Source Populations of Quercus glauca in the Last Glacial Age in Taiwan Revealed by Nuclear Microsatellite Markers

Source Populations of Quercus glauca in

the Last Glacial Age in Taiwan Revealed

by Nuclear Microsatellite Markers

Y

-J

L

, S

-Y

H

, K

-C

H

,

T

-P

L

From the Institute of Plant Biology, National Taiwan University, 1 Roosevelt Road, Section 4, Taipei 106, Taiwan (Lee and Lin); the Graduate Institute of Biotechnology, Chinese Culture University, Yangmingshan, Taipei 116, Taiwan (Hwang); and the Department of Life Sciences, National Taiwan University, 1a Roosevelt Road, Section 4, Taipei 106, Taiwan (Ho). Address correspondence to T.-P. Lin at the address above, or e-mail: tpl@ntu.edu.tw.

Abstract

In this work, we attempted to study genetic differentiation between populations of Quercus glauca in Taiwan using nuclear microsatellite markers to infer the potential refugium in the last glaciation stage. Four microsatellite loci for 20 individuals each in 10 populations of Taiwan were analyzed. We found that Q. glauca has relatively high within-population diversity (HE5

0.741) and low population differentiation (FST5 0.042) but shows isolation by distance. The most divergent populations,

according to the average FSTfor individual populations in comparison with every other population, were found in populations

Cy, Sa, and Hy in southern Taiwan and Pa in north-central Taiwan. Moreover, populations Cy, Sa, and Pa were recognized as being the source populations for gene recolonization after the last glaciation stage. In addition, the three sites of Wu, Ym, and Cy exhibited the highest gene diversities that coincided with populations with the highest chloroplast DNA variations. This may have resulted from an admixture of colonization routes. In conclusion, observations of the most divergent populations and source populations suggest that southern and probably north-central Taiwan may have potentially been refugia for Q. glaucain the last glaciation. This agrees with the possible refugium in southern Taiwan revealed by a previous study using chloroplast DNA markers.

The distribution of genetic variation within a species is shaped by processes such as ecological events, selective forces, pat-terns of genetic exchange, mating systems, and patpat-terns of historical relationships (Schaal et al. 1998). The levels of ge-netic diversity, gene flow, gege-netic differentiation, and post-glacial colonization in subtropical trees are of central interest to subtropical conservation biologists. In recent years, there has been an increase in the number of studies looking not only for present-day population genetic structure but also for the history of plant migration, isolation, and divergence among population lineages since the last glaciation in Taiwan (Cheng et al. 2005; Chiang et al. 2004; Huang et al. 2002, 2004; Hwang et al. 2003). Although limited knowledge has been obtained from these studies, some preliminary results point to a common evolutionary history for various plant species. For example, isozyme studies have revealed that the highest level of genetic diversity was found in Nantou County in central Taiwan, which was a major diversity center harboring the greatest genetic variation for many plant spe-cies (Lin 2001) in several studies including Cunninghamia konishii (Lin et al. 1998), Trochodendron aralioides (Wu et al. 2001), Alnus formosana (Sue et al. 2000), Taiwania cryptomerioides

(Lin et al. 1993), and Castanopsis carlesii (Cheng YP, Hwang SY, and Lin TP, unpublished data).

Lying between the Asian continent and Philippine Sea ba-sin, the island of Taiwan is commonly regarded geographi-cally as a constituent of the island-arc system along the western edge of the Pacific Ocean. Geologically, this island connects the Ryukyu Island-arc to the north and the Philip-pine Island-arc to the south. The Central Mountain Range (CMR), the lofty backbone range runs basically following the axis of the island (see Figure 1, the area above altitude of 2000 m). Within this range are numerous peaks above 3000 m in elevation (Editorial Committee of the Flora of Taiwan, 2nd ed 1994). The pollen record from a lake core at an altitude of 745.5 m from Jih-Yueh Tan (23
49#E, 120
53#N) in central Taiwan during the last glacial maximum suggested that the temperature during the maximal last glacial stage was 8.0
C–11.0
C cooler than today. Lowland vegetation was dominated by cool-temperate forest com-posed of conifers mixed with deciduous hardwood species (Tsukada 1966, 1967). The evidence of paleontology indi-cated a once-downward migration of tree species during gla-cial maximum from higher altitude. Events of recolonization

doi:10.1093/jhered/esj030

Advance Access publication April 13, 2006

of higher altitude after glacial maximum are possible. There-fore, Quercus glauca (or Cyclobalanopsis glauca, Fagaceae), a warm-temperate species, could be scarce or extinct in cen-tral Taiwan at that time. Migration of Q. glauca to southern Taiwan is expected.

Q. glaucais an evergreen broadleaf tree found not only in Taiwan but also in China, Korea, and Japan. Q. glauca is the most commonly occurring plant among the 50 native species of the family Fagaceae in Taiwan. Q. glauca is a medium-sized evergreen tree growing up to 20 m high and 0.6 m in diameter breast height. Machilus-Castanopsis is one of the two major types of evergreen broadleaf forests in Taiwan occurring at 400–1500 m. A large number of tree species appear in these forests, the most prominent among them belonging to the families Fagaceae and Lauraceae. Q. glauca is found from sea level up to 1700 m and is one of the dominant spe-cies of Machilus-Castanopsis forests (Hsieh et al. 1994). No

pure stand has been found, but populations may form aggre-gates on steep hillsides as one of the pioneer species. With respect to the usage of Q. glauca, aboriginal people have used it in construction, in mushroom culture, and for making household items, such as wheels, various poles, etc. The only man-made plantations were established in mixed forests in selected sites in order to build fire-deterrent belts in areas with pine forests and a long dry season. Q. glauca is selected as study material because of its island-wide distribution and representative as a dominant subtropical tree species; in ad-dition, the methodology and primers for the exploitation of Quercusspp. are available and have been tested in numerous European oak species. Moreover, the population genetics of Q. glauca using nuclear markers have never been studied in Taiwan. Using chloroplast DNA, Huang et al. (2002) ex-amined the phylogeography of Q. glauca. Together with the highest haplotype and nucleotide diversities, they suggested Figure 1. Map of Taiwan showing the locations of sampling sites. For population abbreviations see Table 1. Elevations of 100, 1000, 2000, and 3000 m are shown. CMR, Central Mountain Range; HR, Hsueshan Range.

the possibility that the southeastern part of Taiwan could have been a refugium during the last glaciation. A starlike genealogy was characterized in Q. glauca, as a general outcome of population expansion from a bottleneck.

The aims of this work were (1) to estimate the genetic diversity within and between populations of Q. glauca in Tai-wan and to test, in particular, whether central TaiTai-wan is the area with the highest genetic diversity and (2) to examine pat-terns of genetic divergence for each population in order to infer the possible refugia in the last glaciation.

Materials and Methods

Plant Material

Samples of Q. glauca consisting of mature leaves were col-lected from an average of 20 nonadjacent trees on 10 sites in Taiwan (Table 1, Figure 1). The population sites mostly overlapped with the populations used in a previous study (Huang et al. 2002), but the individuals were not the same. Leaves were enclosed in a sealed polyethylene bag and carried to the laboratory. After cleaning with water and drying on a bench top, leaves were conserved at
30
C prior to DNA extraction. In some cases, leaves were dried in silica gel in the sealed polyethylene bag.

DNA Extraction

Total DNA was extracted using the cetyltrimethyl ammonium bromide procedure (Murray and Thompson 1980) from 500 mg of fresh or dried leaves ground by hand. The quantity of DNA was measured by a spectrophotometer (Beckman Coulter, Fullerton, CA). DNA was diluted to 10 ng/ll.

Nuclear Microsatellite Primer Design and Analysis Four polymorphic nuclear microsatellites (Table 2) were selected among the nine that were originally designed for Quercus myrsiniflolia (Isagi and Shuhadono 1997). Polymerase chain reaction (PCR) products of these primers were se-quenced to find the simple sequence repeat (SSR) for each locus. We redesigned the primer pair for fluorescence labeling, and the primer sequences are presented in Table 2. PCR ampli-fications were performed in 10-ll reactions containing 1  PCR buffer [10 mM Tris-HCl (pH 8.3), 50 mM KCl, 100 lM of each dNTP, 0.02% Triton X-100, and 0.01% gelatin], 15 mM MgCl2, 0.25 U Taq polymerase, 0.2 lM of

each primer, and 5–10 ng of template DNA. The PCR con-ditions were 3 min at 94
C and 30 cycles of 30 s at 94
C, 30 s at a primer-specific annealing temperature (50
C), and 30 s at 72
C, with a final extension at 72
C for 7 min. The

Table 1. Genetic variation in individual Quercus glauca populations or in each group of population analyzed using microsatellite loci. AR, allele richness; H, genetic diversity; FIS, Wright’s inbreeding coefficient; and S, private allele. The Geo-group refers to population

groupings according to geographic location

Population Location Sample size

No. of alleles

per locus S AR H FIS

1. Wushe (Wu) 121.08
E, 24.03
N 20 7.25 1 6.42 0.765 0.215**

2. Paling (Pa) 121.23
E, 24.38
N 20 6.75 0 5.82 0.721 0.299**

3. Yuli (Yu) 121.15
E, 23.19
N 19 6 0 5.45 0.747 0.236**

4. Chilan (Ch) 121.33
E, 24.38
N 20 6.75 0 5.58 0.703 0.287**

5. Taroko (Ta) 121.37
E, 24.11
N 20 6.75 0 5.70 0.729 0.282**

6. Shanping (Sa) 120.40
E, 23.00
N 20 7 1 6.24 0.739 0.443**

7. Chinshuiying (Cy) 120.43
E, 22.25
N 20 6 0 5.54 0.758 0.492**

8. Hungyeh (Hy) 121.03
E, 22.53
N 20 6.5 2 5.77 0.733 0.422**

9. Wulai (Wl) 121.33
E, 24.53
N 20 5.75 0 5.08 0.682 0.392**

10. Yangmingshan (Ym) 121.33
E, 25.08
N 12 6 1 6.00 0.756 0.367**

Geo-group

A (1, 2, 4) North-central 60 8.75 — 7.78 0.731 0.122**

B (5, 3, 8) Eastern 59 8.25 — 7.68 0.741 0.132**

C (6, 7) South-western 40 8 — 7.65 0.753 0.286**

D (9, 10) Northern 32 7.25 — 7.25 0.704 0.179**

* Significant at the 5% level, ** Significant at the 1% level.

Table 2. Core sequence for each microsatellite locus and primer sequence. The annealing temperatures for all were 50
C

Locus Core sequence Primer sequence

Mic57 (CCA)n 5#-GCTAAGATTTATCGCAGCCATAGG-3#

5#-TGAGGAGGTTGGTGGAGAAAA-3#

Mic67 (CCA)n þ (CCG)n þ (CCT)n 5#-TGGCTTATCCAATGTTTGTGATT-3#

5#-CGGCTTAGAGATTGGTGTCAAAG-3#

Mic51 (CA)n þ (TA)n 5#-CAAAAACCTAAACCTACAAACGCTAAA-3#

5#-AATAGCAAGAGAGAAGATGTTGCAAC-3#

Mic69 (TGG)n þ (TTG)(TGG)2CGG(TGG)2 5#-CACAATCTGCCCACATCATC -3#

PCR products were separated using a 3100 Genetic Ana-lyzer with GeneScan software (Applied Biosystems, Foster City, CA). About 40% of total samples (72 of 181 individ-uals) were selected to examine the variation in the flanking and core sequences. Three to five clones were sequenced for each individual by a cloning technique (Promega pGEM-T Easy Vector System I, Madison, WI).

Data Analysis

For genetic diversity analysis, number of alleles per locus, al-lelic richness (AR), observed heterozygosity (HO), expected

heterozygosity (HE), genetic diversity (H) (Nei 1987), and

in-breeding coefficient (FIS) were included. These statistics were

calculated for each locus, each population, and for each group of populations. Genetic differentiation, FST, was tested

using FSTAT V. 2.9.3. (Goudet 1995). The levels of signif-icance were adjusted for multiple tests according to the Bonferroni criteria. We computed Wright’s inbreeding coef-ficient (FIS) with FSTAT and tested the deviation of the

ge-notypic frequencies from Hardy-Weinberg proportions with the program GENEPOP v. 3.4 (Raymond and Rousset 1995) to check whether inbreeding had occurred. A permutation procedure (1,000 times) was employed to test the significance of the FISdeviation from a random mating system.

Compu-tation of allelic richness for specified sample sizes was based on the rarefaction method developed by Hurlbert (1971). Al-lelic richness corresponds to the number of different alleles found in the specified sample size. It is sampled at the locus in question and was measured as described by El Mousadik and Petit (1996). Briefly, the rarefaction method is used to estimate allelic richness at a locus for a fixed sample size.

The spatial genetic structure was investigated by testing for a pattern of isolation by distance. A Mantel test (Mantel 1967) with 1,000 random permutations was performed be-tween the matrix of pairwise genetic differentiations bebe-tween populations, using FST/(1
FST) and the matrix of the

nat-ural logarithm of the geographic distance. These analyses were performed with PASSAGE (Rosenberg 2001) on the entire data set. The conventional genetic distance, FST,

according to allelic frequency for population subdivision was estimated using the ARLEQUIN program (Schneider et al. 2000). In addition, the level of divergence for each pop-ulation from the remaining poppop-ulations was calculated as mean values of pairwise FST for each population against

the rest of the populations. To identify the source population in the last glaciation, a partial Mantel test was carried out us-ing the software zt (Bonnet and Van de Peer 2002). The con-cept of source population is derived from the study of fulmar (Burg et al. 2003; Fisher 1966). Fisher (1966) envisaged a ‘‘stepping stone’’ model of expansion of fulmar, relying on the assumption that a new colony was founded by indi-viduals from the nearest existing colony. F statistics were then used to examine the levels of population differentiation and to indirectly obtain rough approximations for the levels of gene flow between sites.

Population relationships were analyzed by drawing dendrograms with Nei’s standard genetic distance (Tamura

and Nei 1983) and Dml (Goldstein et al. 1995). These dendrograms were drawn using either unweighted pair-group method using arithmetic averages (UPGMA) or the Neighbor-joining (NJ) method (Saitou and Nei 1987) and were constructed using POPULATIONS version 1.2.28 software (http://www.cnrsgif.fr/pge/bioinfo/populations/). Data were bootstrapped 2,000 times to estimate the node sig-nificance of the tree.

Results

Microsatellite Variations

Traditionally, length variations of microsatellites are the most conspicuous and usually the sole criterion employed to char-acterize allelic diversity as they display variable numbers of tandem repeats. Fine structural analyses of variations among microsatellite alleles have shown unexpected complexities in length variations. Genetic diversity in the microsatellite region of many species has been found not to be solely due to var-iations in the tandem repeat motif. Nucleotide substitutions and indels in the region flanking the repeat motif also contrib-uted to the electrophoretic size variation detected on sequence gels (England et al. 2002; Garza et al. 1995; Grimaldi and Crouau-Roy 1997; Guillermo et al. 1997; Matsuoka et al. 2002; Orti et al. 1997). Thus, microsatellite alleles of identical size are not necessarily identical within species.

We analyzed the flanking region and found substitutions and indels. Four of 144 alleles (in 72 individuals) showed var-iations in length. One allele has a 2-bp deletion in the flanking region (181 bp) of a SSR from the 5# end of locus Mic69 in the 19 individuals of the Yuli population. In addition to this, we found that one sample has an additional TTG in the core sequence of Mic69 of the Yuli population. In the case of flanking sequence variations, only the length of the core se-quence was considered. Providing for 2.8% length variations in flanking regions of 72 individuals (40% of the total sam-ples), less than 2% of the total alleles may have been wrongly identified after subtracting the four alleles mentioned above. Microsatellite Diversity within Populations

If the population was grouped using geographical approxi-mations, it could be divided into group A (Wu, Pa, and Ch), group B (Yu, Ta, and Hy), group C (Sa and Cy), and group D (Wl and Ym) (Table 1). North-central group (A), including populations Wu, Pa, and Ch, is located within the Hsueshan Range (HR). Populations Yu, Ta, and Hy (B) are located on the eastern side of the CMR, while popula-tions Sa and Cy on the southwestern side of the CMR. Group C of southern Taiwan harbored the highest genetic diversity but not the greatest number of alleles or allelic richness. The fixation indices statistically deviated from zero at the group level due to Wahlunds’ effect.

Averaged over the four loci, the number of alleles per locus was 5.75–7.25 at the population level and 7.25–8.75 at the group level (Table 1). Allelic richness (AR) was

the Wu population and a minimum value of 5.08 in the Wl population. Coincidently, the Wu population harbored the highest H value (0.765), while population Wl exhibited the lowest value (0.682). The fixation index, a measure of hetero-zygotic deficiency, statistically deviated from zero in all pop-ulations. Only a few private alleles, the alleles unique to one population, were found, and the south populations, Hy and Sa, have 50% of total private alleles.

The four SSR loci investigated in the present study were polymorphic in all populations. The number of alleles ob-served per locus in samples of 191 individuals ranged from 8 to 17, with an overall total of 45 alleles scored over the four loci (Table 3). Among the four loci, Mic67 was the most poly-morphic. The size ranges of PCR products corresponding to these alleles were roughly between 145 and 268 nucleotides (Table 3). Overall genetic diversities (HE) were similar for

each locus and ranged from 0.643 to 0.820. The average genetic diversity (HE, 0.741) was higher than HO (0.612).

This was also reflected in the positive FISvalues indicating

the existence of more homozygotes in the population than expected under Hardy-Weinberg equilibrium (Table 3). A low variation in FST and only 4.2% of the total variation

existed among populations.

Genetic Differentiation and Isolation by Distance

Genetic differentiation between populations was low as re-vealed by SSR markers, and FSTvaried from 0.011 to 0.041

(data not shown). The unexpectedly high level of gene flow as revealed by the low FSTvalue was found between Pa and

Wu, Wl and Ym, Yu and Ch, and Yu and Ta. Interestingly, through the calculation of the average FSTfor individual

pop-ulations in comparison with every other population, we found that Cy was genetically the most distinct population, followed by Sa and Hy (Figure 2), all situated in the south. The Pa population in the north-central area had a value comparable to Hy.

A Mantel test for SSRs also showed a significant corre-lation between popucorre-lation differentiations measured as FST/

(1
FST), and this significantly increased with the natural

logarithm of the geographical distance between populations (r 5 0.5652; P 5 .007 with 1,000 permutations). To test the correlation between geographical distances and genetic dis-tances for a specific population against the remaining

pop-ulations, the r value (correlation coefficient) was used to find the source population (Burg et al. 2003). We have obtained R2 values from calculations of the simple Mantel test generated by zt, a software tool for performing simple and partial Mantel tests (Bonnet and Van de Peer 2002). According to the R2value for each population, all the regres-sions were significant. We found that four populations, i.e., Pa, Cy, Sa, and Wl, had values 0.8 (Table 4) and can be considered source populations for gene recolonization after the last glaciation stage. Populations Pa and Wl are located in the northern portion of the CMR, and Sa and Cy are in the south.

Genetic Relationships of Populations

The population NJ trees according to Nei’s D (Tamura and Nei 1983) and Goldstein’s Dml (1995) are shown in Figure 3a and b, respectively. The first major cluster included populations Hy, Sa, and Cy. The second cluster included populations Ch and Yu. The third included Wl, Ym, Wu, Pa, and Ta.

Table 3. Allelic diversity of microsatellite loci scored in Quercus glauca. Size distribution (nt), range of sizes of PCR products; HO,

observed heterozygosity; HE, expected heterozygosity; FIS,Wright’s inbreeding coefficient; and FST, relative differentiation according to

allele identity

Locus Size distribution (nt) No. of alleles No. of heterozygotesa HO HE FIS FST

Mic67 220–262 17 115 0.604 0.643 0.028* 0.033*

Mic57 145–166 9 109 0.572 0.767 0.221** 0.042*

Mic51 250–268 8 88 0.461 0.737 0.332** 0.063**

Mic69 221–251 11 154 0.811 0.820
0.019 0.030*

Mean — — 116.5 0.612 0.741 0.140 0.042

a_{The number of plants showing heterozygotic alleles in a total number of 191 individuals.}

* Significant at the 5% level, ** Significant at the 1% level.

Figure 2. Plot of the average pairwise FSTvalues for each

population in comparison with the remaining populations against latitude in Quercus glauca. Cy (22.25
N, population no. 7) was genetically the most distinct population, followed by Sa (23.00
N, no. 6) and Hy (22.53
N, no. 8), all located in southern Taiwan. For population abbreviations see Table 1.

These two dendrograms, however, are not the same in terms of population grouping, and no high bootstrap values exist. Some alleles present in both eastern and western popu-lations of the CMR can be observed in Table 5. For example, in locus Mic57, allele 223 is present in populations Yu, Wl, and Ym; allele 226 is present in populations Sa and Hy; and allele 232 is present in populations Pa, Ta, and Cy. Similar conditions could be observed for the three other loci (data not shown). This indicated that the CMR is not a barrier to gene flow among populations. However, this pattern also could be due to convergence because of the high mutation rates of the microsatellite loci.

Discussion

Null Allele Might Cause the Deviation of Fixation Index FIS for each population significantly deviated from zero

(Table 1). Possible causes for deviation from Hardy-Weinberg equilibrium may be the locus being under selection, presence of null alleles in the level of inbreeding (mating among rel-atives), and the presence of population substructure leading to Wahlunds’ effect. Although the vast majority of microsat-ellites are believed to be neutral, linkage of those markers to selected loci is present. In addition, strong selective pressure may cause the fixation of alleles in different populations. However, this is unclear in the loci studied in this paper. In-breeding may be common in the population. If inIn-breeding is frequent in Q. glauca, differentiation among population will be strengthened. On the contrary, very low differentiation was observed between populations (Table 3). Inbreeding should have the same effect on every locus tested. This is also not supported by our data.

Null alleles may be present which are leading to a false observation of excess homozygotes. Locus Mic51 and fol-lowed by Mic57 but not the other two loci have a strong FIS, suggesting that putative null alleles might occur at these

two loci. It could be short allele dominance as a source of heterozygote deficiency as has been reported previously (Wattier et al. 1998).

Genetic Relationships of Populations

These two dendrograms of NJ trees based on Nei’s D and Goldstein’s Dml (Figure 3) do not agree completely with the geographic groupings in Table 1. Only groups of populations Sa and Cy in the south and Wl and Ym in the north were closely related geographically. The UPGMA method also generated different groupings of populations except popula-tions Sa and Cy that were always grouped together (data not shown). This indicated that the south group was slightly iso-lated, and the rest of the populations have gene interchange from different directions. Nuclear gene interchanges have oc-curred between populations despite the presence of the CMR as low genetic differentiation (FST) was observed for Q. glauca

(Figure 2). The pollen-seed flow ratio was estimated for the species of Quercus to be from 190 to 500 (Squirrell et al. 2001). In consequence, the dendrograms were not highly supported, and spatial genetic pattern was less clear than that revealed by chloroplastic DNA markers (Huang et al. 2002).

Table 4. Isolation by distance correlations for 10 populations, calculated separately based on the genetic distance (FST) and

geographical distances between a given population and all other populations. The R2values were derived from calculations by simple Mantel test generated by zt, a software tool for simple and partial Mantel tests (Bonnet and Van de Peer 2002)

Population R2value P value

Wu 0.7263 ,.001 Pa 0.8603 ,.001 Yu 0.3879 .002 Ch 0.7662 ,.001 Ta 0.7104 ,.001 Sa 0.8374 ,.001 Cy 0.7965 .002 Hy 0.6092 ,.001 Wl 0.9444 ,.001 Ym 0.7712 ,.001

Figure 3. Unrooted NJ tree of the 10 Quercus glauca populations drawn using (a) Nei’s D (Tamura and Nei 1983) and (b) Goldstein’s Dml(Goldstein et al. 1995) by NJ

methods. The numbers in the figures are percentage values over 2,000 bootstrap replicates. Only bootstrap values over 50% are presented.

The NJ method does not assume a constant rate of evo-lution (Nei 1987). Longer branches in the NJ tree suggested by populations Cy, Sa, Hy, and Ym might indicate a faster evolutionary rate, higher differentiation, and possibly popu-lation fragmentation.

Potential Refugia of C. glauca Predicted from Microsatellite Analysis

The degree of the average FSTvalue of each population in

comparison with the remaining populations can be used to examine the consequence of historical and contemporary geographical population subdivisions on evolutionary pro-cesses (Johnson et al. 2000) and is important for reconstruct-ing the phylogeographical history that has evolved durreconstruct-ing pre- and postcolonization events (Grant and Grant 1997). In the European oak [ Q. petraea (Matt.) Liebl.], the highest differentiation values were found among the most ancient populations, close to refugia (Zanetto and Kremer 1995). It was found that in the common ivy (Hedera sp.) in Europe, differentiation of each population from the remaining ones revealed a latitudinal pattern, with populations from the south being significantly more differentiated from the pooled remaining populations than the central or northern popula-tions (Grivet and Petit 2002). Thus, population divergence or genetic differentiation can be a useful criterion for locating a region of glacial refugium. Petit et al. (2003) tested the hy-pothesis that glacial refuge areas harbor a large fraction of intraspecific diversity. They concluded that plant populations in refuge areas have high genetic divergence and uniqueness rather than a high number of haplotypes.

Considering the existence of unrelated haplotypes and the highest nucleotide diversities, Huang et al. (2002) proposed that the southeastern part of Taiwan (in the region around the Taimali, Hy, Cy, and Laiye populations) was a potential re-fugium for Q. glauca in the last glaciation. This was further

supported by this study as we found that the populations in southern Taiwan (populations Sa, Cy, and Hy) were the most genetically divergent among the populations tested (Figure 2). The south probably provided a warmer environ-ment for Q. glauca to survive in the last glaciation stage. Un-fortunately, no climatic/historic or pollen record data are available to support this observation. In southern Taiwan, group C (populations Sa and Cy) was found to have relatively high genetic diversity among the groups of populations (Table 1). A second refugium might have been located in the north-central part (Paling population) to the west of the CMR because a peak FSTvalue was observed in this area

(Figure 2). This was not mentioned in the chloroplast DNA study (Huang et al. 2002).

Moreover, according to R2of the Mantel analysis for each population, four putative source populations for gene flow, i.e., Cy, Sa, Pa, and Wl, were identified which coincide with the refuge populations mentioned above except population Wl. Wl, in fact, has the lowest allele per locus, AR, and H. The

two dendrograms do not support this conclusion either. Wl also had a low genetic divergence when chloroplastic DNA data (Huang et al. 2002) was calculated as the average FSTof

each population in comparison with the remaining (data not shown). The high R2of Wl at the present time is unclear. The dendrograms (Figure 3b) imply that colonization from the source population Pa in north-central Taiwan to the west of the CMR might have proceeded to population Wu and northward, even to Ta population, the east of the CMR, while source populations from the south might have proceeded to populations Yu and Ch along the east side of the CMR.

Among the 10 populations in this study, Wu, Cy, and Ym had the highest genetic diversity, and they were found to have high allelic richness. These three same populations coincided with the ones with the highest genetic diversity in terms of cpDNA sequences (Huang et al. 2002). In Wu and probably Ym, high gene diversity might have resulted from

Table 5. Representatives of allele-size distributions for locus Mic57 in each population of Quercus glauca

Locus Mic57 allele size Population Wu Pa Yu Ch Ta Sp Cy Hy Wl Ym 220 0 0 0 0 0 0.025 0 0 0 0 223 0 0 0.053 0 0 0 0 0 0.05 0.083 226 0 0 0 0 0 0.075 0 0.05 0 0 229 0.15 0.05 0.026 0.125 0.075 0.05 0.3 0.15 0.025 0.083 231 0.025 0.05 0 0 0 0 0 0.1 0 0 232 0 0.05 0 0 0.025 0 0.025 0 0 0 235 0 0.025 0.026 0 0.025 0.025 0.05 0 0.025 0.042 238 0.325 0.25 0.421 0.55 0.55 0.5 0.35 0.325 0.525 0.375 241 0.2 0.175 0.105 0.05 0.025 0.075 0.025 0.075 0.075 0 244 0.075 0.275 0.289 0.2 0.2 0.125 0.1 0.175 0.15 0.292 247 0.1 0.05 0.053 0.025 0.025 0.1 0.1 0.025 0.1 0 250 0.025 0.025 0.026 0.025 0.075 0 0.05 0.05 0 0.042 253 0.1 0 0 0 0 0.025 0 0 0 0 256 0 0.05 0 0 0 0 0 0 0 0.042 259 0 0 0 0.025 0 0 0 0 0.05 0 261 0 0 0 0 0 0 0 0.05 0 0 262 0 0 0 0 0 0 0 0 0 0.042

admixtures of colonization after the retreat of the last gla-ciation because Wu and Ym had the lowest average genetic differentiation in comparison with the remaining popula-tions (Figure 2). Wu and Ym also have a low average pair-wise FSTaccording to cpDNA markers. Thus, microsatillite

data are consistent with the conclusions from cpDNA markers. In fact, Wu is located in Nantou County, central Taiwan, which is considered to be the major diversity center for several forest tree species (Lin 2001).

In conclusion, in this paper we investigated nuclear mi-crosatellite variations of Q. glauca in Taiwan. We found that the southern part of Taiwan showed the highest popula-tion genetic divergence compared to all the others, which suggests the existence of a potential major refugium. This is in agreement with a previous report according to chloro-plastic markers (Huang et al. 2002). Even though a much higher gene flow was revealed by nuclear markers among populations, this study implies that a correlation still exists between maternal lineage (cpDNA) and nuclear marker var-iations of Q. glauca in Taiwan.

Acknowledgments

This investigation was funded by grants (NSC92-2313-B-002-117 and NSC92-2621-B-002-010) from the National Science Council, Executive Yuan, Taiwan.

References

Bonnet E and Van de Peer Y, 2002. zt: a software tool for simple and partial Mantel tests. J Stat Softw 7:1–12.

Burg TM, Lomax J, Almond R, Brooke MdeL, and Amos W, 2003. Un-ravelling dispersal patterns in an expanding population of a highly mobile seabird, the northern fulmar (Fulmarus glacialis). Proc R Soc Lond B Biol Sci 270:979–984.

Cheng YP, Hwang SY, and Lin TP, 2005. Potential refugia in Taiwan revealed by the phylogeographical study of Castanopsis carlesii Hayata (Fagaceae). Mol Evol 14:2075–2085.

Chiang TY, Hung KH, Hsu TW, and Wu WL, 2004. Lineage sorting and phylogeography in Lithocarpus formosanus and L. dodonaeifolius (Fagaceae) from Taiwan. Ann Mo Bot Gard 91:207–222.

Editorial Committee of the Flora of Taiwan, 2nd ed, 1994. Flora of Taiwan, vol 1. Taipei, Taiwan: Editorial Committee of the Flora of Taiwan, 2nd ed. El Mousadik A and Petit RJ, 1996. High level of genetic differentiation for allelic richness among populations of the argan tree [Argania spinosa (L.) Skeels] endemic to Morocco. Theor Appl Genet 92:832–839.

England PR, Usher AV, Whelan RJ, and Ayre DJ, 2002. Microsatellite diversity and genetic structure of fragmented populations of the rare, fire-dependent shrub Grevillea macleayana. Mol Ecol 11:967–977.

Fisher J, 1966. The fulmar population of Britain and Ireland, 1959. Bird Study 13:5–45.

Garza JC, Slakin M, and Freimer NB, 1995. Microsatellite allele frequencies in humans and chimpanzees, with implications for constraints on allele size. Mol Biol Evol 12:594–603.

Goldstein DB, Linares AR, and Cavalli-Sforza LL, 1995. An evaluation of genetic distances for use with microsatellite loci. Genetics 139:463–471. Goudet J, 1995. FSTAT (version 1.2): a computer program to calculate F-statistics. J Hered 86:485–486.

Grant PR and Grant BR, 1997. Genetics and origin of bird species. Proc Natl Acad Sci USA 94:7768–7775.

Grimaldi MC and Crouau-Roy B, 1997. Microsatellite allelic homoplasy due to variable flanking sequences. J Mol Evol 44:336–340.

Grivet D and Petit RJ, 2002. Phylogeography of the common ivy (Hedera sp.) in Europe: genetic differentiation through space and time. Mol Ecol 11:1351–1362.

Guillermo O, Devon EP, and Avise JC, 1997. Phylogenetic assessment of length variation at a microsatellite locus. Proc Natl Acad Sci USA 94:10745–10749.

Hsieh CF, Shen CF, and Yang KC, 1994. Introduction to the flora of Taiwan, 3: floristics, phytogeography, and vegetation. In: Flora of Taiwan, 2nd ed, vol 1. (Editorial Committee of the Flora of Taiwan, ed). Taipei, Taiwan: Editorial Committee of the Flora of Taiwan; 7–18.

Huang SF, Hwang SY, and Lin TP, 2002. Spatial pattern of chloroplast DNA variation of Cyclobalanopsis glauca in Taiwan and eastern Asia. Mol Ecol 11:2349–2458.

Huang SF, Hwang SY, Wang JC, Tung WF, and Lin TP, 2004. Phylogeog-raphy of Trochodendron aralioides (Trochodendraceae) in Taiwan and its adja-cent areas. J Biogeogr 31:1251–1259.

Hurlbert SH, 1971. The nonconcept of species diversity: a critique and alternative parameters. Ecology 52:577–586.

Hwang SY, Lin TP, Ma CS, Lin CL, Chung JD, and Yang JC, 2003. Post-glacial population growth of Cunninghamia konishii (Cupressaceae) inferred from phylogeographical and mismatch analysis of chloroplast DNA varia-tion. Mol Ecol 12:2689–2695.

Isagi Y and Shuhadono S, 1997. PCR primers amplifying microsatellite loci of Quercus myrsinifolia Blume and their conservation between oak species. Mol Ecol 6:897–899

Johnson KP, Adler FR, and Cherry JL, 2000. Genetic and phylogenetic consequences of island biogeography. Evolution 54:387–396.

Lin TP, 2001. Allozyme variations in the Michelia formosana (Kanehira) Masa-mune (Magnoliaceae), and the inference of a glacial refugium in Taiwan. Theor Appl Genet 102:450–457.

Lin TP, Lu CS, Chung YL, and Yang JC, 1993. Allozyme variation in four populations of Taiwania cryptomerioides in Taiwan. Silvae Genet 42:278–284. Lin TP, Wang CT, and Yang JC, 1998. Comparison of genetic diversity between Cunninghamia konishii and C. lanceolata. J Hered 89:370–373. Mantel N, 1967. The detection of disease clustering and a generalized regres-sion approach. Cancer Res 27:209–220.

Matsuoka Y, Mitchell SE, Kresovich S, Goodman M, and Doebley J, 2002. Microsatellites in Zea—variability, patterns of mutations, and use for evolu-tionary studies. Theor Appl Genet 104:436–450.

Murray MG and Thompson WF, 1980. Rapid isolation of high molecular weight plant DNA. Nucleic Acids Res 8:4321–4325.

Nei M, 1987. Molecular evolutionary genetics. New York: Columbia Univer-sity Press.

Orti G, Pearse DE, and Avise JC, 1997. Phylogenetic assessment of length variation at a microsatellite locus. Proc Natl Acad Sci USA 94: 10745–10749.

Petit RJ, Aguinagalde I, de Beaulieu JL, Bittkau C, Brewer S, Cheddadi R, Ennos R, Fineschi S, Grivet D, Lascoux M, Mohanty A, Muller-Starck G, Demesure-Musch B, Palme A, Martin JP, Rendell S, and Vendramin GG, 2003. Glacial refugia: hotspots but not melting pots of genetic diversity. Science 300:1563–1565.

Raymond M and Rousset F, 1995. An exact test for population differenti-ation. Evolution 21:259–274.

Rosenberg MS, 2001. PASSAGE. Pattern analysis, spatial statistics and geo-graphic exegesis. Version 1.0. Tempe, AZ: Department of Biology, Arizona State University.

Saitou N and Nei M, 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 4:406–425.

Schaal BA, Hayworth DA, Olsen KM, Rauscher JT, and Smith WA, 1998. Phylogeographic studies in plants: problems and aspects. Mol Ecol 7: 465–474.

Schneider S, Roessli D, and Excoffier L, 2000. Arlequin ver. 2.000: a software for population genetic data analysis. Geneva, Switzerland: Genetics and Biometry Laboratory, University of Geneva.

Squirrell J, Hollingsworth PM, Bateman RM, Dickson JH, Light MH, MacConaill M, and Tebbitt MC, 2001. Partitioning and diversity of nuclear and organelle markers in native and introduced populations of Epipactis helleborine(Orchidaceae). Am J Bot 88:1409–1418.

Sue CY, Fang K, and Huang S, 2000. Patterns of genetic variation of Alnus formosanain Taiwan. Taiwania 45:95–106.

Tamura K and Nei N, 1983. Estimation of the number of nucleotide substitutions in the control region of mitochondrial DNA in humans and chimpanzees. Mol Biol Evol 10:512–526.

Tsukada M, 1966. Late Pleistocene vegetation and climate in Taiwan (Formosa). Proc Natl Acad Sci USA 55:543–548.

Tsukada M, 1967. Vegetation in subtropical Formosa during the Pleistocene glaciation and the Holocene. Palaeogeogr Palaeoclimatol Palaeoecol 3:49–64. Wattier R, Engel CR, Saumitou-Laprade P, and Valero M, 1998. Short allele dominance as a source of heterozygote deficiency at microsatellite loci: ex-perimental evidence at the dinucleotide locus Gv1CT in Gracilaria gracilis (Rhodophyta). Mol Ecol 7:1569–1573.

Wu E, Huang S, Wang JC, and Tong WF, 2001. Allozyme variation and the genetic structure of populations of Trochodendron aralioides, a monotypic and narrow geographic genus. J Plant Res 114:45–57.

Zanetto A and Kremer A, 1995. Geographical structure of gene diversity in Quercus petraea(Matt.) Liebl. I. Monolocus patterns of variation. Heredity 75:506–517.

Received April 27, 2005 Revised October 24, 2005