Introduction

Cotton is grown in more than 80 countries, and contributes to the world economy as a leading natural fiber crop in the textile industry and a source of oil and protein from cottonseeds. The genus Gossypium consists of approximately 45 diploid (2n = 2x = 26) and 5 tetraploid (2n = 4x = 52) species (Percival et al. 1999), including four cultivated species, G. arboreum L., G. herbaceum L., G. barbadense L. and G. hirsutum L.. Among the four cultivated species, G. hirsutum L., commonly known as upland cotton, is the most important species and provides about 95 % of the world’s cotton fiber production.

Genetic improvement of fiber yield is the top priority goal in cotton breeding program. However, with the demands for higher quality cotton fiber to produce more competitive products and increase the manufacture efficiency, cotton breeders have also spent much effort to improve fiber quality. Due to the negative association between fiber yield and quality (Culp and Lewis 1973) and the narrow genetic base of modern cotton cultivars (Iqbal et al. 2001; Rungis et al. 2005; Lacape et al. 2007), the progress to continuously increase fiber productivity and simultaneously improve its fiber quality only using conventional breeding methods has been limited (Smith and Coyle 1997). Thus, the innovative breeding approaches have to be incorporated.

Great advances in molecular marker technologies make it possible for breeders to find a rapid and precise alternative approach to conventional breeding selection schemes (Tanksley and Hewitt 1988). Based on the detailed genetic maps, quantitative trait locus (QTL) controlling fiber quality traits could be precisely mapped on cotton genome and apply to marker-assisted selection (MAS) in breeding projects. Since the first genetic map was reported in cotton (Reinisch et al. 1994), numerous genetic maps have been developed and used to identify QTL. Up to now, 726 QTL related to fiber quality traits have been mapped on 42 molecular maps (Said et al. 2013). While most of these QTL were mapped based on interspecific populations (G. barbadense × G. hirsutum), issues with sterility, cytological abnormality, extremely late flowering, and distorted segregation limited their application, such as fine mapping the gene underlying these QTL and MAS (Lacape et al. 2010), which suggests that QTL mapping using the upland cotton intraspecific population is more practical. However, due to the narrow genetic background of modern upland cotton cultivars, QTL mapping based on intraspecific populations were of low resolution, far from being satisfactory for practical application. In addition, complicated gene expression in different development stages of cotton fiber (Lee et al. 2007; Taliercio and Boykin 2007; Hovav et al. 2008; Al-Ghazi et al. 2009; Paterson et al. 2012) and QTL meta-analysis (Lacape et al. 2010; Said et al. 2013) indicated that only a few QTL related to fiber development have been mapped. Predictably, much more new or elite QTL will be identified with new divergent mapping parents.

In the present study, we constructed a genetic map based on an upland cotton recombinant inbred line (RIL) population and used this map to detect QTL for fiber quality traits. The results were expected to be valuable for research on upland cotton genome structure and fiber development molecular mechanism and quality improvement through MAS.

Materials and methods

Mapping population and fiber quality traits evaluation

Two upland cotton cultivars, CCRI 35 and Yumian 1 were chosen to produce the segregating population. CCRI 35, a high yield and disease resistance cultivar, was widely planted in China in the last decade. Yumian 1, a high fiber quality cultivar, especially characterized with high fiber strength, was developed from multiple-cultivar intermating program (Zhang et al. 2009). The two parents were crossed in the summer of 2005 at Southwest University, Chongqing, China. F1 individuals were self-pollinated to produce F2 seeds in the winter of 2005 in Hainan, China. F2 seeds were planted at Southwest University and a total of 180 F2 individual plants were randomly selected in the summer of 2006. One hundred eighty F2-derived lines were self-pollinated for four generations to produce F2:6 seeds during 2006 and 2007. F2:6 seeds were planted by lines in single-row plot (0.8 m wide and 5 m long, for 15 plants) in the summer of 2008 at Southwest University and one individual plant in each family line was randomly selected to form a population. From 2009 to 2012, 180 RIL lines were randomly planted by single-row plot (0.7 m wide and 5 m long, for 15 plants) during the summer season at Southwest University. All the naturally-opened bolls from the RIL population and parents were hand-harvested to gin fiber. Fiber samples were evaluated for fiber quality traits, using the high volume instrument (HVI) spectrum, at the Supervision Inspection and Testing Cotton Quality Center, Anyang, China. Data were collected on fiber elongation (FE, %), fiber upper half mean length (FL, mm), fiber micronaire (FM), fiber strength (FS, cN/tex), and fiber length uniformity ratio (FU, %).

Assay of DNA markers

Total genomic DNA from fresh young leaves of the parents and the 180 lines were extracted according to the modified CTAB method (Zhang et al. 2005).

A total of 25,313 simple sequence repeat (SSR) primer pairs were employed in the present study and they were synthesized by Invitrogen Co. Ltd. (Shanghai, China). Among these SSR primers, 18,358 primer pairs were downloaded from Cotton Marker Database (http://www.cottonmarker.org/), including BNL, CIR, CM, DOW, Gh, HAU, JESPR, MGHES, MON, MUSB, MUCS/MUSS, NAU, NBRI (renamed by Tang et al. 2014), and TMB. The other 6,955 primer pairs were designed in our laboratory, including 5,000 PGML primer pairs, 1,592 SWU primer pairs, and 363 SWU06-/07-primer pairs (Tang et al. 2014).

All these primer pairs were first screened for polymorphism between the mapping parents and the primer pairs showing clear polymorphism were used to genotype the RIL population. PCR amplification and product test were performed according to the procedures by Zhang et al. (2005). Clear polymorphic DNA bands on the gels were used for scoring and genotyping. Loci detected were named with the primer name. For multiple polymorphic loci revealed by a same primer pair, an extra letter was added to the primer name, such as a/b/c, indicating the molecular size from the smallest to the largest.

Genetic map construction

JoinMap 4.0 (Van Ooijen and Voorrips 2006) was served to primarily group and order all the polymorphic loci with a LOD threshold from 4 to 8 according to shared markers from the previous maps (Guo et al. 2008; Xiao et al. 2009; Zhang et al. 2009; Yu et al. 2011; Blenda et al. 2012; Zhang et al. 2012). Linkage groups belonging to a given chromosome were then treated as separate data sets and reordered at a LOD values between 1 and 4. Map distances were calculated using Kosambi’s mapping function.

QTL mapping

Multiple QTL mapping of MapQTL 6.0 (Van Ooijen 2009) was implemented to identify QTL and estimate their effects. The LOD threshold of significant QTL was calculated by 1,000 permutation tests, with a genome-wide significance level of P < 0.05. The QTL with the LOD value between 2.5 and the LOD value evaluated by permutation test were declared as putative QTL in the present study. Additive effects were defined with respect to the alleles of CCRI 35. Therefore, the positive genetic effect of each QTL indicated that the allele of CCRI 35 increased the phenotypic value, whereas the negative effect indicated that the allele of Yumian 1 increased the phenotypic value. QTL name was started with ‘q’, followed by a trait abbreviation (FL for fiber upper half mean length, FU for fiber length uniformity, FS for fiber strength, FE for fiber elongation and FM for fiber micronaire) and the chromosome number, and then followed by the number of QTL controlling the same trait on the chromosome. The graphic representation of genetic map and QTL bars representing 1-LOD drop intervals was carried out with Map Chart 2.2 (Voorrips 2006).

Results

Primer polymorphism and marker analysis

Among the 25,313 SSR primer pairs used to screen for polymorphism between the two parents, 1,333 polymorphic markers were yielded, accounting for 5.3 % of the total primer pairs. The polymorphic markers were used to genotype the RIL population, and produced 1,382 loci. For each locus, χ2 test was performed to determine if the allele frequency was deviated from the expected Mendelian segregation ratio. Of the 1,382 loci, 518 loci showed significant segregation distortion (P < 0.05), accounting for 37.5 % of the total loci. Among the distorted loci, 452 loci were biased in favor of Yumian 1 alleles, whereas other 66 loci were biased in favor of CCRI 35 alleles.

Genetic map

Based on the linkage analysis of all the 1,382 loci, a genetic map with 1,274 loci was constructed (Fig. 1; Table 1), and the map spanned 3,076.4 cM with an average distance of 2.41 cM between two adjacent markers. At-subgenome comprised of 500 loci and spanned 1,462.6 cM with an average distance of 2.93 cM between two adjacent markers. Dt-subgenome contained 774 loci and spanned 1,613.8 cM with an average distance of 2.09 cM between two adjacent markers.

Fig. 1
figure 1figure 1figure 1

The genetic map and QTL controlling fiber quality traits from upland cotton (CCRI 35 × Yumian 1) RIL population. The marker with asterisk was distorted locus. The QTL controlling fiber quality traits and the bars representing 1-LOD likelihood intervals are beside the linkage group. QTL are shown as FL for fiber upper half mean length, FU for fiber length uniformity, FS for fiber strength, FE for fiber elongation, and FM for fiber micronaire

Full size image
Table 1 Distribution of loci and distortion loci on the genetic map from the RIL population
Full size table

On the whole, all mapped loci were well-proportioned distributed across the entire genome, but still some chromosomes had more loci or fewer loci than other chromosomes. For example, Chr. 20 was mapped with 112 loci, whereas Chr. 12 was mapped with only 12 loci. Chromosome with the longest recombination length was Chr. 16 which spanned 216.4 cM, and the shortest one was Chr.12 which spanned only 65.0 cM. There were four large gaps (adjacent marker interval >20 cM), including two, one and one distributed on Chr. 11, Chr. 13 and Chr. 21, respectively.

The phenotypic analysis of fiber quality traits

The phenotypic data of the five fiber quality traits were summarized in Table 2. The two parents were only significantly different at fiber strength. All the five fiber quality traits of the RIL population showed a wide range of variation. Skewness and kurtosis test showed that these traits were approximately in normal distribution. The variance analysis result showed that fiber quality traits possessed of significant environmental and genetic efforts except for fiber elongation (P < 0.05) in the RIL population (Table 3). Correlation analysis showed that complex significant correlation exists among fiber quality traits (Table 4). Fiber strength was significantly positive-correlated with fiber length and uniformity. Fiber micronaire was significantly positive-correlated with fiber length and strength. Fiber micronaire was significantly negative-correlated with fiber uniformity. Fiber elongation was significantly negative-correlated with fiber length and strength.

Table 2 Phenotypic data analysis of five fiber quality traits for the parents and the RIL population
Full size table
Table 3 Variance analysis of fiber quality traits in the RIL population
Full size table
Table 4 Correlation coefficients among fiber quality traits in 180 recombinant inbred lines
Full size table

QTL identified for fiber quality traits

With the multiple QTL mapping method, a total of 59 QTL were detected for five fiber quality traits based on the phenotype data from 5 environments, including 28 significant QTL and 31 putative QTL (Table 5). These QTL were mapped on 23 chromosomes (Fig. 1; Table 5), with 23 QTL distributed on At-subgenome chromosomes and 36 QTL distributed on Dt-subgenome chromosomes. Parent CCRI 35 conferred 22 favorable alleles, whereas Yumian 1 conferred 37 favorable alleles.

Table 5 QTL controlling fiber quality traits identified from the RIL population
Full size table

For fiber upper half mean length, 15 QTL (5 significant QTL and 10 putative QTL) were identified and mapped on 14 chromosomes (Fig. 1; Table 5), explaining 6.1–13.4 % of the phenotypic variation. qFL07.1 and qFL19.1 were identified in three environments, qFL08.1 and qFL17.1 were identified in two environments, and all other QTL could be detected only for one environment. Among these QTL, 5 favorable alleles were contributed by CCRI 35 and all other favorable alleles were contributed by Yumian 1.

For fiber length uniformity, 10 QTL (6 significant QTL and 4 putative QTL) were identified and mapped on 10 chromosomes (Fig. 1; Table 5), explaining 6.3–11.4 % of the phenotypic variation. Only qFU08.1 was identified in two environments and all other QTL were identified in only one environment. Among these QTL, 4 favorable alleles were contributed by CCRI 35, whereas other favorable alleles were contributed by Yumian 1.

For fiber strength, 9 QTL (5 significant QTL and 4 putative QTL), were identified and mapped on 8 chromosomes (Fig. 1; Table 5), explaining 6.1–26.5 % of the phenotypic variation. qFS07.1 and qFS14.1 were identified in five and three environments, respectively, showing that they were very stable QTL. Besides qFS05.1 identified in two environments, all other QTL could be detected only in one environment. Six favorable alleles including all the stable QTL were contributed by Yunmian1 and 3 favorable alleles were contributed by CCRI 35.

For fiber elongation, 10 QTL (5 significant and 5 putative QTL) were identified and mapped on nine chromosomes (Fig. 1; Table 5), explaining about 4.8–11.1 % of the phenotypic variation. All the 10 QTL were identified in only one environment.

For fiber micronaire, 15 QTL (7 significant QTL and 8 putative QTL) were identified and mapped on 13 chromosomes (Fig. 1; Table 5), explaining between 6.3 and 15.4 % of the phenotypic variation. qFM07.1 and qFM08.1 were identified in four and three environments, respectively, indicating that they were stable QTL. qFM22.1 was identified in two environments, and all other QTL were detected only in one environment. Among these QTL, 5 favorable alleles decreasing the trait value were conferred by CCRI 35, whereas other favorable alleles were conferred by Yumian 1.

Discussion

High-density genetic map

The genetic map constructed in the present study represents the most saturated upland cotton intraspecific genetic maps to date. It spanned about 70.0 % of the entire recombination length of tetraploid cotton genome estimated to be 4,400–4,660 cM (Lacape et al. 2003; Rong et al. 2004; Yu et al. 2011). The marker polymorphic ratio was much lower than that in the previous reports (Lin et al. 2009; Zhang et al. 2009). The advance of this genetic map was mainly attributed to the large amount of SSR primer pairs and the approximately even distribution of loci across the entire genome. The chromosomes with fewer loci and the gaps identified in our study might suggest that there was low marker diversity in these chromosomes or regions between the two parents. The recombination length of some chromosomes was much shorter than other chromosomes and the similar result was also reported in the other studies (Zhang et al. 2009; Yu et al. 2011; Zhao et al. 2012; Zhang et al. 2012). Regarding to the physical length of their homologous chromosomes in D genome (Paterson et al. 2012), it seems that lower recombination rate in these chromosomes/regions played a major role leading to the result. In addition, more loci were distributed on Dt-subgenome than At-subgenome, which was consistent with the reports in the other studies (Yu et al. 2011; Zhang et al. 2012; Zhao et al. 2012), and the possible reason was that D/Dt genome was more divergent than A/At genome among cotton species (Guo et al. 2008).

Segregation distortion

Segregation distortion is widespread in plant populations, and is regarded as the source and force of plant evolution (Taylor and Ingvarsson 2003). The percentage, degree, origin and genetic effects of segregation distortion vary significantly with species, population types, crosses and marker types in plants (Xu et al. 1997). Previous studies considered that higher segregation distortion in RIL population may mainly result from genetic drift (Zhang et al. 2009), genetic incompatibility and genome instability (Zhang et al. 2009) and the divergence between species (Paterson et al. 1988). The segregation distortion ratio in this study is high and most distorted loci skewed to Yumian 1 alleles, the same phenomenon was found in the other studies with Yumian 1 as one parent (Hu et al. 2008; Zhang et al. 2009), and the results suggested that Yumian 1 alleles probably played a critical role in segregation distortion. Considering the mapping parent Yumian 1 with complicated parentage (Zhang et al. 2009), we deduce that both the parentage of parent Yumian 1 and the population type contributed to the high segregation distortion in the present study. Similarly, most segregation distortion loci occurred in clusters, and it is consistent with the result found in the interspecific populations (Reinisch et al. 1994; Lacape et al. 2003; Guo et al. 2007; Zhang et al. 2008; Yu et al. 2011), and the intraspecific populations (Shen et al. 2005, 2007; Lin et al. 2009; Zhang et al. 2009). These results indicated that genetic hitchhiking effects commonly occurred in cotton. Furthermore, many segregation distortion regions were close to or even overlapping with the QTL intervals and this phenomenon implies that some relationship exists between the alleles underlying these QTL and the alleles causing segregation distortion.

Common or stable QTL across multiple populations and environments

Due to the most detailed intraspecific genetic map constructed in the present study, 59 QTL controlling fiber quality traits were detected. Among these QTL detected, only 11 QTL were detected in two or more environments and some QTL with large effects were detected in merely one environment. The same results were also found in the other studies (Shen et al. 2007; Zhang et al. 2009; Zhang et al. 2012), and these results further proved that environmental factor played an important role in QTL expression. However, according to the common shared markers in the QTL-regions, 16 QTL detected in the present study were identified in the other populations and these QTL included qFE07.1 (Sun et al. 2012; Wang et al. 2013), qFE15.1 (Sun et al. 2012), qFE24.1 (Shen et al. 2007), qFL07.1 (Sun et al. 2012; Wang et al. 2013), qFL14.1 (Sun et al. 2012), qFL17.1 (Wang et al. 2013), qFM05.1 (Sun et al. 2012), qFM07.1 (Sun et al. 2012), qFM14.1 (Sun et al. 2012), qFM16.1 (Sun et al. 2012), qFM16.2 (Wang et al. 2013), qFS02.1 (Wang et al. 2013), qFS07.1 (Sun et al. 2012; Wang et al. 2013), qFS23.1 (Shen et al. 2007), qFS24.1 (Shen et al. 2007) and qFU07.1 (Sun et al. 2012). Furthermore, 21 QTL detected in the present study were also identified in the populations with the same mapping parents in our previous studies and these QTL included qFE07.1 (Ni et al. 2011; Zhang et al. 2012), qFE11.1 (Ni et al. 2011), qFE15.1 (Zhang et al. 2012), qFE24.1 (Zhang et al. 2012), qFE16.1 (Zhang et al. 2012), qFL01.1 (Ni et al. 2011), qFL07.1 (Chen et al. 2008; Ni et al. 2011), qFL08.1 (Zhang et al. 2012), qFL12.1 (Zhang et al. 2012), qFL17.1 (Zhang et al. 2012), qFL19.1 (Zhang et al. 2012), qFL24.1 (Zhang et al. 2012), qFM07.1 (Ni et al. 2011; Zhang et al. 2012), qFM08.1 (Ni et al. 2011), qFM17.1 (Zhang et al. 2012), qFM24.1 (Zhang et al. 2012), qFS07.1 (Chen et al. 2008; Ni et al. 2011; Zhang et al. 2012), qFS14.1 (Ni et al. 2011), qFS23.1 (Zhang et al. 2012), qFU07.1 (Zhang et al. 2012) and qFU24.1 (Zhang et al. 2012). The QTL identified across multiple environments and populations revealed that they were important for fiber quality traits, even though some of them were largely affected by environmental factors.

Favorable QTL allele origin

In the present study, although the two parents only had significant difference in fiber strength, there are 22 favorable alleles originated from CCRI 35 and 37 favorable alleles originated from Yumian 1. Meanwhile, cultivar Yumian 1 with high fiber strength had more favorable fiber strength alleles than cultivar CCRI 35 with low fiber strength. The same results were also found in the previous studies (Zhang et al. 2009; Sun et al. 2012; Zhang et al. 2012; Wang et al. 2013). Additionally, favorable alleles for fiber elongation and micronaire were also detected in the populations developed from the parents that didn’t have significant difference in these traits (Shen et al. 2005; Sun et al. 2012). These results confirmed that different cultivars comprised of different favorable alleles for the same traits at different position on the genome.

QTL-rich regions

The phenomenon of QTL-rich regions for at least three fiber traits was observed on Chr07, Chr08, Chr17, Chr19 and Chr24, and the similar result was also reported in the previous studies (Paterson et al. 2003; Zhang et al. 2009; Sun et al. 2012; Zhang et al. 2012; Yu et al. 2013). These regions were also near or within the QTL hotspots (Rong et al. 2007; Said et al. 2013).The recent study found that large numbers of coordinately regulated genes existed near ‘hotspots’ for cotton fiber QTL (Paterson et al. 2012), and this discovery seems to imply that the QTL-rich regions maybe result from the closely-linked alleles. However, we couldn’t exclude the possibility that these QTL-rich regions are contributed by pleiotropic alleles, especially for those QTL with overlapping intervals to date.

In conclusion, the genetic map constructed in the present study is the most detailed upland cotton intraspecific map based on SSR markers to date. It generally reveals the upland cotton genome structure and could be used to construct detailed consensus map or as reference genetic map for tetraploid cotton genome assembly. The large number of QTL detected and their distribution on entire genome indicated that regulation of cotton fiber quality traits were complicated with genetic and environmental factors. Stable QTL, especially the qFS07.1 and qFS14.1, reflect some important and favorable alleles shaping fiber strength, and they could be the candidate alleles for MAS breeding projects as well as for gene function research to reveal the molecular regulation mechanism of fiber strength.