Abstract
Benevolent social behaviours, such as parental care, are predicted to relax selection against deleterious mutations, enabling them to persist. We tested this prediction experimentally using burying beetles Nicrophorus vespilloides, which make an edible nest for their larvae, whom they nourish and defend. For 20 generations, we allowed replicate experimental burying beetle populations to evolve either with post-hatching care (‘Full Care’ populations) or without it (‘No Care’ populations). Lineages were seeded from these experimental populations and then inbred to expose differences in their mutation load. Outbred lineages served as controls. Half the lineages received post-hatching care, half did not. We found that inbred lineages derived from the Full Care populations had lower breeding success and went extinct more quickly than lineages derived from the No Care populations – but only when offspring received no post-hatching care. We infer that Full Care lineages carried more recessive deleterious mutations. When parents provided care, the developmental environment was sufficiently benign that broods had higher survival, whether the population had a high mutation load or not. We suggest that the increased mutation load caused by parental care increases a population’s dependence upon care. This could explain why care is seldom lost once it has evolved.
Introduction
Classical population genetics models imagine that populations attain an equilibrium level of genetic variation (known as mutation-selection balance [1-5]). New genetic mutations arise spontaneously, through diverse mechanisms, and increase genetic variation in the population [e.g. 5, 6]. However, since the majority of new mutations are mildly deleterious [e.g. 5, 6], they are quickly purged by natural selection. Mutation-selection balance is theoretically achieved when the rate of input of new genetic variants through spontaneous mutation is perfectly balanced by the rate of their elimination by selection [1-5].
The concept of mutation-selection balance has long been used as a theoretical reference point for understanding the effects of mutation rate on the health of human populations, partly because it is recognised that humans can modify their own environment and so change the forces of natural selection to which they are exposed [1-5]. Better quality housing, improved diets, and benevolent social activities, such as a welfare state or the universal provision of medical care, are suggested to have been particularly influential in preventing natural selection from purging deleterious mutations in human populations [1, 2, 4, 5]. Consistent with this suggestion, recent comparative genomic analyses have revealed a greater incidence of genetic pathologies in western industrialised populations than in traditional, pre-industrial human societies which are more exposed to natural selection [4, 5, 7-9]. Nevertheless, it is impossible to demonstrate that a more benign physical and social environment, in which selection is relaxed, has caused this difference.
Elaborate architecture, enhanced access to resources and benevolent social behaviours are relatively commonplace in other animals too, especially among the many bird, mammal and insect species that cooperate with each other and live socially [10]. For these species, any causal effects of this social and physical environment on genetic variation can more easily be investigated. A complication in other animal societies, however, is that additional factors might perturb the mutation-selection balance. For example, animals that breed cooperatively also tend to produce fewer, larger offspring. This life history strategy is known to reduce genetic diversity [11] and could potentially oppose, or even conceal, any increases in genetic variation that are due to cooperation buffering the effects of natural selection. Cooperative animal societies are also commonly associated with a high incidence of reproductive skew. Since only a few dominant individuals are typically able to reproduce, the effective population size is greatly reduced [12]. This can lead to a reduction in the efficiency of natural selection and a greater influence of genetic drift [6], potentially confounding any increases in genetic variation that are due solely to relaxed selection. Similarly, animal societies typically comprise related individuals that derive kin-selected benefits from their cooperative social interactions. Theoretical analyses have shown that kin selection acts more weakly than direct selection [13]. Consequently, loci under kin selection are predicted to harbour more sequence variation than loci under direct selection [3, 13].
We tested the effect of kin-selected cooperative actions on the maintenance of genetic variation by focusing on parental care, a widespread form of cooperation [14]. Since care is commonly exhibited by pair-breeding individuals, this form of cooperation is unlikely to change effective population size – eliminating this potentially confounding effect. By building protective nests, defending their brood from attack and nourishing them, animal parents shield their young from environmental stressors [15] and weaken the correlation between the phenotypic variation seen by selection and the underlying genetic variation [3, 16]. In these ways, parents relax selection on the offspring phenotype [3, 15-18] theoretically allowing mildly deleterious alleles to accumulate [3, 6, 15]. Previous experimental work with insects has shown that parental care relaxes selection sufficiently that new mutants [15] and inbred offspring [19] can survive - at least for one generation. However, it is not yet known whether parental care also enables mildly deleterious mutations to persist over multiple generations.
We tested this prediction by evolving replicate laboratory populations of burying beetles Nicrophorus vespilloides under sharply contrasting levels of parental care, for 20 generations. Comparing populations within species also allowed us to eliminate the confounding effect of propagule number on genetic diversity [11]. Burying beetles breed on the body of a small dead vertebrate [20], which the parents jointly convert into a carrion nest by removing the fur or feathers, rolling the flesh into a ball, covering it with anti-microbial anal exudates, and burying it. This is pre-hatching parental care [21]. Parents also guard and feed larvae after hatching, though larvae can survive in the lab with no post-hatching care at all [22]. In two of our evolving populations, larvae were able to receive both pre-hatching and post-hatching parental care (these were called the ‘Full Care’ lines) while in two other populations we prevented parents from supplying any post-hatching care by removing them before the larvae hatched, after the carrion nest was complete (these were called the ‘No Care’ lines). During the first 20 or so generations of experimental evolution, No Care lines rapidly adapted to a life without parental care [23], through divergent phenotypic change in both larval [e.g. 24]) and parental [21] traits.
To determine whether parental care causes deleterious genetic variation to accumulate over the generations, we inbred sub-populations, each derived from the replicate experimental evolving populations, for 8 successive generations (we called this The Evolutionary History Experiment). For these 8 generations, we measured the extent to which inbreeding reduced measures of reproductive success in comparison with control outbred populations. To determine whether parental care could temper the rate of extinction (as implied by [19]), in half of all our treatments parents were allowed to provide care after their offspring hatched, while in the remainder they were prevented from supplying post-hatching care. This generated 8 different treatments in total (see Supplementary Figure 1, for the design of the Evolutionary History Experiment).
We used the data from the Evolutionary History Experiment, to test three predictions: (1) that the No Care environment is harsher than the Full Care environment. (2) That the more benign conditions of the Full Care environment relax selection and promote the survival of more genetic variants. (3) That inbred populations from the Full Care lines should exhibit greater inbreeding depression than inbred populations from the No Care lines (having accumulated a greater number of deleterious recessive mutations under relaxed selection). The outbred populations acted as a control treatment for tests of all three predictions.
Methods
Nicrophorus vespilloides natural history
The common burying beetle N. vespilloides breeds on a small dead vertebrate (like a songbird or mouse). The larvae hatch from eggs laid nearby in the soil and crawl to their carrion nest, which they can feed upon themselves [20]. Once at the carcass, larvae receive post-hatching biparental care. Parents supply fluids to their offspring through oral trophallaxis, and defend their brood and the carrion nest from attack by predators, microbes and rival beetles [20]. The duration and extent of post-hatching care are highly variable, however. For example, when wild beetles are brought into the lab to breed, roughly 5% of larvae receive no post-hatching care at all, yet larvae can still survive to become reproductively competent adults (e.g. [22, 25]. Within roughly a week of hatching, the larvae complete development and at this point (which we refer to as ‘dispersal), they start to crawl away from the scant remains of the carcass to pupate in the soil. The parents, meanwhile, fly off in search of a new carcass.
Experimental evolution
The experimental lines used in this work have been described in detail elsewhere [e.g. 21, 24]. In brief, we established a large founding population of N. vespilloides by interbreeding wild-caught individuals from four different woodlands. This was then divided into four experimental lines. In two lines, larvae experienced ‘Full Care’ at each generation, with both parents staying in the breeding box throughout the breeding bout and able to provide post-hatching care as well as pre-hatching care. In the other two ‘No Care’ lines, parents engaged in pre-hatching care but at each generation they were removed from the breeding box around 53 h after they were paired, so that they never interacted with their larvae. The work reported here began when these lines had been exposed to 20 generations of experimental evolution under these contrasting regimes of care.
Evolutionary History Experiment Preparatory common garden generation
The experiment began by exposing individuals drawn from the four lines (Full Care replicated twice and No Care replicated twice) to a common garden Full Care environment for one generation (N = 60 pairs for each No Care line (to counter-balance the slightly lower breeding success caused by the No Care environment) and N = 50 pairs for each Full Care line). In this way, we minimised any potentially confounding transgenerational effects prior to starting the Evolutionary History Experiment.
Overview (see Supplementary Figure 1)
Broods from the common garden generation were used to seed lineages in the experimental treatments: broods derived from the Full Care populations (FCPOP) founded lineages that were Inbred or Outbred, in either a Full Care (FCENV) or No Care (NCENV) environment, and the same was true for broods derived from the No Care populations (NCPOP). Thus, for each experimental line of origin, individuals in the different treatments came from a similar genetic pool. From Generation 1 onwards, half of the beetles drawn from each line were exposed to continuous inbreeding (full-sibling crosses) for up to 8 generations (by which point all the inbred lineages had gone extinct) (N = c. 45 crosses per treatment at Generation 1). The remaining beetles were outbred in identical conditions to provide a control baseline for comparison with the inbred lineages (N = c. 35-40 crosses per treatment, per generation). Half of all inbred lineages, and half of the outbreeding populations, were allowed to provide post-hatching care for their young (Full Care environment), while the remaining beetles were only allowed to provide pre-hatching care (No Care environment). The experiment therefore had a 2 × 2 × 2 design, with 8 treatments in all (Full Care versus No Care line of origin; Inbred versus Outbred; Full Care environment versus No Care environment), with each treatment replicated twice due to replicate Full Care and No Care populations (Supplementary Figure 1).
Detailed methods
Beetle maintenance was carried out following standard protocols [23]. Briefly, adult beetles were kept individually in plastic boxes (12 × 8 × 6cm) filled with moist soil and fed twice a week with raw beef mince. Adults were bred at 2-3 weeks post-eclosion in a breeding box (17 × 12 × 6cm) with soil and a mouse carcass (11-13 g for all treatments except for the individuals derived from the Full Care lines, that were outbred under Full Care conditions (8-14 g)). To ease the considerable burden of work, data for broods in this treatment were collected from the ongoing experimental evolution lines in the laboratory. Carcass size was included, where appropriate, as a factor in the statistical analyses (see below).
For the inbreeding treatments, we paired full siblings (one pair per family) whereas for the outbreeding treatments we paired males and females at random and did not pair siblings or cousins. Each pair was given a breeding box with a dead mouse sitting on soil, and the breeding boxes were placed in a dark cupboard to simulate natural underground conditions. For broods assigned to a No Care environment, parents were removed around 53 h after pairing. Eight days after pairing (which is when the larvae have completed their development and start to disperse away from the carcass) we scored two standard measures of reproductive success in burying beetles [21]: brood success (fail = no larvae produced; success = some larvae produced) and brood size at dispersal. Larvae were then placed into cells (2 × 2 × 2cm) in an eclosion box (10 × 10 × 2cm), with one eclosion box per brood, which was filled with soil until larvae had developed into sexually immature adults (about 18 days after dispersal). At this point, adults were transferred to individual boxes until they reached sexual maturity roughly 2 weeks later. Both the eclosion boxes and the individual boxes were kept on shelves in the laboratory at 21°C on a 16L:8D hour light cycle.
Statistical Analyses
All statistical tests were conducted in R version 3.5.1 [26]. Data handling and visualisation were carried out using the ‘tidyverse’ [27] and ‘survminer’ [28] R packages. All data and code presented in the manuscript is available through: https://github.com/r-mashoodh/nves_MutationLoad.
Testing predictions (1) and (2)
To test predictions (1) and (2) we focused on the data collected from Generation 1. Using a binomial generalised linear model (GLM) in the base ‘statistics’ package in R, we tested the effect of evolutionary history (i.e. derived from a No Care evolving population or from a Full Care evolving population), current care environment (i.e. experienced No Care or Full Care during Generation 1), and inbreeding (i.e. inbred or outbred) on brood success. We defined brood success at dispersal in the following way: broods that produced at least one larva that survived to breed were defined as successful (following [21, 23]) whereas those that did not produce any surviving young were classified as failures.
We subsequently ran analyses separately for the inbreeding and outbreeding conditions to examine any interactions between evolutionary history (i.e. derived from a No Care evolving population (NCPOP) or from a Full Care evolving population (FCPOP)) and the current environment (i.e. experienced No Care (NCENV) or Full Care (FCENV) during Generation 1) dropping non-significant interaction terms where appropriate. We included block and carcass weight as covariates to ensure any effects we detected occurred over and above any variation in these variables.
Testing prediction (3)
Calculation of inbreeding depression
For direct comparison with previous work [19], we calculated the inbreeding depression rate: δ =(wo−wi)/wo, where wo and wi are respectively the number of surviving outbred and inbred offspring at dispersal. We combined data from both blocks.
Survival analysis across generations
To determine the effect of evolutionary history (i.e. derived from a No Care evolving population or from a Full Care evolving population), and current care environment (i.e. experienced No Care or Full Care during Generation 1) on the survival of the different lineages in the Evolutionary History Experiment (Supplementary Figure 1), we fit accelerated time hazard models with a log-logistic distribution using the ‘survival’ R package [27]. Again carcass weight and block were included as covariates. A lineage was considered to be extinct if it did not survive to reproduce in the subsequent generation. We additionally used the non-parametric Kruskal Wallis test to determine if median survival times of each inbred lineage differed, by comparing the effect of evolutionary history (i.e. derived from a No Care evolving population or from a Full Care evolving population) in separate analyses, one for each current care environment (No Care versus Full Care). Model diagnostics were checked visually.
Results
To test predictions (1) and (2) we initially focused on the data collected from the first generation of breeding in the Evolutionary History Experiment. In support of prediction (1), we found that exposure to a No Care environment reduced reproductive success, regardless of the evolutionary history of the lineage (Figure 1, Table 1). However, in support of prediction (2), we found that a supply of post-hatching care enabled more broods to survive, even if they were inbred -and regardless of the evolutionary history of their lineage (Figure 1, Table 1), replicating previous work [19].
To test prediction (3), we continued to examine inbred families in the first generation of breeding in the Evolutionary History Experiment. In this generation, we found an interaction between evolved history and the current environment in inbred but not outbred lineages (Table 2). We split the dataset by the current level of care supplied, to be able to examine the effect of evolutionary history in more detail. In support of prediction (3), we found inbred families derived from the Full Care populations had lower brood survival than inbred families drawn from the No Care populations (log(OR) =1.12 [0.49-1.80], z=3.42, p<0.001) – though only when broods were raised in a No Care current environment. No equivalent differences were observed in the Full Care current environment (log(OR)=0.20 [-1.4,1.9], z=0.25, p=0.80). For the outbred families, the evolutionary history of the lineage had no effect on breeding success, though broods were in general less successful when they received no post-hatching care (Table 2).
To further test prediction (3), we expanded our analyses to consider all generations, beginning by calculating the extent of inbreeding depression at each generation. Inbreeding depression was greater in Generation 1 for families descended from the Full Care evolving populations than the No Care evolving populations, in the No Care current environment (Figure 2A). We found the same pattern in the Full Care current environment – though here the differences between lineages were first seen at Generation 3 (Figure 2A).
Finally, we compared the survival of all lines across generations in the Evolutionary History Experiment, by fitting accelerated failure time hazard models (Figure 2B). A lineage was considered extinct if none of its members survived to reproduce in the subsequent generation. Whilst all inbred lineages in our experiments eventually went extinct, outbred lineages were still reproducing successfully at the point at which the experiment was terminated (Supplementary Figure 2, Supplementary Table 2).
For the inbred lineages, there was once again an interaction between the evolutionary history of a population and current care received (Supplementary Table 1). This resulted, in part, from a No Care current environment causing particularly rapid extinction (Figure 2B; Supplementary Table 2). When there was No Care, inbred lineages seeded from the Full Care evolving populations had significant lower median survival than inbred lineages seeded from the No Care evolving populations (Estimate=0.20 [0.05-0.36], p<0.01; Figure 2B). Lineages seeded from the Full Care evolving populations reached 50% extinction one generation sooner under a No Care environment than inbred lineages seeded from the No Care evolving populations (non-parametric Kruskal Wallis test: H(1)=4.59, p=0.03; Supplementary Table 2). In a Full Care environment, by contrast, we could detect no equivalent difference in lineage survival between the No Care and Full Care populations (Estimate=-0.01 [-0.13-0.10], p=0.85; Figure 2B; Supplementary Table 2).
Discussion
Burying beetles care for their offspring by making a nest for them to inhabit during development, providing them with plentiful carrion to feed upon and defending them from attack by rival microbes and animals [20]. Our experiments show that the supply of post-hatching care is sufficient to perturb the mutation-selection balance by relaxing selection -as predicted generally by evolutionary theory [1-5]. We cannot tell from our experiments whether selection is relaxed because the primary beneficiaries of care are kin [3] or because parental care more generally buffers against harsh environments and so weakens the effects of natural selection [14], or both.
By supplying care, parents shield their young from relatively harsh environmental conditions: larvae receiving parental care had higher survival than those that had no care. Indeed, we found that when parents provided care, the developmental environment was sufficiently benign, and the strength of selection then sufficiently weak, that diverse genetic variants were able to survive - even those that were inbred, just as previous work has shown [19]. Consequently, after 20 generations of experimental evolution in these contrasting environments, we found that the Full Care populations carried a greater mutation load than the No Care populations (confirmed in a companion paper [29] which uses SNPs to quantify the extent of genetic variation in the two types of experimental population). The difference between the populations was especially pronounced during the first generation of inbreeding, and most readily detectable when inbred individuals were prevented from supplying care. This suggests that some of the additional mutations present in the Full Care populations were recessive and / or only mildly deleterious [5]. Given the relatively short timeframe of this experiment, we presume that these mutations were present in the founding populations of wild-caught beetles but were removed from the No Care populations by selection acting more strongly against them. In this sense, our findings are similar to previous work on Tribolium which found that deleterious genetic variation was purged when populations were exposed experimentally to more intense sexual selection [30].
Although it is now well-understood why individuals evolve cooperative behaviour, the mechanisms that cause cooperation to persist and diversify remain relatively unclear [31]. Recent theoretical work suggests that positive feedback cycles could play a key role in entrenching cooperation, following its initial evolution [32]. Cooperative social interactions facilitate the transfer of beneficial microbes, for example, upon which social partners might then become dependent over evolutionary time, ensuring that cooperation must persist [e.g. 33-36]. Likewise, cooperative interactions can promote the division of labour between social partners, causing a degree of interdependence that ensures cooperation must continue [37]. Our results, together with those obtained by Pilakouta et al. [19], suggest a third mechanism through which cooperation can become entrenched, hinted at originally by Crow [2]. We have shown that parental care creates a problem (increased mutation load: our results) for which it is also the solution (enhanced survival of all genetic variants: [19], our results). By relaxing selection, parental care causes an increase mutation load which increases the population’s dependence upon care. Care ensures that the diverse genetic variants, whose existence it has facilitated, are able survive until the end of development. This could explain why parental care has evolved more frequently than it has been evolutionarily lost [14]. As Crow [2] put it: ‘there is no turning back…A return to the original conditions leads to the immediate full impact of all the mutants that have accumulated during the period of improved environment”. In principle, this reasoning can be extended to any form of cooperation that relaxes selection. Indeed, Crow [2] made the argument originally in the context of environmental improvements in human societies and their effect on genetic variation.
Finally, we have focused on the immediate effects of parental care on genetic variation, but the longer-term consequences are still unclear and need not match the effects seen in the short-term. For example, although greater intensity of intrasexual selection is beneficial in the short term, because it purges deleterious mutations from the population [30], in the longer run more intense intrasexual selection can make lineages more prone to extinction [38]. This might be due to a lack of beneficial genetic diversity. Likewise, although parental care enables mildly deleterious mutations to persist in the short-term, perhaps in the longer-term it builds up genetic diversity that could be beneficial and underpin rapid evolution, especially if environmental conditions change suddenly, or if mutations promote novelty through compensatory evolution [26]. In future work, it would be interesting to isolate the longer-term effects of parental care on genetic diversity and the effects it might have on the evolutionary resilience of wild populations in a changing world [39].
Funding statement
This project was supported by a Consolidator’s Grant from the European Research Council (310785 Baldwinian_Beetles), by a Wolfson Merit Award from the Royal Society, The Leverhulme Trust (RPG-2018-232) and The Isaac Newton Trust (18.23(q)), each to RMK. RM was supported by a Biotechnology and Biological Sciences Research Council Future Leaders Fellowship (BB/R01115X/1). HS was partially supported by the Japan Society for the Promotion of Science (KAKENHI Grant Number: JP19K21569).
Author contributions
Conceived the idea and designed the Evolutionary History Experiment: RMK, SP; Performed the Evolutionary History Experiment: SP; Collected data from Experimentally evolving populations: RM; Analysed the data: RM, HS; Wrote the manuscript: RMK, RM. All authors discussed the results and commented on the manuscript.
Competing interests
The authors declare no competing financial or nonfinancial interests.
Data and materials availability
Data and code needed to evaluate the conclusions in the paper are available from https://github.com/r-mashoodh/nves_MutationLoad.
Acknowledgements
We thank Sue Aspinall and Chris Swannack for helping with beetle maintenance. We also thank Benjamin Jarrett and Darren Rebar for collecting data from the experimental evolution.
Footnotes
Sonia Pascoal Sonia.pascoal{at}bioresource.nihr.ac.uk
Hideyasu Shimadzu H.Shimadzu{at}lboro.ac.uk
supplementary information updated (outdated version uploaded previously)