Copyright 0 1997 by the Genetics Society of America

Microsatellite Polymorphism in Natural Populations of the Wild Plant Arabidqsis thuliana

Hideki Innan,*’t Ryohei Terauchi:.’ and Naohiko T. Miyashita*

*Laboratory of Plant Genetics, Graduate School of Agriculture, Kyoto University, Sakyo-ku, Kyoto 606-01, Japan, tDepartment of Biological Sciences, Graduate School of Science, The University of Tokyo, Bunkyo-ku, Tokyo 113, Japan and

fLaboratory of Plant Systematics, Faculty of Science, Kyoto University, Sakyo-ku, Kyoto 606-01, Japan Manuscript received December 20, 1996 Accepted for publication April 14, 1997

ABSTRACT Variation in repeat number at 20 microsatellite loci of Arabidopsis thaliana w a s studied in a worldwide

sample of 42 ecotypes to investigate the pattern and level of polymorphism in repetitive sequences in natural plant populations. There is a substantial amount of variation at microsatellite loci despite the selfing nature of this plant species. The average gene diversity was 0.794 and the average number of alleles per locus was 10.6. The distribution of alleles was centered around the mean of repeat number at most loci, but could not be regarded as normal. There was a significantly positive correlation between the number of repeats and the amount of variation. For most loci, the observed number of alleles was between the expected values of the infinite allele and stepwise mutation models. The two models were rejected by the sign test. Linkage disequilibrium was detected in 12.1% of the pairwise comparisons between loci. In phylogenetic tree, there was no association between ecotype and geographic origin. This result is consistent with the recent expansion of A. thaliana throughout the world.

M ICROSATELLITES comprise a class of variable number of tandem repeats (VNTR) with a repeat unit of a few nucleotides (LITT and LUTY 1989). They are found in eukaryotic nuclear genomes (TAUTZ and RENZ 1984) and in the chloroplast genome of some plant species (POWELL 1995). The general function of VNTR loci is unclear, although their involvement in gene regulation, signaling for gene conversion, and re- combination has been suggested (WANG et al. 1979; SHEN et al. 1981).

It is known that VNTR loci in animals have a substan- tial level of repeat number polymorphism, which is manifested as a large number of alleles per locus (TAUTZ and RENZ 1984; EAIRD et al. 1986; CLARK 1987; CHAKRABORTY et al. 1991; DEKA et al. 1991; EDWARDS et al. 1992; VALDES et al. 1993; DI RIENZO et al. 1994; E% TOUP et al. 1995a,b; MICHALAKIS and VEUILLE 1996). This characteristic makes them useful as genetic mark- ers for genome mapping, paternity testing, individual identification, and so on UEFFREYS et al. 1985; NAKA- MURA et al. 1987; SILVER 1992). Accordingly, a number of VNTR loci have been characterized, especially in humans. However, the molecular mechanisms for pro- ducing new alleles are not completely understood. It has been suggested that replication slippage, sister chromatid exchange, unequal crossing over, and gene conversion may cause variations (DRAKE et al. 1983;

Corresponding author: Naohiko T. Miyashita, Laboratory of Plant Genetics, Graduate School of Agriculture, Kyoto University, Sakyo- ku, Kyoto 60601, Japan. E-mail: arabis@kais.kyoto-u.ac.jp

‘Present address: Biozentrum, Johann-Wolfgang-Goethe Universitat, Mane Curie Str. 9/N200, D-60439, Frankfurt, Germany.

Genetics 146: 1441-1452 (August, 1997)

TAUTZ and RENZ 1984). Among these mutational mech- anisms, replication slippage seems to play a major role in producing new alleles at VNTR loci (LEVINSON and GUTMAN 1987a; WOLFF et al. 1991). The relationship between repeat number and amount of variation is also unclear. In humans, a positive correlation between re- peat number and amount of variation was observed by WEBER (1990) and HUDSON et al. (1992), but VALDES et al. (1993) found no relationship between the repeat number and amount of variation.

To explain the pattern and level of polymorphism in VNTR, two mutation models have been examined: the infinite allele model (KIMURA and CROW 1964) and the stepwise mutation model (OHTA and KIMURA 1973; KI- MURA and OHTA 1978). These two models were origi- nally proposed to explain allozyme polymorphism. The infinite allele model assumes that each mutation pro- duces a new allele that did not exist in the population before, while the stepwise mutation model assumes that mutation changes the allelic state back and forth, and does not necessarily create a new allele. CLARK (1987) found an acceptable fit of the infinite allele model to human VNTR loci by analyzing the allelic frequency spectrum. EDWARDS et al. (1992) and VALDES et al. (1993) applied the two models to human microsatellite data and concluded that the stepwise mutation model explained the allelic frequency distributions better than did the infinite allele model. ESTOUP et al. (1995a,b) showed a better fit of the infinite allele model to micro- satellite polymorphism in the honeybee with respect to the number of alleles. DI RIENZO et al. (1994) intro- duced the two-phase mutation model as an extension

1442 H. Innan, R. Terauchi and N. T. Miyashita

of the stepwise mutation model, which combines one- step changes and rare large jumps in repeat number. They found a reasonable fit of this modified version of the stepwise mutation model to human data. However, a clear consensus has not been reached regarding the applicability of the theoretical mutation models.

In plants, VNTR (microsatellite) polymorphism is also abundant and has been studied mainly for agro- nomical purposes (see review of POWELL et al. 1996), especially to identify cultivars (grapevine: THOMAS and S C O ~ 1993; soybean: RONGWEN et al. 1995) and con- struct linkage maps for crop species (Brassica: LA- GERCRANTZ et al. 1993; soybean: AKKAYA et al. 1992; MOR- GANTE and OLMERI 1993; rice: ZHAO and KOCHERT 1993; KUN-SHENG and TANKSLEY 1993; maize: SENIOR and HEUN 1993). Recently, by taking advantage of the high level of variation at the microsatellite loci, the genetic structure (parentage and gene flow) of natural plant populations has been investigated (CONDIT and HUBBELL 1991; TERAUCHI 1994; TERAUCHI and KONUMA 1994; TODOKORO et al. 1995; VAN TREUREN et al. 1997). For Arabidopsis thaliana, BELL and ECKER (1994) charac- terized 30 microsatellite loci and detected size variation among laboratory strains. Using the microsatellite markers developed by BELL and ECKER (1994), TODO- KORO et al. (1995) studied the structure of Japanese natural populations of A. thaliana and showed that there is a large divergence among populations but no intrapopulational variation. These works on A. thaliana microsatellites did not provide information on popula- tion genetic parameters. To understand the evolution- ary history of A. thaliana and the maintenance mecha- nisms of genetic polymorphism at microsatellite loci, the pattern and level of polymorphism need to be stud- ied. Here, using the same marker system, a worldwide sample of A. thaliana was analyzed from a viewpoint of population genetics.

MATERIALS AND METHODS

Plant materials. Seeds of 31 ecotypes of A. thalialza were obtained from Professor N O B U M u GOTO, Arabidopsis Stock Center, Miyagi University of Education, Sendai, Japan (1 -31 in Table 1). Microsatellite data for 11 Japanese ecotypes (32- 42 in Table 1) were obtained from TODOKORO et al. (1995). A total of 42 ecotypes sampled worldwide were analyzed for 20 microsatellite loci.

DNA isolation and detection of microsatellite variation: Total DNA was extracted from mature plants by a modified CTAB method (TERAuCHI and KONUMA 1994) and used as a template for PCR amplification. The PCR primers for the microsatellite loci were obtained from Research Genetics Inc., Huntsville, Alabama, and had been developed by BELL and ECKER (1994). Twenty microsatellite loci investigated were selected at random. The motifs of investigated loci are shown in Table 2. All the loci have repeat units of 2 base pairs (bp) . The microsatellite loci studied in this report are located on four of the five chromosomes in A. thalzana (Table 2). The following procedure used to detect repeat number polymor- phism at microsatellite loci was essentially the same as in TO DOKORO et al. (1995). PCR amplification was carried out in a

10 pl volume containing 10 ng of template DNA, 1 PM of each of the primers, 0.15 mM of dNTP, 1 X reaction buffer, and 0.05 unit of Taq polymerase (AGS GmbH). Before the amplification, the 5' primer was labeled with [y3*P] dATP (Amersham) using T4 polynucleotide kinase (Bio Labs). Ki- nase reaction was conducted at 37" for 30 min with one unit of T4 polynucleotide kinase for 40 PM of primer. PCR ampli- fication consists of 25 cycles of a three-step process: 15 sec at 94", 15 sec at 55", and 30 sec at 72". Radioactive PCR products were denatured for 3 min at 90" and electrophoresed on 6% polyacrylamide sequencing gel for 3-4 hr. To determine the length of PCR product, the M13 sequence ladder was used as a size marker. After drying for 1 hr, the gel was exposed to an X-ray film for 3 days. From the autoradiogram, the length of PCR product was scored. The number of repeats for each allele was determined by comparing the size of the PCR product with that of the Col-0 ecotype whose repeat number was characterized by BELL and ECKER (1994). Esti- mated repeat number was used in the following analyses.

Estimation of expected values of gene diversity and number of alleles: An estimate of gene diversity (or expected hetero- zygosity, H) was calculated according to NEI (1973): n(1 - Cp') / ( n - I) , where n is the number of samples and p is the frequency of an allele. The expected number of alleles under the infinite allele model was estimated following EWENS (1972): 1 + M / ( M + 1) + M / ( M + 2) + * * * + M / ( M + n - l) , where n. is the number of samples and M = 4Nv (N, the effective population size; v, the mutation rate per locus). M was estimated as M = H/(1 - H ) , where H was the esti- mated gene diversity. The bias of overestimation was cor- rected by using the equation ( A 2 ) of CHAKRABORTY and GRIF- FITH (1982). Under the stepwise mutation model (one-step model), estimation of the expected number of alleles fol- lowed Equation 18 of WMURA and OHTA (1975).

RESULTS

The distribution of repeat number at microsatellite loci of A. thuliana: A total of 216 alleles were detected at 20 microsatellite loci in 42 ecotypes (Table 3). Each ecotype represents a single allele for a locus, since all the ecotypes showed only a single band for each pair of PCR primers: the investigated plants were all homozy- gous. When all the alleles were considered (Table 3), each ecotype had a distinct genotype (haplotype) indi- cating that microsatellites are useful for ecotype identi- fication even for the selfing plant A. thaliana.

Variation in repeat number at microsatellite loci of A. thaliana is summarized in Table 4. Although BELL and ECKER (1994) originally selected these 20 microsat- ellite loci based on the judgement that a locus was more than 20 nucleotides (10 repeats) long, the size of micro- satellite loci (the average number of repeats) varies greatly. An analysis of variance (ANOVA) indicated that the difference in size of the loci was statistically signifi- cant ( F = 21.3; d.f. = 19, 152; P < 0.0001). The loci nga126 and ngal72 have -30 repeats on average, while the locus nga280 has only 6.6 repeats. The effect of chromosome (1, 3 and 5) on locus size was tested, but no association was found between chromosome and size of the locus ( F = 1.36; d.f. = 2, 16; P > 0.2). In this study, four different motifs [(AT) n, (AG) n, (CT) n and (GA)n] were studied. However, motifs (AG) and (CT)

Microsatellites in A. thalzana 1443

TABLE 1

hbidopsis thalianu ecotypes analyzed

Name code Accession no." Origin Name code Accession no." Origin

1 Aa-0 2108 FRG

26 Pog-0 2951 Canada 5 B1-1 1440 Italy 25 MV-0 2916 USA 4 Ba-1 1441 UK 24 Mt-0 926 Libya 3 Al-0 1587 Denmark 23 La- 1 905 Poland 2 Ag-0 1660 France 22 Kn-0 2821 Lithuania

7 Bor-0 2175 Algeria 28 Yo4 2856 USA 8 Br-0 2912 Czecho 29 Tsu-0 958 Japan 9 Bs-1 2836 Switzerland 30 Hiroshima Japan

10 Bur4 1625 Ireland 31 Shokei Japan 11 Bus4 2176 Norway 32 Akita Japan 12 Can4 2850 Canary Island 33 Sakata Japan

6 Bla-1 0 1661 Spain 27 Ri-0 2844 Canada

13 Chi4 281 1 Russia

38 Fukui Japan 17 ES-0 1831 Finland 37 Ishikawa Japan 16 Col-0 2797 USA 36 Toyama Japan 15 co-1 2224 Portugal 35 Nigata Japan 14 Ci-0 2846 UK 34 Yamagata Japan

18 Gr-1 1717 USA

42 Fukuoka Japan 21 Kas-1 2136 India 41 Tokushima Japan 20 Ita-0 1659 Morocco 40 Kyoto Japan 19 Gre-0 2910 USA 39 Mie Japan

n = 42 species. "Accession number of the Sendai Arabzdopsis Seed Stock Center.

are regarded as the same. There was no relationship between motif and size of the locus ( F = 1.27; d.f. = 2, 17; P > 0.3). It is also of interest to see if ecotype has some effect on locus size, but an ANOVA showed that

TABLE 2 Microsatellite loci

Chromosome Locus Chromosome Repeat position"

ATHCHIB 3 (AT) n 36.3 ATEATl 1 (AT) n 72.2 ATHATPASE 1 (AG) n 80.6 ATHCTRl 5 (AG) n 43.8 nga 59 1 (CT)n 77.0 nga 63 1 (GA)n 68.2 nga 106 5 (GA) n 13.0 nga 111 1 (GA) n 74.5 nga 126 3 (AGh 47.1 nga 128 1 (AG) n 22.0 nga 129 5 (GAM 70.0 nga 151 5 ( C m 23.4 nga 158 5 (GA) n 27.9 nga 162 3 (GA) n 35.0 nga 168 2 ( W n - nga 172 3 (GA) n 58.8 nga 225 5 (CT)n 38.8 nga 248 1 (CT)n 33.0 nga 249 5 (CT)n 25.2 nga 280 1 (AG) n 22.0

n = 20 loci. a Distance from the centromere (in cM).

the effect of ecotype was not significant ( F = 0.77; d.f. = 41, 711; P > 0.80). This result implies that ecotypes cannot be discriminated on the basis of locus size. Taken together, these results suggest that locus size varies by chance or that the variation in locus size is caused by some factor(s) not studied here.

The distribution of alleles (repeat number) for each locus is shown in Figure 1. As locus size varies among loci, the variance in repeat number at each locus also varies greatly (2.04-179.98). A highly significant heter- ogeneity in the variance of repeat number was detected by the Bartlett test (x2d.f.,19 = 529.28; P < 0.001: SNEDE- COR and COCHRAN 1978, pp. 296-298). This variation in the variance of repeat number is not due to the chromosome on which the microsatellite locus is lo- cated. Heterogeneity of the weighted average of vari- ances was not detected among chromosomes (x2d.f.,2 = 1.16; P> 0.25). Although chromosome 3 has a relatively high variance, this is due to its locus ngu226. The type of motif does not seem to be the cause of this variation either. Although the motif AG (CT) seems to have a large variance, there was not significant heterogeneity among motifs ( x * ~ . ~ , , ~ = 0.62; P > 0.75). These results may suggest that the range of distribution at microsatel- lite loci vanes by chance.

To test whether the distribution of the number of repeats is normal, the skewness and kurtosis were cal- culated (Table 4). The obtained values indicated that none of distributions can be regarded as normal (skew- ness = 0 and kurtosis = 3) . It was difficult to find any

1444 H. Innan, R. Terauchi and N. T. Miyashita

TABLE: 3

S u m m a r y of the number of repeats at each locus for the 42 ecotypes

Locus

Aa-0 Ag-0 Al-0 Ba-0 B1-1 Bla-10 Bora Br-0

Bur-0

Can-0 Chi-0 Ci-0 co-1 Col-0 Es-0 Gr-1 Gre-0 I ta-0 Kas-1 Kn-0 La-1 M t-0 Mv-0

Ri-0 Ye0 TSU-0 Hiroshima Shokei Akita Sakata Yamagata Nigata Toyama Ishikawa Fukui Mie Kyoto Tokushima Fukuoka

Bs-1

BUS-0

Pog-0

13 7 10 11 22 13 19 NS NS 13 NS 23 18 12 17 29 13 NS 12 6 17 8 10 11 15 13 13 NS 43 21 13 32 13 14 17 31 16 24 12 6 12 6 10 10 14 NS 11 NS 38 17 11 27 14 11 18 35 20 16 12 6 13 7 10 10 20 13 12 17 27 17 14 15 15 14 18 32 18 22 12 6 14 7 10 12 18 13 12 NS 43 17 14 32 15 13 20 38 12 23 12 6 12 6 10 10 19 13 14 NS 43 15 14 24 11 9 18 28 NS 17 12 6 NS NS 10 10 NS NS NS NS NS NS 2 NS NS NS 18 35 NS NS 11 6 14 7 10 11 16 13 14 27 25 19 14 25 14 14 18 30 20 22 12 6 13 6 10 9 NS 13 14 14 44 17 NS 14 8 12 17 20 22 27 12 6 17 7 10 10 14 13 21 22 37 17 9 26 14 10 18 30 18 19 12 6 9 6 10 11 NS 12 14 26 41 13 11 NS 11 9 19 27 12 22 12 6 6 6 10 NS NS NS 20 NS NS NS NS NS 11 9 19 22 NS NS 12 6

12 6 10 9 14 12 12 27 31 21 13 NS 13 12 17 35 16 24 13 6 6 6 13 7 NS 12 13 27 16 NS 16 41 10 21 17 32 22 NS 13 7

10 6 10 NS NS 13 13 NS 42 15 NS 28 12 18 18 25 19 22 12 6 14 11 18 16 19 23 26 16 31 16 NS 31 13 21 25 29 19 24 15 15 16 6 10 9 20 13 12 22 NS 16 19 23 14 12 17 28 15 22 12 6 16 6 10 11 14 13 13 NS 42 16 14 26 NS 13 17 35 33 23 21 6 8 6 10 10 16 12 13 16 12 16 14 18 14 18 18 32 20 22 15 6

13 6 10 NS NS 12 12 28 NS 10 11 NS 10 9 21 29 15 17 12 7 14 6 14 9 23 12 12 NS NS 19 14 22 14 9 18 27 NS 24 12 6 12 6 10 10 22 13 9 13 43 16 NS 23 NS 11 18 22 17 26 12 11 NS 6 10 9 21 12 NS 23 NS 16 11 NS 11 15 16 20 15 20 12 6 12 6 10 12 14 13 21 22 29 17 9 26 15 9 19 21 NS 19 12 6 15 6 10 10 14 13 24 NS NS 11 NS 14 14 9 17 30 26 19 12 6 6 6 10 11 NS 11 13 16 NS 21 NS NS 14 12 18 35 28 NS 12 6

12 6 10 10 17 13 9 NS 43 16 NS 23 15 11 18 22 16 19 12 11 7 6 10 9 20 13 14 22 17 17 10 NS 15 13 17 32 14 20 12 6

16 7 10 10 21 13 16 20 30 11 NS 19 17 14 20 29 21 19 12 6 9 8 10 9 15 13 13 NS 10 11 10 18 10 14 17 36 14 17 14 7

15 6 10 10 NS 13 16 24 31 17 NS 14 9 10 17 30 21 26 12 6 14 9 10 NS 55 11 26 39 37 18 13 14 39 12 18 40 22 19 10 6 8 9 10 6 19 12 18 39 9 10 12 13 9 15 16 29 11 11 13 7 8 11 10 6 18 13 17 39 9 10 12 13 9 14 16 39 13 11 13 7 8 9 10 10 35 12 9 14 16 22 15 30 34 17 16 18 12 NS NS 6 7 12 10 10 19 13 10 35 NS 21 15 18 11 15 17 39 8 21 10 6 8 12 10 10 20 13 10 35 NS 20 14 18 11 15 17 39 8 21 10 6 8 12 10 10 20 12 10 35 9 20 15 18 11 14 17 40 8 21 10 6 8 10 10 9 20 12 17 35 9 10 12 13 9 13 16 38 13 11 13 7

15 9 10 8 19 11 9 24 NS 19 16 14 16 11 13 20 8 19 23 6 8 10 10 6 20 12 17 35 NS 10 12 12 9 13 16 37 11 NS NS 6 8 11 10 7 20 12 17 39 9 10 12 13 9 16 16 39 12 11 16 7

~~ ~

NS. not scored.

association between the shape of the distribution and chromosome or motif. The shape of the distribution also seems to vary by chance. The most common allele had the number of repeats near the median, although there were a few clear exceptions ( n g u l l l , ngul.26, and ngul51). The distribution was unimodal for most of the loci, while some ( n g u l l l , ngu1.26, ngul51, and ngul7.2) showed a clear deviation from unimodality.

Notable was the locus ngu126, which had peaks at both ends of the distribution, exhibiting a U-shape. Figure 1 also shows that the presence of outlier alleles was not common, except for loci ngu59, ngu63, and ngu158. Locus ngu59 had an allele with 55 repeats, much more than the other alleles of this locus, and the other two loci had a few alleles that deviated nota- bly from the median.

Microsatellites in A. thalzuna

TABLE 4

Summary of the amount of variations at 20 microsatellite loci

Locus Sample no. Average" Variance" Gene diversity Skewness Kurtosis

ATHCHIB 40 11.33 11.66 0.899 -0.023 1.667 ATEATl 41 7.54 4.21 0.709 1.048 2.655 ATHATPASE 42 10.36 2.04 0.139 4.333 21.838 ATHCTRl 38 9.68 3.36 0.782 0.422 5.747 nga 59 33 19.79 55.86 0.890 3.475 16.518 nga 63 39 12.77 3.24 0.579 4.805 28.365 nga 106 40 14.63 19.99 0.918 0.996 3.415 nga 111 29 25.90 72.95 0.926 0.186 1.788 nga 126 29 28.14 179.99 0.936 -0.300 1.526 nga 128 39 15.85 13.61 0.893 -0.291 2.039 nga 129 31 12.61 8.71 0.886 1.242 6.900 nga 151 34 21.18 51.24 0.936 0.646 2.886 nga 158 39 13.62 35.72 0.892 2.990 12.555 nga 162 41 13.00 9.45 0.900 0.753 3.407 nga 168 42 17.62 3.17 0.775 1.509 9.157 nga 172 42 30.57 39.96 0.944 -0.279 2.113 nga 225 37 16.43 32.42 0.952 0.707 3.531 nga 248 35 20.00 17.77 0.904 - 0.770 3.169 nga 249 40 12.70 6.22 0.624 2.740 11.168 nga 280 42 6.62 2.97 0.398 3.610 16.040

"Average and variance are represented by the number of repeat units.

1445

The ievel of polymorphism at microsatellite loci in A. thlianu: The level of polymorphism at microsatellite loci can be expressed in terms of the number of alleles and the gene diversity (the expected heterozygosity). The estimate of gene diversity for each locus is shown in Table 4. The average gene diversity over the 20 loci was 0.794 with a range of 0.139-0.952. The locus ATHATPASE exhibited the lowest gene diversity ( H = 0.139). The other loci were highly polymorphic ( H = 0.398-0.952). The average gene diversity of A. thaliana is comparable to estimates obtained for animal and plant species (DEKA et al. 1991; EDWARDS et al. 1992; VALDES et al. 1993; DI RIENZO et al. 1994; TERAUCHI and KONUMA 1994; ESTOUP et al. 1995a,b; TODOKORO et al. 1995). Despite the selfing nature of A. thaliana, this plant species contains a large amount of variation at microsatellite loci. An ANOVA indicated that there was no association between gene diversity and chromosome ( F = 2.34; d.f. = 2, 16; P > 0.10) or motif ( F = 0.54; d.f. = 2, 17; P > 0.50).

Table 5 lists the number of alleles at each locus. The average number of alleles per locus was 10.6 with a range of 4-17. Three loci (ATHATPASE, nga63, and nga280) had only four alleles each, but most of the other loci had more than 10 alleles. The loci on chro- mosome I had fewer alleles (7.9 on average) than those on chromosomes 3 and 5, and this difference was statis- tically significant ( F = 3.90; d.f. = 2, 16; P < 0.05). It is not clear why the loci on chromosome 1 had fewer alleles. BELL and ECKER (1994) reported the difference in the level of polymorphism among different repeat units (motif). In the present study, although (GA)n had

the highest number of alleles (1 1.3 on average), there was no clear difference in the number of alleles among motifs ( F = 0.23; d.f. = 2, 17; P > 0.80).

The expected number of alleles under the infinite allele and stepwise mutation models are shown in Table 5. For 12 of the 20 loci, the observed number of alleles was between the expected values of the two models. Gen- erally, the infinite allele model predicted a larger ex- pected number of alleles than did the stepwise mutation model, because the infinite allele model requires more alleles than the stepwise mutation model for a given gene diversity. Under the infinite allele model, 17 loci had fewer alleles than expected, while under the stepwise mutation model, 15 loci had more alleles than expected. To compare the two models, the sign test (ZAR 1974, pp. 290-291) was conducted. Both models were rejected at the 1% level (x2 = 11.84; d.f. = 1 for the infinite allele model, and x' = 8.00; d.f. = 1 for the stepwise mutation model). This result indicates that neither model explains the variation satisfactorily at microsatellite loci of A. thali- ana observed in this study.

Factors affecting the repeat number variation at mi- crosatellite loci: Variation at microsatellite loci is thought to be caused by several molecular mechanisms, mainly replication slippage and unequal crossing over. If replication slippage is important, a longer repeat would tend to have a larger variation, since the chance of replication mistakes is higher for a longer sequence ( LEVINSON and GUTMAN 1987a, b; WOLFF et al. 199 1 ) . To test this prediction, the relationship between amount of variation and locus size (average number of repeats) was studied. First, the relationship between average

1446 H. Innan. R. Terauchi and N. T. Miyashita

1 l! 8 1( - 0

z 0 5 0

E I t 0 b: 1(

0 5

- 0

z 0

v) 0

E 8 1c - 0

z 0 5

0

v) 0

E l5 8 1c .c 0

z 0 5

0

v) 0 E lE b: 1c - 0

z 0 5

0

ATHHCHIB

nga 59

5 10 15 20 25 30 35 40 45 50 55

No. of repeats

nga 106

0

0 5 z 0

5 10 15 20 25 30 35 40 45 50 55 v)

5 0 l5 8 10 nga 111

nga 126

z 0 5 0 I L I

5 10 15 20 25 30 35 40 45 50 55

8 1 5 1 b: 0 10 nga 128

No. of repeats

FIGURE 1.-Distribution of repeat number for each of 20 microsatellite loci in A. thaliana. The Xaxis indicates the number of repeats, and the Y axis the number of alleles.

number of repeats and number of alleles was investi- gated (Figure 2a). A positive and significant relation- ship was observed ( T = 0.74; P < 0.01). Figure 2b shows the relationship between average number of repeats and standard deviation of the repeat number. The cor- relation was again positive and significant ( r = 0.80; P < 0.01). However, when the relationship between coefficient of variation (standard deviation divided by the average) and locus size (average number of repeats) was tested, the correlation was still positive but became nonsignificant ( r = 0.30; P > 0.05), suggesting that variation per repeat is similar for different loci. This result implies that the heterogeneous variances over loci detected above is partly due to the increase in vari- ance associated with increased size. These results indi- cate that loci with more repeats tend to have a larger amount of variation in repeat number in A. thalzana and are consistent with the idea that replication slippage is

an important factor in generating variation at microsat- ellite loci in A. thaliana.

If unequal crossing over is important, it would be expected that a locus located in a region of high recom- bination rate would have a larger variation than a locus in a region of reduced recombination rate. As an at- tempt, the relationship between the amount of varia- tion and the chromosomal locations of microsatellite loci (centimorgans from the centromere) was investi- gated. No obvious relationship was detected between number of alleles and chromosomal position or be- tween standard deviation of repeat number and chro- mosomal position (data not shown). This result does not support unequal crossing over as a mechanism to generate variation in microsatellites; however, due to the small number of loci on each chromosome, these data are not conclusive.

Linkage disequilibrium between microsatellite loci:

Microsatellites in A. thaliana 1447

F lE v) a 0

z

8 1c

0 5

e

0

B lE v) 0 8 1c 3

z 0 5 0

v) g 15

8 1c 8 - 0

0 5 z 0

g 1: 8 8 1(

v)

e 0

z 0 5 0

v) E 0 If 8 1c 5

z 0 5 0

nga 129

15c. 5 10 15 20 25 30 35 40 45 50 55 nga 151

L 5 10 15 20 25 30 35 40 45 50 55 nga 158 c 5 10 15 20 25 30 35 40 45 50 55 nga 162

1 5 10 15 20 25 30 35 40 45 50 55

No. of repeats

B l5 H 10 nga 172 z 0 5

0 I 5 10 15 20 25 30 35 40 45 50 55

15

8 8 10 - 0

nga 225

z 0 5 0

5 10 15 20 25 30 35 40 45 50 55

nga 248 e 0

z 0 5 0

5 IO 15 20 25 30 35 40 45 50 55 v) g 15 24 s 8 10 nga 249 e 0 0 5 z

0 5 10 15 20 25 30 35 40 45 50 55

v) g 15 32 8 8 10

nga 280 u- 0

z 0 5 0 - I r

5 10 15 20 25 30 35 40 45 50 55

No. of repeats

FIGURE 1 .- Continued

Because A. thaliana is a selfing plant with a low rate of outcrossing (ABBOTT and GOMES 1989), its effective population size is expected to be smaller than the effec- tive population sizes of outcrossing plant and animal species. Even without any epistatic selection, a small effective size would result in linkage disequilibrium (HILL and ROBERTSON 1968; OHTA and KIMURA 1969). It is of interest from the viewpoint of population genet- ics to investigate how much linkage disequilibrium ex- ists between microsatellite loci in this selfing plant spe- cies. To test linkage disequilibrium using the present data, each allele at each locus was classified as common or non-common. Thus, nonrandom association in a simple two-locus/two-allele model was tested using a chi-square for most of the loci. For some loci, three alleles were considered, i.e., when two major alleles were equally common. There were 190 pair-wise compari- sons, of which 23 were significant at the 5% level. The percentage of significant pairs (12.1 %) was higher than that expected by chance (5%) and higher than that

obtained for honeybee microsatellites (6.4%: ESTOUP et al. 1995a). Many of the significant pairs were detected between chromosomes, especially between the chromo- some I and the other chromosomes, and two adjacent loci on a chromosome did not necessarily have nonran- dom association. By looking at genotype data for each locus, it was noted that Japanese ecotypes had distinct two-locus genotypes contributing to high chi-square val- ues. In other words, allelic frequency differences in Jap- anese A. thaliana may be the cause of the high propor- tion of significant linkage disequilibrium. The distinc- tiveness of Japanese ecotypes was also detected in the phylogenetic analysis described below. Although the ex- act genetic mechanism (epistatic selection or genetic drift) of significant nonrandom association cannot be determined from the present data, it is clear that A. thalzana contains a high level of linkage disequilibrium at microsatellite loci.

Phylogenetic relationship of A. thaliana based on microsatellite polymorphism: Microsatellite loci have

1448 H. Innan, R. Terauchi and N. T. Miyashita

TABLE 5 Observed and expected number of alleles

No. of alleles

Expected

Locus n Observed IAM SMM

ATHCHIB 40 11 16.7 11.0 ATEATl 41 7 8.8 5.0 ATHATPASE 42 4 1.8 1.4 ATHCTRl 38 8 10.5 6.2 nga 59 33 12 14.6 10.0 nga 63 39 4 6.2 3.8 nga 106 40 14 18.4 12.9 nga 111 29 13 16.1 12.7 nga 126 29 16 16.9 13.8 nga 128 39 11 16.0 10.5 nga 129 31 10 13.9 9.5 nga 151 34 17 18.7 14.6 nga 158 39 13 16.0 10.4 nga 162 41 11 17.0 11.1 nga 168 42 8 10.7 6.1 nga 172 42 17 22.3 17.0 nga 225 37 16 21.8 17.9 nga 248 35 11 16.0 11.2 nga 249 40 9 7.0 4.1 nga 280 42 4 4.0 2.8

IAM, infinite allele model; SMM, stepwise mutation model.

been successfully used to investigate phylogenetic rela- tionships and population structures of animal and plant species because of their high level of polymorphism (BOWCOCK et al. 1994; ESTOUP et al. 1995a; TODOKORO et al. 1995). It has been suggested that the present wide distribution of A. thaliana in the world was established only recently by the rapid migration from the original distribution area in the Himalayas (KING et al. 1993; PRICE et al. 1994; TODOKORO et al. 1995; INNAN et al. 1996). The investigated 42 ecotypes of A. thaliana were sampled worldwide. Therefore, our sample should pro- vide a good opportunity to investigate the evolutionary history of this plant species. To construct the neighbor- joining (NJ) tree, the average squared difference in repeat number was used. Squared difference in repeat number between a pair of ecotypes, divided by the vari- ance in repeat number, was obtained for each locus. The average over all the loci was used as the measure of distance for each pair of ecotypes (Figure 3) . This measure was shown to be linear with time under the stepwise mutation model (GOLDSTEIN et al. 1995). There is no clear association between ecotype and geo- graphic origin. Although there are several weak clus- ters, ecotypes in a cluster came from different parts of the world, except for the Japanese ecotypes. As shown in Figure 3, bootstrap probabilities for some ofJapanese ecotypes are only meaningful. This result is consistent with the idea that A. thaliana has spread over the world recently, as shown in previous studies. Most of Japanese

I . I/..

0 0 10 20 30

Average number of repeats

1 4 I

0 10 20 30 4 0 Average number of repeats

-I

FIGURE 2.-Correlation between repeat number and varia- tion in repeat number. (a) Relationship between average number of repeats and number of obsewed alleles. (b) Rela- tionship between average number of repeats and standard deviation of the number of repeats.

ecotypes formed one cluster, with several exceptional ecotypes (Nigata, Kyoto, Tsu-0, Akita, and Shokei). The clustering of Japanese ecotypes may suggest that this plant was introduced into Japan from the Western countries, and/or that the effective size of Japanese A. thaliana populations is small.

DISCUSSION

The pattern and level of microsatellite polymor- phism: This study demonstrated that microsatellite loci in A. thaliana are highly variable at the species level, despite selfing nature of this plant species (ABBOTT and GOMES 1989). First, microsatellite loci in A. thaliana were shown to vary in size (average number of repeats) considerably. Second, the variance of repeat number

Microsatellites in A. thaliana 1449

87 lshikawa 1 Yo-0 (USA)

Ci-0 (UK) Co-1 (Portugal)

Gre:O (USA). - La-1 (Poland)

BUS-0 (Norway) Ita-0 (Morocco)

Can-0 (Canary Island) 17 Mt-0 (Libya)

Akita

Bs-1 (Switzerland) 1 Shokei

Hiroshima

60 - Sakata 33 - 80

A Mie 6

m I

D

1

I - 25

r

Chi-0 (Russia) /- COI-0 (USA)

I AI-0 (Denmark) Ag-0 (France)

BI-1 (Italy) Br-0 (Czecho) I Es-0 (Finland) - Kas-1 (India) - Aa-0 (FRG)

Bur-0 (Ireland) Mv-0 (USA)

TSU-0 , Bla-10 (Spain)

FIGURE 3.-Neighbor-joining tree of A. thaliana ecotypes, showing bootstrap values (100 replicates). The distance measure was the squared difference in repeat number between a pair of ecotypes divided by the variance.

(the range of the distribution) and the shape of the distribution (skewness and kurtosis) varied among loci. The distribution was not regarded as normal. Third, the level of polymorphism was high. The average gene diversity (expected heterozygosity) was 0.794, and the average number of alleles per locus was 10.6. None of the measures of variation were associated with the chromosome of the locus or with motif, except that the

number of alleles on chromosome 1 was significantly smaller than on chromosomes 3 and 5. Although we may not be studying the most pertinent factors, this result may suggest that microsatellite loci vary in repeat number only by chance. To test this hypothesis, a ran- dom sample of microsatellite loci must be studied in the future.

The estimates of polymorphism (number of alleles

1450 H. Innan, R. Terauchi and N. T. Miyashita

per locus and gene diversity) at the species level of A. thaliana were comparable to those obtained in animal and plant species (CLARK 1987; EDWARD et al. 1992; DI RIENZO et al. 1994; ESTOUP et al. 1995a,b; TODOKORO et al. 1995; MICHALAKIS and VEUILLE 1996). On the other hand, an allozyme study of this plant in the United Kingdom indicated that local populations had an ex- tremely low level of heterozygosity (ABBOTT and GOMES 1989). By using the same microsatellite markers used here, TODOKORO et al. (1995) showed no variation within three local Japanese populations, but a great divergence among populations. These results suggest that selfing reduces the level of genetic variation within a population (a micro-habitat) . Although there are sev- eral reports that indicated the existence of genetic varia- tion with respect to life history characters in local popu- lation in the United Kingdom (WESTERMAN and LAW- RENCE 1970; JONES 1971a-c; WESTERMAN 1971a-c), these studies did not give estimate of variation at the genic level. An explanation for the high level of micro- satellite polymorphism at the species level (divergence among local populations) is the high mutation rate at microsatellite loci. A mutation rate of lo-* per locus per generation has been suggested for microsatellites (DALLAS 1992; WEBER and WONG 1993). High mutation rate and selfing (small effective size) would result in divergence among subpopulations and low polymor- phism within a population (CROW and & M U M 1970). However, if a local population is stable over a long time, a high mutation rate would also cause polymorphism within a local population, unless the effective size is very small. Therefore, it is necessary to consider the ecology of A. thaliana. This plant is regarded as a fugi- tive species, invading and occupying disturbed habitats (S. KAWANO, personal communication), and is thought to undergo rapid expansion and extinction. It is un- likely that a local population would be stable for a long time. A high mutation rate, selfing, and ecology of this plant species may explain the pattern of microsatellite polymorphism in natural populations of A. thaliana. In this study we could not examine population structure (distribution of microsatellite variation within and be- tween populations) of A. thaliana, because only one sample was used from a location. To test the high muta- tion rate hypothesis on the high level of microsatellite polymorphism at the species level, it is certainly neces- sary to study population structure of this plant species, especially level of polymorphism within local popula- tions, in the future.

Molecular mechanisms responsible for microsatellite variation: Although BELL and ECKER (1994) did not detect any clear correlation between the locus size (av- erage number of repeats) and number of alleles in six laboratory strains, this study showed that locus size is associated with the amount of variation in nature. Sig- nificant and positive relationships between average number repeats and number of alleles (Figure 2a) and

between average number of repeats and the standard deviation (Figure 2b) were detected. These results indi- cate that microsatellite loci with more repeats have greater variation, and agree with the idea that replica- tion slippage plays a major role in the generation of new alleles at microsatellite loci (LEVINSON and GUT- MAN 1987a; WOLFF et al. 1991). On the other hand, variation in repeat number may not be associated with recombination rate, based on the absence of a correla- tion between chromosomal location and level of varia- tion. This result is inconsistent with the idea that un- equal crossing over is an important mechanism in the generation of variation at microsatellite loci UEFFREYS et al. 1985). However, these results provide only circum- stantial evidence of molecular mechanisms. A more di- rect approach using a larger number of loci on each chromosome is necessary.

Evolutionary history of A. thuliana revealed by micro- satellite polymorphisms: In this study, investigated mi- crosatellite loci were located on four of the five chromo- somes in A. thaliana and were shown to be as variable as those in other investigated organisms. Therefore, it was expected that analysis of these microsatellites as markers covering most of the genome would reveal the evolutionary history of A. thaliana. Previous studies of restriction fragment length polymorphism (RFLP) (KING et al. 1993) and the Adh sequence (INNAN et al. 1996) in A. thaliana failed to show any concordant rela- tionship between ecotype and sampling locality, which could be due to the low sensitivity of RFLP analysis and the low variability of a specific DNA sequence. However, it was shown again in this report that there was no association between ecotype and geographic origin. Ex- cept for the Japanese ecotypes, most of the ecotypes sampled worldwide seemed to cluster without any rules. In the phylogenetic tree, the clusterings were not strong, as indicated by low bootstrap probabilities. From these results, it can be concluded that the present wide distribution of A. thaliana in the world was established recently, as suggested by previous studies (KING et al. 1993; PRICE et al. 1994; INNAN et al. 1996).

The laboratory strain Col-0 has a long branch in phy- logenetic tree (Figure 3). Cold has unique alleles with a long repeat at six loci among 19 (one unstudied lo- cus), of which five are the longest for a locus (Table 3). This characteristic obviously contributed to high val- ues in the distance estimate (average squared differ- ence) and resulted in the long and isolated branch. Col-0 was established by selecting an individual from offspring of the original Laibach Landsberg by GEORGE DEI (Nottingham Arabidopsis Stock Centre Home Page 1994). In the sequencing study of the Adh region (INNAN et al. 1996), Col-0 was shown to be a recombi- nant between Landsberg and another recombinant de- tected in the study. The selection procedure from Landsberg may be related to the occurrence of these unique and long microsatellite alleles in Col-0. A com-

Microsatellites in A. thaliana 1451

parison of microsatellite patterns with the Landsberg may provide an answer to the origin of these alleles. Unfortunately, the Landsberg was not studied here.

Comparison between the infinite allele and stepwise mutation models: To compare the infinite allele and stepwise mutation models, the expected number of al- leles was estimated under the two models. For 12 of the 20 loci, the observed number of alleles was between the expected values of the two models. A non-parametric test (the sign test) rejected both models at the 1% level. This result indicates that neither mutation model is satisfactory to explain the observed pattern of polymor- phism at the microsatellite loci investigated.

There may be some problems in applying these theo- retical models to microsatellite polymorphism in A. thal- iana. Both models assume stationarity of population and neutral mutations. The phylogenetic tree obtained in this study suggest that the present A. thaliana popula- tion was established recently throughout the world. Based on sequence analysis of the Adh gene, INNAN et al. (1996) suggested that this plant migrated from the Himalayas from at most 0.1 million years ago. There- fore, the present population of this plant may not be in equilibrium at the species level. Rather, this plant seems to be expanding its habitat in the world ( R ~ D E I 1969). It may be difficult to test these theoretical models in an expanding population. In addition, it is unlikely that all the microsatellite loci investigated are free from natural selection. Because of homozygosity of all the individuals investigated in this study, conventional p o p ulation genetic tests were not applied to see if there was any deviation from neutrality. If some of the loci are involved in an important function (e.g., gene regula- tion), they are unlikely to be selectively neutral. One suggestive result is that some loci (ATHATPASE, nga63, and nga280) have low variance and high kurtosis. In other words, these loci have only a small number of alleles (four alleles) compared to the other loci. Al- though the low number of alleles at these loci may reflect an exceptionally low mutation rate or a relatively young age of the loci, it is also possible that natural selection is keeping the level of variation low at these loci. On the other hand, there is no reason to reject some form of balancing selection on microsatellites. The high number of alleles at most of the loci may be due to positive selection favoring a large diversity of repeat number. Before testing theoretical models in future experiments, it may be important consider these possibilities for microsatellites.

The authors thank N. GOTO for A. thaliana seeds, and T. ENDO, S. UWANO, S. NASUDA, F. TAJIMA and Y. YASUI for comments and suggestions. This is contribution 543 from the Laboratory of Plant Genetics, Graduate School of Agriculture, Kyoto University, Japan.

LITERATURE CITED ABBOTT, R. J., and M. F. GOMES, 1989 Population genetic structure

and outcrossing rate of Arabidopsis thaliana (L.) Heynh. Heredity 62: 411-418.

AKXAYA, M. S., A. A. BHACWATT and P. B. CREGAN, 1992 Length poly- morphisms of simple sequence repeat DNA in soybean. Genetics

BAIRD, M., I. BALAZS, A. GIUSTI, L. MIYAZAKI, K. WEXLER et al., 1986 Allele frequency distribution of two highly polymorphic DNA sequences in three ethnic groups and its application to the deter- mination of paternity. Am. J. Hum. Genet. 39: 489-501.

BELL, C. J., and J. R. ECKER, 1994 Assignment of 30 microsatellite loci to the linkage map of Arabidopsis. Genomics 19: 137-144.

BOWCOCK, A. M., A. RUIZ-LINARES, J. TOMFOHRDE, E. MINCH, J. R. KIDD et al., 1994 High resolution of human evolutionary trees with polymorphic microsatellites. Nature 368: 455-457.

CHAKRABORTY, R., and R. C. G R I ~ T H 1982 Correlation of heterozy- gosity and number of alleles in different frequency classes. Theor. Popul. Biol. 21: 205-218.

CHAKRABORTY, R., M. FORNAGE, R. GUEGUEN and E. BOERWINKLE, 1991 Population genetics of hypervariable loci: analysis of PCR based VNTR polymorphism within a population, pp. 127-143 in DNA Fingoprinting: ApProaches and Applications, edited by T. BURKE, G. DOLF, A. J. JEFF- and R. WOLFF, Birkhauser Verlag, Basel.

CLARK, A. G., 1987 Neutrality tests of highly polymorphic restriction fragment length polymorphisms. Am. J. Hum. Genet. 41: 948- 956.

CONDIT, R., and S. P. HUBBELL, 1991 Abundance and DNA se- quence of two-base repeat regions in tropical tree genomes. Ge- nome 34: 66-71.

CROW, J. F., and M. KIMURA, 1970 An Introduction to Population Genet- ics Theory. Harper & Row, New York, NY.

DALLAS, J. F., 1992 Estimation of microsatellite mutation rates in recombinant inbred strains of mouse. Mamm. Genome 3: 452- 456.

DEKA, R., R. CHAKRABORTY and R. E. FERRELL, 1991 A population genetic study of six VNTR loci in three ethnically defined popula- tions. Genomics 11:- 83-92.

DI RIENZO, A., A. C. PETERSON, J. C. GRAZA, A. M. VALDES, M. SLATKIN, et al., 1994 Mutational processes of simple-sequence repeat loci in human populations. Proc. Natl. Acad. Sci. USA 91: 3166- 3170.

DRAKE, J. W., B. W. GLICKMAN and L. S. RIPLEY, 1983 Updating the theory of mutation. Am. Sci. 71: 621-630.

EDWARDS. A,, H. A. HAMMOND, L. JIM, C. T. CASKEY and R. CHAKRA- BORTY 1992 Genetic variation at five trimeric and tetrameric tandem repeat loci in four human population groups. Genomics

ESTOUP, A., L. GARNERY, M. SOLICNAC and J.-M. CORNUET, 1995a Microsatellite variation in honey bee (Apis m,ell@ra L.) popula- tions: hierarchical genetic structure and test of the infinite allele and stepwise mutation models. Genetics 140: 679-695.

ESTOUP, A., C. TAILLIEZ, J.-M. CORNUET and M. SOLICNAC, 1995b Size homoplasy and mutational processes of interrupted micro- satellite in two bee species, Apis mell@ra and Bombus tmestris (Apidae). Mol. Biol. Evol. 12: 1074-1084.

EWENS, W. J. 1972 The sampling theory of selectively neutral alleles. Theor. Popul. Biol. 3: 87-112.

GOLDSTEIN, D. B., A. R. LINARES, L. L. CAVALLI-SFORZA and M. W. FELDMAN. 1995 An eduat ion of genetic distances for use with microsatellite loci. Genetics 139: 475-485.

HILL, W. G., and A. ROBERTSON, 1968 Linkage disequilibrium in finite populations. Theor. Appl. Genet. 38: 226-231.

HUDSON, T. J., M. ENGELSTEIN, M. K LEE, E. C. Ho, M. J. RUBENFERD et al., 1992 Isolation and chromosomal assignment of 100 highly informative human simple sequence repeat polymor- phisms. Genomics 13: 622-629.

INNAN, H., F. TAJIMA, R. TERAUCHI and N. T. MIYASHITA, 1996 Intra- genic recombination in the Adh locus of a wild plant Arabidopszs thaliana. Genetics 143: 1761-1770.

JEFFREYS, A. J., V. WILSON and S. L. THEIN, 1985 Individual-specific “fingerprints” of human DNA. Nature 316: 76-79.

JONES, M. E., 1971a The population genetics of Arabidopszs thaliana. I. The breeding system. Heredity 27: 39-50.

JONES, M. E., 1971b The population genetics of Arabidopsis thaliana. 11. Population structure. Heredity 27: 51-58.

JONES, M. E., 1971c The population genetics of Arabidopszs thaliana. 111. The effect of vernalisation. Heredity 27: 59-72.

132 1131-1139.

12: 241-253.

1452 H. Innan, R. Terauchi and N. T. Miyashita

KIMURA, M., and J. CROW, 1964 The number of alleles that can be maintained in a finite population. Genetics 49: 725-738.

KIMURA, M., and T. OHTA, 1975 Distribution of allelic frequencies in a finite population under stepwise production of neutral al- leles. Proc. Natl. Acad. Sci. USA 72: 2761-2764.

KIMuRA, M., and T. OHTA, 1978 Stepwise mutation model and distri-

Acad. Sci. USA 75: 2868-2872. bution of allelic frequencies in a finite population. Proc. Natl.

KING, G., J. NIENHUIS and C. HUSSJZY, 1993 Genetic similarity among ecotypes of Arabidopsis thalianaestimated by analysis of restriction fragment length polymorphisms. Theor. Appl. Genet. 86: 1028- 1032.

KUNSHENG, W., and S. D. TANKSLEY, 1993 Abundance, polymor- phism and genetic mapping of microsatellites in rice. Mol. Gen. Genet. 241: 225-235.

LAGERCRANTZ, U., H. ELLERCREN and L. ANDERSON, 1993 The abundance of various polymorphic microsatellite motifs differs between plants and vertebrates. Nucleic Acids Res. 21: 1111- 1115.

L I ~ , M., and J. A. Lun, 1989 A hypervariable microsatellite re- vealed by in vitro amplification of dinucleotide repeat within the cardiac muscle actin gene. Am. J. Hum. Genet. 44: 388-396.

LEVINSON, G., and G. A. GUTMAN, 1987a Slipped-strand mispairing: a major mechanism for DNA sequence evolution. Mol. Biol. Evol.

LEVINSON, G., and G. A. GUTMAN, 1987b High frequency of short bacteriophage M13 in Escherichia coli K-12. Nucleic Acids Res. 15:

MICHAIAKIS, Y., and M. VEUILLE, 1996 Length variation of CAG/ CAA trinucleotide repeats in natural populations of Drosophila melanogasterand its relation to the recombination rate. Genetics 143: 1713-1725.

MORGANTE, M., and A. M. 0 1 . ~ ~ ~ 1 , 1993 PCR-amplified microsatel- lites as markers in plant genetics. Plant J. 3: 175-182.

NAKAMURA, Y., M. LEPPERT, P. O'CONNELL, R. WOLFF, T. HOLM et al., 1987 Variable number of tandem repeat (VNTR) markers for human gene mapping. Science 235: 1616-1622.

NEI, M., 1973 Analysis of gene diversity in subdivided populations. Proc. Natl. Acad. Sci. USA 70: 3321-3323.

Nottingham Arabidopsis Stock Centre Home Page, 1994 The origin of Landsberg erecta, Columbia and C34. http://nasc.life.ac.uk/ centre.htm1.

OHTA, T., and M. KIMURA, 1969 Linkage disequilibrium due to ran- dom genetic drift. Genet. Res. 13: 47-55.

OHTA, T. and M. KIMURA, 1973 The model of mutation appropriate to estimate the number of electrophoretically detectable alleles in a genetic population. Genet. Res. 2 2 201-204.

POWELL, W. G., M. MORGANTE, J. J. DOYLE, J. W. MCNICOL, S. V. TINGEY and A. J. RAFALSKI, 1995 Polymorphic simple sequence repeat regions in chloroplast genomes: applications to the popu- lation genetics of pines. Proc. Natl. Acad. Sci. USA. 92: 7759- 7763.

POWF.I.I., W., G. C. MACHRAY and J. PROVAN, 1996 Polymorphism revealed by simple sequence repeats. Trends Plant Sci. 1: 215- 222.

PRICE, R. A,, J. D. PALMER and I. A. AI-SHEHBAZ, 1994 Systematic relationship of Arabidopsis: a molecular and morphological per- spective, pp. 7-19. in Arabidopsis, edited by E. MEYEROWITZ and C. R. SOMMERVILLE. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.

=DEI, G. P., 1969 Arabidopsis thaliana (L.) Heynh. A review of the genetics and biology. Bibliogr. Genet. 21: 1-151.

RONGWEN, J., M. S. AKKAYA, A. A. BHAGWAT, U. LAVI and P. B. CKECAN,

4: 203-221.

5322-5338.

1995 The use of microsatellite DNA markers for soybean genc- type identification. Theor. Appl. Genet. 90: 43-48.

SENIOR, M. L., and M. HEUN, 1993 Mapping maize microsatellites and polymerase chain reaction conformation of the targeted repeats using CT primer. Genome 36: 884-889.

SHEN, S. H., J. L. SLICHTON and 0. SMITHYES, 1981 A history of the human fetal globin gene duplication. Cell 26: 191-203.

SIIXR, L. M. 1992 Bouncing off microsatellites. Nat. Genet. 2: 8- 9

SNEDECOR, G. W., and W. G. COCHRAN, 1978 Statistical Methods, The Iowa State University Press, Ames, Iowa, pp. 296-298.

TAUTZ, D., and M. RENZ, 1984 Simple sequences are ubiquitous repetitive components of eukaryotic genomes. Nucleic Acids Res. 12: 4127-4138.

TERAUCHI, R., 1994 A polymorphic microsatellite marker from the tropical tree Lhyobalanops lanceolatu (Dipterocarpaceae). Jpn. J. Genet. 69: 567-576.

TERAUCHI, R., and A. KONUMA, 1994 Microsatellite polymorphism in Dioscmea tokmo, a wild yam species. Genome 37: 794-801.

THOMAS, M. R., and N. S. SCOTT, 1993 Microsatellite repeats in grapevine reveal DNA polymorphisms when analysed as se- quence-tagged sites (STSs). Theor. Appl. Genet. 8 6 985-990.

TODOKORO, S., R. TEKAUCHI and S. KAWANO, 1995 Microsatellite polymorphisms in natural populations of Arabidopsis thaliana in Japan. Jpn. J. Genet. 70: 543-554.

VALDES, A. M., M. SLATKIN and N. B. FREIMER, 1993 Allele frequen- cies at microsatellite loci: the stepwise mutation model revisited. Genetics 133: 737-749.

VAN TREUREN, R., H. KUITTINEN, K. W I N E N , E. BAENA-GONZALEZ and 0. SAVOWNEN, 1997 Evolution of microsatellites in Arabis petraea and Arabis lyrata, outcrossing relatives of Arabidopsis thali- ana. Mol. Biol. Evol. 14: 220-229.

WANG, A.J. H., G. J. QUIGLEY, F. J. KOLPAK, J. L. CROWFORD, J. H. VAN BOON et al., 1979 Molecular structure of a left-handed double helical DNA fragment at atomic resolution. Nature 282: 680- 686.

WEBER, J. L., 1990 Informativeness of human (dGdA)n-(dGdT)n polymorphisms. Genomics 7: 517-524.

WEBER, J. L., and C. WONG, 1993 Mutation of human short tandem repeats. Hum. Mol. Genet. 2 1123-1128.

WESTERMAN, J. M., 1971a Genotype-environment interaction and developmental regulation in Arabidopsis thaliana. 11. Inbred lines; analysis. Heredity 2 6 93-106.

WESTERMAN, J. M., 1971b Genotype-environment interaction and developmental regulation in Arabidopsis thaliana. 111. Inbred lines; analysis of responce to photoperiod. Heredity 26: 373- 382.

WESTERMAN, J. M., 1971c Genotype-environment interaction and developmental regulation in Arubidopsis thaliana. N . Wild mate- rial; analysis. Heredity 26: 383-395.

WFSTERMAN, J. M., and M. J. LAWRENCE, 1970 Genotype-environ- ment interaction and developmental regulation in Arabidopsis thaliana. I. Inbred lines; description. Heredity 2 5 609-627.

WOLFF, R. K., R. PLAEKE, A. J. JEFFREYS and R. WHITE, 1991 Unequal crossing over between homologous chromosomes is not the ma- jor mechanism involved in the generation of new alleles at VNTR loci. Genomics 5: 382-384.

W w , A. R., and R. WHITE, 1980 A highly polymorphic locus in human DNA. Proc. Natl. Acad. Sci. USA 77: 6754-6758.

ZAR, J. H., 1974 Biostatistical Analysis. Prentice-Hall, Inc. Englewood Cliffs, NJ.

ZHAO, X., and G. KOCHERT, 1993 Phylogenetic distribution and ge- netic mapping of a (GGC)n microsatellite from rice ( W z a sativa L.). Plant Mol. Biol. 21: 607-614.

Communicating editor: A. G. CIARK