HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by Zhang, J. Articles by Wang, T. Search for Related Content PubMed Articles by Zhang, J. Articles by Wang, T. Agricola Articles by Zhang, J. Articles by Wang, T. Related Collections Crop Genetics Cotton

Linkage Analysis between Gametophytic Restorer Rf₂ Gene and Genetic Markers in Cotton

^a Dep. of Mathematical Sciences, Box 30001, New Mexico State Univ., Las Cruces, NM 88003
^b Dep. of Crop, Soil and Environmental Sciences, 115 Plant Science Building, Univ. of Arkansas, Fayetteville, AR72701

^* Corresponding author (jinzhang{at}nmsu.edu).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION APPENDIX REFERENCES

 
In heterozygous fertility restored F₁ plants of the CMS-D8-Rf₂ gametophytic restoration system of cotton (Gossypium hirsutum L.), only the pollen grains with the Rf₂ allele are functional; consequently, all F₂ plants are fertile. Our objective was to develop a statistical method to estimate the recombination fraction (r) between Rf₂ and genetic markers linked in repulsion or coupling phases in fertile F₂ populations. Alleles linked to Rf₂ are preferentially transmitted through pollen, whereas the transmission of alleles linked to rf₂ is reduced. Thus, linked genes give distorted segregation ratios depending on the linkage strength. Genes independent of Rf show normal segregation. The traditional 3:1 or other appropriate ratios can be used to test the linkage between a marker and the Rf locus in an F₂ population. Examples are given for two crosses: (D8R x T586) F₂ for repulsion linkage and (D8R x T582) F₂ for coupling phase linkage. The morphological data confirmed that Rf₂ is not linked to nine dominant genes in T586 or to five recessive genes in T582. However, a RAPD marker, UBC188₅₀₀, present in D8R and absent in nonrestoring lines, exhibited extremely skewed segregation in the D8R x T586 F₂ population with only two plants without UBC188₅₀₀ in a population of 76 plants. The recombination frequency between Rf₂ and this marker is 5.26%, which agrees with our previous estimate from a testcross, (D8R x T586) x H1330. This indicates that the proposed method is a valid alternative for mapping gametophytic Rf₂. Its advantages and limitations are discussed.

Abbreviations: CMS, cytoplasmic male sterility • ML, maximum likelihood • PCR, polymerase chain reaction • RAPD, random amplified polymorphic DNA • SD, segregation distorter

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION APPENDIX REFERENCES

 
LINKAGE TESTS require normal segregation of alleles at the target loci. Many different statistical methods can be employed to test whether two or more loci are linked (Liu, 1998). For example, the Chi-square test is a traditional method that determines if an observed segregation ratio fits an expected ratio such as 9:3:3:1 for F₂ populations and 1:1:1:1 for a testcross. The test will fail if strong gametophytic or sporophytic selection occurs. Distorted segregation of genetic markers is often encountered in progenies of interspecific or intraspecific hybrids when molecular linkage maps are constructed (Wendel and Parks, 1984; Altaf et al., 1998; Faris et al., 1998; Virk et al., 1998). The differential transmission of markers is attributed to competition among gametes or to the abortion of the gametes or zygotes. Genetic studies have revealed that the differential transmission of genes is controlled by nuclear genes (Wendel et al., 1987). For example, in Drosophila melanogaster, segregation distorter (SD), a meiotic drive system, results in transmission of the SD chromosome in vast excess over the normal SD⁺ homolog in heterozygous SD/SD⁺ males (Powers and Ganetzky, 1991). Some evidence indicates that nucleocytoplasmic interactions can also affect gametophyte competition (Faris et al., 1998). An extreme example is the gametophytic sterility found in some cytoplasmic male sterility (CMS) systems in plants. When a CMS plant is crossed with its restorer (RfRf), the F₁ heterozygote Rfrf (for one-locus model) produces one-half normal and one-half aborted pollen grains. Subsequent F₂ plants are all fertile since only the Rf pollen participates in fertilization. This type of gametophytic restoration has been found in CMS-S of maize, Zea mays L. (Laughnan and Gabay-Laughnan, 1983), CMS-boro of rice, Oryza sativa L. (Varmani, 1996), IS1112C of sorghum, Sorghum bicolor (L.) Moench (Tang et al., 1998), and CMS-D8 of cotton (Stewart, 1992; Stewart et al., 1996). In such cases, the F₂ population cannot be used to determine the genetics of restoration because there is no segregation in male fertility. The fertile F₂ population also complicates linkage tests between Rf and other genetic markers. Usually for these tests, the heterozygous restored plants with CMS cytoplasm are used as females to cross with normal cultivars without Rf gene(s).

Self-pollination is much easier than crossing to generate large segregating populations, and in working with CMS-D8, an F₂ population is more convenient for conducting linkage tests. Previously, we demonstrated that restoration of male fertility in CMS-D8 was controlled by a single dominant gene, Rf₂, from the D8 restorer (Stewart et al., 1996; Zhang and Stewart, 2001a). Despite the lack of fertility segregation in the F₂, the random transmission of other genes or DNA sequences through pollen is expected to be unaffected unless they are linked to the restorer allele (Rf₂). Therefore, it is feasible to use a fertile F₂ population to conduct linkage tests; genes independent of Rf₂ will transmit normally and show normal segregation ratios. Conversely, alleles linked to Rf₂ will be preferentially transmitted through pollen, whereas pollen transmission of rf₂–linked alleles will be reduced depending on their genetic distance. Thus, Rf₂–linked genes or markers will give distorted segregation ratios. This situation has not been taken into consideration in most statistical packages for linkage analysis, and no methods are available for estimating recombination fraction (r) between a gametophytic Rf gene and its linked genetic markers when a fertile F₂ population is used. Our objective was to develop a statistical method with which F₂ progeny can be used to estimate the linkage between genetic markers and gametophytic restorer genes.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION APPENDIX REFERENCES

 
Plant Materials and Segregating Populations
The homozygous D8 restorer selection G7-5 with CMS-D8 cytoplasm and restorer gene (Rf₂) was crossed as female to T586 and T582, respectively. G7-5 was developed by repeatedly backcrossing fertile plants of a hexaploid (D₈AD) derived lineage as female to Upland cotton. G7-5 has a normal Upland cotton type in that it has green leaves, normal leaf and branch types, glanded plant body, normal bracts, cream colored flowers and pollen, pubescent plant parts, seed fuzz, and white fiber. T586 is a multiple dominant marker line carrying nine dominant mutations, while T582 is a multiple recessive marker line carrying five recessive mutations (Table 1; Endrizzi et al., 1984). In total, these markers are located on at least nine chromosomes. In the greenhouse, the two fertile F₁s were self-pollinated to generate F₂ populations. The F₁ from G7-5 x T586 was testcrossed as female with a normal nonrestoring genotype, H1330 (Bourland, 1996), while the F₁ from G7-5 x T582 was used as female in a backcross with T582. The F₂ and testcross populations were grown in the field at the University of Arkansas Main Experiment Station at Fayetteville in 1998.

DNA Extraction and RAPD Analysis
To investigate the linkage between Rf₂ and genetic markers, polymerase chain reaction (PCR)-based RAPD (random amplified polymorphic DNA) markers were employed to screen the F₂ population of G7-5 x T586. The DNA from individual F₂ plants was extracted according to the rapid DNA isolation method described by Zhang and Stewart (2000). The PCR reaction conditions were as reported by Zhang and Stewart (2000). A specific primer, UBC188 (5'-GCTGGACATC-3'), was used since it amplifies a genomic sequence closely linked to Rf₂ (Zhang and Stewart, 2004).

Statistical Method—Repulsion Linkage
Consider a restorer line (R) with the CMS cytoplasm and a dominant restorer gene (Rf), such as the D8 restorer line G7-5. To test if Rf is linked to a recessive gene (m), the restorer (RfRfmm) is used as female in a cross with a maintainer line (B) having the nonrestorer gene allele (rf) and a dominant allele (M), for example, marker gene alleles in T586. The heterozygous F₁ produces four types of gametes: RfM, Rfm, rfM, and rfm. The two male gametes, rfM and rfm, can be ignored because the nonrestorer allele (rf) cannot be transmitted via the male gamete. Gamete frequencies are listed in Table 2, together with the F₂ progenies and their genotype frequencies. Under complete dominance, only six genotypes and two phenotypes (Rf_M_ and Rf_mm) are produced, compared with nine genotypes and four phenotypes in the normal situation (Table 3).

Homozygous dominant

Heterozygous dominant

Recessive

Normally, whether the Rf locus is linked to a marker or not, the three genotypes, RfRf, Rfrf, and rfrf, follow a normal 1/4:1/2:1/4 segregation ratio. In the case of gametophytic restoration, the genotype rfrf does not exist since the male gamete carrying rf allele is nonviable. The frequencies for RfRf and Rfrf in the locus Rf are 1/2:1/2, that is, the frequency of RfRf increases to 1/2 from the normal 1/4. The frequencies for MM, Mm, and mm in the marker locus are 1/2r:1/2:1/2(1 – r). Although the frequency of the heterozygous genotype (Mm) remains unchanged (i.e., 1/2, same as for Rfrf), the frequencies for the two homozygous genotypes depend on the recombination fraction [1/2 r for MM and 1/2(1 – r) for mm]. The frequency for Rf_M_ and Rf_mm has a linear relationship with the recombination fraction. The frequency of Rf_mm decreases linearly from 1/2 (when r = 0) to 1/4 (when r = 1/2); the frequency for Rf_M_ increases linearly from 1/2 (when r = 0) to 3/4 (when r = 1/2). That is, when Rf and m are cosegregating (r = 0), genotype MM will not be produced, while genotype mm will be 1/2. On the other hand, when the two loci are independent (r = 1/2), the segregation in locus M will be normal (3:1). In normal repulsion linkage the relationships between the frequencies for the normal four phenotypes and the recombination fractions are all nonlinear (Table 3).

To estimate the recombination fraction r, three different estimators can be simply derived from Table 3,

and

The estimators for the recombination fraction r can also be obtained by using the maximum likelihood (ML) method (see the Appendix for details). The ML estimator for r is

Because f(mm) = 1 – f(M_), the r_ML given above can be also rewritten as

or

The significance of linkage can be tested by the goodness of fit ² test against 3:1 (or 1:2:1) ratio. If a significant deviation from 3:1 is encountered, the estimated recombination fraction r_ML and its variance can be estimated.

From the Appendix, we know that the information content per observation for r is

The variance for r_ML is

Statistical Method—Coupling Linkage
Assume a restorer R line with the CMS cytoplasm and a dominant restorer gene (Rf), such as the D8 restorer line G7-5. To test the linkage between Rf and any dominant genetic marker (MM), the restorer (RfRfMM) is used as female in a cross with a tester (rfrfmm), such as marker gene alleles in T582. Tables 4 and 5 summarize the expected genotypes and their frequencies for F₂ progeny in coupling linkage phase. As in Table 2, the male gametes rfM and rfm cannot participate in fertilization.

Since P(M_) = 1/2 (2 – r) and P(mm) = 1/2 r (see Table 5), r can be estimated by solving the equations

From the Appendix, the ML estimator of r is

Since n = n_{M_} + n_mm, the other estimator for r is

The information content per observation for r is

and the variance of r_ML is

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION APPENDIX REFERENCES

 
Morphological Markers
In the triploid interspecific hybrid (AD₁ x D₈) between G. trilobum L. containing the D₈ cytoplasm and Upland cotton (AD₁), an average of 12 bivalents involving D₈–D_h combinations forms during meiosis (Phillips, 1975; Umbeck and Stewart, 1985). Therefore, when the original hexaploid 2(AD₁ x D₈) was crossed with tetraploid cotton, genetic recombination involving D8 chromosomes would mostly occur with D-subgenome chromosomes. We assume that the D8 restorer gene Rf₂ and its tightly linked or cosegregated DNA fragments should be introgressed into the D-subgenome in the D8 restorer line during several generations of repeated backcrossing with Upland cotton. T586 and T582 have a total of 14 morphological markers, among which the loci H₂, P₁, Y₁, R₂, Lc₁, N₁, and fg are located on A subgenome chromosomes, and the loci L^o, R₁, Lg, v₁, and cl₁ are on D-subgenome chromosomes. As shown in Table 6, except for the locus N₁, all loci had a 3:1 segregation ratio with a recombination fraction (r) between 0.4 and 0.64 in the F₂ population between D8R as female and T586. This indicates that these eight morphological markers segregate normally in CMS-D8 cytoplasm and are not linked to the restorer gene Rf₂. Surprisingly, more plants with fuzzy seed (n₁n₁) than expected were observed, which would indicate linkage between Rf₂ and N₁, with a recombination fraction of 0.32. However, since N₁ is located on chromosome 12 of the A subgenome (Endrizzi et al., 1984), it is unlikely that it is linked to Rf₂ unless Rf₂ was introgressed into the A subgenome from D₈ chromosomes during the development of the D8 restorer. Even though genetic recombination between nonhomologous chromosomes can occur, the possibility that Rf₂ exists in the A subgenome is very low.

With regard to the pollen color locus (P₁), only two classes were observed in the testcross: cream pollen-fertile (a parental type) and yellow pollen-fertile (a recombinant type). Pollen color could not be scored in sterile anthers because of the absence of pollen, so that yellow pollen-sterile plants were indistinguishable from cream pollen-sterile plants. It seemed that r could not be calculated. However, statistically, since fertile plants make up 1/2 of the population, and the yellow pollen-fertile plant recombinant type accounts for 1/2 r, the recombination fraction (r) can be easily computed according to the proportion of the new type in the population, r = 2 P(P₁P₁). Out of 39 fertile plants, 21 had yellow pollen, while 18 had cream pollen. The segregation agreed with a 1:1 ratio . The yellow-pollen plants accounted for 26.6% of the population (39 fertile and 40 sterile plants), giving 0.532 as the estimator for r. The results confirmed no linkage between Rf₂ and P₁.

In coupling phase, the F₂ population from D8R x T582 gave a normal 3:1 segregation ratio for all five characteristics tested (Table 8), indicating that the Rf₂ gene is not linked to any of the five loci (v₁, cl₁, cu, fg, and gl₁). The results were confirmed by the testcross that was derived from the backcross of D8R/T582 as female with T582. It produced four phenotypes and genotypes: two parental types (FgfgRf₂rf₂ and fgfgrf₂rf₂) and two recombinants (e.g., Fgfgrf₂rf₂ and fgfgRf₂rf₂). The Chi-square test showed that the two new types were equal in number to the parental types with the estimator of recombination fraction ranging from 0.468 to 0.574, indicating no linkage between Rf₂ and the morphological markers scored (Table 9).

RAPD Marker
A RAPD marker, UBC188₅₀₀, is tightly linked to the restorer gene, Rf₂, with recombination fractions ranging from 2.7 to 4.9% in three testcross populations (Zhang and Stewart, 2004). This marker segregated normally (1:1), as expected. The RAPD marker is only present in the D8 restorer and absent in nonrestoring cottons including T586. In the present study, this marker was used to screen an F₂ population of G7-5 x T586 grown in the field. All the F₂ plants were fertile, and among 76 plants that produced the PCR product, UBC188₅₀₀ was present in 74 plants and absent in two. The segregation of UBC188₅₀₀ deviated significantly from an expected 3:1 ratio , indicating linkage between the Rf₂ allele and UBC188₅₀₀. According to the statistical method for computing recombination frequency (r) proposed previously for coupling linkage, r equals twice the frequency of the number of the plants without the UBC188₅₀₀ fragment. Thus, the observed r between Rf₂ and UBC188₅₀₀ is 5.26% with a standard error of 2.56%. This value is close to that (4.9%) obtained for the testcross when (D8R x T586)F₁ was crossed as female to H1330, that is, ARK8518 (Zhang and Stewart, 2004).

Advantages and Limitations in Using an F₂ Population
One of the advantages of using an F₂ population in the linkage test is that the linkage between Rf₂ and pollen color can be evaluated by testing its segregation ratio against the expected 3:1. A distorted segregation ratio in a pollen trait could be an indication of linkage with the gametophytic fertility restorer gene. However, in the testcross situation, pollen color in sterile plants could not be scored because pollen in dominant P₁_ and recessive p₁p₁ plants was absent. No viable pollen grains were produced in sterile plants, and the sterile anthers did not accumulate the yellow carotenoid pigments. Consequently, any pollen traits could not be evaluated for linkage analysis using a traditional testcross population.

Another advantage of the F₂ system is that it requires a similar number of plants to conduct the linkage analysis, compared with a backcross population. To obtain a significant level for a linkage test, the minimum plant number (n) can be computed according to the following formula

or

where z and ² are -level critical values of normal and Chi-square distributions, respectively, at a significance level of = 0.10, 0.05, or 0.01.

Statistically, the two formulas give the same results. The minimum number of plants can be computed, given certain recombination fractions at 0.05 and 0.01 significance levels. At the significance level = 0.05, when r < 0.1, no more than 20 F₂ plants are required; when r is between 0.2 to 0.3, 35 to 72 F₂ plants will be sufficient for linkage tests. The number of plants scored in the present study was 76 to 153 for the cross involving T586, and 35 for the cross involving T582. Thus, the populations used were sufficiently large to detect linkage between Rf₂ and the markers within a genetic distance of 20 to 30 cM.

In a backcross (BC) population, the variance of r_ML (see Appendix) for the case where = r, is

Comparing the standard error, S(r), of r for F₂ and BC populations, when n = 100, if r < 0.2, S(r) for F₂ in coupling phase and BC are close to each other, ranging from 0.022 (when r = 0.05) to 0.040 (when r = 0.20) for BC, and from 0.031 (when r = 0.05) to 0.060 (when r = 0.20) for F₂. Therefore, both populations will give narrow confidence interval estimators for r. When the linkage in coupling phase is not tight, the difference in S(r) becomes obvious: for example, 0.046 for BC and 0.071 for F₂ at r = 0.30, and 0.049 and 0.080, respectively, at r = 0.40. The interval estimates for r will be slightly wider only when the genetic distance is >30 cM if the F₂ population is used. Thus, statistically, F₂ and BC populations are almost equally reliable with similar accuracy in estimating r when the Rf gene is close to a marker. However, the standard error of r for F₂ in repulsion phase will be higher, ranging from 0.099 (when r = 0.10) to 0.092 (when r = 0.40) when n is 100.

Another advantage in using an F₂ population for gametophytic Rf gene mapping is that any markers that are polymorphic between two closely related parents and segregate favorably with the R line genotype in a small sample of the F₂ mapping population, potentially could be useful markers for the Rf gene. The extreme example would be the absence of recombination in all fertile progeny (cosegregation with the R line restorer allele). This would be an indication of complete linkage between Rf and the marker.

One note of caution should be made concerning the proposed method: distorted segregation of DNA markers is commonly encountered, especially in interspecific crosses, and most of them are not associated with a fertility restorer gene such as Rf₂. To control for this possibility, marker segregation could be tested in a control F₂ population carrying the Rf gene in the absence of CMS. In such a control population, the candidate-linked markers should display a normal nondistorted Mendelian segregation. The proposed method may not be convincingly used to replace testcrosses. We suggest that for high resolution mapping using a large F₂ population, a small testcross population should be tested for normal segregation of the candidate-linked markers. In our case, even though the D8 restorer line was developed by many generations of backcrossing with Upland cotton as the recurrent parent, one could argue that its complex genetic background with substantial genetic contribution, including Rf₂ gene, from the wild and distant G. trilobum species could cause distorted segregations of other genes or DNA markers not related to preferential transmission of the Rf₂ allele. However, of 14 polymorphic RAPD markers that were not linked to Rf₂, all but one exhibited dominant inheritance and had an expected 1:1 segregation ratio in the testcross population, (D8R x T586)F₁ x ARK8518 (Zhang and Stewart, 2004).

In conclusion, whether a marker is linked to a gametophytic restorer gene (Rf) or not can be tested according to whether segregation of the marker is skewed or not in an F₂ population. In repulsion linkage, the frequency for the recessive genotype of the marker will be significantly increased if Rf is linked to the recessive allele of the genetic marker. On the contrary, the recessive genotype of the marker will be significantly reduced if Rf is linked to the dominant allele of the marker. The magnitude of the deviation from 1/4 (the expected frequency for the recessive genotype) reflects the degree of linkage. Therefore, a binomial distribution or ² test could be used to determine if the segregation of a marker fits a simple 3:1 ratio. After incorporation of theoretical considerations in the present study, estimators for recombination fraction between Rf and a marker have been proposed. Empirical data on fertility restoration to CMS-D8 by the gametophytic restorer gene Rf₂ were collected and served as an example to demonstrate the usefulness of the methods. Fourteen morphological markers controlled by nine dominant genes and five recessive genes were tested for linkage with Rf₂ and found to have no association. A RAPD marker with distorted segregation in the F₂ population was calculated to be linked to Rf₂ with a recombination frequency of 5.26%. The results were confirmed by segregation analysis from two testcrosses, and were also consistent with the report published for the D2 restorer gene, Rf₁, located on the same chromosome as Rf₂ with a genetic distance of <1 cM between them (Zhang and Stewart, 2001b).

	APPENDIX

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION APPENDIX REFERENCES

 
The Maximum Likelihood Estimator of r
Let

Consider the experiment of total plant count n = n_{M_} + n_mm, where n_{M_} and n_mm are plant counts for phenotypes M_ and mm, respectively. Assume that P(M_) and P(mm) are the same for each plant and all plants are independent. Then, the random valuable of n_mm is binomially distributed with parameters n and , and the likelihood function for is

where

The likelihood function for is equivalent to the likelihood function for r,

Thus, the log-likelihood function of r is

where

By setting the first derivative of l(r) with respect to r, to zero,

the maximum likelihood (ML) estimate for r is

Note that

thus, the r_ML obtained above is indeed the ML estimate of r. The variance of r_ML is

Similarly, in the coupling linkage situation where

the ML estimator of is _ML = n_mm/n, which is equivalent to

The variance of r_ML is

The Fisher Information for r
The information content per observation for r is the Fisher information of the parameter n_mm per observation and is obtained by the distribution of n_mm as follows. The probability mass function of n_mm, when n = 1, is

where n_mm = 0 or 1. Then the logarithm of g(n_mm) is

The Fisher information of r is given by the negative expected value of the second derivative of log [g(n_mm)] with respect to r:

Since E(n_mm) = 1/2(1 – r),

Thus,

that is, the ML estimator of r is also the efficient estimator.

Similarly, in coupling linkage situation where = P(mm) = r/2, the probability mass function of n_mm, when n = 1, is

where n_mm = 0 or 1.

The Fisher information for r can be obtained as

and the variance of r_ML is

Three-Point Linkage Test
This study considers a simplified situation where Rf is linked to a single genetic marker. However, similar deductions can be extended to include three-point linkage analysis. For example, assume that Rf is linked to genetic markers M and N in the order of RfMN (with r₁ between Rf and M, and r₂ between M and N). For the heterozygous trihybrid RfrfMmNn in coupling phase, eight types of female gametes will be produced, including RfMN and rfmn [parental types, each with frequency of 1/2(1 – r₁ – r₂ + r₁r₂), assuming no interference], RfMn and rfmN [single crossover between M and N, each with frequency of 1/2(r₂ – r₁r₂)], Rfmn and rfMN [single crossover between Rf and M, each with frequency of 1/2(r₁ – r₁r₂)], and RfmN and rfMn (double crossover, each with frequency of 1/2 r₁r₂). However, male gametes with the recessive rf allele are sterile and unable to participate in fertilization. Even though this will result in only 19 genotypes in the F₂ instead of 27 genotypes under a normal linkage situation, the recombination fraction between Rf and the two markers can be easily determined. For example, if r₁ between Rf and M is 0.05, r between Rf and N is 0.15, and r₂ between M and N is 0.10, then the three two-point linkage analyses place M between Rf and N. As such, the linear relationship between Rf and linked markers can be resolved. Therefore, the models presented here can be extended for high-resolution mapping of gametophytic Rf gene(s). Statistical packages for genome-wide mapping can be modified to take this situation into consideration. If male fertility is controlled by two gametophytic restorer genes (e.g., A3 in sorghum), or by one gametophytic Rf gene and one sporophytic Rf gene (as in cotton), an extension can be easily made based on the models presented here. However, the question as to whether or not male gamete selection related to the gametophytic Rf affects the results of genome-wide mapping remains an unanswered question.

Received for publication April 7, 2004.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION APPENDIX REFERENCES

 

• Altaf, M.K., J.F. Zhang, J.McD. Stewart, and R.G. Cantrell. 1998. Integrated molecular map based on a trispecific F2 population of cotton. p. 491–492. In Proc. Beltwide Cotton Conf., San Diego, CA. 5–9 Jan. 1998. Natl. Cotton Council Am., Memphis, TN.
• Bourland, F.M. 1996. Registration of ‘H1330’ cotton. Crop Sci. 36:813.[Free Full Text]
• Endrizzi, J.E., E.L. Turcotte, and R.J. Kohel. 1984. Qualitative genetics, cytology and cytogenetics. p. 81–129. In R.J. Kohel and C.F. Lewis (ed.) Cotton. Agron. Monogr. 24. ASA, CSSA, and SSSA, Madison, WI.
• Faris, J.D., B. Laddomada, and B.S. Gill. 1998. Molecular mapping of segregation distortion loci in Aegilops tauschii. Genetics 149:319–327.[Abstract/Free Full Text]
• Kohel, R.J., J.E. Quisenberry, and R.E. Dilbeck. 1984. Linkage analysis of the male fertility restorer gene, Rf, in cotton. Crop Sci. 24:992–994.[Abstract/Free Full Text]
• Laughnan, J.R., and S. Gabay-Laughnan. 1983. Cytoplasmic male sterility in maize. Annu. Rev. Genet. 17:27–48.[ISI][Medline]
• Liu, B.H. 1998. Statistical genomics: Linkage, mapping and QTL analysis. CRC Press, Boca Raton, FL.
• Phillips, L.L. 1975. Cotton. p. 111–133. In R.C. King (ed.) Handbook of genetics. Vol. 2. Plenum Press, New York.
• Powers, P.A., and B. Ganetzky. 1991. On the components of segregation distortion in Drosophila melanogaster. V. Molecular analysis of the SD locus. Genetics 129:133–144.[Abstract]
• Stewart, J. McD. 1992. A new cytoplasmic male sterile and restorer for cotton. p. 610. In Proc. Beltwide Cotton Conf, Nashville, TN. 6–10 Jan. 1992. Natl. Cotton Council Am., Memphis, TN.
• Stewart, J.McD., C.E. Black-Brown, and J.F. Zhang. 1996. Sporophytic and gametophytic male dysfunction conditoned by the D-8 cytoplasm of cotton. p. 86. In Agronomy abstracts. ASA, Madison, WI.
• Tang, H.V., R. Chang, and D.R. Pring. 1998. Cosegregation of single genes associated with fertility restoration and transcript processing of sorghum mitochondrial orf107 and urf209. Genetics 150:383–391.[Abstract/Free Full Text]
• Umbeck, P.F., and J.McD. Stewart. 1985. Substitution of cotton cytoplasms from wild diploid species for cotton germplasm improvement. Crop Sci. 25:1015–1019.[Abstract/Free Full Text]
• Varmani, S.S. 1996. Hybrid rice. Adv. Agron. 57:378–462.
• Virk, P.S., B.V. Ford-Lloyd, and H.J. Newbury. 1998. Mapping AFLP markers associated with subspecific differentiation of Oryza sativa (rice) and investigation of segregation distribution. Heredity 81:613–620.[ISI]
• Wendel, J.F., M.D. Edwards, and C.W. Stuber. 1987. Evidence for multilocus genetic control of preferential fertilization in maize. Heredity 58:297–301.
• Wendel, J.F., and C.R. Parks. 1984. Distorted segregation and linkage of alcohol dehydrogenase genes in Camellia japonica (Theaceae). Biochem. Genet. 22:739–748.[ISI][Medline]
• Zhang, J.F., and J.M. Stewart. 2000. Economical and rapid method for extracting cotton genomic DNA. J. Cotton Sci. 4:193–201.
• Zhang, J.F., and J.McD. Stewart. 2001a. CMS-D8 restoration in cotton conditioned by one dominant gene: Further evidence. Crop Sci. 41:283–288.[Abstract/Free Full Text]
• Zhang, J.F., and J.McD. Stewart. 2001b. Inheritance and genetic relationships of the D8 and D2–2 restorer genes for cotton cytoplasmic male sterility. Crop Sci. 41:289–294.[Abstract/Free Full Text]
• Zhang, J.F., and J.McD. Stewart. 2004. Identification of molecular markers linked to the fertility restorer genes for CMS-D8 in cotton. Crop Sci. 44:1209–1217.[Abstract/Free Full Text]

This Article Abstract Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by Zhang, J. Articles by Wang, T. Search for Related Content PubMed Articles by Zhang, J. Articles by Wang, T. Agricola Articles by Zhang, J. Articles by Wang, T. Related Collections Crop Genetics Cotton