Allozyme Variability among Populations of Three Species of Brush-Furred Mice (Lophuromys,Rodentia, Muridae) from the Bale Mountains National Park (Ethiopia)

Allozyme variability was examined in populations of three endemic species of the species complex Lophuromys flavopunctatussensu lato: L. chrysopus, L. brevicaudus, and L. melanonyx. These species replace each other in adjacent altitudinal belts of the Bale Massif in Ethiopia. A deficit of heterozygotes at several loci was found in most samples of all species studied. Moreover, the samples included animals homozygous for two or three minor alleles and heterozygous for alleles that are rare and unique for the given species. It is suggested that the Bale Massif are inhabited by numerous genetically isolated populations of eachLophuromys species, which exchange genes at an extremely low rate. Genotypic disequilibrium observed in most samples is explained by the fact that most sampling localities comprise ranges of two and more micropopulations. In our view, microgeographic subdivision of the populations is caused by recurrent fragmentation of habitats during the Pleistocene glaciation of the Bale Massif and subsequent prolonged isolation of local populations. Gene drift accompanying these processes resulted in high genetic differentiation of the local populations, which probably persisted until the present. Geographical isolation of the Bale Massif, its uniquely diverse ecological conditions, and extraordinary allozyme structure of the Lophuromys populations suggest that these populations represent remnants or direct descendants of relic local populations.     

A River Metapopulation Structure of a Japanese Freshwater Goby, Odontobutis Obscura, Deduced from Allozyme Genetic Indices

Gene products of 18 allozyme loci from 1268 individuals of a Japanese freshwater goby called donko, Odontobutis obscura (Odontobutidae; Gobioidei), from 33 localities in the Koya River, Yamaguchi Prefecture, Japan, were investigated to determine the extent of genetic divergence and gene flow within a river metapopulation. Genetic indices including GST(mean FST 0.182), FIT(mean 0.192) and D(mean 0.015) indicated a considerable divergence of local populations in the river. The genetic distance (D) and channel distance between pairs of populations did not show a good correlation, and geographical neighbors were not always genetic neighbors. Therefore, the genetic divergence of populations is attributable to independent genetic drift with restricted gene flow among populations. The agricultural dams and weirs constructed across the river must be responsible for the restricted gene flow. The metapopulation structure of O. obscura in the Koya River may be barely sustained by one-way gene flow only from the upper to the lower populations. An occasional artificial transplantation of some individuals from the lower to the upper populations may be one alternative to maintain a river metapopulation structure safely.     

Biochemical characterization of aspartate aminotransferase allozymes from common wheat

Six allozymes of aspartate aminotransferase (AAT, EC 2.6.1.1): three plastidial (AAT-2 zone) and three cytosolic (AAT-3 zone) were isolated from common wheat (Triticum aestivum) seedlings and highly purified by a five-step purification procedure. The identity of the studied proteins was confirmed by mass spectrometry. The molecular weight of AAT allozymes determined by gel filtration was 72.4±3.6 kDa. The molecular weights of plastidial and cytosolic allozymes estimated by SDS-PAGE were 45.3 and 43.7 kDa, respectively. The apparent Michaelis constant (K _m) values determined for four substrates appeared to be very similar for each allozyme. The values of the turnover number (k _cat) and the k _cat/K _m ratio calculated for allozymes with L-aspartate as a leading substrate were in the range of 88.5–103.8 s^?1/10,412–10,795 s^?1 M^?1 for AAT-2 zone and 4.6–7.0 s^?1/527–700 s^?1 M^?1 for AAT-3 zone. These results clearly demonstrated much higher catalytic efficiency of AAT-2 allozymes. Therefore, partial sequences of cDNA encoding AATs from different zones were obtained using the RT-PCR technique. Comparison of the AAT-2 and AAT-3 amino acid sequences from active site regions revealed five non-conservative substitutions, which impact on the observed differences in the isozymes catalytic efficiency is discussed.     

Evolutionary implications of allozyme and RAPD variation in diploid populations of dioecious buffalograss Buchloë dactyloides

Buffalograss, Buchloë dactyloides, is widely distributed throughout the Great Plains of North America, where it is an important species for rangeland forage and soil conservation. The species consists of two widespread polyploid races, with narrowly endemic diploid populations known from two regions: central Mexico and Gulf Coast Texas. We describe and compare the patterns of allozyme and RAPD variation in the two diploid races, using a set of 48 individuals from Texas and Mexico (four population samples of 12 individuals each). Twelve of 22 allozyme loci were polymorphic, exhibiting 35 alleles, while seven 10-mer RAPD primers revealed 98 polymorphic bands. Strong regional differences were detected in the extent of allozyme polymorphism: Mexican populations exhibited more internal gene diversity (H_e= 0.20, 0.19) than did the Texan populations (H_e= 0.08, 0.06), although the number of RAPD bands in Texas (n= 62) was only marginally smaller than in Mexico (n= 68). F-statistics for the allozyme data, averaged over loci, revealed strong regional differentiation (mean F_RT=+ 0.30), as well as some differentiation among populations within regions (mean F_PR=+ 0.09). In order to describe and compare the partitioning of genetic variation for multiple allozyme and RAPD loci, we performed an Analysis of Molecular Variance (AMOVA). AMOVA for both allozyme and RAPD data revealed similar qualitative patterns: large regional differences and smaller (but significant) population differences within regions. RAPDs revealed greater variation among regions (58.4% of total variance) than allozymes (45.2%), but less variation among individuals within populations (31.9% for RAPDs vs. 45.2% for allozymes); the proportion of genetic variance among populations within regions was similar (9.7% for RAPDs vs. 9.6% for allozymes). Despite this large-scale concordance of allozyme and RAPD variation patterns, multiple correlation Mantel techniques revealed that the correlations were low on an individual by individual basis. Our findings of strong regional differences among the diploid races will facilitate further study of polyploid evolution in buffalograss.     

Microsatellite DNA markers for two endemic ground beetles: Carabus punctatoauratus and C. solieri

We isolated and characterized nine and five polymorphic microsatellite loci in the respective ground beetles Carabus punctatoauratus and C. solieri. We tested cross‐species amplification of all these loci plus six isolated in C. solieri and for which primers sequences were soon published. From these combined analyses, we obtained 14 and 17 polymorphic markers, respectively, for C. punctatoauratus and C. solieri.     

Microsatellite polymorphism and genetic differentiation in three Norwegian populations of Elymus alaskanus (Poaceae)

 Variation at seven microsatellite loci was investigated in three local E. alaskanus populations from Norway and microsatellite variation was compared with allozyme variation. The percentage of polymorphic loci was 81%, the mean number of alleles per polymorphic locus was 5.7 and expected heterozygosity was 0.37. An F-statistic analysis revealed an overall 48% deficit of heterozygotes over Hardy-Weinberg expectations. Gene diversity is mainly explained by the within population component. The averaged between population differentiation coefficient, F _st , over 7 loci is only 0.13, which accounts for only 13% of the whole diversity and was contrary to allozyme analysis. The mean genetic distance between populations was 0.12. However, a χ² -test showed that allele frequencies were different (p < 0.05) among the populations at 5 of the 7 loci. In comparison with the genetic variation detected by allozymes, microsatellite loci showed higher levels of genetic variation. Microsatellite analysis revealed that population H10576 possesses the lowest genetic variation among the tested three populations, which concur with allozyme analysis. The dendrogram generated by microsatellites agreed very well with allozymic data. Our results suggest that natural selection may be an important factor in shaping the genetic diversity in these three local E. alaskanus populations. Possible explanations for deficit heterozygosity and incongruence between microsatellites and allozymes are discussed. Received November 6, 2001; accepted April 24, 2002 Published online: November 14, 2002 Addresses of the authors: Genlou Sun (e-mail: Genlou.sun@STMARYS.CA), Biology Department, Saint Mary's University, Halifax. Nova Scotia, B3H 3C3, Canada. B. Salomon, R. von Bothmer, Department of Crop Science, The Swedish University of Agricultural Sciences, P.O. Box 44, SE-230 53, Alnarp, Sweden.     

A comparison of genetic structuring of Yellowcheek Darters (Etheostoma moorei) using AFLPs and allozymes

The goals of this study were to characterize the genetic structure of 6 populations of Etheostoma moorei (Yellowcheek Darter), endemic to the Little Red River watershed of central Arkansas, to estimate the levels of gene flow within isolated streams, and to compare AFLP genetic diversity and distance data to our previously published allozyme data. The Yellowcheek Darter is a candidate species for listing under the Endangered Species Act. This darter is found in previously connected headwater streams presently isolated and partially inundated downstream by Greers Ferry Reservoir. AFLP data for the Yellowcheek Darter was concordant with previous work utilizing allozymes (r_s = 0.682; p < 0.01), yet genetic differences among populations were greater in magnitude. Genetic diversity (polymorphism = 92.7; heterozygosity = 0.496) is higher for the Yellowcheek Darter than would be expected for a species in decline, and greater for AFLP versus allozyme data. Genetic structuring among streams was also more evident using AFLP data. Gene flow levels are indicative of a metapopulation structure within streams (F_ST = 0.003 − 0.010), with genetic structuring indicating distinct populations among streams.     

Amplified fragment length polymorphism (AFLP) analysis of the genetic structure of the zebra mussel, Dreissena polymorpha, in the Mississippi River

1. We predicted that zebra mussel, Dreissena polymorpha (Pallas), genetic structure in the Mississippi River would follow a model of invasive species genetics, which predicts low genetic structure among populations of recently established species. This prediction was upheld in our previous genetic study using allozymes, however, one locus yielded anomalous results. 2. We employed amplified fragment length polymorphism (AFLP) analysis as a neutral marker to assess the amount of genetic structure within and among populations, and as a test of expected population structure from both invasion genetic theory, and the results from our previous study. 3. There was greater spatial differentiation, as measured by F_st, observed using AFLP's than for allozymes (P < 0.001). There was no evidence that AFLP variation conformed to an isolation by distance model, and genetic relationships of populations, as measured by AFLP markers, were not similar to those detected in our allozyme survey. 4. The lack of concordance between these two genetic marker systems probably reflects their differential responses to drift, migration, and selection occurring during this rapid invasion. Strong population structure is counter to predictions that populations of invasive species will not be differentiated, as with observations based on allozyme markers. Therefore, newly established species may require genetic surveys using multiple marker systems to evaluate population structure.     

The structure of morphometric and allozyme variation in relict populations of Gypsophila fastigiata (Caryophyllaceae) in Sweden

Patterns of variation and the structuring of diversity in seed phenotype and allozymes were investigated in Sweden Gypsophila fastigiata, a perennial herb with a disjunct ‘Late Glacial relict’ distribution in eastern and northern Europe. The overall patterns of variation in allozymes and seed morphology are congruent and are significantly correlated with geographic distance. However, most of the congruence between the distance matrices based on allozymes and seed morphology is attributable to regional differentiation between the isolated Öland and Dalarna metapopulations. On a local scale, within the two metapopulations, seed shape variation is partially correlated with geography whereas allozyme variation is unrelated to the geographic disposition of the sampling sites. Despite the fact that regional metapopulations can be discriminated on the basis of seed shape and allozyme frequencies, relatively little of the total diversity in seed morphology and allozymes is due to differentiation between regions. Partitioning of diversity into its within- and among-regional, site and individual components, showed that the majority of diversity (52% for seed shape and 91% for allozymes) is stored within sites. Seventy-three percent of the variation in allozymes is due to variation within individuals (i.e. heterozygosity) whereas phenotypic variation in seed shape within individuals is low (0.4% of the total diversity). The events that led to the large scale disjunction of the Late Glacial distribution of G. fastigiata may have shaped the pattern of differentiation observed among the regional isolates of the species. However, the species' history of local population disjunction during post-glacial and historic times has not had a substantial impact on the spatial organization of allozyme and morphometric diversity within regions. Extensive local gene flow may have allowed the regional metapopulations to have functioned as extended effective populations–on a time scale of tens or hundreds of years–and may have hindered the loss of within-site diversity.     

Heterozygosity and effective size in laboratory populations ofAphidius ervi [Hym.: Aphidiidae]

Decline of allozyme variability in 7 laboratory populations ofAphidius ervi Haliday demonstrates that the effective population size is approximately one half the number of individuals used to renew the cultures each generation. Unequal reproductive contribution by individual females to subsequent generations is the most likely cause for this phenomenon.

---

Résumé La diminution de la variabilité des allozymes dans 7 populations de laboratoire deAphidius ervi Haliday montre que l'importance des effectifs des populations correspond approximativement à la moitié du nombre d'individus utilisés pour renouveler l'élevage à chaque génération. Une participation inégale à la reproduction des individus femelles aux générations successives est la cause la plus probable de ce phénomène.

     

EFFECTS OF POSTGLACIAL RANGE EXPANSION ON ALLOZYME AND QUANTITATIVE GENETIC VARIATION OF THE PITCHER-PLANT MOSQUITO,WYEOMYIA SMITHII

We determined allozyme variability of 34 populations of the pitcher-plant mosquito, Wyeomyia smithii, from Florida (30°N) to northern Manitoba (54°N) and compared allozyme variability with the additive genetic variance for preadult development time and photoperiodic response determined previously for six populations over a similar range (30–50°N). Phylogenetic analysis of allozymes shows a well-defined split between Gulf Coast and lowland North Carolina populations, similar to previously observed phylogeographic patterns in a wide variety of taxa. A deeper split in the phylogeny of W. smithii coincides with the location of the maximum extent of the Laurentide Ice Sheet. Furthermore, both average heterozygosity and patterns of isolation-by-distance decline in populations north of the former glacial border. It is likely that northern populations are the result of a range expansion that occurred subsequent to the late-Wisconsin retreat of the Laurentide Ice Sheet and that these populations have not yet reached a drift-migration equilibrium. The northern decline in allozyme heterozygosity contrasts sharply with the northern increase in additive genetic variance of development time and photoperiodic response found in previous studies. These previous studies also showed that the genetic divergence of populations has involved stochastic variation in the contribution of dominance and epistasis to the genetic architecture underlying demographic traits, including preadult development time, and photoperiodic response. When taken together, the present and prior studies identify the genetic processes underlying the lack of concordance between geographic patterns of allozyme and quantitative genetic variation in natural populations of W. smithii. In the presence of nonadditive genetic variation, isolation and drift can result in opposite patterns of genetic variation for structural genes and quantitative traits.     

Studies of esterase 6 in Drosophila melanogaster. XVIII. Biochemical differences between the slow and fast allozymes

Most natural populations of Drosophila melanogaster are polymorphic for two major electrophoretic variants at the esterase-6 locus. The frequency of the EST 6F allozyme is greatest in populations in warmer latitudes, whereas the EST 6S allozyme is predominant in colder latitudes. Latitudinal clines in electromorph frequencies are found on three continents. Purified preparations of the allozymes have been characterized for their pH optimum, substrate specificity, organophosphate inhibition, alcohol activation, thermal stability, and kinetic parameters. These and previous analyses of the EST 6 allozymes reveal that the two variants have differences in their physical and kinetic properties that may provide a basis for the selective maintenance of the polymorphisms and an explanation of the clinal variation observed in natural populations.      

Genetic variation within and among North Atlantic and Baltic populations of the benthic alga Phycodrys rubens (Rhodophyta)

Genetic variation was examined within and among North Atlantic, North Sea and Baltic populations of the benthic red alga Phycodrys rubens using allozymes and random amplified polymorphic DNA (RAPD) markers. On western and eastern North Atlantic coasts distinct allozyme types were found, with the exception of western Newfoundland where East and West Atlantic types co-occur. Along the European coasts, two genetic groups were distinguished by fixed allelic differences: an outer oceanic group and a North Sea/Baltic group. The two genetic types co-occur in the Skagerrak and Kattegat region. Reproductive isolation between the two types is suggested by the lack of hybrids in the overlap zones, and they may therefore represent sibling species. Unexpectedly, an analysis of RAPD variation was unable to recover the two cryptic species identified using allozymes. Within-population RAPD variation was similar to or greater than between-population variation. The lack of structure in the RAPD data cannot be attributed solely to technical artefacts of the method but appears to reflect real biological variability. Within-population genomic polymorphisms caused by frequent mutational events are discussed, as are high amounts of genetic drift and possible disruptive selection brought about by stressed habitats. Finally, Baltic and extra-Baltic salinity ecotypes are known to exist in P. rubens. However, no correlation between ecotypic variation and allozyme groups was detected.     

ALLOZYME VARIATION AMONG POPULATIONS OF BIOMPHALARIA CAMERUNENSIS (BOETTGER, 1941) (GASTROPODA: PLANORBIDAE) IN CAMEROON

Allozyme electrophoresis was used to study genetic diversityamong populations of the freshwater, hermaphroditic snail Biomphalariacamerunensis in Cameroon. Five of 19 loci studied in 15 enzymesystems were polymorphic. Intrapopulation variation was evidentin 8 of 12 populations sampled and heterozygotes were presentin 6 of these. Two populations with large sampled sets werepanmictic. Although results for the remaining 4 populationswith heterozygotes present were inconclusive, our findings supportthe hypothesis that Biomphalaria are out-breeders. The allelesfor two loci, asparate aminotransferase-1 (AAT-1) and 6-phosphogluconatedehydrogenase (6-PGD), showed distribution patterns that suggestthey may be limited by climate or habitat. The slow migratingAAT-1 was the only allele found in the mountainous regions ofthe southwest. This region falls under the Cameroon-type climateand is the wettest region in the country. The faster allelewas restricted to the lowland rain forest of the south. For6-PGD the fast allele was found throughout the range of B. camerunensis,but the slower allele was found only in the mountainous regionswithin the Cameroon-type climate. (Received 27 April 1989; accepted 9 September 1989)     

Geographic variability in expression of the sex-linked AAT-1 gene of the bell-ring frog, Buergeria buergeri

Cytological observation and artificial crossing experiments were used to examine the geographic differences in the sex-determining mechanism and mode of inheritance of the sex-linked AAT-1 gene in the bell-ring frog, Buergeria buergeri. The AAT-1 phenotypes were also examined by allozyme analysis using field-caught females and males collected from 19 populations from the Honshu, Shikoku, and Kyushu islands of Japan, in order to comprehensively elucidate the geographic variability in the expression of the sex-linked AAT-1 gene of B. buergeri. The results showed that the Aomori population of B. buergeri from the northern end of Honshu was female heterogametic in sex determination, that chromosome No. VII was a sex chromosome of the ZZ/ZW type, and that the sex-linked AAT-1 gene was expressed on both the Z and W chromosomes. This mode of AAT-1 expression in the Aomori population was different from that in the Hiroshima population from western Honshu, in which the AAT-1 gene was expressed on the Z chromosome but not on the W chromosome. The results also showed that there was no differentiation among populations in the expression of the AAT-1 genes on the Z chromosome, whereas two populations, the Hiroshima and Aomori frogs, exhibited distinct modes of expression of the AAT-1 gene on the W chromosome. These two modes of expression may be widely distributed in western and eastern Japan, and coexist in the central part of Honshu.     

Analysis of allozyme variability in populations of three species of brush-haired mice of species Lophuromys (Rodentia, Muridae) from the Bale Mountains National Park in Ethiopia

Allozyme variability was examined in populations of three endemic species of the species complex Lophuromys flavopunctatus sensu lato: L. chrysopus, L. brevicaudus, and L. melanonyx. These species substitute each other in adjacent latitudinal belts of the Bale Massif in Ethiopia. A deficit of heterozygotes at several loci was found in most samples of all species studied. Moreover, the samples included animals homozygous for two or three minor alleles and heterozygous for alleles that are rare and unique for the given species. It is suggested that the Bale Massif are inhabited by numerous genetically isolated populations of each Lophuromys species, which exchange genes at an extremely low rate. Genotypic disequilibrium observed in most samples is explained by the fact that most sampling localities comprise ranges of two and more micropopulations. In our view, microgeographic subdivision of the populations is caused by recurrent fragmentation of habitats during the Pleistocene glaciation of the Bale Massif and subsequent prolonged isolation of local populations. Gene drift accompanying these processes resulted in high genetic differentiation of the local populations, which probably persisted until the present. Geographical isolation of the Bale Massif, its uniquely diverse ecological conditions, and extraordinary allozyme structure of the Lophuromys populations suggest that these populations represent remnants or direct descendants of relic local populations.     

ASSOCIATIONS BETWEEN ALLOZYME FREQUENCIES AND SOIL CHARACTERISTICS IN GAILLARDIA PULCHELLA (COMPOSITAE)

Selection favoring different alleles in different environments frequently has been suggested as an explanation for allozyme variation within and among populations. This hypothesis predicts that allozyme frequencies will be correlated with environmental variables. Previous studies on allozyme frequency-environment covariation in plants often have relied on qualitative assessments of the environment and have emphasized highly autogamous species. We have examined allozyme frequency-soil associations in Gaillardia pulchella, an obligately outcrossed annual plant, by regressing the frequencies of 15 common allozymes representing six polymorphic enzyme loci on principal components from a set of 20 quantitative soil variables. Fifty-one populations, representing four taxonomic varieties, were included in the analysis. Among the 26 populations representing the var. pulchella, allozymes Adh-2f and Pgm-1c were significantly associated with a block of highly inter-correlated soil characteristics which serve to discriminate between soils derived from calcareous vs. non-calcareous rock types. This geographically complex pattern of allozyme frequency-soil covariation is not likely to be spurious and, thus, indicates the presence of adaptively differentiated soil races, or ecotypes. However, these results are not sufficient to conclude that the allozyme frequency divergence between ecotypes was mediated by selection, either directly or through genetic hitchhiking. The pattern of allozyme frequency-soil covariation within var. pulchella was not found among the other taxonomic varieties. Patterns of genotype-environment covariation often may be recognizable only within geographically or environmentally restricted groups of populations because of the confounding influences of other environmental variables.     

Genetic variation and its evolutionary implications in a Mediterranean island endemic lizard

The peculiar bioclimatic and geographic features of Corso–Sardinian islands may provide an ideal scenario for investigating microevolutionary processes, given their large heterogeneity of environments, which could affect dispersal and gene flow among populations, as well as processes of local adaptation. The genetic variation and differentiation among populations of the endemic lizard Archaeolacerta bedriagae were studied by allozyme electrophoresis at 20 presumptive loci. The genetic structure of this species is characterized by relatively high levels of polymorphism and low differentiation among populations. The pattern of genetic differentiation cannot be explained by genetic drift as a function of geographic distance. Genetic distance data show that genetic variation is distributed into three geographically coherent population groups and suggest a recent (Late Pleistocene) origin for the observed geographic fragmentation. The analysis of environmental correlates of allozymic variation indicates a strong correlation of the Idh-1 locus with climatic variables. The frequency of the Idh-1¹⁰⁶ allele is negatively correlated with annual temperature, and positively correlated with annual precipitation. In addition, the observed heterozygosity at this locus decreases towards more arid climatic regimes. The results obtained support the assumption of differential selection acting on Idh-1 allozymes under diverse climates. An association between Idh-1 allozymes and local bioclimatic regimes was also observed for the sympatric lizard Podarcis tiliguerta , suggesting a key role for such selective agents on Idh-1 polymorphism in these two Corso–Sardinian lacertids.  © 2009 The Linnean Society of London, Biological Journal of the Linnean Society , 2009, 98 , 661–676.     

The genetic effective size of a metapopulation

The structure of a population over time, space and categories of social and sexual role governs its ability to retain genetic variation in the face of drift. A metapopulation is an extreme form of spatial structure in which loosely coupled local populations 'turnover', that is, suffer extinction followed by recolonization from elsewhere within the metapopulation. These local populations turn over with a characteristic half-life. Based on a simulation model that incorporates both realistic features of population ecology and population genetics, the ability of such a metapopulation to retain genetic variation, which may be defined as proportional to its so-called effective population size, denoted N_e(^meta), can be one to two orders of magnitude lower than the maximum total number of individuals in the system. N_e(meta) depends on the persistence time associated with longevity of local populations (the turnover half-life), the average number of local populations extant in the metapopulation and the gene flow between local populations. Habitat fragmentation, which can create a metapopulation from a formerly continuously distributed species, may have unappreciated large genetic consequences for species impacted by human development.     

Parallel microgeographic patterns of genetic diversity and divergence revealed by allozyme, RAPD, and microsatellites in Triticum dicoccoides at Ammiad, Israel

The levels of genetic diversity were compared by means of 35 allozyme, 60 RAPD, and 25 microsatellite (SSR) markers for 75–175 individuals of tetraploid wild emmer wheat (Triticum dicoccoides) collected in 1993 from a microgeographic microsite, Ammiad, north of the Sea of Galilee, Israel. This microsite included four major habitats, which showed highly significant differentiation in ecological factors, in particular with respect to rock cover, proximity and height, and surface soil moisture in the early growing season of T. dicoccoides. Higher within-subpopulation genetic diversity was found in the primarily non-coding DNA regions (RAPD and SSR) rather than in the protein-coding (allozymes) regions. However, much larger gene differentiation (G _ST) among the subpopulations was observed in the protein-coding allozymes than in the RAPDs and SSRs. Larger genetic distance was found at SSR loci, followed by allozyme and RAPD loci. The subpopulations in drier habitats tend to have higher allozyme, RAPD and SSR diversities (He), the relatively wet Karst subpopulation showed only about half He of the other relatively drier habitats. The subpopulations with larger difference of soil moisture between habitats tend to show larger genetic distances at allozyme, RAPD and SSR loci. These results suggest that climatic selection through aridity stress may be an important factor acting on both structural protein-coding and presumably partly regulatory non-coding DNA regions, resulting in microscale adaptive patterns, although hitchhiking and random drift may also intervene. These results have profound implications for genetic conservation both in situ and ex situ.