Philosophical Transactions of the Royal Society B: Biological Sciences
You have accessResearch articles

Queen loss increases worker survival in leaf-cutting ants under paraquat-induced oxidative stress

Megha Majoe

Megha Majoe

Institute of Organismic and Molecular Evolution, Johannes Gutenberg University Mainz, Hanns Dieter Hüsch Weg 15, 55128 Mainz, Germany

Institute for Biology I (Zoology), University of Freiburg, Hauptstrasse 1, D-79104 Freiburg, Germany

[email protected]

Google Scholar

Find this author on PubMed

,
Romain Libbrecht

Romain Libbrecht

Institute of Organismic and Molecular Evolution, Johannes Gutenberg University Mainz, Hanns Dieter Hüsch Weg 15, 55128 Mainz, Germany

Google Scholar

Find this author on PubMed

,
Susanne Foitzik

Susanne Foitzik

Institute of Organismic and Molecular Evolution, Johannes Gutenberg University Mainz, Hanns Dieter Hüsch Weg 15, 55128 Mainz, Germany

Google Scholar

Find this author on PubMed

and
Volker Nehring

Volker Nehring

Institute for Biology I (Zoology), University of Freiburg, Hauptstrasse 1, D-79104 Freiburg, Germany

Google Scholar

Find this author on PubMed

    Abstract

    Longevity is traded off with fecundity in most solitary species, but the two traits are positively linked in social insects. In ants, the most fecund individuals (queens and kings) live longer than the non-reproductive individuals, the workers. In many species, workers may become fertile following queen loss, and recent evidence suggests that worker fecundity extends worker lifespan. We postulated that this effect is in part owing to improved resilience to oxidative stress, and tested this hypothesis in three Myrmicine ants: Temnothorax rugatulus, and the leaf-cutting ants Atta colombica and Acromyrmex echinatior. We removed the queen from colonies to induce worker reproduction and subjected workers to oxidative stress. Oxidative stress drastically reduced survival, but this effect was less pronounced in leaf-cutting ant workers from queenless nests. We also found that, irrespective of oxidative stress, outside workers died earlier than inside workers did, likely because they were older. Since At. colombica workers cannot produce fertile offspring, our results indicate that direct reproduction is not necessary to extend the lives of queenless workers. Our findings suggest that workers are less resilient to oxidative stress in the presence of the queen, and raise questions on the proximate and ultimate mechanisms underlying socially mediated variation in worker lifespan.

    This article is part of the theme issue ‘Ageing and sociality: why, when and how does sociality change ageing patterns?’

    1. Introduction

    Advanced eusocial insects live in societies that contain many closely related individuals of several generations that exhibit negligible within-group conflict [1]. Queens (and kings in termites) reproduce, whereas workers perform all other tasks necessary to raise the brood. In solitary insects, reproduction is typically traded off with lifespan [24], whereas the opposite is true in social insects: the reproductive castes live much longer than workers in all social insects [5]. Yet, workers of many social insects retain the ability to reproduce under certain circumstances. Most of the ant workers are unable to mate, but can lay haploid eggs, which develop into males. However, worker reproduction in queenright colonies is rare, as there are fitness incentives for workers to refrain from reproduction in the presence of a fertile queen [68]. Workers that attempt to develop their ovaries and lay eggs are policed by other workers who physically attack them or destroy their eggs, and these fights reduce colony efficiency [913]. Worker life histories are thus optimized for their lifetime contribution to colony fitness rather than for direct reproduction—which has led to extreme forms of self-sacrifice [1316].

    If the queen dies, the fitness landscape of a worker changes drastically. Since no more female brood can be produced, the residual lifespan of a colony is now limited to that of the remaining workers. After queen loss, remaining female larvae that have not reached the point of caste differentiation in their development will typically be raised to become queens instead of workers, as a form of terminal investment [1719]. Rearing these larvae might not exhaust the lifetime work capabilities of the remaining workers. In this case, policing is relaxed in many ant species and workers turn to direct reproduction. However, this option is not available for ant workers of all species. In advanced eusocial species, typically with extremely large colonies and strong morphological caste differentiation, workers are invariably sterile in an evolutionary sense, as they cannot lay eggs that develop into fertile males (e.g. army ants, Atta leaf-cutting ants and many invasive ants [8,2025]).

    The ability of workers to develop their ovaries is also influenced by age and task, as younger workers are more likely to develop their ovaries than older workers [6]. Age often predicts worker behaviour as well: young workers typically start out as nurses inside the colony and then later transition to outside behaviours such as foraging [26]. This age polyethism is thought to be caused by variation in the extrinsic mortality among tasks; workers that carry out risky tasks, such as leaving the colony or defending the nest, are more likely to be killed in the process. If young workers performed these risky tasks, they would endanger their long residual life- and workspan, during which these workers could still contribute to the colony's fitness. Old workers, in contrast, have a shorter residual lifespan so their death impacts the colony's fitness less than the young workers' [27,28].

    Interestingly, recent studies on termites, bees and ants have shown that worker reproduction can have an effect on worker lifespan as well: when the queen is removed and queenless workers begin to reproduce, they live longer than workers in colonies with a queen [2933]. In bumblebees, it is the long-lived workers that are more likely to lay eggs [34]. From an evolutionary perspective, living longer increases the period in which workers can reproduce, and hence increases the lifetime fitness [33]. In accordance with this reasoning, totipotent workers of wood-dwelling termites (Cryptotermes secundus) live long and invest in anti-ageing pathways [35,36]. The hypothesis that selection on the individual level has caused the evolution of longer worker lifespans at the cost of the colony has not been tested so far. It would predict that only the lives of those workers that can indeed reproduce are prolonged, but not of workers that are too old for egg production, or of workers from species with sterile workers. There are also alternative explanations for the observed patterns: colony lifespan is limited by the lifespan of the remaining workers—and if the latter is prolonged, the colony will survive for longer as well. The longer the colony lives, the higher the chance that any sexual offspring raised from the remaining queen brood live to adulthood, long enough to take part in the next mating flight. This would predict that all workers prolong their lives, not just the fecund ones.

    The idea that reproductive workers have longer lifespans than sterile ones is in line with the finding that highly fecund social insect queens live much longer than the typically sterile workers [37,38]. One way to prolong lifespan is by investing into body maintenance. There is no consistent evidence of whether and how queens invest more in body repair or resilience than workers [3942], but they can handle some stressors better than workers [43,44] One potential stress factor is oxidation. Imperfect mitochondrial respiration leaks reactive oxygen species (ROS) into cells, where they damage macromolecules such as proteins and lipids. Such damage can be avoided if organisms produce antioxidants that neutralize ROS before they do any critical damage, or if they repair the damage after it has occurred [4548].

    Here, we investigated resistance to oxidative stress as a potential mechanism by which ant worker lives are prolonged when workers become fertile. We subjected queenright and queenless workers to paraquat-induced oxidative stress. We predicted that queenless workers that activate their ovaries are better able to withstand oxidative stress. We further investigated whether the consequences of oxidative stress differed between workers collected from inside the colony and those performing outside tasks. Usually, only inside workers can become fertile and would thus be able to reap the benefits of a longer life through direct reproduction. We studied three ant species of the subfamily Myrmicinae: Temnothorax rugatulus and the two leaf-cutting ants Acromyrmex echinatior and Atta colombica. Atta colombica workers are sterile, which allowed us to test whether it is indeed direct fitness prospects that caused selection for the longer lives of orphaned ant workers.

    2. Material and methods

    To test whether fertility increases worker resistance to oxidative stress, we split colonies of three Myrmicine ant species (Ac. echinatior, At. colombica, T. rugatulus) into queenright and queenless half-colonies for 10–15 weeks. From these half-colonies, we created small subcolonies consisting of equal numbers of marked inside and outside workers. Workers in each subcolony were subjected to either paraquat-induced stress or a control treatment for two weeks (figure 1).

    Figure 1.

    Figure 1. Overview of the experimental design. Source colonies were split into queenright and queenless half-colonies. Ten to fifteen weeks later, we created subcolonies from each half-colony and they were subjected either to paraquat-induced oxidative stress or to a control treatment. Each subcolony contained equal numbers of ‘outside’ and ‘inside’ workers. Treatments were administered to individual workers that were isolated in wells of cell-culture plates for 4 h and then returned to the subcolonies.

    (a) Study species, ant collection and maintenance

    We selected three Myrmicine species with known reproductive potential of workers. Acromyrmex echinatior and Atta colombica are Neotropical leaf-cutting ants that typically have a single, multiply mated queen [49,50]. Temnothorax rugatulus is a Nearctic species with small monogynous or polygynous colonies and singly inseminated queens [51].

    Acromyrmex and Temnothorax workers rarely lay eggs in the presence of the queen, but queen loss triggers ovary development in young workers and the production of eggs that remain unfertilized and develop into haploid male offspring [9,17,32,52]. By contrast, Atta workers are sterile in an evolutionary sense, as worker groups rarely produce any males even months after queen loss, and the few males that mature are smaller and probably infertile [21]. To confirm these species-specific differences in worker reproduction in response to queen removal, we dissected the ovaries of 30 Ac. echinatior workers and 46 At. colombica workers from inside the fungus gardens of the experimental half-colonies at the time of the experiment. We found that queenless workers of Ac. echinatior were more likely to have developed eggs in their ovaries (13/15) than queenright workers (7/15; Fisher's exact test p = 0.05). By contrast, virtually, no At. colombica workers developed eggs (1/22 and 0/24 for queenright and queenless conditions, respectively; Fisher's exact test p = 0.48). In T. rugatulus, queen removal stimulated ovarian development and egg production in nurses (6 weeks after queen removal, M. Choppin 2021, unpublished data; 53 weeks after queen removal, [53]).

    The Ac. echinatior and At. colombica colonies were collected in Gamboa, Panama, in the years 2017 and 2018. In the laboratory, the leaf-cutting ants' fungus gardens grew in beakers. These were covered by upturned plastic flowerpots placed in fluon-lined plastic boxes (57 × 51 × 28 cm), the latter acting as a foraging arena. They were housed in a climate chamber at 25–27°C, 65–70% humidity with a 12 h light–dark cycle and provided with bramble, rice and apple slices twice a week, but received no additional water.

    The T. rugatulus colonies were collected in the Chiricahua Mountains in Arizona, USA, in 2018. All the colonies were habituated to laboratory conditions. The T. rugatulus colonies nested in plastic boxes (52 × 30 × 18 mm) with plaster at the base and an aluminium foil-lined lid. This box was placed in a (95 × 95 × 65 mm) closed square plastic container with small holes in the lid. All these boxes were placed in a 22–25°C climate chamber at ambient humidity in a 12 h light–dark cycle. They had ad libitum access to water and were fed with mealworms and honey once a week.

    (b) Queen removal experiments

    All source colonies were equally split into experimental colonies that either contained a queen (queenright) or no queen (queenless) and maintained this way for 10–15 weeks. Acromyrmex echinatior and At. colombica colonies are very large; they are housed in fungus gardens, containing many thousands of workers [5456]. Thus, it is difficult to count the exact number of workers. Therefore, we broke the 1.5–2 l of fungus garden into smaller pieces and redistributed the parts while keeping the vertical structure of the garden intact, distributing workers and brood equally within both halves. We split two source colonies of Ac. echinatior and four source colonies of At. colombica, creating from each colony a queenright and a queenless half-colony. After 13–15 weeks, we set up two or four subcolonies from each half-colony, resulting in 12 subcolonies for Ac. echinatior and 32 subcolonies for At. colombica. Workers from half of the subcolonies were subjected to a paraquat treatment, the other half to a control treatment. Each subcolony contained 10–12 inside and 10–12 outside workers. All workers used in this experiment were of median size (head width 1.36 ± 0.06 mm for Ac. echinatior and 1.35–2.69 mm in At. colombica based on [5759]). We collected inside workers from the fungus garden and picked outside workers from the foraging area, but not from the waste-disposal pile. The cuticle coloration—light for inside and dark for outside workers—was used to collect Ac. echinatior workers according to age, as shown before for Acromyrmex octospinosus [60]. Inside and outside workers were marked with randomly selected colours (Edding 750 paint markers). Subcolonies were housed in Petri dishes (90 mm diameter). They were provided with some fungus bereft of any eggs, larvae or workers, one single, size-standardized bramble leaf and a moist cotton pad to sustain the fungus that the workers could access ad libitum. Workers that died before the onset of the oxidative stress treatment were removed from the analysis. Our total sample size comprised 248 Ac. echinatior workers (12 subcolonies × 10–12 inside and 10–12 outside workers) and 641 At. colombica workers (32 subcolonies × 10–12 inside and 10–12 outside workers).

    We split six T. rugatulus source colonies with each half-colony receiving exactly the same number of eggs, larvae and workers. These half-colonies were maintained for 10 weeks. Thereafter, we created two subcolonies from each of them, each subcolony composed of about five inside workers (collected near the brood-pile) and five outside workers (collected from the feeding arena). Temnothorax rugatulus workers are monomorphic. Subcolonies were housed in plaster lined Petri dishes (55 mm diameter), capable of holding moisture. Each subcolony was provided with honey-soaked pieces of sponge on the first day and subsequently had access to water only through the wet sponge. In total, 208 T. rugatulus workers were included in the analysis after some mortality between marking and treatments (24 subcolonies × 5 inside and 5 outside workers).

    (c) Oxidative stress treatment

    One day after subcolony set up, each ant in half of the subcolonies was subjected to a 0.46 M solution of paraquat dichloride (CH3(C5H4N)2CH3 · 2Cl) dissolved in millipore water (oxidative stress treatment). Paraquat is a herbicide that induces oxidative stress through superoxide formation [61]. Since injecting paraquat might lead to injury, and administration by food cannot be controlled, especially in fungus-consuming leaf-cutting species, we applied the paraquat solution to each worker's head with a size 2 Vernisage™ paintbrush such that it covered the entire surface. Based on preliminary experiments using different paraquat concentrations (1.25, 0.85, 0.46 M), we chose a dose of 0.46 M that had killed approximately half of the workers within 6–8 days. The brushing procedure ensured that the amount of solution the ants received was roughly equivalent to the size of the ant as workers of the three species varied in size. To avoid transfer of the liquid through trophallaxis and to facilitate self-grooming and ingestion, each ant was placed in an isolated well of a Falcon™ cell-culture plate. All workers were isolated from each other for 3–4 h, after which all survivors were returned to their Petri dishes (figure 1). Workers of the remaining half of the subcolonies were treated in the same manner, but here we applied pure millipore water instead of the paraquat solution (control treatment).

    We noted that ants would groom themselves and consume the solution during isolation. We did not observe any attacks or special interest by the nest-mates following the return of the treated workers into the subcolonies. We found no dead individuals with signs of fights (e.g. missing appendages) inside the subcolonies. Mortality was noted once before the treatment and once after the isolation period on treatment days. During the isolation, 0.01–10% of the workers died in the control treatment and 10–22% in the paraquat treatment, depending on species. The paraquat treatment was repeated six more times on every other day. The final treatment was administered on day 13 and the final survey of the subcolonies happened at the end of this day.

    (d) Statistical analyses

    We used the coxme v. 2.2–16 package [62] in R 3.5.2 [63] to build cox-regression mixed-effect models for each species separately. The presence of queen (queenright versus queenless), treatment (paraquat versus water) and the location of workers (outside versus inside), as well as all possible interactions, were entered as fixed factors, and source colony ID as a random factor. Hypothesis testing was done with the ‘Anova’ function from the package car [64] for each species separately (electronic supplementary material, table S1). The function ‘ggsurvplot_facet’ built under the ggplot2 v. 3.3.0 package [65] was used to plot the Kaplan–Meir survival curves. As a post hoc analysis, we also ran the same model without the treatment factor for both treatments separately (oxidative stress, control) in each species to illustrate the interactions of other factors with treatment (electronic supplementary material, table S1). If workers lost their colour marks, they were not included in the analyses (NA). A few workers escaped during the treatment procedure (in total n = 2, 4 and 27 individuals in Ac. echinatior, At. colombica and T. rugatulus, respectively); these data were censored in the analyses.

    3. Results

    Workers of all species died earlier when subjected to paraquat-induced oxidative stress compared to the control treatment (p < 0.001 for each species, figures 2 and 3; statistical details: electronic supplementary material, table S1a–c). The effects of the other factors varied with species.

    Figure 2.

    Figure 2. Survival of ant workers of three Myrmicine species (a,b: Ac. echinatior, c,d: At. colombica, e,f: T. rugatulus), under control (a,c,e) and oxidative stress treatment (b,d,f). Under oxidative stress (b,d), queenless Ac. echinatior (total N = 248, evenly distributed over the groups) and At. colombica (total N = 641) workers survive better than queenright workers do. This effect is absent in the control treatment (a,c) and in T. rugatulus (e,f; total N = 208). Significance values shown in the plots are for comparisons of queenright with queenless workers, calculated post hoc from a reduced dataset limited to the treatment (control or oxidative stress) in question (electronic supplementary material, table S1).

    Figure 3.

    Figure 3. Survival of ant workers of three Myrmicine species (a,b: Ac. echinatior, c,d: At. colombica, e,f: T. rugatulus), under control (a,c,e) and oxidative stress treatment (b,d,f). Inside workers from Ac. echinatior survive better than outside workers under control conditions (a), but not under oxidative stress (b; total N = 248, evenly distributed over the groups), while in At. colombica, the same effect is visible independent of treatment (c,d; total N = 641). The effect is weak or absent in T. rugatulus (e,f; n = 208). The figure is based on the same dataset as figure 2. Significance values in the plots are for a comparison of inside with outside workers and were calculated post hoc from a reduced dataset limited to the treatment (control or oxidative stress) in question (electronic supplementary material, table S1).

    In Ac. echinatior, worker survival depended on two two-way interactions. First, the effect of queen removal depended on the treatment (queen absence: oxidative stress treatment χ12=5.242, n = 248 p < 0.05; electronic supplementary material, table S1a). A post hoc analysis revealed that the survival of control workers was not affected by the queen's absence (figure 2a; electronic supplementary material, table S1a), but inducing oxidative stress caused disproportionally high mortality in queenright workers when compared with queenless workers (figure 2b). Second, the effect of location also depended on treatment (location: oxidative stress treatment, χ12=10.745, n = 248, p < 0.001 electronic supplementary material, table S1a). Outside workers died earlier than inside workers in control subcolonies (figure 3a; electronic supplementary material, table S1a), but this effect disappeared when workers were under oxidative stress (figure 3b; electronic supplementary material, table S1a).

    In At. colombica, the queen's absence (queen absence χ12=9.093, n = 641, p < 0.01) and location (location χ12=57.499, n = 641, p < 0.001) had independent effects on worker survival (see electronic supplementary material, table S1b for details). There was no clear evidence for an interaction (queen absence: treatment χ12=2.664, p = 0.102; electronic supplementary material, table S1b). However, post hoc tests revealed that workers without the queen survived better than queenright workers under oxidative stress (electronic supplementary material, table S1b; figure 2d), but that this was not the case in the control treatment (electronic supplementary material, table S1b; figure 2c). Inside workers survived better than outside workers under both control (figure 3c) and oxidative stress treatments (figure 3d; electronic supplementary material, table S1b).

    In T. rugatulus, inside workers outlived outside workers (location χ12=9.093, n = 208, p = 0.037; electronic supplementary material, table S1c), but queen absence had no effect on worker survival (queen absence χ12, p = 0.322 figure 2e,f; electronic supplementary material, table S1c) and there were no significant interactions between any of the factors (electronic supplementary material, table S1c; p > 0.29 for all possible interactions in the full model). Post hoc tests showed a tendency for inside workers to survive better than outside workers under control treatment (figure 3e; electronic supplementary material, table S1c) and this trend was more visible under oxidative stress (figure 3f; electronic supplementary material, table S1c).

    4. Discussion

    We experimentally analysed the influence of queen presence, oxidative stress and their interactions on worker survival in three Myrmicine ant species. We found queenless workers of two of the species to be more resilient to paraquat-induced oxidative stress than queenright workers. An increase in resistance to oxidative stress in orphans either via the production of antioxidants or by repairing molecular damage induced by ROS could potentially explain how queen removal can prolong the lives of workers. The higher survival rate of inside workers is likely owing to their younger age and was independent of paraquat-induced oxidative stress, suggesting that resilience to oxidative stress does not change with age.

    (a) Queenless workers deal better with oxidative stress

    In all three species, paraquat-induced oxidative stress caused a higher mortality than the control treatment and killed most workers in all experimental groups, indicating that paraquat and the dose we used are suited to induce stress and mortality. Queenless workers have previously been found to live longer than queenright workers in some ant species [12,3032,53,66] and we had hypothesized that this effect may in part be caused by individuals being more resilient to stressors such as ROS. We found evidence for this in the two leaf-cutting ant species we tested. In Ac. echinatior and At. colombica, workers survived the oxidative stress treatment better when they came from queenless colonies.

    (b) Proximate and ultimate causes of long-lived orphans

    Our results demonstrate that increased resilience to oxidative stress could be responsible for the prolonged lives of orphaned workers in ants. A simple explanation would be that orphaned workers upregulate their investment into the production of antioxidants [67], which totipotent termite workers do [35,36], and which has also been shown in Temnothorax ants, where workers can only have male offspring [53]. Alternatively, the oxidative stress resilience and short-term increase in survival of orphaned workers might be a mere by-product of a physiology modified for reproduction. Evidence from honeybees suggests that the lipoprotein vitellogenin could cause such effects. This yolk precursor required for egg production also has antioxidant properties in bees [48,68]. Vitellogenin is preferentially oxidized and thus protects other proteins from ROS in many tissues, including the honeybee brain [69], where involvement in reproduction is unlikely. Honeybees with high vitellogenin titres survive experimentally induced oxidative stress better, and vitellogenin knockdown leads to a lower oxidation tolerance [68]. High expression of vitellogenin in queens and to a lesser extent also in young nurses can explain their decreased intrinsic mortality compared to foragers. This may mean that any egg-producing individual is also less susceptible to oxidative stress, resulting in a longer life. It is as yet unclear whether vitellogenin plays a similar role in ants [44,48,67,70], where the vitellogenin gene underwent subfunctionalization after duplication, with different copies taking over different functions in reproduction and division of labour [7173].

    If worker resilience is not a mere by-product but an adaptive trait, it may be caused by a shift in the evolutionary interests of workers. In our focal species, there is little incentive for workers to reproduce in the presence of their queens [7,74]. However, this situation drastically changes when queen reproduction is no longer an option. Then, producing their own offspring will be the most efficient way of gaining fitness [75]. In this scenario, increased resilience, and the longer lives of queenless workers demonstrated by other studies [51,76], may have been selected for in workers because this would give them more time to produce offspring.

    Selection for longer lives should be restricted to those workers that can still lay eggs. In most social insects, workers transition from inside to outside tasks throughout their lives. Incidentally, worker reproduction is typically restricted to younger workers, while older workers often resorb their ovaries, as in our focal species [6,18,57,77]. Such a pattern should then lead to young workers, but not older workers, prolonging their lives after queen loss. In none of our focal species did we find any indication of such an effect. There were no interactions between queen presence and location, which is a proxy for worker age, which would be needed to support such a hypothesis (figure 3; electronic supplementary material, table S1). In particular, within the oxidative stress treatment, the effect of queen presence was not restricted to inside workers. In addition, At. colombica workers were more resilient in the absence of the queen, even though these workers are sterile in an evolutionary sense. Therefore, increased lifetime fecundity cannot be the reason for better resilience in this species. However, an onset of ovarian development, even if it is only for the production of trophic non-viable eggs, might still contribute to a longer lifespan.

    If increased resilience is not restricted to potentially fecund individuals but applies to all orphaned workers, it may be caused by a different effect: upon queen loss, a colony's workforce cannot be replenished because workers can only lay haploid eggs that develop into males, but not into workers. At the same time, leaf-cutting ant colonies are always full of queen-produced brood in different stages, many of which can still develop into queens that can go on to found daughter colonies. This development will take a few months [57,78] and it is thus crucial that enough workers are available throughout to take care of the brood and to forage. Further, the young queens cannot simply leave the nest and found a colony whenever they have eclosed, but they depend on mating flights, which occur only once or twice a year. Thus, it would mean a great increase of fitness if workers were able to extend their lives long enough to sustain the colony until the next mating flight. This reasoning based on indirect fitness gains would also explain the life-prolonging effect of queen loss in sterile At. colombica workers and of older workers.

    Finally, a potential proximate reason for sterile At. colombica workers to increase oxidative stress resistance in response to queen loss might be that the physiological changes associated with queen loss can prolong lifespan without realized fecundity. The molecular networks that translate the social environment into reproductive activity may be conserved, and perhaps cannot be rapidly modified by evolution. Such physiological changes upstream of egg production, for example, linked to the insulin signalling, juvenile hormone and vitellogenin pathways [36,79,80], may result in an extended lifespan via higher resistance to oxidative stress [4,8185].

    (c) Inside workers survive better

    Overall, inside workers outlived outside workers in all three species and this was largely independent of the oxidative stress treatment. The exception was Ac. echinatior, where oxidative stress seemed to eradicate the survival difference. The observation that no induced oxidative stress was required for outside workers to die earlier indicates that control workers also experienced some sort of stress during their daily lives or owing to our handling. The most likely cause for the survival difference between inside and outside workers is that the former are younger than the latter, and that older workers have a higher intrinsic mortality rate. In many advanced eusocial insects, workers undergo a temporal polyethism. They start out as young workers conducting inside tasks such as brood care, and only later move outside the nest, e.g. to forage [26,71,8688]. Since old workers are more likely to leave the nest and forage, they are also more exposed to predators and parasites than the inside workers. Such a higher risk of extrinsic mortality would select for a lower investment into body maintenance in old workers and consequently a higher intrinsic mortality. Any investment into body repair processes is less likely to pay off with increasing chances that an individual will be killed [26,28,44,67,89].

    (d) No effect of queen loss on Temnothorax rugatulus workers

    Our result that queen loss did not improve the resilience of T. rugatulus workers to oxidative stress is surprising, because unchallenged workers have been previously shown to indeed live longer in queenless colonies of Temnothorax longispinosus and T. rugatulus [32,53]. It is possible that the prolonged lives of T. rugatulus workers are caused by a different mechanism than in the leaf-cutting ants, or that the differences in lifestyle cause the leaf-cutting ants to be more sensitive to oxidative stress in general (e.g. the fungus garden resource [52], or metabolic rates [90]). Leaf-cutting ant workers exhibit much shorter lifespans (around 3–7 months [57,78]; H. Simon, M.M. and V.N. 2021, unpublished data) than Temnothorax workers, which can live for 2–3 years [33], supporting the notion that Temnothorax workers are more resilient. Indeed, both Acromyrmex and Atta workers typically die within a few days without food (V.N. 2021, unpublished data), while T. rugatulus workers can live for months without nutrition [91].

    5. Conclusion

    We show here that queenless leaf-cutting ant workers are more resistant to paraquat-induced oxidative stress than queenright workers are. This indicates that queen loss leads to physiological changes in orphaned workers that increase their resilience to oxidative stress, and potentially lengthen their lifespan. These changes may be directly caused by fecundity, e.g. because vitellogenin synthesized for egg production can act as an antioxidant, as it does in honeybees. Workers may also upregulate their body maintenance in other ways to increase their chances of future direct reproduction. However, we did not find evidence that life-prolonging effects are restricted to workers that can indeed reproduce. It is thus possible that the effect is ultimately caused by workers becoming more valuable because they cannot be replaced any more, or by fundamental physiological constraints. Queen loss had no effect on the resilience of T. rugatulus workers, although these are known to live longer when orphaned. To elucidate the causes of these interspecific differences, comparative studies including more species [39] and detailed research into the physiological and molecular mechanisms underlying worker lifespan are needed.

    Ethics

    Ant collection and export permits were obtained for the leaf-cutter ants from the Autoridad Nacional del Ambiente (ANAM) in Panama by J.J. Boomsma, and for T. rugatulus by S.F. from the Coronado National Forest via the SWRS. Import and export licences are not required for the transport of our study species. We followed the DFG Animal welfare guidelines and the local laws.

    Data accessibility

    All raw data presented here are provided in the electronic supplementary material.

    Authors' contributions

    All authors designed the experiment. Temnothorax rugatulus ant collection was done by S.F. with help of Barbara Feldmeyer and Marina Choppin. M.M. conducted the experiments and survival analyses. V.N. and R.L. helped M.M. in the statistical analyses. M.M. wrote a first draft and all authors revised it. V.N., S.F. and R.L. co-supervised the project.

    Competing interests

    We have no competing interests.

    Funding

    Funding came from the German Research Foundation (DFG grant nos. NE1969/4-1, FO 298/26-1, LI 3051/3-1).

    Acknowledgements

    We are indebted to J.J. Boomsma for donating the leaf-cutting ant colonies and to all members of the research unit FOR2281, Judith Korb, in particular for discussions and inspiration. Rebecca Endermann and Maarten Wissink assisted with the care of the ants. Barbara Feldmeyer and Marina Choppin helped in T. rugatulus ant collection. M. Choppin also provided information on ovary development in orphaned brood-care workers of T. rugatulus. Thanks to Judith Korb and Jürgen Heinze for inviting this contribution.

    Footnotes

    One contribution of 16 to a theme issue ‘Ageing and sociality: why, when and how does sociality change ageing patterns?

    Electronic supplementary material is available online at https://doi.org/10.6084/m9.figshare.c.5300840.

    Published by the Royal Society. All rights reserved.

    References