Sex ratio selection and the evolution of environmental sex determination in laboratory populations of Menidia menidia.
When should ESD as opposed to GSD evolve? GSD presumably evolves because, as Fisher (1930) first pointed out, the evolutionary stable strategy for many outbreeding species is equal investment in production of sons and daughters. If one sex is less numerous, genes that produce the minority sex increase until the sex ratio is balanced. Mechanisms of GSD, such as heterogamety, that ensure production of a balanced sex ratio may thus be explained. ESD evolves when: i) the environment that offspring experience affects the fitness of each sex differently; ii) the environment that offspring (1) Present address: National Oceanic and Atmospheric Administration, National Marine Fisheries Service, 1335 East-West Highway, Silver Spring, MD 20910 USA. enter cannot be chosen; and iii) offspring from different environments mate with one another (Charnov and Bull, 1977). Hence, when the environment is patchy, ESD ensures that offspring become the sex with higher fitness given the environmental patch encountered. As long as the environmental determinant of sex correlates with, or is a reliable cue of, the type of environment that offspring experience, ESD is superior to GSD because individuals with GSD would often develop arbitrarily into the sex with lower fitness. Analytical models of this evolutionary process suggest, however, that there is a drawback to ESD that depends on the variability of the environment (Bull, 198 1 a, 1981b; Bulmer and Bull, 1982; Bull and Bulmer, 1989). Environmental fluctuations cause the population sex ratio to vary in species with ESD, resulting in frequency-dependent Fisherian selection for GSD. Moderate fluctuations result in mixed ESD-GSD systems while larger fluctuations may ultimately lead to complete loss of ESD (this outcome also depends on life span; see Bull and Bulmer, 1989).
Direct empirical tests of the evolutionary processes described above have been nonexistent, chiefly because they require a species with genetic variation in ESD and a life cycle that is amenable to manipulation. Many species appear to have either strict GSD (e.g., birds, mammals) or ESD (e.g., reptiles) with little among or within population variability in the degree of environmental versus genetic control (exceptions include Conover and Heins, 1987a, 1987b; Lagomarsino and Conover, 199 3; Naylor et al., 1988; Blackmore and Charnov, 1989). Hence, there is little direct evidence that ESD is capable of evolving.
One species ideally suited for testing sex determination theory is the Atlantic silverside, Menidia menidia. This species has temperature-dependent sex determination but displays variation in the level of ESD and GSD both within and among local populations. Conover and Van Voorhees (I 990) exploited the unique characteristics of this natural system to provide the first direct demonstration of Fisher's sex ratio principle. Laboratory populations of M. menidia were forced to undergo frequency-dependent sex ratio selection by rearing each over successive generations in artificial constant-temperature environments that, in some, initially caused highly biased sex ratios. In all cases, increases in the proportion of the minority sex occurred until a balanced sex ratio was established thus confirming Fisher's argument.
Here we address the issue of whether (or not) the sex ratio changes in the laboratory populations described above were accompanied by concomitant changes in the level of ESD. How did the response of sex ratio to temperature change in each population? Was the level of ESD altered in a manner predicted by sex determination theory? We also provide data from additional generations further confirming our original conclusion that balanced sex ratios were established and then maintained for several generations until termination of the experiment.
We can envision two general ways in which the sex determining mechanism of M. menidia might change to shift the sex ratio from an unbalanced to a balanced condition. First, there may be selection for temperature-insensitive genotypes, leading to a decline in the level of ESD. For example, those few fish that remain female despite being reared at high temperature may possess temperature-insensitive, female-determining genes. Temperature-insensitivity would therefore increase along with the frequency of female determining genes in subsequent generations. If so, then the sex ratio produced at different temperatures should converge. Ultimately, a sex ratio of 0.5 at all temperatures should arise. This outcome is the one predicted by sex determination theory (Bull, 1983).
An alternative way to achieve a balanced sex ratio at a given temperature is to adjust the relationship between temperature and sex ratio without necessarily altering the level of thermal plasticity in sex determination. For example, fish that become female at high temperature may be those that possess genes with a higher temperature threshold for female determination. Hence, temperatures that are female-producing increase in subsequent generations. Selection is for genes with sensitivity to a different range of temperatures rather than lack of sensitivity to temperature. In this case we expect to see that an increase in sex ratio at one temperature also increases sex ratio proportionally at other temperatures. Hence, ESD is retained rather than being converted to GSD.
MATERIALS AND METHODS
The Atlantic silverside (Pisces: Atherinidae) is a common marine fish inhabiting bays, estuaries and salt marshes of the Atlantic coast of North America (Conover and Ross, 1982). Its one-year life cycle in nature can be completed in only six to nine months under laboratory conditions, making it especially well suited to multigenerational studies.
The sex of silversides is determined by an interaction between temperature and genotype during a specific period of larval development (Conover and Kynard, 1981; Conover and Fleisher, 1986). Previous experiments have verified that temperature during development directly influences primary sex determination rather than causing differential mortality between the sexes (Conover and Kynard, 1981). In general, low temperatures produce mostly female offspring and high temperatures produce mostly male offspring. But strong genetic influences, probably involving a few genes with major effects, are indicated by the following (based on Conover and Heins, 1987a, 1987b): 1) at least a few members of both sexes are produced even at the most extreme temperatures encountered in nature; 2) there are large nonadditive effects of parentage on family sex ratio; and 3) family sex ratios within temperatures tend to vary in a discrete Mendelian pattern (e.g., 0, 0.25, 0.5, 1.0), at least in northern populations (see also Lagomarsino and Conover, 1993).
The temperature-genotype interaction differs markedly among populations from different latitudes. The sex ratio of fish from Nova Scotia tends to be near 0.5 at all temperatures, indicating near complete genotypic control of sex determination. Temperature-dependent genotypes exist in Nova Scotia but are rare (Lagomarsino and Conover, 1993). The sex ratio of fish from South Carolina, however, changes by as much as 70% with temperature (Conover and Heins, 1987b; Lagomarsino and Conover, 1993). Fish from New York display an intermediate temperature-sex ratio response.
The primary objective was to test Fisher's theory that balanced sex ratios evolve by frequency-dependent selection of the minority sex. As mentioned above, this was done by establishing five separate laboratory populations of M. menidia in different constant temperature environments that would create highly skewed ratios in the first generation. Shifts in sex ratio (if any) were measured in subsequent generations. The general plan has been thoroughly described in Conover and Van Voorhees (1990). Key features are repeated below.
Two populations were founded in 1984 with several thousand embryos collected from a mass spawning site in South Carolina (field collection procedures and specific locations are the same as reported in Conover and Heins, 1987b). After hatching, the larvae were randomly subdivided into two groups. One group was reared in temperature-controlled seawater baths at 28 [degrees] C (SC-H) and the other at 17 [degrees] C (SC-L) during the temperature-sensitive period of sex determination (for rearing techniques see Conover and Fleisher, 1986). These two temperatures were chosen because they represent levels that both produce skewed sex ratios and also allow sufficiently high growth and survival. Afterward, juveniles were transferred to large (800 liter) rearing tanks at 20-25 [degrees] C. Excess juveniles were culled periodically and roughly 100 fish were reared to adult size. Maturity was then induced by photoperiod manipulations (Conover and Fleisher, 1986).
Silversides typically spawn in large schools, broadcasting their eggs over filamentous algae (Middaugh, 1981; Conover and Kynard, 1984). In the laboratory, mass spawnings occur daily and we collected eggs on mops of yam placed in each tank. The progeny from the SC-L and SC-H lines were then transferred to the seawater baths and reared at the same temperature during the sensitive period as were their parents. To achieve a representative sample of offspring from each parental generation, several thousand eggs were allowed to collect on the yam mops for several days before removal, and at least three to five such batches of offspring were combined to form the next generation. After the temperature-sensitive period, excess juveniles were preserved to estimate the sex ratio and again [is greater than or equal to] 100 fish were raised to maturity. All succeeding generations were treated in the same manner as described above. All fish were sexed by dissection and direct examination of gross morphology under a microscope using validated techniques described in Conover and Fleisher (1986).
Two other populations were founded in 1985 from a common pool of embryos collected from New York: NY-H, reared at 28 [degrees] C; and NY-L, reared at 17 [degrees] C during the sensitive period each generation. The fifth population was established in 1985 with embryos collected from Nova Scotia. This lab population (NS-H) was reared at 28 [degres] C each generation.
The mean (and range) of the number of breeding adults in each generation for each population was: SC-H = 151 (62-220), SC-L = 114 (48-220), NY-H = 153 (41-246), NY-L = 119 (42-198), and NS-H = 165 (50-428). Mean effective population sizes N,) over the course of the experiment for each population were SC-H = 87; SC-L = 60; NY-H = 48; NY-L = 80; NS-H = 113. Effective population sizes were generally less than number of breeding adults because of the effect of skewed sex ratios. These values of [N.sub.e] should be considered maximum estimates because they assume that all fish exposed to photoperiod manipulation contribute equally to spawning.
The initial sex ratios (F/F + M) in the five populations were: SC-H = 0.18; SC-L = 0.70; NY-H = 0.05; NY-L = 0.29; and NS-H = 0.53. Hence, the five populations began with varying degrees of sex ratio bias. Note that because each of the three donor populations have different sex ratio responses to temperature, the direction of initial sex ratio biases are not the same in each temperature regime: e.g., SC-L starts with an excess of females, but NY-L starts with a male excess. NS-H starts with a balanced sex ratio and therefore serves as a control. This set of initial conditions potentially allows separation of the effects (if any) of common thermal environment from the frequency-dependent effects of sex ratio biases.
Measuring Changes in ESD
Changes in the level of environmental versus genotypic control of sex determination were measured each generation by rearing some offspring from each selected line at the temperature opposite to that experienced by their parents. For example, some embryos from the SC-H population were reared at 17 [degrees] C instead of 28 [degrees] C during the temperature-sensitive period. Likewise, some NY-L offspring were reared at 28 [degrees] C, and so on for the other three populations. All such offspring not raised at the same temperature as their parents were sacrificed as juveniles and did not contribute to future generations. The sex ratio at the alternative temperature therefore has no influence on sex ratio evolution in the five laboratory populations. The difference in sex ratio between temperatures provides an index of the "Ievel of ESD" within each generation of a given selected line.
RESULTS AND DISCUSSION
Changes in Sex Ratio
All four populations that began with biased sex ratios showed increases in the proportion of the minority sex until a balanced sex ratio was established it the temperature where each line was maintained (i.e., the trajectories depicted with solid lines in Fig. 1). Thereafter, sex ratios fluctuated around 0.5. This occurred independent of the temperature each population experienced. For example, the SC-L and NY-L lines began with sex ratios skewed to opposite extremes but both ultimately produced balanced sex ratios. Moreover, the NS-H population maintained a sex ratio at or near 0.5 throughout the experiment (Fig. 1).
The agreement of these results with Fisher's prediction was considered extensively in Conover and Van Voorhees (1990) and that discussion will not be repeated here. We wish to emphasize, however, that perhaps the strongest evidence of frequency-dependent selection in this experiment is the direction of sex ratio change following each generation where the sex ratio differed significantly from 0.5. Each such case is an opportunity for frequency-dependent selection to operate. Summed over all five populations, there were 26 times when the sex ratio deviated significantly from 0.5. The minority sex increased in the next generation on 25 of these occasions. Clearly, the changes in sex ratio among these populations are not random but instead correspond closely with what Fisher predicted.
Changes in ESD
There were substantial changes in the response of sex ratio to temperature among the four lines subjected to frequency-dependent selection. The level of ESD in the founders of the two SC populations was about 0.5. In the first few generations of selection, the SC-H population showed an initial decline in the level of ESD: i.e., as the male-biased sex ratio at 28 [degrees] C increased to 0.5, the female-biased sex ratio at 17[degrees] C declined, thus causing a decrease in the level of ESD (Figs. 1A, 2A). After the sex ratio at 28 [degrees] C reached 0.5 (generation 4), the level of ESD increased and then declined again to a level where the difference in sex ratio between temperatures was about 0.10-0.25 over the last five generations. In contrast, the sex ratio of the SC-L line declined greatly at both temperatures after only one generation of selection (Fig. 1B). The sex ratio at 28 [degrees] C in the SC-L line became nearly all male and remained so even after the sex ratio at 17 [degrees] C increased in later generations and began to oscillate around 0.5. Hence, the level of ESD in the SC-L line at the end of the experiment was similar to the level at the experiment's inception, i.e., about 0.5 (Fig. 2B).
The level of ESD among the founders of the two NY populations was about 0.25. In NY-H, the selected sex ratio at 28 [degrees] C became 0.5 after only one generation of selection (Fig. 1C). The sex ratio at 17 [degrees] C in NY-H fish was also very near to 0.5 from the second generation onward. Hence, the NY-H fish appear to have lost most of their sensitivity to temperature: in three of the last four generations, the difference in sex ratio among temperatures was very close to zero (Fig. 2C). In contrast, the NY-L line consistently produced a sex ratio of 0.5 from the fourth generation onward, but the level of ESD did not change. Although the sex ratio at 28 [degrees] C in NY-L fish was increased somewhat (Fig. ID), the difference in sex ratio with temperature remained constant at about 0.25 (Fig. 2D).
In the NS-H line, sex ratios at each temperature tracked one another closely and fluctuated around 0.5 throughout the experiment (Fig. 1E). There was evidence of only minimal thermal influence: sex ratio did not differ significantly with temperature in 8 of 11 comparisons (Fig. 2E) and 95% binomial confidence limits overlapped substantially in all but one generation. A test of combined probabilities showed that there was a significant overall effect of temperature on sex ratio (Fisher's combined test, P < 0.01; Sokal and Rohlf, 1981), suggesting the existence of a very low but significant level of ESD in NS fish. Similar results were reported for NS fish from nature by Lagomarsino and Conover (1993).
It is clear from our results that ESD is capable of evolving in a manner that produces a balanced sex ratio after an environmental shift. At the end of the experiment, for example, the temperature that produces a balanced sex ratio in the SC-L line is 17 [degrees] C, whereas the SC-H line produces a balanced sex ratio at 28 [degrees] C. Correspondingly, the sex ratio produced at 28 [degrees] C in SC-H fish ( 0.5) was, after selection, very different from the sex ratio produced at 28 [degrees] C in the SC-L line ( 0.0). Given that these two populations initially had identical sex ratio-temperature responses, the results provide dramatic evidence of the capacity for ESD to evolve. Similar conclusions can be drawn from contrasts between the NY-L versus NY-H populations.
Although all four populations that started with skewed sex ratios ultimately produced a balanced sex ratio, the manner in which ESD changed among them was not consistent. SC-H and especially NY-H both showed loss of ESD, with sex ratios at each temperature tending to converge toward one another. In SC-L and NY-L, however, there was at least as much ESD at the end of the experiment as there was at the beginning. These latter two populations achieved a balanced sex ratio by an adjustment in the response of sex ratio to temperature rather than an overall reduction in temperature sensitivity. Hence, in the process of producing a balanced sex ratio, both of the predicted modifications in ESD were observed within each of the two founding populations that originally had ESD.
Both lines that showed a reduction in the level of ESD were high temperature treatments and the two that did not were from low temperature environments. This tentatively suggests that although changes in sex ratio at the selected temperature were a direct response to frequency-dependence (i.e., all populations showed increases in the minority sex irrespective of temperature), evolution of the sex determining mechanism was affected directly by the thermal environment. The reason for this is not clear, but perhaps it reflects an asymmetry in the way temperature and genotype interact to determine sex at different temperatures. The genes that determine sex at one temperature, for example, may not be the same as those that affect sex at another temperature.
In natural populations of the Atlantic silverside, ESD appears to be adaptive (Conover, 1984): most fish born during low temperatures that prevail at the beginning of the growing season become female whereas those born late in the breeding season experience higher temperatures that cause them to become male. By virtue of a longer growing season, females tend to be larger than males and large size seems to enhance the reproductive success of females more than males. Changes in the level of ESD across latitudes appear to be correlated with differences in the length of the breeding season that affect the relative benefit of ESD (Conover and Heins, 1987b). The design of our experiment, however, eliminated the adaptive value of ESD (because the environments that males and females enter are identical) while simultaneously magnifying the major drawback of ESD; i.e., the skewed ratios that occur under environmental extremes. Why then was a shift from ESD to GSD not consistently observed?
One reason may be that balanced sex ratios were achieved too quickly in our laboratory populations. The experiment empirically modeled the processes that occur when a population with ESD is introduced to an extreme but constant environment. In constant environments, there is no further selection for or against ESD or GSD after a sex ratio of 0.5 is produced. This may explain why ESD declined initially in some lines but then rebounded somewhat after a sex ratio of 0.5 was reached. It may be that repeated perturbations from a balanced sex ratio are generally required to eliminate ESD. Experiments could be designed where selection against ESD was maintained by repeatedly shifting populations among extreme environments before a balanced sex ratio was established.
Adaptive variation in ESD appears to occur among natural populations of several species (Naylor et al., 1988; Blackmore and Charnov, 1989), including M. menidia, but our results are the first to demonstrate that mechanisms of ESD are capable of evolving. Thus differences in ESD among natural populations may, in fact, be the result of selection and adaptation to local environments. The most compelling evidence is, perhaps, the changes that occurred in the NY-H population, which appears to have converted from a moderate ESD to a GSD system like that found in fish from Nova Scotia. If such transitions can occur in the laboratory, then it is certainly plausible that variation in ESD and GSD among natural populations of M. menidia and other organisms is a function of climate-mediated sex ratio selection in local environments.
We thank S. Heins, R. McBride, R. Nyman, L. Chiarella, M. Meade, M. Green, T. Present, 1. Lagomarsino, F. Juanes, R. Marks, R. G. Rowland, K. Sosebee, J. Schreer, and R. Tegge for assistance in rearing the fish. This research was supported by the National Science Foundation through grants nos. BSR84-15878 and BSR87-17315 to D.O.C. This is Contribution 844 of the Marine Sciences Research Center, State University of New York, Stony Brook, NY USA.
Adams, J., P. Greenwood, and C. Naylor. 1987. Evolutionary aspects of environmental sex determination. Int. J. invertebr. Reprod. Dev. 11:123-136. Blackmore, M. S., and E. L. Charnov. 1989. Adaptive variation in environmental sex determination in a nematode. Am. Nat. 134:817-823. BULL, J. J. 1981a. Sex ratio evolution when fitness varies. Heredity 46:9-26. --. 1981b. Evolution of environmental sex determination from genotypic sex determination. Heredity 47:173-184. --. 1983. Evolution of Sex Determining Mechanisms. Benjamin/Cummings, Menlo Park, CA USA. Bull, J. J., and M. G. Bulmer. 1989. Longevity enhances selection of environmental sex determination. Heredity 63:315-320. Bull, J. J., and E. L. Charnov. 1989. Enigmatic reptilian sex ratios. Evolution 43:1561-1566. Bulmer, M. G., and J. J. Bull. 1982. Models of polygenic sex determination and sex ratio control. Evolution 36:13-26. Charnov, E. L., and J. J. Bull. 1977. When is sex environmentally determined? Nature 266:828-830. Conover, D. O. 1984. Adaptive significance of temperature-dependent sex determination in a fish. Am. Nat. 123:297-313. Conover, D. O., and M. Fleisher. 1986. The temperature-sensitive period of sex determination in Menidia menidia. Can. J. Fish. Aquat. Sci. 43:514-520. Conover, D. O., and S. W. Heins. 1987a. The environmental and genetic components of sex ratio in Menidia menidia. Copeia 1987:732-743. --. 1987b. Adaptive variation in environmental and genetic sex determination in a fish. Nature 326: 496-498. Conover, D. O., and B. E. Kynard. 1981. Environmental sex determination: interaction of temperature and genotype in a fish. Science 213:577-579. --. 1984. Field and laboratory observations of spawning periodicity and behavior of a northern population of the Atlantic silverside, Menidia menidia (Pisces: Atherinidae). Environ. Biol. Fishes 11:161-171. Conover, D. O., and M. R. Ross. 1982. Patterns in seasonal abundance, growth, and biomass of the Atlantic silverside, Menidia menidia, in a New England estuary. Esutaries 5:275-286. Conover, D. O., and D. A. Van Voorhees. 1990. Evolution of a balanced sex ratio by frequency-dependent selection in a fish. Science 250:1556-1558. Deeming, D. C., and M. W. J. Ferguson. 1988. Environmental regulation of sex determination in reptiles. Phil. Trans. R. Soc. London B 322:19-39. Fisher, R. A. 1930. The Genetical Theory of Natural Selection. Clarendon, Oxford, UK. Haed, G., R. M. May, and L. Pendleton. 1987. Environmental determination of sex in the reptiles. Nature 329:198-199. Janzen, F. J., and G. L. Paukstis. 1991a. A preliminary test of the adaptive significance of environmental sex determination in reptiles. Evolution 45: 435-440. --. 1991b. Environmental sex determination in reptiles: Ecology, evolution, and experimental design. Q. Rev. Biol. 66:149-179. Lagomarsino, I., and D., O. Conover. 1993. Variation in environmental and genotypic sex determining mechanisms across a latitudinal gradient in the fish Menidia menidia. Evolution. 47:In press. Middaugh, D. P. 1981. Reproductive ecology and spawning periodicity of the Atlantic silverside, Menidia menidia (Pisces: Atherinidae). Copcia 1981: 766-776. Naylor, C., J. Adams, and P. J. Greenwood. 1988. Variation in sex determination in natural populations of a shrimp. J. Evol. Biol. 1:355-368. Sokal, R. R., and F. J. Rohlf. 1981. Biometry. W. H. Freeman and Co., San Francisco, CA USA. Sullivan, J. A., and R. J. Schultz. 1986. Genetic and environmental basis of variable sex ratios in laboratory strains of Poeciliopsis lucida. Evolution 40:152-158.
|Printer friendly Cite/link Email Feedback|
|Author:||Conover, David O.; VanVoorhees, David A.; Ehtisham, Amir|
|Date:||Dec 1, 1992|
|Previous Article:||Are trioecy and sexual liability in Atriplex canascens genetically based?: Evidence from clonal studies.|
|Next Article:||Effects of progeny and size of the pollen load on progeny performance in Campanula americana.|