Nuclear Gene Genealogies Reveal Historical, Demographic and Selective Factors Associated With Speciation in Field Crickets

Corresponding author: Richard E. Broughton, University of Oklahoma, 111 E. Chesapeake St., Norman, OK 73019., rbroughton{at}ou.edu (E-mail)

Communicating editor: J. HEY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Population genetics theory predicts that genetic drift should eliminate shared polymorphism, leading to monophyly or exclusivity of populations, when the elapsed time between lineage-splitting events is large relative to effective population size. We examined patterns of nucleotide variation in introns at four nuclear loci to relate processes affecting the history of genes to patterns of divergence among natural populations and species. Ancestral polymorphisms were shared among three recognized species, Gryllus firmus, G. pennsylvanicus, and G. ovisopis, and genealogical patterns suggest that successive speciation events occurred recently and rapidly relative to effective population size. High levels of shared polymorphism among these morphologically, behaviorally, and ecologically distinct species indicate that only a small fraction of the genome needs to become differentiated for speciation to occur. Among the four nuclear gene loci there was a 10-fold range in nucleotide diversity, and patterns of polymorphism and divergence suggest that natural selection has acted to maintain or eliminate variation at some loci. While nuclear gene genealogies may have limited applications in phylogeography or other approaches dependent on population monophyly, they provide important insights into the historical, demographic, and selective forces that shape speciation.

PHYLOGENIES of alleles or haplotypes (gene genealogies) can reveal the sequence and timing of speciation (lineage-splitting) events and the geographic context in which they occurred (HUDSON 1990 ; TEMPLETON 1998 ; WAKELEY and HEY 1998 ; KLIMAN et al. 2000 ). However, random sorting of ancestral polymorphisms, hybridization and introgression, and natural selection obscure the signature of previous history; therefore, gene trees and species trees may not be congruent (NEIGEL and AVISE 1986 ; PAMILO and NEI 1988 ; TAKAHATA 1989 ; WU 1991 ; HUDSON 1992 ; NICHOLS 2001 ). For any single locus, inferences of recent evolutionary history are most reliable if coalescence times for samples of alleles are less than the time between lineage-splitting events. Under these circumstances, diverging lineages become exclusive groups (groups in which all members are more closely related to each other than to any individual outside of the group) before further lineage-splitting events occur.

Population bottlenecks or persistent small N_e will reduce coalescence times for all loci; diverging lineages will become exclusive groups more quickly. Fixation of new variants via directional selection ("selective sweeps") will also reduce coalescence times, but only for markers tightly linked to the selected locus. In contrast, balancing selection promotes the maintenance of polymorphism and thus increases coalescence times; lineages become exclusive groups more slowly. Of course, the footprint of natural selection in the genome depends on the strength of selection and the local recombination rate; if selection is strong and recombination rates are low, then a selective sweep may carry to fixation a relatively large genomic region (KAPLAN et al. 1989 ), and variation at linked markers will reflect the fixation event. Weak selection and high rates of recombination mean that only a very small region will be affected by a selective sweep or by balancing selection (TING et al. 2000 ).

The relative merits of nuclear vs. mitochondrial loci for evolutionary studies have been much discussed (e.g., WU 1991 ; MOORE 1995 ; HOELZER 1997 ; AVISE 2000 ; HARE 2001 ). That coalescence times for mitochondrial genes will be only one-quarter those for nuclear genes (BIRKY 1991 ) is an argument in favor of using such genes for inferring the sequence of recent lineage-splitting events (MOORE 1995 ). However, the absence of recombination in mtDNA means that a selective sweep acting anywhere in the mitochondrial genome will carry a single haplotype to fixation, reducing variation across the entire genome. Amounts of variation will then no longer reflect historical effective population sizes, and information on population history is lost.

Among nuclear genes, there is also considerable variation with respect to the probability that they will provide clear resolution of past lineage-splitting events. Both persistent ancestral polymorphisms and introgression subsequent to secondary contact may produce nuclear gene genealogies that are incongruent with each other and with relationships defined by morphology. For divergent lineages in which barriers to gene exchange are incomplete, boundaries are often semipermeable or porous (HARRISON 1991 ; WU 2001 ; MACHADO et al. 2002 ). That is, the genetic architecture of reproductive isolation has important consequences for patterns of differentiation and introgression. Therefore, gene genealogies for closely related "species" will vary with gene region (see HEY 1994 ; WANG et al. 1997 ; TING et al. 2000 ), and in regions closely linked to genes that have experienced recent selective sweeps and/or genes that contribute to reproductive isolation, genealogies will more likely reveal the two forms as exclusive groups (TING et al. 2000 ).

Comparative studies of gene genealogies for multiple loci are still not common in nonmodel organisms, and therefore the data to test the ideas discussed above are few. Here we analyze sequence variation in introns of four nuclear gene loci in North American field crickets (genus Gryllus) and compare these data with already published data on mtDNA sequence variation in these insects. Our goal is to elucidate genealogical patterns within and among this group of closely related species and to explain observed patterns in terms of a recent history of demographic change and natural selection.

North American crickets in the genus Gryllus are a group of closely related species that have been well characterized morphologically, behaviorally, ecologically, and genetically (ALEXANDER 1957 , ALEXANDER 1968 ; HARRISON 1979 , HARRISON 1983 , HARRISON 1986 ; Rand AND HARRISON 1989 ; HARRISON and BOGDANOWICZ 1995 , HARRISON and BOGDANOWICZ 1997 ; WILLETT et al. 1997 ). From 1915 until 1957 all field crickets in North America were considered to be a single species, on the basis of REHN and HEBARD's (1915) assessment that different morphological forms represented the adaptation of a very plastic species to local environmental conditions. Although morphologically similar, the crickets have clearly diverged in ecology (habitat, life cycle) and behavior (calling song), and variation in these traits formed the initial basis for recognizing multiple independent lineages (ALEXANDER 1957 ). The most distinctive mtDNA haplotypes among those sampled from all of the North American species are only 6.8% (HUANG et al. 2000 ) or 8% (HARRISON and BOGDANOWICZ 1995 ) divergent, suggesting recent (Pliocene/Pleistocene) diversification.

MtDNA restriction site and sequence data indicate that the three egg-overwintering species (Gryllus pennsylvanicus, G. firmus, and G. ovisopis) are all very closely related (sequence divergence of <1%). G. ovisopis is found in woodlands in northern Florida and southern Georgia. Both G. pennsylvanicus and G. firmus are widespread in disturbed habitats. They overlap and hybridize along an extensive zone that runs from southern Connecticut, across the Hudson Valley, and then along the eastern front of the Appalachian and Blue Ridge Mountains (HARRISON and ARNOLD 1982 ). G. pennsylvanicus is found in inland and upland localities while G. firmus is found along the coastal plain. Although individuals of mixed ancestry are found in the hybrid zone, allele frequency differences at three allozyme loci and diagnostic haplotypes for mtDNA and anonymous nuclear restriction fragment length polymorphisms (RFLPs) are characteristic of allopatric populations (HARRISON 1986 ; HARRISON et al. 1987 ; Rand AND HARRISON 1989 ; HARRISON and BOGDANOWICZ 1997 ). MtDNA sequences from the COI-COII region distinguish four clades that correspond to northern and southern populations of the two species. However, relationships among the four clades are not resolved (WILLETT et al. 1997 ), and the two species are not clearly exclusive groups for mtDNA. On the basis of cytochrome b sequences, HUANG et al. 2000 suggested that G. pennsylvanicus and G. firmus are not even sister species, that G. ovisopis is sister to G. pennsylvanicus. However, they sampled only a single locality and one or two individuals for each species. What is clear is that the three egg-diapausing species have all diverged recently and that G. pennsylvanicus and G. firmus have subsequently come together to form an extensive hybrid zone.

Of the four nuclear gene loci that we examined [calmodulin (Cam), cytochrome c (Cyt-c), elongation factor 1 (Ef1), and phosphoglucose isomerase (Pgi)], only Pgi has been characterized previously in field crickets. Allozyme studies revealed a Pgi polymorphism shared among several species (HARRISON 1977 , HARRISON 1979 ; HARRISON and ARNOLD 1982 ). Not only do G. pennsylvanicus, G. firmus, and G. veletis share the same array of electromorphs, but in the broadly sympatric (but distantly related) G. pennsylvanicus and G. veletis, the frequency of the most common Pgi electromorph varies in parallel (HARRISON 1977 ). The patterns of shared electrophoretic alleles and parallel variation led to the suggestion that balancing selection may be acting at this enzyme locus. Comparison of DNA sequences in G. pennsylvanicus and G. veletis revealed that electrophoretic mobility classes are heterogeneous, i.e., that the same shifts in electrophoretic mobility have apparently occurred independently both within and between species (KATZ and HARRISON 1997 ). This is not consistent with a trans-specific polymorphism maintained by balancing selection. However, the DNA sequence data also suggest that Pgi is not evolving neutrally. Comparisons of patterns of variation at the Pgi locus with those observed for other nuclear genes may shed additional light on the forces acting to maintain the polymorphism.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Cricket samples:
Samples of G. firmus and G. pennsylvanicus were each obtained from three localities in the eastern United States. Individuals of G. pennsylvanicus were collected from Ithaca, New York; Sharon, Connecticut; and Harrisonburg, Virginia. G. firmus were collected from Guilford, Connecticut; Saybrook Point, Connecticut; and Seaside Park, New Jersey (Fig 1). Collecting sites were chosen on the basis of previous collecting experience, and crickets were identified on the basis of differences in body length, hind wing length, ovipositor length, and tegmina color (HARRISON 1986 ). G. ovisopis was obtained from a culture of crickets from Gainesville, Florida and G. veletis was collected from Sharon, Connecticut. Crickets were transported alive to the laboratory where they were frozen at -80° until DNA isolation.

View larger version (17K): In this window In a new window Download PPT slide   Figure 1. Collection localities for G. firmus and G. pennsylvanicus in eastern North America. Solid boxes indicate G. firmus samples; open boxes indicate G. pennsylvanicus samples.

PCR amplification, cloning, and sequencing:
Sequences of the Cam, Cyt-c, and Ef1 introns were obtained by PCR amplification, using primers located within adjacent exons but near the exon-intron boundaries (PALUMBI 1996 , p. 240). Primers were selected from the insect nuclear primer set assembled by B. Crespi and obtained from J. Hobbs at the University of British Columbia, Vancouver, Canada (Table 1). For Pgi, primers were designed on the basis of sequences in KATZ and HARRISON 1997 with one primer in an exon and the second within the intron itself to avoid sequencing through a tandem repeat. Sequences of Cyt-c, Cam, and Ef1 were obtained directly from PCR products. Reactions contained 1x Taq buffer (GIBCO/BRL, Gaithersburg, MD), 1.5 mM MgCl, 200 mM each dNTP, 0.5 µM each primer, 1 unit Taq polymerase, and 100 ng template DNA in a 50-µl volume. Cycling parameters were 35 cycles of 94° for 15 sec, 48°–54° for 15 sec, and 72° for 30–90 sec (60 sec/kb of product length). Templates were prepared for sequencing with QIAquick PCR product purification columns (QIAGEN, Valencia, CA). All sequencing reactions employed the ThermoSequenase ³³P-labeled terminator kit (USB) according to manufacturers' recommended protocols. Products were separated on manual sequencing gels and visualized with autoradiography. Both strands of each intron were sequenced.

Direct sequencing of PCR products revealed many nucleotide sites at which individuals were heterozygous. Although there are methods for estimating individual haplotypes from population data (e.g., CLARK 1990 ), it is not always possible to fully determine gametic phase without additional experimental data. Cam and Ef1 had low levels of heterozygosity and haplotypes could be determined unambiguously. Cyt-c had somewhat higher levels of heterozygosity so unknown haplotypes were estimated on the basis of the presence of observed homozygotes. This eliminated most ambiguity in haplotype assignments. High levels of heterozygosity made it particularly difficult to define haplotypes for Pgi. In this case, purified PCR products were cloned into the pGEM T-vector (Promega, Madison, WI) and transformed into Escherichia coli DH5 cells. Plasmids were purified via standard alkaline lysis methods and sequenced using universal primers. Typically, two to four clones from each individual cricket were sequenced, allowing gametic phase and allele identities to be determined. Of all the analyses performed, only the genealogical and recombination analyses would be affected by the lack of unambiguous haplotype assignment for Cyt-c and even in these cases, the consequences are expected to be slight given that most of the relevant sites are singleton variants.

Data analysis:
Sequences were aligned with Sequencher (Gene Codes, Ann Arbor, MI) and modified manually. The number of polymorphic sites and their distribution among species was determined. Within species, nucleotide variability was measured using , the proportion of polymorphic sites in a sample (WATTERSON 1975 ), and , the average number of nucleotide differences among sequences in a sample (NEI and LI 1979 ). Tajima's D (TAJIMA 1989 ) and Fu and Li's D* (FU and LI 1993 ) were calculated to test for departures from neutrality. Tajima's D is based on the expectation that under mutation-drift equilibrium, and should estimate the same parameter, 4N_eµ, so significant differences in these estimates may indicate departures from neutrality. Fu and Li's D* compares differences between estimates of 4N_eµ, on the basis of h_s (the number of singleton mutations) and S (the number of segregating sites).

The neutral theory also predicts that loci with high polymorphism within species will exhibit high divergence between species and loci with low polymorphism will exhibit low divergence. Thus if loci are behaving neutrally, the ratio of polymorphism to divergence should be similar for each locus. Polymorphism and divergence were compared between loci using the Hudson-Kreitman-Aguadé (HKA) test (HUDSON et al. 1987 ). Measures of variation and standard HKA tests were performed with the computer program DnaSP version 3.14 (ROZAS and ROZAS 1999 ). Multiple-locus HKA tests based on coalescent simulations were performed with the computer program HKA (provided by Jody Hey, Rutgers University). To test for spatial heterogeneity in the ratio of polymorphic sites to fixed differences, we employed the DNA Slider program (MCDONALD 1998 ).

Genealogical patterns were visualized via neighbor-joining trees constructed from Jukes-Cantor distances with the computer program MEGA2 (KUMAR et al. 2001 ) and by statistical parsimony networks (TEMPLETON et al. 1992 ) as implemented by the computer program TCS (CLEMENT et al. 2000 ). The regions sequenced for at least two of the loci appear to have been subject to recombination (KAPLAN and HUDSON 1985 , see Table 2) so portions of the introns may reflect different historical patterns. We employed a distance-based approach, which groups alleles on the basis of overall similarity; therefore, resulting trees cannot be assumed to represent the true genealogy, but they will reflect approximate patterns of relatedness. The statistical parsimony approach allows reticulations (if necessary) to reflect uncertainty generated by recombination and homoplasy (TEMPLETON et al. 1992 ). Potential population structure was investigated by comparing pairwise values (with Jukes-Cantor distances) among sample localities. Significance of these differences was assessed by permutation tests.

Analysis of nucleotide variation with respect to speciation has recently been applied in a hypothesis-testing framework where empirical frequencies of polymorphisms and fixed differences are compared to a null model (the isolation species model) using coalescent simulations (WAKELEY and HEY 1997 , WAKELEY and HEY 1998 ; WANG et al. 1997 ). The isolation species model assumes that variation is neutral and that there has been no gene flow beyond a point in the past at which the species diverged. The fit of the data from G. firmus and G. pennsylvanicus to the isolation model was tested with the program WH (provided by Jody Hey, Rutgers University).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Polymorphism and divergence:
The number of sequences obtained ranged from 34 for Pgi to 40 for Cam (see Fig 2 and Fig 3). Sequences of Cyt-c and Cam were only 298 and 309 bp in length; sequences of Ef1 and Pgi were longer, 616 and 749 bp, respectively. Levels of variation differ substantially among loci (Table 2). For G. firmus and G. pennsylvanicus, polymorphism as measured by and differs by more than an order of magnitude between the two extreme values for Ef1 and Pgi. At each locus, values of and are similar, and neither Tajima's D nor Fu and Li's D* differed significantly from the neutral expectation of zero. There were no sites within species with more than two nucleotide states so that S, number of segregating sites, is equal to h, the minimum number of mutations; consequently Fu and Li's D* is identical to their other statistic, F*. Although polymorphism varied substantially among loci, levels of variability are similar between the two species G. firmus and G. pennsylvanicus. Levels of polymorphism were not correlated with intron length. The lowest and highest levels of polymorphism were found in the two longest intron sequences (Ef1 and Pgi; see Table 2).

View larger version (52K): In this window In a new window Download PPT slide   Figure 2. Polymorphic nucleotide sites in Ef1, Cam, and Cyt-c introns. Sequence identification symbols are as follows: species are identified as f (firmus), p (pennsylvanicus), ovi (ovisopis), or vel (veletis) followed by a two-letter locality symbol [SA (Saybrook Point), GU (Guilford), SP (Seaside Park), IT (Ithaca), SH (Sharon), and HA (Harrisonburg)], an individual number, and letters a and b indicating different alleles within an individual. Numbering indicates position in whole sequence.

View larger version (55K): In this window In a new window Download PPT slide   Figure 3. Polymorphic nucleotide sites in the Pgi intron. Symbols are as in Fig 2.

The estimated minimum number of recombination events based on the presence of all four possible recombination products (Table 2) was proportional to the number of segregating sites. Therefore, the number of observed recombination events is related to the number of opportunities to detect recombination and provides no evidence that recombination rates differ among loci.

Interspecific divergence is summarized in Table 3. Divergence between G. firmus and G. pennsylvanicus was uniformly low, although average number of pairwise sequence differences is relatively high for Cyt-c and Pgi because of the large number of haplotypes. There were no fixed differences between these species at any locus and about one-half of all polymorphisms were shared between them (36 across all loci and 25 for Pgi alone), the other sites being polymorphic in one or the other species but not both. There were no fixed differences between G. firmus, G. pennsylvanicus, and G. ovisopis at Ef1, but variability at this locus was low. The similarity of these species was also reflected in the presence of three polymorphisms shared among all three species (one for Cyt-c and two for Pgi). In contrast, G. veletis has diverged substantially from the other species, consistent with previous phylogenetic analyses based on mtDNA (HARRISON and BOGDANOWICZ 1995 ; WILLETT et al. 1997 ; HUANG et al. 2000 ).

Average pairwise divergence () among G. firmus and G. pennsylvanicus sample localities was generally greater for interspecific comparisons than among intraspecific populations. However, in only one case was divergence significant (Pgi for G. firmus Saybrook Point vs. G. pennsylvanicus Harrisonburg) and in that case there were only two haplotypes for the Harrisonburg sample so the test result is suspect. This suggests little population subdivision within species.

Genealogical patterns:
Neighbor-joining trees for each locus are shown in Fig 4. Because recombination has affected Pgi and Cyt-c (Table 2), we cannot treat the trees as true genealogies in the sense of estimating a single history for each allele; rather these trees are approximations of genealogy based on overall similarity. For each locus, some alleles in one species are at least as similar to alleles in another species as they are to other alleles within the same species. Although haplotypes defined for Cyt-c are somewhat arbitrary, because we could not directly assess gametic phase when heterozygotes had more than one segregating site, alternative haplotype assignments do not change the fundamental characteristics of the gene genealogies. Thus, alleles in G. firmus and G. pennsylvanicus do not form exclusive groups for any locus. Furthermore, for three of the four loci (all but Cam) G. firmus and G. pennsylvanicus do not form an exclusive group with respect to G. ovisopis. This suggests that coalescence times for Ef1, Cyt-c, and Pgi predate the common ancestor of all three species. Coalescence for Cam appears to occur after divergence of G. ovisopis but before divergence of G. firmus and G. pennsylvanicus. Although introgression could also result in shared alleles, the geographic distribution of alleles is not consistent with this hypothesis (see below). Networks produced under statistical parsimony (not shown) were consistent with patterns from neighbor-joining trees and it was not possible to draw nonoverlapping clades corresponding to G. firmus and G. pennsylvanicus for any locus. The genealogies also indicate little geographic structure at the level of individual alleles. In numerous instances, alleles from the same population appeared in divergent portions of a gene network. In addition, alleles from geographically distant localities often cluster together.

View larger version (46K): In this window In a new window Download PPT slide   Figure 4. Neighbor-joining trees using Jukes-Cantor distances of individual alleles at each of four loci. To emphasize the distribution of alleles among species, G. pennsylvanicus alleles are indicated by dark background with white lettering; G. ovisopis alleles are indicated by light gray background.

A model of speciation:
We examined the fit of polymorphism data to that predicted by the isolation species model of WAKELEY and HEY 1997 . Observed counts of sites exhibiting polymorphism shared by two species, polymorphic in species 1 but monomorphic in species 2, polymorphic in species 2 but not in 1, and fixed differences were used to estimate the population parameter for each species (₁ and ₂), the population parameter for the ancestral species (_A), and the time (T) in 2N_e generations to the common ancestor. Parameters were as follows: ₁ = 9.166, ₂ = 7.937, _A = 18.100 (95% C.I. = 6.84–48.24), and T = 0.050 (95% C.I. = 0.0–0.487) (where G. firmus is population 1 and G. pennsylvanicus is population 2). These results suggest that the ancestral population was approximately double the size of either descendant population. This is consistent with a speciation scenario in which a large ancestral population becomes subdivided, as would be expected by a vicariance event, and the resulting daughter species each fill part of the ancestral species range. Parameter estimates were compared to expectations derived from 1000 coalescent simulations. The frequency of simulations with higher values of the WH test statistic than the observed was 0.51, so we cannot reject the model. Although G. firmus and G. pennsylvanicus hybridize where they come into contact, there is little evidence of introgression outside of the hybrid zone for many markers. A lack of ongoing gene flow would be consistent with the isolation speciation model. However, the estimate of the time to the common ancestor is T = 0.05 (in units of 2N_e). We do not have good estimates for N_e in crickets, but if N_e is 10⁴–10⁵, the T would be only 10³–10⁴ generations. Thus, if effective gene flow is absent, the high levels of shared polymorphism would indicate a very recent origin of these two cricket species. Some shared polymorphism and low sequence divergence suggest that the common ancestor of G. firmus, G. pennsylvanicus, and G. ovisopis also existed fairly recently.

Tests of natural selection:
HKA tests compare levels of polymorphism within a species and divergence between that species and a close relative. We examined polymorphism to divergence ratios for all six possible locus pairs for G. firmus compared to G. pennsylvanicus, G. ovisopis, and G. veletis, and G. pennsylvanicus compared to G. ovisopis and G. veletis. For comparisons between G. firmus and G. pennsylvanicus, polymorphism/divergence values were similar across loci. However, in comparisons of each of these species with G. ovisopis and the more distantly related G. veletis, polymorphism/divergence values were generally low for Cam and high for Pgi. While the Cam-Pgi comparisons yielded HKA test P values greater than but close to 0.05, none of the tests indicated statistical departure from neutrality in comparisons between individual pairs of loci or when all loci were combined in a single test. Because G. firmus and G. pennsylvanicus exhibited no fixed differences and many shared haplotypes, we combined these two species into a single taxon for additional HKA tests vs. G. veletis. In this case, the Pgi vs. Cam comparison did indicate departure from neutrality (P = 0.040). G. firmus and G. pennsylvanicus differ little with respect to allele frequencies yet they are clearly not a single equilibrium population, so the statistical validity of combining them for this test is not clear. However, it does highlight the fact that the relationship between polymorphism and divergence differs substantially between Pgi and Cam.

The ratio of polymorphism to divergence for the Pgi intron is also significantly higher than the corresponding ratio for mitochondrial coding sequences (1621 bp of COI and COI; WILLETT et al. 1997 ). HKA tests (accounting for reduced N_e in mtDNA) comparing G. firmus and G. pennsylvanicus with G. veletis were both significant (P = 0.016 and P = 0.020, respectively). On the basis of comparisons with mtDNA sequences, Pgi exons have previously been shown to exhibit evidence of balancing selection (KATZ and HARRISON 1997 ). Because it appears that the intron has been subject to recombination, we might expect a portion proximal to the selected site(s) to show a more extreme relationship between polymorphism and divergence. Hence, there may be heterogeneity in the ratio of polymorphic sites to fixed differences along the sequence (MCDONALD 1998 ). However, we found no significant heterogeneity in the Pgi intron (Kolmogorov-Smirnov test statistic 0.04, P = 0.22). Thus we could not identify specific variants in the intron that might be linked to presumptive selected sites in the coding region.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Gene genealogies and diversification within the genus Gryllus:
Speciation involves the splitting of a single interbreeding population into two differentiated populations between which there are intrinsic barriers to gene exchange (HARRISON 1998 ). Observed patterns of variation in the daughter populations reflect the operation of evolutionary forces over recent history and potentially provide insights into how speciation occurred. The amounts of polymorphism and the structure of gene genealogies will depend on the timing of speciation events, historical population sizes, the impact of natural selection and recombination, and the extent of hybridization and introgression. For the closely related field crickets examined in this study, levels of intron sequence variation differ substantially among loci, much of the polymorphism appears to predate divergence of two or even three of the species, and many shared alleles (or groups of similar alleles) show little or no geographical structure.

That ancestral polymorphism has persisted through the divergence of G. firmus, G. pennsylvanicus, and G. ovisopis is consistent with the notion that speciation has occurred rapidly, that N_e has remained large, and that a small number of genes are responsible for divergence and reproductive isolation in these crickets. If alternate fixation of alleles at relatively few loci is sufficient for reproductive isolation to occur, then at most loci polymorphisms may persist for substantial lengths of time after interbreeding ceases (HILTON et al. 1994 ; TING et al. 2000 ; WU 2001 ). Even substantial gene flow might not affect species integrity so long as there is no introgression of "speciation genes." Only after substantial time has passed from the speciation event [typically >2N_e generations (MADDISON 1997 )] will taxa approach reciprocal monophyly for variation at most loci (TAJIMA 1983 ; NEIGEL and AVISE 1986 ; HARRISON 1991 ; HEY 1994 ).

An alternative explanation for shared haplotypes is that gene flow subsequent to initial divergence has erased patterns of exclusivity for some loci. If ongoing interspecific gene flow were occurring we would expect greater similarity among alleles from geographically close populations. That is, under a stepping-stone population model (which is appropriate for taxa of low vagility such as crickets), local demes should be more similar to those with which they are directly exchanging genes than to more distant demes with which they would have little or no contact. Thus we would expect, for example, that northern G. firmus samples would be more similar to northern G. pennsylvanicus samples than to more distant G. firmus due to isolation by distance. Consequently, in this case with no correlation between genetic distance and geographic proximity, polymorphism maintained from the common ancestor is the most plausible explanation for the observed deep coalescence of nuclear loci. Although our results suggest that divergence (cessation of gene flow) between the species occurred recently, we cannot readily determine whether this was the initial divergence event or the cessation of gene flow between populations that had diverged previously.

The estimated time since divergence between G. firmus and G. pennsylvanicus (0.05 x 2N_e generations) suggests that branch lengths are substantially shorter than those that would be necessary for drift to eliminate shared variation (i.e., 2N_e generations). Furthermore, PALUMBI et al. 2001 have suggested that there is a relatively high probability of exclusivity for nuclear genes only if the length of the branch leading to an exclusive mtDNA clade is more than three times the diversity within the clade. The ratio of branch length to diversity for mitochondrial sequences in G. firmus and G. pennsylvanicus ranges from 0.5 to 2 (mtDNA data from WILLETT et al. 1997 ) and thus does not satisfy the "three times rule." Therefore, polyphyly for nuclear loci is not unexpected, given the mtDNA tree.

Of course, genetic differences must underlie observed morphological, behavioral, ecological, and reproductive differences between the closely related cricket species. Previous surveys of genetic variation across the hybrid zone between G. firmus and G. pennsylvanicus have revealed evidence of such differentiation, but the overall impression from multiple surveys is that these two species have diverged very little. Allozyme surveys (25 loci) of species in the genus Gryllus revealed no fixed differences between G. firmus and G. pennsylvanicus, with a Nei's D (NEI 1972 ) of only 0.02–0.03 (HARRISON 1979 , HARRISON 1986 ). Three loci (Esterase, Peptidase-1, and Peptidase-3) showed significant allele frequency differences. The hybridizing cricket species can be distinguished on the basis of mtDNA haplotype (HARRISON et al. 1987 ; WILLETT et al. 1997 ), but as discussed above, neither of the species appears to be an exclusive group, although regional groups of populations within species are exclusive for mtDNA. Finally, previous efforts to characterize nuclear gene markers involved RFLP analysis of anonymous nuclear regions. Of 52 pUC clones used as hybridization probes to digests of cricket genomic DNA, many appeared to exhibit evidence of shared polymorphisms (for presence/absence of restriction sites) and only three uncovered fixed or nearly fixed differences (HARRISON and BOGDANOWICZ 1997 ). Taken together, these results support the impression derived from the intron sequence data that G. firmus and G. pennsylvanicus have diverged very little and that for much of the genome, ancestral polymorphisms are still segregating in the cricket populations. Thus, evidence from Gryllus is consistent with the theoretical expectation that a large portion of the genome frequently remains common to closely related species well after speciation has occurred.

Patterns of polymorphism and divergence in several groups of Drosophila are very similar to those seen in the field crickets. Many nuclear loci appear to be paraphyletic or polyphyletic with respect to one or more recognized species in the Drosophila melanogaster species group (HEY and KLIMAN 1994 ; KLIMAN et al. 2000 ), D. pseudoobscura and close relatives (WANG et al. 1997 ; MACHADO et al. 2002 ), and the D. virilis species group (HILTON and HEY 1996 ). However, in contrast to patterns seen for other loci, variation at a putative speciation gene (OdsH) has been shown to resolve D. simulans and its close relatives D. seychellia and D. mauritiana as monophyletic species (TING et al. 2000 ). The expectation is that such regions are also present in the Gryllus genome and mark sites of recent lineage-specific selective sweeps and/or loci that contribute to reproductive isolation.

Patterns of variation and the role of natural selection:
Levels of polymorphism varied 10-fold among the four loci we sampled. However, on average, amounts of variation in Gryllus appear to be comparable to those found in Drosophila. MORIYAMA and POWELL 1996 summarized available information from D. melanogaster, D. simulans, and D. pseudoobscura. Noncoding autosomal regions from D. melanogaster exhibited values between 0.0005 and 0.0265 with a mean of 0.0117. Although fewer noncoding regions were available, diversity was somewhat higher in D. simulans (mean = 0.0189, range = 0.0058–0.0338) and D. pseudoobscura (mean = 0.0170, range = 0.0100–0.0209). Two of the four cricket intron sequences (Pgi and Cyt-c) fall within observed ranges of diversity values for Drosophila, while the other two cricket loci are less variable.

Population sizes and rates of diversification are characteristics of lineages and therefore cannot explain differences in patterns of variation among loci within a lineage. To explain locus- or genomic region-specific properties, we must invoke differences in mutation rates among gene regions or differences in the history of selection. The neutral theory predicts that if population size is constant, the number of alleles present at any given time will depend on the mutation rate, but that the time to coalescence will not. Variation among loci in (4N_eµ) within a species may indicate variation in the mutation rate if population size is assumed to be similar for all autosomal loci. Population subdivision can elevate effective population size (WRIGHT 1943 ) and can stretch out coalescence times (WAKELEY 2000 ). This would violate the assumption of panmixia made by the isolation species model and could be an additional explanation of the high degree of shared polymorphism. However, we found no evidence of population subdivision for these loci.

Natural selection also affects levels of polymorphism. Balancing selection can maintain ancestral polymorphism, causing deep coalescence. In contrast, directional selection will cause reduced variation through selective sweeps. If mutation rates are equal for different genomic regions, the depth of coalescence will vary depending on the type and intensity of selection, the rate of recombination, and the proximity of the site(s) under selection to the gene region used in genealogical analysis.

One way to distinguish selection from mutation rate is with the HKA test. Neutral theory predicts that polymorphism will be proportional to divergence regardless of mutation rate. Hence significant differences in such ratios are indicative of selection. On the basis of early allozyme studies (HARRISON 1977 ) and subsequent characterization of coding sequences for Pgi in G. veletis and G. pennsylvanicus (KATZ and HARRISON 1997 ), we started with the expectation that the Pgi locus might provide evidence for balancing selection. For most comparisons between intron sequences, Pgi has the highest ratio of polymorphism to divergence, which can be explained by a history of balancing selection. Comparisons with mtDNA also indicate a significantly greater polymorphism/divergence ratio, again consistent with the balancing selection scenario. Presumably Pgi and Cam represent ends of a continuum with respect to levels of variability, and these loci may well have been subject to opposing selective forces.

Our failure to find compelling evidence for selection on Pgi is perhaps not entirely surprising. In other organisms, balancing selection has been difficult to demonstrate (KREITMAN and AKASHI 1995 ; HEY 1999 ). Departures from neutrality consistent with balancing selection are evident from patterns of variation for several Drosophila genes (e.g., AGUADE et al. 1992 ; KIRBY and STEPHAN 1995 ; ODGERS et al. 1995 ; EANES et al. 1996 ; WAYNE et al. 1996 ), but the power of available statistical tests for selection is generally weak (SIMONSEN et al. 1995 ). Moreover, alternative tests (e.g., HKA, Tajima's D, and McDonald-Kreitman) often yield conflicting results for a particular locus (see MORIYAMA and POWELL 1996 ). So while it is likely that many loci have been subject to balancing selection, Adh remains one of the few clear examples (HUDSON et al. 1987 ).

Pgi has been examined for evidence of selection in Drosophila. HKA tests involving the noncoding region of Pgi in Drosophila yield significant differences from a few other loci and the difference from Adh is highly significant (<0.005; MORIYAMA and POWELL 1996 ). However, in Drosophila the differences are due to a lack of polymorphism at Pgi, which is one of the least variable loci in both D. melanogaster ( = 0.0018) and D. simulans ( = 0.0058). Given the extreme functional conservation of the glycolytic pathway, Pgi almost certainly performs the same function in both taxa. Thus, the striking difference in levels of polymorphism between crickets and flies may reflect different selective regimes.

An alternative to balancing selection is that the elevated polymorphism level at Pgi is related to a higher recombination rate and the lower polymorphism observed for other loci could be due to lower recombination rates in combination with selective sweeps. Under this scenario, polymorphism at Pgi could be neutral but a greater recombination rate allows it to escape the effects of selective sweeps. We cannot discount this hypothesis entirely, although in our view the bulk of the evidence points to balancing selection.

Conclusions:
Given sufficient time between divergence events, recognized species or populations will become monophyletic or exclusive groups. However, when populations have diverged recently or when divergence events have occurred in rapid succession, patterns of exclusivity are not expected, particularly for nuclear loci. While often considered to be an annoyance from a purely phylogenetic perspective, this phenomenon allows for important inferences about very recent evolutionary events. In this case, genealogical analysis suggests that three species of field crickets diverged recently, that most of these species' genomes have diverged little from their common ancestor, and that relatively few genes have been involved directly in the speciation process. In addition, extensive polymorphism and deep coalescence at one locus (Pgi) may reflect maintenance by balancing selection, while low polymorphism characteristic of a selective sweep is evident at a second locus (Cam), in spite of the fact that the limited variation appears to predate the G. firmus-G. pennsylvanicus speciation event. Thus, nuclear gene genealogies may have limited utility in resolving closely related taxa, yet they can provide important insights into the historical, demographic, and selective forces that shape speciation and ultimately give rise to broader phylogenetic patterns.

	FOOTNOTES

Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AY181260–AY181421.

	ACKNOWLEDGMENTS

We thank S. Bogdanowicz for technical assistance, members of the Harrison lab for helpful discussions, and J. Feder and an anonymous referee for thoughtful reviews of an earlier version of the manuscript. This work was supported by a National Science Foundation (NSF)/Alfred P. Sloan Foundation postdoctoral fellowship in molecular evolution to R.E.B. and by NSF population biology grant DEB-981530 to R.G.H.

Manuscript received June 5, 2002; Accepted for publication January 8, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AGUADÉ, M., N. MIYASHITA, and C. H. LANGLEY, 1992  Polymorphism and divergence in the Mst26A male accessory gland gene region in Drosophila.. Genetics 132:755-770.[Abstract]

ALEXANDER, R. D., 1957  The taxomony of the field crickets of the eastern United States (Orthoptera: Gryllidae: Acheta). Ann. Entomol. Soc. Am. 50:584-602.

ALEXANDER, R. D., 1968  Life cycle origins, speciation, and related phenomena in crickets. Q. Rev. Biol. 43:1-41.[Medline]

AVISE, J. C., 2000 Phylogeography: The History and Formation of Species. Harvard University Press, Cambridge, MA.

BIRKY, C. W., 1991 Evolution and population genetics of organelle genes: mechanisms and models, pp. 112–134 in Evolution at the Molecular Level, edited by R. K. SELANDER, A. G. CLARK and T. S. WHITTAM. Sinauer, Sunderland, MA.

CLARK, A. G., 1990  Inference of haplotypes from PCR-amplified samples of diploid populations. Mol. Biol. Evol. 7:111-122.[Abstract]

CLEMENT, M., D. POSADA, and K. A. CRANDALL, 2000  TCS: a computer program to estimate gene genealogies. Mol. Ecol. 9:1657-1660.[Medline]

DUDA, T. F. and S. R. PALUMBI, 1999  Developmental shifts and species selection in gastropods. Proc. Natl. Acad. Sci. USA 96:10272-10277.[Abstract/Free Full Text]

EANES, W. F., M. KIRCHNER, H. YOON, C. H. DIERMANN, and I. N. WANG et al., 1996  Historical selection, amino acid polymorphism and lineage-specific divergence at the G6pd locus in Drosophila melanogaster and D. simulans.. Genetics 144:1027-1041.[Abstract]

FU, Y.-X. and W.-H. LI, 1993  Maximum likelihood estimation of population parameters. Genetics 134:1261-1270.[Abstract]

HARE, M. P., 2001  Prospects for nuclear gene phylogeography. Trends Ecol. Evol. 16:700-706.

HARRISON, R. G., 1977  Parallel variation at enzyme locus in sibling species of field crickets. Nature 266:168-170.[Medline]

HARRISON, R. G., 1979  Speciation in North American field crickets: evidence from electrophoretic comparisons. Evolution 33:1009-1023.

HARRISON, R. G., 1983  Barriers to gene exchange between closely related cricket species. I. Laboratory hybridization studies. Evolution 37:245-251.

HARRISON, R. G., 1986  Pattern and process in a narrow hybrid zone. Heredity 56:337-349.

HARRISON, R. G., 1991  Molecular changes at speciation. Annu. Rev. Ecol. Syst. 22:281-308.

HARRISON, R. G., 1998 Linking evolutionary pattern and process: the relevance of species concepts for the study of speciation, pp. 19–31 in Endless Forms: Species and Speciation, edited by D. J. HOWARD and S. H. BERLOCHER. Oxford Unversity Press, New York.

HARRISON, R. G. and J. ARNOLD, 1982  A narrow hybrid zone between closely related cricket species. Evolution 36:535-552.

HARRISON, R. G. and S. M. BOGDANOWICZ, 1995  Mitochondrial DNA phylogeny of North American field crickets: perspectives on the evolution of life cycles, songs, and habitat association. J. Evol. Biol. 8:209-232.

HARRISON, R. G. and S. M. BOGDANOWICZ, 1997  Patterns of variation and linkage disequilibrium in a field cricket hybrid zone. Evolution 51:493-505.

HARRISON, R. G., D. M. RAND, and W. C. WHEELER, 1987  Mitochondrial DNA variation in field crickets across a narrow hybrid zone. Mol. Biol. Evol. 4:144-158.

HEY, J., 1994 Bridging phylogenetics and population genetics with gene tree models, pp. 435–449 in Molecular Ecology and Evolution: Approaches and Applications, edited by B. SCHIERWATER, B. STREIT, G. WAGNER and R. DESALLE. Birkhauser, Boston.

HEY, J., 1999  The neutralist, the fly and the selectionist. Trends Ecol. Evol. 14:35-38.[Medline]

HEY, J., and R. M. KLIMAN, 1994 Genealogical portraits of speciation in the Drosophila melanogaster species complex, pp. 208–216 in Non-neutral Evolution, Theories and Molecular Data, edited by B. GOLDING. Chapman & Hall, New York.

HILTON, H. and J. HEY, 1996  DNA sequence variation at the period locus reveals the history of species and speciation events in the Drosophila virilis group. Genetics 144:1015-1025.[Abstract]

HILTON, H., R. M. KLIMAN, and J. HEY, 1994  Using hitchhiking genes to study adaptation and divergence during speciation within the Drosophila melanogaster complex. Evolution 48:1900-1913.

HOELZER, G. A., 1997  Inferring phylogenies from mtDNA variation: mitochondrial-gene trees versus nuclear-gene trees revisited. Evolution 51:622-626.

HUANG, Y., G. ORTI, M. SUTHERLIN, A. DUHACHEK, and A. ZERA, 2000  Phylogenetic relationships of North American field crickets inferred from mitochondrial DNA data. Mol. Phylogenet. Evol. 17:48-57.[Medline]

HUDSON, R. R., 1990  Gene genealogies and the coalescent process. Oxf. Surv. Evol. Biol. 7:1-44.

HUDSON, R. R., 1992  Gene trees, species trees and the segregation of ancestral alleles. Genetics 131:509-512.[Medline]

HUDSON, R. R., M. KREITMAN, and M. AGUADÉ, 1987  A test of neutral molecular evolution based on nucleotide data. Genetics 116:153-159.[Abstract/Free Full Text]

KAPLAN, N., R. R. HUDSON, and C. H. LANGLEY, 1989  The "hitchhiking effect" revisited. Genetics 123:887-899.[Abstract/Free Full Text]

KAPLAN, N. L. and R. R. HUDSON, 1985  The use of sample genealogies for studying a selectively neutral m-loci model with recombination. Theor. Popul. Biol. 28:382-396.[Medline]

KATZ, L. A. and R. G. HARRISON, 1997  Balancing selection on electrophoretic variation of phosphoglucose isomerase in two species of field cricket: Gryllus veletis and G. pennsylvanicus.. Genetics 147:609-621.[Abstract]

KIRBY, D. A. and W. STEPHAN, 1995  Haplotype test reveals departure from neutrality in a segment of the white gene of Drosophila melanogaster.. Genetics 141:1483-1490.[Abstract]

KLIMAN, R. M., P. ANDOLFATTO, J. A. COYNE, F. DEPAULIS, and M. KREITMAN et al., 2000  The population genetic of the origin and divergence of the Drosophila simulans complex species. Genetics 156:1913-1931.[Abstract/Free Full Text]

KREITMAN, M. and H. AKASHI, 1995  Molecular evidence for natural selection. Annu. Rev. Ecol. Syst. 26:403-422.

KUMAR, S., K. TAMURA, I. B. JAKOBSEN, and M. NEI, 2001  Molecular evolutionary genetics analysis software. Bioinformatics 17:1244-1245.[Abstract/Free Full Text]

MACHADO, C. A., R. M. KLIMAN, J. A. MARKERT, and J. HEY, 2002  Inferring the history of speciation from multilocus DNA sequence data: the case of Drosophila pseudoobscura and close relatives. Mol. Biol. Evol. 19:472-488.[Abstract/Free Full Text]

MADDISON, W. P., 1997  Gene trees in species trees. Syst. Biol. 46:523-536.

MCDONALD, J. H., 1998  Improved tests for heterogeneity across a region of DNA sequence in the ratio of polymorphism to divergence. Mol. Biol. Evol. 15:377-384.[Abstract]

MOORE, W. S., 1995  Inferring phylogenies from mtDNA variation: mitochondrial-gene trees versus nucelar-gene trees. Evolution 49:718-726.

MORIYAMA, E. N. and J. R. POWELL, 1996  Intraspecific nuclear DNA variation in Drosophila. Mol. Biol. Evol. 13:261-277.[Abstract]

NEI, M., 1972  Genetic distance between populations. Am. Nat. 106:283-292.

NEI, M. and W.-H. LI, 1979  Mathematical model for studying genetic variation in terms of restriction endonucleases. Proc. Natl. Acad. Sci. USA 76:5269-5273.[Abstract/Free Full Text]

NEIGEL, J. E., and J. C. AVISE, 1986 Phylogenetic relationships of mitochondrial DNA under various demographic models of speciation, pp. 515–534 in Evolutionary Processes and Theory, edited by S. KARLIN and E. NEVO. Academic Press, New York.

NICHOLS, R., 2001  Gene trees and species trees are not the same. Trends Ecol. Evol. 16:358-364.[Medline]

ODGERS, W. A., M. J. HEALY, and J. G. OAKESHOTT, 1995  Nucleotide polymorphism in the 5' promoter region of esterase 6 in Drosophila melanogaster and its relationship to enzyme activity variation. Genetics 141:215-222.[Abstract]

PALUMBI, S. R., 1996 Nucleic acids II: the polymerase chain reaction, pp. 205–247 in Molecular Systematics, edited by D. M. HILLIS, C. MORITZ and B. K. MABLE. Sinauer, Sunderland, MA.

PALUMBI, S. R., F. CIPRIANO, and M. P. HARE, 2001  Predicting nuclear gene coalescence from mitochondrial data: the three-times rule. Evolution 55:859-868.[Medline]

PAMILO, P. and M. NEI, 1988  Relationships between gene trees and species trees. Mol. Biol. Evol. 5:568-583.[Abstract]

RAND, D. M. and R. G. HARRISON, 1989  Ecological genetics of a mosaic hybrid zone: mitochondrial, nuclear, and reproductive differentiation of crickets by soil type. Evolution 43:432-449.

REHN, J. A. G. and M. HEBARD, 1915  The genus Gryllus (Orthoptera) as found in America. Proc. Acad. Nat. Sci. Phila. 67:292-322.

ROZAS, J. and R. ROZAS, 1999  DnaSP version 3: an integrated program for molecular population genetics and molecular evolution analysis. Bioinformatics 15:174-175.[Abstract/Free Full Text]

SIMONSEN, K. L., G. A. CHURCHILL, and C. F. AQUADRO, 1995  Properties of statistical tests of neutrality for DNA polymorphism data. Genetics 141:413-429.[Abstract]

TAJIMA, F., 1983  Evolutionary relationship of DNA sequences in finite populations. Genetics 105:437-460.[Abstract/Free Full Text]

TAJIMA, F., 1989  Statistical method for testing the neural mutation hypothesis by DNA polymorphism. Genetics 123:585-595.[Abstract/Free Full Text]

TAKAHATA, N., 1989  Gene genealogy in three related populations: consistency probability between gene and population trees. Genetics 122:957-966.[Abstract/Free Full Text]

TEMPLETON, A. R., 1998  Nested clade analyses of phylogeographic data: testing hypotheses about gene flow and population history. Mol. Ecol. 7:381-397.[Medline]

TEMPLETON, A. R., K. A. CRANDAL, and C. F. SING, 1992  A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping and DNA sequence data. III. Cladogram estimation. Genetics 132:619-633.[Abstract]

TING, C.-T., S.-C. TSAUR, and C.-I. WU, 2000  The phylogeny of closely related species as revealed by the genealogy of a speciation gene, Odysseus.. Proc. Natl. Acad. Sci. USA 97:5313-5316.[Abstract/Free Full Text]

VILLABLANCA, F. X., G. K. RODERICK, and S. R. PALUMBI, 1998  Invasion genetics of the Mediterranean fruit fly: variation in multiple nuclear introns. Mol. Ecol. 7:547-560.[Medline]

WAKELEY, J., 2000  The effects of population subdivision on the genetic divergence of populations and species. Evolution 54:1092-1101.[Medline]

WAKELEY, J. and J. HEY, 1997  Estimating ancestral population sizes. Genetics 145:847-855.[Abstract]

WAKELEY, J., and J. HEY, 1998 Testing speciation models with DNA sequence data, pp. 157–175 in Molecular Approaches to Ecology and Evolution, edited by R. DESALLE and B. SCHIERWATER. Birkhauser, Boston.

WANG, R.-L., J. WAKELEY, and J. HEY, 1997  Gene flow and natural selection in the origin of Drosophila pseudoobscura and close relatives. Genetics 147:1091-1106.[Abstract]

WATTERSON, G. A., 1975  On the number of segregating sites in genetical models without recombination. Theor. Popul. Biol. 10:256-276.

WAYNE, M. L., D. CONTAMINE, and M. KREITMAN, 1996  Molecular population genetics of ref(2)P, a locus which confers viral resistance in Drosophila. Mol. Biol. Evol. 13:191-199.[Abstract]

WILLETT, C. S., M. J. FORD, and R. G. HARRISON, 1997  Inferences about the origin of a field cricket hybrid zone from a mitochondrial DNA phylogeny. Heredity 79:484-494.

WRIGHT, S., 1943  Isolation by distance. Genetics 28:114-138.[Free Full Text]

WU, C.-I., 1991  Inferences of species phylogeny in relation to segregation of ancient polymorphisms. Genetics 127:429-435.[Abstract]

WU, C.-I., 2001  The genic view of the process of speciation. J. Evol. Biol. 14:851-865.

This article has been cited by other articles:

T. Stadler, U. Arunyawat, and W. Stephan Population Genetics of Speciation in Two Closely R