Causes and consequences of the evolution of reproductive mode in Caenorhabditis nematodes

Abstract

Reproduction is directly connected to the suite of developmental and physiological mechanisms that enable it, but how it occurs also has consequences for the genetics, ecology, and longer-term evolutionary potential of a lineage. In the nematode C. elegans, anatomically female XX worms can self-fertilize their eggs. This ability evolved recently and in multiple Caenorhabditis lineages from male-female ancestors, providing a model for examining both the developmental causes and longer-term consequences of a novel, convergently evolved reproductive mode. Here we review recent work that implicates translation control in the evolution of XX spermatogenesis, with different selfing lineages possessing both reproducible and idiosyncratic features. We also discuss the consequences of selfing, which leads to a rapid loss of variation and relaxation of natural and sexual selection on mating-related traits, and may ultimately put selfing lineages at a higher risk of extinction.

Developmental causes and organismal consequences of reproductive mode

Since its rebirth roughly 30 years ago^2,3, a major goal of evolutionary developmental biology (EDB here, but often referred to as ‘evo-devo’) has been the discovery of the empirical developmental and genetic details of how anatomical novelty arises. This is difficult work, and convincing connections between the evolution of DNA sequence and morphology have only been forged in the last decade. With the first apparent victories now won, general principles are being inferred^7–9. Much of recent work on animal EDB has been focused on somatic traits, but germ cells also evolve, and merit study for several reasons. First, changes in germ cell properties often underlie many fascinating adaptations¹¹. These include abrupt shifts in reproductive mode, such as loss of a feeding larva¹⁵ or changes in breeding systems¹⁷. Second, germ cell development is unusually dependent upon post-transcriptional gene regulation^18,19, in contrast to the transcriptional and coding changes that have been repeatedly implicated in the evolution of somatic traits⁹. These shifts also have large and immediate impacts on transmission genetics, effective population size, ecology, and macro-evolutionary potential. As a result, research models developed for investigating the EDB of reproductive evolution can also be used to study its influence on other biological processes.

Caenorhabditis nematodes are well suited for investigating the causes and consequences of reproductive mode shifts because C. elegans is an intensely studied model organism, and its androdioecious mating system (males and self-fertile hermaphrodites; see Glossary Box) evolved recently from gonochoristic (male and female) ancestors²⁶. Androdioecy has evolved at least three times in the genus (Figure 1), and in each case the overt anatomical modifications required for evolution of self-fertility were confined to the germ line. Therefore, Caenorhabditis can serve as a tractable model for how germ-line adaptation works and for addressing repeatability in adaptive evolution. In addition, t he advanced state of Caenorhabditis genomics^33,34 supports investigation of the role sexual mode plays in shaping genome content and variation. Below, we describe recent advances in both of these areas. First, we summarize recent studies that point to an important role for the control of mRNA translation in the evolution of XX spermatogenesis. Then we address other new work that highlights the major behavioral and genomic consequences of the evolution of self-fertility in Caenorhabditis.

Figure 1. Phylogenetic relationships of Elegans group Caenorhabditis.

The most parsimonious scenario posits that androdioecy evolved convergently in three lineages from a common gonochoristic ancestor. This is supported by the polarity of changes in sex-related traits (see Table 1). The outgroup species, C. japonica, belongs to the Japonica group. Cladogram is modified from Ref. ²⁶.

Developmental causes of self-fertility in Caenorhabditis

Caenorhabditis employs genetic sex determination; males have a single X chromosome and the XX sex is either female or hermaphroditic, depending upon the species. Hermaphrodite spermatocytes differentiate in a typically female somatic gonad and maintain their XX karyotype. A compelling developmental biology question, therefore, is how limited spermatogenesis is achieved in any one hermaphrodite species. From the standpoint of EDB, however, answering this question immediately leads to several others: Which aspects of the mechanisms required for limited XX spermatogenesis first evolved in hermaphroditic lineages? Which of these mechanisms are ancestral? Are those changes that are associated with self-fertility similar or different in convergent species? These questions have stimulated comparative studies of germline sex determination in Caenorhabditis.

Of the many C. elegans mutations with germline-specific sex determining phenotypes (see Box 1), only those that affect hermaphrodites are of particular interest here. The characterization of these mutants revealed a major role for translational regulation in germline sex determination. The feminizing tra-2(gf) and masculinizing fem-3(gf) alleles alter small regulatory elements in the 3’ untranslated region (UTR) of each gene, the Direct Repeat Element (DRE)³² and Point Mutation Element (PME)²⁹, respectively. The loss-of-function phenotypes of the RNA-binding proteins that recognize these elements (GLD-1 for tra-2³⁵, FBF for fem-3⁴⁰) are consistent with their repressive function. The F-box protein FOG-2 has been indirectly implicated in mRNA regulation, as it complexes specifically with GLD-1⁴¹. These factors define a set of elements that are necessary for limited hermaphrodite spermatogenesis in C. elegans, and thus are prime subjects for comparative studies.

BOX 1: An overview of germ-line sex determination in C. elegans.

The study of C. elegans sex determination helped pioneer developmental genetics¹. Decades of work has shown that a two-fold difference in X dosage is first converted to a binary signal by xol-1⁴, and through a series of interactions this is converted into differential abundance of the zinc-finger transcription factor, TRA-1^5,6. In males, TRA-1 levels are kept low by the combined action of FEM-1, FEM-2, and FEM-3, which form a complex that targets TRA-1 for ubiquitination and degradation by the proteosome¹⁰. TRA-1 abundance is sufficient to explain essentially all sexually dimorphic somatic traits, but in the germ line things become more complicated. First, male tra-1 mutants cannot sustain spermatogenesis and frequently produce oocytes^12,13. In addition, tra-1; fem double mutants have an unexpected feminized (Fem) phenotype in the germ line (but not the soma)^13,14. These results are not fully understood, but are consistent with tra-1 having both positive and negative effects on spermatogenesis, and with it being only one of a larger set of germline sex determiners.

Germline-specific sex determination mutations have also been discovered¹⁶, generally falling into “no sperm” (feminization of germline, or Fog) and “too many sperm” (masculinization of germline, or Mog) categories. Loss-of-function mutations in fog-1 and fog-3 eliminate both male and hermaphrodite sperm^20–22, whereas fog-2 mutations only feminize hermaphrodites²³. In contrast, the mog-class loss-of-function mutations produce a vast excess of hermaphrodite sperm^24,25. Loss of gld-1 and fbf-1/fbf-2 activity also ablates or overproduces XX sperm, respectively, but with pleiotropic defects in hermaphrodites and no effect on male sex determination^27,28. Through clever genetic screens, gain-of-function mutations affecting the 3’ UTRs of tra-2 and fem-3 with germline-specific effects were also found^29–32. Eventually it was discovered that the loss-of-function gld-1 and fbf-1/2 phenotypes stemmed from an inability to bind RNA motifs defined by the tra-2 and fem-3 gain-of-function alleles^35,36. FOG-1^21,22 and several of the MOG proteins^37–39 are also implicated in mRNA regulation.

The global sex determination pathway of C. elegans is conserved across at least the Elegans group of Caenorhabditis, as are sperm-promoting factors acting downstream of tra-1⁴². An exception, however, is fog-2, a recently duplicated gene specific to C. elegans⁴³. FOG-2 is unique among its closest C. elegans paralogs in affecting sex determination, and also in possessing a GLD-1-binding domain in its divergent C-terminus⁴³. Presumably FOG-2 is involved in the post-transcriptional regulation of tra-2 mRNA. Thus, fog-2 is a new gene with a new function, and it is a strong candidate for a key step in the gain of XX spermatogenesis in the C. elegans lineage. Interestingly, the C. briggsae-specific F-box gene she-1 is also necessary for hermaphrodite spermatogenesis at elevated temperatures⁴⁴. In contrast to FOG-2, however, SHE-1 does not interact with C. briggsae GLD-1. Thus, although fog-2 and she-1 are both implicated in hermaphrodite spermatogenesis, they probably exert their roles through different mechanisms. Characterization of these mechanisms is an important area for future research.

Functional evolution of RNA-binding proteins and translational control

As noted above, C. elegans hermaphrodite germline sex determination requires the translational control of tra-2 and fem-3 expression. Early comparative studies suggested that the 3’ UTR elements required for these controls were conserved, even in species that were gonochoristic^45–47. More recent work has focused on the evolution of the trans-acting RNA-binding proteins that recognize these elements. As a result, an evolutionary picture of the translational control networks that regulate germline sex determination is emerging, one that suggests both quantitative and qualitative changes matter for producing novel phenotypes.

Besides promoting XX spermatogenesis, C. elegans gld-1 is also necessary for germline tumor suppression, oocyte maturation and meiotic progression^28,48. It has recently been shown that gld-1 plays these latter roles in females of gonochoristic species (e.g. C. japonica, C. brenneri, C. remanei) as well, but it has no discernible impact on sex determination⁴⁹. This suggests the ancestral role of gld-1 was to regulate oogenesis, and that it was been coopted into sex determination recently. Remarkably, C. briggsae gld-1 has also acquired a role in hermaphrodite sexual patterning, but as a limiter of sperm production^43,49. Thus, gld-1 has been coopted into hermaphrodite sex determination at least twice, which suggests some reproducibility in evolution. However, it functions in opposite manners in two species, which indicates an important role for chance or the influence of initial conditions in incipient hermaphroditic lineages.

How can Ce-gld-1 and Cbr-gld-1 have opposite effects on germline sex? Cross-species rescue experiments rule out the possibility that gld-1 itself is the locus of functional evolution⁴⁹. One clue is that Ce-GLD-1 associates strongly with tra-2 mRNA in vivo, as expected, but Cbr-GLD-1 does not⁴⁹. Furthermore, the Cbr-tra-2 transcript lacks DREs⁴⁹ and has fewer GLD-1-binding elements than Ce-tra-2^32,50. This suggests that gld-1 has divergent functions, at least in part, because of lineage-specific evolution of target mRNA 3' UTRs. The robust association of GLD-1 with tra-2 mRNA in C. elegans may synergize with the ability of its cofactor, FOG-2, to recruit an E3 ubiquitin ligase complex¹⁰ (Figure 2). This has been proposed to target translation-stimulating factors associated with the 3’ end of the mRNA for degradation⁵¹.

Figure 2. Evolution of the tra-2 translational control module in C. elegans hermaphrodite sex determination.

Orthologs of tra-2 are found across Caenorhabdtis, and some level of GLD-1 binding to tra-2 mRNA may be ancestral^45,47. This starting condition is depicted in the upper panel, in which GLD-1 is bound, but does not interfere with the interaction between poly-A-binding protein (PABP) and eIF4F, the protein complex that binds the 7-methylguanosine cap (hexagon). As a result, translation of tra-2 occurs at an appreciable level in germ cells. In C. elegans (lower panel), the tandemly duplicated Direct Repeat Elements³² encode 3 STAR-binding elements (SBEs)⁵⁰. The DREs are unique to Ce-tra-2 and may stabilize the GLD-1-tra-2 mRNA complex in vivo⁴⁹. Synergizing with this is a C. elegans-specific F-box-containing cofactor, FOG-2^41,43, which likely recruits an E3 ubiquitin ligase. Though the targets of FOG-2-mediated ubiquitination remain unknown, one plausible scenario is that they are translation-promoting factors, such as PABP. As a result, poly-A-tails are shortened and translation is strongly inhibited. Loss of FOG-2, GLD-1, or the SBEs thus leads to major TRA-2 over-expression and germline feminization.

In C. elegans, the factors regulating fem-3 translation are the PUF family RBPs FBF-1 and FBF-2⁴⁰. FBF-1 and FBF-2 also indirectly promote female fate by serving as negative regulators of GLD-1 expression. C. briggsae lacks strict fbf orthologs, but members of the related PUF-2 sub-family (Cbr-puf-1.2 and Cbr-puf-2) have nevertheless been implicated in C. briggsae hermaphrodite germline sex determination. Pair-wise RNAi knockdown of either Cbr-puf-1.2 and Cbr-puf-8⁴⁹ or Cbr-puf-1.2 and Cbr-puf-2⁵² both feminize the hermaphrodite germ line. This is in complete opposition to the roles of fbf-1, fbf-2, and Ce-puf-8, which are redundantly necessary for hermaphrodite oogenesis in C. elegans^40,53. The unexpected sperm-promoting roles of Cbr-PUF-2 and Cbr-PUF-1.2 can be explained by their ability to negatively regulate the Cbr-gld-1 transcript, which (as noted above) promotes female fate in C. briggsae⁵². Translation of gld-1 appears to be regulated by PUF proteins in all Caenorhabditis species⁵², probably indicative of an ancestral role in the regulation of germ cell proliferation (rather than sex). Thus the FBF and PUF-2 sub-families were apparently “coopted by association” in each hermaphrodite due to their long-standing linkage with gld-1.

Possible genetic complexity of self-fertility in Caenorhabditis

The above studies have deepened our understanding of how hermaphrodite development evolved in the C. elegans and C. briggsae lineages, and they have even guided a successful effort to engineer selfing in C. remanei females⁵⁴. However, they have not defined the historical genetic variants distinguishing females from hermaphrodites with certainty. More generally, the comparative candidate-gene approach is problematic when the trait is not completely understood in the model system¹⁶ and when taxa are substantially diverged⁵⁵. Because simple genetic changes can produce dramatic sexual transformations, and because self-fertility has evolved repeatedly (Figure 1), it may be expected that only a few changes are necessary for hermaphroditic self-fertility to evolve. However, it is also possible that hermaphroditism evolves through multiple changes of moderate to minor effect. Quantitative variation in self brood size^56–59 and cryptic variation in somatic sex determination⁶⁰ exist in C. elegans. Furthermore, C. elegans sex determination mutations that disrupt wild-type self-fertile hermaphroditism on their own can produce fertile animals in combination³⁰. Thus, the use of natural variants is a desirable alternative in determining the causes of self-fertility.

Recent work on interspecies Caenorhabditis hybrids with divergent reproductive modes has shown the utility and limitations of this approach. The F1 hybrid progeny of C. briggsae and C. sp. 9 (a recently discovered gonochoristic species) are fertile females⁶¹. That self-fertility is recessive indicates that no C. briggsae gene is sufficient for self-fertility at a single dose, and that that the C. sp. 9 XX germline may be highly canalized in the female state. An important aspect of the evolution of self-fertility, therefore, may have been a weakening of the commitment to female germ-line sexual identity⁶². Certain informative backcrosses in the C. briggsae-C. sp.9 system are inviable, but those that can be performed are unable to regenerate self-fertility⁶¹. This result is consistent with a requirement for homozygosity of C. briggsae alleles of multiple unlinked genes. Hybridization therefore eliminates self-fertility, suggesting that incipient hermaphrodites achieved the requisite fixation of multiple genetic variants promoting XX spermatogenesis via isolation and inbreeding. A quantitative genetic approach for understanding the evolution of self-fertility may be a fruitful one for future studies.

Consequences of the evolution of selfing in Caenorhabditis

Self-fertility may provide a way to invade a new ecological niche or survive drastic times, but it also has major consequences for genome content and sexual traits. Research that collectively highlights these consequences is summarized below.

Population genetics

Selfing leads to progressive homozygosis, which in turn unmasks deleterious recessive mutations and often results in a decline in fitness called inbreeding depression⁶³. In order to survive, an incipient selfing population must rid themselves of residual deleterious mutation through a process called purging. Because the fitness effects of alleles often depend upon epistatic effects of alleles at other loci, this process is expected to eventually produce unique homozygous genotypes. Consistent with this prediction, selfing Caenorhabditis nematodes do not suffer from general inbreeding depression^64,65, but instead suffer from outbreeding depression⁶⁶. This is in stark contrast to their gonochoristic congeners, which display severe inbreeding depression^66,67.

The succession of inbreeding depression and purging of deleterious mutations predict that self-fertilizing species should retain less genetic diversity than dioecious counterparts. Indeed, higher selfing rates lead to a decrease of the effective population size for neutral alleles of autosomal genes⁶⁸. The neutral theory of evolution predicts that outcrossing species will display at least twice as much genetic diversity as selfing species⁶⁹, and other factors are likely to reduce genetic diversity in selfing species further^70–72. For example, severe population bottlenecks are facilitated by the fact that any single individual can found a new population. Moreover, progressive homozygosis leads to extensive linkage disequilibrium across the genome of selfing species. The impact of natural selection is then limited because both deleterious and advantageous loci are effectively linked to one another.

The above dynamics might lead to reduced selection against weakly deleterious mutations and thereby accelerate rates of evolution in sequences located in non-recombining parts of the genome. By contrast, selection for a new beneficial mutation would be expected to sweep a large swath of the genome along with it, lowering population-level variation. Consistent with the latter, dramatically less nucleotide diversity is observed in the genomes of selfing Caenorhabditis compared with gonochoristic Caenorhabditis^73–75 and is lowest in the low-recombination central domain of the C. elegans chromosomes⁷⁶. Similar findings have been described in plants (e.g.^77,78).

In addition to lower genetic diversity, the smaller effective population sizes of selfing should lead to weaker selection against various forms of selfish elements, which would in turn inflate genome size⁷⁹. This is consistent with observations of the abundance of transposable elements (TE) in low-recombining regions of the Drosophila melanogaster genome⁸⁰ and a higher copy number of certain TEs in Arabidopsis thaliana⁸¹ and C. elegans⁸² compared to their obligately outcrossing congeners. However, more recent studies contradict some of the above predictions. In Arabidopsis, the genome size of the selfing A. thaliana is smaller than that of outcrossing relatives^83–85. Furthermore, in the obligately outcrossing A. lyrata, TEs are more active and younger than in the selfing A. thaliana⁸⁶. In Caenorhabditis, the genome sequences of C. elegans and C. briggsae also harbor fewer repeats than those of three gonochoristic species⁸⁷.

Ten Caenorhabditis species have or will soon have their genomes completely sequenced, annotated and available^34,88. Preliminary assemblies of the gonochoristic C. japonica, C. remanei and C. brenneri genomes suggests they have larger genome sizes than C. elegans and C. briggsae, even when the substantial amounts of heterozygosity retained in the sequenced strains⁸⁹ is considered. While genome sizes and the emergence of selfing could be unrelated, the intriguing possibility of repeated genome shrinkage in selfing lineages remains. If this is borne out by more direct measures, it may be driven by an interaction between partial selfing and the preferential segregation o f deletion-bearing autosomes with the X chromosome in male meiosis⁹⁰.

Role of males and outcrossing

Though neither C. elegans nor C. briggsae hermaphrodites need to mate with males to reproduce, males exist in wild populations. They are produced spontaneously from meiotic X chromosome non-disjunction or through fertilization of XX individuals by males, the latter allowing sex ratios of up to 50%. In laboratory cultures of the C. elegans reference strain, N2, the male frequency is close to that of X chromosome non-disjunction rate^64,91. Nevertheless, outcrossing can be detected in wild populations of C. elegans and C. briggsae^92–96, and in some isolates of C. elegans males persist at a higher rate than in N2 and vary widely in their abilities to mate^57,91,96. At the same time, a natural population of C. elegans carrying a mutation in mab-23, which renders the males unable to mate, has been reported^57,97, illustrating the fact that males are not needed, at least for short-term survival of the population.

Why males are still retained in androdioecious species is a subject of open debate^64,98–100. In laboratory environments, C. elegans hermaphrodites can invade gonochoristic populations generation after generation, even when starting with a population composed of nearly 50%males^64,91,99 (though this may not be true for C. briggsae⁴⁴). While this suggests that androdioecious males are often useless and perhaps on their way out of existence, another theory is emerging. When environmental conditions are not optimal (i.e., scarce food resources, overpopulation and unfavorable temperatures), Caenorhabditis nematodes have the ability to go into a resistant larval phase called the dauer. Males survive dauer arrest better than hermaphrodites, and outcrossing rates increase after starvation¹⁰¹. In addition, in populations subjected to environmental stresses or an increased mutation load, the progeny resulting from crosses between males and hermaphrodites have a higher fitness than those from selfing¹⁰². Thus, outcrossing allows both avoidance of inbreeding depression and more rapid adaptation to new environments for C. elegans - as is the case for most species. This also squares well with theoretical calculations that suggest C. elegans would go extinct in a few thousand years were it to lose all ability to outcross¹⁰³.

Degradation of mating behavior—the selfing syndrome

Mating in C. elegans relies on sex-specific anatomy and on chemosensory and neuronal signals. Briefly, upon contact between the rays and ventral sensilla of the male tail with the hermaphrodite cuticle, the male presses his tail against the hermaphrodite‘s body and starts scanning the hermaphrodite to find the vulva. When the male tail eventually makes contact with the vulva, the male stops moving and inserts his spicules in the vulval slit. The male then ejaculates and deposits a gelatinous copulatory plug. Larger, male-derived sperm is stored in the hermaphrodite spermathecae along with the hermaphrodite‘s own sperm, where it has a competitive advantage because of its larger size^104,105. In the absence of a mate, males wander away from food source, presumably in search of mates¹⁰⁶.

Because mating is not required for reproduction in androdioecious species, mating-related traits should be under weaker stabilizing and sexual selection. Indeed, in C. elegans and C. briggsae, both sexes have lost their ability to behave as reliable mating partners in comparison to gonochoristic Caenorhabditis species (Table 1)^{64,99,107,108}. Most laboratory crosses between males and hermaphrodites in C. elegans or C. briggsae use mobility-impaired hermaphrodites or a large excess of males to make up for the low frequency of successful matings. Furthermore, males from androdioecious species do not discriminate between the sexes, and tend to initiate mating behaviors upon contact with any cuticle, including their own¹⁰⁹.

Table 1.

Comparison of mating-related traits in androdioecious and gonochoristic Caenorhabditis.

	gonochoristic	androdioecious
mate discrimination	excellent109	poor109
pheromone potency	high107,110	low107,110
female immobilization	yes109	no109
mating length	~40 min.64	2 min. 64
mating efficiency	100%64	5% 64
copulatory plug	always	polymorphic57,118

Several steps in the mating process a re degraded in androdioecious species. Females from dioecious species attract conspecific males as well as C. briggsae and C. elegans males, while hermaphrodites do not elicit such a reaction from males from any species^64,107,109. C. elegans hermaphrodites do produce a mixture of characteristic glycosides, the ascarosides, that attract both conspecific males and males from other species to varying lesser extent¹¹⁰. Females from C. remanei and C. brenneri secrete a sex pheromone to attract males that seems to be either not produced by hermaphrodites or is less efficient¹⁰⁷. It is unclear whether the attractants secreted by the gonochoristic females are chemically similar to those secreted by C. elegans hermaphrodites. In addition, both males and females in dioecious species wander in the absence of potential mate, while in selfing species hermaphrodites do not search for mates¹¹¹.

Once males from gonochoristic species have found a female partner, they are able to immobilize it to facilitate spicule insertion and subsequent copulation. It has been suggested that males produce a ‘soporific factor’ that allows them to immobilize the female and facilitates spicule insertion by widening the vulval opening¹⁰⁹. However, hermaphrodites seem to be unresponsive to this signal, and most of the matings fail because the hermaphrodite moves away from its partner before copulation occurs. Interestingly, C. briggsae males induce the same inactive behavior in females from both gonochoristic species tested, whereas C. elegans males do not¹⁰⁹. Similarly, a strain of C. briggsae otherwise able to mate seems to lack sex-drive and displays neither mating nor self-plugging behaviors¹⁰⁹. In addition, hermaphrodites have been observed to eject male sperm if their reproductive tracts already contain their own, particularly if they are young¹¹². This behavior is currently unexplained, and stands at odds with the well-documented fertilization advantage experienced by male sperm in a mixed brood^105,113. However, both may true, especially if ejaculate ejection can be overcome by repeated mating.

The above studies indicate that androdioecious s pecies have lost (or are in the process of losing) characteristics that would otherwise make them efficient cross-fertilizers. The genes underlying these traits and/or their regulation may have evolved in response to relaxation of selection¹⁰⁸ and some are still responsive to selection¹¹⁴. However, given that androdioecious species have clear mating defects, some historically important genes associated with mating success may be more readily identified in a gonochoristic species such as C. remanei, work that is currently underway (CGT and ESH, unpublished).

Macroevolutionary consequences

The complex molecular biology and repeated evolution of self-fertility in Caenorhabditis suggests that it is an adaptation, perhaps driven by selection for reproductive assurance at low density. Selfers can also enjoy a large boost in intrinsic growth rate once males, who cannot produce eggs, are largely eliminated from the population. The power of this can be seen in laboratory cultures of the obligately outcrossing C. elegans fog-2 mutant, which have been repeatedly observed to revert to selfing by rare gene conversion events that repair the mutation¹¹⁵ and restore self-fertility.

Despite the above advantages, however, across the nematode family Rhabditidae multi-taxon clades of selfing species are rarely seen¹¹⁶, and within Caenorhabditis all three known cases are clearly recently evolved from gonochorism^26,72. This suggests that conversion to selfing has powerful short-term benefits to species that can evolve XX spermatogenesis and overcome the resulting inbreeding depression, but in the long run extinction generally occurs before speciation. A similar macroevolutionary cost of selfing has been observed in plants of the nightshade family¹¹⁷ and may occur in many taxa that colonize disturbed or ephemeral habitats. The above results suggest the paradoxical result that selfing nematodes need to maintain the ability to outcross to withstand stress and on going mutation, yet at the same time, standing variation falls to very low levels and mating-related traits are gradually degraded. The eventual endpoint may be selfing species that are so poor at mating they cannot respond to an environmental shift and succumb. From this macroevolutionary perspective, the impressive achievement of hermaphrodite development appears to be a Faustian bargain.

Concluding Remarks

In this review, we have shown how Caenorhabditis nematodes represent a useful experimental system for the integrative study of shifts in sexual mode. Early comparative studies on the EDB causes of selfing focused on the genetic control of sex determination, and important insights were gleaned from this work. A promising new area will be to examine the consequences of selfing, both for the genome overall and for reproductive traits. As the natural history and ecological niche of the genus is being refined²⁶, we can also start to address another outstanding question—what ultimately explains the phylogenetic distribution of selfing. An adaptationist hypothesis would be that it is a facile transition driven by ecological shifts to a particular lifestyle. However, it is also possible that gonochoristic lineages have a more or less constant probability of spinning off selfing derivatives, which derive short-term gains but then go extinct. In any case, we can look forward to many interesting connections being made.

Glossary Box

Androdioecy

a sexual system with males and hermaphrodite sexes.

Ascarosides

nematode-specific glycosides (modified sugars) that serve as signals for various reproductive decisions.

Congener

member of the same genus.

Conspecific

member of the same species.

Convergent evolution

evolution of an outwardly similar trait in multiple lineages, whose common ancestor did not possess the trait.

Cooption

use of a preexisting feature (gene, organ, etc.) for a new purpose.

Gonochory

a sexual system with separate male and female sexes.

Selfing

hermaphroditic reproduction through union of gametes from the same individual.

Outbreeding and inbreeding depression

loss of viability or fecundity of progeny of mating between distantly or closely related individuals, respectively.