From phenotype to genotype: the precursor hypothesis predicts genetic influences that facilitate transitions in social behavior

Abstract

Parental care is expected to be one of the key evolutionary precursors to advanced social behavior. This suggests that there could be common genetic underpinnings to both parental care and sociality. However, little is known of the genetics underlying care. Here, we suggest that ethological predictions of behavioral precursors to care along with a genetic toolkit for behavior provide testable hypotheses and a defined approach to investigating genetics of sociality. We call this the ‘precursor hypothesis’.

Introduction

Fascination among biologists with the evolution of social behavior can be traced back to Darwin. Social behaviors, especially those appearing to be altruistic, presented a special problem for the theory of natural selection based on costs and benefits to an individual [1]. Furthermore, social behavior is often associated with elaborate structures and interactions, traits that led him to refine the theory of natural selection by developing the concept of sexual selection [2]. Since then, these theoretical considerations have closely dovetailed with the persistent desire to understand our own species; social interactions more than any other trait define humans, and in explaining their prevalence in animals we can inch closer to comprehending the forces responsible for molding our own natures.

As with other complex biological phenomena, the study of social behavior can be broken down into two separate but intertwined questions: why did it evolve, and by what mechanisms is it generated? Regarding the first question, others have written extensively on the nature of selection and ecological conditions for social evolution [3–7], and we touch on these thoughts here only when they inform specific predictions. Instead, we are emphasizing the proximate genetic changes associated with evolution along a social continuum. Our logic is simple: without understanding of the genetic basis of sociality, we lack any understanding of how selection on social behavior is filtered through inheritance to result in evolution. We adopt an explicitly hierarchical framework in approaching this problem. Sociality is an evolutionary outcome, which is generated through interactions of several distinct behavioral modules (e.g. brood care, resource defense, mating, reproduction) [3,5]. In turn, each of these behaviors is itself a composite of many individually varying phenotypes, comprising even simpler behavioral traits and accompanying physiologies. Studying these granular phenotypes offers two clear advantages. First, they likely represent the true targets of natural selection leading to sociality. Second, they are necessarily more generalizable than higher-level traits, and thus amenable to a priori predictions about specific genes and mechanisms that might follow from ultimate evolutionary explanations.

Sociality begins (typically) with parental care

There is healthy debate over the different levels of sociality, and there can be confusion over what is and is not a social insect [8–10]. In its most restricted and least helpful usage, social insects refer specifically to a subset of Hymenoptera, and perhaps termites that have evolved eusociality. A more inclusive use allows other groups to be social and expands the understanding of sociality to different levels. This also opens up the potential use of multiple and diverse species to understand the molecular basis of social evolution [11]. Here, we follow the suggestion of Costa [10] and use social insects to refer to any taxon of insect where there are prolonged social interactions beyond mating.

Although social systems are characterized by an array of specializations in different categories of behavior, we focus on parental care for several reasons. Regardless of the scheme for classifying levels of sociality, parental care is often considered to be one of the first steps in the evolution of eusociality [3–5]. It is relatively easy to envision how parental behavior, simply through its expression in siblings and other individuals, could evolve toward the cooperative brood care generally required by eusociality [11]. Furthermore, there are strong hypotheses for why parental care has evolved. Specifically, parental care is thought to evolve as a result of selection for parents to counter environmental adversity and defend essential resources, typically food, for offspring [5], generating an inclusive fitness benefit to those offspring. Thus, understanding the evolution of parenting may provide a roadmap for understanding the evolution of both relatively simple and more complex forms of sociality.

The behavioral precursors of parenting

If we are to understand the social transition leading to parenting, it is first worth considering the nuances of how this behavior is constructed at the phenotypic level. Tallamy [12,13] argues that rather than being a ubiquitous response to adverse environmental conditions, parental care only evolves when there are specific behavioral precursors such as nest building, defensive postures, or aggressive behavior, already in place. This suggests parental care evolved as a behavioral ‘panda’s thumb’, through modification of existing traits rather than de-novo novelty.

The hypothesis that the evolution of parenting is contingent on precursor behaviors leads to clear predictions about the molecular foundations of parenting (Table 1, Figure 1). First, it suggests that no new genes or significant coding mutations affecting gene function need arise in a population, but rather that existing genetic variation and genotypes may be modified and selected. Currently, the strongest hypothesis for such a process is heterochrony, wherein the timing of gene expression evolves leading to the production of old behaviors in novel contexts [14,15].

Table 1.

Examples of genes in pathways predicted to be co-opted in the evolution of parental care

Food seeking/feeding	Social behavior/aggression	Mating/reproduction	Timing
Pkg1 (foraging)	It (inotocin [oxytocin])	Dsx (double sex)	tim
Pkg2	Itr (inotocin receptor)	Fru (fruitless)	(timeless)
Mvl (malvolio)	5-Htr-1 (serotonin receptor 1)	Tra2 (transformer 2)	per
Vit (vitellogenin)	5-Htr-2 (serotonin receptor 2)	Sifa (SIFamide)	(period)
Ast-B (allostatin B)	5-Htr-3 (serotonin receptor 3)	Hex (heramerin)
Ast-C (allostatin C)	Aad (aromatic L-amino acid decarboxylase)	Akh (adipokinetic hormone)	Clk (clock)
Ilr (insulin-like receptor)	Ada2A (adrenergic receptor)	Vit (vitellogenin)	cyc
Ilp-1 (insulin-like peptide 1)	Dnat (dopamine N-acetyltransferase)	Pvk (periviscerokinin)	(cycle) dbt (doubletime)
Ilp-2 (insulin-like peptide 2)	Oar-1 (octopamine receptor 1)	sNpf (short neuropeptide F)	cry
Npf (neuropeptide F)	Oar-2 (octopamine receptor 2)		(cryptochrome)
Tor (target of rapamycin)	Oar-3 (octopamine receptor 3)
Abl (tyrosine-protein kinase Abl)	Tbh (tyramine b-hydroxylase)
Akt1 (RAC serine-threonine kinase)	Dopr1 (dopamine receptor 1)
Chrb (charybde)	Dopr2 (dopamine receptor 2)
Sclla (scylla)
C190 (clip190)

Figure 1.

One set of hypotheses regarding the evolution of care via precursors, and its relationship to social classification. The dashed arrow represents the hypothesis that cooperative brood care evolved directly from parental care. Although this figure concerns only parental care, similar paths could be drawn for other categories of social behavior that define subsocial and eusocial insects.

The precursor hypothesis not only leads to predictions about the types of genetic changes that underlie parental care, but also the specific genes themselves [16] (Table 1, Figure 1). We expect that parental care evolves by coopting the genes or genetic pathways that influence these precursors [15,17^•]. For example, if parental care evolves first with defense of a food resource and mating around or on the resource, it would therefore follow that genes from aggression and defensive behavior pathways as well as genes from mating and reproduction pathways will be coopted. If further elaboration of parental care involves feeding your young rather than yourself, we predict that genes underlying the food acquisition pathway will be expressed differently to influence parenting. Additional precursors related to parenting could involve circadian behavior, sensory behavior, and locomotion, and their corresponding genetic systems. Precursor behaviors are necessarily more generalizable than the social behaviors they compose, and therefore allow us to draw on a broad range of empirical genetic results to inform a priori hypotheses about the involvement of specific genes. Each of the behaviors mentioned above has been studied thoroughly at the genetic level in the non-social model insect Drosophila melanogaster in addition to more isolated studies in non-model insects. Following this, we present a list of potential candidate genes for individual precursor behaviors (Table 1).

Tests of the precursor hypothesis: coopting reproductive and feeding systems

One of the first and most durable predictions of how evolution coopts preexisting phenotypes to evolve sociality is West-Eberhard’s Ovarian Ground Plan Hypothesis [18]. This hypothesis suggests that decoupling ovarian physiological cycles from behavior has allowed for the evolution of helping behavior in social insects; that is, resources and energy spent on reproduction is reallocated to helping. This was later generalized and expanded to include genetic changes [19] and named the Reproductive Ground Plan Hypothesis (RGPH). The RGPH specifically suggests regulatory gene networks underlying reproduction are co-opted to facilitate social behavior in worker castes. The RGPH provided the basis for suggesting the gene coding for the yolk precursor protein, vitellogenin (Vg), would be co-opted to affect behavior in non-reproductive social insects. Support for this prediction is strong, and, importantly, Vg expression is linked to parental care in both subsocial insects [20,21] as well as cooperative brood care in several lineages of eusocial insects (wasps — [22,23]; ants — [24^••,25^•,26^••]; bees — [27]). Correlation does not necessarily imply cooption; however, two results implicate a causal role for Vg on parental behavior. First, in burying beetles, gene expression changes are seen not only during female parental care but during male parental care as well, where, we posit, Vg is not likely influencing reproduction [21]. More directly, Kohlmeier et al. [24^••] knock down Vg using RNAi and observe a corresponding reduction of nestmate care behavior. Beyond Vg, recent support for the RGPH comes from a South African honey bee population where a naturally occurring nonsocial worker morph differs from social workers at the Ethr locus, a gene implicated in ovarian function [28^•].

The relationship between precursor behaviors and their underlying molecular pathways need not be one-to-one. The tool-kit hypothesis predicts that with reproduction, food acquisition should be a second pathway altered in social behavior [16,29]. This leads to a prediction that insulin signaling will be a part of the genetic tool-kit underlying parenting. Insulin is expected to be associated with feeding and energy acquisition behavior [30]. Variation in expression of genes in the insulin pathway is indeed associated with eusocial evolution in bees and ants [31,32^••], although it is unclear if insulin signaling plays a causal role in parental care specifically. Hunger and feeding behavior also have a well-documented relationship with neuropeptide signaling pathways distinct from (but not unrelated to) the insulin signaling pathway [33]. In burying beetles, multiple neuropeptides with feeding-related functions are related to parenting [17^•,34,35^••]. One of these, NPF [34], was previously implicated in honey bee social behavior [36], and behavioral roles for other neuropeptides continue to be demonstrated in other eusocial systems [26^••,37].

The precursor hypothesis raises a related question; while the pathways may be repurposed, how is it that the recipient of behavior changes from ‘self’ to ‘non-self’? That is, how does feeding oneself become feeding another, and how does directing aggression to protect oneself change to directing aggression to protect another? Chandra et al. [32^••] suggest that this change requires recognition and response to different cues, which further suggests that there must be changes in neural circuits controlling the link between cues and behavior. Thus, rather than responding to some internal signal of hunger, a parent must respond to signals of hunger from larval offspring. The response to these novel cues is expected to lead to alterations of neuropeptide expression. Consistent with this hypothesis, we see that neuropeptides are the most consistent genetic changes underlying all forms of variation or transitions in parenting in the burying beetle Nicrophorus vespilloides independent of cues of parental hunger [17^•,34,35^••,38]. Chandra et al.’s [32^••] prediction of the shifts in signals from self to larvae also provides a description for how further elaboration from subsocial to social to eusocial may occur; in this case, the ‘non-self’ cues recognized by mothers simply need to become recognizable by siblings or other nestmates.

Future research

The successes of the precursor approach to date suggest that the cooption of preexisting behavioral and physiological pathways play a role in facilitating transitions to parental behavior. However, the powerful a priori nature of this approach can be overestimated. Parental care, like other behaviors, is likely to be highly polygenic; is it possible that different genes in unexpected pathways contribute just as much or more to social phenotypes as predictable precursors [39]? What, if any, is the role of novel genes?

There is certainly evidence that multiple forms of genetic novelty can also play a role. Duplication of a gene may lead to subfunctionalization and neofunctionalization, and such changes can influence behavior [40]. We have found the gene Malvolio, which influences feeding decisions [41] and is known to influence honey bee social behavior [42], is duplicated in N. vespilloides. The two copies have different expression patterns during parenting and in different tissues [43]. More broadly, duplicated [44] and taxonomically restricted genes [45–47] appear to be overrepresented among genes related to social function, although-specific behavioral functions were not ascribed in these studies.

Unraveling the relative importance of novel genes versus co-opted genes will require a shift in methodological effort. Functional annotation of genes is paramount; the utility of genetics as a tool to understand the evolution of social transitions depends on the ability to classify genes, even coarsely, by their roles and activities. Currently, the number of unknown proteins and unannotated genes identified by RNA-seq experiments represent an untapped source of potential hypothesis tests about novel and predictable genes. Additionally, most studies to date have understandably been correlational, relying heavily on RNA-seq and measured expression of single genes by qRT-PCR. Moving forward, manipulating candidate genes and assessing their effects on specific behavioral components will be the strongest arbiter of that gene’s role in a social transition.

Finally, one of the most important considerations in investigating how the evolution of parental care relates to the evolution of sociality is in defining and appropriately phenotyping parental care. The precursor hypothesis implies a multivariate nature of parental care, but parenting is often described as a singular trait. There are several approaches that can help define parenting as a quantifiable phenotype rather than a description of an outcome. One we have taken is to examine genetic influences on natural variation in an isolated component of parental behavior [35^••], a method that has also been adopted for colony-level phenotypic variation in eusocial insects [48]. Another that has yielded promising results is manipulation of the social environment to induce changes in specific aspects of parental behavior [26^••,37]. Warner et al. [49^••] cleverly quantify gene expression simultaneously in caretakers and offspring in order to find genes specifically related to the components of offspring provisioning. Regardless of the approach, the usage of careful experimentation and nuanced phenotyping should improve the ability to determine if there are common sets of traits and genes involved in the evolution of parental care. By defining the universal mechanistic aspects of parental care in subsocial and eusocial insects, and the constituent traits, we will have taken a key step in understanding how parental care evolves and leads to further elaboration and sociality.