Dvi D.Fng , Lingh Zng , Grgory N.Thyssn c, Christophr D.Dlhom , Efrm Bchr ,Don C.Jons, Ping Li

a Cotton Fiber Bioscience Research Unit, USDA-ARS-SRRC, New Orleans, LA 70124, USA

b Crop Genetics Research Unit, USDA-ARS, Stoneville, MS 38776, USA

c Cotton Chemistry & Utilization Unit, USDA-ARS-SRRC, New Orleans, LA 70124, USA

d Cotton Structure and Quality Research Unit, USDA-ARS-SRRC, New Orleans, LA 70124, USA

e Cotton Incorporated, Cary, NC 27513, USA

Keywords:

ABSTRACT Previously we identified a major cotton fiber strength QTL(qFS-c7-1)on chromosome A07 using a multiparent advanced generation intercross (MAGIC) population.To assess the stability and transferability of this QTL and its utility in cotton breeding,we made ten new populations.These populations were developed from crosses between MAGIC recombinant inbred lines,or between cotton cultivars that are different from the MAGIC parents.A total of 2801 F2 plants were grown and their fiber quality traits were measured.We also selected a subset of F3 seeds from two populations,and grew F3 progeny plots to further evaluate the stability of this QTL.Our results showed that the peak of qFS-c7-1 is at 70–72 Mb region.This QTL had a major effect on fiber strength explaining 21.9% phenotypic variance.Its effect on other fiber quality attributes such as micronaire, short fiber content, length and uniformity varied between populations, and no effect on fiber elongation was observed.The QTL effects were stable in the populations analyzed, and in different generations of the same population.The SSR and SNP markers near and within the QTL peak reported herein will assist selecting superior fiber quality traits in breeding, with a recommendation that the parental cotton lines should be analyzed using the seven DNA markers within the QTL peak before fully implementing marker assisted selection in a cotton breeding program.

1.Introduction

The physical attributes contributing to cotton (Gossypiumspp.)fiber quality are multiple traits controlled by many genes, and often times affected by environmental factors [1–4].Improving fiber quality while maintaining or increasing yield has been a major breeding goal for a majority of cotton breeding programs in the world.Although incremental gains in fiber quality attributes such as strength and length have been achieved during the past decade via traditional breeding, those gains have been small and become more difficult to improve further without new breeding technologies.Marker-assisted-selection (MAS) is considered one such technology.In cotton, MAS has been successfully used to improve disease resistance such as blue disease,bunchy top,bacterial blight,rootknot and reniform nematodes[5–9].Applications of MAS to improve complex polygenic traits such as fiber quality and yield have started but are still very limited [4,10–12].

Since the first cotton genetic map was published in 1994 [13],cotton scientists have been developing molecular markers associated with fiber quality quantitative trait loci (QTL) using a variety of mapping populations [5]with the aim to improve fiber quality.So far, ＞3000 fiber QTL have been reported in the literature[3,14–17].However, the effects of many of these QTL could not be replicated in different environments or populations [18], which significantly reduced their value in breeding.This also partially explains why use of molecular markers to assist selection of a quantitative trait such as fiber quality in cotton breeding is still at its early stage.This low level of repeatability of fiber QTL effects may be due to but not limited to the following three factors:1)the majority of mapping populations previously used were interspecific crosses betweenGossypium hirsutumL.(commonly called Upland cotton) andG.barbadenseL.(commonly called Pima, Egyptian or Sea island cotton).A fiber QTL fromG.barbadensemay not be present inG.hirsutum; 2) markers showing polymorphism between two species may be monomorphic withinG.hirsutum, which in turn makes the markers not useful in Upland cotton breeding; 3)the QTL effect is not stable, or its effect is so small that it would be nullified under a new environment or in a new genetic background.In realization of limitations 1 and 2,cotton scientists have been usingG.hirsutumintra-specific populations or a pool (usually ＞100) of Upland cotton cultivars to identify fiber quality QTL [11,19–23].Although low intra-species polymorphism is a challenge, it can be overcome by using the combination of: a) a population derived from multiple (e.g.＞10) parents to ensure a greater genetic diversity and an increased polymorphism frequency in the population, or a pool of cotton cultivars; and b)sequencing the whole genomes of progenies or cultivars to identify large number of DNA sequence variants.Novel fiber QTL have been identified via this approach [24,25].However, studies to address the limitation 3, i.e., the stability and transferability of a fiber QTL in different genetic backgrounds and/or environments,are still scarce.

In our previous studies, we used a multi-parent advanced generation intercross (MAGIC) population to identify DNA markers associated with fiber quality QTL.This MAGIC population was developed from half-diallel crosses between 11 Upland cotton lines(Acala Ultima, Coker 315, Deltapine Acala 90, Fibermax 966,M240RNR, Paymaster HS26, Phytogen PSC 355, Stoneville 474,Stoneville 825, Suregrow 747, and Tamcot Pyramid).These cotton lines include 10 cultivars and one breeding germplasm(M240RNR).They represent a diverse group of non-related cotton lines developed from major breeding programs across the United States cotton belt.Five cycles of random-mating and six cycles of self-pollination were completed to produce a MAGIC population consisting of 550 recombinant inbred lines(RILs)[26,27].We have identified QTL for fiber length,strength,short fiber content,uniformity, and elongation using simple sequence repeat (SSR) markers[28], genotype-by-sequencing based single nucleotide polymorphisms (SNPs) [27]and whole genome sequencing [24].The fiber strength QTL (qFS-c7-1) on chromosome A07 was consistently detected in all 14 environments [24,28].In order to evaluate the stability of the QTLqFS-c7-1under different environments and its transferability in populations with different genetic backgrounds,we made ten new F2populations.Four were developed from crosses between a MAGIC RIL with strong fiber and a RIL with weak fiber while the other six were developed from crosses between cotton cultivars that are different from the 11 MAGIC parental lines.We grew the progeny plants in Stoneville,MS,USA to obtain fibers.In addition,we selected a subset of F2plants from two populations,and grew F3progeny plots in a field to evaluate the stability of the QTLqFS-c7-1across generations.Our goals were to evaluate whetherqFS-c7-1is present in new populations, and to assess its utility in a breeding program.

2.Materials and methods

2.1.Population development, field experiments and fiber quality measurement

Development of the Upland cotton MAGIC population was briefly described above.A detailed description about the MAGIC population development was reported by Jenkins et al.[26]and Islam et al.[27].The 11 cotton lines used to develop the MAGIC population were selected based on following criteria: not related in pedigree,developed by 10 independent breeding programs that were located in different regions spanning the entire US cotton belt,and cultivated varieties(except M240RNR which is a rootknot nematode resistant germplasm).More details about these 11 parental lines are described by Fang et al.[28]and Jenkins et al.[26].The MAGIC RILs and their 11 parents were grown in three locations between 2009 and 2017, in a total of 14 environments[24].

Ten new F2populations were developed for this study(Table 1).These populations were developed for two purposes:populations 1 through 4 were used for evaluating the stability of the fiber strength QTLqFS-c7-1, and populations 5 through 10 were used to analyze the transferability of this QTL in different genetic backgrounds.For populations 1 through 4,we chose four RILs from the MAGIC population as parents to make crosses.RIL119 had very weak fiber (28.26 g tex-1).Its fiber strength consistently fell into the lowest fifth percentile of all 550 RILs across 14 environments.In addition, RIL119 had the same marker genotype (designated‘B’ genotype) as the weakest parent Tamcot Pyramid after analyzing DNA markers flanking theqFS-c7-1QTL region (63–73 Mb region on chromosome A07).RIL009, RIL303 and RIL037 were top three RILs with strongest fiber in all environments with average fiber strength of 38.66,37.65,and 37.07 g tex-1, respectively.Furthermore,their DNA marker genotypes at the QTL region were the same as the strongest parent Acala Ultima (designated ‘A’ genotype).Crosses were made in 2018 in a greenhouse in New Orleans,LA, USA.F1plants were grown in the same greenhouse during the winter to produce F2seeds.F2seeds were planted in a field in Stoneville, MS in 2019.

Is the fiber strength QTLqFS-c7-1present in cotton lines with different genetic backgrounds from the eleven MAGIC parental lines? If the answer is positive, then what is this QTL’s effect on fiber strength and other fiber traits in these cotton lines? To answer these questions, we made six new F2populations, i.e.,Pop 5 through Pop 10(Table 1),derived from crosses between cotton cultivars with different genetic backgrounds.All crosses were made in a greenhouse in Stoneville, MS.F2plants were grown in fields in Stoneville, MS.Their respective planting years are shown in Table 1.The parental cultivars were selected based on their genetic background differences from the 11 MAGIC parents and fiber strength (Table 2).TAM182-34ELS (PI 654362) [29]is an extra-long staple cotton line from Texas A&M University, while AR9317-26 is a semi-naked breeding line from University of Arkansas.Previously, we [30]used these two cotton lines to make two reciprocal crosses in a research of net ginning energy requirement.MD52ne (PI 634930) is a cotton line with good fiber strength developed by USDA-ARS in Stoneville, MS [31].FM832 (PVP 9800258) is an okra leaf cotton cultivar with good fiber quality[32].MD15 (PI 642769) [33]has good fiber quality especially strength and was developed from a cross between FM832 and MD51ne.TAM98D-99ne (PI 636491) [34]is a nectariless cotton line released by Texas A&M University.Acala 1517-80 is a high fiber quality cotton cultivar developed by New Mexico State University [35].JJ1145ne is a high yield and average fiber quality germplasm derived from a cross between cotton cultivars JAJO 9596 and JAJO 9550 (Jack Jones, JAJO Genetics, Baton Rouge, LA;personal communication on 03/21/2012).UA48 (PI 660508) [36]is a cultivar developed by University of Arkansas,and characterized as superior fiber quality and high yield.The fiber quality measurements of all parental lines are shown in Table 2.Except TAM182-34ELS and AR9317-26, all parental lines were genotyped with at least one SSR marker closely associated withqFS-c7-1locus to ensure polymorphism between two parental lines before a cross was made.All F2populations were planted in Stoneville,MS.Their respective planting years are shown in Table 1.Two weeks after emerging, plants were thinned down to a density of one plant per 30 cm.Plants were grown in 12.2 m×1.0 m single-row plots.Standard field practices were applied in all growing seasons.

In order to further evaluate the stability ofqFS-c7-1, we grew a subset of F3plants from Pop 8 and Pop 10 in 2019 in Stoneville,MS.For Pop 8, we selected 107 F2plants: 30 plants with strong fiber(＞42.35 g tex-1), 30 with weak fiber (＜35.5 g tex-1), and 47 with intermediate fiber strength (＞36.0 and ＜42.3 g tex-1, average 39.8 g tex-1).For Pop 10, we selected 60 F2plants: 30 plants with strong fiber(＞32.6 g tex-1),and 30 with weak fiber(＜27.0 g tex-1).We made selections solely based on fiber quality measurements of F2plants although a majority of the selected plants demonstrated correlation between their fiber strength and QTL marker genotypes.Seeds of each F2plant were planted as one F3progeny plot of 12.2 m×1.0 m.

Table 1 Populations used in this study.

Table 2 Fiber quality measurements of parental lines used in this study.

Ten to fifteen naturally open bolls were manually harvested from each F2plant for fiber quality measurement.Thirty bolls were harvested from each F3progeny plot.Cotton bolls from the populations 1 through 4 were ginned using a roller gin,while bolls from the populations 5 through 10, and from the F3plants were ginned using a laboratory saw gin.Bundle fiber strength (STR, g tex-1),elongation (ELO, %), micronaire (MIC), short fiber content (SFC,%),upper-half mean fiber length(UHM,mm),and uniformity index(UI, %) were measured using a High Volume Instrument (USTER Technologies Inc., Knoxville, TN, USA) in the Cotton Structure and Quality Testing Laboratory, USDA-ARS-SRRC, New Orleans, LA.Each fiber sample was measured five times, and data presented in this paper is the mean of the five measurements.

2.2.DNA isolation and molecular marker analysis

Young leaves were collected from parental lines and F2plants inall 10 populations.Total DNA was extracted from the leaves following a CTAB-based protocol described before [7]with an additional RNAase A digestion step before binding DNA to a column.DNA quantity and quality were measured using a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) as well as on a 1.5% agarose gel.

SSR, InDel and SNP marker analysis was according to the protocols described earlier [7,37].Primer sequences for the markers used in this study are shown in Table S1.

2.3. Statistical analyses and QTL mapping

Of the 11 MAGIC parents, Acala Ultima has the strongest fiber while Tamcot Pyramid has the weakest fiber.In addition, their marker genotypes around theqFS-c7-1region are completely different.Thus, one-way ANOVA was used to compare fiber quality means between three groups of samples possessing Acala Ultima type (A), Tamcot Pyramid type (B), and heterozygous (H) marker genotypes according to the Tukey-Kramer method (JMP Genomics 9.0; SAS, Cary, NC, USA).In a one-way ANOVA figure, if two circles do not connect, or the intersection angle of two circles is ＜90°, the group difference is considered significant.For more detail, refer to the user manual of JMP Genomics 9.0 [38].The TM-1 reference genome at the QTL region is ‘B’ type.Pearson coefficient between the fiber quality traits was calculated using Microsoft Excel.

The software program MapQTL 6 [39]was used to map QTL using the combined phenotypic and marker data of populations 1 through 4 based on the multiple-QTL models (MQM).The cofactor was determined using the Automatic Cofactor Selection built in the MapQTL 6.

3.Results

3.1.Fiber trait variation and Pearson coefficients between six traits

We collected fibers and measured six fiber quality attributes:STR, ELO, MIC, SFC, UHM, and UI.The fiber quality measurements of all parental lines and F2progeny plants are shown in Tables 2 and S2, respectively.For the populations 1 through 8, the mean values of the progenies were similar to or slightly better (i.e.,higher STR, ELO, UHM, UI, and lower MIC or SFC) than the mean values of their respective parental lines for all traits.For the Pop 9 and Pop10,the parents had slightly better mean values than their F2progenies.With little exception, the standard deviations, minimum and maximum values in each F2population were broader than their respective parents, indicating transgressive segregation(Tables 2 and S2).We calculated Pearson coefficients between six fiber quality traits using either the data of each individual population or combined data of a few populations.Table S3 shows Pearson coefficients between the fiber traits based on the combined data of populations 1 through 4, of Pop 5 and Pop 6, and of populations 7 through 10.We observed significant correlation between STR and SFC, UHM, or UI in all data sets.ELO and MIC were also correlated with the other four traits but correlation coefficient values were lower.

We observed fiber quality differences of the cotton cultivars TAM 182-34 ELS and AR 9317-26,and their F2plants grown in different years.The overall fiber quality was better in 2016 than in 2012.

3.2.The effect of qFS-c7-1 is stable in the four RIL×RIL populations

Previously we [24,27,28]reported that the QTLqFS-c7-1is around 70–72 Mb region according to the TM-1 reference genome released by Zhang et al.[40], herein referred to as NAU genome.The representative markers include SSR markers DPL0757,DPL0852, C2-0114 and SNP marker A7_7115 (Table 3).We developed more DNA markers covering the genomic region between 56 and 77 Mb on chromosome A07 based on the whole genome sequences of the MAGIC RILs.The primer sequences of the markers are listed in Table S1.We compared the marker orders with the newly-published TM-1 genome (referred to as HAU genome) by Wang et al.[41].Both TM-1 genomes released by two different research groups had the identical order in this region although the marker interval sizes showed some differences (Table S1).

We analyzed 1107 F2progenies of populations 1 through 4 using nine markers.MQM-based QTL mapping using the combined marker and phenotypic data of these four populations identified the QTL peak around the location of the marker A7_7115 (71 Mb region) with a LOD score of 53 (Fig.S1).QTL analysis showed thatqFS-c7-1explained 21.9%fiber strength variance in these four populations.The QTL has additive effect of 1.85 and dominance effect of 0.43 on strength.The QTL peak area and its effect on fiber strength in the four RIL×RIL F2populations are in agreement with our previous results obtained based on the genome wide association study of the MAGIC population [24].

A QTL for MIC,SFC,UHM,and UI was also identified at the same location as theqFS-c7-1but with lower LOD scores (Fig.S1).Because fiber strength has high correlation with SFC, UHM and UI, it is not clear whether this is due to the effect ofqFS-c7-1on other fiber traits, or due to the presence of a fiber QTL cluster.No ELO QTL was identified around this region even though a correlation between STR and ELO was present in these populations(Table S3).

Table 3 Physical map of the QTLqFS-c7-1region on chromosome A07.

We conducted one-way ANOVA by comparing the group means according to three marker genotypes.The results with marker DPL0757 as an example are shown in Figs.1 and S2.F2plants with‘B’ genotype had significantly (P= 0.01) lower values in STR, UHM and UI,and higher values in MIC and SFC than the plants with‘A’or‘H’ genotypes.Progeny plants with ‘H’ genotype had significantly lower values in STR and UI than the‘A’genotype plants,but no difference in MIC and UHM values were observed between the‘A’and‘H’ genotype plants (Fig.S2).

Fig.1.One-way ANOVA of fiber strength of four RIL ×RIL F2 populations based on the DPL0757 marker genotypes.STR, fiber strength (g tex--1).The green line is the group mean.Tukey-Kramer method was used with α=0.01.A=Acala Ultima-high strength type, B = Tamcot Pyramid-low strength type, H = heterozygous.

In conclusion,qFS-c7-1had stable and strong effect on STR,SFC,UHM and UI in four populations derived from crosses between a MAGIC RIL with strong fiber (i.e., possessing favorable QTL allele)and a RIL with weak fiber (without favorable QTL allele).Progeny plants with ‘A’ genotype in theqFS-c7-1QTL region had superior fiber.The progeny plants having‘B’genotype had inferior fiber and plants with ‘H’ genotype being intermediate.

3.3.qFS-c7-1 is present in cotton cultivars with different genetic backgrounds

The populations 5 through 10 were used to test whetherqFS-c7-1is present in cotton lines with different genetic backgrounds from the eleven MAGIC parental lines.We genotyped each F2progeny with three or more DNA markers around the QTL locus, and grew the populations to obtain fiber phenotypes.The results are described below:

For Pop 5 (TAM182-34ELS×AR9317-26) and Pop 6 (AR9317-26×TAM182-34ELS), we genotyped F2plants with markers COT133, BNL1604, A7_7069 and DPL0119.The parent TAM182-34ELS possesses favorable QTL allele.Markers COT133, BNL1604,and A7_7069 showed very good association with the target QTL.However, the marker DPL0119 had poor association with the QTL.One-way ANOVA indicated that F2progenies with ‘B’ genotype at marker A7_7069 locus had significantly (P= 0.01) inferior fiber quality(lower STR,UHM,UI,and higher MIC,SFC)to the progenies with ‘H’ or ‘A’ genotype.‘A’ genotype plants had significantly superior STR and MIC to the ‘H’ genotype plants.It is worthy to note that the effect of the QTLqFS-c7-1on UHM and SFC was not as pronounced as in the RIL populations mentioned previously.TheqFS-c7-1did not have an effect on ELO.Fig.S3 shows the ANOVA results of the Pop 6 using the marker A7_7069 as an example.Although the fiber measurement values of Pop 5 were overall lower than those of Pop 6 (Table S2), the QTL effect remained the same in both populations.

For Pop 7(MD52ne×FM832),the two parents had similar fiber quality measurements (Table 2), and the favorableqFS-c7-1allele is present in MD52ne.We genotyped the F2plants with three markers: DPL0852, A7_7069 and A7_7115.The F2plants with ‘B’genotype had significantly (P=0.05) inferior STR, MIC, SFC, UHM,and UI values to the plants with ‘A’ or ‘H’ genotypes.Although‘H’ genotype plants had lower STR than the ‘A’ genotype plants,the difference was not significant atP=0.05 level.Fig.S4 shows the ANOVA results using the marker A7_7115 as an example.

For Pop 8 (MD15×TAM98D-99ne), MD15 possesses favorable allele ofqFS-c7-1.We genotyped the F2plants with three markers:CGR6764,A7_7069,and DPL0852.Each demonstrated close association with the QTL.For all three markers, ‘B’ genotype plants had significantly (P=0.01) inferior STR, SFC, UHM and UI to the ‘A’ or‘H’ genotype.No significant difference between ‘A’ and ‘H’ genotype was observed for any traits.Fig.S5 shows the ANOVA results using the marker DPL0852 as an example.

For Pop 9 (Acala 1517–80×JJ1145ne), we analyzed five SSR markers: CGR6764, DPL0852, C2-0114, HAU1399 and DPL0119.The markers CGR6764, HAU1399,and DPL0119 were polymorphic while DPL0852 (within the QTL peak) and C2-0114 (closest to the QTL peak)were monomorphic between the two parents.The polymorphic markers that are out of theqFS-c7-1peak region implied that Acala 1517-80 was likely ‘A’ genotype while JJ1145ne might be‘B’genotype.None of these three polymorphic markers showed meaningful correlation with fiber quality traits.SNP markers A7_7069 and A7_7115 did not reveal polymorphisms between the two parents.The QTLqFS-c7-1was not detected in this population.

For Pop 10 (TAM98D-99ne×UA48), we genotyped F2plants using three markers (DPL0852, A7_7115 and DPL0757) located within the QTL peak area.We observed that UA48 was heterogenous in the QTL region based on the marker analysis.We chose UA48 plants with ‘A’ genotype as a parent to make crosses.TheqFS-c7-1significantly (P=0.05) affected STR, SFC and UI, but had no effect on ELO, MIC or UHM (Fig.S6).The ‘H’ genotype plants did not have significantly superior fiber strength to the ‘B’ genotype plants.

In summary,the QTLqFS-c7-1is present in cotton cultivars with different genetic backgrounds from the MAGIC parents.In general,this QTL had stable effect on STR across different genetic backgrounds, but had no effect on ELO.Its effect on MIC, SFC, UHM,and UI varied between populations, and usually was less pronounced than on STR if and when such effect was present.In most cases,the‘A’genotype plants did not have significantly better fiber attributes than the ‘H’ genotype plants.

3.4.The effect of qFS-c7-1 is stable in F3 plants

We selected 107 and 60 F2plants from Pop 8 and Pop 10,respectively, based on the fiber strength measurements.We grew the selected F3seeds in a field in Stoneville, MS in 2019 to test whether the effect ofqFS-c7-1would behave similarly in F2and F3plants.In both populations, the QTL effect on each fiber trait in the F3plants matched its effect in F2plants except that the effect magnitude was larger in F3(Fig.2 and Fig.S7) than in F2(Fig.S5).For example, in the F3of Pop 8, the mean fiber strength difference between the plants with‘A’genotype and plants with‘B’genotype was 7.76 g tex-1, while in its F2population, the STR difference between these two groups of progeny plants was only 3.22 g tex-1.The same trend was observed in Pop 10.The mean fiber strength difference between ‘A’ and ‘B’ genotype plants was 2.89 g tex-1in F3, and 1.25 g tex-1in F2.This indicates that in F3and possibly any further generations, selectingqFS-c7-1using the closely associated DNA markers could improve fiber quality.

4.Discussion

Fig.2.One-way ANOVA of fiber strength of the MD15×TAM98D-99ne F3population based on the DPL0852 marker genotypes.STR, fiber strength (g tex -1)-.The green line is the group mean.Tukey-Kramer method was used with α = 0.01.A = Acala Ultima-high strength type, B = Tamcot Pyramid-low strength type,H = heterozygous.

Based on the genome wide association study of 419 cotton cultivars,Ma et al.[3]identified a fiber strength QTL on chromosome A07 and listedGh_A07G1769as a potential candidate gene for superior fiber strength.We believe that this is the same QTLqFSc7-1as what we are reporting in this study.Previously, Zhang et al.[21]confirmed that this QTL was present in the 0–153×sG K9708 population and explained about 14% of fiber strength phenotypic variance.More recently,the same research group[11]further confirmed their findings with more phenotypic and marker data.It is clear based on our previous [24,27,28]and present research and multiple reports from other cotton researchers that the fiber strength QTLqFS-c7-1on chromosome A07 is stable in different genetic backgrounds.The favorable QTL allele had major effect on improving fiber strength, and was present in two of the 11(19%)MAGIC parental lines.We could not determine how many cultivars possess this favorable QTL out of the 419 cultivars in Ma et al.’s report [3]as such data were not available or difficult to obtain from this publication.In our previous report[42],we genotyped 194 cotton cultivars collected from the world using the marker DPL0852 that is within the QTL peak area.The marker data indicated that 58 (29.8%) cultivars potentially possess favorable allele of this QTL.We also genotyped 493 different cotton lines with the marker DPL0852 (data not shown), and 126 (25.5%) of them may contain the favorable QTL allele.This indicates thatqFS-c7-1may be widely present in cotton cultivars and germplasm.

Presence of a fiber QTL cluster in a genomic region poses a challenge in precision cotton fiber QTL mapping and gene identification.In cotton, such clusters were reported many times in the literature [11,14,28].In the present study, we observed that QTL for STR, SFC, UHM, and UI may be present in the same genomic region based on the MQM QTL mapping using a combined phenotypic and DNA marker data from populations 1 through 4.Because cotton fiber quality attributes are inter-correlating with each other,it is also possible that the same fiber strength QTLqFS-c7-1affects all other fiber traits.The latter hypothesis is supported by the results from Pop 5, Pop 6 and Pop 8, but not by Pop 7 and Pop 10.In Pop 7 and Pop 10, the effects ofqFS-c7-1on SFC, UHM or UI were not pronounced (Fig.S4).A further dissection of the QTL and identification and characterization of the underlying genes will provide insights to the question whether the same fiber strength QTL affects multiple fiber traits or a fiber QTL cluster is present.

SSR markers CGR6764 and HAU1399 that flankqFS-c7-1revealed polymorphisms between the cotton lines Acala 1517-80 and JJ1145ne.However, none of the markers within the QTL peak area were polymorphic between these two cotton lines.This rendered Pop 9 not useful to study the effect ofqFS-c7-1.We are not clear whether this is due to: a) the QTLqFS-c7-1is not present in either parent even though the flanking markers are polymorphic;or b) the QTLqFS-c7-1is present in both parents, and the markers within the QTL peak area are monomorphic.In our previous research using an F2population from a cross between MD52ne and its near-isogenic line MD90ne, we failed to detectqFS-c7-1as the two parents were monomorphic at the QTL region [43].Re-checking our DNA marker results implied that both MD52ne and MD90ne might possess favorable alleles ofqFS-c7-1.This fact alerted us that when using DNA markers to assist selection of this QTL in cotton breeding,the potential parental lines need to be analyzed using the markers within the QTL peak area, i.e.the 70–72 Mb region of chromosome A07 [40].

We used a classic breeding technique by selecting a subset of F3seeds based on the fiber strength measurements of the F2plants.In general, the F3plants performed similarly to their respective F2progenitors in fiber quality measurements.The overall fiber strength mean of F3plants was very close to the mean of the F2plants in both populations.However, if making selections based on the DNA marker genotypes, the F3plants with ‘A’ genotype would be 7.76 g tex-1higher than the F3plants with ‘B’ genotype in Pop 8, and 2.9 g tex-1higher in Pop 10.The fact that theqFSc7-1QTL is less effective in Pop 10 than in Pop 8 indicates that UA48, a highly successful cotton cultivar in modern US cotton breeding, may possess other un-identified fiber strength QTL.It could be intriguing to pursue such novel QTL.

It has been reported that fiber quality measurements are affected by the ginning methods [44].It would be ideal to gin all samples using the exact same ginning equipment.However, often times this is not possible especially when a large number of samples are required to be ginned in a short period of time after harvesting.To minimize potential impacts of ginning methods on fiber quality measurements, we ginned all samples of the same population using the same gin equipment.It is difficult to determine whether ginning methods (roller ginvssaw gin) had any impact on the fiber quality measurements used in this study as we did not do a comparative analysis of the same boll samples but ginned using different gin methods.As a precaution, we did not combine fiber data of populations 1 to 4 with data from populations 5 to 10 to conduct analysis in this report.

5.Conclusions

The peak of the fiber strength QTLqFS-c7-1is around the 70–72 Mb region.Representative DNA markers include SSR DPL0852, DPL0757, and SNP A7_7115.This QTL had a major effect on fiber strength, and no effect on ELO.Its effect on MIC, SFC, UHM or UI varied between populations.This QTL was stable in the populations analyzed, and in different generations of the same population.Selection based on the DNA markers associated with this QTL could improve fiber strength and likely MIC, SFC, UHM and UI depending on the populations.It is recommended that the parental cotton lines should be analyzed using DNA markers within the QTL peak area before fully implementing marker assisted selection of this QTL in a practical cotton breeding program.

Author contributions

David Fang conceived and designed experiments, and wrote the manuscript.Linghe Zeng made crosses and conducted field experiments.Gregory Thyssen analyzed DNA sequences, identified SNP markers, and designed primers for marker analysis.Christopher Delhom measured fiber quality attributes.Efrem Bechere made two populations.Don Jones supported the project and critically edited the manuscript.Ping Li helped developing populations and genotyped plants with DNA markers.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This research was funded by the USDA-Agricultural Research Service CRIS projects 6054-21000-018-00D, and Cotton Incorporated project #19-916.We thank Mrs.Holly King at Cotton Structure and Quality Research Unit in New Orleans for measuring the fiber quality attributes using a high volume instrument.Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the USDA which is an equal opportunity provider and employer.

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2020.06.016.