Why Are There Males in the Hermaphroditic Species Caenorhabditis elegans?

Corresponding author: King L. Chow, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong., bokchow{at}ust.hk (E-mail)

Communicating editor: D. CHARLESWORTH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The free-living nematode worm Caenorhabditis elegans reproduces primarily as a self-fertilizing hermaphrodite, yet males are maintained in wild-type populations at low frequency. To determine the role of males in C. elegans, we develop a mathematical model for the genetic system of hermaphrodites that can either self-fertilize or be fertilized by males and we perform laboratory observations and experiments on both C. elegans and a related dioecious species C. remanei. We show that the mating efficiency of C. elegans is poor compared to a dioecious species and that C. elegans males are more attracted to C. remanei females than they are to their conspecific hermaphrodites. We postulate that a genetic mutation occurred during the evolution of C. elegans hermaphrodites, resulting in the loss of an attracting sex pheromone present in the ancestor of both C. elegans and C. remanei. Our findings suggest that males are maintained in C. elegans because of the particular genetic system inherited from its dioecious ancestor and because of nonadaptive spontaneous nondisjunction of sex chromosomes, which occurs during meiosis in the hermaphrodite. A theoretical argument shows that the low frequency of male mating observed in C. elegans can support male-specific genes against mutational degeneration. This results in the continuing presence of functional males in a 99.9% hermaphroditic species in which outcrossing is disadvantageous to hermaphrodites.

THE stability of a dioecious species with equal numbers of males and females requires explanation. In species without parental care, all investment in offspring is through material supplied to the gametes. Females invest significantly more resources in their gametes than males do in theirs, yet males contribute an approximately equal number of genes. A parthenogenetic female or a self-fertilizing hermaphrodite that produces no male offspring could potentially double her number of grandchildren (MAYNARD SMITH 1978 ) and invade a dioecious population unless there are opposing selective forces. The problem of the maintenance of sexual outcrossing in a dioecious species is to identify and quantify these opposing forces. Mainstream theories divide possible forces into two broad categories: Either sexual outcrossing produces recombinant types that are better able to adapt to a changing environment, or sex more efficiently eliminates deleterious mutations (CROW 1994 ; HURST and PECK 1996 ).

Although general theories for the maintenance of outcrossing are desirable, nature consists of special cases and it may be fruitful to study thoroughly particular examples. Natural species may not only confirm or refute various theories, but may also add important details missed by the wide theoretical brush. The phylum nematoda provides us with two closely related species that may be relevant to the problem of the stability of the dioecious mating system: Caenorhabditis remanei and C. elegans. The former nematode worm is dioecious with equal numbers of males and females, and the latter is a self-fertilizing hermaphrodite with males present at low frequency (BRENNER 1974 ; BAIRD et al. 1994 ). In fact, the ability of C. elegans to both self-fertilize and outcross has rendered it a powerful genetic model for developmental studies, and well-established laboratory procedures as well as readily available mutants facilitate experimentation.

These two related species, C. remanei and C. elegans, are indistinguishable by their gross morphology, and their 18S rDNA sequences differ by only 1.2% (FITCH et al. 1995 ; D. H. A. FITCH, personal communication). Phylogenetic evidence suggests that their common ancestor was dioecious (FITCH and THOMAS 1997 ) so that an evolutionary study of these two species may serve as a useful model for understanding the maintenance of outcrossing against invasion by selfing hermaphrodites. Here, as a step in this direction, we consider the evolutionary status of C. elegans. Are C. elegans hermaphrodites descended from modified females capable of spermatogenesis that successfully invaded the ancestral dioecious species? If so, then why are there still males present in the C. elegans species?

Sex is determined in Caenorhabditis by the ratio of sex chromosomes to autosomes; females and hermaphrodites are XX and males are XO. With normal meiosis, females and hermaphrodites fertilized by males produce 50% males; self-fertilized hermaphrodites produce only hermaphrodites. Spontaneous nondisjunction of the sex chromosomes, however, can occur during meiosis (resulting in XX and O gametes) so that a self-fertilized wild-type hermaphrodite will produce males at low frequency. As observed in nature and in the laboratory, C. elegans males are present in populations at low frequencies, implying small levels of outcrossing between hermaphrodites and males. If males and hermaphrodites rarely mate, then why has natural selection not completely eliminated males from the species?

Two distinct possibilities present themselves. The first is that only a small amount of outcrossing may be required for sexual reproduction to yield an advantage (HURST and PECK 1996 ). A selfing hermaphrodite that produces no males may be selected against in competition with hermaphrodites that produce males at low frequency and occasionally outcross. The second possibility is that C. elegans is essentially a self-fertilizing hermaphroditic species descended from protohermaphrodites that successfully invaded a dioecious ancestor. Here, nondisjunction and the production of males at low frequency by the hermaphrodite are nonadaptive to the hermaphrodite. We can rephrase these two possibilities as the following question: Is the frequency u of male offspring produced by nondisjunction in a selfing hermaphrodite optimum or minimum? An analogous question has also been asked of mutation rates in general (MAYNARD SMITH 1978 ). If u is optimum, then natural selection has fine tuned its value to maximize the hermaphrodite's reproductive potential. Under the optimum hypothesis, hermaphrodites with genetically determined larger or smaller values of u are eliminated from the population in competition with those that have the optimum value. If u is minimum, however, then natural selection has favored hermaphrodites with genes that reduce u to a value as small as possible without incurring excessive costs in resources.

It is difficult to determine directly whether the value of u in C. elegans is optimum or minimum. In this article, however, guided by a mathematical model of the C. elegans genetic system, we perform laboratory observations and experiments that show that hermaphrodites have lost their attractiveness to males, resulting in a low frequency of outcrossing. We further argue that this implicates C. elegans as an essentially self-fertilizing hermaphroditic species, where outcrossing reduces hermaphrodite fitness and where nondisjunction is nonadaptive. Furthermore, any males present in the C. elegans population are under selection pressure to successfully mate with hermaphrodites, and we present theoretical results showing that the amount of outcrossing observed experimentally is sufficient to support male-specific genes against degeneration by deleterious mutations.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Worm cultures:
C. elegans strains were maintained and crosses were performed according to the standard laboratory procedures described by BRENNER 1974 . N2 was used as wild type. CB4088: him-5 (e1490) V, which gives a high incidence of male progeny, was used to obtain sufficient male worms for mating assays (HODGKIN et al. 1979 ). Both unc-24 (e138) and unc-17 (e245) mutants were used as markers in some of the assays. Genetic variants of C. elegans used in this study include AB1 (Australia). C. remanei (EM464, BAIRD et al. 1994 ) was used as the dioecious species for comparisons of mating behavior and efficiency.

Fecundity score:
Different genotypes of C. elegans were transferred individually onto an empty plate (one per plate). The brood size was scored over a period of 6 days with daily transfer of the parent. The number of progeny produced by each individual was averaged over the total test worms to reflect the fecundity of the parental worms of specific genotype.

Equilibrium cultures:
An equilibrium culture of him-5 or N2 worms was obtained by chunking a block of agar containing worms from an old plate onto a new plate seeded with bacteria once every 3–4 days over a period of >1 month. The incidence of males was scored over an additional six repeated passages of worms.

Observation of matings:
Mating behavior in an equilibrium population of C. elegans him-5 mutants and C. remanei was monitored by direct observation every 5 min under a dissecting microscope. Observed matings were timed. To obtain an estimation of the duration of matings, we doubled these timings because, on average, the observation period began halfway through the mating. For a further recording of the C. elegans male-hermaphrodite or C. remanei male-female association, five males were put on a fresh mating lawn (0.5 cm diameter) with five hermaphrodites or females. Their associations were timed and the averages from multiple observations were recorded.

Mating efficiency test for C. elegans:
Standard crosses were set up under two different conditions to address the impact of population density: (1) One male was paired with 20 hermaphrodites on a 5-mm diameter spot bacterial lawn (high density cross) or (2) the same combination of worms was cultured on a 9-cm² bacterial lawn (low density cross). The hermaphrodites and males were picked as young adults less than half a day old, unless otherwise noted. Mating tests were performed using active and nonactive (unc-24) hermaphrodites to address the impact of hermaphrodite mobility on mating success. These matings were allowed to take place for 2 days; subsequently, the male or non-Unc progeny was scored.

Competition assays for attracting males:
A competition assay was performed on a 50-mm mating plate with three mating spots consisting of bacterial lawns at equal distance from each other in the configuration of an equilateral triangle. A C. remanei female was put on one spot, a C. elegans unc-24 hermaphrodite on a second, and the third spot was left empty. Ten to 12 young males (C. elegans him-5 or C. remanei) were transferred to the center of the plate equally distant from the three bacterial spots. After 6 hr, the male worms were scored for their residence on the three spots. Plates where the female or hermaphrodite migrated from her home spot were discarded. Additional competition assays were also performed on 50-mm plates with two mating spots. Worms to be tested for the competition were placed on one of the two bacterial spots. The competitions were between (1) one C. remanei female vs. 10 C. elegans unc-24 hermaphrodites; (2) a dead C. remanei female vs. an empty spot; (3) one C. elegans unc-24 hermaphrodite vs. an empty spot; and (4) 10 C. elegans unc-24 hermaphrodites vs. 10 C. elegans unc-24; him-5 males. Ten to 12 C. remanei or C. elegans him-5 males were placed midway between the two spots and after 6 hr their final positions were recorded. In these experiments, uncoordinated hermaphrodites were employed to inhibit migration. Some of the experiments were repeated using unc-17 or wild-type hermaphrodites to confirm that the results were independent of the specific Unc mutants employed in the assays.

Competition experiment between him-5 and wild type:
The population competition experiment on C. elegans was performed in triplicate with the starting culture composed of 100 him-5 mutant hermaphrodites on a 100-mm culture plate. Two N2 wild-type hermaphrodites were added on day 0. After intervals of 2–3 days, when the food was finished or the cultures had a large number of L1 larvae, the worms were washed off the plate into a 1.5-ml Eppendorf tube. The mixed culture was allowed to stand for 15–20 sec so that a majority of the old or dead adults settled to the bottom. The young smaller larvae and actively wiggling adults still in suspension were transferred to a new tube and washed twice with M9 buffer. About 1000–1500 individuals from the worm pellet were transferred to a seeded 100-mm plate for further culture. This procedure was repeated multiple times and the presence of males in the cultures was scored every 2–3 days over a period of 2¹/₂ months until the incidence of males had fallen to a level consistent with the entire population being wild type.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A mathematical model for the mating system of C. elegans:
We first develop a general mathematical model for the C. elegans mating system that will be useful in suggesting and interpreting subsequent experiments. Consider a worm population consisting of H hermaphrodites and M males; the population frequencies of hermaphrodites P and males S are given by

(1)

Furthermore, we denote the average number of fertilized eggs per hermaphrodite by h and the average number of fertilizing sperm per male by m. An important model parameter will be the mating efficiency

(2)

where b is unity for a dioecious species with hermaphrodites replaced by females and males and females in equal numbers. As a consequence of selfing, the total number of fertilizing male sperm mM is less than or equal to the total number of fertilized eggs hH, so that (1) and (2) imply bS/P 1.

A deterministic model is constructed for the evolution of the frequency of males in the population under the usual simplifying assumptions of very large population size and discrete generations. We assume that one-half the offspring from male-fertilized eggs and a fraction u from self-fertilized eggs are male. To further simplify the model, we make the reasonable assumption that selfed offspring of genotype other than XO and XX (e.g., XXX), which occur at low frequency due to nondisjunction of the sex chromosomes, are nonviable. The fractions of hermaphrodite and male zygotes from selfed and male-fertilized eggs are presented in Table 1. We further assume that offspring from self-fertilized eggs have fitness 1 - d relative to those from male-fertilized eggs. The parameter d models inbreeding depression (CHARLESWORTH and CHARLESWORTH 1978 ), and in general 0 d 1.

The evolution equation for the male frequency, with a prime denoting the frequency in the next generation, is determined to be

(3)

with . Our model (3) differs significantly from that of HEDGECOCK 1976 in that no a priori equilibrium assumption has been made in its construction. In (3), a population equilibrium is obtained by iteration once u, d, b, and the initial male frequency are specified.

The mating efficiency b is in general expected to depend on the male frequency S when males compete for mates. Male competition acts to reduce b by decreasing the average number m of fertilizing sperm per male.

The mating efficiency b, valid only for equilibrium populations, may be determined from (3) with :

(4)

We note that the functional dependence for nonequilibrium populations may be quite different from (4).

We can further determine the conditions under which males will be maintained in the worm population in the absence of sex chromosomal nondisjunction . If males are present in the population at low frequencies, then b can be taken to be independent of S. An equilibrium solution of (3) with u = 0 is S = 0. When this solution is stable, males are eliminated from the population; otherwise males are maintained. The stability of the solution S = 0 is readily determined, and we find that males will be eliminated from the worm population when

(5)

The above stability condition is similar to one found previously (OTTO et al. 1993 ) in a study of the mating system of the clam shrimp Eulimnadia texana, in which both selfing and outcrossing occurs, and our interpretation of (5) follows theirs. The factor of ¹/₂ can be attributed to the cost of males. Males can be maintained if inbreeding depression is absent and males are more than twice as productive as hermaphrodites (b > 2) or if males and hermaphrodites are equally productive and inbreeding depression is large (d > ¹/₂).

When sex chromosomal nondisjunction occurs (u > 0) and the inequality (5) is satisfied, males are maintained in the population at a frequency proportional to u. For poor mating efficiency (b << 1), the male equilibrium frequency S is given to leading order in b as

(6)

When u, d << 1, S is insensitive to b over a wide range of values. Larger values of u, however, increase the sensitivity of S to b, and this observation will be helpful in an experimental determination of the mating efficiency.

Inbreeding depression is not found in laboratory C. elegans strains:
Inbreeding depression can be caused either by loss of heterozygosity at individual loci or by deleterious mutations being made homozygous. Inbreeding depression has been clearly demonstrated in Drosophila (HOLLINGSWORTH and MAYNARD SMITH 1955 ). Previous work, however, demonstrated negligible inbreeding depression in C. elegans (JOHNSON and HUTCHINSON 1993 ). We made additional tests by crossing two isolates of C. elegans: the AB1 strain from Australia and N2 from Bristol, England. Using the average self-fertilized brood size as a parameter for comparison, we scored the number of progeny of the parental F₀, F₁ heterozygous, and F₂ variants. Our parental strain, N2, had an average brood size of 195 ± 26 (134–266, N = 40), and AB1 had an average of 182 ± 27 (148–251, N = 50). The average brood size of heterozygous AB1/N2 F₁ worms was 175 ± 24 (138–227, N = 40); the F₂ worms had an average brood size of 176 ± 45 (71–269, N = 50). From these results, the F₁ worms inheriting one chromosome from each parental strain did not have an increased fecundity, and neither did the genetically heterogeneous F₂ worms. Considering only brood sizes of selfed hermaphrodites, outcrossing seemed to be of little benefit to hermaphrodites, and we subsequently assume d = 0 in the mathematical model.

Measurement of u and S and determination of b for a mutant him-5 strain:
To obtain an estimate of the mating efficiency b used in the mathematical model, we took advantage of the mutant him-5 strain, which had been reported to produce 30% males among the progeny of selfed hermaphrodites due to a high nondisjunction rate (HODGKIN et al. 1979 ). We measured u by counting 5695 males out of 17,488 offspring from virgin hermaphrodites, resulting in u = 0.3257 ± 0.0035. The errors here and subsequently are estimated using the binomial distribution, i.e., , where N is the total number of worms counted. We then counted 9929 males out of 30,008 total worms present in equilibrium cultures, yielding a value of , slightly higher than the value of u as expected if there is only a small amount of outcrossing. Using these values of u and S in (4), taking d = 0, and propagating the errors in the usual way results in . Since the mating efficiency b is much less than unity, C. elegans males are not nearly as productive as the males of a dioecious species. The inefficiency of C. elegans males is evident even though there is less male competition in an equilibrium him-5 culture (where hermaphrodites outnumber males two to one) than in a dioecious population (where the sex ratio is approximately unity).

Measurement of S and determination of u for a wild-type N2 strain:
Male worms were counted in a number of parallel cultures of C. elegans wild-type N2 worms. Twenty-two males were recorded after counting 28,473 worms, yielding an equilibrium male frequency of , or 0.077 ± 0.016%. A somewhat larger value of approximately 0.2% was reported previously for the equilibrium male frequency of N2 (HODGKIN 1983 ). Assuming the same value of the mating efficiency b for N2 as him-5, we can estimate u from the N2 population by solving (4) for u (with d = 0),

(7)

the last approximation being valid for S << 1. We find , which is only slightly lower than the value of S because of the rarity of outcrossing.

In Fig 1, using (4) we plot the values of b, which could occur in equilibrium populations, vs. S, for , and . Also shown by the dotted line is the measured value . From the figure it is evident that b is best measured using the him-5 strain.

View larger version (14K): In this window In a new window Download PPT slide   Figure 1. The mating efficiency b vs. the male frequency S for populations in equilibrium. Solid curves represent the equilibrium solutions for N2 wild type with and him-5 mutant with u = 0.3257. The dashed line is the value for b obtained by measuring u and S from a him-5 mutant population.

Direct observation of mating:
To confirm that outcrossing was infrequent, matings were observed in an equilibrium him-5 mutant culture. Occasional matings were observed with males displaying an active vulva searching behavior when associated with hermaphrodites. When the number of males and hermaphrodites was sampled, we found 9.1% and 4.0% of males and hermaphrodites, respectively, engaged in mating. When the duration of the mating events was estimated, males and hermaphrodites stayed together an average 2.14 min (measured timings from 10 sec to 4 min, N = 40). Out of these 40 matings, copulation was observed in only 2, both of which had their measured times exceeding 2 min (very few matings lasted for >2 min). Thus, although multiple matings were observed, copulation was less frequent. To further investigate the lack of mating success, we examined the impact of density as well as hermaphrodite inactivity on matings. When 1 single male was used to mate with 20 active hermaphrodites on a 9-cm² lawn, successful matings occurred in only 6 out of 80 tests, with an average cross brood of 11 ± 10, estimated by doubling the number of male offspring. The overall average cross brood among the 80 tests was 0.83. When the mating lawn area was reduced to a 0.5-cm diameter circle, however, the successful matings increased to 17 out of 80 tests, with the average cross brood 10 ± 9 per successful cross. The overall average cross brood was more than doubled to 2.2. The mating success of males was further improved when 20 uncoordinated hermaphrodites were used to mate with a single male. In a 9-cm² lawn mating test, there were 23 successful matings in 30 tests, with an average cross brood per mating of 55 ± 23 and an overall average of 42. In a parallel test with a mating lawn of 0.5 cm diameter, 100% successful matings were achieved (N = 30), with an average cross brood of 120 ± 45. It should be mentioned that active mating behavior exhibited by males could be affected by their age. While the above tests were done with young males <1 day old, additional tests done with males older than 3 days showed extremely inefficient mating.

All of these results imply that C. elegans males cannot mate efficiently in a normal cultured equilibrium population. A similar observation was recently reported by STEWART and PHILLIPS 2002 . The inability of uncoordinated hermaphrodites to migrate actively within the mating area enhances mating success 40- to 50-fold, and an increased density of hermaphrodites enhances mating success by 2-fold. Because the hermaphrodites exhibit no active seeking behavior, our mating efficiency results suggest that hermaphrodites are passive mating partners. The success of mating is determined primarily by the seeking behavior of the males, which is affected by the density of hermaphrodites on a plate as well as the ability of hermaphrodites to move.

C. remanei mates more efficiently than C. elegans:
In contrast to a hermaphroditic species, a dioecious species depends on successful male-female matings. A lack of mating vigor comparable to C. elegans would likely drive a dioecious species extinct. Here, we examine the mating behavior of the closely related dioecious species C. remanei. In equilibrium C. remanei populations, we determined 33.4% (N = 2136) males and 30.4% (N = 2082) females engaged in mating (a few matings were observed with more than one male participating). Thus the encounters required for the initiation of successful mating appeared to be more frequent in the dioecious species. For successful mating, the duration of coupling between the two sexes is important. In this regard, C. remanei outperformed C. elegans by >15-fold. The dioecious species had an average mating time of 41.6 ± 33.2 min, and copulation was observed in 100% (40/40) of the matings. In fact, in synchronized C. remanei populations, almost all females were found to be mated (by scoring the presence of a mating plug) within 1 hr after reaching maturity.

C. elegans males mate effectively with C. remanei females:
Our observations make it evident that the dioecious species C. remanei is more efficient at mating than C. elegans. Since both species are believed to share a common dioecious ancestor (FITCH and THOMAS 1997 ), this suggests that a genetic mutation might have occurred during the evolution of C. elegans that reduced its mating efficiency. To explore this possibility, we conducted cross-species mating tests between males and females or hermaphrodites. Five C. elegans males were placed with five C. remanei females on mating plates, and their mating frequencies and duration of matings were scored. By observing the females at an instant in time, matings were found in 24.6% of the worms (N = 275). In opposite mating pairs with five C. remanei males and five C. elegans hermaphrodites, only 3.0% of the hermaphrodites were observed to be mating (N = 201). The duration of matings was estimated for observed couplings. The C. elegans male-C. remanei female pairs mated for an average of 34.0 ± 19.6 min (N = 25) and the C. remanei male-C. elegans hermaphrodite pairs mated an average of 1.4 ± 0.6 min (N = 5). The mating duration of the C. elegans male-C. remanei female pair is similar to that recorded for intraspecies C. remanei couplings, and the mating duration of the C. remanei male-C. elegans hermaphrodite pairs is similar to that for intraspecies C. elegans couplings. The differences in mating durations are thus determined by the C. remanei female or the C. elegans hermaphrodites and not by the males. The surprising result here is that C. elegans males mate more effectively with C. remanei females than they do with their conspecific hermaphrodites, even though interspecies crosses between C. elegans and C. remanei result in no viable progeny (BAIRD et al. 1994 ).

C. elegans hermaphrodites no longer attract males:
To account for the difference in mating durations, we reason that male attraction to females might be due to a chemotactic process relying on sex pheromones produced by females and not by hermaphrodites. To explore this possibility, we first performed an assay for attracting males with a three-way competition between an empty spot with only a bacterial lawn, a second spot with a C. elegans hermaphrodite, and a third spot with a C. remanei female. The number of males attracted to each spot is shown in Table 2A. Both C. elegans and C. remanei males showed a preference to the spot where the C. remanei female was placed over the spot where the C. elegans hermaphrodite was placed, which in turn was slightly more attractive than the empty spot. In similar assays, two-way competitions were performed to compare directly the preference between sources of attractant. As shown in Table 2B, a single C. remanei female was much more attractive to both C. elegans and C. remanei males than were 10 C. elegans hermaphrodites, implying that a female produces at least an order of magnitude more attracting substance than a hermaphrodite. The null hypothesis that this result was due to a random choice between females and hermaphrodites (exact binomial test) is rejected by the very small P value. Also, no statistically significant difference (goodness-of-fit test: P = 0.76) was observed between C. elegans and C. remanei males in this assay. Furthermore, as shown in Table 2C, when the C. remanei female was killed by a hot worm pick the carcass did not elicit the same attraction, suggesting that the attractant was emitted only by live females.

Additional experiments were done to determine if C. elegans hermaphrodites secreted any attractive substance. Two two-way competition experiments were performed: one C. elegans hermaphrodite vs. an empty spot, presented in Table 2D, and 10 C. elegans hermaphrodites vs. 10 C. elegans males, presented in Table 2E. Our results indicate a small preference for live worms over an empty spot (the P values indicate only marginal significance), but no preference for hermaphrodites over males. In fact, on plates cultured with him-5 mutants it is common to observe males mating with other males. All of these competition results considered together, as well as our direct observations of matings, strongly suggest that C. elegans hermaphrodites lack an attractant (perhaps a sex pheromone) expressed by C. remanei females.

Why are C. elegans males still attracted to C. remanei females if their conspecific hermaphrodites are no longer attractive? Maintaining this attraction would appear to be maladaptive to C. elegans males since copulation with C. remanei females is unproductive. Here, we can suggest only three possible reasons. First, a very small amount of attractant undetectable by our assays is produced by hermaphrodites; second, C. elegans male attraction to C. remanei females is maintained as a slightly deleterious side effect to some other more advantageous but unknown function; or third, continuing male attraction to C. remanei females is an evolutionary relic that has not been significantly selected against. Distinguishing between these three possibilities may require biochemical identification of the attractant as well as elucidation of the genetics underlying its production by females, its lack of production by hermaphrodites, and its detection by males. Of particular interest is whether a gene required for pheromone production in C. remanei females is mutated in C. elegans hermaphrodites.

A competition experiment between him-5 and wild type:
With negligible inbreeding depression and b small, the mathematical model predicts that natural selection should act to lower the nondisjunction rate and the corresponding u to as low a value as possible. As a test of this hypothesis, we conducted a competition experiment using a wild-type N2 strain to compete against a mutant him-5 strain. The brood sizes of selfing N2 and him-5 strains were measured prior to the experiment and found to be approximately equal: For N2, the brood size was 195 ± 26 (134–266, N = 40); for him-5, the brood size was 197 ± 28 (118–238, N = 27). (This result in itself is puzzling because the substantially larger nondisjunction rate for him-5 should result in autosomal nondisjunctions as well as zygotes with sex chromosome genotypes O, XXX, and XXXX. In fact, XXX genotypes should occur with the same frequency as XO males, but in our him mutants very few XXX dumpy worms are observed.) Three competition experiments were initiated with 100 him-5 and 2 wild-type hermaphrodites, and subsequent measurements of average male frequency vs. the elapsed number of days are shown as the data points in Fig 2. The error bars represent the standard deviation determined from the three separately evolving cultures. It is observed that the him-5 mutation was essentially eliminated from the populations in 70 days (23 generations).

View larger version (12K): In this window In a new window Download PPT slide   Figure 2. Results of a competition experiment between N2 wild type and him-5 mutant. Male frequency S (points with error bars) is plotted vs. the elapsed number of days. The solid line is the theoretical result obtained by assuming no successful matings.

The experimental results may be compared to a relatively simple mathematical model. We consider the dynamics of a population consisting of a mixture of him mutants and wild-type worms, assuming that both phenotypes have the same brood size and differ only in the frequency of selfed spontaneous males. The genotypes, associated phenotypes expressed as the spontaneous male frequency u, and corresponding genotype frequencies are shown in Table 3.

In general, a mating table may be constructed and evolution equations for the male and hermaphrodite frequencies determined. The resulting model, with parameters b, d, u_AA, u_Aa, and u_aa, consists of five independent evolution equations and requires numerical solution. Since the observed value b = 0.06 is small, however, we make the simplifying assumption that no mating occurs (b = 0), which permits an analytical solution of the model equations.

The assumption of no mating as well as the initial absence of heterozygous genotypes leads to the evolution equation

(8)

which permits analytical solution. The corresponding solution for the male frequency S at generation n > 0 is determined to be

(9)

where c is the initial value of P_AA/P_aa when n = 0. In a competition between AA wild type and aa him-5 mutants, we can approximate u_AA = 0 and u_aa = U, so that (9) simplifies further to

(10)

The result (10), with U = 0.3257 and n = D/3, where D is the number of elapsed days in the experiment, 3 days is the estimate for the generation length of C. elegans, and c = 0.02 (2 wild-type and 100 him-5 hermaphrodites at n = 0), can be compared directly to the laboratory measurements and is the solid curve in Fig 2. Good agreement is observed between the experimental data and the theoretical curve. Our results exemplify the simple idea that without a substantial number of matings between males and hermaphrodites, a hermaphrodite that produces a male instead of a hermaphrodite reduces its effective fecundity.

Why has the male phenotype not degenerated?
The loss of sex pheromone expression in hermaphrodites may be due to natural selection favoring hermaphrodites that self-fertilize. Natural selection, however, also favors males that mate efficiently since this is the only route for male genes to pass into the next generation. Interestingly, the two sexes of C. elegans have conflicting interests: Hermaphrodites want to self and males want to mate. This is in contrast to a dioecious species where males and females must cooperate to reproduce.

Here, we further consider the evolution of genes that are expressed only in males, e.g., the genes needed to build the C. elegans male sensory rays. All genes face a constant mutation pressure and those genes required to build the sensory rays, say, can be maintained only if they enhance the reproductive success of males. Complex adaptations that no longer contribute to an organism's reproductive success degenerate over evolutionary time: An often-cited example is eye loss in cave-dwelling animals (FONG et al. 1995 ). If, for instance, C. elegans males never successfully mate with hermaphrodites, then an adaptive organ such as the sensory rays would be expected to degenerate, and C. elegans might evolve into a hermaphroditic species without males.

C. elegans males, however, do occasionally fertilize hermaphrodites and obviously the genes specifically expressed in males have not degenerated. The question we address here is: How often must males successfully mate with hermaphrodites to prevent male genes from degenerating? An answer requires determining the relationship between a critical deleterious mutation rate v_c for a male-expressed gene and the mathematical model parameters u, b, and d, such that for a deleterious mutation rate v > v_c the male gene becomes nonfunctional, resulting in male sterility. For simplicity, we consider only a deterministic model with infinite population size. Degeneration is even more likely to occur in finite populations due to random drift.

Accordingly, we consider a wild-type gene A expressed only in males such that male genotypes AA average m fertilizing sperm per male, and male genotypes Aa and aa average none; i.e., these genotypes are sterile because of degeneration. Here, we assume a dominant mutation to simplify the algebra, but additional numerical computations (not shown here) demonstrate the same final result (13) for a recessive mutation. We further assume that hermaphrodites of all genotypes average h fertilized eggs per hermaphrodite and that the chance of an AA male fertilizing an oocyte is independent of the genotype of the hermaphrodite. To develop our model equations, consider a worm population consisting of H_AA, H_Aa, and H_aa hermaphrodites of genotype denoted by the subscript and M_AA, M_Aa, and M_aa males. With H = H_AA + H_Aa + H_aa and M = M_AA + M_Aa + M_aa, the population frequencies of hermaphrodites and males of the various genotypes are given by

(11)

The number of zygotes from different matings can be calculated. As an example, the number of self-fertilized zygotes from AA hermaphrodites is given by

where the first and second terms are the total number of zygotes from AA hermaphrodites and the number that are male-fertilized, respectively. Division by the total number of fertilized zygotes hH determines the various zygote frequencies. These frequencies and the frequencies of the various progeny are given in Table 4, where as before we define . Finally, we assume that wild-type gene A has probability v of mutating to gene a each generation. For further simplicity, we assume that v << 1 so that terms of order v² are negligible.

General evolution equations for the various genotype frequencies may now be constructed. Here, however, we are interested in the stability of a population in which all genes A have mutated; that is, , with all other genotype frequencies equal to zero. We thus perturb this equilibrium state and linearize the evolution equations about the infinitesimal frequencies P_AA, S_AA, P_Aa, and S_Aa, and the deviations of P_aa and S_aa from equilibrium. Only the evolution equations for P_AA, S_AA, and P_Aa are coupled, and these equations written in matrix form are given by

(12)

The eigenvalues of the coupling matrix may be computed; the degenerated equilibrium state is unstable when an eigenvalue becomes larger than unity. Setting the maximum eigenvalue to unity and solving for v determines the critical mutation rate v_c. To leading order in u, the analytical result is

(13)

When v > v_c, the equilibrium state where all genes A have mutated to a is stable; all males are sterile in this population. When v < v_c, at least some males in the population are fertile. With , and , we obtain the numerical value . An estimate for the deleterious mutation rate per haploid genome in C. elegans is (KEIGHTLEY and CABALLERO 1997 ), and with an estimate of 19,000 genes (C. ELEGANS SEQUENCING CONSORTIUM 1998), the mutation rate per gene is 0.0026/19,000 1.4 x 10^-7, almost two orders of magnitude smaller than our estimate for v_c.

A further simplifying approximation considers all the male-only expressed genes to be tightly linked. For this approximation, male fertility will be maintained only if males express < 1.1 x 10^-5/1.4 x 10^-7 80 genes that have no phenotypic expression in hermaphrodites. For the more realistic situation of less tightly linked genes, substantially more male-only expressed genes could be maintained. Recent work by JIANG et al. 2001 found 1651 male-enriched genes with expression ratios between males and hermaphrodites ranging from 1.5 to 110. Our model applies to male genes that have no phenotypic expression in hermaphrodites. The unc-68 gene known to affect both males and hermaphrodites has an expression ratio of 5.3 and the pkd-2 gene known to be expressed only in males has an expression ratio of 7. Accordingly, we assume that genes with an expression ratio >6 are expressed only in males. If we count only those genes thought to be expressed in the male soma (478) rather than in spermatogenesis (also required by hermaphrodites), then an estimate of 50–60 male-only genes is obtained. It would thus appear that the amount of gene flow from males to future generations is sufficient to prevent the mutational degeneration of the male phenotype even for the limiting case of tight linkage.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We begin with some words of caution. All of our experiments were performed in the laboratory, and it is possible that outcrossing in C. elegans occurs more frequently in the natural environment. Mating, however, is likely to be easier for males in the two-dimensional environment of the culture plate with relatively higher worm densities than in the three-dimensional natural environment, where worm density seldom reaches a level >100 worms/cm³ (D. H. A. FITCH, personal communication). It is also possible that we have missed some advantage of outcrossing that appears intermittently and perhaps only in the natural environment. This advantage, however, would have to be large to have an impact, and the low incidence of C. elegans males typically found in natural compost piles (D. H. A. FITCH, personal communication) argues against it.

We now review the laboratory facts before addressing the question posed by this article's title:

1. No obvious fitness advantage of outcrossed offspring is observed. Brood sizes from self-fertilizing hermaphrodites are independent of whether these hermaphrodites developed from self-fertilized or outcrossed zygotes. Self-fertilized offspring from C. elegans appear to suffer little inbreeding depression, certainly not the factor of two required to repay the cost of males.

2. The mating efficiency of C. elegans is poor. In our laboratory culture of him-5 mutants, each male had on average six productive sperm for every 100 fertilized hermaphrodite eggs, even though hermaphrodites outnumber males by two to one in the population. By direct observation, successful copulation by males is very infrequent.

3. Direct comparison of mating behavior shows that the frequency is much lower and the duration of mating for C. elegans is much shorter than that for C. remanei.

4. Competition assays between C. elegans hermaphrodites and C. remanei females to attract males of both species demonstrate that C. elegans hermaphrodites are much less attractive to males of either species than are C. remanei females. In fact, C. elegans males are significantly more attracted to C. remanei females than they are to hermaphrodites of their own species.

Some additional experimental facts are also relevant:

5. Hermaphrodites produce fewer sperm than oocytes. Self sperm are utilized with almost 100% efficiency and the additional oocytes produced are laid unfertilized unless the hermaphrodite is mated. As a consequence, under laboratory conditions hermaphrodites that both self and outcross produce substantially more progeny than hermaphrodites that only self (WARD and CARREL 1979 ).

6. Natural selection acts more strongly on earlier than later produced progeny. A mutation that results in a 50% increase in hermaphrodite sperm production is outcompeted by wild-type worms. Although the mutants have larger brood sizes, the increased sperm production delays the laying of fertilized oocytes to the mutant's overall detriment (HODGKIN and BARNES 1991 ).

It may be impossible to reconstruct the evolutionary history of C. elegans. That said, it is still illuminating to speculate on a plausible evolutionary path from the ancestral dioecious species to C. elegans that is consistent with the observations and experiments. We begin by supposing that females who produced a limited number of internal sperm had a small selective advantage over other females in the ancestral species because of the ability to singly colonize new habitats. These protohermaphrodites still required male sperm to fertilize the bulk of their oocytes, so natural selection would support hermaphrodite attractiveness. These protohermaphrodites may have evolved, however, to self-fertilize more of their own oocytes as the deleterious effects of inbreeding depression diminished, and selection for attractiveness could have turned from positive to negative. This reversal in selection is due to an important trade-off. Attractive hermaphrodites more efficiently obtain sperm from conspecific males and can produce larger brood sizes than nonattractive hermaphrodites. Attractive hermaphrodites, however, may be mated quickly with the disadvantage that early progeny are 50% males, whereas early progeny of unattractive hermaphrodites are likely to be selfed and are thus 100% hermaphrodites. As soon as hermaphrodite attractiveness became disadvantageous, any mutation that eliminated sex pheromone expression would have rapidly swept through the hermaphrodite population. Although the optimum reproductive strategy for a C. elegans hermaphrodite is to first self-fertilize its oocytes until all internal sperm are used and then outcross, natural selection acting on attractiveness alone cannot attain this optimum. A single loss-of-function mutation could conceivably result in loss of attractiveness, whereas delaying the onset of attractiveness may require a specific regulatory mutation, if such a mutation exists at all, with much lower probability of occurring. Loss of attractiveness is likely to have become fixed in the population before the regulatory mutation occurred, and it would then be impossible to move from the adaptive peak of no attraction to the theoretical optimum peak of a well-timed attraction.

With the frequency of male-hermaphrodite matings suppressed, males exist mainly because of the nondisjunction of the sex chromosomes in self-fertilizing hermaphrodites. Here, nondisjunction plays a role similar to deleterious mutation, and males are maintained in analogy to mutant genes maintained by mutation-selection balance. In fact, if there were no successful matings between males and hermaphrodites, the male phenotype would be genetically lethal and the frequency of males in the population would be identical to the rate at which males are born to selfing hermaphrodites. With small levels of outcrossing, the male phenotype is maintained in the population at slightly higher frequencies.

Furthermore, we hypothesize that the nondisjunction rate found in C. elegans is as low as possible without incurring excessive costs. A possible test of this hypothesis would be a comparison of X chromosome nondisjunction rates in C. elegans with that in its dioecious relative C. remanei. If natural selection favored hermaphrodites with an increased level of X chromosome nondisjunction, resulting in the birth of a significant number of spontaneous males, then the rate of nondisjunction in C. elegans hermaphrodites should be significantly higher than that found in C. remanei females, for which X chromosome nondisjunction has no obvious adaptive purpose. On the other hand, if the two nondisjunction rates are similar, then the reasonable inference is that the rate of nondisjunction has been minimized by natural selection in both species.

We are unaware, however, of any data on nondisjunction rates in C. remanei or other dioecious nematodes. A relevant study, however, of spontaneous X chromosome nondisjunction in Drosophila melanogaster was undertaken almost 40 years ago (MERRIAM and FROST 1964 ). Using the X-linked recessive mutation y (yellow) and X-linked dominant mutation B (Bar) as markers, XX females (genotype y/y) were crossed with XO males (genotype yB/O) and 26 yellow-non-Bar females out of a total of 45,112 female progeny were scored. The X chromosome nondisjunction rate in D. melanogaster was thus determined to be 0.00058 ± 0.00011, where we have estimated the error from the binomial distribution. Our result is for the frequency u of spontaneous male births by C. elegans hermaphrodites. Male births may be due to either a nullo-X egg fertilized by a normal X sperm or a normal X egg fertilized by a nullo-X sperm. Since we do not know the relative frequency of nondisjunction in the formation of eggs or sperm, we assume here that these frequencies are equal. Therefore, for comparison, the X chromosome nondisjunction rate in C. elegans is . In mammals, however, the nondisjunction rates are substantially higher. In humans, aneuploidy—the most common types being trisomy 21 (causing Down's syndrome) and sex chromosome trisomies—occurs in 0.3% of all newborns, and it is estimated that >>5% of all human conceptions are aneuploid (HASSOLD and HUNT 2001 ). The closeness of the two nondisjunction rates for the worm and the fly, both below that of mammals, lends some support to our claim that the nondisjunction rate in C. elegans is not significantly elevated. We further note that the observed nondisjunction rate in C. elegans is at a level similar to errors occurring in other cellular functions, e.g., an error rate of 0.03–1% for amino acid incorporation during translation (FREIST et al. 1998 ).

Despite the obstacles imposed by indifferent hermaphrodites, natural selection still favors males that successfully mate. In this battle between the sexes, males manage to fertilize hermaphrodites often enough to support a small number of male-only genes against degeneration by deleterious mutations.

We have thus argued that males are not present in the C. elegans species because of any advantage to outcrossing, as is usually supposed. Rather, hermaphrodites no longer attract males and obtain higher fitness by selfing.

	ACKNOWLEDGMENTS

The authors thank Dr. Man-Yu Wong for statistical advice, colleagues in the laboratory for critical comments on this manuscript, and an anonymous referee for his extensive comments and important suggestions. The wild isolates and different genetic strains used in this study were kindly provided by the Caenorhabditis Genetics Center, Minnesota, which is supported by the National Institutes of Health National Center for Research Resources. This study was supported by a grant from the Research Grants Council, Hong Kong.

Manuscript received September 5, 2001; Accepted for publication December 18, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BAIRD, S. E., D. H. FITCH, and S. W. EMMONS, 1994  Caenorhabditis vulgaris sp.n. (Nematoda: Rhabditidae): a necromenic associate of pill bugs and snails. Nematologica 40:1-11.

BRENNER, S., 1974  The genetics of Caenorhabditis elegans.. Genetics 77:71-94[Abstract/Free Full Text].

Genome sequence of the nematode C. elegans: a platform for investigating biology. (1998) Science 282:2012-2018[Abstract/Free Full Text].

CHARLESWORTH, B. and D. CHARLESWORTH, 1978  A model for the evolution of dioecy and gynodioecy. Am. Nat. 112:975-997.

CROW, J. F., 1994  Advantages of sexual reproduction. Dev. Genet. 15:205-213[Medline].

FITCH, D. H. A., and W. K. THOMAS, 1997 Evolution, pp. 815–850 in C. elegans II, edited by D. RIDDLE, T. BLUMENTHAL, B. MEYER and J. PRIESS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

FITCH, D. H. A., B. BUGAJ-GAWEDA, and S. W. EMMONS, 1995  18S ribosomal RNA gene phylogeny for some Rhabditidae related to Caenorhabditis.. Mol. Biol. Evol. 12:346-358[Abstract].

FONG, D. W., T. C. KANE, and D. C. CULVER, 1995  Vestigialization and loss of nonfunctional characters. Annu. Rev. Ecol. Syst. 26:249-268.

FREIST, W., H. STERNBACK, I. PARDOWITZ, and F. CRAMER, 1998  Accuracy of protein biosynthesis: quasi-species nature of protein and possibility of error catastrophes. J. Theor. Biol. 193:19-38[Medline].

HASSOLD, T. and P. HUNT, 2001  To err (meiotically) is human: the genesis of human aneuploidy. Nat. Rev. Genet. 2:280-291[Medline].

HEDGECOCK, E. M., 1976  The mating system of Caenorhabditis elegans: evolutionary equilibrium between self- and cross-fertilization in a facultative hermaphrodite. Am. Nat. 110:1007-1012.

HODGKIN, J., 1983  Male phenotypes and mating efficiency in Caenorhabditis elegans.. Genetics 103:43-64[Abstract/Free Full Text].

HODGKIN, J. and T. M. BARNES, 1991  More is not better: brood size and population growth in a self-fertilizing nematode. Proc. R. Soc. Lond. Ser. B 246:19-24[Medline].

HODGKIN, J., H. R. HORVITZ, and S. BRENNER, 1979  Nondisjunction mutants of the nematode Caenorhabditis elegans.. Genetics 91:67-94[Abstract/Free Full Text].

HOLLINGSWORTH, M. J. and J. MAYNARD SMITH, 1955  The effects of inbreeding on rate of development and on fertility in Drosophila subobscura.. J. Genet. 53:295-314.

HURST, L. D. and J. R. PECK, 1996  Recent advances in understanding of the evolution and maintenance of sex. Trends Ecol. Evol. 11:46-52.

JIANG, M., J. RYU, M. KIRALY, K. DUKE, and V. REINKE et al., 2001  Genome-wide analysis of developmental and sex-regulated gene expression profiles in Caenorhabditis elegans.. Proc. Natl. Acad. Sci. USA 98:218-223[Abstract/Free Full Text].

JOHNSON, T. E. and E. W. HUTCHINSON, 1993  Absence of strong heterosis for life span and other life history traits in Caenorhabditis elegans.. Genetics 134:465-474[Abstract].

KEIGHTLEY, P. D. and A. CABALLERO, 1997  Genomic mutation rates for lifetime reproductive output and lifespan in Caenorhabditis elegans.. Proc. Natl. Acad. Sci. USA 94:3823-3827[Abstract/Free Full Text].

MAYNARD SMITH, J., 1978 The Evolution of Sex. Cambridge University Press, Cambridge, UK.

MERRIAM, J. R. and J. N. FROST, 1964  Exchange and nondisjunction of the X chromosomes in female Drosophila melanogaster. Genetics 49:109-122[Free Full Text].

OTTO, S. P., C. SASSAMAN, and M. W. FELDMAN, 1993  Evolution of sex determination in the Conchostracan shrimp Eulimnadia texana.. Am. Nat. 141:329-337.

STEWART, A. D. and P. C. PHILLIPS, 2002  Selection and maintenance of androdioecy in Caenorhabditis elegans.. Genetics 160:975-982[Abstract/Free Full Text].

WARD, S. and J. S. CARREL, 1979  Fertilization and sperm competition in the nematode Caenorhabditis elegans.. Dev. Biol. 73:304-321[Medline].

This article has been cited by other articles:

				C. G. Artieri, W. Haerty, B. P. Gupta, and R. S. Singh Sexual Selection and Maintenance of Sex: Evidence from Comparisons of Rates of Genomic Accumulation of Mutations and Divergence of Sex-Related Genes in Sexual and Hermaphroditic Species of Caenorhabditis Mol. Biol. Evol., May 1, 2008; 25(5): 972 - 979. [Abstract] [Full Text] [PDF]

				A. D. Cutter, J. D. Wasmuth, and N. L. Washington Patterns of Molecular Evolution in Caenorhabditis Preclude Ancient Origins of Selfing Genetics, April 1, 2008; 178(4): 2093 - 2104. [Abstract] [Full Text] [PDF]

				J. R. Chasnov, W. K. So, C. M. Chan, and K. L. Chow The species, sex, and stage specificity of a Caenorhabditis sex pheromone PNAS, April 17, 2007; 104(16): 6730 - 6735. [Abstract] [Full Text] [PDF]

				L. R. Garcia, B. LeBoeuf, and P. Koo Diversity in Mating Behavior of Hermaphroditic and Male-Female Caenorhabditis Nematodes Genetics, April 1, 2007; 175(4): 1761 - 1771. [Abstract] [Full Text] [PDF]

				A. D. Cutter, S. E. Baird, and D. Charlesworth High Nucleotide Polymorphism and Rapid Decay of Linkage Disequilibrium in Wild Populations of Caenorhabditis remanei Genetics, October 1, 2006; 174(2): 901 - 913. [Abstract] [Full Text] [PDF]

				S. C. Weeks, C. Benvenuto, and S. K. Reed When males and hermaphrodites coexist: a review of androdioecy in animals Integr. Comp. Biol., August 1, 2006; 46(4): 449 - 464. [Abstract] [Full Text] [PDF]

				A. D. Cutter, M.-A. Felix, A. Barriere, and D. Charlesworth Patterns of Nucleotide Polymorphism Distinguish Temperate and Tropical Wild Isolates of Caenorhabditis briggsae Genetics, August 1, 2006; 173(4): 2021 - 2031. [Abstract] [Full Text] [PDF]

				A. D. Cutter Nucleotide Polymorphism and Linkage Disequilibrium in Wild Populations of the Partial Selfer Caenorhabditis elegans Genetics, January 1, 2006; 172(1): 171 - 184. [Abstract] [Full Text] [PDF]

				M. Haber, M. Schungel, A. Putz, S. Muller, B. Hasert, and H. Schulenburg Evolutionary History of Caenorhabditis elegans Inferred from Microsatellites: Evidence for Spatial and Temporal Genetic Differentiation and the Occurrence of Outbreeding Mol. Biol. Evol., January 1, 2005; 22(1): 160 - 173. [Abstract] [Full Text] [PDF]