Character discriminatory power, character-set congruence, and the classification of individuals from hybrid zones: an example using stone crabs (Menippe).
Evolutionary biologists often use a combination of distinct, presumably genetically independent traits to classify individuals from hybrid zones (e.g., allozymes, chromosomes, extranuclear DNA, morphology, behavior). Typically, they analyze character sets separately and use various methodologies to combine the results from each analysis to classify each individual in their samples (e.g., Dowling et al. 1989; Rand and Harrison 1989). This approach is similar to the method of taxonomic congruence employed by some systematic biologists for phylogenetic analyses (e.g., Nelson 1979; Hillis 1987; Swofford 1991); in this method, the integrity of different character sets is retained in the data analysis and the congruence (consensus) among sets is subsequently examined. An alternative approach to phylogenetic analysis advocated by other systematists is that of character congruence (total evidence), whereby data from various character sets are combined before statistical analysis and the designation of phylogenetic relationships is founded upon the analysis of the combined data (e.g., Miyamoto 1985; Kluge 1989; Barrett et al. 1991). The objective of phylogenetics is similar to that of the classification of individuals for hybrid-zone studies; the best ("true") classification of operational taxonomic units (terminal taxa for phylogenetic analysis and specific individuals for the analysis of hybrid zones) is sought using the maximum number of informative characters available. Thus, a total-evidence approach should also be relevant to determining the discriminatory power of characters used to distinguish hybridizing species and to classify individuals from their hybrid zones.
We investigated the effects of using the principles of taxonomic congruence versus those of total evidence to assign individuals from a hybrid zone to genotypic classes using stone crabs Menippe adina and M. mercenaria. These two species hybridize extensively in the marine waters off northwest Florida (Bert 1986; Williams and Felder 1986; Bert and Harrison 1988). A broad range of hybrid forms, as well as both parental species, occupy the hybrid zone (Bert and Harrison 1988). We evaluated the utility of two classes of morphological traits (coloration and mensural) and two classes of molecular traits - allozymes and mitochondrial DNA (mtDNA) restriction-fragment-length polymorphisms (RFLPs) - considered both independently and in combination, for distinguishing parental species individuals and for classifying individuals from the hybrid zone. Our goal was to identify the combination of molecular and morphological characters that provided the best discrimination between the parental species and that identified a broad range of hybrid forms from the hybrid zone.
The characters we chose for the three traits coded by nuclear genes were those that had previously been targeted as being taxonomically informative. Both Bert (1986) and Williams and Felder (1986) described diagnostic differences in body coloration for M. adina and M. mercenaria; Bert and Harrison (1988) quantified the range of coloration seen in the two species and in hybrid forms. Williams and Felder (1986) investigated morphometric variation between the species and found relationships between the carapace and anterolateral teeth (protrusions along the anterolateral edge of the carapace) that they considered diagnostic. Bert and Harrison (1988) used three allozyme loci to distinguish between species and identify hybrids; two loci were nearly homozygous for a different allele in each species, and a third was nearly homozygous for one allele in M. adina and polymorphic for that allele in M. mercenaria. We evaluated the utility of mtDNA because it is an extranuclear genetic marker with a different mode of inheritance and other unique properties (Moritz et al. 1987) useful for species discrimination. Schneider-Broussard (1993) characterized a 580 bp region of the large-subunit mitochondrial ribosomal-RNA gene from several crabs collected from within areas of allopatry of the parental species, the northwest Florida hybrid zone, and the Atlantic region of hybridization (Bert and Harrison 1988), but Schneider-Broussard's sample sizes were small (three individuals per location), the geographic coverage was restricted, and the survey was limited to a single conserved gene. We utilized mtDNA RFLPs, which to some degree represent the entire mtDNA genome, and we assayed a large number of individuals from several locations.
Here we use the multiple character sets, both in combination (a total-evidence approach) and independently (a taxonomic-congruence approach) to statistically examine the level of discriminatory power attained for the two stone crab species, and we assess the efficacy of using the combined versus the independent character sets to classify stone crabs from the hybrid zone. Simultaneously, we evaluate the discriminatory power of each character within the framework of the character set(s) being analyzed. We also assess the pattern and degree of concordance among the four character sets. Finally, we discuss the general utility of a taxonomic-congruence approach versus a total-evidence approach in studies involving hybrid zones.
MATERIALS AND METHODS
Field and Laboratory Procedures
Collections of subadult and adult stone crabs ([greater than or equal to]50 mm carapace width [CW]) were made within the ranges published for Menippe adina (four sites) and M. mercenaria (three sites), and from the northwest Florida hybrid zone (two sites) [ILLUSTRATION FOR FIGURE 1 OMITTED]. All samples were obtained by trapping with standard commercial stone crab or blue crab traps, except the Turkey Point sample (TKP, [ILLUSTRATION FOR FIGURE 1 OMITTED]), which was collected by hand. For each individual, carapace color, carapace markings, claw color, claw markings, leg color, and color and intensity of leg banding were numerically scored following Bert and Harrison (1988). Also recorded were sex, type of each claw (original crusher, original pincer, regenerated) and, to the nearest 0.1 mm, CW, carapace length (CL), right and left propodus lengths (RPL and LPL), and lengths of right anterolateral teeth 3 and 4 (ALT3 and ALT4). The relationships of CL to ALT3, CL to ALT4, and ALT3 to ALT4 are considered to be diagnostic for the two species (Williams and Felder 1986). The crabs were dissected alive to obtain samples of hepatopancreas, gill, gonad, somatic muscle, and heart for protein electrophoresis and mtDNA RFLP analysis; the tissue samples were frozen immediately in liquid nitrogen and stored at -80 [degrees] C.
Protein electrophoresis of the three diagnostic presumptive genetic loci (an alkaline phosphotase [ALP-2], superoxide dismutase [SOD], and an isocitrate dehydrogenase [IDH-2]) was conducted on all individuals following the procedures of Bert (1986). The mtDNA RFLP analysis was conducted on 178 individuals collected from the nine sampling sites. For this analysis, total cellular DNA was obtained from approximately 100-200 mg of gonad, gill, or somatic muscle tissue of each individual. The tissues were ground to powder in liquid nitrogen and were then incubated at room temperature for 15 min in 10 ml of a solution (pH 9.0) containing 2 M sucrose, 50 mM EDTA, and 100 mM Tris, and to which 25 [[micro]liter] of 20% SDS was added. To precipitate the SDS and cellular debris, we added 0.15 ml of 8 M potassium acetate, cooled the solution for 30 min at - 20 [degrees] C, and centrifuged for 10 min. The remaining proteins were removed by general phenol-chloroform extraction (Maniatis et al. 1982). The resultant supernatant was mixed with 100% isopropanol (1:1 vol:vol) and stored at -20 [degrees] C overnight to precipitate the DNA. The DNA was pelleted by centrifugation, air-dried, and redissolved in 100 [[micro]liter] of 10 mM Tris, 1mM EDTA (pH 7.5). Restriction digests were performed according to the manufacturer's directions (New England Biolabs or Boeringer Mannheim) using 2-5 [[micro]liter] of the total DNA per digest. We initially assayed a representative number of individuals from each location, using 19 restriction enzymes: BclI, BstBI, BstNI, HaeIII, HindIII, MspI, BgIII, EcoRI, PvuII, ApaI, BamHI, ClaI, EcoRV, PstI, SacI, SacII, SalI, XbaI, and XhoI. Of these, the first nine listed cut the mtDNA more than once, and the first six were polymorphic. These six were used to assay the 178 individuals.
The DNA fragments resulting from the digests were electrophoretically separated on 0.8% agarose gels using TAE buffer and were Southern-transferred to Zetabind[R] filters (Maniatis et al. 1982). The DNA fragments were bound to the filters by baking (30 min, 120 [degrees] C). We hybridized the mtDNA to a CsCl-purified, 32p-labeled, Menippe-mtDNA probe using random priming procedures. The filters were then washed three times in 0.2 x SSC + 0.1% SDS at 65 [degrees] C. Restriction-fragment patterns were revealed by autoradiography; X-ray film exposure times ranged from several hours to five days. Distinct mtDNA restriction patterns produced by each restriction enzyme were designated by specific letters; for each enzyme, "A" represented one common pattern and "B" a second common pattern; other patterns were labeled alphabetically as detected.
For analysis of the allozyme data, a numerical genetic-index score was calculated for each of the three loci. Each [TABULAR DATA FOR TABLE 1 OMITTED] allele at each locus was assigned a numerical allele diagnostic value (ADV; Bert et al. 1993; Bert and Arnold 1995) calculated as [f.sub.a] - [f.sub.m], where [f.sub.a] and [f.sub.m] are the mean frequencies of the allele in M. adina and M. mercenaria, respectively (see Table 1). Then, for each individual, a genetic-index score for each locus was calculated as the sum of the two ADVs derived from the individual's genotype for the locus. This method follows the logic of Sage and Selander (1979), Howard (1986), and Bert and Harrison (1988) except that here the relative diagnostic capability of each allele is considered. The greater the frequency difference of an allele between two species, the closer its ADV value is to + 1 or - 1.
All analyses involving claw measurements were performed only on individuals possessing original claws because regenerated claws are more variable in size and coloration. Because we suspected that claw size is sexually dimorphic, we performed a series of size-corrected principal components analyses (PCAs) (Burnaby 1966) on the morphometric data, using various combinations of the mensural, sex, and claw-type variables. The Burnaby method removes variation described by the first PC axis, the axis usually dominated by differences in size. The effects of differences in shape can then be viewed using the other PC axes. Rohlf and Bookstein (1988) support the use of the Burnaby adjustment rather than shearing (Humphries et al. 1981; Bookstein et al. 1985), if the purpose of the adjustment is size correction.
Discriminatory Power. - to identify the relative contribution of the allozyme, coloration, and mensural characters in discriminating between the two species, an iterative process using PCA was employed. For each individual from the pure-species and the hybrid-zone samples, the three genetic-index scores, six coloration scores, and six morphometric scores were entered without transformation into standard PCAs performed as follows: (1) on each character set individually, (2) on all pairwise combinations of character sets, and (3) on all three character sets. In each analysis, the eigenvector coefficient for each character in the principal component (PC) that best separated the species (always PC1 or PC2) revealed the relative importance of that character in explaining the variance described by the eigenvector. Additional PCAs following the protocol described above were performed using various combinations of the characters that had comparatively high coefficients in the PC that best separated the species. From these analyses, various subsets of characters from the individual and combined character sets could be ranked by comparing the percentages of the total PC score ranges that separated the species in the analyses, and sets of diagnostic characters for the individual and combined character sets could be selected. This approach allowed us to determine the percentage of the total PC space separating the species (discriminatory power) in each PCA and to assess the percentages of individuals from the hybrid zone whose PC scores either fell within the ranges of the two species (i.e., could be classified as pure parentals) or were between the species' PC score ranges (i.e., could be classified as hybrids). All individuals from the reference pure-species samples and from the hybrid-zone samples scored for the characters under consideration were used in each PCA.
Although discriminant-function analysis (DFA) or its analogue for exploratory ordination, canonical-variates analysis (James and McCulloch 1990), is frequently used to describe the interrelationships of variables between species and their hybrids (e.g., Williams and Felder 1986; Lamb and Avise 1987; Dillon and Manzi 1989; Karakousis and Skibinski 1992), PCA is more appropriate because it does not require a priori definition of groups (Neff and Smith 1978; James and McCullock 1990). In the PC that best discriminates between the species, PCA yields a statistic for each individual that reflects the proportional admixture of parental genes in that individual. In addition, the coefficient of each character on the axis that best discriminates the species corresponds to the character's relative contribution to the variation along that axis (Wayne et al. 1989). PCA can be used to identify new combinations of uncorrelated variables for further study, as we did here. However, discrete variables (e.g., allozyme data) may not be well summarized by the method (James and McCulloch 1990), and the correlation matrix (which we used) must be used when characters among character sets are not measured on the same scale (Neff and Marcus 1980). In addition, some individuals will inevitably be misclassified (Neff and Smith 1978) because not all genes in the genome can be examined. Nevertheless, although conservative (Dowling et al. 1989), PCA is the preferred method for determining the characters that are informative from a discriminatory perspective and for objectively inferring the relative contribution of parental genes in individuals of mixed ancestry.
TABLE 2. Allozyme genotypes used to group individuals from the Menippe adina-M. mercenaria northwest Florida hybrid zone into allozyme-defined genotypic classes. Alleles a and b for each locus are defined in Table 1. - indicates locus not considered.
Genotype(1) Genotype class ALP-2 SOD IDH-2
M. adina bb bb bb bb bb - bb - bb - bb bb bb - - - bb - M. adina-like hybrid bb ab ab or bb ab bb ab or bb bb bb aa or ab bb ab - bb - ab ab bb - - bb ab Intermediate hybrid ab ab ab or bb(2) ab bb aa aa bb ab or bb bb ab aa bb aa ab or bb ab ab -(2) ab - ab or bb(2) aa bb - bb aa - bb - aa - ab ab or bb(2) - bb aa ab - -(2) - ab -(2) M. mercenaria-like hybrid aa ab aa, ab, or bb aa bb aa ab aa aa, ab, or bb ab ab aa bb aa aa aa ab - ab aa - ab - aa - ab aa M. mercenaria aa aa aa, ab, or bb aa aa - aa - aa, ab, or bb - aa aa, ab, or bb aa - - - aa - - - aa
1 Because all loci did not resolve in all individuals, we included two-locus and single-locus combinations as well as the three-locus combinations in our defined genotypic classes. The following single-locus combinations were too ambiguous to warrant assignment to a genotypic class: - - ab, - - bb.
2 [F.sub.1] hybrid genotypes.
To facilitate direct comparisons of our morphometric data with those of Williams and Felder (1986), we also performed regression analyses on pairwise combinations of mensural characters and compared the resulting regression lines using ANCOVA. However, we did not perform a stepwise DFA on morphometric ratios as did Williams and Felder (1986), for the reasons described above, and because many researchers advise against using ratios (James and McCulloch 1990; Rohlf 1990).
We did not include our mtDNA data in the iterative PCA analysis. Because heterozygosity cannot be expressed, hybrid individuals could never be intermediate to the parental species. To evaluate the utility of mtDNA RFLPs for species discrimination, we calculated the frequencies of mtDNA haplotypes in each species and tested for significant differences in frequency between species using the 2 x 2 G-test for independence (Sokal and Rohlf 1981). We also examined the pattern of mtDNA haplotype variation between species and the distribution of mtDNA haplotypes in individuals from the hybrid zone.
Congruence. - Using for each character set only characters deemed to be diagnostic as determined by the iterative PCAs, we examined patterns of congruence between pairwise combinations of the three nuclear-gene character sets by calculating the percentages of individuals that fell within the range of PC scores of the pure species in the PCA of one character set and into the hybrid range or range of the alternative species of a second character set. To show the relationship of the distribution of mtDNA haplotypes in hybrid-zone individuals to that of pure-species individuals, we coded mtDNA haplotypes onto bivariate plots of pairwise combinations of the three nuclear-gene character sets, using the diagnostic characters. To examine congruence between mtDNA haplotype and nuclear-gene markers, we used the R x C G-test (Sokal and Rohlf 1981) to test for independence of mtDNA haplotype and allozyme-defined genotypic class (Table 2), and we examined the extent and pattern of cytonuclear linkage disequilibrium between allozyme genotype and mtDNA haplotype (Arnold 1993, and references therein).
Populational and species allele frequencies and the ADV values for the three diagnostic allozyme loci are presented in Table 1. Allele frequencies for pure-species samples were generally very similar, but those from the two hybrid-zone samples differed notably. The hybrid zone is highly asymmetrical, and throughout most of the zone, Menippe mercenaria alleles predominate (Bert, unpubl. data). Nevertheless, [TABULAR DATA FOR TABLE 3 OMITTED] the hybrid-zone samples provided us with an adequate representation of a broad spectrum of genotypes.
The Burnaby size-corrected PCAs revealed that, in both species, the sexes are separable by PCA when claw measurements are included in the analyses [ILLUSTRATION FOR FIGURE 2A OMITTED]. However, no sexual dimorphism was discernible when the claws were removed from the analyses [ILLUSTRATION FOR FIGURE 2B OMITTED]. Therefore, for the iterative PCA, the sexes were analyzed separately only when claw measurements were included in an analysis.
All PCAs performed in our iterative procedure produced eigenvectors that weighted the allozyme and coloration characters in the same relative fashion in the PCs that best separated the two species (representative examples are given in Table 3). Eigenvector coefficients were consistently high for all allozyme loci, and as might be expected, the coefficients for the diagnostic loci in which allele frequency differences were nearly fixed (ALP-2 and SOD) were higher than those for the semidiagnostic (Bert and Arnold 1995) IDH-2 locus. The coefficients of five color characters were also consistently high and approximately equal; leg color has locality-specific variation (Bert, unpubl. data) that weakened the diagnostic property of this character in the PCAs. The coefficients for the morphometric character set varied more among analyses than did those of the other character sets, and the interactions of the morphometric characters with each other and with other characters were complex, probably because the distribution of the morphometric characters differs from that of the allozyme and coloration characters.
The combination of all allozyme characters plus the five diagnostic coloration characters best separated M. adina and M. mercenaria in the PCAs; the two species were separated by 54% of the total PCA score range [ILLUSTRATION FOR FIGURE 3 OMITTED]. Other character combinations of this combined character set yielded nearly the same level of resolution. For example, the percentage by which the character combination ALP-2, SOD, and all diagnostic colors separated the species was 52%; that of ALP-2, SOD, and the three color patterns was 51%; and that of the three allozyme loci and all six coloration characters was 51%. Both the allozyme and coloration character sets alone adequately distinguished the two species, but species discrimination was notably improved when the two character sets were combined (Table 3).
In the PCAs that best discriminated between species and that included morphometric characters, the eigenvector coefficients for ALT3 and ALT4 were high (Table 3). ALT4 coefficients were usually higher than those of ALT 3; however, discriminatory power was improved when both ALT3 and ALT4 were included in the morphometric analyses, along with some other body dimension. In any PCA in which mensural variables were involved, the inclusion of RPL and LPL necessitated separate analyses of the sexes, and the inclusion of CW resulted in a size influence in the PC that best separated the species (results not shown). Therefore, CL was the appropriate body dimension to include with ALT3 and ALT4 in the suite of diagnostic morphometric characters. The regressions of ALT4 on CL and of ALT3 on ALT4 [ILLUSTRATION FOR FIGURE 4 OMITTED] differed significantly between species (ALT4 on CL: [F.sub.MS] = 1.25, [F.sub.slope] = 33.23, [F.sub.intercept] = 371.67; ALT3 on ALT4: [F.sub.MS] = 1.34, [F.sub.slope] = 11.93, [F.sub.intercept] = 566.93; P [less than] 0.001 for all values); nevertheless, many pure-species individuals were indistinguishable from individuals of the alternate species. Both the standard and Burnaby size-corrected PCAs conducted on the diagnostic morphometric characters only partially distinguished the species [ILLUSTRATION FOR FIGURES 2-3 OMITTED]; the standard PCA PC2 scores of 27% of the M. adina and 48% of the M. mercenaria were within the region of overlap. Separating the sexes in the PCAs increased the distinction between species, but the eigenvector coefficients in the PCs that best separated the species differed between sexes (Table 3). When character sets were combined in an analysis, the morphometric character coefficients in the PC with the best discriminatory power were usually less than those of other character coefficients, indicating that the morphometric characters contributed less to those eigenvectors than did most allozyme or coloration characters. Thus, when other characters were combined with morphometric characters, discriminatory power improved compared to that obtained from the morphometric data alone. However, including morphometric characters in a combined character set typically reduced discriminatory power compared to that obtained when morphometric characters were omitted from the analysis.
The mtDNA restriction-site patterns produced by the six informative enzymes were clearly linked, and two common patterns occurred. Most individuals that exhibited the A common pattern had the six-enzyme composite haplotype AAAAAA (BclI, BstBI, BstNI, HaeIII, HindIII, and MspI). Five A-type haplotypes (AAABAA, AAADAA, AAAAAD, AACAAC, and ACAAAA) that appeared in single individuals differed from the six-A pattern by a single restriction site and were included in our A-type category. Two common B patterns occurred: BBBBBB and BBBBBE; only a single individual (BBBCBE) deviated from one of these patterns, again by a single restriction site. Because the MspI E pattern was always linked to the B pattern of the other five restriction enzymes, we considered both the six-B and the five-B, E patterns as belonging to the same mtDNA genome and included both, and the single individual with the alternative B haplotype, in our B-type category.
The frequencies of the A and B mtDNA haplotypes in M. adina (1.00 and 0.00, respectively; n = 41) and M. mercenaria (0.71 and 0.29, respectively, n = 45) were significantly different (G = 18.9, df = 1, P = 0.001). However, the utility of mtDNA haplotype for species discrimination in these stone crab species is restricted. The possession of a B mtDNA haplotype indicates that the individual is not a member of M. adina, but the possession of an A mtDNA haplotype gives no indication of species affinity [ILLUSTRATION FOR FIGURE 5A OMITTED].
Classifying Individuals from Hybrid Zones
When the statistical separation between the parental species is large, a high percentage of individuals can be classified as hybrid forms, and the probability that backcross individuals will be classified as pure-species individuals is reduced. The PC1 score ranges of the two stone crab species as defined by the diagnostic allozyme-plus-coloration character set differed in such a way that 61% of the crabs from the hybrid zone were classified as individuals of mixed ancestry (Table 4, [ILLUSTRATION FOR FIGURE 3 OMITTED]). The character sets that included either diagnostic coloration characters only or the diagnostic characters from all three character sets also identified high percentages of individuals from the hybrid zone as hybrids. However, the diagnostic morphometric characters alone did not allow for any individuals from the hybrid zone to be unambiguously classified as hybrids. The PC score of every individual from the hybrid zone fell within the range of scores of at least one of the parental species [ILLUSTRATION FOR FIGURE 3 OMITTED], and the PC scores of 50% of the hybrid-zone individuals fell within the region of overlap between the parental species' PC scores (Table 4). The PCAs performed on the diagnostic allozyme-plus-morphometric characters and the diagnostic coloration-plus-morphometric characters did identify hybrid-zone individuals as hybrids, but the percentages of those individuals were small (Table 4).
Depending on mating patterns and the fitness of different cytonuclear gene combinations, individuals in a hybrid zone may have the mtDNA haplotype of only one parental species (e.g., Patton and Smith 1993) or some combination of mtDNA haplotypes of both parental species (e.g., Edwards and Skibinski 1987; Forbes and Allendorf 1991). Individuals from the stone crab hybrid zone characterized as hybrids or as M. mercenaria based on allozyme characters carried either the A or B mtDNA haplotype, but individuals characterized as M. adina had only the A mtDNA haplotype [ILLUSTRATION FOR FIGURE 5B OMITTED]. The mtDNA polymorphism in both nuclear-marker-defined M. mercenaria and hybrids seriously limits the utility of this system for discriminating among genotypic classes. Individuals possessing the B mtDNA haplotype could be members of either the M. mercenaria or hybrid genotypic class, and individuals having the A mtDNA haplotype could belong to any of the three genotypic classes.
Neither coloration and allozymes nor coloration and mensural variables were completely congruent in classifying individuals from the hybrid zone [ILLUSTRATION FOR FIGURES 6A, C OMITTED]. A substantially higher percentage of individuals have coloration PC scores indicative of mixed ancestry and allozyme or morphometric scores indicative of M. mercenaria (37% and 18%, respectively) than have coloration PC scores indicative of mixed ancestry and allozyme or morphometric scores indicative of M. adina (1% and 2%, respectively; Table 5). Thus, some individuals that might be classified as M. mercenaria using allozymes or mensural variables alone might be classified as hybrids (predominantly M. mercenaria-like hybrids) using coloration in combination with either of these character sets. Allozymes and mensural variables showed greater congruence in the PC score ranges into which hybrid-zone individuals were categorized [ILLUSTRATION FOR FIGURE 6B OMITTED]; only 2% were classified morphometrically as M. mercenaria and allozymically as M. adina or hybrid, and only 6% were classified morphometrically as M. adina and allozymically as M. mercenaria or hybrid (Table 5). The difference in expression of the characters in the three character sets also affects the percentages of hybrid-zone individuals classified as parental types. The diagnostic allozyme, morphometric, and allozyme-plus-morphometric character sets all classified high proportions of those individuals as M. mercenaria, whereas the coloration and coloration-plus-morphometric character sets classified relatively high percentages as M. adina (Table 4). Nevertheless, for each pairwise character-set combination, the two character sets concurred in classification of approximately 50% of all hybrid-zone individuals (Table 5, sums of the first three values in each column).
The mtDNA haplotype distribution of the hybrid-zone individuals in relation to that of the pure-species individuals as defined by coloration and morphometric diagnostic characters is illustrated in Figure 5B. In no case did the mtDNA B haplotype occur in any individual from the hybrid zone that had a PC score within the range of M. adina. However, the B haplotype occurred with equivalent frequency in all types of hybrids, including M. adina-like hybrids. The frequencies of the A and B mtDNA haplotypes did not differ among allozyme-defined genotypic classes, but, using allozymes, we classified as M. adina only a few individuals for which we also analyzed mtDNA haplotype (Table 6A). Also, of the four allozyme-defined [F.sub.1] hybrids in the sample, three had the A mtDNA haplotype and one had the B haplotype. Moreover, the frequencies of cytonuclear genotypes observed were very close to those expected for all three loci (Table 6B), and no nuclear-cytoplasmic disequilibrium values were found to be significant using either the G-test (Table 6C) or exact test (Avise et al. 1990). However, all [D.sub.1] values were negative and all [D.sub.2] values were positive, suggesting that mating or survival of cytonuclear types is asymmetrical (Arnold 1993).
TABLE 4. Proportions of individuals from the Menippe northwest Florida hybrid zone whose principal component (PC) scores were contained within the range of one of the parental species or were intermediate between the ranges of both parental species (i.e., within a PC-score range designated as hybrid). The PC scores were obtained using diagnostic characters from three character sets (allozyme loci [A], coloration [C], morphometrics [M]). The PCs upon which these percentages were based and the characters used are listed in Table 3. The heading (n) indicates number of individuals.
Percentage in each range of PC scores Character-set combination (n) M. adina M. mercenaria hybrid
A (229) 7 80 13 C (307) 18 40 42 M(1) (340) 59 92 0 A + C (200) 5 34 61 A + M (227) 7 88 5 C + M (305) 30 51 19 A + C + M (199) 5 44 51
1 The PC score ranges of the two species overlapped. Therefore, 9% and 42% of the individuals from the hybrid zone could be unambiguously identified as M. adina or M. mercenaria, respectively; 50% of the hybrid-zone individuals could not be distinguished from either M. adina or M. mercenaria. TABLE 5. Percentages of individuals from the Menippe northwest Florida hybrid zone grouped within the ranges of principal components (PC) scores indicative of M. adina (Ma), M. mercenaria (Mm), or hybrids (hy) independently by three diagnostic character sets (allozymes [A], coloration [C], morphometrics [M]). For each pair of character sets, the taxonomic classification of individuals agreed in the combinations Ma-Ma, Mm-Mm, and hy-hy; the two-character sets disagreed in all other taxonomic groupings. The percentages were calculated using the distributions of individuals in the various quadrats of the bivariate plots depicted in Figure 6. Because the morphometric character set did not completely separate the two species, the hy category represents a region of overlap rather than a region of intermediate PC scores. The percentages for the Ma and Mm groupings were calculated using only individuals outside this overlap region; percentages in the overlap region were considered "hybrid" percentages; (n) number of individuals.
Taxonomic classification Character sets (1, 2)
Character Character A, C A, M C, M set 1 set 2 % % %
Ma Ma 6 3 7 Mm Mm 35 41 21 hy hy 7 9 24 Ma Mm 1 1 1 Mm Ma 8 4 0 Ma hy 1 4 10 Mm hy 37 35 16 hy Ma 4 2 2 hy Mm 1 1 18 (n) (200) (227) (305)
[TABULAR DATA FOR TABLE 6 OMITTED]
Species Discrimination and the Classification of Hybrid-Zone Individuals
Our results and those of previous studies demonstrate that an advantage can be gained by using multiple data sets when classifying individuals from closely related species and attempting to discern their hybrids. When multiple character sets are used, phylogenetic information is maximized (Hillis 1987), more individuals can be identified as hybrids than can be identified using only a single character set (Lamb and Avise 1987; Baker et al. 1989; Dowling et al. 1989), the directionality and intensity of introgression can be established, the probability of obtaining insight into genetically based interactions relevant to the speciation process is enhanced, the joint relationships of intercorrelated data can be discerned (Hillis 1987; James and McCulloch 1990), and estimates of the reliability of each data set can be made by comparing the levels of resolution of the various data sets and examining their congruence (Lanyon 1993). Using even a clearly defined genetic character set alone to classify individuals can result in the gross misclassification of extreme backcross individuals.
Including a genetic component in the suite of traits used is crucial. Erroneous conclusions or misinterpretations of hybrid-zone interactions and selection can result when individuals are identified using only morphology (Paige and Capman 1993). The use of molecular data can facilitate detection of introgression, which may be missed if only morphological characters are used (DePamphilis and Wyatt 1990). Sometimes phylogenetic differences clearly indicative of speciation or past hybridization are detected only by analyzing the rapidly evolving mtDNA molecule (Avise et al. 1986; Forbes and Allendorf 1991; Dowling and Childs 1992; Dowling and DeMarais 1993). However, conflict among molecular markers can exist, particularly in putative closely related taxa (e.g., Karl and Avise 1992), and sometimes patterns differ among all character sets used (e.g., Moore et al. 1991). Nevertheless, in general, as the number of independent traits one uses increases, the confidence in identification increases, provided that discordance among markers is not too great and is understood. In addition, it is important to survey a number of populations. Particularly for geographically subdivided hybridizing species, the evolutionary history of the populations can influence their genetic complement such that character expression and concordances or discordances among traits differ regionally (e.g., Bert and Harrison 1988; Wayne et al. 1991).
Of the characters we considered, allozymes and color characters were far more diagnostically informative than were mensural variables or mtDNA RFLPs. Because their genetic basis can usually be understood, diagnostic allozyme loci are frequently used to differentiate parental species individuals and their hybrids. Whether all allozyme loci are free from the influence of selection is debatable (Karl and Avise 1992), but the vast majority seem to be (Barton and Hewitt 1989; Skibinski et al. 1993). In stone crabs, the three diagnostic loci shift in frequency more or less concurrently through the northwest Florida hybrid zone, and in allopatry, species allele frequencies vary little geographically (Bert and Harrison 1988). If selection is operating on these presumably independent loci, the same selective forces would have to be influencing allele frequencies at each locus (or the genetic linkage group it represents) in the same manner.
Morphological analysis has experienced a resurgence, as exemplified by recent interest in assessing the relative value of morphological and molecular data in constructing phylogenies and, in applied biology, in subdividing fisheries stocks (e.g., Fornier et al. 1984; Davidson et al. 1985; Allendorf et al. 1987; Hillis 1987; DeSalle and Grimaldi 1991; Patterson et al. 1993; Thomas and Hunt 1993; Kinsey et al. 1994). However, the use of morphological traits has potential shortfalls that can render the exercise phylogenetically problematic. For some groups, coloration can be strongly influenced by selection (e.g., in birds [Joseph and Moritz 1993] and butterflies [Mallet and Barton 1989]). However, in deca-pod crustaceans, coloration has proven to be a reliable discriminator between species (Knowlton 1986, 1993). Some interspecific differences in coloration are subtle (Boileau 1991; Knowlton and Mills 1992). In Menippe adina and M. mercenaria, however, differences in both background color and color pattern are quite distinct (Bert 1985, 1986; Williams and Felder 1986), and variation between species is much greater than variation within species [ILLUSTRATION FOR FIGURE 3 OMITTED]; in hybrids, color is expressed as complex mixtures of colors and patterns (Bert and Harrison 1988) as is shown here. However, in Menippe, the use of coloration for taxonomic classification is limited, in that background color and color pattern change dramatically and complexly through early ontogeny in both species (Bert 1985, 1986). Thus, only crabs [greater than or equal to]50 mm CW can be reliably scored for coloration (Bert, pers. observ.).
The efficacy of morphometrics can be compromised by transient phenotypic plasticity, which allows for ecophenotypic variation that may falsely suggest hybridization between differentiated populations (Davidson et al. 1985; Allendorf et al. 1987; Kinsey et al. 1994). Increased phenotypic variability of hybrids, mosaic character expression, or selectively mediated canalization can decrease the power of morphometrics to discriminate hybrids from parentals (Lamb and Avise 1987; Dowling et al. 1989; Carson et al. 1989). Finally, in species that are morphometrically similar (such as M. adina and M. mercenaria), measurement error can lead to substantial misclassification of individuals (Francis and Mattlin 1986). The poor resolution of our morphometric data may have been, in part, a consequence of the measurements used. We used the mensural variables Williams and Felder (1986) described as those that best discriminated between M. adina and M. mercenaria. However, they reported neither all of the measurements they tested nor the measurement strategy they used. Most morphologists recommend employing a truss network (Humphries et al. 1981; Strauss and Bookstein 1982) in the measurement design and a multivariate analysis that examines shape change in relation to size change. A comprehensive truss- or landmark-based study might yield greater diagnostic resolution between the two stone crab species.
Cryptic taxa or intraspecific, genetically distinct subgroups that otherwise might not be detected can be differentiated by analysis of mtDNA (e.g., Kessler and Avise 1985; Reeb and Avise 1990; Moore et al. 1991). However, in some cases, the discriminatory power of mtDNA is no better or is poorer than that of allozymes (Edwards and Skibinski 1987; Dowling and Brown 1989). Because of its essentially matriarchal, clonal mode of inheritance (Moritz et al. 1987), mtDNA can take an independent evolutionary pathway that is concordant or discordant with that of nuclear traits (Barton and Jones 1983; Avise and Saunders 1984; Takahata and Slatkin 1984; Barton and Hewitt 1989; Moore et al. 1991; Degnan 1993). Together, M. adina and M. mercenaria possess two quite distinct mtDNA haplotypes (A and B). However, because only M. adina is monomorphic for one haplotype, the unambiguous diagnostic resolving power of mtDNA is limited to eliminating M. adina as a candidate for classification of any individual possessing a B haplotype.
The polymorphism of mtDNA in M. mercenaria could reflect extreme introgression (Dowling and Hoeh 1991), possibly facilitated by asymmetrical mating preferences (Lamb and Avise 1986), incomplete lineage sorting (Neigel and Avise 1986; Schneider-Broussard 1993), or an independent mutational event (Harrison 1990). Because it segregates independently of all nuclear genes, mtDNA can introgress to a greater degree than nuclear traits (Barton and Jones 1983; Harrison et al. 1987; Harrison 1989; Forbes and Allendorf 1991; Dowling and Hoeh 1991; Ballinger et al. 1992). The northwest Florida stone crab hybrid zone is highly asymmetrical; although very rare, nuclear-gene-coded characteristics reminiscent of M. adina can be identified in individuals even from southernmost Florida (Bert, unpubl. data). The M. adina haplotype may have introgressed into the range of M. mercenaria to a greater extent than have the nuclear-gene markers we monitor. However, cytonuclear analysis provides only qualified evidence that the observed mtDNA pattern is due to asymmetrical interspecific mating preferences. Although no D values were significant, the uniformly negative and positive signs of [D.sub.1] and [D.sub.2], respectively, suggest that there is a disproportionate tendency for M. adina males to mate with M. mercenaria females (Arnold 1993). On the contrary, the high frequency of the M. adina mtDNA haplotype within the range of M. mercenaria suggests that the progeny of matings between M. adina females and M. mercenaria males (B-ab cytonuclear types) have a greater viability than the progeny of reciprocal matings (A-ab cytonuclear types). There is some evidence for selection for the B haplotype; the highest frequency of that haplotype occurs in the hybrid zone (Bert, unpubl. data). If selection on hybrids is strong, a unique mtDNA haplotype can nearly replace a common haplotype within a few generations (Aubert and Solignac 1990). However, the absence of significant differences in the frequencies of the A and B haplotypes among the various genotypic classes suggests that within the region of haplotype variation, mating is essentially random and survival of progeny with respect to haplotype is equal (Dowling et al. 1989). Overall, the evidence for selection is not overwhelming.
The evidence that the observed mtDNA patterns are due to lineage sorting is also qualified. Sequence divergence is 2.0% (calculated using the method of Nei and Li , as formulated in the RESTSITE software program [Nei and Miller 1990], on the 43 restriction sites generated from the nine restriction endonucleases that cut the mtDNA more than once). This corresponds to an estimated mtDNA lineage divergence time of approximately 770,000-910,000 ybp (based on the mtDNA molecular clock estimated for snapping shrimp [Knowlton et al. 1993]), representing nearly 300,000 generations of stone crabs (based on 3 yr per generation [Bert et al. 1986]). Achieving a high probability of mtDNA lineage monophyly for species with large population sizes and carrying capacities (corresponding to data type 3 of Neigel and Avise 1986), such as the stone crab species, should take 2-3 x N generations ([ILLUSTRATION FOR FIGURE 9 OMITTED], Neigel and Avise 1986). For M. adina and M. mercenaria, this translates into at least 16-24 x [10.sup.6] generations (based on an N of at least 8 x [10.sup.6] crabs, estimated from fishery landings data [Simonson and Hochberg 1992; Bolden 1993]). Thus, lineage sorting may not be completed. However, the number of generations theoretically required to completely sort lineages seems unrealistically large, and if, as we suspect, the actual population size of stone crabs is substantially larger than we have estimated, then, hypothetically, it is highly improbable that monophyly would ever be achieved. Nevertheless, this exercise does provide for the possibility that insufficient time has elapsed for complete lineage sorting.
Incongruity in the expression of traits in hybrid-zone individuals reflects the presence of trailing neutral alleles in a moving hybrid zone (Dowling and Hoeh 1991), or the differential action of selection on the various characters or the gene associations they represent (introgression). Of the characters coded by nuclear genes in stone crabs, mensural characters and allozymes are largely congruent in their expression in hybrid-zone individuals, but coloration is expressed somewhat differently, indicating that the evolutionary forces acting in the hybrid zone affect coloration differently than they affect body shape and metabolic enzymes or that the mode of inheritance of coloration differs from that of mensural variables and allozymes. Nevertheless, there is an overall concordance in the complex geographic pattern of variation in the clearly diagnostic traits coloration and allozymes, indicating that these traits share a phylogenetic history (Bert and Harrison 1988). Therefore, the expression of coloration reminiscent of one species in an individual with mensural or allozyme characteristics of the alternate species is an indicator of gene flow between species and not of intraspecific color variation without phylogenetic significance. If coloration was not among the suite of traits used to classify individuals in the Menippe hybrid zone, the classification of hybrids would be overly conservative.
Applications of Total Evidence and Taxonomic Congruence
In the total-evidence approach, all data available are analyzed as an unpartitioned set (Jones et al. 1993). One character is pitted against another, patterns among characters are assessed, and the congruence among characters is used to derive a conclusion (Kluge 1989; Eernisse and Kluge 1993). In the taxonomic-congruence approach, patterns among character sets from different traits are assessed independently, and a consensus hypothesis is obtained (Bull et al. 1993) in which the criterion for resolution is agreement among the character sets.
For classifying individuals from zones of hybridization, it is the patterns among characters that are important. All informative evidence should be analyzed simultaneously using appropriate multivariate statistics. Analyzing all characters together allows the mathematical process to weigh the characters for their relative contribution toward discriminating between species. For example, the PCA correlation coefficients and eigenvector coefficients for the PC that best separates the species provide necessary information on the co-relationships among characters and the relative value of each character for taxonomic identification. Using PCA and a total-evidence approach, we can select a combination of influential characters that best distinguishes the species, independent of their inclusion in character sets. However, because the distributions and covariances of the data can be distorted by using the correlation matrix in PCA (which usually will be necessary), the behavior of the data from some character sets may be influenced when the underlying distributions of the data sets differ (e.g., continuously distributed mensural data vs. discretely distributed coloration data). In our case, this may have contributed to the variation in the morphometric eigenvector coefficients among PCAs. Nevertheless, reanalyzing the morphometric data using a size-corrected PCA based on covariances did not improve their discriminatory power.
Using a taxonomic-congruence approach to classify the genetic admixtures of individuals necessitates, in the absence of an agreement among character sets, a method for determining a consensus among the character sets. Confidence in the assignment of a taxonomic classification by any unweighted consensus method would be weakened because less-resolved and well-resolved character sets would be given equal weight. Many individuals classified as hybrids using a well-resolved character set would probably be erroneously classified as pure-species individuals using a less-resolved character set. Any method of weighting the character sets would require justification. This would be difficult if the discriminatory power of some characters in the set was great but overall the set was weak. The level of species discrimination that could potentially be provided by the highly diagnostic characters would be weakened by their a priori association with a suite of weaker characters.
Some criticisms levied against the total-evidence method in its phylogenetic application are the following: (1) When character sets are highly unequal in number of characters, the results will be biased by the larger data set (Lanyon 1993); (2) Including many characters increases the chance that the positive effect of reliable characters may be diluted by errors from less reliable characters (Bull et al. 1993); and (3) Some data sets are not amenable to combining (Lanyon 1993). In our application of the method, the first criterion should not be important unless character expression is highly incongruent among character sets; it is the pattern of covariation, not the number of characters alone, that generates the numerical PC score upon which an individual's taxonomic classification is based. Nevertheless, to prevent one data set from swamping another, the number of characters in the character sets should be relatively balanced (DeSalle and Grimaldi 1991; Honeycurt and Adkins 1993) if they are not completely congruent. The second criticism also is not a concern because the iterative PCA enables us to eliminate weak characters and character sets, thereby minimizing the potential for diluting discriminatory power. The third criterion is of concern in our application. For example, because mtDNA inheritance patterns are different from those of our other three traits, we could not combine the mtDNA RFLP data with the nuclear-marker character sets because the effectiveness of the PCA would have been compromised.
Through the use of iterative PCA and certain principles employed by systematists to structure phylogenetic hypotheses, we have attempted to formulate a more rigorous approach to the classification of individuals from hybrid zones. In our application of the total-evidence approach, the character set that provides the best discrimination between species and allows for classification of a broad spectrum of morphogenotypes from hybrid zones can be determined, and hybrid-zone individuals can be ranked by hybridity or classified into groupings for further analyses. Although a total-evidence approach seems to be the most appropriate for assigning individuals from a hybrid zone to taxonomic categories, a taxonomic-congruence approach is also useful in certain contexts because that approach enables researchers to make comparisons among character sets. Studying classes of characters can be informative when discerning historical patterns of species distributions, comparing the extent of introgression or level of discriminatory power among traits, or determining barriers to gene flow that may be important in the speciation process. In formalizing dichotomous approaches to phylogenetic analysis, systematic biologists fortuitously provided students of hybrid zones with a theoretical framework for classifying individuals that can be applied to and benefit future studies of hybrid zones.
We are grateful to the members of the commercial fishing industry and to the Sea Grant Program marine extension agents who assisted us with obtaining samples. We also thank S. Brown, R. Hochberg, J. Stevely, and W. Tweedale for field assistance; W. Moore for introducing T.M.B. to PCA; J. Arnold for providing to us his linkage-disequilibrium computer program and for performing the exact text for us; L. French and J. Leiby for editorial assistance; L. French for assistance with the illustrations; and J. Arnold, W. Arnold, B. Chernoff, T. Lamb, J. Quinn, T. Perkins, and two anonymous reviewers for valuable reviews and comments that improved the manuscript. This work was funded by grants from the Department of Commerce, National Oceanographic and Atmospheric Administration (P.L. 88-305 and P.L. 99-659), and by the State of Florida, Department of Natural Resources and Department of Environmental Protection, Florida Marine Research Institute.
ALLENDORF, F. W., N. RYMAN, AND F. M. UTTER. 1987. Genetics and fishery management. Pp. 1-19 in N. Ryman and F. Utter, eds. Population genetics and fishery management. University of Washington, Seattle.
ARNOLD, J. 1993. Cytonuclear disequilibria in hybrid zones. Annu. Rev. Ecol. Syst. 24:521-554.
ASMUSSEN, M. A., J. ARNOLD, AND J. C. AVISE. 1987. The effects of assortative mating and migration on cytonuclear associations in hybrid zones. Genetics 122:923-934.
AUBERT, J., AND M. SOLIGNAC. 1990. Experimental evidence for mitochondrial DNA introgression between Drosophila species. Evolution 44:1272-1282.
AVISE, J. C., AND N. C. SAUNDERS. 1984. Hybridization and introgression among species of sunfish (genus Lepomis): Analysis by mitochondrial DNA and allozyme markers. Genetics 108:237-255.
AVISE, J. C., G. S. HELFMAN, N. C. SAUNDERS, AND L. S. HALES. 1986. Mitochondrial DNA differentiation in North Atlantic eels: Population genetic consequences of an unusual life history pattern. Proc. Natl. Acad. Sci. USA 83:4350-4354.
AVISE, J. C., W. S. NELSON, J. ARNOLD, R. K. KOEHN, G. C. WILLIAMS, AND V. THORSTEINSSON. 1990. The evolutionary genetic status of Icelandic eels. Evolution 44:1254-1262.
BAKER, R. J., S. K. DAVIS, R. D. BRADLEY, M. J. HAMILTON, AND R. A. VAN DEN BUSSCHE. 1989. Ribosomal-DNA, mitochondrial DNA, chromosomal, and allozymic studies of a contact zone in the pocket gopher Geomys. Evolution 43:63-75.
BALLINGER, S. W., L. H. BLANKENSHIP, J. W. BICKHAM, AND S. M. CARR. 1992. Allozyme and mitochondrial DNA analysis of a hybrid zone between white-tailed deer and mule deer (Odocoileus) in west Texas. Biochem. Genet. 30:1-11.
BARRETT, M., M. J. DONOGHUE, AND E. SOBER. 1991. Against consensus. Syst. Zool. 40:486-493.
BARTON, N. H., AND K. S. GALE. 1993. Genetic analysis of hybrid zones. Pp. 13-45 in R. G. Harrison, ed. Hybrid zones and the evolutionary process. Oxford University Press, New York.
BARTON, N. H., AND G. M. HEWITT. 1989. Adaptation, speciation and hybrid zones. Nature 341:497-503.
BARTON, N. H., AND J. S. JONES. 1983. Mitochondrial DNA: New clues about evolution. Nature 306:317-318.
BERT, T. M. 1985. Geographic variation, population biology, and hybridization in Menippe mercenaria and evolution in the genus Menippe in the southwestern North Atlantic Ocean. Ph.D. diss. Yale University, New Haven.
-----. 1986. Speciation in western Atlantic stone crabs (genus Menippe): The role of geological processes and climatic events in the formation and distribution of species. Mar. Biol. 93:157-170.
BERT, T. M., AND W. S. ARNOLD. 1995. An empirical test of predictions of two competing models for the maintenance and fate of hybrid zones: Both models are supported in a hard clam hybrid zone. Evolution 49:276-289.
BEAT, T. M., AND R. G. HARRISON. 1988. Hybridization in western Atlantic stone crabs (genus Menippe): Evolutionary history and ecological context influence species interactions. Evolution 42:528-544.
BERT, T. M., J. TILMANT, J. DODRILL, AND G. E. DAVIS. 1986. Aspects of the population dynamics and biology of the stone crab (Menippe mercenaria) in Everglades and Biscayne Bay National Parks based on trapping. South Florida Research Center Report SFRC-86-04. Everglades National Park, Homestead, FL.
BERT, T. M., D. M. HESSELMAN, W. S. ARNOLD, W. S. MOORE, H. CRUZ-LOPEZ, AND D. C. MARELLI. 1993. High frequency of gonadal neoplasia in a hard clam (Mercenaria spp.) hybrid zone. Mar. Biol. 117:94-104.
BOILEAU, M. G. 1991. A genetic determination of cryptic species (Copepoda: Calanoida) and their postglacial biogeography in North America. Zool. J. Linn. Soc. 102:375-396.
BOLDEN, S. K. 1993. Summary of the Florida Gulf Coast commercial stone crab fishery 1962-1992. National Marine Fisheries Service, Southeast Fishery Science Center Contribution MIA-92/93-84. Miami, FL.
BOOKSTEIN, F. L., B. CHERNOFF, R. L. ELDER, J. M. HUMPHRIES, JR., G. R. SMITH, AND R. E. STRAUSS. 1985. Morphometrics in evolutionary biology. Spec. Publ. Acad. Nat. Sci. Phila. 15:1-265.
BULL, J. J., J. P. HUELSENBECK, C. W. CUNNINGHAM, D. L. SWOFFORD, AND P. J. WADDELL. 1993. Partitioning and combining data in phylogenetic analysis. Syst. Biol. 42:384-397.
BURNABY, T. P. 1966. Growth-invariant discriminant functions and generalized distances. Biometrics 22:96-110.
CARSON, H. L., K. Y. KANESHIRO, AND F. C. VAL. 1989. Natural hybridization between the sympatric Hawaiian species Drosophila silvestris and Drosophila heteroneura. Evolution 43:190-203.
DAVIDSON, K., J. C. ROFF, AND R. W. ELNER. 1985. Morphological, electrophoretic, and fecundity characteristics of Atlantic snow crab, Chionoecetes opilio, and implications for fisheries management. Can. J. Fish. Aquat. Sci. 42:474-482.
DECNAN, S. M. 1993. The perils of single gene trees: Mitochondrial versus single-copy nuclear DNA variation in white-eyes (Aves: Zosteropidae). Mol. Ecol. 2:219-225.
DEPAMPHILIS, C. W., AND R. WYATT. 1990. Electrophoretic confirmation of interspecific hybridization in Aescusus (Hippocastanaceae) and the genetic structure of broad hybrid zone. Evolution 44:1295-1317.
DESALLE, R., AND D. A. GRIMALDI. 1991. Morphological and molecular systematics of the Drosophilidae. Annu. Rev. Ecol. Syst. 22:447-475.
DILLON, R. T., JR., AND J. J. MANZI. 1989. Genetic and shell morphology in a hybrid zone between the hard clams Mercenaria mercenaria and M. campechiensis. Mar. Biol. 199:217-222.
DOWLING, T. E., AND W. M. BROWN. 1989. Allozymes, mitochondrial DNA, and levels of phylogenetic resolution among four minnow species (Notropis: Cyprinidae). Syst. Zool. 38:126-143.
DOWLING, T. E., AND M. R. CHILDS. 1992. Impact of hybridization on a threatened trout of the southwestern United States. Cons. Biol. 6:355-364.
DOWLING, T. E., AND B. D. DEMARAIS. 1993. Evolutionary significance of introgressive hybridization in cyprinid fishes. Nature 362:444-446.
DOWLING, T. E., AND W. R. HOEH. 1991. The extent of introgression outside the contact zone between Notropis cornutus and Notropis chrysocephalus (Teleostei: Cyprinidae). Evolution 45:944-956.
DOWLING, T. E., G. R. SMITH, AND W. M. BROWN. 1989. Reproductive isolation and introgression between Notropis cornutus and Notropis chrysocephalus (family Cyprinidae): Comparison of morphology, allozymes, and mitochondrial DNA. Evolution 43:620-634.
EDWARDS, C. A., AND D. O. F. SKIBINSKI. 1987. Genetic variation of mitochondrial DNA in mussel (Mytilus edulis and M. gallo-provincialis) populations from south west England and south Wales. Mar. Biol. 94:547-556.
EERNISSE, D. J., AND A. G. KLUGE. 1993. Taxonomic congruence versus total evidence, and amniote phylogeny inferred from fossils, molecules, and morphology. Mol. Biol. Evol. 10:1170-1195.
FORBES, S. H., AND F. W. ALLENDORF. 1991. Associations between mitochondrial and nuclear genotypes in cutthroat trout hybrid swarms. Evolution 45:1332-1349.
FORNIER, D. A., T. D. BEACHAM, B. E. RIDDELL, AND C. A. BUSACK. 1984. Estimating stock composition in mixed stock fisheries using morphometric, meristic, and electrophoretic characteristics. Can. J. Fish. Aquat. Sci. 41:400-408.
FRANCIS, R. I. C. C., AND R. H. MATTLIN. 1986. A possible pitfall in the morphometric application of discriminant analysis: Measurement bias. Mar. Biol. 93:311-313.
GRAHAM, J. H. 1992. Genomic coadaptation and developmental stability in hybrid zones. Acta. Zool. Fenn. 191:121-131.
HARRISON, R. G. 1989. Animal mitochondrial DNA as a genetic marker in population and evolutionary biology. Trends Ecol. Evol. 4:6-11.
-----. 1990. Hybrid zones: Windows on evolutionary processes. Pp. 69-128 in J. Antonovics and D. Futuyma, eds. Oxford surveys in evolutionary biology, vol. 7. Oxford University Press, Oxford.
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.
HEWITT, G. M. 1988. Hybrid zones: Natural laboratories for evolutionary studies. Trends Ecol. Evol. 7:158-167.
HILLIS, D. M. 1987. Molecular versus morphological approaches to systematics. Annu. Rev. Ecol. Syst. 18:23-42.
HONEYCUTT, R. L., AND R. M. ADKINS. 1993. Higher level systematics of Eutherian mammals: An assessment of molecular characters and phylogenetic hypotheses. Annu. Rev. Ecol. Syst. 24:279-305
HOWARD, D. J. 1986. A zone of overlap and hybridization between two ground cricket species. Evolution 40:34-43.
HOWARD, D. J., G. L. WARING, C. A. TIBBETS, AND P. G. GREGORY. 1993. Survival of hybrids in a mosaic hybrid zone. Evolution 47:789-800.
HUMPHRIES, J. M., F. L. BOOKSTEIN, B. CHERNOFF, G. R. SMITH, R. L. ELDER, AND S. G. Poss. 1981. Multivariate discrimination by shape in relation to size. Syst. Zool. 30:291-308.
JAMES, F. C., AND C. E. MCCULLOCH. 1990. Multivariate analysis in ecology and systematics: Panacea or Pandora's box? Annu. Rev. Ecol. Syst. 21:129-166.
JONES, T. R., A. G. KLUGE, AND A. J. WOLF. 1993. When theories and methodologies clash: A phylogenetic reanalysis of the North American ambystomatic salamanders (Caudata: Ambystomatidae). Syst. Biol. 42:92-102.
JOSEPH, L., AND C. MORITZ. 1993. Phylogeny and historical aspects of the ecology of eastern Australian scrubwrens Sericornis spp.: Evidence from mitochondrial DNA. Mol. Ecol. 2:161-170.
KARAKOUSIS, Y., AND D. O. F. SKIBINSKI. 1992. An analysis of allozyme, mitochondrial DNA and morphological variation in mussel (Mytilus galloprovincialis) populations from Greece. Experientia 48:878-881.
KARL, S. A., AND J. C. AVISE. 1992. Balancing selection at allozyme loci in oysters: Implications from nuclear RFLPs. Science 256:100-102.
KESSLER, L. G., AND J. C. AVISE. 1985. Microgeographic lineage analysis by mitochondrial genotype: Variation in the cotton rat (Sigmodon hispidus). Evolution 39:831-837.
KINSEY, S. T., T. ORSOY, T. M. BERT, AND B. MAHMOUDI. 1994. Population structure of the Spanish sardine Sardinella aurita: Natural morphological variation in a genetically homogeneous population. Mar. Biol. 118:309-317.
KLUGE, A. G. 1989. A concern for evidence and a phylogenetic hypothesis of relationships among Epicrates (Boidae, Serpentes). Syst. Zool. 38:7-25.
KNOWLTON, N. 1986. Cryptic and sibling species among the deca-pod Crustacea. J. Crustacean Biol. 6:356-363.
-----. 1993. Sibling species in the sea. Annu. Rev. Ecol. Syst. 24:189-216.
KNOWLTON, N., AND D. K. MILLS. 1992. The systematic importance of color and color pattern: Evidence for complexes of sibling species of snapping shrimp (Caridea: Alpheidae: Alpheus) from the Caribbean and Pacific coasts of Panama. Proc. San Diego Soc. Nat. Hist. 18:1-5.
KNOWLTON, N., L. A. WEIGHT, L. A. SOLORZANO, D. K. MILLS, AND E. BERMINGHAM. 1993. Divergence in proteins, mitochondrial DNA, and reproductive compatibility across the Isthmus of Panama. Science 260:1629-1632.
LAMB, T., AND J. C. AVISE. 1986. Directional introgression of mitochondrial DNA in a hybrid population of tree frogs: The influence of mating behavior. Proc. Natl. Acad. Sci. USA 83:2526-2530.
-----. 1987. Morphological variability in genetically defined categories of anuran hybrids. Evolution 41:157-165.
LANYON, S. M. 1993. Phylogenetic frameworks: Towards a firmer foundation for the comparative approach. Biol. J. Linn. Soc. 49:45-61.
MALLET, J., AND N. H. BARTON. 1989. Strong natural selection in a warning-color hybrid zone. Evolution 43:421-431.
MANIATIS, T. E., E. F. FRITSCH, AND J. SAMBROOK. 1982. Molecular cloning. Cold Springs Harbor Laboratory, Cold Springs Harbor, NY.
MARCHANT, A. D., AND D. D. SHAW. 1993. Contrasting patterns of geographic variation shown by mtDNA and karyotype organization in two subspecies of Caledia captiva (Orthoptera). Mol. Biol. Evol. 10:855-872.
MIYAMOTO, M. M. 1985. Consensus cladograms and general classifications. Cladistics 1:186-189.
MOORE, W. S., J. H. GRAHAM, AND J. T. PRICE. 1991. Mitochondrial DNA variation in the northern flicker (Colaptes auratus, Aves). Mol. Biol. Evol. 8:327-344.
MORITZ, C., T. E. DOWLING, AND W. M. BROWN. 1987. Evolution of animal mitochondrial DNA: Relevance for population biology and systematics. Annu. Rev. Ecol. Syst. 18:269-292.
NEFF, N. A., AND L. F. MARCUS. 1980. A survey of multivariate methods for systematics. American Museum of Natural History, New York.
NEFF, N. A., AND G. R. SMITH. 1978. Multivariate analysis of hybrid fishes. Syst. Zool. 28:176-196.
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.
NEI, M., AND J. C. MILLER. 1990. A simple method for estimating average number of nucleotide substitutions within and between populations from restriction data. Genetics 125:873-879.
NEIGEL, J. E., AND J. C. AVISE. 1986. Phylogenetic relationships of mitochondrial DNA under various demographic models of speciation. Pp. 515-534 in S. Karlin and E. Nevo, eds. Evolutionary processes and theory. Academic Press, New York.
NELSON, G. 1979. Cladistic analysis and synthesis: principles and definitions, with a historical note on Adanson's "Families des Plantes" (1763-1764). Syst. Zool. 28:1-21.
PAIGE, K. N., AND W. C. CAPMAN. 1993. The effects of host-plant genotype, hybridization, and environment on gall-aphid attack and survival in cottonwood: The importance of genetic studies and the utility of RFLPs. Evolution 47:36-45.
PATTERSON, C., D. M. WILLIAMS, AND C. J. HUMPHRIES. 1993. Congruence between molecular and morphological phylogenies. Annu. Rev. Ecol. Syst. 24:153-188.
PATTON, J. L., AND M. F. SMITH. 1993. Molecular evidence for mating asymmetry and female choice in a pocket gopher (Thomomys) hybrid zone. Mol. Ecol. 2:3-8.
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.
REEB, C. A., AND J. C. AVISE. 1990. A genetic discontinuity in a continuously distributed species: Mitochondrial DNA in the American oyster, Crassostrea virginica. Genetics 124:397-406.
ROHLF, F. J. 1990. Morphometrics. Annu. Rev. Ecol. Syst. 21:299-316.
ROHLF, F. J., AND F. L. BOOKSTEIN. 1988. A comment on shearing as a method for "size correction." Syst. Zool. 36:356-367.
SAGE, R. D., AND R. K. SELANDER. 1979. Hybridization between species of the Rana pipiens complex in central Texas. Evolution 33:1069-1088.
SCHNEIDER-BROUSSARD, R. M. 1993. Composition, variation and divergence in the nuclear and mitochondrial genomes of the stone crabs Menippe adina and M. mercenaria. Ph.D. diss. University of Southwestern Louisiana, Lafayette.
SIMONSON, J. L., AND R. J. HOCHBERG. 1992. An analysis of stone crab (genus Menippe) claws and their use in interpreting landings on Florida's west coast. Fla. Mar. Res. Publ. 50:26-35.
SKIBINSKI, D. O. F., M. WOODWARK, AND R. D. WARD. 1993. A quantitative test of the neutral theory using pooled allozyme data. Genetics 135:233-248.
SOKAL, R. R., AND F. J. ROHLF. 1981. Biometry, 2d ed. Freeman, New York.
STRAUSS, R. E., AND F. L. BOOKSTEIN. 1982. The truss: Body form reconstructions in morphometrics. Syst. Zool. 31:113-135.
SWOFFORD, D. L. 1991. When are phylogeny estimates from molecular and morphological data incongruent? Pp. 295-333 in M. M. Miyamoto and J. Cracraft, eds. Phylogenetic analysis of DNA sequences. Oxford University Press, New York.
TAKAHATA, N., AND M. SLATKIN. 1984. Mitochondrial gene flow. Proc. Natl. Acad. Sci. USA 81:1764-1767.
THOMAS, R. H., AND J. A. HUNT. 1993. Phylogenetic relationships in Drosophila: A conflict between molecular and morphological data. Mol. Biol. Evol. 10:362-274.
WAYNE, R. K., B. VAN VALKENBURGH, P. O. KAT, T. K. FULLER, W. E. JOHNSON, AND S. J. O'BRIEN. 1989. Genetic and morphological divergence among sympatric canids. J. Hered. 80:447-454.
WAYNE, R. K., S. B. GEORGE, D. GILBERT, P. W. COLLINS, S. D. KOVACH, D. GIRMAN, AND N. LEHMAN. 1991. A morphologic and genetic study of the island fox, Urocyon littoralis. Evolution 45:1849-1868.
WILLIAMS, A. B., AND D. L. FELDER. 1986. Analysis of stone crabs: Menippe mercenaria (Say), restricted, and a previously unrecognized species described (Decapoda: Xanthidae). Proc. Biol. Soc. Wash. 99:517-543.
|Printer friendly Cite/link Email Feedback|
|Author:||Bert, Theresa M.; McCarthy, Kevin J.; Cruz-Lopez, Hector; Bogdanowicz, Steven|
|Date:||Apr 1, 1996|
|Previous Article:||Mode of sexual selection determined by resource abundance in two sand goby populations.|
|Next Article:||Shell microstructure of gastropods from Lake Tanganyika, Africa: adaptation, convergent evolution, and escalation.|