Prevalent sex ratio bias in Caenorhabditis nematodes 1

1Biodiversity Research Center, Academia Sinica, Taipei, Taiwan 4 2Department of Molecular Genetics and Gene Technology, College of Forestry Biotechnology, Vietnam 5 National University of Forestry, Hanoi, Vietnam 6 3Université Côte d’Azur, CNRS, Inserm, IBV, Nice, France 7 8 #These authors contributed equally 9 *Correspondence to: 10 Biodiversity Research Center, Academia Sinica, Taipei 115, Taiwan 11 yunhuang@gate.sinica.edu.tw 12 johnwang@gate.sinica.edu.tw 13 TEL: +886 2 27871582 14 FAX: +886 2 27871583 15 16


41
Sex ratio theory is at the core of evolutionary biology. First proposed by Charles Darwin (1,42 2), and later formulated by Ronald Fisher (3), sex ratio theory posits that in a large, random mating 43 population, when the sex ratio departs from equality, the rarer sex would have better mating 44 prospects. Thus, genetic elements that favor the rarer sex would be favored by natural selection. 45 Consequently, equal sex ratio should be restored and represents an evolutionary stable strategy. 46 However, as many organisms have unique life histories, social structures and parental behaviors 47 that do not meet Fisher's assumptions, biased sex ratios exist and are intriguing examples to study 48 the evolution of sex ratio (4). 49 Theories have been proposed to explain sex ratio bias based on the asymmetry of sex 50 allocation and reproductive return between the two sexes in different scenarios (5, 6). Hamilton's 51 theory of local mate competition (LMC) predicts that in cases of highly structured populations 52 and local mating, sex ratios would be biased toward females so that just enough males are 53 produced to inseminate their female siblings (4). Similarly, theories predict that sex ratio bias may 54 arise due to local resource competition, where the sex that only consumes local resources is 55 reduced (7). In the case of cooperative breeding, sex ratio would be biased toward the helping 56 sex (local resource enhancement, LRE) (8,9). Moreover, Trivers and Willard proposed that females 57 in good condition or of high social ranking would produce more male offspring as their sons would 58 inherit their advantages and have above-average breeding success (10). These theories have 59 successfully explained many empirical findings of sex ratio bias. 60 Examples of sex ratio bias across the unicellular and metazoan world are plentiful. For 61 instance, female-biased sex ratios are commonly found in Apicomplexan parasites such as 62 Figure 1. Progeny sex ratios and the ancestral states of 23 Caenorhabditis species. (a) The 141 proportion of females (hermaphrodites) in total progeny in the unlimited mating assay. For 142 the species with multiple strains, only one tester strain is included in this figure. Each circle 143 represents a replicate mating pair of the tester strain and the circle size denotes the total 144 number of progeny. Bars represent median sex ratios. The asterisks indicate significant 145 female (hermaphrodite) bias (combined and adjusted P < 0.05, binomial test). (b) The 146 ancestral states of sex ratio bias constructed based on the phylogeny of the 23 species. The 147 black branches represent sex ratio bias in the child nodes whereas the white branches 148 represent equal sex ratio. For the species with multiple strains, the sex ratios from different strains showed 157 consistent bias or non-bias within the species. For example, for the four female-male species 158 (C. nigoni, C. latens, C. remanei, and C. sp. 33), all strains from the respective species showed 159 consistently biased sex ratios (P < 0.05 for the strains), except the C. remanei strain JU1084, 160 which had a marginally non-biased sex ratio (P = 0.058). However, C. nigoni, C. remanei, and 161 C. sp. 33 had significantly different sex ratios between strains (P < 0.05, ANOVA test), whereas 162 C. latens did not have significantly different sex ratios between strains (P = 0.824, ANOVA test). 163 To test the possibility that sex ratio bias was due to sex-specific consequences of 164 inbreeding depression (34), for the four female-male species, for which we had multiple 165 strains, (C. nigoni, C. latens, C. remanei, and C. sp. 33), we performed inter-stain crosses. The 166 inter-strain crosses yielded consistent female-biased sex ratios, congruent with the intra-167 strain experiments above (combined P < 0.05, Table S1). 168 For the three androdioecious species, the multiple tester strains also showed consistent 169 bias (for C. briggsae) or non-bias (for C. elegans and C. tropicalis). All the strains from C. 170 elegans and C. tropicalis, respectively, did not have hermaphrodite-biased sex ratios (P > 0.05 171 for the strains), and the sex ratios were not significantly different between strains for both of 172 these two species (P = 0.658 for C. elegans and P = 0.441 for C. tropicalis, ANOVA test). For 31 173 C. briggsae strains we surveyed in addition to the common lab strain AF16, 19 strains showed 174 significant hermaphrodite-biased sex ratios (combined and adjusted P <0.05, binomial test, 175 Table S2), consistent with AF16. For the other 12 strains that did not have significant sex ratio 176 bias, most of them still had more hermaphrodites than males for all replicates, except 1 out 177 of 4 replicates in BRC20095, 1 out of 5 in BRC20299, 1 out of 5 in BRC20324, 1 out of 5 in 178 BRC20334, 2 out of 5 in BRC20234 and 4 out of 8 in BRC20339. Thus, the majority of the C. 179 briggsae strains had significantly hermaphrodite-biased sex ratios, whereas for the other 180 strains, although there was no significant bias, there was a general tendency toward bias, 181 except for BRC20234 and BRC20339. 182 183 184

Sex ratios in the limited mating experiment 185
To test whether sperm competition between X and nullo-X plays a role in progeny sex 186 ratio bias, we conducted limited mating experiments to assay sex ratios. When exposed to 187 males for only up to 5 hours, the females (or hermaphrodites) sired on average 51% (range 188 20-80%) fewer progeny (outcrossed progeny for androdioecious species) compared to 189 unlimited mating, except for C. sinica and C. latens, which had comparable numbers of 190 progeny between the two mating experiments (Table 1 and 2). Ten of the 23 species examined 191 showed female(hermaphrodite)-biased sex ratios in the limited mating experiment (combined 192 and adjusted P < 0.05, binomial test; Figure 2a, Table 2). Out of the 15 species that showed 193 female-biased sex ratio in the unlimited mating experiment, nine were also found with 194 female(hermaphrodite)-biased sex ratios in the limited mating experiment. The other six 195 species (C. nigoni, C. zanzibari, C. sinica, C. remanei, C. becei, and C. portoensis) did not show 196 significant sex ratio bias in the limited mating experiment. In contrast, C. sp. 41, which did not 197 show sex-ratio bias in the unlimited mating experiment, showed significant female-biased sex 198 ratio in the limited mating experiment (adjusted P = 0.0002  line represents a species. The points represent mean sex ratios per day across replicates and 211 the error bars represent standard error of the mean. The day-by-day sex ratios calculated from 212 less than 10 progeny are excluded. Note that for the red, purple and blue panels, only 4 213 exemplary species are displayed. Other species are presented in Figure S1. We found prevalent female sex ratio bias (15 out of 23) among the Caenorhabditis 234 species and none in the other direction. The ancestral state was inferred to be female biased, 235 with seven transitions from female bias to no bias in the phylogeny. While most species (16 236 out of 23) showed consistent sex ratio bias or no bias between the unlimited mating and 237 limited mating experiments, six species had a female(hermaphrodite)-biased sex ratio in the 238 unlimited experiment but no bias in the limited mating experiment, consistent with sperm 239 competition as a possible explanation. Our genus-wide survey of progeny sex ratios sheds light 240 on the evolution of sex ratio bias in this genus. 241 Female-biased sex ratios are frequently found in parasitic nematodes (15,(35)(36)(37). Theory 242 predicts female-biased sex ratios in parasites due to confined dispersion and high inbreeding 243 (4, 38, 39). We found that the species of the free-living nematode Caenorhabditis also show 244 prevalent female-biased sex ratios. Though being free-living and dwelling in diverse habitats, 245 many Caenorhabditis species probably share the features of life history such as boom-and-246 bust population growth in ephemeral habitats, active dispersal seeking, and strong founder 247 effect followed by population re-expansion, such as found in extensive sampling of C. elegans 248 (40-42). These life history features may result in high inbreeding rates and intense local 249 competition for mating between kin. Hence, a female(hermaphrodite)-biased sex ratio may 250 be favored, according to the theory of LMC. Compared to their parasitic relatives, despite the 251 very distinct life styles, sex ratio bias may have evolved in parallel in the free-living 252 Caenorhabditis species. Alternatively, sex ratio bias is widely conserved across diverse 253 nematode taxa. This hypothesis could be tested with a broader survey of sex ratios in free-254 living nematode species outside of Caenorhabditis genus. 255 Based on the prevalent sex ratio bias detected in these 23 species, the ancestral state was 256 inferred to be female-biased, with seven transitions from female bias to equal sex ratio along 257 the phylogeny. This suggests that sex ratio bias could be a phenotype that can be frequently 258 gained or lost, possibly reflecting the adaptation to the respective habitats and life histories 259 in these species. Despite common features of their life styles, Caenorhabditis species dwell in 260 ecologically diverse habitats, ranging from cattle auditory canals to rotting fruits and man-261 made compost (27). Except for a few species that have been sampled extensively, such as C. 262 elegans, C. briggsae, and C. remanei, (40,(43)(44)(45), the natural habitats of most Caenorhabditis 263 species are largely unknown, mainly because they have been sampled very rarely. More 264 knowledge about the ecology and natural history of Caenorhabditis species might thus be able 265 to explain the driving forces of sex ratio evolution in these species. 266 Of the three androdioecious Caenorhabditis species, C. elegans, C. briggsae, and C. 267 tropicalis, only C. briggsae showed a hermaphrodite-biased sex ratio in out-crossed progeny. 268 As the hermaphrodites can self-fertilize, the role of males in these species is obscure (46) but  269 is likely important for rapid adaptation, such as to pathogens (47). Production of male progeny 270 would take up brood "quota" but does not directly contribute to population growth (48). Field 271 studies of C. elegans have rarely found males in the wild, and wild C. elegans largely suffer 272 from outcrossing depression (49-51). Thus, a hermaphrodite-biased sex ratio may be favored 273 in androdioecious species. However, C. elegans and C. tropicalis showed no hermaphrodite-274 biased sex ratio, suggesting unknown ecological factors or historical contingencies that may 275 be contributing to offspring sex ratios in these hermaphroditic species. 276 277 Our strategy of a broad survey across species with limited diversity of strains within 278 species assumes that the sex ratio status of the tester strain is representative for each species. 279 An alternative scenario is that sex ratio is a trans-species polymorphism in the genus. As we 280 used a single strain for the sex ratio assays for most species (16 of 23), we cannot exclude the 281 possibility that, for some of these species, the sex ratio bias we detected was specific to those 282 strains. However, for the four female-male species for which we did have multiple strains, we 283 tested sex ratios for multiple strains as well as for inter-strain crosses. We found that both the 284 intra-strain and inter-strain crosses yielded qualitatively consistent sex ratio bias within 285 species. For the three androdioecious species, the sex ratios of multiple strains also showed 286 consistent non-bias in C. elegans and C. tropicalis. With an extensive survey of C. briggsae 287 strains, we found that most of the strains had significantly hermaphrodite-biased sex ratios, 288 consistent with the common lab strain AF16. Although the other C. briggsae strains did not 289 have significant sex ratio bias, they mostly had more hermaphrodites than males in the 290 replicates, except a few outlier strains. These results together suggest that the prevalent sex 291 ratio bias in Caenorhabditis nematodes is rather a stable trait within species rather than a 292 trans-species polymorphic trait. 293 Despite the qualitative consistency in female bias, C. nigoni, C. remanei, and C. sp. 33 294 showed quantitative differences in sex ratios between strains, as indicated by ANOVA tests, 295 suggesting a contribution of genetic differences in the female-biased sex ratios. On the other 296 hand, C. latens, C. elegans, and C. tropicalis had quantitatively constant sex ratios among 297 strains, suggesting genetic constraints that govern the bias or non-bias. 298

299
To investigate sperm competition as a potential mechanism underlying the sex ratio bias, 300 as previously found in C. briggsae (33), we conducted mating experiments where mating was 301 limited for a few hours as opposed to the unlimited mating experiments where mating was 302 allowed for the entirety of adulthood. A contrast of female biased sex ratios in unlimited 303 mating experiment and an equal sex ratio in limited mating experiment suggests sperm 304 competition. In the unlimited mating experiments, the couples probably mated repeatedly 305 and the X-bearing sperm would be refilled and therefore X-bearing sperm would always take 306 precedence over the nullo-X sperm, resulting in an overall female-biased sex ratio. In contrast, 307 in the limited mating experiments, the amount of sperm transferred was limited and all sperm 308 were presumably used, resulting in an overall equal sex ratio. Consistent with this sperm 309 competition model, six of the 15 species that showed significant female-biased sex ratio in 310 unlimited mating did not have a sex ratio bias in limited mating. The day-by-day sex ratios of 311 these six species also show female bias in unlimited mating throughout the reproductive 312 period and equal sex ratio in limited mating, consistent with the sperm competition scenario. 313 On the other hand, nine species showed female bias in both mating experiments. The day-by-314 day profile of these nine species in both experiments showed female bias at the beginning 315 and then declined to equal sex ratio or male bias in the following days. These observations 316 suggest that sperm competition plays, at least, a partial role in contributing to female-biased 317 sex ratios in both experiments. In C. briggsae, LaMunyon and Ward (1997) conducted 3-hour 318 mating as well as 8-hour mating experiments and found the overall sex ratio was equal after 319 3-hour mating whereas it was hermaphrodite-biased after 8-hour mating (33). Here, we found 320 a hermaphrodite-biased sex ratio after 5 hours of mating. These together suggest that the 321 amount of sperm ejaculated into the hermaphrodite is a limiting factor for sperm competition 322 and hence sex ratio bias. In addition, we found one species, C. sp. 41, which had no sex ratio 323 bias under unlimited mating conditions while limited mating yielded a female bias. The day-324 by-day sex ratios show a slightly female-biased sex ratio on the first day in both experiments, 325 suggesting sperm competition may be present but very weak in this species. Finally, the 326 remaining nine species did not have sex ratio bias in either mating experiments, and the day-327 by-day sex ratios fluctuated around 50:50. Thus, there is no indication of sperm competition 328 in these species. 329 The presence of selfish genetic elements could provide an alternative mechanism for sex 330 ratio bias where the proportion of X-bearing sperm is selfishly enhanced in some species. 331 Examples include segregation distorter on the X chromosome of Drosophila simulans (52) and 332 asymmetric division in spermatocytes in a Rhabditis nematode (53). So far, several selfish 333 elements have been discovered in Caenorhabditis nematodes, but they all reside on 334 autosomes (54-57). 335 While equal sex ratios are presumably predominant in nature, biased sex ratios may be 336 largely underappreciated. In this study, we carried out a broad survey of sex ratio bias in 337 outcrossed progeny across Caenorhabditis nematodes. Our findings of prevalent sex ratio bias 338 add to the limited knowledge about sex ratio bias in the animal kingdom, and provide 339 evidence that sex ratio bias can evolve rapidly within a single genus. We examined 23 Caenorhabditis species for progeny sex ratio in this study (Table 1) For the female-male species, sex ratios were assayed by intra-strain crosses, i.e., 358 females and males from the same wild-type isofemale strains (tester strains). For most of the 359 female-male species, we had only one tester strain, except for four of the species (C. nigoni, 360 C. latens, C. remanei, and C. sp. 33). For these four species we had more than one strain 361 available at the start of this experiment, so the sex ratios of intra-strain crosses were tested 362 for multiple strains. We also conducted inter-strain crosses for these four species to test the 363 possibility that sex ratio bias was due to sex-specific consequences of inbreeding depression 364 (34). For C. nigoni, for which we had two strains, we intercrossed the two strains and then 365 crossed the heterozygous F1 males with the maternal strain. For the three species, C. latens, 366 C. remanei, and C. sp. 33, for which we had three strains, we first crossed two strains and then 367 crossed the heterozygous F1 males with the third strain (Table S1). We scored the two sexes 368 in the F2 progeny. 369 For the androdioecious species, we crossed males from wild-type strains (tester strain) 370 to hermaphrodite strains carrying a recessive mutation, so that the outcrossed progeny were 371 visually identifiable. The recessive morphological mutant strains had Uncoordinated (Unc) or 372 Dumpy (Dpy) phenotypes: C. elegans (BRC0189, unc-119(ed9)); C. briggsae (BRC0258,373 unc(ant10)); and C. tropicalis (BRC0419, dpy(ant23)). For each of these androdioecious 374 species, we had multiple tester strains, especially for C. briggsae, for which we had many 375 isolates collected in Taiwan. Because androdioecious species are normally inbred, we tested 376 sex ratios using males from different strains but did not generate heterozygous F1 males to 377 test an inter-strain effect. 378

Time limitations for mating 380
We set up mating experiments to assay sex ratio in the same manner across all the 381 species tested. In the "unlimited mating" experiment, we placed one L4 female (or 382 hermaphrodite) and one L4 male on a fresh 55 mm diameter Petri plate and then transferred 383 them together every one or two days to fresh plates until they died or produced no more 384 eggs. When the progeny reached the L4 or adult stage, the numbers of outcrossed females 385 (or hermaphrodites) and males per plate were manually scored under the microscope. For 386 the male-female species, all progeny were counted, while for the androdioecious species, 387 only wild-type cross progeny were counted whereas selfed Unc or Dpy progeny were ignored. 388 Mating pairs with total number of out-crossed progeny smaller than 40 were excluded from 389 further analyses to ensure adequate statistical power. For each test of the strains, we had at 390 least three replicate mating pairs. The counts of progeny of the two sexes for each replicate 391 were used for further statistical analyses (see below). 392 In addition to the unlimited mating experiment, to interrogate the potential role of 393 competition between male X-bearing sperm and the nullo-X counterpart in progeny sex ratio 394 bias, we conducted sex ratio assays by "limited mating" for one tester strain per species. To 395 do so, L4 females (hermaphrodites) and L4 males were isolated one day prior to the cross to 396 ensure their virginity and matured singly overnight. The next day, one male was added to one 397 isolated female (hermaphrodite). The male was removed when mating plugs were observed 398 on the females (hermaphrodites) or after five hours. Mated females (hermaphrodites) were 399 transferred daily to new plates. For C. tropicalis, all pairs failed to mate within five hours, so 400 we crossed one hermaphrodite with three males to increase the chance of mating. The 401 progeny sex ratios were scored as above. 402 403

Statistical analysis 404
For each of the sex ratio assays, unlimited mating or limited mating, we tested 405 whether the progeny sex ratio was biased. For each replicate within the tester strain, the 406 counts of total females (hermaphrodites) and total males were used for the binomial test. As 407 the majority of crosses yielded more female (hermaphrodite) than male progeny, we tested 408 whether the proportions of females (hermaphrodites) significantly exceeded equality (R, 409 binom.test, alternative = "greater"). The P-values of the replicates within a strain were 410 corrected for multiple testing for the numbers of replicates using the Benjamini-Hochberg 411 method (59) and then combined to yield the overall P-value for the strain (Fisher's method,R 412 package metaseqR (60)). For species with only one strain, the strain P-value was used to 413 represent the species. For species with multiple strains, the P-values of the strains were again 414 corrected for multiple strains and then combined to yield the P-value for the species. The 415 species P-values were corrected a final time for multiple testing for the 23 species. Species 416 with the corrected P-values smaller than 0.05 were defined as having sex ratio bias. Based on 417 the states of sex ratio bias or non-bias of the 23 species (unlimited mating) and the 418 phylogenetic tree (61), we inferred the ancestral state of sex ratio bias and the evolutionary 419 transitions between bias and non-bias among these species, using the maximum parsimony 420  Table 1 Figure S1. Day-by-day sex ratios of species not included in Figure 2 454 Table S1. Sex ratios of inter-stain crosses 455 Table S2. Sex ratios of 31 C. briggsae strains 456 Table S3. Day-by-day sex ratios by replicates of each species (unlimited mating) 457  Figure S1. Day-by-day sex ratios of species not included in Figure 2 limited mating