Quantitative trait locus mapping of yield and plant height in autotetraploid alfalfa(Medicago sativa L.)

Fei He,Ruici Long,Tiejun Zhng,Fn Zhng,Zhen Wng,Xijing Yng,Xueqin Jing,Chngfu Yng,Xuxin Zhi,Mingn Li,Longxi Yu,Junmei Kng,*,Qingchun Yng,*

aInstitute of Animal Science,Chinese Academy of Agricultural Sciences,Beijing 100193,China

bPlant and Germplasm Introduction and Testing Research,United States Department of Agriculture-Agricultural Research Service,Prosser,WA 99350,USA

ABSTRACT

Keywords:

1.Introduction

Alfalfa(Medicago sativa L.)is a worldwide forage legume crop with high yield and nutritional value[1].It is used for animal feed and is also a valuable rotation crop owing to its ability to fix atmospheric nitrogen.It has been researched as a source for cellulosic ethanol[2].Alfalfa is a perennial tetraploid species with high heterozygosity,which increases the genetic complexity and challenge of breeding.Biomass yield is the most critical trait for alfalfa production and is controlled by multiple genes and influenced by environmental factors.Plant height is considered a critical indicator for biomass yield in alfalfa[3].Although this trait is desirable for alfalfa cultivars,no genetic characterization has been reported.Identifying the genetic basis of biomass yield and plant height traits could assist in alfalfa breeding for yield[4].

Conventional breeding strategies have typically been used to increase alfalfa yield.One way to improve the efficiency of yield selection is to identify genetic loci associated with yield and plant height and then develop diagnostic markers closely linked with yield loci for marker-assisted selection.Most agronomic traits are quantitative and controlled by many genes[5].Quantitative trait locus(QTL)mapping has been used for mapping genetic loci associated with complex traits in various populations in crops[6]and the technique allows mapping those loci that control traits to their corresponding positions in the genome[7].

QTL associated with major alfalfa agronomic traits such as yield,plant height[8],winter hardiness,fall dormancy[9],lodging resistance,spring vigor[10],plant growth,and forage production[11]have been reported.Li et al.[12]found 15 molecular markers closely associated with yield in an alfalfa breeding population.Using genotyping by sequencing(GBS),they mapped 45 and 35 QTL significantly associated with fall dormancy and winter hardiness,respectively[9].Li et al.[13]found 71 QTL associated with plant height and winter injury,and these QTL were linked to yield of alfalfa in an alfalfa F1population.

The rapid advance of DNA sequencing technology allows genotyping of DNA markers with high density covering the entire genome[14].With sequencing methods,large numbers of single-nucleotide polymorphisms(SNPs)can be obtained at relatively low cost even in species without reference genomes.GBS[15]and restriction site-associated DNA sequencing(RAD-seq)technologies have been widely used for genotyping.Peterson et al.[16]developed DNA markers for QTL mapping.The application of GBS has been reported in many species including soybean[17,18],maize[19],and potato[20].GBS was used in alfalfa by Yu et al.[21].GBS could be used to identify QTL associated with yield and plant height.

The genetic control of yield-related traits of alfalfa is not clear.Few QTL have been mapped on low-density linkage maps of alfalfa.Mapping QTL on a high-resolution genetic map will accurately detect genetic loci and allow functional analysis of these genomic regions.The objective of this study was to investigate the genetic basis of alfalfa yield related traits by using GBS to develop high density linkage maps and mapping QTL associated with yield and plant height in a fullsib population.The results were expected to provide valuable information for marker-assisted breeding for yield potential in alfalfa.

2.Materials and methods

2.1.Plant materials and growth conditions

Two parental genotypes,landraces Medicago sativa Cangzhou(CF000735)(male parent)and the improved cultivar Medicago sativa Zhongmu 1(CF032020)(female parent),were crossed to generate an F1population consisting of 392 progeny.In 2012,seeds of the population were planted in a greenhouse at the Chinese Academy of Agricultural Sciences(CAAS)in Langfang,Hebei province,China(39.59°N,116.59°E).The greenhouse was maintained at 16 h day/8 h night,22 °C and 40% relative humidity.Clones were propagated from individual plants by stem cuttings.During the early branching stage in 2013,the cloned plants were transplanted from propagation flats to the field of the CAAS research station in Langfang.

In the experimental field,the annual average temperature was 11.9°C,and the coldest month was January(-4.7°C),and the hottest month July(26.2 °C).The average annual precipitation was 554.9 mm,with large spatial and seasonal variation.During the summer season(July through September),the area received more than 50% of the annual precipitation.The soil was a medium loam soil containing 1.69% organic matter(pH 7.37).The field trial employed a randomized complete block design with three replications.Every replication had one cloned plant for each individual.The spacing was 100 cm between rows and 80 cm between plants.No fertilizer or irrigation was applied,and weeding was done manually.Since the individual plants were not similar to each other after transplanting,mowing was performed on each individual plant before winter to 5 cm above ground,thus ensuring consistency among individuals.

2.2.Phenotype measurement and data analysis

Phenotypic data were collected for three years from 2017 to 2019.Above ground biomass was harvested and total biomass yield was measured four times per year at the beginning of flowering.Yield was measured after the harvested plant was dried in a forced-air dryer.Before harvest,the plant height of each plant was measured with a centimeter ruler,from the surface of the soil to the top of the main stem.Statistical analysis was performed for each harvest using analysis of variance(ANOVA)[22].A normal distribution test was performed for each harvest time.The yield and plant height of four harvests at Langfang were summed to obtain total biomass yield.The genotypic effects of yield,plant height,and the interaction of genetic effect with environment were estimated with a general linear model using PROC GLM[22].ANOVA and heritability estimates were based on phenotypic data from all three years,including plant height and yield.

2.3.Genotyping

GBS was used to generate sequence data following Elshire et al.[15].A CWBIO plant genomic DNA kit(CoWin Biosciences,Beijing,China)was used to extract DNA from 100 mg of fresh young leaf tissue.DNA was quantified using a Nanodrop 2000 spectrophotometer(Thermo Scientific,Waltham,MA,USA)based on 260 nm absorbance.DNA was then digested with the EcoT22I(ATGCAT)restriction enzyme,and libraries were sequenced on a Hi-Seq2000(Illumina)with 96 lines.The Tassel 3.0 Universal Network Enabled Analysis Kit(UNEAK)pipeline[23]was used to discover SNP without a reference genome.Briefly,the quality check of initial sequences was performed using FastQC.The FASTQ files were recorded into the tagCount file using UFastqToTagCountPlugin.Next,the UTagPairToTBTPlugin was used to tag distributions in all of the taxa.Finally,HapMap files were generated using UMapInfoToHapMapPlugin with the MAF value from 0.05 to 0.5.

2.4.Genetic linkage map

The linkage map information has been reported separately[24].In the UNEAK pipeline[24],the UQseqToTagCountPlugin command line used a minimum of 5 tag numbers and a 64-ntp target length to trim the sequence.SNP markers with more than 50% missing values were removed.Other parameters were set to default values.Single-dose alleles(SDA,AAAB×AAAA)with a segregation ratio＜2:1 in the F1progeny were used to construct a genetic linkage map using JoinMap[25].Linkage-group names were assigned based on chromosome numbers from Medicago truncatula.Four homologous linkage groups were assigned randomly.

2.5.QTL mapping

QTL mapping was performed using yield and plant height data by additive composite interval mapping(ICIM-ADD)in the QTL IciMapping software[26]with the mapping population types P1BC1F1 and P2BC1F1.The ICIM-ADD(inclusive composite interval mapping for additive effect QTL)method was applied as in a previous study[27].QTL were mapped using the BIP function[26]in QTL IciMapping with a LOD threshold of 3.QTL locations were drawn with MapChart[28].

3.Results

3.1.Phenotypic data analysis

Phenotypic data for yield and plant height were analyzed using best linear unbiased prediction(BLUP).The minimum plant height and yields were 35.4 cm and 114 g,respectively.The maximum plant height and yields were 58.2 cm and 1412.4 g,respectively.The range of values for F1plants was wider than that of the parents,reflecting the presence of transgressive segregation(Table 1).Broad-sense heritability(H2)was calculated as described in a previous study[29].Broad-sense heritability(H2)ranged from 0.77 to 0.81.Pearson's correlation between yield and plant height was significant(r=0.898,P＜0.01).Pearson correlation analysis showed a significant correlation(P＜0.01)among yield and plant height traits.Wide variation was observed in both traits(Table 1).Genotypic variation,variation between years,and genotype×year interactions were significant(P＜0.001)for all plant height and yield traits(Table S1).

3.2.Genetic linkage map

The male parent linkage map spanned 4088 cM with 944 mapped markers at a mean marker density of 4.33 cM.The highest number of markers was found in linkage group(LG)7D(186 markers)and lowest in LG 4B(7 markers)(Table S2).The female parent linkage map spanned 4229 cM with 2874 mapped markers at a mean density of 1.47 cM.The highest number of markers was found in LG 1C(188 markers)and the lowest in LG 6D(31 markers)(Table S2).

3.3.Identification of QTL for yield and plant height in the F1 population

Phenotypic data collected in three years were treated as three environments,LF2017,LF2018,and LF2019,and used for QTL mapping.Phenotypic data were analyzed using BLUP,and the BLUP values were used for QTL mapping.Table 2 lists QTL for yield and plant height identified by composite interval mapping(CIM)in the three years.Seven QTL for plant height and nine for yield were detected on chromosomes 1–8(Table 2).The phenotypic variance explained(PVE)ranged from 1.96% to 30.2% and from 3.08% to 21.63% for yield and plant height,respectively.Among all these 16 QTL,six explained more than 10%of the phenotypic variation.Yield and plant height QTL were located on chromosomes 6D(qheight-5,qyield-4,qyield-5)(25.5%,10.32%,and 18.97%),7D(qheight-1 and qyield-2)(32.24% and 21.63%),and 1B(qyield-1)(11.29%).The highest PVE value was 32.24%on chromosome 7D(qheight-1)at 67.5–68.5 cM.Three pairs of QTL were co-located on five chromosomes:qyield-1 and qheight-7,qheight-5 and qyield-4,qheight-6 and qyield-6(Figs.1 and 2).

3.4.Analysis of QTL mapping results

Six QTL were identified in the same location in all three years(Figs.1,2)and a QTL for both plant height and yield was found at two locations on chromosome 7B(qyield-1 and qheight-7)(36 cM)in the male parent(Fig.1).These QTL showed LOD values ranging from 3.28 to 4.07.The PVEs of individual QTL ranged from 3.62% to 11.28%.In the QTL map of the female parent(Fig.2),two QTL for plant height and yield were found in the same locations on chromosome 6D(qheight-5 and qyield-4)(25 cM).These QTL showed LOD values ranging from 9.5 to 28.5 and PVE ranging from 10.32% to 25.5%.Two QTL were found at the same locations on chromosome 7A(qheight-6 and qyield-6)of the female parent(79 cM).These QTL showed LOD values ranging from 5.24 to 6.88 and PVE from 5.52%to 5.58%.Yield QTL overlapped at the positions of chromosomes 7B(qyield-7)and 8C(qyield-8),and the positions were less than 1 cM apart.

Table 1–Statistics of plant height and yield in the F1 population.

Table 2–Quantitative trait loci(QTL)associated with yield and plant height traits identified by inclusive composite interval mapping of additive effects in the paternal parent.

4.Discussion

Fig.1–Locations of yield(red bars)and plant height(green bars)QTL on the male parent linkage map.

In this study,we used a full-sib F1 alfalfa population to map QTL associated yield and plant height.It is worth noting that mapping in an F1 population reflects the variation within each of the parents[30].Therefore,trying to compare whether two parents show significant differences in traits may not explain our linkage map.New allele combinations in progeny lead to new variation in yield traits.The two-parent lines used to create the population map differed in the phenotype of yield traits(Table 1),as the traits did in the population.To reduce the influence of genetic and environmental interaction(G×E),we used BLUP to estimate the phenotypic variation between traits for three years and to identify the feasibility of QTL for each trait analyzed in this study(Table S1).The positive correlation between yield and plant height in the three study years may be due to genetic differences under different climate or environments in those three years.The results showed that alfalfa yield was strongly affected by environment.

Fig.2–Locations of yield(red bars)and plant height(green bars)QTL on the female parent linkage map.

The application of molecular markers in breeding alfalfa has been relatively slow compared with its application in other crops.Alfalfa is an autotetraploid species in which genetic improvement is difficult.Even today,most breeders still use conventional methods(e.g.recurrent selection)for breeding alfalfa,relying heavily on repeated phenotypic selection.This approach works well for simple or highly heritable traits,but not for quantitative traits,especially when heritability is low.QTL analysis of agronomic traits,in contrast,can be used to investigate their genetic bases,assisting in breeding for such traits.The high-density linkage maps obtained in this study were comparable to previous maps[9].However,linkage mapping was performed with software for mapping in diploid mating designs such as F2,recombinant inbred lines,and backcross.It sacrifices allele information from tetraploid alfalfa.Also,because alfalfa is an outcrossing species,selfing may cause severe depression.For this reason,it is best to use an F1population and software developed for F1populations.GACD is software developed for linkage map construction and QTL mapping in an F1population[31]developed from crossing two heterozygous parents.The F1population of alfalfa used in this study could be analyzed using GACD.

QTL associated with alfalfa yield have been reported[8,13,24,32].Zhang et al.[24]performed QTL mapping of yield-related traits with RAD-seq and identified five yield-related QTL.Our study was similar to theirs but employed different methods.Our population was sequenced by GBS and was larger than theirs(392 instead of 149 F1plants).Finally,their study used only first-cut yield and plant height.The QTL found in the present study were highly correlated with each other but different from those found in Zhang et al.[24].We identified yield QTL at 29 cM(qyield-5),45 cM(qyield-7),and 46 cM(qyield-8)within the same intervals(qyield-2 and qyield-4)as those found by Zhang et al.[24].We also identified QTL on chromosomes 1,6,and 7 associated with yield and plant height,whereas Zhang et al.[24]found no QTL on these chromosomes.

In the present study,we identified 16 major QTL associated with yield and plant height in the F1population,including one new QTL on chromosome 1.Some reports on alfalfa have identified loci for winter damage and fall dormancy,but the QTL were widely spaced(＞10 cM)[13]with low resolution.So further studies were needed to narrow the QTL intervals and generate single-dose allele(SDA)maps for tetraploid species.We first sought significant associations among SNP markers to determine QTL of yield and plant height,and then performed CIM to identify more accurate genomic locations for each QTL.We used SDA markers and CIM to reduce the QTL spacing(＜3 cM)in comparison with that in previous studies.

QTL related to yield and plant height identified in this study was located on the same chromosomes as QTL for alfalfa yield[24]and winter injury[33].This consistency further indicates that our QTL results are reliable.The loci found at 36 and 79 cM of chromosome 7(qyield-2 and qheight-6)were comparable with QTL for plant height previously[8]found on chromosome 7 of tetraploid alfalfa,and the loci located at 78–104 cM and 66–98 cM in the same interval were in the same region in the present study.Adhikari et al.[9]located QTL for fall dormancy and winter hardiness at 34.6–48.7 cM on chromosome 7,within the same interval(36 cM)as QTL identified in the present study.The QTL found at the corresponding position on chromosome 1 has not been reported previously.In all,the loci we found were mutually verified by other studies.Some loci were new,indicating the complexity of the genetic structure controlling the yieldrelated traits in alfalfa.

In summary,the plant population showed wide variation in plant height and yield,which can be used to dissect the genetic architecture of yield-related traits.We identified QTL on chromosomes 1,6,and 7 associated with yield and plant height.The identification of identical yield QTL in both previous and present analyses suggests a common genetic base for genetic improvement of the trait.

Declaration of competing interest

The authors declare that there are no conflicts of interest.

Acknowledgments

The authors thank the reviewers for their valuable comments on this manuscript and gratefully acknowledge the financial support for this study provided by grants from the Collaborative Research Key Project between China and EU(granted by the Ministry of Science and Technology of China,2017YFE0111000),the China Forage and Grass Research System(CARS-34),the Agricultural Science and Technology Innovation Program of CAAS(ASTIP-IAS14),and the National Natural Science Foundation of China(31772656).

Author contributions

Junmei Kang and Qingchuan Yang designed the experiments,developed the mapping population,and revised the paper.Fan Zhang,Tiejun Zhang,and Longxi Yu performed the genotyping and constructed genetic maps.Zhen Wang,Xijiang Yang,and Xueqian Jiang collected phenotypic data in Langfang in 2017 and 2018.Changfu Yang,Xuxin Zhi,and Mingna Li managed field work and collected phenotypic data in 2019.Fei He and Ruicai Long performed data analysis and wrote the manuscript.All authors read and approved the final manuscript.

Appendix A. Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2020.05.003.