Brief Communications - Oxford Journals

Brief Communications

Flowering Gene Ppd in Pea: Map Position and Disturbed Segregation of Allele ppd-2 I. C. Murfet and S. A. Taylor Our results show flowering gene ppd in pea (Pisum sativum L.) is located between branching gene rms3 and isozyme locus Aatp near the IA end of a chromosome now known to include linkage groups IA and II. The ppd locus is about 3 cM from rms3 and 5 cM from Aatp. Two mutant alleles of Ppd are known, ppd-1 and ppd-2. Both mutations result in early flowering and loss of ability to respond to photoperiod. In F2 populations segregating for alleles Ppd and ppd-2 we found a significant deficiency of mutant segregants (on average, half the expected 25%). Reciprocal crosses were made between heterozygous Ppd ppd-2 and homozygous ppd2 ppd-2 plants. Segregation was in accordance with a 1:1 ratio when the hybrid plants were used as the female parent but a significant (P , .0001) deficiency of recessive plants occurred (only 24% were ppd-2) when the hybrid plants were used as the male parent. These results suggest that where Ppd and ppd-2 pollen are in competition there is selection against male gametes carrying the ppd-2 allele. The ppd-1 mutation appears less severe than ppd-2 and segregation for ppd-1 was not significantly disturbed. The ability to respond to photoperiod is conferred in pea (Pisum sativum L.) by the complementary action of three major flowering genes Sn ( Barber 1959; Murfet 1971), Dne ( King and Murfet 1985) and Ppd (Arumingtyas and Murfet 1994). The latter authors found evidence of weak linkage (32 cM) between ppd and the anthocyanin gene a in group IA, but the map position was not determined. We present here evidence that ppd is located in group IA be-

548

tween the branching (ramosus) gene rms3 and the isozyme locus Aatp (aspartate aminotransferase). The rms3 locus lies about 7 cM beyond Aatp near the IA end of a chromosome now known to include linkage groups IA and II (Rameau et al. 1998; Weeden et al. 1996). Two mutant alleles of Ppd are known, ppd-1 and ppd-2 (Arumingtyas and Murfet 1994; Taylor and Murfet 1996). In the course of the linkage studies it became clear that there was a consistent deficiency of mutant segregants in F2 populations segregating for Ppd and ppd-2. We therefore made reciprocal testcrosses between heterozygous Ppd ppd-2 and homozygous ppd-2 ppd-2 plants. The results indicate male gametophytic selection against the ppd-2 allele.

Materials and Methods Group IA markers waxy1 ( Kovalenko and Ezhova 1992) and rms3 (Rameau et al. 1998) are located about 30 cM either side of the a locus. Accordingly, we crossed line M2/830 (ppd-2) with lines HL242 (waxy1; cross 846) and WL6042 (rms3-3; cross 847). No linkage was apparent with waxy1, but possible linkage of ppd with rms3 was indicated by an absence of double mutants in the 95 F2 plants of cross 847. Double recessive ppd rms3 plants were obtained in the descendants of F2 plant 847a/2/6 and these were used to make coupling phase crosses as follows. Cross 967: line HL2 (A Ppd Rms3) 3 847a/ 2/6/8 (a ppd rms3); cross 972: line BC1/6 (Aatp-F Ppd Rms3) 3 967d/2/24 (Aatp-S ppd rms3); cross 973: line BC1/6 3 847a/2/6/8/ 4 (Aatp-S ppd rms3); testcross 978: 972 F1 (female parent) 3 967d/2/24/- (male parent); and testcross 979: the reciprocal of testcross 978. All F2 and testcross plants were grown singly in 14 cm slim-line pots under a 10h short-day photoperiod to enable clear separation of late photoperiodic (Ppd) and

day-neutral (ppd) segregants. Under these 10 h conditions most Rms3 plants produced one or two basal laterals, but the rms3 segregants were readily distinguished by their profuse basal and aerial branching, thinner stems, and shorter internodes. Testcrosses 978 and 979 were made by crossing plants growing under an 18 h photoperiod. Additional information on materials and growing conditions is given by Arumingtyas and Murfet (1994) and Taylor and Murfet (1996). Linkage chi squares were calculated from a 2 3 2 contingency table and recombination values estimated from F2 data by the productratio method.

Results and Discussion The results from both F2 and testcross populations ( Table 1) indicate close linkage between ppd and rms3 with a distance of 3 cM. There was also strong linkage between ppd and Aatp with a distance of 5 cM. The distance of 7.5 cM for rms3 and Aatp indicated by the testcross data is near identical to the value of 7.4 cM reported by Rameau et al. (1998); the F2 data gave a slightly higher value of 9.8 cM. We conclude that the ppd locus lies between rms3 and Aatp near the group IA end of a chromosome now known (Weeden et al. 1996) to include linkage groups IA and II. Although we did not obtain significant linkage between ppd and a, the observed distance of 37 6 6 cM ( Table 1) is fairly close to the value of 32 6 5 cM reported by Arumingtyas and Murfet (1994) and the combined data (n 5 249) indicate significant linkage (P , .001) with a distance of 33 6 4 cM. In six different F2 populations segregating for Ppd and recessive mutant allele ppd-2, the percentage of recessive segregants ranged from 18.6 to 5.7 with an average value of 12.6% ( Table 2). While the deviation from a 3:1 ratio was not significant in some populations, the combined

Table 1. Dihybrid segregation data for ppd and group IA markers obtained from F 2 or testcross populations Phenotypea Cross

DD

DR

RD

RR

Total

Linkage x2

P

967 972 973 Comb 972 973 Comb 972 973 Comb 967 967

112 52 49 213 52 46 98 53 48 101 89 90

0 0 0 0 0 3 3 0 3 3 7 6

1 1 2 4 2 2 4 1 0 1 23 23

10 3 6 19 2 6 8 2 6 8 4 4

123 56 57 236 56 57 113 56 57 113 123 123

110.8 41.2 41.1 191.4 27.0 24.5 49.5 36.6 35.8 69.7 1.5 2.1

,.0001 ,.0001 ,.0001 ,.0001 ,.0001 ,.0001 ,.0001 ,.0001 ,.0001 ,.0001 ..2 ..1

Testcross populations Rms3 Ppd 978 979 Comb Rms3 Aatp 978 Ppd Aatp 978

43 60 103 40 42

0 2 2 3 3

2 1 3 3 1

35 17 52 34 34

80 80 160 80 80

72.3 64.1 138.5 57.7 64.8

,.0001 ,.0001 ,.0001 ,.0001 ,.0001

Loci F2 populations Rms3 Ppd

Rms3 Aatp

Ppd Aatp

A Rms3 A Ppd

a

RCV 6 SE

2.7 6 1.1 9.8 6 3.0 5.1 6 2.1 39.2 6 5.9 37.0 6 5.7 2.5 3.8 3.1 7.5 5.0

All crosses are in coupling phase; details are given in the materials section. D 5 dominant or homozygous fast 1 heterozygous for Aatp; R 5 recessive or homozygous slow for Aatp. The first named locus is shown first.

data (n 5 437) reveal a highly significant (P , .0001) deficiency of recessive plants. The results from reciprocal testcrosses ( Table 3) show that when heterozygous (Ppd ppd-2) plants were used as the female parent, the observed segregation was in accordance with a Mendelian 1:1 ratio. In contrast, when the hybrid plants were used as pollen donors with homozygous (ppd-2 ppd-2) plants as the female parent, there was a highly significant (P , .0001) deficiency of recessive offspring. Comparing the two 80-plant testcross populations using a 2 3 2 contingency table confirmed that the proportion of recessive segregants was not independent of cross direction (x2 5 7.18, P , .01). The pollen from hybrid plants appeared fully (.99%) fertile from examination of orcein-stained samples. We conclude there is male gametophytic selection against the ppd-2

allele. Using hybrid plants as the male parent in the testcross produced 23.8% of recessive offspring ( Table 3). Assuming no female gametophytic bias we would expect 11.9% (0.5 3 0.238 3 100) of recessive plants in an F2 population. This is close to the observed mean of 12.6% ( Table 2). Segregation at the rms3 and Aatp loci is normally in accordance with Mendelian expectation (Arumingtyas et al. 1992; Poole et al. 1993; Rameau et al. 1997; Weeden and Marx 1987). However, segregation of these flanking markers was grossly disturbed in crosses involving the ppd-2 allele. For example, the combined F2 data for crosses 972 and 973 ( Table 1) show 8.0% of ppd-2 plants, 10.6% of rms3 plants, and 9.7% of homozygous Aatp slow plants. Table 2 also includes a summary of published segregation data for several other

mutations in the photoperiod response genes Sn, Dne, and Ppd. The combined data for allele ppd-1 fit a 3:1 ratio (P . .7) but differ significantly (P , .001) from the combined data for ppd-2; the proportion of mutant segregants is 23.8% versus 12.6%, respectively. These results further support the conclusion of Taylor and Murfet (1996) that ppd-1 is a less severe mutation than ppd-2. The naturally occurring sn mutation (allele sn-1, type line HL58) consistently displays disturbed segregation with an average of around 20% of mutant F2 segregants (Murfet 1971; Rameau et al. 1998). The data for the induced mutant allele sn-2 are too small to permit a firm conclusion but the value of 15% of recessive segregants is consistent with its severe mutant phenotype. Only one mutant allele is known at the dne locus. The combined F2 data (n 5 240) for dne-1 show only 21% of mutant segregants, but the observed numbers still fit a 3:1 ratio (P . .1). The dne-1 allele is considered leaky (Murfet 1989), and a larger sample size is necessary to establish whether segregation is significantly disturbed. In summary, there is convincing evidence that mutations sn-1 and ppd-2 result in significantly disturbed segregation. In addition, 17 crosses, 3 photoperiod response genes, and 5 mutations are represented in Table 2, but in not one instance did the proportion of recessive F2 segregants reach 25%. It appears mutations that impair the Sn Dne Ppd pathway result in male gametophytic selection against the mutant allele. The product of the Sn Dne Ppd pathway acts as a graft-transmissible flower inhibitor ( Barber and Paton 1952; King and Murfet 1985; Murfet and Reid 1973; Taylor and Murfet 1996). However, this gene system has pleiotropic effects on many traits and the primary role of the product may be to direct assimilate flow

Table 2. Monohybrid segregation in F 2 populations segregating for wild-type and mutant alleles at one or other of the three photoperiod response loci Ppd, Sn, or Dne Phenotype Cross

Allele

Wild type

Mutant

Total

x2 (3:1)

P

Proportion of mutant

M2/830 3 Borek 846 847 967 972 973 Combined data M2/137 3 Borek M2/137 3 HL111 Combined data M2/176 3 Borek Combined (6 crosses) Combined (2 crosses)

ppd-2 ppd-2 ppd-2 ppd-2 ppd-2 ppd-2 ppd-2 ppd-1 ppd-1 ppd-1 sn-2 sn-1 dne-1

39 48 78 113 53 51 382 36 95 131 39 1590 189

8 11 17 10 3 6 55 10 31 41 7 406 51

47 59 95 123 56 57 437 46 126 172 46 1996 240

1.60 1.27 2.56 18.67 11.52 6.37 35.92 0.26 0.01 0.12 2.35 23.11 1.80

..2 ..2 ..1 ,.0001 ,.001 ,.05 ,.0001 ..5 ..9 ..7 ..1 ,.0001 ..1

17.0% 18.6% 17.9% 8.1% 5.7% 10.5% 12.6% 21.7% 24.6% 23.8% 15.2% 20.3% 21.3%

Reference Taylor and Murfet 1996 This article This article Table 1 Table 1 Table 1 Arumingtyas and Murfet 1994 Arumingtyas and Murfet 1994 Arumingtyas and Murfet 1994 Murfet 1971 King and Murfet 1985

Brief Communications 549

tification of a third ppd allele and its physiological action as revealed by grafting. Physiol Plant 97:719–723.

Table 3. Monohybrid segregation in reciprocal testcrosses 978 and 979 Testcross parents

Phenotype

Female

Ppd

ppd

Total

x2 (1:1)

P

45 61

35 19

80 80

1.25 22.05

..2 ,.00001

male

Ppd ppd-2 3 ppd-2 ppd-2 ppd-2 ppd-2 3 Ppd ppd-2

( Beveridge et al. 1992; Kelly and Davies 1986; Murfet and Reid 1993; Reid and Murfet 1984). Sn Dne Ppd activity slows the rate of seed fill ( Duche ˆne 1984), and using both selfing and crossing of Sn and sn isolines Kelly and Spanswick (1997) have shown the final dry weight of the seed is determined by the genotype of the maternal plant not the embryo, thus supporting the hypothesis that the product of the Sn Dne Ppd pathway acts in the vegetative plant to regulate partitioning of assimilates. Activity of the Sn Dne Ppd pathway is down-regulated by light via phytochrome A (Weller et al. 1997a,b) and is only minimally expressed under an 18 h photoperiod (Murfet and Reid 1974). It is curious, therefore, that Ppd pollen had such a strong selective advantage (they were three times as successful as ppd-2 pollen) in plants that were grown and cross-pollinated under an 18 h photoperiod. It is possible that insufficient light reached the pollen tube, but light access to the stigma and style was increased when the keel petals were opened to enable emasculation and cross-pollination. More likely either light is unable to downregulate the Sn Dne Ppd pathway in haploid tissue or only a small amount of activity is sufficient to confer a major selective advantage. This question deserves further exploration. The basis for the male gametophytic selection also remains unclear. Using isolines and homogeneous pollen, Duche ˆne (1984) could not detect any appreciable difference in the rate of pollen tube growth for Sn and sn pollen tubes on either Sn or sn maternal tissue. However, the relative performance of Sn and sn pollen in a mixed population could not be examined in the absence of a suitable marker. From the School of Plant Science, University of Tasmania, GPO Box 252-55, Hobart 7001, Australia. S. A. Taylor is now at the Department of Applied Genetics, John Innes Institute, Colney Lane, Norwich, UK. We thank Ian Cummings, Tracey Jackson, and Jodie van de Kamp for technical assistance and the Australian Research Council for financial support. Address correspondence to I. C. Murfet at the address above or email: [email protected]. q 1999 The American Genetic Association

550 The Journal of Heredity 1999:90(5)

Weeden NF and Marx GA, 1987. Further genetic analysis and linkage relationships of isozyme loci in pea. J Hered 78:153–159. Weeden NF, Swiecicki WK, Timmerman-Vaughan GM, Ellis THN, and Ambrose M, 1996. The current pea linkage map. Pisum Genet 28:1–4.

References

Weller JL, Murfet IC, and Reid JB, 1997a. Pea mutants with reduced sensitivity to far-red light define an important role for phytochrome A in daylength detection. Plant Physiol 114:1225–1236.

Arumingtyas EL, Floyd RS, Gregory MJ, and Murfet IC, 1992. Branching in Pisum: inheritance and allelism tests with 17 ramosus mutants. Pisum Genet 24:17–31.

Weller JL, Reid JB, Taylor SA, and Murfet IC, 1997b. The genetic control of flowering in pea. Trends Plant Sci 2: 412–418.

Arumingtyas EL and Murfet IC, 1994. Flowering in Pisum. A further gene controlling response to photoperiod. J Hered 85:12–17.

Received April 27, 1998 Accepted December 31, 1998 Corresponding Editor: Norman Weeden

Barber HN, 1959. Physiological genetics of Pisum. II. The genetics of photoperiodism and vernalisation. Heredity 13:33–60. Barber HN and Paton DM, 1952. A gene-controlled flowering inhibitor in Pisum. Nature 169:592. Beveridge CA, Ross JJ, and Murfet IC, 1992. Mutant dn influences dry matter distribution, assimilate partitioning and flowering in Lathyrus odoratus L. J Exp Bot 43: 55–62.

Inheritance of Growth Habit– Related Attributes in Red Clover (Trifolium pratense L.)

Duche ˆne C, 1984. Reproductive development in Pisum: the role of genes Sn and Lf (M.Sc. thesis). Hobart, Australia: University of Tasmania.

H. Mirzaie-Nodoushan, I. L. Gordon, and W. B. Rumball

Kelly MO and Davies PJ, 1986. Genetic and photoperiodic control of the relative rates of vegetative and reproductive development in peas. Ann Bot 58:12–21. Kelly MO and Spanswick RM, 1997. Maternal, singlegene regulation of assimilate partitioning in pea. Plant Physiol 114:1055–1059. King WM and Murfet IC, 1985. Flowering in Pisum: a sixth locus, Dne. Ann Bot 56:835–846. Kovalenko OV and Ezhova TA, 1992. Two waxless mutants of somaclonal origin in pea. Pisum Genet 24:60– 63. Murfet IC, 1971. Flowering in Pisum. A three gene system. Heredity 27:93–110. Murfet IC, 1989. Flowering genes in Pisum. In: Plant reproduction: from floral induction to pollination ( Lord EM and Bernier G, eds). Rockville, Maryland: American Society of Plant Physiologists; 10–18. Murfet IC and Reid JB, 1973. Flowering in Pisum: evidence that gene Sn controls a graft-transmissible inhibitor. Aust J Biol Sci 26:669–673. Murfet IC and Reid JB, 1974. Flowering in Pisum: the influence of photoperiod and vernalising temperatures on expression of genes Lf and Sn. Z Pflanzenphysiol 71: 323–331. Murfet IC and Reid JB, 1993. Developmental mutants. In: Peas: genetics, molecular biology and biotechnology (Casey R and Davies DR, eds). Wallingford, UK: CAB International; 165–216. Poole AT, Murfet IC, and Vaillancourt RE, 1993. Ramosus loci rms3 and rms4 are in linkage groups 1 and 7, respectively. Pisum Genet 25:40–42. Rameau C, Bodelin C, Cadier D, Grandjean O, Miard F, and Murfet IC, 1997. New ramosus mutants at loci Rms1, Rms3 and Rms4 resulting from the mutation breeding program at Versailles. Pisum Genet 29:7–12. Rameau C, Deńoue D, Fraval F, Haurogne´ K, Josserand J, Lancou V, Batge S, and Murfet IC, 1998. Genetic mapping in pea. 2. Identification of RAPD and SCAR markers linked to genes affecting plant architecture. Theor Appl Genet 97:916–928. Reid JB and Murfet IC, 1984. Flowering in Pisum: a fifth locus veg. Ann Bot 53:369–382. Taylor SA and Murfet IC, 1996. Flowering in pea: iden-

There are two divergent types of red clover, prostrate and erect. Prostrateness of the first type may be helpful in changing the architecture of commercial cultivars to increase their persistence. Generation mean experiments were carried out to investigate the function and number of genes controlling the two growth habit– related attributes, prostrateness and stem thickness, in red clover. To achieve these estimates, two pairs of parent plants (one erect and one prostrate in each pair) were used to produce F1, F2, BC1, and BC2 populations. Three-parameter, six-parameter, and best-fit models were presented for each attribute. Prostrateness was partially dominant to erectness. Thick stems were partially to completely dominant to thin stems. There was strong evidence for dominance 3 dominance epistasis controlling prostrateness and additive 3 additive epistasis controlling stem thickness. Both characters seemed to be controlled by few genes. In high rainfall areas, a red clover (Trifolium pratense L.) with good persistence would be a major asset to farmers. Lack of persistence is a major limitation to the widespread acceptance of this forage legume (Smith and Bishop 1993). Its low persistence is mostly due to susceptibility to root rot and an inherently short life span. Its persistence decreases even further in natural pastures when grazing pressure is high. Changing the architec-

ture of the existing commercial cultivars may help to increase the persistence of red clover. There are two extreme discrete phenotypic classes in red clover, erect and prostrate. The first type is characterized by large leaves and upright and thick stems on which few adventitious roots grow. The second type is characterized by small leaves, creeping thin stems, with more adventitious roots on the nodes. The second type is more persistent (Smith and Bishop 1993). It may also be useful for crossing with erect commercial cultivars to increase their persistence. What is the nature of gene action controlling these attributes? How many genes control these traits? To answer these questions, it was decided to carry out several sets of generation mean analysis, assuming the parent plants are homozygous for those loci controlling the attributes of the two diverse types. Red clover is a self-incompatible and completely cross-pollinated plant. As a result, individual plants of red clover are heterozygous and its populations are heterogeneous. Consequently the starting points for genetic analysis in red clover are heterogeneous varieties or heterozygous individual plants. This may complicate the theoretical basis for the analysis of both qualitative and quantitative variation. Thus using generation mean analysis in a plant population like red clover might be impossible. Although in the crosspollinated plant populations, pure lines could be produced through full-sib mating or other procedures which take a longer time than through selfing, but the requirement of inbred lines in practice may lead to unwanted complications of inbreeding depression on the mean of the populations. However, divergent populations may well be homozygous for the genes controlling the contrasting morphological types. This particularly happens when populations have been isolated and selected naturally for the contrasting traits for a long time. In this case it is valid to use individual plants from each population as parents, followed by generation mean analysis to investigate the inheritance of the character(s) under study.

Materials and Methods Two pairs of parent plants (one erect and one prostrate in each pair) were used in this study. Parent plants were induced to flower by providing artificial long days and warm temperatures during the winter season. Crosses were made by hand pol-

lination between contrasting parent plants. F1 seeds were kept on wet filter paper, in a cool room, at 48C for 10 days as a prechilling treatment. F1 plants were backcrossed to their parent plants (P1 and P2) to produce the backcross generations, BC1 and BC2. Because of selfincompatibility, F2-generation seeds were produced through full-sib mating of F1 plants. Progeny Tests Parents, F1, F2, BC1, and BC2 seeds were pregerminated and transplanted into jiffy peat pots in a glasshouse. Intact jiffy pots were transplanted into the field. A randomized complete block design with three replications was used for both sets of crosses. Field plots consisted of eight plants in a single 4 m row with 50 and 60 cm spaces between individual plants within and between rows, respectively. The number of rows per plot differed according to segregating and nonsegregating generations. Measurements were made on all of the individual plants. Prostrateness was scored from 1 to 5. Score 1 was given to plants that were completely erect and score 5 to plants that were completely prostrate. Half scores were used as required. Stem thickness was measured in millimeters on three random main stems per plant at the third internode at the median flowering stage. The average of the three measurements was used for analysis. Statistical Analysis An ordinary randomized complete block design analysis of variance was conducted for the characters to determine whether differences exist among generation means. Since some generations were segregating while others were not, heterogeneity of the within-plot variances was expected. Therefore a weighted analysis was done, utilizing the inverse of the plant-to-plant within-plot variances. Genetic Analysis Function of the genes. Since the analysis of variance of the field experimental design indicated significant differences between generation means, a generation mean analysis was carried out for the characters for the two sets. The prostrate parent was designated P1 and the erect parent P2 regardless of the mean values of parents in the two characters. Usually the parent with the greater value is designated P1 to get a positive value for d. In this experiment prostrateness in the prostrate par-

Table 1. Degrees of freedom and mean squares from the weighted analyses of variance of parents, F1, F2, BC 1, and BC 2 for prostrateness and stem thickness in two sets of crosses (erect 3 prostrate)

Set

Source

df

Prostrate- Stem ness thickness

1

Replication Generation Error Replication Generation Error

2 5 10 2 5 10

0.41* 34.63** 0.16 0.11* 36.38** 0.20

2

0.76* 13.81** 0.37 0.67* 17.25** 0.38

* Nonsignificant at P # .05. ** Significantly greater than the error mean square at P , .01.

ent plant (P1) had a greater value than the erect parent plant. In contrast the erect parent plant had a greater value in stem thickness than the prostrate parent plant. Weighted generation mean analysis was done using the QBASIC program (MirzaieNodoushan 1993) in which the generation variances were weighted and analyzed following the procedure presented by Mather and Jinks (1982). The genetic analysis was done in three separate stages with three different models: the three-parameter model, or additive-dominance model, the six-parameter model, and the best-fit model. Estimating the number of genes. Estimating the number of genes contributing to the variance of quantitative characters in plant populations is fundamental for the study of mechanisms of heredity. According to Lande (1981), if one line is fixed with alleles decreasing the value of a character of interest and the other line is fixed with alleles increasing its value, the number of genes, or the minimum number of effective factors, can be estimated. The data recorded on these crosses are typical of crosses between populations that differ greatly in some of the quantitative characters as a result of natural or artificial selection. Generation mean analysis on the diverse attributes in the two types provided us with a possibility of estimating the number of genes or factors that control the attributes. The procedure presented by Lande (1981) and elaborated by Cockerham (1986) was used in order to estimate the number of genes and the standard error.

Results The generation means differed significantly in both crosses ( Table 1). The observed generation means and their within-plot variances for the two sets of crosses are


Table 2. Observed generation means, their within-plot variances, and estimated number of genes

Table 3. Gene effects estimated for prostrateness using three- and six-parameter models on means and their variances of parents, F1, F 2, BC1, and BC 2 in the crosses between erect and prostrate plants Cross 2

Cross 1 Generation or parameter

P1 (prostrate) P2(erect) F1 F2 BC1 BC2 No. of genes

a b

Prostrateness (score 1–5)

Stem thickness (mm)

Model

Effect

Estimate

SE

Estimate

SE

Set 1

Set 2

Set 1

Set 2

Three-parameter model

4.95 (0.023)a 1.86 (0.053) 3.46 (0.189) 3.32 (0.548) 3.99 (0.429) 2.57 (0.253) 2.55* (0.37)b

4.93 (0.031) 1.87 (0.051) 3.48 (0.211) 3.39 (0.449) 3.91 (0.333) 2.63 (0.247) 3.27* (0.52)

3.07 (0.153) 4.87 (0.265) 3.79 (0.310) 3.42 (0.441) 3.12 (0.247) 4.00 (0.416) 2.08* (0.83)

2.85 (0.109) 4.55 (0.253) 3.56 (0.200) 3.20 (0.334) 2.82 (0.242) 3.84 (0.376) 2.17* (0.77)

m d h A B C x2 P m d h i j l m d h i j l x2 P

3.387** 1.550** 20.031a 20.430a 20.185a 20.453a 6.924 ,.005 3.569** 1.546** 20.885a 20.162a 20.245a 0.778a 3.412** 1.534** 20.456** 0 0 0.505** 1.45 .5–.25

60.029 60.029 60.068 60.184 60.133 60.325

3.386** 1.514** 20.006a 20.589** 20.091a 20.185a 13.5 ,.005 3.896** 1.531** 21.592* 20.495a 0.498** 1.175** 3.401** 1.531** 20.457** 0 20.444** 0.535** 2.48 .25–.1

60.028 60.028 60.065 60.161 60.126 60.303

Simple scaling test

Joint scaling test Six-parameter model

Best-fit model

Within-plot variance in parentheses. Standard error for the estimated number of genes.

presented in Table 2. Within-plot variances were usually greater for the segregating generations ( F2, BC1, and BC2) than for the parents and F1 generations. The three- and six-parameter and bestfit models are presented in Tables 3 and 4. The results of the simple scaling test showed the presence of nonallelic interactions in both traits. The joint scaling test verified the adequacy of the three parameters: m, d, and h. This test also revealed a lack of goodness-of-fit and the results were in a good agreement with the simple scaling tests for both attributes. Based on the results of the tests for goodness-of-fit and the presence of epistasis, the six-parameter model results were presented and the removal of the nonsignificant components—i, j, and l—caused a considerable reduction in both the standard errors of the remnant components and in the chi-square test results.

Discussion Prostrateness In the first cross, the simple and joint scaling tests indicated the presence of epistasis ( Table 3). In the six-parameter model, additive 3 dominance and additive 3 additive nonallelic interactions (or epistasis) were not significant. Removal of the two components, i and j, led to a fourparameter model. A negative value for h indicated a partial dominance of prostrateness over erectness. The parameters h and l had different signs in both sets of crosses. This indicated duplicate epistasis (Mather and Jinks 1982). In the second cross, both simple and joint scaling tests showed the presence of nonallelic inter-


60.349 60.032 60.829 60.348 60.205 60.508 60.031 60.030 60.194 0 0 60.216

60.316 60.030 60.743 60.314 60.180 60.455 60.030 60.030 60.182 0 60.177 60.206

* Significantly different from zero at 5% probability level. ** Significantly different from zero at 1% probability level. a Nonsignificant.

action. In the six-parameter model additive 3 additive interaction was not significant ( Table 3). Deleting this interaction from the model greatly increased the precision of estimates, as shown by the reduced standard errors. A remarkable drop in the chi-square value suggested that the five-parameter model explained most of the genetic variation in this trait.

Stem Thickness In the first cross, the simple and joint scaling tests suggested the existence of nonallelic interaction effects. In the sixparameter model, m, d, and i were significant. The removal of additive 3 dominance and dominance 3 dominance nonallelic interactions (or epistasis)—j and l—gave the best-fit model and the pa-

Table 4. Gene effects estimated for stem thickness using three- and six-parameter models on means and their variances of parents, F1, F 2, BC1, and BC 2 in a cross between erect and prostrate plants Cross 2

Cross 1 Model

Effect

Estimate

SE

Estimate

SE

Three-parameter model

m d h A B C x2 P m d h i j l m d h i j l x2 P

3.760** 20.847** 20.276** 20.616** 20.669** 21.820** 31 ,.005 3.432** 20.900** 20.390a 0.535* 0.052a 0.750a 3.005** 20.883** 0.740** 0.927** 0 0 2.26 .25–.1

60.061 60.057 60.113 60.177 60.197 60.347

3.509** 20.856** 20.191* 20.777** 20.428** 21.731** 46 ,.005 3.172** 20.849** 20.289a 0.526* 20.349* 0.697a 2.774** 20.839** 0.759** 0.898** 0.365* 0 2.33 .25–.1

60.052 60.050 60.090 60.154 60.171 60.292

Simple scaling test

Joint scaling test Six-parameter model

Best-fit model

* Significantly different from zero at 5% probability level. ** Significantly different from zero at 1% probability level. a Nonsignificant.

60.323 60.074 60.785 60.315 60.234 60.501 60.153 60.057 60.220 60.172 0 0

60.290 60.063 60.708 60.283 60.210 60.445 60.126 60.063 60.173 60.144 0.210 0

rameters m, d, h, and i gave the best explanation of stem thickness gene functions in this cross ( Table 4). Since the h value is very similar to d, but with negative sign, the function of the genes controlling for thickness of stem seems to be partially dominant to stem thinness. In the second cross, the simple additivedominance model failed to explain gene function, and both the simple and the joint scaling tests indicated the existence of at least one kind of nonallelic interaction effects. The removal of the least important component of the model, l, produced a good fit with the rest of the parameters, that is, m, d, h, i, and j. Also, in this cross the results indicated that the genes controlling thickness are partially dominant to those for thinness of stem. Although in this set j was significant at the 5% level, the other parameters in the best-fit model were very similar to those of cross 1. This confirms that in such a situation—extreme phenotypic classes in two diverse populations—generation mean analysis could be applied in order to investigate the dominance and epistasis relationships of the genes involved.

Minimum Number of Genes The number of effective factors or the minimum number of genes estimated in set 1 and 2 indicated that these attributes are oligogenic (controlled by a few genes; Table 2). The corresponding standard errors for the estimated number of genes were very small. This indicates that when the assumptions for this procedure are approximately satisfied, the estimates of the number of genes would be reasonably accurate. From the Research Institute of Forests and Rangelands, P.O. Box 13185-116, Ministry of Jahade Sazandegi, Tehran, Iran (Mirzaie-Nodoushan), and Massey University (Gordon) and Grassland, Agresearch (Rumball), Palmerston North, New Zealand. q 1999 The American Genetic Association

red clover. Proceedings of XVII International Grasslands Congress. 421–423. Received June 20, 1998 Accepted February 26, 1999 Corresponding Editor: Prem Jauhar

Fr1 (Root Fluorescence) Locus Is Located in a Segregation Distortion Region on Linkage Group K of Soybean Genetic Map W. Jin, R. G. Palmer, H. T. Horner, and R. C. Shoemaker We report the use of bulked segregant SSR analysis for rapid identification of DNA markers linked to the Fr1 locus in soybean. Pooled DNA extracts from 10 homozygous Fr1 Fr1 and 10 fr1 fr1 F2 plants, derived from a msMOS 3 Minsoy cross, were analyzed using 65 SSR markers. Five SSRs produced repeatable polymorphisms between paired bulks. Linkage with the Fr1 locus was tested using these five SSR primers and DNA from individual plants of each bulk. DNA polymorphisms generated by these five primers were linked to the Fr1 locus. Linkage of SSR loci with the Fr1 locus was verified by using an F2 population segregating for Fr1. The five SSR markers and Fr1 are on linkage group K of the USDA ARS/ISU molecular genetic map. The markers flanking Fr1 are Satt337 (11.0 cM) and Satp044 (0.6 cM). Fr1 previously was mapped on linkage group 12 of the classical genetic map. Thus classical genetic linkage group 12 has been correlated to linkage group K of the molecular genetic map. Six SSR markers were chosen on linkage group K to test the segregation ratio. All six SSRs tested were skewed toward the Minsoy genotype, one chi-square value was statistically significant. This suggested that a gametophyte factor may lie in the region close to Fr1 and most likely close to Satt046.

References Cockerham CC, 1986. Modifications in estimating the number of genes for a quantitative character. Genetics 114:659–664. Lande R, 1981. The minimum number of genes contributing to quantitative variation between and within populations. Genetics 99:541–553. Mather K and Jinks JL, 1982. Biometrical genetics. New York: Chapman & Hall. Mirzaie-Nodoushan H, 1993. Quantitative genetics of prostrateness and other related attributes in red clover (Trifolium pratense L.) (PhD dissertation). Palmerston North, New Zealand: Massey University. Smith RS and Bishop DJ, 1993. Astred—a stoloniferous

Soybean root fluorescent mutants are important in characterizing germplasm diversity ( Delannay and Palmer 1982), in tissue culture (Roth et al. 1982), and in genetic linkage studies ( Devine et al. 1993; Griffin et al. 1989; Palmer and Chen 1998). Five loci controlling root fluorescence have been reported ( Delannay and Palmer 1982; Sawada and Palmer 1987) and the genomic locations of several of these loci have been defined. Fr2 was placed on a

molecular genetic map approximately 6.5 cM from RFLP markers pBLT 73 and 6 cM from pBLT 42 ( Devine et al. 1993). Fr1 and Fr3 have been located on the classical genetic map. Fr1 is 41 cM distant from the Ep (seed coat peroxidase level) locus on classical linkage group 12 (Griffin et al. 1989). Fr1 is located on U1c linkage group (Mansur et al. 1996); however, no detailed data were described in the article. Fr3 is mapped on classical linkage group 9 (Palmer and Chen 1998). The linkage relationships of the Fr3, Fr4, and Fr5 loci on molecular genetic maps are unknown. A classical genetic map of soybean has been developed with great effort over many years (Palmer and Hedges 1993). Integration of this map with the soybean molecular map would be useful for designing breeding strategies for soybean improvement and using the molecular map as a bridge to obtain a comprehensive map of important genes (Shoemaker and Specht 1995). In soybean, diversity at the DNA sequence level is low ( Keim et al. 1989). This has made the construction of an RFLP-based genetic map difficult and tedious. Recently, simple sequence repeat (SSR) markers have been developed (Akkaya et al. 1995; Weber and May 1989). Because SSRs are highly polymorphic, stable, and simple, they have great potential for molecular mapping. The combination of SSR polymorphism and bulked segregant analysis (Michelmore et al. 1991) will facilitate the identification of markers that are tightly linked to the gene of interest. Segregation distortion has been reported in wide crosses of rice ( Xu et al. 1997), maize (Pereira and Lee 1995), barley ( Devaux et al. 1995), common bean (Paredes and Gepts 1995), Aegilops tauschii ( Faris et al. 1998), and soybean ( Honeycutt et al. 1990). Many instances of segregation distortion have been reported through studies of isozymes ( Ishikawa et al. 1987a,b; Wu et al. 1988) and RFLP alleles (McCouch et al. 1988; Saito et al. 1991). The mechanisms of segregation distortion are not well understood, but they may result from action of gametophytic factors, competition among gametes, or from the abortion of male or female gametes. Studies by Nakagahra (1972, 1986) localized gametophytic gene loci on chromosome 3 based on a locus responsible for the partial or total elimination of gametes carrying one of the alleles at that locus. Faris et al. (1998) reported that there are segregation distortion loci on chromosomes 1D, 3D, 4D, 5D, and 7D of Ae. tauschii.


In the present study we describe the application of bulked segregant analysis as a method for rapidly identifying SSR markers linked with the Fr1 locus. Our goals were to place Fr1 on the molecular genetic map and thus to associate classical linkage group 12 with a molecular linkage group.

Table 1. Segregation of the Fr1 (root fluorescence) locus and six simple sequence repeat markers in an F 2 soybean population from msMOS 3 Minsoy

Material and Methods Plant Material msMOS (a male-sterile line obtained from Midwest Oilseed, Adel, Iowa; Fr1 Fr1) as a female parent, was crossed with Minsoy (fr1 fr1) in Ames, Iowa, in the summer of 1995. The resulting F1 seeds were planted at the University of Puerto Rico, Iowa State University Soybean Nursery, Isabela Substation, Isabela, Puerto Rico. An F2 population of 111 individuals was used to test linkage between SSR markers and the Fr1 locus. F2 seeds were planted on germination paper for scoring root fluorescence. Fluorescence was observed by irradiating roots with ultraviolet light ( Delannay and Palmer 1982). After scoring, F2 seedlings were transferred to a growth chamber, leaf material was harvested, and DNA was isolated. F2 plants were single-plant threshed, and F2 seeds were planted in Puerto Rico to generate the F3 generation. F3 seeds were planted on germination paper for scoring root fluorescence to determine F2 genotypes (Fr1 Fr1, Fr1 fr1, fr1 fr1). DNA Extraction and Bulk DNA Preparation DNA isolation was conducted as described by Keim et al. (1988). Equal quantities of DNA were combined from each of 10 homozygous F2 Fr1 Fr1 plants and from each of 10 homozygous F2 fr1 fr1 plants to obtain two DNA bulks. SSR/BSA (Bulked Segregation Analyses) SSRs were developed and initially placed on linkage groups as reported by Cregan et al. (in press). Primer sequences specific to each SSR can be obtained through Soybase, the soybean genome database at http://129.186.26.94. For SSR/BSA analyses, PCR amplifications were carried out in a total volume of 20 ml containing 60 ng of soybean genomic DNA, 1.5 mM Mg21, 0.3 mM of sense and antisense primers, 200 mM of dNTPs, 13 PCR buffer, and 2.5 units of Taq DNA polymerase. Cycling consisted of 30 s at 948C, 30 s at 478C, and 30 s at 688C for 40 cycles on a Perkin-Elmer 960 thermal cycler. PCR


No. of F2 plants

Observed no.

Trait or markers

AAa

HHa

BBa

x2 (1:2:1)

Satt137 Satt055 Satt337 Satt046 Fr1 Satp044 Satt326

78 80 80 79 80 80 78

13 12 12 10 14 13 14

40 42 41 41 42 43 43

25 26 27 28 24 24 21

3.75 4.10 5.67 8.30b 2.70 3.70 2.07

Genotypes: AA 5 msMOS; HH 5 heterozygous; BB 5 Minsoy. b Significant deviation from 1:2:1 ratio at 5% level.

a

products were run on 2.5% Metaphor ( FMC) agarose gel in TBE (0.089 M Trisborate, 0.089 M boric acid, 0.002 M EDTA) buffer with ethidium bromide incorporated into the gel. Initial screening was performed on samples of Fr1 Fr1 (msMOS), fr1 fr1 (Minsoy), pooled DNA from 10 Fr1 Fr1 (pool Fr1) genotypes, and pooled DNA from 10 fr1 fr1 (pool fr1) genotypes. Linkage Analysis The segregation ratios of fr1 and each molecular marker in the F2 population were tested for fitness to a 1:2:1 genotypic ratio by chi-square test. After finding that three SSRs (Satt055, Satp044, Satt337) showed segregating excess BB class and less AA class, six more SSR markers (Satt137, Satt247, Satt046, Satp116, Satt326, Satt001) from linkage group K were chosen to investigate segregation ratios. Three (Satt137, Satt046, Satt326) out of these six markers showed polymorphisms and were analyzed further. The MapMaker program ( Lander et al. 1987) was used to construct a linkage map by using F2 segregation data. A logarithm of odds ratio ( LOD) score of 3 was used as the lower limit for accepting linkage between two markers. Distances between markers were calculated in centiMorgans (cM) derived with the Kosambi function ( Kosambi 1944).

Results The root fluorescence test on the F3 progenies, derived from 80 F2 plants, yielded 14 homozygous Fr1 Fr1, 42 segregating, and 24 homozygous fr1 fr1 families ( Table 1). This segregation fit the expected ratio of 1:2:1 (x2 5 2.7). Ten homozygous Fr1 Fr1 and 10 homozygous fr1 fr1 F2 plants were selected for preparation of the two DNA bulks.

Figure 1. SSR markers detecting polymorphisms between DNA bulks of genotypes Fr1 Fr1 and fr1 fr1. The first two lanes contain parental DNA from msMOS (Fr1 Fr1) and Minsoy (fr1 fr1). The third lane contains bulked DNA from the homozygous Fr1 Fr1 individuals, and the fourth lane contains bulked DNA from the homozygous fr1 fr1 individuals.

In total, 65 SSR markers distributed on the 23 linkage groups of the soybean molecular genetic map (Cregan et al., in press) were tested with DNA from Fr1 Fr1 parent, fr1 fr1 parent, a pool of 10 Fr1 Fr1 genotypes, and a pool of 10 fr1 fr1 genotypes ( Figure 1). Of these, 4 (6%) did not produce any amplification product and were not analyzed further. The pool size of 10 genotypes for BSA analyses was determined on the basis of detection limit, desired interval of the genome to be covered (Giovannoni et al. 1991; Michelmore et al. 1991), and our population size. Six of the remaining 65 SSRs detected polymorphisms between pooled Fr1 and fr1 DNAs. These polymorphisms were repeated to verify reproducibility. Primers for these six SSRs were used to amplify DNA from each of the 80 F2 progeny. Again, all six markers tested were skewed toward the Minsoy genotype, and one chi-square value was statistically significant ( Table 1, Figure 2). In all cases, fewer msMOS gametes were transmitted than expected. Segregation analysis with 80 F2 progenies determined that these two SSRs (Satp044 and Satt337) were linked to the Fr1 locus ( Figure 3) with LOD scores of 30 and 16, respectively. The two markers flanking Fr1 locus were Satp044 and Satt337. The map locations of the two flanking loci agreed with the soybean molecular genetic map (Cregan et al., in press). The fragment amplified by Satp044 in the Fr1 Fr1 bulk and Fr1 Fr1 parent was present in all 14 Fr1 plants and was not present in any fr1 individuals. The fragment amplified in the

Figure 2. Effect of map distance (cM) from Satt046 locus on segregation ratios (chi square). * indicates chi-square value of Fr1.

fr1 bulk and fr1 parent was present in all 24 homozygous fr1 fr1 plants. For Satt337, the fragment amplified in the Fr1 Fr1 bulk and the Fr1 Fr1 parent was present in 13 of 14 plants scored as Fr1 Fr1 and was present in three of 24 plants scored as fr1 fr1. The fragment amplified in the fr1 fr1 bulk and fr1 fr1 parent was present in 23 of 24 plants scored as fr1 fr1. The closest linkage (0.6 cM) was found with SSR marker Satp044. The SSR profiles of some of the individuals amplified with Satp044 primers ( Figure 4). The two SSR markers flanking the Fr1 locus are on linkage group K in USDA ARS/ISU map (Cregan et al., in press).

Discussion Mapping genes with molecular markers can be laborious and costly. One of the most time-consuming aspects of mapping, the screening of the entire mapping population with every probe or primer to be tested, can be eliminated by applying bulked segregant analysis (Michelmore et al. 1991). Bulked segregant analysis simultaneously provides information on polymorphisms between the parents and possible linkage between a marker and a

Figure 3. Position of the Fr1 locus on molecular genetic linkage group of soybean. (A) A part of the soybean genetic group K showing the linkage map around the Fr1 locus obtained from analysis of 80 F2 plants derived from the cross between msMOS (Fr1 Fr1) and Minsoy (fr1 fr1). (B) A linkage map of linkage group K constructed from the Glycine max 3 Glycine soja population (Shoemaker et al. 1996). Linkage group K extends beyond the hash marks.

targeted gene. This process can reduce cost by several-fold, particularly when used with PCR-based techniques such as SSR and RAPD analyses. Bulked segregant analysis combined with RAPD also has been used to detect markers linked to many traits including disease resistance genes (Mouzeyar et al. 1995; Yaghoobi et al. 1995), a male-sterile gene in rice ( Zhang et al. 1994), and a fertility restoration gene in rapeseed ( Delourme et al. 1994). To our knowledge this is the first

report using bulked segregant analysis with SSR for molecular mapping in plants. SSR/BSA efficiently identified markers linked to the gene of interest and allowed their rapid placement on a genetic map. In this study the combination of SSR markers and bulked segregant analysis was faster and less expensive than using RFLPs to find markers flanking the Fr1 locus. We tested 61 polymorphic SSR markers with pools of DNA from 10 Fr1 Fr1 and 10 fr1 fr1 genotypes and found that the Fr1 locus mapped to linkage group K of the USDA ARS/ISU map and was flanked by two SSR markers, Satt337 and Satp044. There were consistently skewed segregation ratios within an interval on linkage group K ( Table 1). Based on chi-square analysis of SSR segregation data, chisquare values peaked in the interval between Satt046 and Satt337 ( Figure 2). This suggests the presence of a gametophytic factor (Redei 1965; Li 1967; Gonella and Peterson 1975; Rashid and Peterson 1992) in the interval affecting transmission of the msMOS gametes. Segregation distortion is expected to be more extreme if the distance between a marker locus and the gametophytic factor gene is very close. Development of the classical genetic map has proceeded slowly in soybean because it is difficult to make crosses and generate large numbers of hybrid seeds, and there are few cytogenetic markers and low genetic variation in the germplasm ( Keim et al. 1989). To exploit fully the potential of a molecular genetic map, molecular and conventional markers must be integrated into one linkage map. Integration of the maps can also be pursued by screening near-isogenic lines ( NIL) ( Young et al. 1988). According to Shoemaker and Specht (1995), about half of the 19 soybean classic linkage groups have been associated with their corresponding molecular genetic linkage groups. But classical linkage group 12 has not been asso-

Figure 4. SSR profiles of some of the F2 individual genotypes of the pools with Satp044. * indicates recombinants. P1 5 parent 1 (msMOS), P2 5 parent 2 (Minsoy).


ciated with any of the molecular genetic linkage groups. Fr1 is known to reside on classical linkage group 12 (Griffin et al. 1989). Our results have shown that the Fr1 locus is located on molecular genetic linkage group K. Therefore classical linkage group 12 is now integrated into molecular genetic linkage group K. From the Interdepartmental Plant Physiology Program and Department of Botany (Jin and Horner), and the Departments of Agronomy and Zoology/Genetics and the USDA Agricultural Research Service CICGR (Palmer and Shoemaker), Iowa State University, Ames, IA 50011. We gratefully acknowledge Drs. David Grant and Arla Bush for helpful discussions during the course of this project, and Drs. David Grant, John Imsande, and Joanne Labate for critically reviewing the manuscript. This is a joint contribution: journal paper J-17701 of the Iowa Agriculture and Home Economics Experiment Station, Ames, Iowa (projects 3236 and 3352) and the U.S. Department of Agriculture, Agricultural Research Service, Corn Insects and Crop Genetics Research Unit. The mention of a trademark or proprietary product does not constitute a guarantee or warranty of the product by Iowa State University or the USDA and does not imply its approval to the exclusion of other products that may also be suitable. Address correspondence to H. T. Horner at the address above. q 1999 The American Genetic Association

References

Ishikawa R, Kinoshita T, and Morishima H, 1987b. Trisomic analysis of genes for isozymes: location of Cat-1, Acp-1 and Pox-2 on chromosomes. Rice Genet Newslett 4:75–76. Keim P, Olson TC, and Shoemaker RC, 1988. A rapid protocol for isolating soybean DNA. Soybean Genet Newslett 15:150–152. Keim P, Shoemaker RC, and Palmer RG, 1989. Restriction fragment length polymorphism diversity in soybean. Theor Appl Genet 77:786–792. Kosambi DD, 1944. The estimation of map distances from recombination values. Ann Eugen 12:172–175. Lander ES, Green P, Abrahamson J, Barlow A, Daly MJ, Lincoln SE, and Newburg L, 1987. MAPMAKER: an interactive computer program for constructing genetic linkage maps of experimental and natural populations. Genomics 1:174–181. Li SL, 1967. A new segregation distorted factor in Arabidopsis. Arabidopsis Inf Serv 4:5–6. Mansur LM, Orf JH, Chase K, Jarvik T, Cregan PB, and Lark KG, 1996. Genetic mapping of agronomic traits using recombinant inbred lines of soybean. Crop Sci 36: 1327–1336. McCouch SR, Kochert G, Yu ZH, Wang ZY, Khush GS, Coffman WR, and Tanksley SD, 1988. Molecular mapping of rice chromosomes. Theor Appl Genet 76:815– 829. Michelmore RW, Paran I, and Kesseli RV, 1991. Identification of markers linked to disease-resistance genes by bulked segregation analysis: a rapid method to detect markers in specific genomic regions by using segregating populations. Proc Natl Acad Sci USA 88:9828– 9832.

Akkaya MS, Shoemaker RC, Specht JE, Bhagwat AA, and Cregan PB, 1995. Integration of simple sequence repeat DNA markers into a soybean linkage map. Crop Sci 35: 1439–1445.

Mouzeyar S, Roeckel-Drevet P, Gentzbittel L, Philippon J, Tourvieille De Labrouhe D, Vear F, and Nicolas P, 1995. RFLP and RAPD mapping of the sunflower PI1 locus for resistance to Plasmopara halstedii race 1. Theor Appl Genet 91:733–737.

Cregan PB, Jarvik T, Bush AL, Shoemaker RC, Lark KG, Kahler AL, VanToai TT, Lohnes DG, Chung J, and Specht JE, in press. An integrated genetic linkage map of the soybean. Crop Sci.

Nakagahra M, 1972. Genetic mechanism of the distorted segregation of marker genes belonging to the eleventh linkage group in cultivated rice. Jpn J Breed 22: 232–238.

Delannay X and Palmer RG, 1982. Four genes controlling root fluorescence in soybean. Crop Sci 22:278–281.

Nakagahra M, 1986. Geographic distribution of gametophyte genes in wide crosses of rice cultivars. In: Rice genetics. Manila, Philippines: IRRI; 73–82.

Delourme R, Bouchereau A, Hubert N, Renard M, and Landry BS, 1994. Identification of RAPD markers linked to a fertility restorer gene for the Ogura radish cytoplasmic male sterility of rapeseed (Brassica napus L.). Theor Appl Genet 88:741–748. Devaux P, Kilian A, and Kleinhofs A, 1995. Comparative mapping of the barley genome with male and female recombination-derived, doubled haploid populations. Mol Gen Genet 249:600–608. Devine TE, Weisemann JM, and Matthews BF, 1993. Linkage of the Fr2 locus controlling soybean root fluorescence and four loci detected by RFLP markers. Theor Appl Genet 85:921–925. Faris JD, Laddomada B, and Gill BS, 1998. Molecular mapping of segregation distortion loci in Aegilops tauschii. Genetics 149:319–327. Giovannoni JJ, Wing RA, Ganal MW, and Tanksley SD, 1991. Isolation of molecular markers from specific chromosomal intervals using DNA pools from existing mapping populations. Nucleic Acids Res 19:6553–6558.

Palmer RG and Chen XF, 1998. Assignment of the Fr3 locus to soybean linkage group 9. J Hered 89:181–184. Palmer RG and Hedges BR, 1993. Linkage map of soybean [Glycine max ( L.) Merr.] (2n 5 40). In: Genetic maps (O’Brien SJ, ed). Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press; 6.136–6.148. Paredes OM and Gepts P, 1995. Segregation and recombination in inter-gene pool crosses of Phaseolus vulgaris L. J Hered 86:98–106. Pereira MG and Lee M, 1995. Identification of genomic regions affecting plant height in sorghum and maize. Theor Appl Genet 90:380–388. Rashid A and Peterson PA, 1992. The RSS system of unidirectional cross-incompatibility in maize: I. Genetics. J Hered 83:130–134. Redei GP, 1965. Non-Mendelian megagametogenesis in Arabidopsis. Genetics 51:857–872.

Gonella J and Peterson PA, 1975. Gametophytic factor (ga 10) on chromosome 5 distal to A2. Maize Genet Coop Newslett 4:71–73.

Roth EF, Weber G, and Lark KG, 1982. Use of isopropylN (3-chlorophenyl) carbamate (CIPC) to produce partial haploids from suspension cultures of soybean (Glycine max). Plant Cell Rep 1:205–208.

Griffin JD, Broich SL, Delannay X, and Palmer RG, 1989. The loci Fr1 and Ep define soybean linkage group 12. Crop Sci 29:80–82.

Saito A, Shimosaka E, Hayano Y, Saito K, Matuura S, and Yano M, 1991. Molecular mapping on chromosome 3 of rice. Jpn J Breed 41:158–159.

Honeycutt RJ, Newhouse KE, and Palmer RG, 1990. Inheritance and linkage studies of a variegated leaf mutant in soybean. J Hered 81:123–126.

Sawada S and Palmer RG, 1987. Genetic analyses of nonfluorescent root mutants induced by mutagenesis in soybean. Crop Sci 27:62–65.

Ishikawa R, Sato TI, and Morishima H, 1987a. Abnormal segregation of isozyme genes in the Indica 3 Japonica crosses of the common rice. Jpn J Breed 37:188–189.

Shoemaker RC and Specht JE, 1995. Integration of the soybean molecular and classical genetic linkage groups. Crop Sci 35:436–446.


Shoemaker R, Polzin K, Lorenzen L, and Specht J. 1996. Molecular mapping in soybean. In: Soybean: genetics, molecular biology and biotechnology ( Verma DPS and Shoemaker RC, eds). Wallingford, UK: CAB International; 37–56. Weber JK and May PE, 1989. Abundant class of human DNA polymorphisms which can be typed using the polymerase chain reaction. Am J Hum Genet 44:388– 397. Wu KS, Glaszmann JC, and Khush GS, 1988. Chromosomal locations of ten isozyme loci in rice (Oryza sativa L.) through trisomic analysis. Biochem Genet 26: 303–320. Xu Y, Zhu L, Xiao J, Huang N, and McCouch SR, 1997. Chromosomal regions associated with segregation distortion of molecular markers in F2, backcross, doubled haploid, and recombinant inbred populations of rice (Oryza sativa L.). Mol Gen Genet 253:535–545. Yaghoobi J, Kaloshian I, Wen Y, and Williamson VM, 1995. Mapping a new nematode resistance locus in Lycopersicon peruvianum. Theor Appl Genet 91:457–464. Young ND, Zamir D, Ganal MW, and Tanksley SD, 1988. Use of isogenic lines and simultaneous probing to identify DNA markers tightly linked to the Tm-2a gene in tomato. Genetics 120:579–585. Zhang QF, Shen BZ, Dai XK, Mei MH, Saghai Maroof MA, and Li ZB, 1994. Using bulked extremes and recessive class to map genes for photoperiod-sensitive genic male sterility in rice. Proc Natl Acad Sci USA 91:8675– 8679. Received September 14, 1998 Accepted February 18, 1999 Corresponding Editor: Fredrick Bliss

Quantitative Genetics of Leaf Morphology in Crepis tectorum ssp. pumila (Asteraceae) S. Andersson The present study of Crepis tectorum ssp. pumila employed the factorial mating design to examine the genetic (co)variance structure of leaf morphology in a population that appears to be homozygous for a major gene causing deeply lobed leaves. One set of factorials was based on widely dispersed plants (different patches, mean crossing distance 24 m), while the second was based on crosses between plants within 0.25 m2 plots (within-patch crosses). Variance component analyses revealed significant levels of additive variation in leaf size and four measures of leaf shape, indicating a potential for further adaptive change—even after the fixation of a major mutation. Offspring from short-range crosses had a significantly shorter leaf perimeter and were more likely to produce weakly lobed leaves (a recessive character) than progeny representing the longer crossing distances, suggesting that this patchily distributed population is subdivided into patches of related individuals.

However, the crossing distance between parents had no consistent effect on the additive and nonadditive genetic variances, and the patterns of association between characters, showing that spatially restricted samples of genotypes capture most of the (co)variation present in the population as a whole. Extreme differences in morphology can be governed by a small number of major genes (Gottlieb 1984), a factor that may increase the rate at which new adaptations evolve in natural populations, particularly if the novel, advantageous allele is dominant ( Haldane 1924). Whether the population can evolve the optimum phenotype also depends on the presence of genes with smaller effects on the phenotype. If some of the variation is polygenic, a novel morphological feature can be refined by selection after the fixation of a major mutation (Gottlieb 1984). To test this possibility, it is necessary to search for particular genes that influence the expression of a major gene (Schat et al. 1993; Smith and Macnair 1998) or to assess patterns and amounts of polygenic variation in populations that have become fixed for a major mutation. Estimates of the genetic (co)variances are valid only for the population in which they have been measured, and most analyses require that this ‘‘base population’’ is close to equilibrium (no inbreeding or linkage) ( Falconer and Mackay 1996; Mitchell-Olds and Rutledge 1986). Hence it may be difficult to infer the response to selection from genetic (co)variances when the source population consists of patches with related or locally adapted individuals, as may be the case in plants and other sedentary organisms ( Epperson 1990; Schaal and Levin 1978; Schemske 1984; Waser 1993). While population structure is a central concern in empirical studies of single-locus diversity ( Epperson 1990), only a few authors have considered the scale of sampling when estimating genetic (co)variances ( Kelly 1993; Tonsor and Goodnight 1997; Waser et al. 1995). Populations of the diploid (2n 5 8), annual plant Crepis tectorum L. (Asteraceae) have diverged for a large number of adaptive traits, particularly in the Baltic area, where the relatively uniform weed ecotype (ssp. tectorum) grades into a series of morphologically distinct rock outcrop forms that are regarded as derived within the species (Andersson 1990). The most distinct taxon, C. tectorum ssp. pumila ( Liljebl.) Sterner, occurs on shallow soil in a

distinct zone around bare rock on the calcareous grasslands (‘‘alvars’’) on the Bal¨ land (southeast Sweden). tic island of O This endemic subspecies is self-sterile (Andersson 1989a) and has deeply divided rosette leaves with leaflets separated to the midrib, contrasting with the simple leaves of populations adapted to more shaded habitats (Andersson 1989b, 1990, 1991a). Seed dispersal is facilitated by the presence of a pappus, but the potential for long-distance gene dispersal may be limited, given the short stature of the plants (Andersson 1990) and the dense grass sward (Festuca ovina, Agrostis stolonifera) surrounding patches of C. tectorum ssp. pumila (personal observation). Populations of C. tectorum ssp. pumila continue to exhibit polygenic variation in leaf shape (Andersson 1991b), despite evidence for monogenic inheritance in interpopulation crosses (Andersson 1991a, 1995) and strong selection at the phenotypic level (Andersson 1992). The aims of the present investigation were twofold: to partition the within-population variation in leaf morphology into additive and nonadditive genetic components, and to compare the quantitative genetic partitioning of variance over two spatial scales. One set of crosses was based on widely dispersed plants (different patches), as in the previous study (Andersson 1991b), while the second was based on crosses within 0.25 m2 plots (within patches). The results are used to discuss the potential for further adaptive change in a population that appears to be fixed for a major gene causing deeply lobed leaves (Andersson 1995).

Materials and Methods Seeds were collected from about 10 plants in each of eight 0.5 m 3 0.5 m plots scattered over an area of 30 m 3 50 m in a patchily distributed population approximately 1.5 km south of Vickleby ( hereafter referred to as the ‘‘Vickleby population’’). The field plots were established at sites where numerous plants of Crepis were found earlier in the summer. Hence the scale of sampling was determined by the spatial distribution of plants in the Vickleby population. Phenotypic data from another study (Andersson 1992) indicate extensive spatial variation in leaf shape over the area from which the parents were sampled. A factorial crossing experiment ( North Carolina Design II; Comstock and Robinson 1948) was performed to assess patterns of (co)variation in leaf morphology.

Five seedlings from each plant were planted individually in pots and assigned to random positions on a greenhouse bench. Once plants began to flower, each family was split into two groups of two or three individuals, and the two groups placed on different benches in the same greenhouse. Within-plot crosses were carried out by reciprocally mating a group of three to five plants from each plot to a different group of three to five plants from the same plot; biparental inbreeding was minimized by restricting matings to plants in different maternal families. This procedure resulted in seven replicate factorials ( hereafter ‘‘blocks’’), each corresponding to a plot in the field. One block was omitted due to low crossing success. Between-plot crosses were performed in a similar way, except that each block represented a mixture of plants from different plots ( Figure 1). Based on the success of each cross ( Figure 1) and the spatial distribution of the field plots, the physical distance between parents in the between-plot experiment ( hereafter ‘‘crossing distance’’) ranged from 10 to 42 m (mean 24 m), contrasting with the much shorter crossing distances in the within-plot treatment (range 0–0.7 m). Six seeds per cross were planted individually into 25 cm2 cells (one seed per cell) in a series of plastic flats (connected to reduce edge effects) on a large greenhouse bench. Seeds from a given cross were randomized across the whole ‘‘planting area.’’ Plants were given supplementary light (12 h/day) to minimize spatial variation in light intensity. In late December, when all plants had attained maximum leaf dissection (cf. Andersson 1989b), one leaf from each plant was pressed. The sampling procedure was standardized by collecting the youngest fully developed leaf from each individual. All leaves were collected within a 2-day period. The dried leaves were digitized with a video camera connected to a Macintosh computer, after which the program OPTILABTM was used to score each leaf for five traits: leaf length, relative leaf perimeter ( length of the outer contour divided by the square root of the area), relative leaf width (ratio of maximum leaf width to leaf length), ‘‘center of gravity’’ (the distance from the leaf apex to the point dividing the leaf into two equal areas, divided by leaf length), and a ‘‘leaf dissection index’’ (ratio of smallest width between two lobes to maximum width). Leaf length and center of gravity were approximately normally


likelihood procedures (using the REML option in SAS PROC VARCOMP) because of the unbalanced experimental design. These analyses partitioned the total variance in each character into variance attributable to block (Vblock), sire and dam within block (Vsire, Vdam), the interaction between sire and dam (Vsire3dam), and differences between plants within full-sib families (Vwithin), all considered random effects. Approximate 95%, 99%, and 99.9% confidence intervals (CIs) of each estimate were computed from the (co)variances for the variance components obtained from the PROC VARCOMP output. The heritability (h2), a standardized measure of the additive genetic variance, was estimated as four times the fraction of variance attributed to sire or dam, and related to the corresponding parameter for the nonadditive component (D), estimated as four times the fraction of variance attributed to the sire 3 dam interaction ( Falconer and Mackay 1996). Paternal half-sib (sire) means were calculated to assess the effect of crossing distance on mean phenotype and on patterns of genetic correlation between leaf characters. To compare individual values of the correlation coefficient, bootstrap procedures with 1,000 resamplings were used to construct approximate CIs for each estimate, using the percentile method ( Efron and Tibshirani 1986). A similar approach was employed to compare the overall level of association in the two matrices, calculated by averaging the absolute values of all correlation coefficients. Permutation procedures ( Dietz 1983) were applied to examine whether corresponding elements were correlated or random with respect to one another. Some plants had exceptionally large values for the leaf dissection index (weakly lobed leaves), resulting in a frequency distribution with a tail drawn out to the right ( Figure 2). The frequency of ‘‘outliers’’ (classified as plants with a leaf dissection index exceeding 0.20) in the two treatment groups was compared using a contingency test.

Results Figure 1. Designs for the two crossing experiments. Rows and columns represent sets of parent plants derived from particular plots in the source area (numbered 1–8). Plants from a given plot represent different maternal families. The symbols indicate whether a cross was successful in one (x) or both (squares) directions.

distributed, while the other characters were log-transformed to achieve normality and homogeneity of variances. A total of 1,060 plants from the between-plot experiment was scored for the five leaf traits,


while the number of progeny in the withinplot category was 1,010. The number of offspring per cross ranged from three to six (variable germination rate). The data were analyzed with maximum

Offspring from between-plot crosses had a significantly longer leaf perimeter than those from within-plot crosses, while differences in the mean failed to reach significance for other characters ( Table 1). Comparison of family-mean correlation matrices ( Table 2) revealed great similarities in patterns of association (r 5 0.92, p

Table 1. Means and standard deviations (SD) for plants derived from crosses within and between plots in a population of C. tectorum ssp. pumila Between plot

Within plot Character Leaf length (mm) Center of gravity Relative leaf perimeter Relative leaf width Leaf dissection index a b

83.5 0.415 1.51 20.688 21.14

SD

Mean

9.11 0.0188 0.0741 0.0439 0.158

83.8 0.418 1.54 20.683 21.16

SD

ta

7.19 0.0156 0.0523 0.0439 0.104

0.249 NS 1.536 NS 3.296b 1.068 NS 1.427 NS

The t value refers to the difference between treatment means. All analyses based on sire means (N 5 163–172). Significance level: p 5 .001.

than for other characters (0.252–0.384). The nonadditive component (D) was more pronounced for degree of leaf dissection than for the other characters ( Table 3). There was a higher frequency of outliers in the within-plot crosses (3.6%) than in the between-plot crosses (1.8%), a significant difference according to a G test of data pooled over families (G 5 6.36, p 5 .012). A similar trend was found when data were collapsed into the number of families with and without outliers (G 5 4.26, p 5 .039). Exclusion of outliers had negligible effects on the results of the variance component analyses (data not shown).

Figure 2. Histogram showing the distribution of leaf dissection in the offspring generation (N 5 2,070, more deeply lobed leaves toward the left).

5 .009, matrix permutation test). Differences between corresponding elements in the matrices were too small to be declared significant (overlapping 95% CI), a conclusion that also applies to the average strength of association in the within-plot matrix (|r| 5 0.376, CI 5 0.296–0.463) and the between-plot matrix (|r| 5 0.375, CI 5 0.295–0.455). Both correlation matrices revealed an association between leaf length, relative leaf perimeter, degree of leaf dissection, and center of gravity (the distance from the leaf apex to the point dividing the leaf into two equal areas). Relative leaf width was positively correlated with degree of leaf dissection, but varied independently of the other traits (Table 2). According to the variance component analyses ( Table 3), the within-family component (Vwithin) always accounted for most of the variation in leaf morphology, contrasting with the low variance attributed to block (Vblock) and the low to moderate variance arising from differences between sires (Vsire) and dams (Vdam), and the sire 3 dam interactions (Vsire3dam). The sire and dam estimates were similar and had overlapping 95% CIs for all traits, despite sizable differences in some cases (degree of leaf dissection in the within-plot crosses). These components generally exceeded the sire 3 dam interaction variance, but differences were too small to be declared significant (p . .05). There was no tendency for variance components to be higher (or lower) for plants derived from betweenplot crosses than for plants derived from within-plot crosses (overlapping 95% CI in all cases). Characters differed slightly in the size of the additive component, with higher h2 for center of gravity (0.503– 0.540) and relative leaf width (0.415–0.481)

Mean

Discussion Quantitative Genetics of Leaf Morphology Finely dissected leaves represent the derived condition in C. tectorum and probably evolved as a mechanism to minimize transpiration and overheating in dry, exposed, and nutrient-poor habitats (Andersson 1989b, 1991a). Previous studies indicated that major mutations have been important for the origin of dissected-leaf populations of C. tectorum in the Baltic region, that a single dominant gene may be responsible for the finely dissected leaves of C. tectorum ssp. pumila, and that the Vickleby population is fixed for this gene

(Andersson 1995). However, leaf morphology is still a target of selection (Andersson 1992), and the present investigation clearly indicates that further adaptive change is possible, as shown by the significant levels of additive variance in leaf size and four measures of leaf shape. In this context, it is necessary to assume that the genetic variation in leaf morphology is expressed in the field and that the sign of the genetic correlations ( Table 2) and the direction of selection converge for most leaf characters. The presence of dominance variance for degree of leaf dissection is consistent with the detection of nonadditive gene action (dominance) in crosses between C. tectorum ssp. pumila and two simple-leafed populations of the same species (Andersson 1991a, 1995). Since deeply lobed leaves behaved as a dominant character in these crosses, it is tempting to interpret the weakly lobed outliers in the withinpopulation analyses as recessive homozygotes, perhaps at the same major locus as that segregating in hybrid populations. The hypothesis that the Vickleby population is not yet fixed for the major gene conferring deeply lobed leaves needs to be confirmed by more detailed crossing experiments, but agrees with the slow elimination of recessive genes in large, outcrossing populations (Johnson 1976)

Table 2. Sire mean correlations for plants of C. tectorum ssp. pumila

Character Leaf length Center of gravity Relative leaf perimeter Relative leaf width Leaf dissection index

Leaf length

Center of gravity

Relative leaf perimeter

Leaf dissection index

0.20a 0.34**b 0.43* 0.68*** 20.07 20.22 20.68** 20.47**

0.59*** 0.35** 0.27 0.12 20.56*** 20.50***

20.01 20.18 20.64*** 20.48**

20.30** 20.41***

The upper value for each trait combination refers to crosses within plots, the lower to crosses between plots. Corresponding elements had overlapping 95% CI in all cases. b Asterisks denote values significantly different from 0, as determined by the 95%, 99%, and 99.9% CI calculated from 1,000 bootstrap samples. Significance levels: *p , .05, **p , .01, ***p , .001.

a


Table 3. Relative variance components (V), narrow-sense heritabilities (h2), and the proportion of variance attributed to dominance effects (D)

a

Type of cross/character

Vblock

Vdam

Vsire

Vsire3dam

Vwithin

h2

D

Within plots Leaf length Center of gravity Relative leaf perimeter Relative leaf width Leaf dissection index

0.000 0.063 0.019 0.032 0.011

0.060*a 0.132** 0.071* 0.120** 0.141**

0.066* 0.138** 0.062* 0.121** 0.017

0.094* 0.069* 0.048 0.040 0.142*

0.781*** 0.661*** 0.820*** 0.720*** 0.701***

0.252 0.540 0.265 0.481 0.315

0.375 0.275 0.191 0.159 0.567

Between plots Leaf length Center of gravity Relative leaf perimeter Relative leaf width Leaf dissection index

0.019 0.003 0.000 0.047 0.000

0.092*** 0.135** 0.074** 0.102** 0.052

0.100** 0.116** 0.083** 0.106** 0.082*

0.040 0.066* 0.014 0.081* 0.102*

0.768*** 0.682*** 0.829*** 0.712*** 0.764***

0.384 0.503 0.313 0.415 0.268

0.161 0.265 0.057 0.323 0.407

Comparison of 95% CI revealed no significant difference between corresponding values in the two treatment groups. Asterisks denote values significantly different from zero, as determined by the 95%, 99%, and 99.9% CI. Significance levels: *p , .05, **p , .01, ***p , .001.

and the low penetrance of dominant alleles conferring deeply lobed leaves in C. tectorum (Andersson 1995). In this context, it is important to emphasize that the frequency of outliers was too low (,4%) to influence the results of the variance component analyses, showing that major gene segregation accounts for only a minor fraction of the within-population variation. Hence the evolutionary response to current selective forces on leaf morphology (if any) probably involves genes with smaller effects on the phenotype, including genetic factors that stabilize or enhance the expression of the major gene (Schat et al. 1993; Smith and Macnair 1998). Scale Effects on Quantitative Genetic Parameters It may be difficult to infer the response to selection from genetic (co)variances if the estimates depend on the scale of sampling in the parent generation. To test for such scale effects, I compared the quantitative genetic partitioning of variance over two spatial scales in the Vickleby population. As no information on pollen and seed dispersal was available, I simulated two contrasting patterns of gene flow by crossing plants derived from the same 0.25 m2 plot or from different plots separated by more than 10 m (mean crossing distance 24 m). These treatments should span the full range of possible mating distances in the outcrossed C. tectorum ssp. pumila. Offspring from crosses between plants from the same plot had a shorter leaf perimeter and were more likely to produce weakly lobed leaves than progenies derived from wider crosses. Since weakly lobed leaves behave as a recessive character, I suggest that leaf shape responded


to ( biparental) inbreeding in the shortrange crosses (inbreeding depression; Falconer and Mackay 1996) and that the Vickleby population is spatially structured with patches of related individuals, as found in other wild plant species (Waser 1993). However, I found no treatment effect on the level of additive and nonadditive genetic variance. Moreover, the matrix comparisons revealed great similarities in patterns of association between traits, suggesting little influence of crossing distance on the correlation structure. Hence there is no evidence from the present investigation to suggest that the size of the mating pool has a strong effect on the amount of variation available to selection in the offspring generation, or that extreme crossing distances give a distorted view of the genetic (co)variance structure in C. tectorum ssp. pumila. Whether these conclusions also apply to other plants is uncertain, given the paucity of studies relating estimates of genetic (co)variances to the physical distance between plants in the parent generation. Waser et al. (1995) found a weak but suggestive effect of crossing distance on the estimated variance structure for seed set and seed mass in Ipomopsis aggregata, while population structure had a negligible effect on the partitioning of phenotypic variance in a population of Plantago lanceolata (Tonsor and Goodnight 1997). It is possible that mutation and migration rates are sufficiently high to maintain most of the variation within subpopulations that would occur in a panmictic population ( Lande 1991, 1992; see also Whitlock et al. 1993) or that this and previous empirical studies lacked the power to detect scale effects on genetic (co)variances at these levels.

From the Department of Systematic Botany, University ¨ . Vallgatan 18-20, S-22361 Lund, Sweden. I of Lund, O thank Rune Svensson for technical assistance in the garden and Karin Ryde for linguistic advice on the manuscript. Financial support was provided by the Swedish Natural Science Research Council. Address correspondence to Stefan Andersson at the address above or e-mail: [email protected]. q 1999 The American Genetic Association

References Andersson S, 1989a. The evolution of self-fertility in Crepis tectorum (Asteraceae). Plant Syst Evol 168:227– 236. Andersson S, 1989b. Variation in heteroblastic succession among populations of Crepis tectorum. Nordic J Bot 8:565–573. Andersson S, 1990. A phenetic study of Crepis tectorum in Fennoscandia and Estonia. Nordic J Bot 9:589–600. Andersson S, 1991a. Geographical variation and genetic analysis of leaf shape in Crepis tectorum (Asteraceae). Plant Syst Evol.178:247–258. Andersson S, 1991b. Quantitative genetic variation in a population of Crepis tectorum subsp. pumila (Asteraceae). Biol J Linn Soc 44:381–393. Andersson S, 1992. Phenotypic selection in a population of Crepis tectorum ssp. pumila (Asteraceae). Can J Bot 70:89–95. Andersson S, 1995. Differences in the genetic basis of leaf dissection between two populations of Crepis tectorum (Asteraceae). Heredity 75:62–69. Comstock RE and Robinson HF, 1948. The components of genetic variance in populations of biparental progenies and their use in estimating the average degree of dominance. Biometrics 4:254–266. Dietz EJ, 1983. Permutation tests for association between distance matrices. Syst Zool 32:21–26. Efron B and Tibshirani R, 1986. Bootstrap methods for standard errors, confidence intervals, and other measures of statistical accuracy. Stat Sci 1:54–77. Epperson BK, 1990. Spatial patterns of genetic variation within plant populations. In: Plant population genetics, breeding, and genetic resources ( Brown AHD, Clegg MT, Kahler AL, and Weir BS, eds). Sunderland, MA: Sinauer; 229–253. Falconer DS and Mackay TFC, 1996. Introduction to quantitative genetics. New York: Longman. Gottlieb LD, 1984. Genetics and morphological evolution in plants. Am Nat 123:681–709. Haldane JBS, 1924. A mathematical theory of natural and artificial selection. Trans Camb Phil Soc 23:19–41. Johnson C, 1976. Introduction to natural selection. Baltimore: University Park Press. Kelly CA, 1993. Quantitative genetics of size and phenology of life-history traits in Chamaecrista fasciculata. Evolution 47:88–97. Lande R, 1991. Isolation by distance in a quantitative trait. Genetics 128:443–452. Lande R, 1992. Neutral theory of quantitative genetic variance in an island model with local extinction and colonization. Evolution 46:381–389. Mitchell-Olds T and Rutledge JJ, 1986. Quantitative genetics in natural plant populations: a review of the theory. Am Nat 127:379–402. Schaal BA and Levin DA, 1978. Morphological differentiation and neighborhood size in Liatris cylindracea. Am J Bot 65:923–928. Schat H, Kuiper E, Ten Bookum WM, and Vooijs R,1993. A general model for the genetic control of copper tolerance in Silene vulgaris: evidence from crosses between plants from different tolerant populations. Heredity 70:142–147. Schemske DW, 1984. Population structure and local se-

lection in Impatiens ( Balsaminaceae), a selfing annual. Evolution 38:817–832. Smith SE and Macnair MR, 1998. Hypostatic modifiers cause variation in degree of copper tolerance in Mimulus guttatus. Heredity 80:760–768. Tonsor SJ and Goodnight CJ, 1997. Evolutionary predictability in natural populations: Do mating system and nonadditive genetic variance interact to affect heritabilities in Plantago lanceolata? Evolution 51:1773– 1784. Waser NM, 1993. Population structure, optimal outbreeding, and assortative mating in angiosperms. In: The natural history of inbreeding and outbreeding: theoretical and empirical perspectives ( Thornhill NW, ed). Chicago: University of Chicago Press; 173–199. Waser NM, Shaw RG, and Price MV, 1995. Seed set and seed mass in Ipomopsis aggregata: variance partitioning and inferences about postpollination selection. Evolution 49:80–88. Whitlock MC, Phillips PC, and Wade MJ, 1993. Gene interaction affects the additive genetic variance in subdivided populations with migration and extinction. Evolution 47:1758–1769. Received September 14, 1998 Accepted March 31, 1999 Corresponding Editor: William F. Tracy

Cloned Microsatellite Repeats Differ Between 4-Base Restriction Endonucleases M. B. Hamilton and R. C. Fleischer Simple sequence repeat (SSR) loci are an important marker type for population genetic studies despite the limitation that development of novel loci requires construction and screening of genomic DNA libraries. The common practice of size fractioning genomic DNA before cloning could lead to differential representation of SSR loci within genomic libraries. In addition, linkage mapping studies have shown that small numbers of SSR markers are not randomly distributed within the genomes from which they are isolated. From attempts to clone five SSR repeat sequences in two wild plant species we show that the numbers and repeat type of potential SSR markers depend on the restriction endonuclease used to sample the genome when constructing DNA libraries. This observation is consistent with unequal sampling of the genome by different restriction enzymes. However, as a group the five SSR repeat sequences are not associated with a given restriction enzyme, suggesting they are not clumped within the genome. Use of multiple restriction enzymes to construct DNA libraries may help ensure that cloned SSR loci are drawn from diverse locations in the genome, helping to meet the assumption of randomly locat-

ed marker loci required for population genetic inferences. Segments of DNA that contain tandemly repeated sequences of 1–6 bp are referred to as simple sequence repeats (SSRs) or microsatellite loci. SSR loci often have high levels of allelic diversity and heterozygosity and have been employed for genetic mapping, parentage analysis, and population structure studies (e.g., Goldstein and Pollock 1997). The main drawback of SSRs is that the development of novel loci requires considerable effort to construct and screen genomic DNA libraries and optimize flanking primers. As with any genetic marker used for population genetic inferences, it is desirable that SSR loci be unlinked and distributed randomly within and among chromosomes. A random distribution would mean that SSR loci would not be clustered at a given location within chromosomes or on a small number of chromosomes. Two observations predict that small numbers of SSR loci may not be randomly distributed within genomes when cloned using common techniques. First, SSR cloning procedures often use a sample of restriction enzyme digested genomic DNA that falls within the 200–1000 bp size range (e.g., Fleischer and Loew 1995). This process of ‘‘size fractioning’’ the genomic DNA before library construction prevents the cloning of inserts that are too large to sequence with terminal vector polylinker primers. However, size fractioning genomic DNA is a sampling process that could lead to differential representation of SSR loci within genomic libraries. Second, recent observations suggest that small numbers of SSR markers are not randomly distributed across the genome, although random distributions have been observed ( Kijas et al. 1997). Thirty-two tetranucleotide microsatellites were found to cluster when mapped within the genome of tomato (Arens et al. 1995). In Arabidopsis, thirty mono- and dinucleotide microsatellites mapped to five chromosomes, but two of these chromosomes contained only one and two microsatellites each ( Bell and Ecker 1994). In three species of birds, CAn microsatellites occurred infrequently on microchromosomes compared to intermediate and macrochromosomes (Primmer et al. 1997). Because wild species are most often not amenable to linkage mapping, there are currently no estimates of how often nonrandom genomic distributions of SSRs might occur in species where a small number of microsatellite markers

are characterized for population genetic studies. In this article we present the results from attempts to clone di-, tri-, and tetranucleotide repeat microsatellites for use in population genetic studies of two wild plant species, Corythophora alta ( Lecythidaceae) and Heliconia acuminata ( Zingiberaceae). The experiment involved constructing two independent SSR-repeatenriched plasmid libraries based on sizefractioned inserts from aTaqI and DpnII digests of total genomic DNA. All enrichment and cloning steps were identical, except the starting DNA insert pool, and were carried out simultaneously. The genomic library construction was based on modifications to procedures to repeat-enrich DNA sequences by subtractive hybridization described in detail elsewhere (Armour et al. 1994; Fleischer and Loew 1995; Kijas et al. 1994). Approximately 5 mg of total genomic DNA was digested with DpnII or aTaqI ( New England Biolabs), dephosphorylated, and then size separated on 1.5% low-melt agarose gels. DNA fragments in the 200–1000 bp size range were cut from the gel and the agarose was digested with Gelase ( Epicenter Technologies). Pairs of linker oligonucleotides, with a 59 overhang complementary to the 59 overhang produced by each restriction enzyme, were ligated ( New England Biolabs) to the digested, size-selected genomic DNA. This linker genomic DNA was not amplified by PCR to avoid potential amplification bias. The linker genomic DNA was hybridized to five SSR sequences (AAT10, CAC10, GT15, CAT10, and GGAT8; each biotinylated) in separate reactions (all carried out simultaneously) at temperatures appropriate for the sequence composition. Streptavidin-coated beads ( Dynal) were added to capture the biotinylated oligonucleotides and any annealed linker genomic DNA. Several stringency washes in 13 SSC, 0.1% SDS removed genomic DNA not bound to the biotinylated oligonucleotides. The resulting repeat-enriched linker genomic DNA was used as template in a PCR reaction with one linker oligonucleotide as a primer. PCR products were digested with the original restriction enzyme to remove the linkers and ligated to BamHI or ClaI (using DpnII and aTaqI digests, respectively) digested, dephosphorylated pBS SK1 (Stratagene). Plasmids were transformed into XL1-blue supercompetent cells (Stratagene) and the transformation mixture was plated onto LBampicillin agar containing X-gal and IPTG. White colonies were transferred to fresh LB-ampicillin master plates and used for


colony lifts. Colony lift filters were hybridized in independent reactions to the same probe sequence used for enrichment (each oligonucleotide was 32P end-labeled) and exposed to X-ray film. Cells from darkly hybridizing colonies were picked from LB plates, transferred to 100 ml aliquots of T.E (10 mM Tris–HCl, 0.1 mM EDTA), and incubated at 1008C for 10 min to lyse cells. One microliter of this cell lysis solution was used as template in a PCR reaction with T7 and T3 polylinker primers to amplify the cloned insert. These PCR products were run on agarose gels to verify amplification of an insert and then purified of primers using Qiagen QiaQuick spin columns. Single-band PCR products were sequenced with ReactionReady FS Taq (ABI) and electrophoresed on a 373 florescent sequencer (ABI). One hundred forty-two colonies were sequenced for at least one strand, although some positive colonies could not be successfully sequenced due to polyphyletic inserts. All sequences were trimmed of trailing ambiguous bases and aligned with Sequencher 3.0 (GeneCodes). Several insert sequences were represented multiple times in the positive colonies, in most cases for GTG/CAC repeat inserts that appeared up to 15 times. Total numbers of positive colonies with unique insert sequences were counted, as well as the total number of colonies screened. Positive colonies that could not be sequenced were assumed to contain unique inserts. A detailed version of the experimental protocol is available via the Internet ( http:// www.si.edu/natzoo/genetics/ and click on the ‘‘Technical Corner’’ button). The number of positive and negative colonies for each SSR-restriction enzyme combination is shown in Table 1. G tests were used to test the null hypothesis that the number of positives of each SSR was the same for each restriction enzyme (Sokal and Rohlf 1981). For some SSR probes no positives were observed for an enzyme even though sample sizes were large. In those cases (mostly for H. acuminata), one was added to the count of positives for both restriction enzymes since G tests cannot be carried out when any frequency is zero. Of five SSR probes used, significantly different numbers of positives for each restriction enzyme were obtained for three in C. alta and one in H. acuminata. The total G for each species is significant, indicating that for all SSR probes considered together, equal numbers of positives for each enzyme is improbable. The pooled G is not significant for both spe-


Table 1. Numbers of hybridization-positive colonies (numbers of negatives in parentheses) resulting from genomic libraries constructed with the restriction enzymes aTaqI and DpnII for two plant species Heliconia acuminata

Corythophora alta SSR

aTaqI

DpnII

G

aTaqI

DpnII

AAT10

12 (383) 17 (379) 6 (341) 3 (393) 0 (296)

0 (495) 6 (489) 11 (484) 6 (489) 9 (486)

15.27**

2 (345) 0 (495) 0 (297) 1 (494) 0 (297)

0 (495) 10 (485) 3 (492) 0 (495) 3 (492)

GTG5 GT15 CAT10 GGAT5 Total Pooled Heterogeneity

8.44* 0.25 0.47 3.96* 28.39** 4.00 24.39**

G 0.79 8.63** 0.28 0.00 0.28 10.00** 3.69 6.31**

Positive counts are adjusted for redundancy of clones detected by sequencing. All colonies screened by hybridization for the presence of SSRs were positive for a plasmid insert by X-gal/IPTG color selection. For each SSR, if no hybridization-positive colonies were observed, one was added to the positive tally of both restriction enzymes in order to calculate G tests. * p , .05; **p , .01.

cies, indicating that the total number of positives detected for each enzyme were equal. The heterogeneity G is significant in each case, indicating significant variation in the number of positives for each enzyme-SSR combination. The pattern of positives observed depends on the interaction of the restriction enzyme and SSR probe, as expected if SSRs are not detected equally with each enzyme. Our results indicate that the number and repeat types of potential SSR markers depend on the restriction endonuclease used to size fraction the genome when constructing plasmid DNA-insert libraries. Restriction digests of genomic DNA are a critical step where sampling of the genome takes place which can cause resulting SSR markers to be differentially sampled from the genome. In addition to the frequency-dependent sampling inherent in the enrichment and ligation process, DNA fragments outside the selected size range are not screened for SSRs. Different restriction enzymes will result in different proportions of the genome within the clonable size range and available for SSR detection. These results suggest that employing several enzymes to restrict genomic DNA in SSR cloning efforts provides a broader sample of the genome within the insert pool and thereby a broader genomic sample of SSR loci for use in population genetic studies. This approach would be effective even when using enrichment techniques that increase the chances of cloning repeat-containing inserts because enrichment only selects from sequences available in the insert pool. Although enzymes such as DpnII, which leave long complementary overhangs to aid ligation

into vectors, are frequently used, the use of multiple restriction enzymes irrespective of overhang may be required to build genomic libraries to provide diverse SSR markers. Ideally a combination of restriction enzymes should be employed that render the entire genome within the clonable size range. Since linkage and chromosomal locations of SSR loci cannot be determined in most wild species (especially long-lived plants such as those used here), the assumption of a random physical distribution of such markers cannot be tested directly. These results provide indirect evidence that these five SSR sequences are not clustered in the same location within the genomes of these two plant species. If all of these SSR repeats were closely associated in the genome we would expect that different restriction enzymes would give similar numbers of positive clones for each repeat. However, the significant heterogeneity G in each case suggests that the pattern of SSR-repeat positives for each enzyme is substantially different. This conclusion does assume that restriction enzyme cut sites are distributed independently of SSR repeat sites. Clones for a single SSR repeat are associated with a given restriction enzyme in 4 of 10 cases by the individual G tests. Physical clustering of these individual SSR loci bearing the same repeat sequence and the sampling effects of size fractioning cannot be distinguished without linkage data. Such data are effectively impossible to collect for these species since generation of near-isogenic parental lines would require five generations of selfing (assuming selfcompatibility) and two generations of

crossing among these lines to create F2 progeny [the progeny design employed by Arens et al. (1995)]. For H. acuminata, this design would take at least 35 years to accomplish, compared to several hundred years for C. alta. Random physical distributions of marker loci is an important assumption of population genetic studies as well as a requirement for high-resolution genetic mapping. For population genetic studies, marker loci should be drawn from as much of the genome as possible so evolutionary events (mutation dynamics, phylogenetic history, selection, and linkage) are averaged across the genome. The use of multiple restriction enzymes or the reduction of sampling steps when constructing the DNA insert pool may increase the diversity of SSR marker loci available. These results suggest that SSR loci bearing different repeats are not associated within the genome, so the use of SSR marker loci with different repeats may aid in meeting the assumption of randomness for population genetic studies. Although the genomic distribution of SSR loci is less of a concern in parentage analysis applications (as long as markers are unlinked), using multiple restriction enzymes when constructing DNA libraries may increase the diversity of SSRs available. From the Smithsonian Institution, National Zoological Park, Molecular Genetics Laboratory, Washington, DC. This work was supported by a Smithsonian Institution postdoctoral fellowship to M.B.H. and Friends of the National Zoo. We thank D. Fonseca and L. Shapiro for contributions to the microsatellite cloning methods, I. Jones for help picking colonies, and J. Ballou for discussion of G tests. W. J. Kress provided tissue samples of H. acuminata and helpful discussion. Two anonymous reviewers provided comments that improved the manuscript. Address correspondence to M. B. Hamilton, Georgetown University, Department of Biology, Reiss Building STE 406, Box 571229, Washington DC 20057-1229, or e-mail: [email protected]. q 1999 The American Genetic Association

References Arens P, Odinot P, van Heusden AW, Lindhout P, and Vosman B, 1995. GATA- and GACA-repeats are not evenly distributed throughout the tomato genome. Genome 38:84–90. Armour JAL, Neumann R, Gobert S, and Jeffreys AJ, 1994. Isolation of human simple repeat loci by hybridization selection. Hum Mol Genet 3:599–605. Bell CJ and Ecker, JR, 1994. Assignment of 30 microsatellite loci to the linkage map of Arabidopsis. Genomics 19:137–144. Fleischer RC and Loew S, 1995. Construction and screening of microsatellite-enriched genomic libraries. In: Molecular zoology: advances, strategies and protocols ( Ferraris J and Palumbi S, eds). New York: WileyLiss; 461–468. Goldstein DB and Pollock DD, 1997. Launching microsatellites: a review of the mutation processes and methods of phylogenetic inference. J Hered 88:335–342. Kijas JMH, Fowler JCS, Thomas MR, and Garbett CA,

1994. Enrichment of microsatellites from the citrus genome using biotinylated oligonucleotide sequences bound to streptavidin-coated magnetic particles. BioTechniques 16:657–662. Kijas JMH, Thomas MR, Fowler JCS, and Roose ML, 1997. Integration of trinucleotide microsatellites into a linkage map of Citrus. Theor Appl Genet 94:701–706. Primmer CR, Raudsepp T, Chowdhary BP, Moller AP, and Ellegren H, 1997. Low frequency of microsatellites in the avian genome. Genome Res 7:471–482. Sokal RR and Rohlf FJ, 1981. Biometry, 2nd ed. New York: W.H. Freeman. Received May 13, 1998 Accepted March 31, 1999 Corresponding Editor: Norman F. Weeden

Female and Hermaphrodite Flowers on a Chimeric Gynomonoecious Silene vulgaris Plant Produce Offspring with Different Genders: A Case of Heteroplasmic Sex Determination? H. Andersson In gynodioecious plant species, individuals have either female or hermaphrodite flowers. However, individuals with both types of flowers, that is, ‘‘gynomonoecious’’ or ‘‘partially male steriles,’’ are sometimes found. The standard explanation of gynomonoecious individuals is that their male-sterilizing cytoplasm is incompletely restored by nuclear male fertility genes. Silene vulgaris, the bladder campion, is usually referred to as being gynodioecious, though gynomonoecious individuals have frequently been observed. In an attempt to determine if the segregation of flower types on gynomonoecious plants of S. vulgaris is associated with a difference in offspring gender, a series of crosses was performed. The results show that female and hermaphroditic flowers on the same plant produce offspring of all three types, but the frequencies with which they do so differ. If incomplete restoration was the relevant explanation in this case, these results would not have occurred. Instead an intraindividual segregation of one or many genetic factors that affect the sexual phenotype of flowers and their subsequent offspring is proposed. Three different genders are frequently found in gynodioecious species: plants with only female flowers, plants with only hermaphrodite flowers, and plants with

both types of flowers ( Baker 1966; Brockman and Bocquet 1978; Burrows 1960; Dulberger and Horovitz 1984; Horovitz and Gallil 1972; Jolls and Chenier 1989; Koelewijn and Van Damme 1995a; Philipp 1980). The last type is generally called ‘‘gynomonoecious’’ or ‘‘partially male sterile.’’ The term ‘‘gynodioecy’’ was originally coined to describe the plant reproductive system with only females and hermaphrodites ( Darwin 1877). Formally speaking, populations containing individuals of the third type should therefore be said to have a gynodioecious-gynomonoecious breeding system, but normally the occurrence of such plants is disregarded and the breeding system is still said to be gynodioecious. The standard explanation of gynomonoecious plants is that they represent the developmental outcome of an incomplete restoration by nuclear male fertility genes of male-sterilizing cytoplasms ( Koelewijn and Van Damme 1995b; Louis and Durand 1978; Van Damme 1983; Vranceanu and Stoenescu 1978; Wickersham and Patterson 1980). Silene vulgaris (Moench) Garcke (Caryophyllaceae), a common weed in ruderate habitats throughout Europe, is usually referred to as being gynodioecious, though gynomonoecious individuals have frequently been observed ( Brockman and Bocquet 1978; Charlesworth and Laporte 1998; Dulberger and Horovitz 1984; Jolls and Chenier 1989). The full genetic details of the sex determination system in the species is not yet fully understood. There is, however, strong evidence that both nuclear and cytoplasmic factors are involved in the determination of sex (Charlesworth and Laporte 1998; Marsden-Jones and Turill 1957). In a survey of S. vulgaris plants in natural populations from southern Sweden, gynomonoecious individuals were found together with females and hermaphrodites. Seeds were collected from the populations and plants were raised under standard conditions in a greenhouse. At flowering the frequencies of gynomonoecious, female, and hermaphroditic plants were determined. In the different populations gynomonoecious plants ranged in frequency from 0 to 25%, while female plant frequency ranged from 0 to 90%. Three types of flower stalks could be observed on the gynomonoecious individuals: stalks with only female flowers, stalks with only hermaphrodite flowers, and stalks with a mixture of both types of flowers. The ratio of female to hermaphrodite flowers on the gynomonoecious plants varied, as did the


Table 1. The gender of offspring to four sets of crosses between a chimeric gynomonoecious plant (A) and two hermaphrodite test plants (M and N) of S. vulgaris Offspring gender

Figure 1. Typical flower arrangements in chimeric branches of gynomonoecious S. vulgaris plants. Filled circles represent hermaphroditic flowers; unfilled circles represent female flowers.

degree of maleness of the individual hermaphroditic flowers (as judged by the number and size of anthers). The segregation of female and hermaphroditic flowers on chimeric stalks often showed a strikingly symmetrical pattern ( Figure 1). Even though the ratio of female to hermaphroditic flowers on gynomonoecious plants varied from one period of flowering to the next, their gender expression was stable over years. To determine if the segregation of flowers on gynomonoecious plants of S. vulgaris is associated with a difference in the offspring types produced, a series of crosses was performed. A typical gynomonoecious individual (A) and two hermaphroditic test plants (M and N) were used as parents in the crosses. The paternal plants were chosen from the greenhouse-grown material described above.

The Crossing Experiments Four sets of crosses were performed, identified as 1, 2, 3, and 4 in Table 1. Seeds were obtained from all crosses except from hermaphroditic flowers on chimeric stalks (4b). The derived seeds were germinated and cultivated in the greenhouse and the offspring plants could, when flowering, be classified into the same three gender classes as the plants from the natural populations. When female and hermaphroditic flowers on plant A were fertilized with pollen from the same plant (set 1, Table 1), both female and hermaphroditic flowers produced offspring of all three gender classes, but they did so in different proportions. The most obvious difference was that hermaphroditic flowers produced many more gynomonoecious offspring than female flowers, while female flowers produced more female offspring than hermaphroditic flowers. The same result was obtained in the second set of crosses, where female and hermaphroditic flowers on A were fertilized with pollen from test plant M, but here the recorded differences were even greater. Also, in the third cross, which was


Crosses

Female

Gynomo- Hermanoecious phrodite

1. A (self-fertilized) a. Female flowers b. Hermaphroditic flowers

50 44

6 28

8 8

2. A 3 M a. Female flowers b. Hermaphroditic flowers

64 32

2 22

3. A 3 N a. Female flowers b. Hermaphroditic flowers

60 52

4. A (self-fertilized) a. Female flowers b. Hermaphroditic flowers

49 —

Test of homogeneity x2

p

64 80

13.0

,.01

30 46

96 100

30.6

,.001

2 15

37 32

99 99

10.9

,.01

20 —

6 —

75 —

—

—

Total

In crosses 1, 2, and 3 the female flowers came from female stalks and the hermaphroditic flowers from hermaphrodite stalks. Castrations were made in crosses 2 and 3 to avoid self-fertilization. The homogeneity between offspring classes was investigated with a chi-square test with two degrees of freedom. In the fourth cross, the flowers used to set seeds came from chimeric stalks; no seeds were produced by the hermaphroditic flowers. The pollen used for self-fertilizing plant A (crosses 1 and 4) came from hermaphroditic flowers on hermaphroditic flower stalks.

analogous to the second cross but in which plant N was used as the pollen donor, the same differences in offspring gender were seen. In general, the proportion of hermaphrodites among the offspring increased when plants M and N were used for fertilization, in comparison to when plant A was self-fertilized. The fourth set of crosses were identical to the first crosses, except that the female and hermaphrodite flowers that were fertilized with A pollen this time came from chimeric flower stalks. In this case only the female flowers set any seeds. The selffertilized female flowers on chimeric stalks and the self-fertilized female flowers on nonchimeric stalks (results from experiment 1a) produced progeny that differed significantly in gender proportions (x2 5 7.0, df 5 2, p , .05). The difference was primarily due to the flowers on chimeric branches producing a higher frequency of gynomonoecious offspring.

Interpretation These results show that female and hermaphroditic flowers on nonchimeric stalks produce offspring that differ significantly from each other. Both female and hermaphroditic flowers can produce offspring of all three types, but the frequencies with which they do so differ. The presence of hermaphroditism appears to be associated with the production of gynomonoecious offspring. For example, the greatest difference between hermaphroditic flowers on hermaphroditic stalks and female flowers on female stalks is that the

hermaphroditic flowers produce more gynomonoecious offspring than female flowers do. When the flowers from chimeric stalks are studied, it is seen that female flowers from such stalks produce more gynomonoecious offspring than female flowers on stalks with only females. If the occurrence of gynomonoecious plants in S. vulgaris were strictly due to incomplete restoration of a male-sterile cytoplasm by one or many nuclear loci, as has been suggested for other species ( Koelewijn and Van Damme 1995b; Louis and Durand 1978; Van Damme 1983; Vranceanu and Stoenescu 1978; Wickersham and Patterson 1980), the observed differences between the offspring to the two flower types would not have existed. Nuclear genes are inherited on the female side in the same way, irrespective of whether they are transmitted via a female or a hermaphroditic flower. The frequency of offspring genotypes with an unstable sex expression should therefore have been the same for the two kinds of flowers. On the other hand, cytoplasmic factors are capable of segregating within single individuals. Since a nucleocytoplasmic determination of sex is proposed for S. vulgaris (Charlesworth and Laporte 1998; Marsden-Jones and Turill 1957), intraindividual segregation of one or many genetic factors affecting sex determination is a possible explanation of the results reported here. It has the advantage of explaining both the mosaic of flowers on the chimeric individuals and the differences among offspring from the different flower types. If

a fertilized egg is heteroplasmic for maledeveloping and/or male-sterilizing mitochondrial factors, then successive cell divisions could lead to a sorting-out process where mitochondria of different types end up in different parts of the plant. The symmetrical pattern often seen in chimeric stalks could be the result of such segregational cell division events. Thus it seems as if some plants consist of a mixture of mitochondrial genomes, a part of which signals the development of male function in the given nuclear background, while other parts inhibit it. Progressively during growth the proportion of the two mitochondrial types change, leading to different sexual phenotypes in different parts of the plant, as determined by the most common mitochondrial genotype. Presumably female flowers on chimeric stalks contain a higher proportion of male-fertile mtDNA molecules than, for example, female flowers on purely female stalks. This effect is seen in the present experiment as the difference between crosses 1 and 4. Of course, various complicated systems invoking epigenetic inheritance, imprinting, and/or maternal effects could also be proposed to explain the results. However, all such explanations are difficult to reconcile with the fact that female flowers on chimeric stalks produce offspring that are more similar to that of hermaphroditic flowers on hermaphroditic stalks than to that of female flowers on female stalks. Segregating heteroplasmy opens up the possibility of intraindividual competition between different types of cytoplasm with far-ranging effects on sex determination and sex ratio evolution (see, e.g., Godelle and Reboud 1997). It would be of interest to investigate if the surprising effect reported herein also occurs in other species with gynomonoecious plants. From the Department of Genetics, Lund University, So¨lvegatan 29, S-223 62 Lund, Sweden. Address correspondence to Helene Andersson at the address above or email: [email protected]. Thanks are due to Bengt O. Bengtsson for helpful comments on the manuscript. This research was supported by grants from the Jo¨rgen Lindstro¨m Fund to H. Andersson and the Swedish Natural Science Research Council to B. O. Bengtsson. q 1999 The American Genetic Association

References Baker H G, 1966. The evolution of floral heteromorphism and gynodioecism in Silene maritima. Heredity 21:689–692. ¨ kologische einflu¨sse Brockman I and Bocquet G, 1978. O auf die geschlechtsverteilung bei Silene vulgaris (Moench) Garcke (Caryophyllaceae). Ber Deutsch Bot Ges Bd 91:217–230.

Burrows C J, 1960. Studies in Pimeliea I—the breeding system. Trans R Soc N Z 88:29–45. Charlesworth D and Laporte V, 1998. The male-sterility of Silene vulgaris: analysis of genetic data from two populations and comparison with Thymus vulgaris. Genetics 150:1267–1282. Darwin C, 1877. Different forms of flowers on plants of the same species. London: John Murray. Dulberger R and Horovitz A, 1984. Gender polymorphism in flowers of Silene vulgaris (Moench) Garcke (Caryophyllaceae). Bot J Linn Soc 89:101–117. Godelle B and Reboud X, 1997. The evolutionary dynamics of selfish replicators: a two-level selection model. J Theor Biol 185:401–413. Horovitz A and Galil J, 1972. Gynodioecism in east Mediterranean Hirschfeldia incana, Cruciferae. Bot Gaz 133: 127–131. Jolls CL and Chenier TC, 1989. Gynodioecy in Silene vulgaris (Caryophyllaceae): progeny success, experimental design, and maternal effects. Am J Bot 76:1360– 1367. Koelewijn HP and Van Damme JMM, 1995a. Genetics of male sterility in gynodioecious Plantago coronopus. I. Cytoplasmic variation. Genetics 139:1749–1758. Koelewijn HP and Van Damme JMM, 1995b. Genetics of male sterility in gynodioecious Plantago coronopus. II. Nuclear genetic variation. Genetics 139:1759–1775. Louis JP and Durand B, 1978. Studies with the dioecious angiosperm Mercuroalis annua L. (2n 5 16): correlation between genic and cytoplasmic male sterility, sex segregation and feminizing hormones (cytokinins). Mol Gen Genet 165:309–322. Marsden-Jones EM and Turill WB, 1957. The bladder campions. London: Ray Society. Philipp M, 1980. Reproductive biology of Stellaria longipes goldie as revealed by a cultivation experiment. New Phytol 85:557–569. Van Damme JMM, 1983. Gynodioecy in Plantago lanceolata L. II. Inheritance of three male sterility types. Heredity 50:253–273. Vranceanu AV and Stoenescu FM, 1978. Genes for pollen fertility restoration in sunflowers. Euphytica 27: 617–627. Wickersham DS and Patterson FL, 1980. Male-fertility in crosses of R5 with soft red winter wheats. Crop Sci 20: 100–102. Received January 25, 1999 Accepted May 3, 1999 Corresponding Editor: Norman F. Weeden

Evidence for TemperatureDependent Selection for Malate Dehydrogenase Allele Frequencies in Honeybee Populations S. Hatty and B. P. Oldroyd The MDH-1 genotype and a mitochondrial DNA haplotype was determined for feral honeybees (Apis mellifera L.) collected from 10 sites in southern New South Wales, Australia. The frequency of the Mdh 65 allele was positively correlated, and the Mdh 80 allele negatively correlated with increasing average daily temperature for

July and January (P , .01), whereas no cline was found for the mitochondrial marker. Parallel clines in MDH allele frequencies have now been found on four continents, and the Mdh 80 allele has been shown to be less heat stable in vitro than the other alleles. We conclude that this is very strong evidence that the MDH-1 clines observed in honeybees are due to temperature-dependent selection. Based on morphological and ethological characters, honeybees (Apis mellifera) have been broadly divided into three racial groups, the African, the northern European, and the eastern European (Ruttner 1968). These groupings are now well supported by mitochondrial (Smith 1991; Garnery et al. 1992) and microsatellite ( Estoup et al. 1995) evidence. Two simple molecular markers are useful for distinguishing these groups. First, mitochondria originating from the A. m. mellifera (northern European) and A. m. ligustica (eastern European) subspecies can be distinguished using a DNA restriction fragment length polymorphism (Garnery et al. 1992; Hall and Smith 1991; Nielson et al. 1994). Second, the frequency of the three common alleles at the MDH-1 locus (Mdh65, Mdh80, and Mdh100) is also broadly discriminatory ( Badino et al. 1983; Cornuet 1979). The mellifera subspecies is monomorphic for the Mdh80 allele, whereas this allele is absent from ligustica (Cornuet 1979; Badino et al. 1984). Thus this locus has been frequently used as a molecular marker for hybridization events (e.g., Badino et al. 1984; Del Lama et al. 1990; Lobo et al. 1989; Oldroyd et al. 1992, 1995). Latitudinal clines have been demonstrated at the MDH-1 locus in natural populations in Italy ( Badino et al. 1984) and in introduced populations in North and South America ( Del Lama et al. 1990; Lobo et al. 1989; Nielsen et al. 1994). The populations in these areas have different genetic antecedents and therefore these parallel clines provide strong evidence that MDH-1 experiences natural selection ( Endler 1986; Oakshott et al. 1982). Further evidence for this notion comes from the differences in the thermostability of MDH-1 allozymes (Cornuet et al. 1995). However, the clines have also been regarded as evidence for as yet incomplete hybridization events, not selection ( Badino et al. 1984; Del Lama et al. 1990; Lobo et al. 1989; Smith and Glenn 1995). A. m. mellifera was established in Australia in the early 19th century and A. m.


Figure 1. Honeybee MDH-1 allele and genotype frequencies, and mtDNA haplotype frequencies for 10 sites in southern New South Wales, Australia. The elevations and number of bees analyzed at each site are as follows: Tathra elevation 5 30 m, n 5 45; Bega 30 m, n 5 47; Bemboka 346 m, n 5 46; Brown Mountain 762 m, n 5 48; Nimmitabel 326 m, n 5 42; Dalgety 914 m, n 5 53; Jindabyne 1049 m, n 5 50; Guthega 1660 m, n 5 58; Charlotte’s Pass 1768 m, n 5 19; Thredbo 1433 m, n 5 49.

ligustica in the 1880s ( Hopkins 1886). Here we describe how alleles derived from these populations have formed a temperature-parallel cline for MDH-1 allele frequency in southern New South Wales. Further, using frequencies of an independent mitochondrial marker, we evaluate the competing hypotheses of selection at the MDH-1 locus and hybridization between populations with differing allele frequencies to explain the cline. We then discuss the likely causes of the parallel clines present in North America ( Nielsen et al. 1994), Italy ( Badino et al. 1984), and Brazil ( Del Lama et al. 1990; Lobo et al. 1989).


Materials and Methods Collection of Specimens Worker bees were collected from 10 sites in southern New South Wales ( Figure 1) between March 10–14, 1997. Sites were selected along a transect encompassing a strong temperature gradient. Foraging bees were collected by sweep netting and stored in liquid nitrogen. Samples were taken from many different plants at each site in order to prevent overrepresentation of any single colony. Collections (sample sizes in Figure 1) were made within a 2 km radius of each site, except at Bemboka where two collections separated by

approximately 5 km were made because of a lack of foraging bees. The elevation at each site was measured with an altimeter ( Esdal, Sydney). Mean monthly maximum and minimum temperatures for January and July at the nearest weather station were obtained from the Australian Bureau of Meteorology. Temperatures from the Bega weather station were used for the Brown Mountain site, as samples were collected from the eastern side of the mountain where the conditions were considered more similar to those at Bega than those at a closer weather station at Nimmitabel. Similarly, although Jin-

dabyne was closer, weather data from Cooma were used for the Dalgety site. mtDNA and Protein Extraction and Analysis DNA was extracted following the method of Walsh et al. (1991). A single hind leg was crushed while being held at 2708C. One milliliter of boiling Chelext [5 g Chelext100 resin ( Bio-Rad), 10 ml 1.0 M TE buffer in 100 ml MQ-dH2O] was added to each tube which was then immediately placed in boiling water for 15 min, cooled to room temperature, and stored at 48C. PCR reactions (20 ml vol) contained 2 ml of Tth plus 103 buffer ( Fisher Biotech), 1.5 mM MgCl2, 0.08 mM of each dNTP, 0.4 mM of each of the primers described by Nielsen et al. (1994), 0.8 units of Tth plus DNA polymerase ( Fisher Biotech), and 2 ml of the Chelex extraction. Reactions underwent 1 cycle of 4 min at 948C, 32 cycles of 1 min at 948C, 1 min at 538C, and 2 min at 728C and 1 cycle of 10 min at 728C. The primers used in the reaction amplified a 1 kb region of the large ribosomal subunit ( Hall and Smith 1991). PCR reaction product (5 ml) was digested using EcoRI (Promega) at 378C for 1.5 h and electrophoresed in 1% agarose gels. Two mitochondrial haplotypes were observed after digestion with EcoRI. The first was a pair of bands (one restriction site) 480 bp and 500 bp in size. This haplotype is usual for ligustica populations ( Hall and Smith 1991). The second haplotype, which lacks the EcoRI restriction site, is characteristic of mellifera populations ( Hall and Smith 1991). The MDH-1 genotype was determined using the procedure of Oldroyd et al. (1995).

Results mtDNA RFLPs The frequencies of the two haplotypes were significantly heterogeneous across the 10 sites (G 5 102.33, df 5 9, P , .01) ( Figure 1). The mellifera haplotype occurred with the highest frequency at Guthega, whereas the highest frequency of the ligustica type was at Thredbo. Haplotype frequencies were normalized using the transformation arcsine(Ïy) (Steel and Torrie 1980). There was no significant correlation of transformed frequencies with elevation or temperature ( Table 1). MDH-1 Although other alleles have been reported (Sheppard and Berlocher 1984), only Mdh65, Mdh80, and Mdh100 were observed in

Table 1. Pearson correlation coefficients (probability in brackets) for elevation and mean seasonal temperature (8C) with MDH-1 allele frequency and mtDNA haplotype frequency for 10 sites in New South Wales * indicates a significant correlation after a sequential Bonferroni test of the four factor -1 probabilities Genetic marker mellifera mtDNA Mdh65 Mdh80 Mdh100

Elevation (m)

July minimum

July maximum

January minimum

January maximum

Factor 1

0.078 (0.83) 20.81 (0.005) 0.71 (0.02) 0.17 (0.63)

0.12 (0.74) 0.84 (0.002) 20.73 (0.02) 20.23 (0.51)

0.12 (0.74) 0.84 (0.002) 20.74 (0.01) 20.20 (0.57)

0.11 (0.75) 0.85 (0.002) 20.76 (0.01) 20.18 (0.62)

0.23 (0.52) 0.83 (0.003) 20.69 (0.03) 20.29 (0.42)

0.161 (0.65) 0.83 (0.003)* 20.68 (0.03)* 20.31 (0.39)

this study. Allele frequencies were significantly different across the 10 sites (G 5 117.83, df 5 18, P , .01; Figure 1). MDH-1 allele frequencies were normalized using the transformation arcsin(Ïy). The frequency of the Mdh80 allele was negatively correlated with temperature and positively correlated with elevation, while the reverse was true for the Mdh65 allele ( Table 1, Figure 1). The frequency of the Mdh100 allele showed no environmental correlation ( Table 1). When a large number of tests of individual significance are performed simultaneously, the type 1 error becomes significant. To control the experiment-wise error rate, we performed a principle component analysis on the temperature variables in order to combine the temperature variables into a single index. The correlations between factor 1 of this analysis and the allele and haplotype frequencies were then computed, and a sequential Bonferroni correction (a 5 0.05) was made to the four correlations (Rice 1989). This analysis showed that both Mdh65 and Mdh80 allele frequencies were significantly correlated with temperature ( Table 1). There was no significant association between mtDNA haplotype frequency and the temperature index.

Discussion Our data show a temperature-associated cline in MDH-1 allele frequencies over a distance of less than 300 km. The population of honeybees along our transect is likely to be very large (Oldroyd et al. 1997), and thus the cline is unlikely to be the result of stochastic events such as genetic drift. The New South Wales cline is similar to others reported ( Badino et al. 1984; Del Lama et al. 1990; Lobo et al. 1989; Nielsen et al. 1994), with the Mdh80 allele predominant in colder areas and the Mdh65 and Mdh100 alleles at higher frequency in warmer regions.

If ligustica bees were introduced on the coastal regions and mellifera bees were introduced in the highlands, then ligustica genotypes would be expected to occur with highest frequency on the coast and the mellifera genotypes with the highest frequency in the mountains. The lack of a parallel cline in mtDNA haplotype frequency provides evidence against such a differential introduction ( Figure 1) and supports the notion of selection maintaining the MDH-1 cline despite frequent migration events caused by beekeepers. The apparently random, but heterogenous distribution of haplotype frequencies across the transect indicates that beekeepers are influencing mitotype frequencies by frequent migration of colonies. However, this migration is apparently insufficient to overcome selection at the MDH-1 locus. Moreover, if the cline in southern New South Wales were due to differential introduction of ligustica and mellifera stocks by beekeepers, then we would expect Mdh80 to be at lowest frequency in the presence of the ligustica mitotype. This is clearly not the case. We conclude that there is now overwhelming evidence that MDH-1 allele frequencies are maintained by selection in honeybees. Although correlation does not imply causation, four independent correlations on different continents is much stronger evidence for causation than a single observation. The demonstration that Mdh80 is less heat stable in vitro than the other alleles (Cornuet et al. 1995) provides further compelling evidence that these clines are maintained by selection. Evolutionary and population genetic models often assume enzyme polymorphisms have no functional significance and are selectively neutral (e.g., Harrison et al. 1996). In the light of the now very strong evidence of temperature-dependent selection, the usefulness of the MDH1 locus of A. mellifera for population genetics studies is very questionable.


From the School of Biological Sciences A12, University of Sydney, NSW 2006, Australia. This research was supported by an Australian Research Council small grant. Address correspondence to Dr. B. P. Oldroyd at the address above or e-mail: [email protected].

Rice WR, 1989. Analysing tables of statistical tests. Evolution 43:223–225.

q 1999 The American Genetic Association

Sheppard WS and Berlocher SH, 1984. Enzyme polymorphism in Apis mellifera from Norway. J Apic Res 23: 64–69.

References Badino G, Celebrano G, and Manino A, 1983. Population structure and Mdh-1 locus variation in Apis mellifera ligustica. J Hered 74:443–446. Badio G, Celebrano G, and Manino A, 1984. Population genetics of Italian honey bee (Apis mellifera ligustica Spin.) and its relationships with neighbouring subspecies. Boll Museo Reg Scienze Nat Torino 2:571–584. Cornuet J-M, 1979. The MDH system in honey bees of Guadaloupe. J Hered 70:223–224. Cornuet J-M and Louveaux J, 1981. Aspects of genetic variability in Apis mellifera L. In: Biosystematics of social insects ( Howse PE and Clement J-L, eds). London: Academic Press.

Ruttner F, 1968. Les races d’abeilles. In: Traite de biologie de l’abeille, vol 1 (Chauvin R, ed). Paris: Masson et Cie; 27–44.

Smith DR, 1991. African bees in the Americas: insights from biogeography and genetics. Trends Ecol Evol 6: 17–21. Smith DR, 1995. Allozyme polymorphisms in Spanish honeybees (Apis mellifera iberica). J Hered 86:12–16. Steel RDG and Torrie JH, 1980. Principles and procedures of statistics. Tokyo: McGraw-Hill. Walsh PS, Metzger DA, and Higuchi R, 1991. Chelex as a medium for simple extraction of DNA for PCR-based typing from forensic material. Biotechniques 10:87. Received July 27, 1998 Accepted February 18, 1999 Corresponding Editor: Ross MacIntyre

Cornuet J-M, Oldroyd BP, and Crozier RH, 1995. Unequal thermostability of allelic forms of malate dehydrogenase in honey bees. J Apic Res 34:45–47. Del Lama MA, Lobo JA, Soares AEE, and Del Lama SN, 1990. Genetic differentiation estimated by isozymic analysis of Africanized honey bee populations from Brazil and Central America. Apidologie 21:271–280. Endler J, 1986. Natural selection in the wild. Princeton: Princeton University Press. Estoup A, Garnery L, Solignac M, and Cornuet J-M, 1995. Microsatellite variation in honey bee (Apis mellifera L.) populations: hierarchical genetic structure and test of the infinite allele and stepwise mutation models. Genetics 140:679–695. Garnery L, Cornuet J-M, and Solignac M, 1992. Evolutionary history of the honey bee Apis mellifera inferred from mitochondrial DNA analysis. Mol Evol 1:145–154. Hall HG and Smith DR, 1991. Distinguishing African and European honey bee matrilines using amplified mitochondrial DNA. Proc Natl Acad Sci USA 88:4548–4552. Harrison JF, Nielsen DI, and Page RE Jr, 1996. Malate dehydrogenase phenotype, temperature and colony effects on flight metabolic rate in the honeybee, Apis mellifera. Funct Ecol 10:81–88. Hopkins I, 1886. Illustrated Australasian bee manual, 3rd ed. Auckland, New Zealand: I Hopkins. Lobo JA, Del Lama MA, and Mestriner MA, 1989. Population differentiation and racial admixture in the Africanized honeybee (Apis mellifera L.). Evolution 43:794– 802. Nazi F, 1992. Morphometric analysis of honey bees from an area of racial hybridization in northeastern Italy. Apidologie 23:89–96. Nielsen D, Page RE Jr, and Crosland MWJ, 1994. Clinal variation and selection of MDH allozymes in honey bee populations. Experientia 50:867–871. Oakshott JG, Gibson JB, Anderson PR, Knibb WR, Anderson DG, and Chambers GK, 1982. Alcohol dehydrogenase and glycerol-3-phosphate dehydrogenase clines in Drosophila melanogaster on different continents. Evolution 36:86–96. Oldroyd BP, Sheppard WS, and Stelzer JA, 1992. Genetic characterization of the bees of Kangaroo Island. J Apic Res 31:141–148. Oldroyd BP, Cornuet J-M, Rowe D, Rinderer TE, and Crozier RH, 1995. Racial admixture of Apis mellifera in Tasmania, Australia: similarities and differences with natural hybrid zones in Europe. Heredity 74:315–325. Oldroyd BP, Thexton EG, Lawler SH, and Crozier RH, 1997. Population demography of Australian feral bees (Apis mellifera). Oecologia 111:381–387.


Isozyme Patterns and Their Inheritance in the Tachinid Compsilura concinnata V. Sańchez and R. Carde´ An isozyme survey and mating study was used to investigate the genetic basis of variation in polymorphic enzymes of the parasitic fly Compsilura concinnata (Meigen), an exotic biological control of lepidopteran pests. We investigated the isozyme variation for 34 enzyme systems in three colonies of C. concinnata, and in offspring from matings of two colonies, all reared on larvae of Lymantria dispar L. We found 37 loci, 14% of which were polymorphic. To determine the enzymes’ inheritance we followed the patterns of transmission from parents to offspring for 35 isozymes. We found monomorphic banding patterns in 30 enzyme systems, the majority being single bands that migrated similarly in the gels. Five isozymes (-Est, Fum, Ha, Mpi, Tpi) were polymorphic and their inheritance followed the expectations for simple Mendelian inheritance, confirming them as candidate allozymes of C. concinnata. Estimates of average heterozygosity for laboratory populations of this parasitoid were half those of other Diptera, but the estimates were higher than those of a primary host, L. dispar, and of some parasitic Hymenoptera. Starch-gel electrophoresis, a relatively rapid method of obtaining information on genetic structure, is used extensively to describe molecular genetic variation within and between populations for a variety of

taxa (Selander et al. 1971; Utter et al. 1987; May 1992). Identifying isozymes of a taxon is a first step in the electrophoretic study of a population’s genetic structure. Using statistical methods is one way to infer the pattern of inheritance for allozymes, but breeding studies provide the most direct evidence of enzyme heritability (Wagner and Selander 1974). This is important because isozymes on an electrophoretic gel may be diagnostic of alleles at an enzyme locus, but not all patterns are of genetic origin ( Harris et al. 1977). Generally, allelically determined enzymes, or allozymes, follow patterns of simple Mendelian inheritance ( Hubby and Lewontin 1966; Lewontin 1985). The tachinid fly Compsilura concinnata (Meigen) was one of the first exotic parasitoids released into North America for biological control of the forest pest the gypsy moth, Lymantria dispar L., and the browntail moth, Euproctis chrysorrhoea L. (Schaffner 1934). This tachinid is an important biocontrol agent because of its capacity to respond to and maintain hosts at low densities ( Elkinton et al. 1990; Ferguson et al. 1994; Gould et al. 1990; Sisojeviç 1979; Skinner et al. 1993). There is biological and ecological information on C. concinnata (Schaffner 1934; Sisojeviç 1979), but its genetic structure is unknown. Our objective was to identify the range of isozyme patterns in different colonies of C. concinnata and to determine the mode of inheritance of these isozymes. We used enzyme electrophoresis to identify the range of isozyme patterns in three colonies of C. concinnata. We then examined the resolvable isozymes of breeding pairs and their progeny to determine patterns of isozyme inheritance and identify those that were allozymes.

Materials and Methods Laboratory Culture of C. concinnata and Isozyme Identification Isozymes in adult C. concinnata were surveyed in various generations of three populations reared at the USDA Forest Service Northeastern Center for Forest Health Research, Hamden, Connecticut. Populations 1–5 were the 28th, 29th, 30th, 32nd, and 35th generations, respectively, of a laboratory colony founded by adults reared from late-instar L. dispar collected in 1987 in the Quabbin Watershed near Belchertown in central Massachusetts. Populations 6 and 7 were the fourth and fifth generations, respectively, of a colony founded

with flies reared from L. dispar larvae collected in 1993 in the Rose Lake Resource Area, Shiawasee County, Michigan. Population 8 was the 4th generation of a colony founded in 1993 with flies from hosts collected in the River Road Watershed, Hamden, Connecticut. Flies in all colonies were kept in cages of 150–200 flies of mixed age. The hosts used in all facets of this study were gypsy moth larvae of the New Jersey Standard Strain reared to the 4th instar following the protocols described by ODell et al. (1985) in 0.18 L plastic disposable cups (ù20–40 larvae per cup) with 60 ml of gypsy moth diet ( Bell et al. 1981). Hosts were reared in groups of 8 in 0.18 L plastic cups with 40 ml of diet kept at 25 6 28C, 16 h light:8 h dark, and 60 6 10% RH. Five days after flies had begun to emerge in a population, 40–60 flies were removed, anesthetized in a 2208C freezer (ù20 s), placed individually in labeled 1.8 ml cryotubes filled with liquid nitrogen, and stored at 2708C until electrophoresis. Mating Methods for Following Isozyme Inheritance The inheritance of isozymes in C. concinnata was examined using two colonies reared at Hamden, Connecticut. The NY population was from the fifth generation of a colony collected in lot 20 in Hightower State Forest, Lewis County, New York. The BN population was from the second generation of a colony formed from the cross between the NY and Massachusetts colonies. Following a 9–12 day egg maturation period and daily thereafter for 8 days, mated pairs of adults from each sample colony were transferred to fresh 0.47 L cups containing 15 fourth-instar hosts that were 48–72 h old. Flies demonstrating signs of senescence (e.g., hopping without flight) were removed and stored at 2708C to prevent loss in enzyme activity. The hosts were reared as described earlier and the progeny of flies were collected individually. As the adult progeny emerged, they were stored individually at 2708C until electrophoresis. Isozyme Electrophoresis To identify isozymes and determine the conditions for their optimal resolution, we screened a group of 34 enzymes with four gel-buffer combinations (May et al. 1989; Sańchez 1995) following the protocols for electrophoresis described by May (1992). The two discontinuous gel-buffer systems differed in Tris:citrate concentrations and were R (0.03 M:0.005 M, pH 8.5) of Ridgway

et al. (1970) and 4 (0.008 M:0.003 M, pH 6.7) of Selander et al. (1971). The electrolyte buffers for the R system included a hydroxide-boric acid solution (pH 8.1), while the 4 system had a Tris : citrate ratio of 0.223 M:0.094 M (pH 6.3). The two continuous systems consisted of a citrate (C) electrolyte buffer (0.004 M, pH 6.1) of Clayton and Tretiack (1972) and a Tris : boric acid (M) buffer (0.18 M:0.1 M, pH 8.7) of Markert and Faulhaber (1965). The buffer formulations for the gels of C and M continuous systems, respectively, were a 10% and 25% dilution of each electrolyte buffer. To minimize enzyme degradation, flies were held in ice baths during all phases of preparation for electrophoresis. Sample flies were homogenized individually in 5 ml test tubes with 200 ml of 0.05 M Tris extraction buffer (pH 7.1), centrifuged for 5 min at 1800 rpm, drawn onto Schuller paper wicks ( Norfolk Paper Products), and loaded into gels. A wick saturated with red food dye ( Durkee-French, Inc.) was placed at the first and every 11th gel lane as a migration marker. Electrophoresis ended after approximately 4 h when the dye marker had migrated approximately 7 cm in the gel. To obtain duplicate gel slices for evaluating multiple enzymes for an electrophoretic run, each gel was sliced into 5–6 layers. Individual slices were placed on glass plates and stained as agar overlays except for the enzymes LAP, ACP, and those forming fluorescent products (a-GLU, HA, aMAN, MUP). Fluorescent stains were poured on filter paper that was layered on the gels and left on approximately 15–20 s. All gels were covered with clear, nonreactive plastic film (Saran Wrap) and, except for fluorescent stains, were placed in a dark incubation chamber at 30 6 28C. Fluorescent gels were scored and photographed on a UV table within 10–20 min during the optimum window of resolution. The remaining gels were scored and photographed after their colored products had precipitated fully. Gels were scored by associating single- and multiple-band patterns with putative alleles based on the enzyme’s subunit composition and assignment of the simplest model of Mendelian inheritance. The alleles were numbered consecutively, with 1 as the most common allele. Nomenclature follows May et al. (1989), with names of enzyme systems in normal font and those for loci italicized. Gel profiles with similar substrates were compared to determine whether the same isozyme was resolved on different gels (Richardson et al. 1986).

Electromorph patterns were identified and those with genetically interpretable patterns were presumed to reflect allelic differences at a locus. Isozymic variability is a product of codominant characters, hence the relative allele frequencies were calculated from these zymograms. The relative allele frequencies were evaluated for conformity to Hardy–Weinberg equilibrium ( HWE) expectations under random mating and were used to estimate the levels of heterozygosity or genetic diversity in each colony ( Nei 1987). Enzyme loci with a frequency of x(i) # 0.975 for their most common allele were considered polymorphic, whereas the observed heterozygosity H was calculated as the average heterozygosity over all loci. Analysis of Isozyme Inheritance Isozyme genetics in C. concinnata was investigated by examining the genetic patterns in 24 parent-pair crosses and the inheritance of these isozymes in their progeny. We followed the methods described earlier ( Table 1, excluding SOD, two electromorphs from MDH, one from ADH, one from EST, and all other enzymes with zero resolution) to screen each parent in a pair for 30 enzymes. The isozyme patterns of each parent and all progeny were scored and assigned alleles following the simplest interpretation of Mendelian inheritance. The relative frequency of each genotype class of progeny was compared with those expected under a model of Mendelian segregation with independent assortment. The relative frequencies of alleles at polymorphic loci were then compared for their goodness-of-fit with the proportions expected under HWE. We used the Genes in Populations software developed at the Cornell Laboratory for Ecological and Evolutionary Genetics (May et al. 1992) to compare goodness-of-fit with a single locus model of Mendelian genetics, assuming segregation and independent assortment (Weir 1990), and to investigate linkage disequilibrium and sex linkage among loci.

Results Isozyme Patterns In this survey of 34 enzyme systems among various subpopulations from three C. concinnata colonies, 37 isozymes were resolved ( Table 1) and identified for use in the inheritance study. Four enzymes (ADH, MDH, -MAN, and EST ) had multiple electromorphs on the gels; their mobility


Table 1. Banding patterns resolved (res.) in 34 isozymes of Compsilura concinnata

Enzyme

Abbreviation

Alcohol dehydrogenase Xanthine dehydrogenase Lactate dehydrogenase Glycerate dehydrogenase Hydroxybutyric dehydrogenase Malate dehydrogenase Malic enzyme Isocitrate dehydrogenase Phosphogluconate Glycerol-3-phosphate dehydrogenase Superoxide dismutase Glyceraldehyde-3-phosphate dehydrogenase Diaphorase Aspartate aminotransferase Glucokinase Adenylate kinase Fructose biphosphate Acid phosphatase a-Glucosidase a-Mannosidase Hexoseaminidase Peptidase-GL Leucine aminopeptidase Guanine deaminase Aldolase Fumarase Aconitase Mannosephosphate isomerase Glucosephosphate isomerase Triose phosphate isomerase Phosphoglucomutase Methylumbelliferyl phosphate Esterase Galactosaminidase Total

ADH XDH LDH G2DH HBDH MDH ME IDH PGD G3P SOD GAPDH DIA AAT GK AK FBP ACP a-GLU a-MAN HA PEP-GL LAP GDA ALD FUM AC MPI GPI TPI PGM MUP EST GAM

E.C.a number 1.1.1.1 1.1.1.204 1.1.1.27 1.1.1.29 1.1.1.30 1.1.1.37 1.1.1.40 1.1.1.42 1.1.1.44 1.1.1.8 1.15.1.1 1.2.1.12 1.8.1.4 2.6.1.1 2.7.1.2 2.7.4.3 3.1.3.11 3.1.3.2 3.2.1.20 3.2.1.24 3.2.1.52 3.4.11 3.4.11.1 3.5.4.3 4.1.2.13 4.2.1.2 4.2.1.3 5.3.1.8 5.3.1.9 5.3.11 5.4.2.2

Gelb

n

M R C R R C M 4 4 4 All C 4 C R C C C R R C R 4 M 4 4 C M R R 4 C R R

58 158 198 40 17 138 198 150 157 191 198 40 78 91 196 156 40 192 188 196 194 173 20 98 145 188 198 178 158 198 175 140 50 186

Res. mono. 2 1 0 1 1 1d 1 1 1 1d 1 1 1d 1d,e 1 1 1f 1d,e 1 3 0 1 1 1 1f 0 0 0 1d 0 1 1 2 1 32

Mobilityc 2 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 2 1 1 1 1 1 1

1 1 2 1 1

Res. poly. 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 1 0 0 0 0 1 0 5

Mobility

2

1

1 1

1

Total res. 2 1 0 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 3 1 1 1 1 1 1 0 1 1 0 1 1 3 1 37

Enzyme commission number. Gel buffer with optimal resolution. c Mobility relative to origin as toward anode (1) or cathode (2). d Up to five additional posttranslational product bands observed. e,f Systems with the same letter appear to code for the same locus. a

b

patterns suggested they were products of multiple homozygous loci for the same functional enzyme and probably produced by at least six additional loci ( Table 1). Electromorphs with different migration patterns were observed in the a-EST, bEST, FUM, GPI, a-GLU, a-MAN, MDH-1, MPI, and PGM enzyme systems. The isozyme patterns observed in this analysis were of three types: a single electromorph with fixed migration distance, multiple isozymes migrating different distances, and multiple isozymes with similarities among samples in one or more bands. Of the enzymes surveyed initially, 77% had a one-band pattern and were

scored as products of monomorphic loci. Some enzymes contained distinct multiple electromorphs ( Table 1), which suggested that their isoenzymes either were coded by multiple loci or were posttranslational products. The enzymes GPI, MDH, and DIA were exemplary in having such variable electromorphs, with DIA-stained gels displaying cross-activity for other non-DIA electromorphs ( Figure 1). The enzyme MPI demonstrated duplicate expression with the strongly staining ME. The additional posttranslational electromorphs of ME were not scored on the MPI gel. In 28 enzyme systems, 32 putative loci were found with invariant patterns while 3

Figure 1. Electromorph patterns for 20 C. concinnata samples in gel stained for Diaphorase ( DIA) that also shows numerous non-DIA isoenzymes (e.g., lanes with boxed areas).

Figure 2. Polymorphism in C. concinnata for the Fumarase locus exhibiting narrow patterns characteristic of homozygous individuals (e.g., lanes 1 and 8) and broad patterns characteristic of heterozygotes (e.g., lanes 10 and 20).


of the stain systems could not be resolved. Electromorphs were duplicated in three different enzyme systems and were named for the system in which they appeared consistently: ACP 5 AAT, ALD 5 FBP, and ME 5 one electromorph in MPI. Fumarase provided readily interpretable phenotype patterns for surmising the sample’s allelic constitution ( Figure 2). In the FUM system, C. concinnata exhibited one of three electrophoretic patterns, that is, each homozygous phenotype had only one electromorph, while the heterozygotes had the five-banded pattern expected for enzymes with a tetrameric subunit structure. Putative allele frequencies between sexes of a population demonstrated no systematic differences and so were pooled within a sample to obtain diversity statistics for each of the eight populations (Table 2). A qualitative comparison of the average heterozygosity observed (Ho) did not seem to deviate from the average heterozygosity expected (Hs) under HWE. The

Table 2. Measures of diversity in putative genotype frequencies from isozyme patterns of C. concinnata

a b

Sample

Polymorphica Ho 6 SEb

1 2 3 4 5 6 7 8

0.150 0.136 0.150 0.200 0.125 0.158 0.160 0.174

0.068 0.095 0.090 0.171 0.097 0.064 0.060 0.094

6 6 6 6 6 6 6 6

Hs 6 SE

0.038 0.053 0.056 0.081 0.051 0.048 0.037 0.060

0.062 0.064 0.072 0.104 0.066 0.046 0.045 0.067

6 6 6 6 6 6 6 6

Ht 6 SE 0.034 0.035 0.039 0.047 0.032 0.034 0.024 0.034

0.068 0.064 0.072 0.104 0.062 0.046 0.052 0.067

6 6 6 6 6 6 6 6

N 0.037 0.035 0.040 0.047 0.032 0.030 0.027 0.034

20 40 20 20 20 18 20 20

Percent of loci considered polymorphic if frequency of common allele #0.975. Heterozygosities (H) calculated for frequencies observed (o), expected (s), and total (t).

occasional large differences between Hs and Ho under HWE indicated that loci were biased generally more toward heterozygote individuals. Polymorphism over all populations was in the range of 0.161– 0.135 ( Table 3). Analysis of Isozyme Inheritance: Offspring and Parent Electrophoresis Females in 10 families of the 24 crosses were sufficiently fecund and the pair’s isozymes sufficiently diverse for the electrophoretic analysis of their electromorph inheritance. Thirty-five loci were identified using the 30 enzyme stains and four buffer systems. The zymograms in 30 enzyme

systems of both parents and offspring showed the single bands of even intensity expected for monomorphic loci ( Table 1). Electromorphs in four loci (Adh, a-Est, aMan, Mdh) were resolved consistently with even intensity, were not reproduced on multiple gels using different substrates, and lacked the indeterminacy characteristic of nongenetic posttranslational products. The consistent transmission of each of the 30 patterns from parents to offspring suggests these are single classes of allozymes coded at different monomorphic loci. To compare the ratios expected under HWE, progeny with similar phenotypes

Table 3. Relative gene frequencies and measures of diversity in five putative allozymes of C. concinnata

Enzyme locus b-Est

Mdh-1

Ha

Mpi

Fum

Relative frequencies of allelesa Population

1

2

Ho

Hs

1 5 7 4 6 7 8 2 3 4 6 8 1 2 3 4 5 7 8 1 2 3 4 5 6 7 8

0.438 0.500 0.714 0.500 0.944 0.925 0.950 0.514 0.513 0.500 0.472 0.675 0.600 0.575 0.338 0.583 0.675 0.475 0.500 0.250 0.700 0.471 0.575 0.525 0.156 0.132 0.500

0.563 0.500 0.286 0.500 0.056 0.075 0.050 0.486 0.488 0.500 0.528 0.325 0.400 0.425 0.663 0.417 0.325 0.525 0.500 0.751 0.300 0.529 0.425 0.475 0.844 0.868 0.500

0.375 1.000b 0.286 1.000b 0.111 0.150 0.100 0.811b 0.975b 1.000b 0.833b 0.050b 0.600 0.750b 0.575 0.611 0.650b 0.850b 1.000b 0.389 0.500 0.314b 0.650 0.750b 0.063 0.158 1.000b

0.492 0.500 0.408 0.500 0.105 0.139 0.095 0.500 0.500 0.500 0.498 0.439 0.480 0.489 0.447 0.486 0.439 0.499 0.500 0.375 0.420 0.498 0.489 0.499 0.264 0.229 0.500

Fisc

N

0.238 21.000 0.300 21.000 20.059 20.081 20.053 20.623 20.951 21.000 20.672 0.886 20.250 20.535 20.286 20.257 20.481 20.704 21.000 20.037 20.190 0.369 20.330 20.504 0.763 0.309 21.000

16 20 14 20 18 20 20 37 40 20 18 20 20 40 40 18 20 20 20 18 40 35 20 20 16 19 16

The number of alleles at these loci was based on the simplest interpretation of isozyme patterns observed. The observed heterozygosities (Ho) deviate significantly from those expected (Hs) under Hardy–Weinberg equilibrium. c Fixation index estimating correlation between two uniting gametes relative to the subpopulation.

a

b

were pooled and their genotypic classes were derived for each enzyme system. In the 45 single-locus crosses with 10 families, no new classes of progeny were produced at the five polymorphic loci (b-Est, Fum, Ha, Mpi, and Tpi) that were not expected based on the parental genotypes ( Table 4). The genotypes of progeny from these controlled matings approached ratios expected under HWE in 59% of the crosses and may reflect unknown loss of phenotypes from unobservable mortality inherent in rearing obligate parasitosids. No sex linkage or linkage disequilibrium was apparent among the loci. The double heterozygous cross at the b-Est locus and the homozygous-heterozygous cross at the Tpi gene had a slight though not statistically significant deficiency of heterozygotes in offspring ratios ( Table 4). Two alleles per locus were found in the five polymorphic putative genes, with an average 1.29 alleles per locus across all 35 loci.

Discussion Identifying isozymes and determining their variability is an initial step in conducting a molecular genetic analysis of populations using enzyme electrophoresis ( Harris et al. 1977). Although the genetic controls appear to be understood for most enzymes, these inferences need to be corroborated through systematic analysis. Subsequent analyses of populations can then proceed reliably to identify and draw conclusions about their genetic variation and population structure (Murphy et al. 1990). We identified 35 isozyme loci with 14% polymorphism in C. concinnata and determined their mode of inheritance and allozymic characteristics. To date, no similar studies of a tachinid parasitoid have been reported. Inheritance patterns in the isozymes we surveyed suggest that many loci were fixed for single alleles. This was not unexpected because laboratory rearing can select against genotypes and so fix otherwise rare genetic variants or eliminate relatively common alleles (Mackauer 1976). Variation in these laboratory and near-wild populations of C. concinnata, as measured by estimates of heterozygosity, was about half that in other Diptera (Drosophila spp. H 5 0.150) ( Nevo 1978) and near the level of haploid-diploid wasps (H 5 0.062) (Graur 1985). Because these were laboratory populations without immigration, it also was possible that some loci became fixed through random genetic


Table 4. Genotypes of C. concinnata parents and their progeny from controlled matings evaluated for five polymorphic enzymes Parent genotypesa Enzyme locus b-Est

Fum

Ha

Mpi

Tpi

a b

Clayton JW and Tretiack DN. 1972. Amine-citrate buffers for pH control in starch gel electrophoresis. J Fish Res Bd Can 29:1169–1172.

Progeny count by sex and genotype

Female

Male

Number of families

11 12 11 12 11 11 11 12 22 22 12 12 11 12 11 12 11 11

12 11 11 12 —b 11 12 11 12 11 12 11 12 11 12 12 12 11

3 1 1 2 2 2 3 2 1 1 3 3 3 3 3 3 3 6

letin 1584. Washington, D.C.: U.S. Department of Agriculture; 599–633.

Female 11

12

22

11

12

22

7 2 4 3 8 6 5 4 0 0 0 4 4 4 7 3 10 17

4 2 0 1 3 0 6 1 4 4 1 10 8 9 3 8 2 0

0 0 0 1 0 0 0 0 0 0 5 0 0 0 0 0 0 0

6 2 12 4 12 5 9 11 0 0 1 5 12 14 1 13 19 18

6 5 0 1 2 0 6 4 6 8 12 6 9 9 9 15 4 0

0 0 0 4 0 0 0 0 7 0 1 0 0 0 0 1 0 0

Male

Genotypes are specified with 1 5 most common allele and 2 5 alternate allele. Genotypes could not be resolved; however, based on offspring, they probably were heterozygotes.

Culver JJ, 1919. A study of Compsilura concinnata, an imported tachinid parasite of the gypsy moth and the brown-tail moth. USDA technical bulletin 766. Washington, D.C.: U.S. Department of Agriculture. Elkinton JS, Gould JR, Ferguson CS, Liebhold AM, and Wallner WE, 1990. Experimental manipulation of gypsy moth density to assess impact of natural enemies. In: Population dynamics of forest insects (Watt AD, Leather SR, Hunter MD, and Kidd NAC, eds). Andover, UK: Intercept; 275–287. Ferguson CS, Elkinton JS, Gould JR, and Wallner WE, 1994. Population regulation of gypsy moth ( Lepidoptera: Lymantriidae) by parasites: Does spatial density dependence lead to temporal density dependence? Environ Entomol 23:1155–1164. Futuyma DJ and Peterson SC, 1985. Genetic variation in the use of resources by insects. Annu Rev Entomol 30:217–238. Gould JR, Elkinton JS, and Wallner WE, 1990. Densitydependent suppression of experimentally created gypsy moth, Lymantria dispar ( Lepidoptera: Lymantriidae), populations by natural enemies. J Anim Ecol 59: 213–233. Graur D, 1985. Gene diversity in Hymenoptera. Evolution 39:190–199.

drift ( Lewontin 1985). However, the length of time that the stock strain and the other two near-wild strains have been in culture, 2 years and 3–4 months, respectively, seems insufficient for differentiation from either drift or mutation (Roush 1990). This study enabled a genetic analysis of C. concinnata populations from Michigan, Pennsylvania, New York, Connecticut, and Maine, that found seven polymorphic loci of 35 allozymes surveyed and little sitespecific variability (Sańchez and Carde´ 1998). It should be noted that the gene pool of C. concinnata in North America probably experienced a severe bottleneck before their release in North America, both from the initial collections from a limited number of sites in Europe and from selection or drift during rearing before their release (Culver 1919). After its introduction, C. concinnata spread rapidly throughout the Northeast and was collected from many areas in advance of the arrival of gypsy and browntail moths (Culver 1919; Schaffner 1934). As parasites coevolve with their hosts, genetic variability might approach similar levels in both species ( Futuyma and Peterson 1985). Analyses of allozymes in L. dispar showed that heterozygosity averaged 0.000–0.008, and that single-locus heterozygosity provided sufficient information to identify populations in North America and western Europe ( Harrison et al. 1983). In that same study the North American and western European populations of L. dispar were the most similar and least variable in


a spectrum of genetic diversity for gypsy moth that increases in the Palearctic from west to east toward Asia ( Harrison et al. 1983). The heterogeneity (H 6 SE) expected under HWE for the 35 loci in our laboratory colonies of C. concinnata (0.047 6 0.021) is higher than Harrison et al. (1983) reported for L. dispar and may reflect this parasitoid’s generalist qualities. To improve our understanding of how the population genetics of Palearctic parasitoids introduced into North America changes after their release, we need more information on their genetic diversity in their native Eurasian habitats.

Harris HD, Hopkinson DA, and Edwards YH, 1977. Polymorphism and the subunit structure of enzymes: a contribution to the neutralist-selectionist controversy. Proc Natl Acad Sci USA 74:698–701. Harrison RG, Wintermeyer SF, and ODell TM, 1983. Patterns of genetic variation within and among gypsy moth, Lymantria dispar ( Lepidoptera: Lymantriidae), populations. Ann Entomol Soc Am 76:652–656. Hubby JL and Lewontin RC, 1966. A molecular approach to the study of genic heterozygosity in natural populations. I. The number of alleles at different loci in Drosophila pseudoobscura. Genetics 54:577–594. Lewontin RC, 1985. Population genetics. Annu Rev Genet 19:81–102. Mackauer M, 1976. Genetic problems in the production of biological control agents. Annu Rev Entomol 21:369– 385. Markert CL and Faulhaber I. 1965. Lactate dehydrogenase isozyme patterns of fish. J Exp Biol 159:319–332.

From the USDA Forest Service, Northeastern Center for Forest Health Research, Hamden, CT 06514-1777 (Sańchez) and the Department of Entomology, University of Massachusetts, Amherst, Massachusetts (Carde´). R. Carde´ is currently at the Department of Entomology, University of California, Riverside, California. This work was supported by a grant from the USDA Forest Service’s Northeastern Forest Experiment Station. D. E. Leonard and J. S. Elkinton ( University of Massachusetts), J. R. Powell ( Yale University), and T. M. ODell ( USDA Forest Service, retired) provided valuable insight on this study. The insect rearing staffs at the Northeastern Center for Forest Health Research and the USDA-APHIS Methods Development Center provided the gypsy moths used in this study. K. Hopper, G. S. Walton, and N. R. Dubois reviewed earlier drafts of this manuscript. Address correspondence to V. Sańchez at the address above.

May B, Marsden JE, and Schenck CG, 1989. Electrophoretic procedures, recipes, and nomenclature used in the Cornell Laboratory for Ecological and Evolutionary Genetics. Ithaca, New York: Cornell University Laboratory for Ecological and Evolutionary Genetics.

q 1999 The American Genetic Association

Nei M, 1987. Molecular evolutionary genetics. New York: Columbia University Press.

May B, 1992. Starch gel electrophoresis of allozymes. In: Molecular genetic analysis of populations: a practical approach ( Hoelzel AR, ed). Oxford: Oxford Press; 1–27, 271–280. May B, Kruge B, and Eng E, 1992. Gene in populations. Ithaca, New York: Cornell University Laboratory for Ecological and Evolutionary Genetics. Murphy RW, Sites JW Jr, Buth DG, and Haufler CH, 1990. Proteins I: isozyme electrophoresis. In: Molecular systematics ( Hillis DM and Moritz C, eds). Sunderland, Massachusetts: Sinauer; 45–126.

Nevo E, 1978. Genetic variation in natural populations: patterns and theory. Theor Popul Biol 13:121–177. References Bell RA, Owens CD, Shapiro M, and Tardif JR, 1981. Development of mass-rearing technology. In: The gypsy moth: research toward integrated pest management ( Doane CC and McManus ML, eds). USDA technical bul-

ODell TM, Butt CA, and Bridgeforth AW, 1985. Lymantria dispar. In: Handbook of insect rearing (Singh P and Moore RF, eds). New York: Elsevier; 335–367. Richardson BJ, Baverstock PR, and Adams M, 1986. Allozyme electrophoresis: a handbook for animal sys-

tematics and populations studies. Sydney: Academic Press. Ridgway GJ, Sherburne SW, and Lewis RD. 1970. Polymorphism in the esterases of Atlantic herring. Trans Am Fish Soc 99:147–151. Roush RT, 1990. Genetic variation in natural enemies: critical issues for colonization in biological control. In: Critical issues in biological control (Mackauer M, Ehler LE, and Roland J, eds). Andover, UK: Intercept; 263–288. Sańchez V, 1995. The genetic structure of northeastern populations of the tachinid, Compsilura concinnata (Meigen), an introduced parasitoid of exotic forest defoliators of North America (PhD dissertation). Amherst, Massachusetts: University of Massachusetts. Sańchez V and Carde´ RT. 1998. Allozyme variability and genetic structure of Compsilura concinnata ( Diptera: Tachinidae) populations in the northeastern United States. Ann Entomol Soc Am 91:211–216. Schaffner JV Jr, 1934. Introduced parasites of the brown-tail and gypsy moths reared from native hosts. Ann Entomol Soc Am 27:585–592. Selander RK, Smith MH, Yang SY, Johnson WE, and Gentry J, 1971. Biochemical polymorphism and systematics in the genus Peromyscus. I. Variation in the old field mouse (Peromyscus polionotus). Studies in Genetics VI, University of Texas publication 7103. Austin: University of Texas Press; 49–90. Sisojeviç P, 1979. Interactions in the host parasite system, with special references to the gypsy moth-tachinids (Lymantria dispar L.—Tachinidae). Papers of the Sixth Interbalcanic Plant Protection Conference, Izmir, Turkey, 10–16 October 1977. Izmir, Turkey: Turkish Ministry of Food, Agriculture, and Animal Husbandry; 108–114. Skinner M, Parker BL, Wallner WE, ODell TM, Howard D, and Aleong J, 1993. Parasitoids in low-level populations of Lymantria dispar [Lep: Lymantriidae] in different forest physiographic zones. Entomophaga 38:15– 29. Utter FM, Aebersold P, and Winans G, 1987. Interpreting genetic variation detected by electrophoresis. In: Population genetics and fishery management (Ryman N and Utter F, eds). Pullman, Washington: University of Washington Press; 21. Wagner RP and Selander RK, 1974. Isozymes in insects and their significance. Annu Rev Entomol 19:177–138. Weir BS, 1990. Genetic data analysis: methods for discrete data analysis. Sunderland, Massachusetts: Sinauer. Received June 29, 1998 Accepted February 18, 1999 Corresponding Editor: Ross MacIntyre

Genetics of Creamish White, an Eye Color Mutant in Anopheles stephensi T. Adak, S. Wattal, S. Kaur, and V. P. Sharma Genetic analysis of a new eye color mutant, creamish white (cw), has been described in Anopheles stephensi, a major vector of malaria in the Indo-Pakistan subcontinent. Inheritance pattern revealed that it is sex linked and recessive to wild eye color. Creamish white eye was found to be nonallelic and epistatic to another re-

cessive, sex-linked mutant, red eye (r). The map distance between cw and r was estimated as 41.80 6 0.99.

Introduction Anopheles stephensi is an important vector of malaria on the Indo-Pakistan subcontinent and in parts of the Middle East. Among all the morphological markers in An. stephensi, a total of nine eye color mutants have been described: white eye, w (Aslamkhan 1973); colorless eye, c (Sharma et al. 1977); red eye, r (Sharma et al. 1979); rosy eye, wro (Aslamkhan and Gul 1979); maroon eye, mar (Mahmood and Sakai 1982); chestnut eye, wct (Rathor et al. 1983), scarlet eye, wsca; red-spotted eye, prs; and pigmentless eye, p (Akhtar and Sakai 1985). Among these, w, wro, wct, r, wsca, prs, and p mutants are reported as sex-linked recessive and placed in linkage group I, whereas c and mar are recessive autosomal and are placed in linkage group II. This article describes genetic analysis of a new eye color mutant, creamish-white (cw), and its linkage relationship with r, a sex-linked recessive mutant (Sharma et al. 1979).

Materials and Methods The following mosquito strains were used in the genetic crosses. Wild type (1/1). A wild-type laboratory colony. The eye color from the first instar through the adult stage is blackish brown. Red eye (r). This strain is homozygous for red eye (Sharma et al. 1979). The mutant expresses its phenotype red eye color from the first instar through the adult stage. Creamish-white (cw). A few mutant individuals were isolated from a laboratory colony. The mutant expresses its phenotype as white-colored eyes from the first instar through the adult stage. The rearing of larvae and the handling of adults was done as described previously (Sharma et al. 1977).

Results and Discussion In the cross between the wild female and the cw male, all the F1 progeny were wild type, suggesting that the cw mutant is recessive. In the reciprocal cross, all the female progeny were wild type, whereas all males showed a mutant phenotype, suggesting that the mutation is sex linked.

When F1 progeny from the first cross were inbred, wild females, wild males, and mutant males were obtained in 2:1:1 ratio, whereas inbreeding of F1 progeny of the reciprocal parental cross resulted in wild and mutant females and wild and mutant males in a ratio of 1:1:1:1. Backcrosses of heterozygous F1 females from both the reciprocal crosses to mutant males resulted in wild and mutant female and male progeny in a 1:1:1:1 ratio, while when backcrossed to wild-type males, only wild-type females, wild-type males, and mutant males were obtained in 2:1:1 ratio. The results of inbreeding and backcrosses clearly confirm that the cw mutant is sex linked and recessive to wild type. The cw mutant strain was also reciprocally crossed with another sex-linked mutant strain, red-eye (r) (Sharma et al. 1979), to find its linkage relationship. The results of these crosses are presented in Table 1. When cw females were crossed with r males (cross 1) in F1, only wild females and white eyed males were obtained in a 1:1 ratio, and in the reciprocal cross where r females were crossed with cw males (cross 2) in F1 progeny, all females were wild type and all males were red eye. The presence of only wild-type females in both the reciprocal crosses 1 and 2 clearly suggest that cw is not allelic to r. Heterozygous F1 females from crosses 1 and 2 were backcrossed to r males (crosses 3 and 5) and cw males (crosses 4 and 6), respectively. The resulting progeny from crosses 3 and 5 did not show any white eye females, while crosses 4 and 6 did not show any red eye females, as expected, although sex ratio in these two crosses did not deviate significantly from a 1:1 ratio. However, in all these four crosses (crosses 3–6), relatively fewer numbers of wild males were observed as compared to wild females. This was not due to the differential mating, but represented only one class of crossover (cw1 r1). Similarly, in these crosses red eye males (cw1 r) were fewer compared to other phenotypes, as this category was represented by only parental noncrossover phenotypes, as expected. Among the females, expected recombinants in wild and red eye categories in crosses 3 and 5, and wild and white eye in crosses 4 and 6, would not be distinguishable from parental type. Among the males, expected double mutants (cw r) also could not be distinguished from the parental white eye (cw r1) phenotype in these two crosses. However, double the number of white eye males were observed in both these crosses, suggesting that in the hem-


Table 1. Results of crosses to elucidate the linkage relationship of creamish-white (cw) and red eye (r) in Anopheles stephensi Progeny phenotypes Parental genotype Female 1. 2. 3. 4. 5. 6.

cw1 r cw1 r cw1 r cw1 r cw1 r cw r1 cw1 r cw r1 cw r1 cw1 r cw r1 cw1 r

Male 3 cw1 r 3 cw r1 3 cw r

Wild

No. of families tested Female

White eye Male

Female

x2

Red eye Male

Female

Male

Total

Sex

Linkage

Recombinant (%)

10

373

0

0

392

0

0

765

0.24*

13

895

0

0

0

0

890

1785

0.01*

13

896

359

0

843

817

476

3391

0.37*

79.43**

42.99 6 1.71

13

745

295

713

692

0

435

2870

0.74*

90.43**

40.41 6 1.82

12

455

179

0

470

457

280

1841

0.39*

9.88**

39.00 6 2.28

15

429

204

432

446

0

253

1764

1.00*

90.86**

1

3 cw r1 3 cw1 r 3 cw r

1

44.64 6 2.33 x¯ 5 41.80 6 0.99

* p , .05; ** p . .05.

izygous condition the cw r genotype in males probably expresses its phenotype as white eye and is indistinguishable from the cw r1 genotype. This also indicates that cw is epistatic to r. To test this hypothesis, white eye males (cw r1 or cw r) obtained from cross 3 and 4 were crossed with red eye and white eye females separately. Resultant progeny in some of the isofemale cultures from the cross where red females were used showed wild females and red males, while other isofemale cultures showed red females and red males. However, in other crosses where white females were used, all isofemale cultures showed only white females and white males. This clearly suggests that white males of cross 3 and 4 must be having two types of genotypes, that is, cw r1 (parental) and cw r (crossover), but are identical in phenotypic expression, that is, white eye. This is an example of epistasis in which the cw gene hides the effect of r in the hemizygous condition in males. A similar phenomenon of epistasis was reported between an autosomal 2 marker c and a sex-linked marker r in An. stephensi (Subbarao and Adak 1981). It may be mentioned that in crosses 3– 6 only one type of recombinant, that is, the wild male phenotype (cr1 r1), could be identified unambiguously out of the five progeny categories, while other male and female expected recombinants apparently were indistinguishable from the parental phenotypes. Therefore, to estimate the recombination frequency between cw and r loci, only wild male phenotype numbers were used and the recombination frequency calculated by this was 41.80 6 0.99. It should be mentioned that two other mutants reported in An. stephensi—pig-


mentless eye (p) and white eye (w)—had similar phenotypic expression, that is, white eye color, as in the present new mutant, creamish white (cw). However, the relationship of cw to either the p or w locus is not known, as no test of allelism has been made. From the Malaria Research Centre ( ICMR), 22 Sham Nath Marg, Delhi-110 054, India. Address correspondence to T. Adak, Malaria Research Centre ( ICMR), 2 Nanak Enclave (Radio colony), Delhi-110 009, India. The authors are grateful to Pritam Singh for technical assistance. We also thank Bhopal Ram Arya for help in genetic crosses. q 1999 The American Genetic Association

References Akhtar K and Sakai RK, 1985. Genetic analysis of three new eye colour mutations in the mosquito, Anopheles stephensi. J Trop Med Parasitol 4:449–455. Aslamkhan M, 1973. Sex chromosomes and sex determination in the malaria mosquito, Anopheles stephensi. Pakistan J Zool 5:111–115. Aslamkhan M and Gul R, 1979. Inheritance of the sexlinked mutant, rosy, an allele of white in the malaria mosquito, Anopheles stephensi. Pakistan J Sci 31:245– 249. Mahmood F and Sakai RK, 1982. Genetic analysis of maroon eye in Anopheles stephensi. Mosq News 42:33–35. Rathor HR, Rashid S, and Toqir G, 1983. Genetic analysis of a new sex-linked mutant ‘chestnut eye’ an allele of white eye locus in the malaria vector Anopheles stephensi. Mosq News 43:209–212. Sharma VP, Mani TR, Adak T, and Ansari MA, 1977. Colorless-eye, a recessive autosomal mutant of Anopheles stephensi. Mosq News 37:667–669. Sharma VP, Subbarao SK, Ansari MA, and Razdan RK, 1979. Inheritance pattern of two new mutants, red-eye and greenish brown larva in Anopheles stephensi. Mosq News 39:655–658. Subbarao SK and Adak T, 1981. Linkage relationship between three autosomal mutants and the functional relationship between two eye color mutants in Anopheles stephensi. Indian J Malariol 18:98–102. Received April 21, 1997 Accepted March 31, 1999 Corresponding Editor: Glen E. Collier

Low Abundance of Microsatellite Repeats in the Genome of the Brown-Headed Cowbird (Molothrus ater) J. L. Longmire, D. C. Hahn, and J. L. Roach A cosmid library made from brown-headed cowbird (Molothrus ater) DNA was examined for representation of 17 distinct microsatellite motifs including all possible mono-, di-, and trinucleotide microsatellites, and the tetranucleotide repeat (GATA)n. The overall density of microsatellites within cowbird DNA was found to be one repeat per 89 kb and the frequency of the most abundant motif, (AGC)n, was once every 382 kb. The abundance of microsatellites within the cowbird genome is estimated to be reduced approximately 15-fold compared to humans. The reduced frequency of microsatellites seen in this study is consistent with previous observations indicating reduced numbers of microsatellites and other interspersed repeats in avian DNA. In addition to providing new information concerning the abundance of microsatellites within an avian genome, these results provide useful insights for selecting cloning strategies that might be used in the development of locus-specific microsatellite markers for avian studies. Microsatellites (tandemly repeating DNA sequences containing six or fewer nucleotides per repeat unit) serve as valuable genetic markers because they are widely distributed within the genome, highly

Table 1. Characteristics of the cowbird cosmid library DNA mass cloned

Total number of clones

Cloning efficiency (cfu/mg)

Average insert sizea

Nonrecombinant backgroundb

Cowbird genome equivalents

0.5 mg

779,100

1.65 3 106

31.3 kb

,1%

163

Average insert size was estimated by EcoRI digestion followed by agarose gel electrophoresis and fragment sizing of 15 random clones. b Nonrecombinant background was determined by ligation of cosmid cloning arms without insert DNA, followed by packaging and infection.

a

polymorphic, and easily analyzed using the polymerase chain reaction (PCR; Weber and May 1989). Microsatellites have been critical in the development of highresolution linkage maps of the human genome (Weissenbach et al. 1992). From an evolutionary perspective, microsatellites are also interesting because they appear to be present in the genomes of most, if not all, higher organisms ( Epplen et al. 1991; Tautz and Renz 1984). Investigations into the representation of different microsatellite motifs in unstudied genomes can provide further information concerning the relative abundance and evolutionary distribution of these simple repeat sequences. Using hybridization techniques, the representation of microsatellites can be examined in previously unstudied genomes in a couple of ways. First, dot blots made from genomic DNA can be hybridized to oligonucleotide probes and data resulting from such experiments can be used to quantitate and compare the abundance of distinct repeat motifs within the genomic DNAs that are examined ( Epplen 1988; Hamada et al. 1982; Tautz and Renz 1984). Another approach is to use probes to screen highly representative, genomic libraries ( Baker et al. 1995; Janecek et al. 1993; Van Den Bussche et al. 1995). A distinct advantage of screening libraries is that in addition to providing information concerning the relative abundance of the different repeat motifs, library hybridization serves to identify clones that contain microsatellites. Such clones can be further processed to develop locus-specific markers for use in genetic studies (Couch et al. 1994). Here we report on using microsatellite oligomers to screen a multiple representation cosmid library made from genomic DNA from the brown-headed cowbird (Molothrus ater). This study provides new information in a species that has not previously been examined for microsatellite density. Results from the current study are consistent with and substantiate previous observations that the overall density of

microsatellites within avian genomes is many-fold reduced compared to humans.

Materials and Methods Cosmid Library Construction The cowbird cosmid library was made using methods previously described for the construction of human chromosome-specific libraries ( Longmire et al. 1993). Briefly, a female cowbird genomic DNA sample was partially digested with Sau3AI, dephosphorylated with calf intestinal phosphatase (CIP), and ligated to sCos-1 cloning arms that had been prepared with BamHI cloning ends ( Evans et al.,1989). Ligation reactions were packaged in vitro using Gigapack Gold extracts (Stratagene, La Jolla, CA), followed by infection into Escherichia coli strain DH5aMCR. A portion of the primary library was plated on LB agar plates containing 50 mg/ml kanomycin. Random clones (1,536) were picked into sixteen, 96-well microtiter plates containing 100 ml LB/kanomycin media per well. Following overnight incubation at 378C, 100 ml of 2 3 freezing media ( LB/kanomycin with 40% glycerol) was added to each well and the plates were stored frozen at 2708C. Cosmid Gridding and Hybridizations A Biomek robotic instrument was used to construct a set of eight identical high-density gridded arrays of cowbird cosmid clones on nylon membranes as previously described ( Longmire et al. 1991). Each hybridization grid measured 8 cm 3 12 cm and contained 1,536 cosmids for a total of 47.6 Mbp of cowbird DNA (equivalent to approximately 3.2% of the cowbird genome). Microsatellite oligomers were synthesized using an Applied Biosystems model 394 DNA and RNA synthesizer (Applied Biosystems, Foster City, CA). Concatamers of microsatellite oligonucleotides were made and hybridizations were performed using direct-label enhanced chemiluminescence ( ECL) as previously reported ( Longmire and Ratliff 1994). All hybridizations were carried out at 388C in

ECL-Gold hybridization solution (Amersham, Arlington Heights, IL) containing 0.5 M NaCl, 5% blocking agent, and 10 ng/ml labeled probe. Posthybridization washes were twice for 10 min at 408C in 6 M urea, 1 3 SSC, 0.4% SDS followed by two washes 10 min each at 228C in 2 3 SSC. Chemiluminescent signal was developed according to the protocol supplied with the DirectECL kit. Luminographs were obtained by exposing hybridized membranes to Amersham Hyperfilm for periods ranging from 30 s to 60 min. Hybridization to the 17 different microsatellite probes required seven of the grids to be hybridized twice and one grid to be hybridized three times. Between hybridizations the membranes were stored moist in 2 3 SSC for 2–7 days to allow ECL signal to completely disappear. Clones displaying hybridization signals significantly above background were scored as positives.

Results Characteristics of the Cowbird Cosmid Library A cosmid library containing 779,100 primary clones was constructed by cloning 0.5 mg of partially digested and dephosphorylated cowbird genomic DNA ( Table 1). The nonrecombinant background in the library was less than 1% and the average insert size was 31.3 kb (range 19.1– 53.2 kb). Considering average insert size together with the estimation that avian genomes contain approximately 1.5 3 109 bp (Olmo et al. 1989), this cosmid library was 16-fold representative for sequences present within the cowbird genome. Frequency of Microsatellite Repeats Within the Cosmid Library Taking into account circular permutation of repeat sequences and strand complementarity, 17 microsatellite probes were required to examine all possible mono-, di-, and trinucleotide repeats and the tetranucleotide repeat (GATA)n. Representative results from the microsatellite hybridizations are displayed in Figure 1. The frequency of different microsatellites varied over a wide range and were divided into three groups based on their relative abundance ( Table 2). Group 1 consists of those microsatellites that were most abundant with over 1,000 copies per genome. The most abundant microsatellite was (AGC)n, with 8.2% of the clones being positive. Considering that the average insert size within the library was 31.3 kb, the estimated interspersion frequency ( EIF) of


Table 2. Representation of microsatellite sequences in cowbird genomic DNA cosmid library

Probe

Number positive clones

Frequency in cowbird cosmids

Group 1: Most abundant motifs (.1,000 copies per cowbird genome) 126 0.0820 (AGC)17 3 (GCT )17 82 0.0534 (GT )25 3 (AC)25 76 0.0495 (CCT )17 3 (AGG)17 64 0.0417 (AAT )17 3 (ATT )17 39 0.0254 (GATA)10 3 ( TATC)10 34 0.0221 (CT )25 3 (AG)25

Estimated interspersion frequencya

Estimated copies per genomeb

382 587 633 752 1234 1418

kb kb kb kb kb kb

3,927 2,555 2,369 1,994 1,215 1,058

Group 2: Moderately abundant motifs (100-1000 copies per cowbird genome) 28 0.0182 1722 (GTT )17 3 (AAC)17 25 0.0163 1923 (GAT )17 3 (ATC)17 13 0.0085 3688 (GGT )17 3 (ACC)17 12 0.0078 4019 (C)50 3 (G)50 11 0.0072 4354 (CCG)17 3 (CGG)17 8 0.0052 6028 (A)50 3 ( T )50 7 0.0046 6815 (AAG)17 3 (CTT )17 6 0.0039 8038 (AGT )17 3 (ACT )17

kb kb kb kb kb kb kb kb

871 780 406 373 344 249 220 186

Group 3: Infrequent motifs (,100 copies per cowbird genome) 3 0.0020 (CGT )17 3 (ACG)17 1 0.0007 (AT )25 3 (AT )25c 0 0 (CG)25 3 (CG)25c

15,674 kb 44,781 kb 0 kb

96 33 0

Each microsatellite probe was used to screen a reference set of 1,536 independent cosmid clones. Microsatellites are listed in order of decreasing frequency. Based on an average cosmid insert size of 31,347 bp. b Based on an estimated avian genome size of 1.5 3 109 bp. c Results from using each of the two probes (AT )25 3 (AT )25 and (CG)25 3 (CG)25 are probably invalid since these probes are self-complimentary and therefore are unlikely to reliably hybridize to target sequences within the cosmids due to self-annealing in solution.

a

(AGC)n repeats within the cowbird genome is one repeat cluster per 382 kb (assuming one repeat cluster per cosmid). Given that avian genomes contain approximately 1.5 3 109 bp (Olmo et al. 1989), it can be estimated that the entire cowbird genome contains approximately 3,927 (AGC)n clusters. The remainder of group 1 sequences display EIF values ranging from 587 kb for (GT )n to 1418 kb for (CT )n. Group 2 contains microsatellites that were moderate in abundance (more than 100, but less than 1,000 copies per genome). EIF values for group 2 repeats ranged from 1722 kb for (GTT )n to 8038 kb for (AGT )n. Group 3 includes those repeats that were found to occur either very infrequently or not at all within the cosmid library: (CGT )n and (AT )n displayed EIF values of 15,674 and 44,781 kb, respectively, and (CG)n failed to identify any cosmids. Twenty-three cosmids identified as being (GATA)n-positive by primary hybridization data were selected for restriction analysis, subcloning, and sequencing in order to develop locus-specific PCR primer pairs (data not shown). Of these 23 cosmids, only one was found to not contain a (GATA)n repeat. This result indicates that the primary hybridization data is quite accurate in terms of detecting cosmids that actually contain the repeats that were examined.


Discussion Compared to mammals, avian genomes contain approximately one-half the overall amount of DNA (Olmo et al. 1989), shorter intronic regions ( Hughes and Hughes 1995), and relatively fewer repetitive elements (Olofsson and Bernardi 1983; Quinn and White 1987). Although bird genomes are becoming ever more studied, only a single semicomprehensive survey of bird microsatellites has been published to date that can be used to compare to the cowbird results. Primmer et al. (1997) recently used spot blots to examine microsatellite density in three species: the chicken (Gallus gallus), the white-backed woodpecker (Dendrocopus leucotos), and the barn swallow (Hirundo rustica). Based on their hybridization data, these authors were able to estimate the relative abundance for distinct repeat motifs within the genomes of each of the three species that were examined. They also used these hybridization values to calculate an ‘‘avian average’’ for the abundance of different microsatellites. Although some probes were unique to each study, it is possible to compare some of the cowbird results with results from the Primmer et al. study. For example, the top rank order for the microsatellite motifs that were commonly examined between both studies can be compared. The order for the

five most abundant cowbird microsatellites was (AGC)n, (GT )n, (CCT )n, (AAT )n, and (GATA)n. The ‘‘avian average’’ obtained by Primmer et al. (among commonly used probes) was (GT )n, (CT )n, (ACC)n, (GATA)n, and (AAG)n. Although these rank orders are somewhat different, these results probably reflect the normal variance that is seen when comparing microsatellite rank between species ( Epplen et al. 1991; Tautz and Renz 1984). Another comparison that can be made relates to the overall density of microsatellites within the avian genome. Data generated in the present study allowed us to calculate interspersion frequencies and copy number values for each of the distinct microsatellite motifs that were examined in cowbird DNA (see Table 2). We estimated that the cowbird genome contains approximately 2,500 (GT )n repeats. This number is in fairly close agreement with the 1,500 (GT )n repeats that Primmer et al. (1997) estimated to be present in the chicken genome (GT is the only microsatellite for which these authors derived a copy number value). Most dramatic, however, is the difference in overall microsatellite density seen between avian and human genomes. Investigations of human microsatellite abundance have included hybridization-based studies ( Hamada et al. 1984; Stallings et al. 1991; Sun et al. 1984) as well as database scans (Stallings 1994; Tautz et al. 1986). Similar values have been obtained using both approaches ( Beckman and Weber 1992). In the Genbank study conducted by Beckman and Weber (1992) the overall density of human microsatellites was one repeat per 6 kb. In contrast, estimates for the overall density of all microsatellites within the cowbird genome based on the present study are one microsatellite for every 89 kb. From this comparison, it can be estimated that the cowbird genome contains approximately 15-fold fewer microsatellites than does human DNA. Finally, results from this study can be considered from the perspective of developing locus-specific markers for linkage mapping and for population studies. In humans and other mammals, the frequency of microsatellites is sufficiently high to allow efficient recovery of these repeats from small-insert, plasmid, or M13 libraries (Weissenbach et al. 1992). However, if one desires to isolate microsatellites from avian DNA, in which the density of repeats is much lower, it may be prudent to use alternative cloning strategies or methods that increase the yield of microsatellites

alternative to enriched libraries as a means for isolating polymorphic markers from genomes that are not rich in microsatellites (such as avian genomes). Due to large insert size, one can survey a significant portion of the genome while screening a reasonable number of cosmid clones. In this study we were able to examine 3% of the cowbird genome by hybridizing only 1,536 cosmids. The microsatellite repeats that are identified within cosmids can be sequenced directly ( Baron et al. 1992) or subcloned and sequenced for development of locus-specific PCR primer pairs (Couch et al. 1994). Several of the (GATA)n repeats that we identified in this study have been subcloned, sequenced, and developed into PCR-based markers for studies of cowbird parentage. These markers will be the subject of future reports. From the Genomics Group, MS-M880, Los Alamos National Laboratory, Los Alamos, NM 87545 ( Longmire and Roach), and the Patuxent Wildlife Research Center, U.S. Geological Survey, Laurel, Maryland ( D. C. Hahn). J. L. Roach is currently at the Department of Biology, Colorado State University, Fort Collins, Colorado. Address correspondence to J. L. Longmire at the address above or e-mail: [email protected]. The authors gratefully acknowledge the assistance of several individuals that helped with various aspects of this work. Blood samples were collected by Jon Boone, James Kolozar, James Sedgwick, Michael diDomenico, John Goodell, Laura Williams, and Peter Osenton. Scott White, Mary Maltbie, and Kateryna Makova and two anonymous reviewers provided many helpful comments to the manuscript. This work was conducted under the auspices of the U.S. Department of Energy. q 1999 The American Genetic Association

References Baron B, Poirier C, Simon-Chazottes D, Barnier C, and Guenet J-L, 1992. A new strategy useful for rapid identification of microsatellites from libraries with large insert sizes. Nucleic Acids Res 20:3665–3669. Baker RJ, Longmire JL, and Van Den Bussche RA, 1995. Organization of repetitive elements in the upland cotton genome (Gossipium hirsutum). J Hered 86:178–185. Beckmann JS and Weber JL, 1992. Survey of human and rat microsatellites. Genomics 12:627–631. Couch FJ, Kiousis S, Castilla LH, Xu J, Chandrasekharappa SC, Chamoberlain JS, Collins FC, and Weber BL, 1994. Characterization of 10 new polymorphic dinucleotide repeats and generation of a high-density microsatellite-based physical map of the BRCA1 region of chromosome 17q21. Genomics 24:419–424. Figure 1. Hybridization of microsatellite oligomer probes to high-density gridded arrays of cowbird cosmid clones. High-density gridded arrays of clones from the cowbird cosmid library were hybridized to oligomer probes representing all possible mono-, di-, and trinucleotide repeats and the tetranucleotide motif (GATA)n. Results shown include the following probes: (A) (GTT )17 3 (AAC)17; (B) (AAT )17 3 (ATT )17; (C) (AGC)17 3 (GCT )17; and (D) (GATA)10 3 ( TATC)10. Each grid (measuring 8 cm 3 12 cm) contained 1,536 cosmids representing 47.6 Mbp of cowbird DNA (roughly 3.2% of the cowbird genome).

from libraries (Makova and Patton 1998). One widely used approach is to construct libraries that are enriched for microsatellites ( Edwards et al. 1996; Ostrander et al. 1992). However, enriched libraries are not representative for all sequences present

within the genome and such libraries are therefore not suitable for more broadly based studies (such as isolation of other repetitive elements and single copy genes). Cosmid libraries provide an attractive

Edwards KJ, Barker JHA, Daly A, Jones C, and Karp A, 1996. Microsatellite libraries enriched for several microsatellite sequences in plants. BioTechniques 20:758– 760. Epplen JT, 1988. On simple repeated GAT/CA sequences in animal genomes: a critical reappraisal. J Hered 79: 409–417. Epplen JT, Ammer H, Epplen C, Kammerbauer C, Mitreiter R, Roewer L, Schwaiger W, Steimle V, Zischler H, Albert E, Andreas A, Beyermann B, Meyer W, Buitkamp J, Nanda I, Schid M, Nurnberg P, Pena SDJ, Poche H, Sprecher W, Schartl M, Weising K, and Yassouridis A, 1991. Oligonucleotide fingerprinting using simple repeat motifs: a convenient, ubiquitously applicable method to detect hypervariability for multiple purposes. In: DNA fingerprinting: approaches and applications


( Burke T, Dolf G, Jeffreys A, and Wolff R, eds).Basil: Birkhauser Press; 50–69. Evans GA, Lewis K, and Rothenberg BE, 1989. High efficiency vectors for cosmid microcloning and genomic analysis. Gene 79:9–20. Hamada H, Petrino MG, and Kakunaga T, 1982. A novel repeated element with Z-DNA-forming potential is widely found in evolutionarily diverse eukaryotic genomes. Proc Natl Acad Sci USA 79:6465–6469. Hamada H, Petrino MG, Kakunaga T, Seidman M, and Stroller BD, 1984. Characterization of genomic poly(dTdG)●poly(dC-dA) sequences; structure, organization, and conformation. Mol Cell Biol 4:2610–2621.

Weber JL and May PE, 1989. Abundant class of Human DNA polymorphisms which can be typed using the polymerase chain reaction. Am J Hum Genet 44:388– 396. Weissenbach J, Gyapay G, Did C, Vignal A, Morissette J, Millasseau P, Vaysseix G, and Lathrop M, 1992. A second-generation linkage map of the human genome. Nature 359:794–801. Received August 14, 1998 Accepted March 31, 1999 Corresponding Editor: Stephen J. O’Brien

Hughes AL and Hughes MK, 1995. Small genomes for better flyers. Nature 377:391. Janecek LL, Longmire JL, Whichman HA, and Baker RJ, 1993. Organization of repetitive elements in the genome of the white-footed mouse, Peromyscus leucopus. Mamm Genome 4:374–381. Longmire JL, Brown NC, Ford AA, Naranjo CM, Ratliff RL, Hildebrand CE, Stallings RL, Costa AK, Avdalovic N, and Deaven LL, 1991. Automated construction of highdensity gridded arrays of chromosome-specific cosmid libraries. Lab Robot Automat 3:195–198. Longmire JL, Brown NC, Meincke LJ, Campbell ML, Albright KL, Fawcett JF, Campbell EW, Moyzis RK, Hildebrand CE, Evans GA, and Deaven LL, 1993. Construction and characterization of partial digest libraries made from flow sorted human chromosome 16. Genet Anal Tech Applic 10:69–76. Longmire JL and Ratliff RL, 1994. Enzymatic elongation of microsatellite oligomers for use in direct-label chemiluminescent hybridizations. BioTechniques 17:1090– 1097. Makova KD and Patton JC, 1998. Increased yield of triand tetranucleotide heterospecific microsatellites from unenriched small-insert libraries. BioTechniques 24: 38–43. Olmo E, Capriglione T, and Odierna G, 1989. Genome size evolution in vertebrates: trends and constraints. Comp Biochem Physiol 92B:447–453.

Congenital Myotonic Myopathy in the Miniature Schnauzer: An Autosomal Recessive Trait C. H. Vite, J. Melniczek, D. Patterson, and U. Giger Myotonia is a clinical sign characterized by a delay in skeletal muscle relaxation following electrical or mechanical stimulation. A series of related miniature schnauzer dogs with congenital myotonic myopathy were studied. A composite pedigree of six affected litters and the results of a planned breeding between two affected animals are consistent with an autosomal recessive mode of inheritance.

Olofsson B and Bernardi G, 1983. Organization of nucleotide sequences in the chicken genome. Eur J Biochem 130:241–245.

Table 1. Congenital disease exhibiting myotonia

Ostrander EA, Jong PM, Rine J, and Duyk G, 1992. Construction of small-insert genomic DNA libraries highly enriched for microsatellite repeat sequences. Proc Natl Acad Sci USA 89:3419–3423.

Disease

Primmer CR, Raudsepp T, Chowdhary BP, Moller AR, and Ellegren H, 1997. Low frequency of microsatellites in the avian genome. Genome Res 7:471–482. Quinn TW and White BN, 1987. Identification of restriction fragment length polymorphisms in genomic DNA from the lesser snow goose (Anser caerulescens caerulescens). Mol Biol Evol 4:126–143. Stallings RL, 1994. Distribution of trinucleotide microsatellites in different categories of mammalian genomic sequence: implications for human genetic diseases. Genomics 21:116–121. Stallings RL, Ford AF, Nelson D, Torney DC, Hildebrand CE, and Moyzis RK, 1991. Evolution and distribution of (GT )n repetitive DNA sequences in mammalian genomes. Genomics 10:807–815. Sun L, Paulson KE, Schmid CW, Kadyk L, and Leinwand L, 1984. Non-Alu family repeats in human DNA and their transcription. Nucleic Acids Res 12:2669–2690. Tautz D and Renz M, 1984. Simple sequences are ubiquitous repetitive components of eukaryotic genomes. Nucleic Acids Res 12:4127–4136. Tautz D, Trick M, and Dover GA, 1986. Cryptic simplicity in DNA is a major source of genetic variation. Nature 322:652–656. Van Den Bussche RA, Longmire JL, and Baker RJ, 1995. How bats achieve a low C-value: frequency of repetitive DNA in macrotus. Mamm Genome 6:521–525.


Human Myotonia congenita Thomsen’s myotonia Becker’s myotonia Myotonic dystrophy

Defect

Myotonia is a clinical sign characterized by a delay in skeletal muscle relaxation following the cessation of a voluntary activity or following the cessation of an electrical or mechanical stimulus ( Barchi 1994). The delay in skeletal muscle relaxation is not accompanied by pain or cramping. Various biochemical defects may result in myotonia, including reduced membrane chloride conductance, alterations in the kinetics of sodium channel inactivation, and as yet undetermined membrane abnormalities ( Barchi 1988, 1994). The known congenital diseases exhibiting myotonia are inherited as autosomal dominant or autosomal recessive traits and are summarized in Table 1. In animals, myotonia has been described in the mouse, goat, horse, cat, and dog ( Barchi 1994; Hickford et al. 1998). Except for congenital myotonia in the mouse and goat, due to defective chloride ion conductance across the muscle membrane ( Bryant 1979; Mehrke et al. 1988), few studies characterize the biochemical defect or the mode of inheritance of the disease responsible for myotonia in animals. Defective chloride ion conductance in the goat results from a substitution of proline for alanine in the carboxyl terminus of the goat muscle chloride channel

Mode of inheritance

Reference

Autosomal dominant Autosomal recessive Autosomal dominant

Harper (1995) Hudson et al. (1995) Harper (1995), Hudson et al. (1995)

Sodium channel myotonia

Chloride channel Chloride channel Undetermined membrane abnormalities Sodium channel

Autosomal dominant

Paramyotonia congenita

Sodium channel

Autosomal dominant

Hyperkalemic periodi paralysis Chondrodystrophic myotonia

Sodium channel

Autosomal dominant

Chloride channel

Autosomal recessive

Barchi (1995), Hudson et al. (1995) Barchi (1995), Harper (1995), Hudson et al. (1995) Barchi (1995), Hudson et al. (1995) Adams et al. (1997), Harper (1995), Swash and Schwartz (1981)

Chloride channel

Autosomal recessive

Mehrke et al. (1988)

Chloride channel

Autosomal dominant

Beck et al. (1996), Bryant (1979)

Horse Hyperkalemic periodic paralysis

Sodium channel

Autosomal dominant

Hoffman (1995)

Cat Chow chow Miniature schnauzer

? ? Chloride channel

? Autosomal recessive Autosomal recessive

Hickford et al. (1998) Jones et al. (1977) Vite et al. (1997)

Mouse Arrested development of righting (ADR) Goat