Much of our evolutionary understanding of senescence is based on the principle that organisms experience a decline in the force of selection with age [1, 2] resulting in positive selection for traits that increase early-life survival or fecundity at the expense of late-life survival [3]. Theory further predicts that investing in somatic maintenance to postpone senescence is energetically costly [4]; when selection declines more rapidly with age, organisms should invest less in somatic maintenance and experience more rapid senescence. Differences in mean longevity and senescence rate among organisms should therefore be explained at least in part by differences in the pattern and degree to which selection changes with age.
Social insects, such as honeybees, are excellent model systems for exploring the evolution of senescence because of their large degree of phenotypic plasticity in senescence rate and lifespan among genetically similar individuals [10, 17, 26]. Different workers experience different levels of extrinsic hazards depending on their behavioral role in the colony [27]. In addition, extrinsic mortality, resource availability, and worker behavior vary seasonally, allowing us to examine how senescence in workers is influenced by ecological context.
There has been much theoretical work refining predictions about how extrinsic mortality [28], density-dependence [7], and other ecological factors [29] affect the selection against senescence in individuals. However, it is less straightforward how these ecological factors influence the strength of selection against senescence in social organisms, where individuals have little or no direct reproduction and fitness depends on their contribution to the colony as a whole. Using a simple stage-structured demographic model, we seek to bridge this theoretical gap to explore how ecological context influences selection against worker senescence in honeybees and other eusocial animals.
We find that there are seasonal differences in the strength of selection against senescence in honeybee workers, as measured by the sensitivity of the colony growth rate to age-dependent worker mortality. We find that the colony is more sensitive to changes in both nurse and forager senescence in winter conditions, when resources are scarce and extrinsic mortality is lower, than in summer conditions, when resources are plentiful and extrinsic mortality is high (Figs. 1 and 2). Since colonies cannot easily produce new workers in winter, small increases in the senescence of existing workers have larger effects on the colony. This difference in sensitivity may largely explain why winter honeybee workers have a much lower senescence rate than spring or summer workers [30]. In contrast, colonies are most sensitive to changes in extrinsic mortality (Figs. 1 and 2) in summer when resources are plentiful; this may be because summer workers spend more of their lives in the riskier forager state rather than the more protected nurse state [24].
We also find the seasonal pattern of selection changes with worker life stage. There is much stronger selection against nurse senescence in winter, when most workers remain in the nurse stage, than in summer and spring/fall, both periods when they are likely to transition into foragers sooner (Fig. 1). Since nurses have much lower age-dependent and -independent mortality than foragers, selection against nurse senescence in summer is driven partly by how quickly they transition to the riskier forager state. The selection against foragers senescence, on the other hand, is strongest in winter, but intermediate in spring/fall and lowest in summer (Fig. 2), suggesting that selection on forager senescence decreases as extrinsic mortality increases. This aspect of our results highlights how behavioral role can interact with ecological context to influence how the selection against senescence changes with age.
Overall, our model predicts that the selection against worker senescence should be strongest in winter and weakest in summer. This should lead to the evolution of seasonal differences in worker senescence rate, with the slowest senescence in winter and the fastest in summer. This prediction about the seasonal pattern of senescence rate matches what we observe empirically in temperate honeybee colonies [17, 30, 31]. This model therefore suggests that seasonal changes in the force of selection are important in shaping the phenotypically plastic pattern of senescence in honeybees.
Although the main objective of this model is to estimate how seasonally varying selective pressures affect the evolution of aging in honeybee workers, this method could also be used to predict how anthropogenic sources of mortality will affect the health and survival of honeybee colonies. The European honeybee is an economically important pollinator, whose crop pollination services are worth an estimated at $11.68 billion annually in the United States [32]. Managed honeybees face numerous stressors including parasites, nutrition stress, and pesticide exposure [33]. Because of logistical constraints, the impact of potential threats to honeybee health are usually evaluated at the individual rather than colony level [34]. This model can therefore help predict how changes in individual worker mortality will scale up to colony-level effects, which is important to evaluating threats to honeybee health and also can give clues to the causes of colony declines [35, 36].
In addition, many other social insect species are of great ecological importance as pollinators [37, 38], seed dispersers [39], and ecosystem engineers [40–42]; many of these species’ populations are also threatened or declining [38, 43]. Incorporating an evolutionary perspective on how ecological context shapes resource allocation within colonies can help to better inform management practices for social species of conservation concern.