An application of the neutral model to the evolution of tail length in the genus Euplectes (Aves, Ploceidae)
Male tail length among the 16 species of widowbirds and bishops (genus Euplectes) varies considerably, from 3 to 50 cm (females are cryptically colored and have short tails), and recent evidence from three of these is consistent with the hypothesis that their long tails are adaptations arising from sexual selection (M. Andersson, 1982; Savalli, 1991; S. Andersson, 1992). This variation is within the range that can be accounted for by genetic drift, and suggests that current macroevolutionary models of neutral evolution are only of limited use in excluding drift as an explanation of morphological evolution.
A trait can become genetically fixed in a population by either random genetic drift (Kimura, 1983) or natural selection (Darwin, 1859). These two hypotheses are not mutually exclusive, but drift is generally only of interest when it operates without or against selection; when it works in the same direction as selection the effects of the two forces are identical. Evolution by natural selection can further be subdivided into natural selection that acts directly on a trait because of the fitness benefits of the trait, or indirectly, by acting on another trait that is pleiotropically or genetically linked. When investigating the evolution of a structure or behavior it is important to consider nonfunctional hypotheses (drift or indirect selection) as well as functional ones (Gould and Lewontin, 1979).
In the short term, the rate of morphological evolution due to drift is determined by the effective population size and heritability of the trait (Kimura, 1983). Several quantitative genetic models have been developed to predict the amount of phenotypic change in a population that is possible due to drift (Lande, 1976; Lynch and Hill, 1986; Turelli et al., 1988). The applicability of these methods to many evolutionary questions is limited since data on effective population size, the amount of change in a trait over time, and the heritability of the trait are often not available.
These models have been extended to allow for comparisons among multiple populations, permitting testing of macroevolutionary hypetheses (Lande, 1977; Lynch and Hill, 1986). Over a sufficient length of time the rate of possible morphological evolution per year can be predicted from the rate of input of new genetic variance by polygenic mutation, assuming no gene flow (Lynch and Hill, 1986; Lynch, 1988). This can be used to test if the amount of phenotypic divergence among isolated populations or species can be accounted for by drift, without needing information on the effective population size or heritability (Lynch and Hill, 1986; Turelli et al., 1988). If interpopulation variance is higher than predicted by drift, then directional selection is indicated, while if it is lower, then stabilizing selection must be the cause.
This test only provides a qualitative test of agreement with neutral models because the sampling errors of most of the parameters are unknown (Turelli et al., 1988). Because of the assumptions and inherent errors in estimating the parameters, the usefulness of these models is questionable. A number of studies of morphological evolution in a variety of fossil and extant organisms have failed to exclude drift (reviewed by Turelli et al., 1988), though in none of these was there contradictory evidence for selection operating on the traits.
Here I will apply the test developed by Turelli et al. (1988) to try to exclude genetic drift as a sufficient explanation for the interspecific variation in tail length occurring in widowbirds and bishops. This will determine if the model can provide results consistent with the findings of directional selection on tail length in some widowbirds. Since there is a tendency for the longer tailed species of widowbirds to be larger, which could account for variation in tail length, I will also examine the variation in two measures of size: tarsus length and wing length.
Materials and Methods
The lengths of the longest retrix, unflattened wing cord, and tarsus length (as described in Pyle et al., 1987) were measured for specimens of breeding-plumaged males of all species and subspecies of Euplectes available at the National Museums of Kenya in Nairobi and the American Museum of Natural History in New York.
Although Bjorklund (1991) restricted his similar analysis to the species level, it seems inappropriate to do this for widowbirds since a number of species have populations that are isolated by hundreds or even thousands of kilometers and differ in size and tail length (see Hall and Moreau, 1970; Savalli, 1991) and have undergone separate evolutionary trajectories. On the other hand, it was often not possible to determine if described subspecies represent genetically isolated populations or merely points on a cline. Therefore, I defined a population as all apparently contiguous subspecies of a species, with geographically isolated subspecies or groups of subspecies as separate populations. Geographic isolation was determined from the maps in Hall and Moreau (1970). In all, 22 populations for which sufficient specimens (at least five) were available were used in the analysis, including all species except the northern red bishop, E. franciscanus (Appendix).
I used the test for genetic drift developed by Turelli et al. (1988), following the method used by Bjorklund (1991). Turelli et al. (1988) derived the statistic
[Mathematical Expression Omitted]
where t is the number of generations, [V.sub.m] is the rate of input of new genetic variance due to mutation per year, and [S.sub.2] is the variance between species, determined from a one-way ANOVA using the equation
[Mathematical Expression Omitted]
where [MS.sub.b] is the between-population mean square, [MS.sub.w] is the within-population mean square, and [n.sub.o] is the average sample size (Bjorklund, 1991).
Thus, we can determine the upper and lower 95% confidence limits for the amount of the between-population mean square that can be attributed to drift as
[Mathematical Expression Omitted]
(Turelli et al., 1988; Bjorklund, 1991).
Although [V.sub.m] will vary depending on the trait measured and units used, the ratio [V.sub.m]/[V.sub.e], where [V.sub.e] is the environmental component of variance for a trait, has been found to be relatively constant for a wide variety of organisms, with a range between [10.sup.-4] and 5 x [10.sup.-2] (Lynch, 1988). Assuming a typical heritability of between 0.25 and 0.75, the phenotypic variance, [V.sub.e] can be substituted for [V.sub.e] (Turelli et al., 1988). Because Lynch's measurements of [V.sub.m] include deleterious and advantageous mutations as well as the strictly neutral mutations required by the model, these estimates are conservative upper bounds of the mutational input. In order to use Lynch's constant, we need to divide equation (1) by [V.sub.e] yielding
[Mathematical Expression Omitted]
Since exact divergence times for populations of Euplectes are not known, it is necessary to determine the range of time over which divergence could have occurred. Euplectes is most closely related to the queleas, Quelea spp. DNA hybridization between species in the two genera yields a [DELTA][T.sub.50] H of 2.5 (Sibley and Ahlquist, 1990, p. 867). Recent results suggest that the amount of genetic change necessary for a [DELTA][T.sub.50] H of 1.0 represents about two to three million years of divergence for typical passerines that begin breeding at the age of one or two years (Sibley and Ahlquist, 1990, p. 641). Thus, Euplectes and Quelea probably diverged about seven million years ago. This figure will be used as a maximum estimate of the time since divergence within Euplectes. The most recent divergence of widowbird populations probably occurred during the last expansion of forests in Africa, which would have resulted in a fragmentation of the widowbirds' grassland habitat. This began about 12,000 years ago (Hamilton, 1976, 1982) and this age will be used as a minimum.
Substituting [V.sub.m]/[V.sub.e] = [10.sup.-4], appropriate F-values and t = 7 x [10.sup.6] and t = 1.2 x [10.sup.4] for the upper and lower limits, respectively, into equation (2), we get a maximum amount of interpopulation variation possible due to drift of 2,366 and a minimum value of 4.06. Thus, genetic drift can account for interpopulation variation that ranges from 4.06 to 2,366.
Based on the results of an ANOVA (Table 1), the interpopulation mean square divided by the within-population mean square (as an estimate of [V.sub.e]) for tail length, [S.sub.2]/[MS.sub.w] = 42.97, which is less than the maximum that can be accounted for by drift. Repeating the analysis for the body size measurements, we get [S.sub.2]/[MS.sub.w] = 16.81 for wing length, and [S.sub.2]/[MS.sub.w] = 6.57 for tarsus length. Therefore body size also falls within the range of divergence that can be accounted for by genetic drift alone, though tarsus length approaches the lower limit, suggesting stabilizing selection.
Table 1. - Results from one-way ANOVAs on three male traits, based on data in Appendix. Average Between Within sample -populations populations size ([n.sub.0]) mean square mean square Tail length 17.05 201171 274.303 Wing length 14.09 3895.23 16.375 Tarsus length 13.64 127.325 1.192
The results do not allow us to reject the hypothesis of drift for any of the traits examined. This does not, of course, mean that body size and tail length have not undergone selection, but only that selection was not sufficiently strong to eliminate the possibility of drift. Indeed, it is hard to imagine that the enormous variation in tail length, from 3 to 50 cm (body mass varies only three fold), could be due only to drift. The inability to exclude drift despite this variation and experimental evidence for selection favoring long tails in three species (M. Andersson, 1982; Savalli, 1991; S. Andersson, 1992) is most likely due to the crude estimates for many of the parameters of the model. The time since divergence, for example, represents the oldest possible divergence within Euplectes; no doubt many of the divergences occurred much more recently. Furthermore, selection is unlikely to be constant over time: periods of strong directional selection followed by periods of stabilizing selection would appear to give a weaker response over a long time period, resulting in rates of change indistinguishable from drift. Directional selection acting on all populations (parallel evolution) would also not be detected by this method. Other studies also indicate that most typical rates of morphological evolution are not sufficiently great to allow them to be distinguished from drift; the few that did show selection were in laboratory populations of Drosophila (Turelli et al., 1988). Similar types of analyses of allozyme allele frequencies have also encountered statistical difficulties when selection was other than directional (e.g., Rothman and Templeton, 1980). These observations suggest that the models presented by Lande (1977), Lynch and Hill (1986) and Turelli et al. (1988) have only very limited use in testing macroevolutionary hypotheses and under most circumstances genetic drift remains an untestable hypothesis for explaining macroevolutionary patterns.
It is instructive to examine one study which did claim to find rates of morphological evolution that exceeded that possible by drift. Bjorklund (1991) found that variance in male tail length of grackles (Quiscalus spp., Icterinae) was too large to be accounted for by drift while variance in tarsus length was not. Unfortunately, he made an error by substituting Lynch's estimate of [10.sub.-4] into equation (2) for [V.sub.m], and then comparing to [S.sub.2] derived from an ANOVA of tail length. [S.sub.2] is not a dimensionless number, however, as is [V.sub.m]/[V.sub.e], and its value depends on the units used. It cannot be compared to a value incorporating Lynch's dimensionless (and therefore unit-independent) estimate of [V.sub.m]/[V.sub.e], Bjorklund does not present the details of his ANOVA, so it is not possible to incorporate an estimate of [V.sub.e] to see how his results and conclusions might change.
Andersson, M. 1982. Female choice selects for extreme tail length in a widowbird. Nature 299:818-820. Andersson, S. 1992. Female preference for long tails in lekking Jackson's widowbirds: Experimental evidence. Anim. Behav. 43:379-388. Bjorklund, M. 1991. Sexual dimorphism and mating system in the grackles (Quiscalus spp.: Icterinae). Evolution 45:608-621. Darwin, C. 1859. The Origin of Species by Means of Natural Selection. John Murray, London, UK. Gould, S. J., and R. C. Lewontin. 1979. The spandrels of San Marco and the Panglossian paradigm: A critique of the adaptationist program. Proc. R. Soc. London B 205:581-598. Hall, B. P., and R. E. Moreau. 1970. An Atlas of Speciation in African Passerine Birds. British Museum (Natural History), London, UK. Hamilton, A. C. 1976. The significance of patterns of distribution shown by forest plants and animals in tropical Africa for the reconstruction of upper Pleistocene palaeoenvironments. Palaeoecol. Africa 9:63-97. _____. 1982. Environmental History of East Africa. Academic Press, N.Y., USA. Kimura, M. 1983. The Neutral Theory of Molecular Evolution. Cambridge University Press, UK. Lande, R. 1976. Natural selection and random genetic drift in phenotypic evolution. Evolution 30: 314-334. _____. 1977. Statistical tests for natural selection on quantitative characters. Evolution 31:442-444. Lynch, M. 1988. The rate of polygenic mutation. Genet. Res. 51:137-148. Lynch, M., and W. G. Hill. 1986. Phenotypic evolution by neutral mutation. Evolution 40:915-935. Pyle, P., S. N. G. Howell, R. P. Yunick, and D. F. DeSante. 1987. Identification Guide to North American Passerines. Slate Creek Press, Bolinas, CA USA. Rothman, E. D., and A. R. Templeton. 1980. A class of models of selectively neutral alleles. Theor. Popul. Biol. 18:135-150. Savalli, U. M. 1991. Sexual selection and the evolution of tail length in the yellow-shouldered widowbird. Ph.D. Diss. University of California, Berkeley, USA. Sibley, C. G., and J. E. Ahlquist. 1990. Phylogeny and Classification of Birds. Yale University Press, New Haven, CT USA. Turelli, M., J. H. Gillespie, and R. Lande. 1988. Rate tests for selection on quantitative characters during macroevolution and microevolution. Evolution 42:1085-1089.
|Printer friendly Cite/link Email Feedback|
|Author:||Savalli, Udo M.|
|Date:||Apr 1, 1993|
|Previous Article:||Sexual selection for increased male courtship and acoustic signals and against large male size at sharp-tailed grouse leks.|
|Next Article:||Nepotism in a solitary wasp as revealed by DNA fingerprinting.|