QTL analysis of cotton fiber quality using multiple Gossypium hirsutum x Gossypium barbadense backcross generations.

AS THE LEADING NATURAL FIBER CROP, cotton is an important agricultural commodity, providing income to millions of farmers in both industrial and developing countries. Technological changes in the textile industry mean that priorities concerning fiber properties have also changed (May and Lege, 1999), and cotton breeders have concentrated efforts on different fiber properties as important predictors of yarn performance (May, 2002). Classical breeding studies have shown that these properties tend to be moderately to highly heritable (Meredith and Bridge, 1972). Upland cotton, G. hirsutum, dominates the world's cotton fiber production, representing 90% of the production. Compared with the Upland cotton, the second most cultivated species, G. barbadense, has superior fiber length, strength, and fineness, giving a higher spinning and manufacturing performance. Gossypium hirsutum varieties, however, are usually early maturing and higher yielding. Although bidirectional genome exchanges between the two species are well documented (reviewed in Brubaker and Wendel, 2001) attempts at utilizing deliberate interspecific recombination between G. hirsutum/G, barbadense by conventional breeding have had limited impact on cultivar development (Paterson and Smith, 1999). While the use of DNA markers for marker-assisted selection has received considerable attention among plant and animal breeders in the past 10 to 15 yr, published reports on QTL detection in cotton are recent. These include the mapping of disease resistance genes (Wright et al., 1998; B. Lyon, personal communication, 2001), genes or QTLs affecting leaf pubescence (Wright et al., 1999), fertility restoration (Lan et al., 1999), leaf morphology (Jiang et al., 2000a), stomatal conductance (Ulloa et al., 2000), and traits related to water stress response (Saranga et al., 2001). Studies have reported QTLs related to fiber quality derived both from populations of G. hirsutum (Shappley et al., 1998: Ulloa and Meredith, 2000), and populations from interspecific crosses between G. hirsutum x G. barbadense (Jiang et al., 1998; Yu et al., 1998; Kohel et al., 2001; Paterson et al., 2003; Zhang et al., 2003; Mei et al., 2004). Except for Mei et al. (2004), who detected 5 QTLs for four fiber-related traits, most of these studies showed that genetic control of fiber quality was highly complex, translating into high numbers of QTLs and moderate to low elementary additive QTL effects on trait value. In addition, comparisons between experiments and/or populations have so far been limited by the fact that bridge markers between mapping populations were scarce.

This paper reports on an analysis of cotton fiber QTLs undertaken as part of a marker-assisted backcross introgression scheme (Lacape et al., 2000) based on a combined RFLP-SSR-AFLP genetic map of a G. hirsutum/ G. barbadense backcross (Lacape et al., 2003). We choose a G. barbadense variety as the donor parent for superior fiber quality and a G. hirsutum variety as recurrent parent. The first (one set of phenotypic data) and second (two sets of data) backcrosses to the G. hirsutum recurrent parent were used for QTL detection.

MATERIALS AND METHODS

Plant Material

The plant material included three populations, B[C.sub.1], B[C.sub.2], and B[C.sub.2][S.sub.1]. The initial cross involved "Guazuncho 2' (G. hirsutum) and 'VH8-4602' (G. barbadense) as the female and male parents, respectively. Guazuncho 2 is a modern pure-line G. hirsutum (Gh) variety created at INTA (Instituto Nacional de Technologia Agropecuaria) in Argentina. It was chosen for its good overall agronomic performance. VH8 originates from Antigua and was derived from a cross between two Sea Island G. barbadense (Gb) types. Compared with a G. hirsutum standard, it is characterized by superior values for fiber length (+10 mm in UHML as compared with Guazuncho 2), HVI-strength (+16 cN/tex), and standard fineness (-40 mtex). The first backcross generation involved the [F.sub.1] as the female and Guazuncho 2 as the male parent. Seventy-five B[C.sub.1] plants were grown in a greenhouse at Montpellier, France, during summer 1999 (May-October), and served as female parents for making the second backcross to Guazuncho 2. Fiber samples were taken from all B[C.sub.1]s after ginning the open-pollinated seed cotton on a laboratory roller gin. A set of 600 B[C.sub.2] plants originating from 60 different B[C.sub.1]s was grown under field conditions during summer 2000 (May October) at Montpellier. Among these, 200 individual B[C.sub.2] plants having shown a satisfactory production of B[C.sub.3] seeds and originating from 53 different B[C.sub.1]s were harvested. Open pollinated seeds of these B[C.sub.2] plants, constituting 200 B[C.sub.2][S.sub.1] progenies, were grown during the 2001 2002 (December-April) rainy season under field conditions at Primavera do Leste in Brazil. Each B[C.sub.2][S.sub.1] was planted in two fully randomized replications, each plot (one row) measuring 5 m. Each of the 400 plots was bulk harvested, the seed cotton ginned on a laboratory roller gin and the fiber sampled for analysis.

Analysis of Fiber Quality

All fiber samples were analyzed at the Cirad laboratory at Montpellier. Measurements were made with an HVI line (Zellweger Uster 900, Uster Technologies, Switzerland) including fiber length components (mean length, ML, upper half mean length. UHML, and uniformity index, UI), strength (STR), elongation (ELO), color components (reflectance, Rd. and yellowness index. +b). Maturimeter parameters (FMT3, Shirley Dev Ltd, England) included fiber maturity ratio (MR), micronaire reading (IM), linear fineness (H), and standard fineness (Hs).

SSR/AFLP Analysis of the B[C.sub.2] Population

The B[C.sub.1] population was used to construct a genetic map as described in Lacape et al. (2003). With the new microsatellite markers subsequently developed by Nguyen et al. (2004), the updated version of this map consists of 1160 loci mapped along 5520 cM. The 200 individual B[C.sub.2] plants were analyzed for a selected subset of SSR (81 primer pairs) and AFLP (45 EcoRI/ Msel primer pairs) loci chosen on the basis of their representative distribution across the 26 chromosomes. SSR and AFLP protocols are described in Lacape et al. (2003). The allelic constitution of the 200 B[C.sub.2] plants, either homozygous for G. hirsutum (Gh) alleles, Gh/Gh, or heterozygous for G. hirsutum and G. barhadense (Gh), Gh/Gb, was finally scored for 594 segregating loci.

Statistical Analysis

Heritability and Trait Correlation

Trait heritability was directly estimated from the parent/ offspring regressions of 200 B[C.sub.2]-B[C.sub.2][S.sub.1]. Pearson correlation coefficients were calculated for all trait combinations within each of the 3 phenotypic data sets.

Genetic Mapping

Additional markers scored only in the B[C.sub.2] population were positioned on the B[C.sub.1] map (Nguyen et al., 2004) after constructing a B[C.sub.2] genetic map and aligning the two maps by common loci. The Mapmaker "group," (LOD 5.0 and 30 maximal recombination frequency) "order," and "sequence" commands were used to generate the B[C.sub.2] linkage map from 594 segregating markers. Marker segregations in the B[C.sub.2] were tested for deviation from the expected 3:1 (Gh/Gh: Gh/Gb) ratio by a [chi square] test. All B[C.sub.2] linkage groups were assigned to a corresponding B[C.sub.1] group using loci in common between the 2 maps as bridges. Each additional B[C.sub.2] locus was positioned on the B[C.sub.1] map as a backbone reference map, by extrapolating its position from the positions of shared flanking loci. Hence. the same mapping data, i.e., those obtained in the B[C.sub.1] generation, were used for QTL analysis in the B[C.sub.1] and B[C.sub.2] populations.

QTL Analysis

Excluding the cosegregating markers, 595 B[C.sub.1] and 341 B[C.sub.2] markers were used to conduct three individual QTL analyses using one phenotypic data set in B[C.sub.1] and two phenotypic data sets (B[C.sub.2] and B[C.sub.2][S.sub.1]) in B[C.sub.2]. The association between phenotype and marker genotype was investigated through simple marker analysis (SMA), interval mapping (IM), and composite interval mapping (CIM) using the computer software QTL Cartographer 1.13 (Basten et al., 1999). For each variable, IM using multiple regression of phenotypic data on marker genotypic data over the whole genome was run with 1000 permutations to identify the minimum significant LOD score to be considered. For each individual test, permutation-based thresholds were considered at a 5% risk at the genome level. Composite interval mapping (Jansen and Stam, 1994: Zeng, 1994) was then performed by means of the markers preselected by stepwise regression as cofactors. The results of the QTL position, proportion of phenotypic variance explained (R2), and effect that are reported are those derived from CIM, after checking for their agreement with the SMA results. The QTL positions on the B[C.sub.1] or B[C.sub.2] map are reported on the B[C.sub.1]/B[C.sub.2] consensus map and drawn by MapChart software (Voorrips, 2002).

The effects of QTLs are considered from a product transformation perspective. Therefore, decreases in fiber fineness (H), standard fineness (Hs), micronaire reading (IM), and yellowness index, were indicative of a "positive" contribution conferred by either parent.

RESULTS

Phenotypic Variation

In the present study, the Gh and Gb accessions were characterized by their contrasting fiber properties with differences of 9.7 mm in UHML, 15.9 cN/tex in fiber strength, and 38 mtex in standard fineness, as averaged over three measurements (Fig. 1). The frequency distribution of each parameter in B[C.sub.1], B[C.sub.2], and B[C.sub.2][S.sub.1] populations showed typical quantitative variation (Fig. 1). All variables fitted a normal distribution and none were transformed in further analyses.

Combined AFLP/SSR B[C.sub.2] Map and B[C.sub.1]/B[C.sub.2] Consensus Map

Taken as a whole, and in the absence of any deliberate human selection pressure on the B[C.sub.1]s, the genetic transmission observed in the 200 B[C.sub.2] individuals met the expected 1:3 frequency, 25.4 to 74.6% of Gh/Gb and Gh/Gh allelic combinations, respectively. The segregation of 594 B[C.sub.2] markers revealed 511 loci mapped in 26 linkage groups (not shown). Each B[C.sub.2] group was bridged unambiguously to a corresponding chromosome or linkage group on the B[C.sub.1] map by 373 common markers. The number of bridge AFLP or SSR markers varied between seven for c16 and 26 for A03. Although some bias in the construction of the B[C.sub.2] map is expected from the fact that the B[C.sub.2] individuals were not independent and that B[C.sub.2] data were treated as a B[C.sub.1] in Mapmaker, the maps constructed from the two populations were in full agreement concerning locus orders. Distances between common loci were frequently reduced in the B[C.sub.2] map as compared with the B[C.sub.1] map (not shown). On the basis of the distance covered by the common markers, the B[C.sub.2] map was 15% shorter than the 3300 cM of B[C.sub.1]. A total of 138 loci (133 AFLPs and five SSRs) not scored in the B[C.sub.1], were mapped in the B[C.sub.2] population. From the extrapolated positions of these additional B[C.sub.2] loci onto the backbone B[C.sub.1] map, the B[C.sub.1]/B[C.sub.2] consensus map comprises 1306 loci and spans 5597 cM (Fig. 2). The average interval between two markers of the consensus map is 4.3 cM.

Trait Heritability and Correlation between Traits

On the basis of B[C.sub.2]/B[C.sub.2][S.sub.1] regression, the estimates of narrow sense heritability were highly significant for all individual fiber quality traits. These estimates fell within the range of [h.sup.2] values reviewed in May (1999). The highest [h.sup.2]s were for fiber length (UHML: 0.70: ML: 0.58) and fineness (H: 0.65; Hs: 0.57). The lowest [h.sup.2] estimates were for length uniformity (0.24) and color indices (0.29 for reflectance and yellowness index). In the case of length uniformity, some bias in the [h.sup.2] estimate might result from the fact that UI is calculated as a ratio (UHML/ML), while the lower [h.sup.2] for color indices clearly reflected the greater influence of nongenetic sources of variation as compared with other parameters. The moderate [h.sup.2] (0.40) value for fiber strength in our study was possibly related to the greater variability obtained from the HVI fiber strength measurement (May, 2002). compared with classical types of measurements.

Correlation coefficients between phenotypic traits calculated in each of the B[C.sub.1], B[C.sub.2] and B[C.sub.2][S.sub.1] populations are presented in Table 1. The strongest correlations consistently observed within the three data sets related ML to UHML (0.97 on average), IM to H (0.89) and to MR (0.83), and Rd to +b (-0.74), and (only in the B[C.sub.2][S.sub.1] data set) H to Hs (0.76). These were all expected as inherent to the measurement apparatus and/or physiological definitions of the trait concerned. Another group of correlations of lower magnitude relate length (ML or UHML), strength, and/or fineness components (IM or H), probably reflecting some genetic association of these characteristics as observed in the contrasting parents used (Fig. 1).

For simplicity, we will hereafter consider certain traits as groups of related or correlated traits: length components (LEN), grouping the two variables ME and UHML, maturity/fineness components (FIN), grouping the 4 variables IM, MR, H, and Hs, and color components (COL), grouping the two variables reflectance, Rd, and yellowness index, +b. Fiber strength (further noted STR), elongation (ELO), and length uniformity index (UNI) are considered individually,

QTL Analysis

The highest critical LOD thresholds after permutation tests varied between 3.2 and 4.0 for the various traits except for uniformity index in the B[C.sub.2], and B[C.sub.2][S.subb.1] generations (4.3 and 5.7, respectively) and for color indices in B[C.sub.2] generation (5.0 and 11.1 for Rd and +b respectively). Fifty QTLs met permutation-based LOD thresholds. The results are nevertheless reported for all the QTLs that showed a LOD superior to 2.5 in any of the generations, i.e., a total of 80 QTLs (Table 2). While QTLs detected above this LOD limit of 2.5 but below the permutation-based LOD threshold should be treated cautiously, i.e., putative QTLs, they enabled a more precise comparison of results between generations, between traits for a given chromosomal region, and between our data and other data sets for a given trait. For all but one (QTL for IM on c10) of the 50 QTLs meeting permutation-based threshold detected under CIM, and for 70 of the 80 putative QTL, we also obtained a significant marker-trait association (P > 0.001 of F statistics) using simple marker analysis.

Comparison of QTLs detected in the three populations indicated that, among the 50 significant QTLs, 11 were detected in both B[C.sub.2] and B[C.sub.2][S.sub.1], and six were detected in both B[C.sub.1] and either of B[C.sub.2] or B[C.sub.2][S.sub.1].

Percentages of phenotypic effects (Table 2) ranged between 4.8 and 14.8% for length, 3.1 and 17.9% for length uniformity, 4.4 and 21.3% for strength, 5.4 and 21.2% for elongation, 4.6 and 29.1% for fineness/maturity, and 4.8 and 15.6% for fiber color (yellowness index in B[C.sub.2] discarded from comparisons).

As expected, the respective contributions of the Gh and Gb parents to fiber quality (Table 2) indicated that a majority of Gb parent alleles had a positive influence on fiber length (12 versus 3), strength (8 versus 4), and fineness QTLs (13 versus 8), and a majority of Gh parent alleles had a positive influence on fiber color QTLs (13 versus 3).

Altogether, the 80 QTLs mapped over 22 of the 26 chromosomes (a single QTL on two unassigned linkage groups, NL5 and NL8), with chromosomes 1, 2, 7, and 12 lacking any QTLs (Fig. 2). The QTLs that were accounted for by a positive effect of the Gb alleles on either fiber length, strength, or fineness, mapped to 15 different chromosomes. Cases of colocalization of QTLs exceeded cases of isolated positioning. Such cases of colocalization of QTLs for different fiber properties were often detected in different population data sets. Cases of across-population and across-trait colocalization are illustrated for c17 with a colocalization of a fiber elongation QTL detected in the B[C.sub.1] population with a color QTL detected in the B[C.sub.2], on A02 (fineness QTL in the B[C.sub.1] and color QTL in the B[C.sub.2][S.sub.1]), on D03 (color QTL in the B[C.sub.1] and fineness QTL in the B[C.sub.2][S.sub.1]), and on A03 (uniformity QTL in the B[C.sub.1] and a strength QTL in the B[C.sub.2]).

Significant QTLs were identified for all six groups of fiber traits and in all three populations. After merging data from the three QTL analyses, the total of QTLs varied from six for length uniformity to 21 for the fineness-maturity complex.

For fiber length, nine QTLs met permutation-based thresholds, and six additional QTLs may be considered as putative. For 12 of these 15 QTLs, the positive allele contributing to increased fiber length derived from the Gb parent. The two length QTLs detected on c26 have peak LODs separated by 60 cM. The position and direction of three length QTLs on c5 (highest LOD in B[C.sub.1]), c26, and A01 (highest LOD in B[C.sub.2][S.sub.1]) agreed with QTLs reported in the literature (Table 2).

Three of the six QTLs reported for fiber length uniformity met permutation-based thresholds. Gb and Gh alleles contributed equally at three QTLs each. Two QTLs, on c14 and A03, were detected at positions that have been previously reported.

Six fiber strength QTLs met permutation-based thresholds, and six others were detected at lower LOD values. Eight QTLs resulted from a positive contribution of the Gb parent, five of which (on c3, c25, c23, A01, and A03) mapped at comparable positions in the literature. In addition, we detected two QTLs of a Gh positive contribution on c18, mapping at similar positions to two QTLs described by Paterson et al. (2003), near loci pAR788 and P5-11, but with opposite phenotypic effects.

From a total of 10 elongation QTLs, eight met permutation-based thresholds. Six resulted from a positive contribution of the Gb parent. Two QTLs on c20 are positioned 40 cM apart. The fiber elongation QTLs on c15, c9, c23, and c10 agreed with QTLs from the literature.

Fourteen of the 21 reported QTLs relating to the maturity or fineness complex met permutation-based thresholds of either IM, MR, H, or Hs parameter, and, for most of them, QTLs were detected simultaneously for several traits: (i) two traits: IM and MR on c22 and c6, IM and Hs on c9, Hs and MR on c5 and c18, H and Hs on c6 and c16; (ii) three traits: IM, MR and H on c3 and A02, IM, H, and Hs on D08; and (iii) four traits: IM, MR, H, and Hs on D03.

Thirteen cases were found in which the contribution of the Gb parent to each trait was positive (lower IM, H, and Hs values, and higher MR value). Interestingly, the map locations and effects for seven of the QTLs described in this way, including the strongest ones detected in B[C.sub.2] (c3) and B[C.sub.2][S.sub.1] (D03), corresponded to those for QTLs relating to IM values reported in the literature by Kohel et al. (2001) and Paterson et al. (2003). One QTL detected on D08 mapped at a similar position to a QTL reported in Paterson et al. (2003), near pGH225 (former c20 in the reference paper renamed to D08), but had the opposite effect.

In total, 16 QTLs were detected for the two color indexes, Rd and +b. Half of these QTLs explained variation observed for both indexes. In accordance with parental contributions (Fig. 1), 13 of the 16 detected QTLs resulted from a positive contribution of the Gh parent, conferring higher Rd and/or lower +b values. Five QTLs were reported in Paterson et al. (2003) for yellowness index, with similar map location and effect.

We observed a few cases of QTL-rich regions delimited along homeologous regions of the A- and D-subgenome chromosomes: fiber color QTLs on c6 and c25, both congruent with Paterson et al. (2003); fineness-maturity QTLs on the upper part of c6 and c25; fineness and elongation QTLs at central parts of cl0 and c20; and lastly on the c5-D08 homeologous pair a concentration of "negative" QTLs for length, elongation, fineness, and color.

DISCUSSION

In an attempt to overcome the limitations of conventional breeding for the improvement of cotton fiber quality through interspecific hybridization, we decided to use molecular markers in a marker-assisted selection scheme aimed at improving the efficiency of the introgression of fiber quality traits.

The quality of cotton fiber derives from several traits, the most important being length, fineness, and strength, each of which is the result of complex genetic architecture. This study confirmed the expected genetic complexity of individual cotton fiber components (Paterson et al., 2003; Shappley et al., 1998). Combining all the different traits, we detected a total of 80 QTLs using LOD2.5 as a threshold, of which 50 surpassed the permutation-based LOD thresholds (3.2-5.7). For most of the QTLs related to the most economically important traits, the favorable alleles came from the G. barbadense parent. However, the inferior parent, G. hirsutum, also contributed to fiber quality (Table 2): for length on c5 (length QTL of the highest LOD in B[C.sub.1]), strength on c18, and fineness/maturity on D08, c6, A02, and D03. Such observation supports the proposition that superior allele combinations may be obtained in interspecific progenies of mosaic genome constitution (Tanksley et al., 1996).

The distribution of QTLs between chromosomes or linkage groups assigned to A and D subgenomes shows a slight over-representation of QTLs mapping to the D-subgenome chromosomes (58%) as compared with the A-subgenome. This finding is in agreement with the idea that in AADD tetraploid cottons, the D-subgenome globally contributes to a higher level of trait variability than does the A-subgenome (reviewed by Paterson et al., 2003).

The cases of colocalization of QTLs for different traits (Fig. 2) confirmed the observed phenotypic correlations (Table 1); the presence of alleles of the Gb parent in some QTL-rich chromosome regions contributed simultaneously and positively to several properties among fiber strength, length, and fineness. Of significant interest are the localizations of "positive" QTLs related to traits of greater economic importance, i.e., fiber strength. length, and fineness. Chromosomes 3 and 23 are worth mentioning in this respect: c3 with the strongest length QTLs detected in B[C.sub.1] (LOD 4.43) and in B[C.sub.2][S.sub.1] (LOD 4.99), colocalized with the second strongest strength QTL detected in B[C.sub.1] (LOD 4.51), and -c23 with the strongest strength QTL detected in B[C.sub.2] (LOD 4.78) and B[C.sub.2][S.sub.1] (LOD 4.01) colocalized with a length QTL in B[C.sub.2] (LOD 2.80). Because fiber length, strength, and fineness may be considered as physiologically independent and relating to different spatial and temporal biological processes (Wilkins and Jernstedt, 1998), the colocalization of QTLs may be more indicative of linkage between different genes than of pleiotropy.

We observed a low consistency between the QTLs detected in the three populations (around 10% of QTLs in common between B[C.sub.2] and B[C.sub.2] and 20% between B[C.sub.2] and B[C.sub.2][S.sub.1]). Such apparent instability in QTLs detected under diverse environments is not uncommon (Ribaut et al., 1997: Saranga et al., 2001). In the present case, the factors that could explain such a difference between generations may involve the environmental conditions (greenhouse or field, Montpellier or Brazil), the genetic backgrounds of the B[C.sub.1] and B[C.sub.2] generation, or possible differences in the distribution of QTLs between the 75 plants constituting the B[C.sub.1] population and the subset of 53 B[C.sub.1] plants that effectively served as parents to the 200 B[C.sub.2]s. However, the colocalized QTLs also relate to various combinations of traits and data sets. Chromosomes 6 and 25, for example, contain a high density of QTLs corresponding to various trait or generation combinations: color in B[C.sub.2] and fineness in B[C.sub.1] on the upper part of c6; color in B[C.sub.1], fineness in B[C.sub.1] and/B[C.sub.2][S.sub.1], and length in B[C.sub.2] in the central part of c6; length in B[C.sub.2][S.sub.1] and fineness in B[C.sub.2] on the upper part of c25. Also noteworthy is the fact that the map position and sign of phenotypic effect for 26 (33%) of the 80 QTLs reported in this study confirm data reported from four other independent investigations reporting cotton fiber quality QTLs detected in interspecific G. hirsutum x G. barbadense populations (Jiang et al., 1998; Kohel et al., 2001: Paterson et al., 2003; Mei et al., 2004). To evaluate the consistency of QTLs across different populations, the genetic structure (backcross versus F2-derived populations) of these populations, their relationships (direct genetic relatedness in the case of our two populations), the range of molecular polymorphism (and therefore different genome coverage), as well as QTL x environment interactions must be taken into consideration (Melchinger et al., 1998; Paterson et al., 1991). Our observation that QTLs for cotton fiber quality show some stability across populations, however, provides the first insight on comparative QTL mapping in cotton. Knowledge of QTL-rich chromosome regions that are congruent between mapping population, generations, and locations is highly valuable from a breeding perspective. Such consistency in the detection of QTLs adds confidence in their reality and these QTL-rich chromosomes regions may serve as primary targets in future studies on cotton fiber quality. In particular, attention may be paid to the above-mentioned chromosomes 3 and 23: both are rich in strong QTLs in our study, Paterson et al. (2003) also found a fiber strength QTL in the lower part of c23, and chromosome 3 contained one of the four strength QTLs reported by Kohel et al. (2001) and a fineness QTL common between Kohel et al. (2001) and Paterson et al. (2003).

Among the 80 QTLs detected, the subset of "positive" fiber quality QTLs (i.e., of a G. barbadense positive allelic contribution) is distributed across 19 different chromosome segments, each bearing one of several QTLs, and is located on 15 different carrier chromosomes (Table 3). The 19 regions thus delimit a total length of 636 cM (20% of the carrier genome), or 11.5% of the total genome (Table 3, Fig. 3). In the process of our marker-assisted backcross strategy, these carrier chromosomes and chromosomal regions are now considered as G. barbadense target regions (foreground genome) to be manipulated through successive backcrosses for an introgression into a G. hirsutum background genome. Given such a high number of chromosome regions, we are now concentrating efforts on BC progenies containing subsets of introgressed chromosome segments (generally two to four), with the view toward further genetic fixation and intercrossing of advanced BC progenies for QTL pyramiding. The ongoing development of near isogenic lines, NILs, differing only by the introgression of G. barbadense alleles at a given QTL (QTL-NILs) will prove useful for studying the effect of a single given QTL on the phenotypic value of a plant harboring it. Such material will also be useful for expression studies and for map-based cloning.

Table 1. Intrapopulation correlations coefficients among fiber traits.
Significant (P < 0.01 correlations (r > 0.30 in the B[C.sub.1], and r
> 0.18 in the B[C.sub.2] and B[C.sub.2][S.sub.1]) are shown.

Trait
(†) Population ML UHML UI STR

UHML B[C.sub.1] 0.98
 B[C.sub.2] 0.96
 B[C.sub.2][S.sub.1] 0.98
UI B[C.sub.1] 0.61 0.42
 B[C.sub.2] 0.66 0.43
 B[C.sub.2][S.sub.1] 0.49 0.31
STR B[C.sub.1] -- -- 0.34
 B[C.sub.2] 0.80 0.75 0.59
 B[C.sub.2][S.sub.1] 0.56 0.51 0.44
ELO B[C.sub.1] -- -- 0.40 --
 B[C.sub.2] 0.63 0.48 0.74 0.58
 B[C.sub.2][S.sub.1] -- -- 0.31 --
IM B[C.sub.1] 0.31 -- 0.48 --
 B[C.sub.2] -0.33 -0.44 -- 0.48
 B[C.sub.2][S.sub.1] -0.53 -0.57 -- -0.27
MR B[C.sub.1] 0.46 0.39 0.50 --
 B[C.sub.2] -- -0.28 0.21 -0.39
 B[C.sub.2][S.sub.1] -0.31 -0.33 -- --
H B[C.sub.1] -- -- 0.33 --
 B[C.sub.2] -0.43 -0.52 -- -0.51
 B[C.sub.2][S.sub.1] -0.50 -0.54 -- -0.32
Hs B[C.sub.1] -0.58 -0.56 -0.40 --
 B[C.sub.2] -0.26 -0.24 -0.20 --
 B[C.sub.2][S.sub.1] -0.25 -0.29 -- -0.26
Rd B[C.sub.1] -- -- -- --
 B[C.sub.2] 0.23 -- 0.36 --
 B[C.sub.2][S.sub.1] -- -- -- --
+b B[C.sub.1] -- -- -- --
 B[C.sub.2] -- -- -- --
 B[C.sub.2][S.sub.1] -- -- -- --

 Trait
(†) Population ELO IM MR

UHML B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
UI B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
STR B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
ELO B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
IM B[C.sub.1] 0.52
 B[C.sub.2] --
 B[C.sub.2][S.sub.1] --
MR B[C.sub.1] 0.39 0.94
 B[C.sub.2] -- 0.88
 B[C.sub.2][S.sub.1] -- 0.67
H B[C.sub.1] 0.56 0.89 0.67
 B[C.sub.2] -- 0.91 0.62
 B[C.sub.2][S.sub.1] 0.21 0.88 0.24
Hs B[C.sub.1] -- -0.48 -0.75
 B[C.sub.2] -- -- -0.46
 B[C.sub.2][S.sub.1] 0.18 0.37 -0.44
Rd B[C.sub.1] -- -- --
 B[C.sub.2] -- -- --
 B[C.sub.2][S.sub.1] -0.19 -- --
+b B[C.sub.1] -- -- --
 B[C.sub.2] -- -0.27 -0.27
 B[C.sub.2][S.sub.1] -- -- --

 Trait
(†) Population H Hs Rd

UHML B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
UI B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
STR B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
ELO B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
IM B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
MR B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
H B[C.sub.1]
 B[C.sub.2]
 B[C.sub.2][S.sub.1]
Hs B[C.sub.1] --
 B[C.sub.2] 0.40
 B[C.sub.2][S.sub.1] 0.76
Rd B[C.sub.1] -- --
 B[C.sub.2] -- -0.34
 B[C.sub.2][S.sub.1] -0.20 --
+b B[C.sub.1] -- -- -0.84
 B[C.sub.2] -- -- -0.77
 B[C.sub.2][S.sub.1] -- -- -0.60

(†) Traits are: UHML, upper half mean length; ML, mean length;
UI, uniformity index; STR, strength; ELO, elongation; IM, micronaire
reading; MR, maturity ratio; H, linear fineness; [H.sub.s], standard
fineness; Rd, reflectance; +b, yellowness index.

Table 2. Synthesis of the QTLs influencing properties related
to fiber quality, as detected by composite interval mapping of
B[C.sub.1], B[C.sub.2] or B[C.sub.2][S.sub.1], populations.
LOD and [R.sup.2] that are underlined are values meeting
permutation-based thresholds.

 LOD max in population

 Nearest cM from top in
QTL-chrom locus consensus map B[C.sub.1]

Length

LEN-c14 E7M1_307 44
LEN-1-c3 E8M1_400 55
LEN-2-c3 CIR084b 129 4.43 (‡)
LEN-c5 A1135b 31 4.49 (‡)
LEN-c6 E3M4_410 141
LEN-1-c25 E4M3_268 0
LEN-2-c25 E3M5_162 51
LEN-c23 BNL1414a 125
LEN-c10 ElM3_370 113
LEN-c20 E4M3_108 256
LEN-1-c26 E4M4_230 76
LEN-2-c26 E8M2_173 138
LEN-A01 E5M8_400 41
LEN-c18 BNL193 142
LEN-D02 BNL3171 193

Uniformity Index

UNI-c14 E7M4_360 85
UNI-DO8 BNL2448a 111
UNI-c20 EIM6_149 117
UNI-c18 E3M8_400 62
UNI-AO3 CIR416 217 3.96 (‡)
UNI-NL5 E7M7_184 43 5.36 (‡)

Strength

STR-c3 BNL3989 90 4.51 (‡)
STR-1-c5 BNL852b 101 5.61 (‡)
STR-2-c5 E6M6_430 334
STR-c25 E3M5_162 51
STR-c16 E4M6_312 65
STR-1-c23 E8M8_350 57
STR-2-c23 BNL1414a 125 3.61 (‡)
STR-AO1 E4M6_232 191
STR-1-c18 ElM5_410 53
STR-2-c18 E3M4_350 108
STR-A03 BNL3442 217
STR-D02 BNL3171 193

Elongation

ELO-c15 BNL3090b 103 3.10 (‡)
ELO-c17 pAR172b 32 4.91 (‡)
ELO-c22 G1058a 24 3.94 (‡)
ELO-DO8 E8M8_124 189
ELO-c9 BNL1414b 208
ELO-c23 BNL1161a 30
ELO-c10 EIM3_370 113
ELO-1-c20 BNL2570 97
ELO-2-c20 BNL3948 132
ELO-NL8 A1826 0 5.42 (‡)

Micronaire (IM), maturity ratio (MR) and fineness (H/[H.sub.s])

FIN-1-c3 E5M7_218 30
FIN-2-c3 CIR084b 129 2.80 Hs
FIN-c4 E3M4_210 106
FIN-c22 BNL4030a 112
FIN-1-c5 CIR102 29 3.91 (‡) Hs
FIN-2-c5 E6M8_237 271
FIN-1-D08 BNL3452 32
 2.73 (‡) Hs
FIN-2-D08 CIR398u 107 3.62 (‡) IM
FIN-3-D08 E8M8_124 189
FIN-1-c6 CIR298 0 5.87 (‡) IM
 4.18 MR
FIN-2-c6 E6M5_145 154
 4.44 H
FINc-c6 CIR179 278 5.31 (‡) H
FIN-c25 E4M3_268 0
FIN-c16 E7M1_167 81
FIN-c9 E5M5_158 295
FIN-1-c10 A1158a 3 4.31 IM
FIN-2-c10 BNL3895 78
FIN-c20 BNL946 143
FIN-c18 pAR788a 33 4.68 (‡) MR
 3.47 (‡) Hs
FIN-A02 E1M7_219 183 4.26 (‡) MR
 41.08 (‡) IM
FIN-D03 BNL3860 0

Color (reflectance, Rd, and yellowness, +b)

COL-c3 BNL3408b 0
COL-c17 E3M5_134 37
COL-1-D08 BNL852a 84
COL-2-D08 E8M8_124 189
COL-1-c6 BNL3359b 11
COL-2-c6 E2M2_158 126 4.84 (‡) +b
 2.99 (‡) Rd
COL-c25 E2M7_89 73 4.37 (‡) Rd
 4.57 (‡) +b
COL-c16 BNL2986 57
COL-c9 E5M5_158 295 2.95 (‡) Rd
 3.61 (‡) +b
COL-c23 E8M8_350 57
COL-c10 E8M8_490 90
COL-c26 E4M4_230 76
COL-A01 E7M1_186 162
COL-c18 BNL193 142
COL-A02 E6M6_550 187
COL-D03 BNL3860 0 4.19 (‡) Rd
 3.04 (‡) +b

 LOD max in population

QTL-chrom B[C.sub.2] B[C.sub.2][S.sub.1]

Length

LEN-c14 3.22 (‡)
LEN-1-c3 4.99 (‡)
LEN-2-c3
LEN-c5
LEN-c6 2.89 (‡)
LEN-1-c25 3.12 (‡)
LEN-2-c25 3.38 (‡)
LEN-c23 2.80 (‡)
LEN-c10 2.72 (‡)
LEN-c20 2.77 (‡)
LEN-1-c26 4.00 (‡) 2.86 (‡)
LEN-2-c26 3.58 (‡)
LEN-A01 4.86 (‡) 4.19 (‡)
LEN-c18 4.28 (‡)
LEN-D02 3.02 (‡)

Uniformity Index

UNI-c14 2.52 (‡)
UNI-DO8 2.79
UNI-c20 3.73 (‡) 3.97 (‡)
UNI-c18 5.68 (‡) 7.75 (‡)
UNI-AO3
UNI-NL5

Strength

STR-c3
STR-1-c5
STR-2-c5 2.53
STR-c25 2.56 (‡)
STR-c16 3.40 (‡)
STR-1-c23 2.63 (‡)
STR-2-c23 4.78 (‡) 4.01 (‡)
STR-AO1 2.86 (‡)
STR-1-c18 3.43 (‡)
STR-2-c18 3.20 (‡)
STR-A03 3.19 (‡)
STR-D02 3.75 (‡)

Elongation

ELO-c15
ELO-c17
ELO-c22
ELO-DO8 3.63 (‡) 9.72 (‡)
ELO-c9 2.81 (‡)
ELO-c23 4.83 (‡)
ELO-c10 4.75 (‡)
ELO-1-c20 4.31 (‡)
ELO-2-c20 5.69 (‡)
ELO-NL8

Micronaire (IM), maturity ratio (MR) and fineness (H/[H.sub.s])

FIN-1-c3 5.17 (‡) MR
 3.76 (‡) IM 3.95 (‡) IM
 3.52 (‡) H
FIN-2-c3
FIN-c4 3.39 (‡) Hs
FIN-c22 3.24 (‡) IM
 3.27 MR
FIN-1-c5
FIN-2-c5 5.51 (‡) Hs
FIN-1-D08 3.99 (‡) H
 2.67 (‡) IM
FIN-2-D08
FIN-3-D08 3.11 (‡) HS
FIN-1-c6
FIN-2-c6 4.46 (‡) Hs
 2.68 (‡) H
FINc-c6
FIN-c25 2.96 (‡) Hs
FIN-c16 3.64 (‡) H
 2.86 (‡) Hs
FIN-c9 2.99 Hs
FIN-1-c10
FIN-2-c10 2.52 (‡) Hs
FIN-c20 3.42 (‡) H
FIN-c18 3.19 (‡) MR
FIN-A02
FIN-D03 7.72 (‡) H
 6.06 (‡) IM
 4.92 (‡) Hs

Color (reflectance, Rd, and yellowness, +b)

COL-c3 2.84 (‡) +b
COL-c17 5.44 (‡) Rd
COL-1-D08 4.63 Rd
COL-2-D08 4.13 (‡) Rd
COL-1-c6 6.99 +b
COL-2-c6
COL-c25 4.63 (‡) Rd
 3.49 (‡) +b 7.02 (‡) +b
COL-c16 2.57 Rd
COL-c9 4.35 (‡) Rd
 6.33 +b
COL-c23 4.56 (‡) Rd
 2.91 (‡) +b
COL-c10 3.61 (‡) +b
COL-c26 4.11 (‡) +b 3.25 (‡) +b
COL-A01 2.93 (‡) Rd
COL-c18 4.91 (‡) Rd 3.37 (‡) Rd
 6.20 +b
COL-A02 5.29 (‡) Rd
 3.47 (‡) +b
COL-D03

 [R.sup.2] % var explained

QTL-chrom B[C.sub.1] B[C.sub.2] B[C.sub.2][S.sub.1]

Length

LEN-c14 6.3
LEN-1-c3 10.5
LEN-2-c3 14.8
LEN-c5 14.4
LEN-c6 5.3
LEN-1-c25 6.1
LEN-2-c25 7.6
LEN-c23 5.4
LEN-c10 5.8
LEN-c20 4.8
LEN-1-c26 13.4 7.9
LEN-2-c26 6.7
LEN-A01 12.0 10.9
LEN-c18 9.7
LEN-D02 3.8

Uniformity Index

UNI-c14 4.8
UNI-DO8 3.1
UNI-c20 8.7 10.6
UNI-c18 10.7 17.3
UNI-AO3 12.5
UNI-NL5 17.9

Strength

STR-c3 16.2
STR-1-c5 21.3
STR-2-c5 4.4
STR-c25 6.4
STR-c16 7.6
STR-1-c23 5.4
STR-2-c23 12.2 10.0 7.8
STR-AO1 5.0
STR-1-c18 6.7
STR-2-c18 11.8
STR-A03 5.6
STR-D02 8.0

Elongation

ELO-c15 10.0
ELO-c17 16.7
ELO-c22 13.0
ELO-DO8 5.8 17.2
ELO-c9 12.5
ELO-c23 8.8
ELO-c10 9.2
ELO-1-c20 8.3
ELO-2-c20 12.1
ELO-NL8 21.2

Micronaire (IM), maturity ratio (MR) and fineness (H/[H.sub.s])

FIN-1-c3 11.3
 7.0 10.5
 7.5
FIN-2-c3 8.9
FIN-c4 8.2
FIN-c22 6.9
 9.0
FIN-1-c5 12.5
FIN-2-c5 22.2
FIN-1-D08 7.9
 9.7
 6.4
FIN-2-D08 11.7
FIN-3-D08 5.9
FIN-1-c6 20.5
 14.4
FIN-2-c6 15.1
 17.4 8.8
FINc-c6 21.5
FIN-c25 7.6
FIN-c16 16.5
 9.1
FIN-c9 29.1
FIN-1-c10 14.3
FIN-2-c10 4.6
FIN-c20 5.8
FIN-c18 16.9 7.5
 11.3
FIN-A02 15.8
 13.4
FIN-D03 17.6
 16.3
 12.6

Color (reflectance, Rd, and yellowness, +b)

COL-c3 5.6
COL-c17 12.8
COL-1-D08 8.1
COL-2-D08 6.3
COL-1-c6 39.9
COL-2-c6 15.7
 9.2
COL-c25 13.7 8.5
 14.9 9.4 14.1
COL-c16 4.8
COL-c9 8.7 15.6
 10.3 38.2
COL-c23 7.9
 6.4
COL-c10 8.3
COL-c26 40.4 7.0
COL-A01 4.9
COL-c18 10.4 6.7
 38.3
COL-A02 8.2
 7.6
COL-D03 12.7
 9.1

 Genetic effect
QTL-chrom Gh parent Ref Note (†)

Length

LEN-c14 -0.70
LEN-1-c3 -0.90
LEN-2-c3 -1.50
LEN-c5 1.60 (§) 1
LEN-c6 -1.20
LEN-1-c25 -0.80
LEN-2-c25 -1.42
LEN-c23 -1.30
LEN-c10 -1.20
LEN-c20 -1.10
LEN-1-c26 -2.10
LEN-2-c26 -0.70 (§) 2
LEN-A01 -1.80 (¶) 3
LEN-c18 1.80
LEN-D02 1.70

Uniformity Index

UNI-c14 -0.50 (¶) 4
UNI-DO8 3.30
UNI-c20 -1.50
UNI-c18 2.10
UNI-AO3 -1.00 (§) 5
UNI-NL5 1.13

Strength

STR-c3 -2.20 (§) 6
STR-1-c5 -2.80
STR-2-c5 0.90
STR-c25 -2.30 (¶)(#) 7
STR-c16 -1.10
STR-1-c23 -2.20
STR-2-c23 -2.90 (¶) 8
STR-AO1 -0.76 (¶) 9
STR-1-c18 0.97
STR-2-c18 3.43
STR-A03 -1.97 (¶) 10
STR-D02 2.85

Elongation

ELO-c15 0.25 (¶) 11
ELO-c17 -0.33
ELO-c22 -0.29
ELO-DO8 0.41
ELO-c9 0.28 (‡) 12
ELO-c23 0.60 (¶) 13
ELO-c10 -0.50 (¶) 14
ELO-1-c20 -0.22
ELO-2-c20 -0.51
ELO-NL8 -0.35

Micronaire (IM), maturity ratio (MR) and fineness (H/[H.sub.s])

FIN-1-c3 0.07 (§)(¶) 15
 0.22
 8.62
FIN-2-c3 17.8
FIN-c4 20.3 (¶ 16
FIN-c22 0.41 (¶ 17
 0.08
FIN-1-c5 -21.6
FIN-2-c5 -25.6
FIN-1-D08 -8.6
 -17.7
 -0.19
FIN-2-D08 -0.37
FIN-3-D08 12.9
FIN-1-c6 0.49
 0.08
FIN-2-c6 19.7
 13.9
FINc-c6 -15.2
FIN-c25 21.3
FIN-c16 15.0 (¶ 18
 16.2
FIN-c9 -46.5 (¶ 19
FIN-1-c10 0.40
FIN-2-c10 10.2
FIN-c20 6.9 (¶ 20
FIN-c18 -0.09
 19.3
FIN-A02 -0.08
 -0.38
FIN-D03 -14.0 (¶ 21
 0.04
 -17.0

Color (reflectance, Rd, and yellowness, +b)

COL-c3 -0.29
COL-c17 3.03 (¶ 22
COL-1-D08 2.88
COL-2-D08 0.91
COL-1-c6 -2.51
COL-2-c6 -0.88 (¶ 23
 2.06
COL-c25 2.60
 -0.86 (¶ 24
COL-c16 -2.25
COL-c9 1.57
 -2.49
COL-c23 1.04
 -0.36
COL-c10 0.36
COL-c26 -2.44
COL-A01 -1.83 (¶ 25
COL-c18 3.52
 -2.82
COL-A02 0.90 (¶ 26
 -0.31
COL-D03 2.46
 -0.66

(†) QTL denomination, marker locus, and position
reference relatively (= when similar) to corresponding QTL:
1-Lf-2(t)-BNL3029 (+50 cM), 2-Lf-3(t)-C59E1 (=), 3-LGA01-pAR338 (=),
4-Chr14G1147 (+25 cM), 5-LGA03-pAR570a (=), 6-Sf-1(t)-BNL3259
(+20 cM), 7-Chr25-pGH309 (+10 cM), 8-Chr23-pAR209 (=),

9-LGA01-pAR238 (+50 cM), 10-LGA03-pAR570a (=), 11-Chr15-pAR400b
(=), 12-Chr9-accacc6 (+40 cM), 13-Chr23-P12-12 (+20 cM),
14-Chr10-pVNC163b (-30 cM), 15-Ff-3(t)-P09-53 LGA06-pGH364 (=),
16-Chr04-pAR138 (-40 cM), 17-LGD07-A1152 (=), 18-LGD01-G1158a (=),
19-Chr09-P10-62 (-80 cM), 20-LGD04-pAR430 (+30 cM), 21-LGD03-pAR248
(=), 22-Chr17-pGH861 (-40 cM), 23-Chr06-A1208b (=), 24-Chr25-pGH309
(=), 25-LGA01-G1125b (=), 26-LGA02-pGH232b (-40 cM).

(‡) Refers to QTLs also detected by simple
marker analysis (P < 0.01)

(§) Kobel et al. (2001).

(¶) Paterson et al. (2003).

(#) Jiang et al. (1998).

(‡) Mei et al. (2004)

Table 3. Synthesis of 19 targeted regions mapped on 15 different
chromosomes. The targeted intervals are defined as situated between
the two loci flanking the QTL peak LOD position at a 1 LOD confidence
interval and thus identify the segment of the G. barbadense
donor genome for possible introgression into a G. hirsutum genetic
background. All QTLs for fiber quality traits are of a "positive"
contribution from the G. barbadense allele, except for "negative"
cases indicated in parentheses.

 Chromosome Target
Chromosome length interval

 cm

C14 197 28-57
c3 153 32-67
 90-138
c4 190 102-118
c22 139 112-139
c5 360 78-101
c6 296 137-144
c25 183 44-73
c16 168 65-117
c23 173 45-66
 113-135
c10 192 0-21
 78-120
c20 268 88-161
c26 195 67-143
A01 233 16-54
 171-209
c18 158 32-46
A03 271 209-234
Total carrier chromosomes 3176 Total target regions

 Target
Chromosome size Trait

 cm

C14 29 Length
c3 35 Length, fineness
 48 Length, strength, fineness
c4 16 Fineness
c22 27 Fineness
c5 23 Strength
c6 7 Length, fineness
c25 29 Length, strength
c16 52 Strength, fineness, color
c23 21 Strength (elongation-, color-)
 22 Length, strength
c10 21 Fineness
 42 Length, fineness, color
c20 73 Elongation, fineness
c26 76 Length (color-)
A01 38 Length
 38 Strength
c18 14 Fineness
A03 25 Strength, uniformity
Total carrier chromosomes 636

REFERENCES

Basten, C., B. Weir, and Z.-B. Beng. 1999. QTL cartographer, Version 1.13. Dep of Statistics, North Carolina St Univ. Raleigh, NC.

Brubaker, C.L., and J.F. Wendel. 2001. RFLP diversity in cotton, p. 81-101. In J. N. Jenkins and S. Saha (ed.) Genetic improvement of cotton, emerging technologies. Science Publisher, Inc, Enfield, NH.

Jansen, R., and P. Stare. 1994. High resolution of quantitative traits into multiple loci via interval mapping. Genetics 136:1447-1455.

Jiang, C., R.J. Wright, K.M. El-Zik, and A.H. Paterson. 1998. Polyploid formation created unique avenues for response to selection in Gossypium (cotton). Proc. Natl. Acad. Sci. USA 95:4419-4424.

Jiang, C., R.J. Wright, T. Delmonte, and A.H. Paterson. 2000a. QTL analysis of leaf morphology in tetraploid Gossypium (conon). Theor. Appli. Genet. 100:409-418.

Jiang, C., P. Chee, X. Draye, P. Morrell, C.W. Smith, and A.H. Paterson. 2000b. Multilocus interactions restrict gene introgression in interspecific populations of polyploid Gossypium (Cotton). Evolution 54(3):798-814.

Kohel, R.J., J. Yu, Y.-H. Park, and G. Lazo. 2001. Molecular mapping and characterization of traits controlling fiber quality in cotton. Euphytica 121:163-172.

Lacape, J.-M., T.B. Nguyen, L. Grivet, and B. Hau. 2000. Use of DNA molecular markers in cotton breeding. Development of a marker-assisted selection program at CIRAD/France. W71. In Plant and Animal Genome Conference VIII, San Diego CA. 9-12 Jan. 2000. Scherago Int., New York.

Lacape, J.-M., T.B. Nguyen, S. Thibivilliers, B. Courtois, B.M. Bojinov, R.G. Cantrell, B. Burr, and B. Hau. 2003. A combined RFLPSSR-AFLP map of tetraploid cotton based on a Gossypium hirsutum x Gossypium barbadense backcross population. Genome 46:612-626.

Lan, T.-H., C.G. Cook, and A.H. Paterson. 1999. Identification of a RAPD marker linked to a male fertility restoration gene in cotton (Gossypium hirsutum L.). J. Agric. Genomics 4 (hnp://www.cabipublishing.org/gateways/jag/papers99/paper299/indexp299.html; verified 11 August 20(/4).

May, L.O. 1999. Genetic variation in fiber quality, p. 183-229. In A. S. Basra (ed.) Cotton fibers. Food Products Press, Binghampton, NY.

May, L.O. 2002. Quality improvement of upland cotton (Gossypium hirsutum L.). J. Crop Prod. 5:371-394.

May, L.O., and K.E. Lege. 1999. Development of the world conon industry, p. 65 97. In C.W. Smith and J.T. Cothren (ed.) Cotton, origin, history, technology and production. John Wiley & Sons, New York.

Mei, M., N.H. Syed, W. Gao, P.M. Thaxton, C.W. Smith, D.M. Stelly, and Z.J. Chen. 2004. Genetic mapping and QTL analysis of fiber-related traits in cotton (Gossypium). Theor. Appl. Genet. 108:280-291.

Melchinger, A.E., H.F. Utz, and C.C. Schon. 1998. Quanlitative trait locus (QTL) mapping using different testers and independent population samples in maize reveals low power of QTL detection and large bias in estimates of QTL effects. Genetics 149:383-403.

Meredith. W., and R.R. Bridge. 1972. Heterosis and gene action in cotton, Gossypium hirsutum L. Crop Sci. 12:304-310.

Nguyen, T.B., M. Giband. P. Brottier, A.-M. Risterucci, and J.-M. Lacape. 2004. Wide coverage of tetraploid cotton genome using newly developed microsatellite markers. Theor. Appl. Genet. 109:167-175.

Paterson, A.H., and R. Smith. 1999. Future horizons: Biotechnology for cotton improvement, p. 415-432. In C.W. Smith and J.T. Cothren (ed.) Cotton. Origin, history, technology, and production. John Wiley & Sons, New York.

Paterson, A.H., S. Damon, J.D. Hewitt, D. Zamir, and H.D. Rabinowitch. 1991. Mendelian factors underlying quantitative traits in tomato: Comparison across species, generations, and environments. Genetics 127:181-197.

Paterson, A.H.. Y. Saranga, M. Menz, C. Jang, and R.J. Wright. 2003. QTL analysis of genotype x environment interactions affecting cotton fiber quality. Theor. Appl. Genet. 106:384-396.

Ribaut, J.M., C. Jiang, D. Gonzalez-de-Leon, G.O. Edmeades, and D. Hoisington. 1997. Identification of quantitative trait loci under drought conditions in tropical maize. 2. Yield components and marker-assisted selection strategies. Theor. Appl. Genet. 94:887-896.

Saranga, Y., M. Menz, C-X. Jiang, R.J. Wright, D. Yakir. and A.H. Paterson. 2001. Genomic dissection of genotype x environment interactions conferring adaptation of cotton to arid conditions. Genome Res. 11:1988-1995.

Shappley, Z., J.N, Jenkins, J. Zu, and J.C.J. McCarty. 1998. Quantitative trait loci associated with agronomic and fiber traits of Upland cotton. J. Cotton Sci. 2:153-163.

Tanksley, S.D., S. Grandillo, T.M Fulton, D. Zamir, Y. Eshed, V. Petiard, J. Lopez, and T. Beck-Bunn. 1996. Advanced backcross QTL analysis in a cross between an elite processing line of tomato and its wild relative L. pimpinellifolium. Theor. Appl. Genet. 92:213-224.

Ulloa, M., and W. Meredith. 2000. Genetic linkage map and QTL analysis of agronomic and fiber quality, traits in intraspecific population. J. Cotton Sci. 4:161-170.

Ulloa, M., R.G. Cantrell, R.G. Percy, E. Zeiger, and Z. Lu. 2000. QTL analysis of stomatal conductance and relationship to lint yield in an interspecific cotton. J. Cotton Sci. 4:10-18.

Voorrips, R. 2002. MapChart: Software for the graphical presentation of linkage maps and QTLs. J. Hered.:77-78.

Wilkins, T.A., and J. Jernstedt. 1998. Molecular genetics of developing cotton fibers, p. 231-269. In A.S. Basra (ed.) Cotton fibers. Developmental biology, quality improvement, and textile processing. Food Products Press, New York.

Wright, R.J., P.M. Thaxton, K.M. El-Zik, and A.H. Paterson. 1998. D-subgenome bias of Xcm resistance genes in tetraploid gossypium (cotton) suggests that polyploid formation has created novel avenues for evolution. Genetics 149:1987-1996.

Wright, R.J., P.M. Thaxton. K.M. El-Zik, and A.H. Paterson. 1999. Molecular mapping of genes affecting pubescence of cotton. J. Hered. 90:215-219.

Yu, J., Y.-H. Park, G. Lazo, and R.J. Kohel. 1998. Molecular mapping of the cotton genome:QTL analysis of fiber quality properties, p. 485. In P. Dugger and D.A. Richter (ed.) Proc Belt. Cot. Prod. Conf., San Diego, CA. 5-9 Jan. 1998. Natl. Cotton Council, Memphis, TN.

Zeng, Z. 1994. Precision mapping of quantitative trait loci. Genetics 136:1457-1468.

Zhang, T.Z., Y. Yuan, J. Yu, W. Guo, and R.J. Kohel. 2003. Molecular tagging of a major QTL for fiber strength in Upland cotton and its marker-assisted selection. Theor. Appl. Genet. 106:262-268.

Jean-Marc Lacape, * Trung-Bieu Nguyen, Brigitte Courtois, Jean-Louis Belot, Marc Giband, Jean-Paul Gourlot, Gerard Gawryziak, Sandrine Roques, and Bernard Hau

J.-M. Lacape, T.-B. Nguyen, B. Courtois, M. Giband, J.-P. Gourlot, G. Gawryziak, S. Roques and B. Hau, CIRAD, Avenue Agropolis, 34398 Montpellier, Cedex 5, France: J.-L. Belot, CIRAD, Projet Cone Sud, SHIS/QI15, Conjunto 15. Casa 3, Lago Sul, 71635-350, Brasilia DF, Brazil. Received 26 Feb. 2004. * Corresponding author ([email protected]).