In Kanagawa Prefecture and its surrounding area, two types of body color populations (DA: dark and BL: blue) of Japanese freshwater crab, Geothelphusa dehaani, are distributed parapatrically, suggesting some extent of genetic differentiation between them. In this study we examined the genetic variation of G. dehaani in Kanagawa Prefecture and Tokyo by means of electrophoretic analysis on 5 enzyme loci. The results clearly show the existence of substantial genetic differentiation among different color type populations. In addition, some extent of differentiation is also observed within the BL populations.
Geothelphusa dehaani (White) is a freshwater crab inhabiting widely in the mainland of Japan (Honshu, Shikoku, Kyushu), Yakushima Island and Tanegashima Island . Remarkable variation in body color has been observed [4, 5, 8, 13–15, 18] in natural populations of this crab. Chokki [4, 5] classified the body color types into three major categories, DA (dark), BL(blue) and RE (red), and examined the geographic distributions of these types in the northern and middle parts of Honshu. In the northern part of Honshu (Tohoku District) only DA crabs were distributed, whereas in the middle part of Honshu (Kanto and Tokai District) all three color types were distributed allopatrically or parapatrically. A distinct distribution pattern of different color type populations (RE and BL) was also observed in the southern part of Japan (Kagoshima Prefecture, Kyushu District), by Suzuki and Tsuda . These observations lead us to think that some extent of genetic differentiation exists between different color type populations of G. dehaani.
Reports have been made on various morphological and electrophoretical studies intended to reveal genetic differentiation among local populations of G. dehaani [8, 13, 14]. However, little is yet known about detailed geographic or genetic structuring of natural populations in connection with body color polymorphism.
Recently, Suzuki  made extensive analyses of coloration and distribution of G. dehaani in the rivers of the Tanzawa Mountains (Kanagawa Prefecture) and found that populations of the two color types, DA and BL, were distributed parapatrically. At every river, DA crabs inhabit the upper reaches and BL crabs inhabit the lower reaches. Replacement of the distribution from one color type to the other was found to occur abruptly within a very narrow geographic range. Since the boundaries between the DA and BL populations at the Tanzawa Mountains were so clear, we planned to elucidate a detailed genetic or geographic structuring of G. dehaani populations in Kanagawa Prefecture and its surrounding area.
In the present study, using electrophoretic analysis, we studied genetic variation in natural populations of G. dehaani in Kanagawa Prefecture and Tokyo. The results clearly demonstrate the existence of substantial genetic differentiation associated with body color polymorphism.
MATERIALS AND METHODS
A total of 544 crabs were collected from 16 populations in Kanagawa Prefecture and 2 populations in Tokyo (Table 1 and Fig. 1). Crab's body color is known to vary depending on developmental stage and physiological conditions [4, 14, 15]. Suzuki  found that in Kanagawa Prefecture body color finally stabilized to either DA or BL when crabs developed more than about 20 mm in carapace width. Therefore, to characterize the populations with color types correctly and study the relationship between color polymorphism and isozyme variation, we tried to collect crabs larger than 20 mm in carapace. However, at collection sites where enough number of large crabs was not available, some small crabs were also collected to keep the sample size more than 20 for isozyme analysis.
Collection information of G. dehaani examined in the present study
Livers and muscles were extracted and homogenized with cold 40% sucrose solution. The supernatants of homogenates by centrifugation (12,000 rpm, 20 min.) were stored at −40°C before electrophoresis. Polyacrylamide gel  and horizontal starch gel electrophoresis  were performed for detection of isozyme variation at the five enzyme loci; acid phosphatase (Acph), amylase-3 (Amy-3), muscle esterase (Est-M), liver esterase (Est-L) and superoxide dismutase (Sod).
The number of crabs collected and frequencies of different color types in each population are shown in Table 1. In all populations, smaller crabs of less than 20 mm in carapace width showed a variety of body colors that could be designated as DA, BL, TC (two-toned), PB (purple-brown), and GP (green-brown) according to Suzuki's classifications , whereas most of the larger crabs showed DA or BL body colors. Considering the observations that color types other than DA and BL are transitional in Kanagawa Prefecture and its surrounding region ( and our unpublished data), 16 out of 18 populations were characterized unambiguously as DA (7 populations: all or most crabs were DA, but none were BL) or BL (9 populations: high proportion of BL and TC, but no DA crabs) populations. In the remaining two populations, Nos. 7 and 9, the majority of crabs were DA, or BL and TC, but a 7 and 9, the majority of crabs were DA, or BL and TC, but a few BL or DA crabs were contaminated, respectively. The color types of these two populations were then designated as DA′ (No. 7) and BL′ (No. 9).
Table 2 gives the allele frequencies, average heterozygosities and number of crabs examined for 5 different enzyme loci. Alleles at each locus are designated through numbering in increasing order of electrophoretic mobilities (Fig. 2). Most of the populations studied were polymorphic (i.e., the most common allele showed a frequency of less than 0.95) for all enzyme loci. Exceptional monomorphism was observed at the Est-M locus for BL populations Nos. 16, 17 and 18, and for this reason average heterozygosity tended to be lower in these three populations (H = 0.246 to 0.332) than in the other populations. Although in many other populations the degree of polymorphism did not differ significantly, population No. 7 (color type DA′) showed an unusually high level of polymorphism. The value of average heterozygosity for this population was 0.542, whereas the unweighted mean of H in all populations was 0.387 ± S.E. 0.016.
Allele frequencies at 5 enzyme loci for 18 natural populations of G. dehaani in Kanagawa Prefecture and Tokyo
The allele frequencies at the Acph locus were very similar throughout the populations. At the remaining four loci, however, remarkable differences in allele frequency were detected between DA and BL populations. At the Est-L locus, two alleles, Est-L2 and Est-L4, were predominant in all the DA populations. Meanwhile, Est-L3 and Est-L5 were found in relatively high frequencies in all the BL populations. At the Sod locus, the allele Sod1 was the most common throughout the BL populations, ranging from 0.523 to 0.903, while this allele was rare or absent in most of the DA populations. The allelic differentiation between different color type populations was more pronounced at the Amy-3 locus. Furthermore, at this locus, in addition to the considerable variation between the DA and BL populations, local differences in allele frequencies were also observed within the BL populations. Two alleles, Amy-33 and Amy-35, were commonly predominant at the Amy-3 locus in the DA populations, but a different allele, Amy-32, which was very rare in the DA populations, occurred in high frequencies in all BL populations. Moreover, in the BL populations of Nos. 16, 17 and 18, allele Amy-34 was observed in high frequencies (0.419–0.629), whereas this allele was almost absent in other populations irrespective of body color type. A similar tendency was observed at the Est-M locus. In this case, BL populations Nos. 16, 17, and 18 were characterized by the fixation or near fixation of allele Est-M3, which was found in moderate frequencies in other BL populations but was almost absent in the DA populations. In Figure 1a and 1b, graphic descriptions of frequency distributions at Sod and Amy-3 loci are presented.
Based on the allele frequencies shown in Table 2, Nei's standard genetic distance, D , was estimated to evaluate genetic differentiation in natural populations of G. dehaani in relation to body color polymorphism (Table 3). Table 4 summarizes the mean values of D with standard errors by classifying the population comparisons into several categories. In this Table, BL populations are divided into BL1 (population No. 16–18) and BL2 (population No. 9–15) because of some extent of local differentiation as mentioned above. It is apparent from Table 4 that different color type populations, DA and BL, differ significantly in their allelic configurations. The mean value of D between DA and BL populations, 0.887 ± 0.027, is several ten times larger than those between populations of the same color type. Between BL1 and BL2 the value is estimated to be 0.266 ± 0.017, or that is several times larger than those within the BL1 or BL2 populations (0.028 and 0.057, respectively).
Genetic distance among 18 populations of G. dehaani in Kanagawa Prefecture and Tokyo
Average genetic distance (D) between populations of G. dehaani
A dendrogram of genetic relatedness among the populations was constructed using the neighbor-joining method  from the matrix of genetic distance (Table 3) as shown in Figure 3. In the tree, the large two differentiated clusters of populations that correspond perfectly to the grouping of body colors are presented. The cluster of BL populations is further divided into two distinct subclusters, BL1 and BL2. Within the BL2 cluster, population No. 15 is somewhat distantly related from the other BL2 populations.
A species is usually divided in its distribution into local populations among which some genetic differentiation can be recognized by means of several methods of investigation. To understand the genetic basis of evolutionary process, and particularly of speciation, patterns of genetic structuring of local populations provide invaluable information . However, dispersal and gene flow act as factors that override genetic differentiation among geographical populations . Studies on genetic differentiation in Japanese natural populations of animals with a high dispersal capability, such as some species of Drosophila, have sometimes failed to detect clear genetic structuring ([6, 16], but see ). The habitat of G. dehaani is restricted to small streams or swamps with clean freshwater. Juveniles of this crab develop from eggs without passing through a ‘swimming’ larval stage. In consideration of these characteristics, the dispersal capability of this crab is thought to be very low, and marked regional genetic differentiation, as suggested by the observation of body color polymorphism, is expected. Indeed, in the present study, using electrophoretic analysis, we found considerable differences in allele frequency among local populations of G. dehaani within a very restricted area, Kanagawa Prefecture and a part of Tokyo. Allozyme variation in Japanese natural populations of G. dehaani was also reported by Sugawara and Gamo  and Nakajima and Masuda . Although these studies coincidentally suggested the existence of genetically differentiated local populations in relation to body color polymorphism, numbers of populations or isozyme loci examined were not sufficient to clarify the detailed genetic organization of G. dehaani in Japan.
Our study of genetic differentiation is based on the extensive analyses of body color polymorphism by Suzuki . As shown in Table 1 and Figure 1, two types of color population, DA and BL, exist parapatrically in Kanagawa Prefecture, and the boundaries between these color types are very clear. Although the causative genetic or ecological mechanisms for body color polymorphism have not been completely resolved [7, 13], the pattern of distribution in Kanagawa Prefecture strongly suggests the existence of extensive genetic differentiation between the two color type populations. This prediction has turned to be correct. In Figures 1 and 3, substantial genetic differentiation among populations in complete association with color types is clearly indicated. Moreover, the BL populations can be divided into two further differentiated groups, BL1 and BL2 (Fig. 3), and the genetic division of BL1 and BL2 is associated with geographic location (Fig. 1b) The BL1 populations (No. 16, 17 and 18), characterized by somewhat low heterozygosities due to a lack of variation at the Est-M locus and by a unique allele of the Amy-3 locus (Table 2), lie east of Kanagawa Prefecture or southwest of Tokyo, while all the BL2 populations lie west part of Kanagawa Prefecture (Fig. 1). Together with our unpublished data on Amy-3 and Sod loci for several other populations in Kanagawa Prefecture, the presumable boundary between BL1 and BL2 is suggested to be Sagami River (Fig. 1). These results indicate that isozyme differentiation of G. dehaani occurs within a very narrow geographical range with a lack of association of morphological variation. Population No. 15 in BL2 is positioned slightly distant from other BL2 populations in the dendrogram (Fig. 3). The mean genetic distance between No. 15 and other BL2 populations is 0.123 ± 0.015, whereas the value among BL2 populations, including No. 15, is 0.057 ± 0.010 (Table 3, 4). Geographically, population No. 15 is the westernmost in this examination. The different allelic configuration observed in this population may indicate the occurrence of the 3rd genetically structured area of BL populations.
Population No. 7 is an exceptional case in the DA populations from where a small number of BL crabs were collected (Table 1), so we distinguished this population from the other DA populations, as DA′. Genetically, this population was found to be the most polymorphic (H = 0.542) of all those examined (Table 2). In the dengrogram (Fig. 3), No. 7 is slightly differentiated from a cluster consisting of all other DA populations.
Since population No. 7 seems to locate on the boundary between the DA and BL2 populations (Fig. 1), possible explanations for the coexistence of different color types, the high level of isozyme variability and slight genetic differentiaion from DA populations may be “contaminantion” of BL crabs into DA population or “hybridization” between DA and BL crabs. Examining the allelic configuration (Table 2) and the observed genotypes of isozyme loci for individual crabs (Table 5), some extent of gene flow by hybridization at population No. 7 is suggested. The alleles that exist in high frequencies in BL2 populations, i. e., allele Amy-32, and Sod1, tend to occur more frequently in population No. 7 than in the other DA populations (Table 2). The exceptional BL crabs in this population, M4 and F1, did not necessarily show the genotypes typically observed in BL2 populations (Table 5). At the Amy-3 locus, for example, both crabs were heterozygous for alleles Amy-32 and Amy-35. This genotype was very rare in BL2 populations. At the Est-L locus, crabs M4 and F1 were homozygous for Est-L4 allele which was predominant in all DA populations but was rare in BL2 populations (Table 2). Furthermore, among the DA crabs in population No. 7, some had unique genotypes. For example, F2, F13 and F15 crabs were heterozygous for alleles Est-M3 and Est-M4. This genotype was never found in other DA populations because of complete lack of allele Est-M3. From these observations, it is reasonable to think that the coexistence of different color types and the characteristic genetic structure at population No. 7 are due to hybridization between DA and BL2 populations. However, the present data is not enough for any statistical test to confirm the hypothesis.
Isozyme genotypes (5 loci) for individual crabs collected from Population No. 7 and 9
From population No. 9, several DA crabs were collected in the majority of BL crabs (Table 1). The genotypes of isozyme loci for individual crabs at this population are shown in Table 5. The exceptional DA crabs, F6, F12, F14 and F15, showed genotypes which were typically found in BL crabs. The coexistence of different color crabs in population No. 9 can, therefore, be also explained by hybridization. In this case, however, the extent of gene flow by hybridization seems to be less extensive than in population No. 7, because the allele frequencies and average heterozygosity of population No. 9 are fairly typical of BL2 populations (Table 2, Fig. 3).
Although the value of genetic distance between DA and BL populations is very large compared to those within DA or BL populations (Table 4), and the presumable hybrid zone between different color types seems to be restricted within a very narrow range, it is still uncertain because of a small number of isozyme loci examined that these two color type populations are differentiated to the species level. Studies on a greater number of isozyme loci or on other genetic characteristics are needed. Whatever the taxonomic status is, however, considerable genetic divergence between local populations of G. dehaani within a very narrow geographic range is interesting for studies on the evolutionary forces of the speciation process.