Discussion
While there is substantial evidence of actuarial senescence in insects and some evidence for reproductive senescence (Nussey et al.2013; Zajitschek et al. 2019), the drivers of variation in senescence patterns are not fully understood. Here we manipulated both age at mating and nutrition to quantify reproductive senescence in tsetse, a viviparous fly with high maternal allocation and iteroparous reproduction. Both offspring weight and starvation tolerance declined with maternal age, after a peak, yet these patterns of senescence were similar across treatments. Contrary to predictions from life history theory, therefore, neither changes in maternal allocation nor resources affected the timing and rate of reproductive senescence, in terms of offspring quality.
We did observe a steeper increase in the hazard of mortality with age for nutritionally stressed mothers (S5 File). This suggests that females do not have a survival benefit from reduced reproductive effort, as nutritionally stressed mothers had higher probability of abortion and produced smaller offspring at any age compared with mothers in the control and mating delay groups. This contrasts with findings from studies of other insects where nutritional stress both reduced reproductive output and either maintained or even extended lifespan relative to a control group (De Sousza Santos & Begon 1987; Ernsting & Isaaks 1991; Kaitala 1991). Taken together, it may be that factors other than the direct costs of reproduction through physiological damage, or indirectly through resource allocation trade-offs, impose higher mortality rates. Alternatively it may be that, given the extreme maternal allocation in tsetse, even though females produce relatively smaller offspring they still pay a high cost of reproduction in terms of physiological damage; and females on a poor quality diet experience this cost to a greater extent in terms of impact on mortality. This is supported by data from field-caught tsetse, where smaller females invest relatively more of their fat in their offspring, even though their offspring were smaller (Hargrove et al. 2018).
We find that females experiencing nutritional stress have a relatively higher rate of abortion. Hargrove and Muzari (Hargrove & Muzari 2015), using field collected G. pallidipes , showed that transfer of the majority of fat to the larva occurs only after c. 80% pregnancy has been completed. Therefore, a female could potentially abort a larva if there are not enough fat reserves for a full-term pregnancy. Evolutionary models tailored to tsetse life-history, with high investment in single offspring across multiple reproductive bouts, could yield insights into whether such spontaneous abortion is an adaptive strategy to retain reserves for future reproduction, or a result of physiological constraints that limit the reserves available (McNamaraet al. 2009).
Our study highlights the benefits gained from individual-level data to understand senescence. The bell-shaped relationship of offspring quality with age may have contributed to the relatively small effects of age evident in previous studies where grouped ages and mean values were used, rather than tracking reproductive output from individual females (Langley & Clutton-Brock 1998; McIntyre & Gooding 1998). The bell-shaped pattern observed here is strikingly similar in shape across treatments and reflects the general bell-shaped pattern of reproductive senescence observed across diverse taxonomic groups e.g. (Velando et al. 2006; Sharp & Clutton-Brock 2010). As summarised by (Monaghanet al. 2020), a bell-shaped relationship between maternal age and reproductive output, or offspring quality, can arise from population-level effects, due to selective disappearance, but also from individual-level effects. By tracking individual mothers, here we provide evidence that, for tsetse, the bell-shaped relationship between offspring wet weight and maternal age is a consequence of individual effects. The initial increase in offspring wet weight could be a consequence of mothers accruing reserves each time she takes a bloodmeal, so that at each sequential reproductive event she has more energy reserves that can be provided to the offspring. The subsequent decline could then be explained by senescence.
Tracking individual mothers provided insights into individual variation in maternal allocation. There was marked variation in offspring wet weight among mothers, particularly for the mating delay treatment and variation in senescence patterns for wet weight, particularly for nutritionally stressed mothers. Some females in the mating delay treatment consistently produced smaller than average offspring, across all ages, and these females contributed more to the individual heterogeneity than those producing consistently heavier offspring. For nutritionally stressed mothers, variation between individuals in offspring wet weight increased as mothers aged. These observations suggest that variation in offspring quality is affected not only by mother size but also unmeasured aspects of her condition. The large amount of variation between individuals, in offspring weight and changes in the extent of variation with maternal age in this study, was unexpected, suggesting that future studies quantifying the relative roles of mother size and physiological condition on offspring size across different ages would be valuable.
Offspring from young mothers that were nutritionally stressed had the lowest starvation tolerance. We also found that maternal age affected offspring starvation tolerance independently of age-effects mediated through wet weight, suggesting that there may be other factors associated with maternal age that influence the quality of offspring. More subtle effects of maternal age on the quality of resources transferred to offspring warrant further investigation in tsetse and other species. For tsetse, during late stages of pregnancy females not only transfer fat but also amino acids. It may be that young nutritionally-stressed females are limited in these amino acids. High amounts of tyrosine in the gut contents of third instar larvae, which is involved in tanning of larval and adult cuticles have been observed (Cmelik et al. 1969). The authors reasoned that the tyrosine and phenylalanine obtained from a single bloodmeal is unlikely to be sufficient to meet the amount required by offspring and that a surplus stored from previous bloodmeals may be required. This suggests that size may be limited when nutrients are limited, irrespective of the amount of energy available. It also demonstrates that resource allocation processes are likely more complex, and more nuanced studies on the effects of the quality of resources as well as quantity may be required to understand the ageing process.
We focused on reproductive senescence in this study. One limitation is that we did not continue the experiment beyond 100 days to quantify more fully mother survival for all three treatments. Our evidence of actuarial senescence in the nutritional stress treatment supports analysis of mark-recapture studies of G. m. morsitans in the field (Hargrove et al. 2011) which also showed an increase in mortality as a function of age. Considering this, an additional experiment to test whether females mated later experience a delayed onset in actuarial senescence would be informative, given the similar rates of reproductive senescence observed across treatments.
We also note that our conclusions with respect to reproductive senescence are constrained to this 100-day period. While, in principle, it would be optimal to run the experiment until all females had died, the >1/4 of offspring being aborted and fewer females surviving would result in relatively small sample sizes beyond this time point. However, we are confident that our study timeframe captures an ecologically relevant period to study age-dependent allocation in tsetse, given that field studies show that 90% of females have died before 100 days (Hargrove et al. 2011). Lastly, we acknowledge that our conclusions may not necessarily extend to wild flies, and further study of tsetse in the field would need to be carried out to confirm this.
To conclude, our results provide evidence of a bell-shaped relationship between maternal age and offspring quality at the individual-level for the iteroparous tsetse fly. This complements the existing body of work from other species, that has predominantly shown a similar curve for the number of offspring produced as mothers age (Monaghan et al.2020). Contrary to predictions from life history theory, however, our study did not find evidence of a decline in offspring quality as a function of direct costs of reproduction or resource allocation trade-offs.