R:TP2G (N = 6); C:TP1G (N = 6), C:TP2G (N = 4); ovaries: R:TP1G (N = 6), R:TP2G (N = 6); C:TP1G (N = 6), C:TP2G (N = 6)

Fig. 6

Age-related gene expression patterns compared between Bombus terrestris queens (current study) and ageing-related genes. a Comparison with TI-J-LiFe network; each row represents an individual gene from Drosophila melanogaster included in the TI-J-LiFe network by Korb et al. [34] that has a single-copy orthologue in B. terrestris; b comparison with enzymatic antioxidant gene set; each row represents an individual gene from D. melanogaster included in the enzymatic antioxidant gene set by Kramer et al. [36] that has a single-copy orthologue in B. terrestris. In a and b, each column shows the age-related expression status of the focal genes in a given treatment and tissue in the experimental B. terrestris queens. Vertical breaks separate the three tissues (brain, fat body and ovaries). The dendogram at left groups genes that cluster according to their gene expression patterns. Sample sizes: brain: R:TP1G (N = 6), R:TP2G (N = 5); C:TP1G (N = 6), C:TP2G (N = 5); fat body: R:TP1G (N = 6), R:TP2G (N = 6); C:TP1G (N = 6), C:TP2G (N = 4); ovaries: R:TP1G (N = 6), R:TP2G (N = 6); C:TP1G (N = 6), C:TP2G (N = 6)