Introduction
Across the natural world, it is commonly observed that as individuals get older, they are more likely to die and have lower reproductive output (Nussey et al. 2013; Hoekstra et al. 2019; Zajitschek et al. 2019). Broad patterns of actuarial and reproductive senescence have been well described for a wide range of taxa (Nussey et al. 2013; Hoekstra et al. 2019; Zajitscheket al. 2019), particularly birds and mammals (Gaillard & Lemaître 2020). There is also extensive variation in the onset and rate of ageing, both within and across populations (Holand et al.2016; Rodríguez-Muñoz et al. 2019; Cayuela et al. 2020). The role of individual heterogeneity, and subsequent variation in the onset and rate of ageing among individuals, however, has been neglected (Gaillard & Lemaître 2020).
A central tenet of the evolution of senescence is that the strength of selection declines with age (Hamilton 1966), with consequences in terms of antagonistic pleiotropy (Williams 1957) or the accumulation of somatic damage (Kirkwood 1977). Life-history theory suggests both antagonistic pleiotropy and the accumulation of somatic damage can be explained by trade-offs between reproduction and other physiological processes (Kirkwood 1977; Boggs 2009; Baudisch & Vaupel 2012; Davisonet al. 2014). A key assumption underlying this theory is that resources are limited and must be allocated either to reproduction or somatic maintenance (Partridge 1987; Boggs 2009). Reproductive senescence could also occur through physiological damage incurred directly from reproduction, or can be an adaptive strategy to prolong survival (McNamara et al. 2009).
Experimental manipulation of access to mates or food can yield insights on the factors driving variation in senescence, including costs of reproduction and allocation of resources. For such experiments, insects are useful organisms as they have relatively short generation times, can be reared in large numbers, and access to mates and resources can be easily manipulated. Experiments with Lepidoptera, for example, have shown that delayed mating reduces fecundity but extends longevity (Unnithan & Paye 1991; Jiménez-Pérez & Wang 2009). A large body of studies have also shown that females with access to fewer resources have lower overall reproductive output but longer lifespan (De Sousza Santos & Begon 1987; Ernsting & Isaaks 1991; Kaitala 1991; Chippindaleet al. 1993; Tatar & Carey 1995; Curtis Creighton et al.2009). These studies support the theory that senescence is, at least in part, caused by direct costs of reproduction through physiological damage, or arises as an indirect consequence of resource availability. However, experimental manipulation of both age at reproduction and nutrition is required to tease apart the contributions of reproductive history and resource availability to senescence.
Here, we present such an experimental study, focussing on reproductive senescence in tsetse (Glossina morsitans morsitans ). Tsetse are vectors of human and animal trypanosomiasis in Africa. They give birth to a single live larva weighing the same as the mother (Hargrove & Muzari 2015; Haines et al. 2020), approximately every nine days. Immature stages receive energy and nutrients from the mother only. Adults do not increase in size after emergence and have a relatively long lifespan for their small size, living for weeks rather than days (Hargrove et al. 2011).
Evidence for age-related changes in reproductive output in field and laboratory tsetse is mixed (Jordan et al. 1969; Langley & Clutton-Brock 1998; McIntyre & Gooding 1998). Key limitations to these studies are that flies were kept only under optimal laboratory conditions, not tracked individually and frequently grouped across ages. Preliminary data from the tsetse colony at the Liverpool School of Tropical Medicine (LSTM) provided evidence of reproductive senescence (Supporting Information, S1 File), but these data did not allow for teasing apart of within-individual from among-individual patterns (Monaghan et al. 2020). In this study, we therefore used a novel method of housing tsetse females to track individual mothers and their offspring from this colony.
Our first objective was to confirm reproductive senescence in tsetse, as shown by a decline in the number and quality of offspring with maternal age. Our second objective was to determine the effect of maternal nutrition and the physiological costs of reproduction on age-dependent patterns of maternal allocation. To achieve this aim, we manipulated nutritional stress by feeding adult females on a high- or low-quality diet and we varied potential costs of reproduction by delaying the age at which females were mated. Our third objective was to assess variation in reproductive senescence patterns among individuals, due to unmeasured heterogeneity in factors including size, condition and mate quality.
We hypothesised that: i) females mated later would experience a delayed and/or slower increase in the probability of abortion and decline in offspring quality, due to physiological costs of reproduction, resource allocation trade-offs, or both; and either: ii) nutritionally stressed mothers would have an earlier, and potentially steeper increase in probability of abortion and decline in offspring quality, due to resource allocation-trade-offs; or iii) nutritionally stressed mothers may senescence more slowly, due to lower physiological costs of reproduction, if stressed females have lower reproductive output.