QTL mapping and QTL×environment interaction analysis of multi-seed pod in cultivated peanut(Arachis hypogaea L.)

Ling Wng,Xinlei Yng,Shunli Cui,Guojun Mu,Xingming Sun,Lifeng Liu,*,Zicho Li,*

a Key Lab of Crop Heterosis and Utilization,Ministry of Education,Beijing Key Lab of Crop Genetic Improvement,China Agricultural University,Beijing 100193,China

b College of Agronomy,Hebei Agricultural University,North China Key Laboratory for Crop Germplasm Resources of Education Ministry,Baoding 071001,Hebei,China

c Crop Research Institute,Xinjiang Academy of Agricultural and Reclamation Science,Shihezi 832000,Xinjiang,China

Keywords:Best linear unbiased prediction BLUP QTL×environment interaction Ratio of multi-seed pod RMSP

A B S T R A C T To dissect the genetic mechanism of multi-seed pod in peanut,we explored the QTL/gene controlling multi-seed pod and analyzed the interaction effect of QTL and environment.Two hundred and forty eight recombinant inbred lines(RIL)from cross Silihong×Jinonghei 3 were used as experimental materials planted in 8 environments from 2012 to 2017.Three methods of analysis were performed.These included individual environment analysis,joint analysis in multiple environments,and epistatic interaction analysis for multi-seed pod QTL.Phenotypic data and best linear unbiased prediction(BLUP)value of the ratio of multi-seed pods per plant(RMSP)were used for QTL mapping.Seven QTL detected by the individual environmental mapping analysis and were distributed on linkage groups 1,6,9,14,19(2),and 21.Each QTL explained 4.42%-11.51%of the phenotypic variation in multi-seed pod,and synergistic alleles of 5 QTL were from the Silihong parent.One QTL,explaining 4.93%of the phenotypic variation was detected using BLUP data,and this QTL mapped in the same interval as qRMSP19.1 detected in the individual environment analysis.Seventeen additive QTL were identified by joint analysis across multiple environments.A total of 43 epistatic QTL were detected by ICIM-EPI mapping in the multiple environment trials(MET)module,and involved 57 loci.Two main-effect QTL related to multi-seed pod in peanut were filtered.We also found that RMSP had a highly significant positive correlation with pod yield per plant(PY),and epistatic effects were much more important than additive effects.These results provide theoretical guidance for the genetic improvement of germplasm resources and further fine mapping of related genes in peanut.

1.Introduction

Peanut is a very important oil crop and edible protein source in many parts of the world.It is the second largest oil crop in China,with a cultivation area of about 4.67×106ha per year.Its total yield,yield per hectare,and oil content of the seeds were ranked first among oil crops[1].Multi-seed peanut is one of the four major varietal types in China[2].It is characterized by high protein,low oleic acid and high linoleic acid contents,very early maturity to avoid drought,and high resistance to the diseases rust,early spot and late spot[3].Multi-seed cultivars are widely planted in both northeastern China,an early maturity peanut-producing area and the northwest,an inland production area.

Seed number per pod,which is related to pod and seed size and maturity is an important agronomic trait[4].In the early studies Badami[5]and Balaiah et al.[6]found that seed number per pod of≥3 was dominant to the seed number per pod of＜3,but the relationship not consistent with the studies of gene number controlling the trait.For example,these researchers and Branch et al.[7]considered the seed number per pod was controlled by 3,1,and 2 pairs of genes,respectively.However,Schilling et al.[8]considered that seed number per pod was controlled by non-additive genetic effects.In order to resolve these different findings the construction of high density linkage maps and QTL mapping is required to revealing the true genetic basis of the multiseed pod trait and for molecular breeding.

In the early 1990s,the genetic linkage map of peanut was mainly based on RFLP,RAPD,and AFLP molecular markers in wild type,which was due to the polymorphism of these markers in wild type peanut was more rich than that in cultivated peanut[9-11].The first partial linkage map of molecular markers in peanut constructed by Halward et al.[12]based on an F2population of A.stenosperma×A.cardenasii.The map contained 117 RFLP markers in 11 linkage groups and covered a total distance of 1063 cM;three genes related to fat synthesis enzymes were mapped.Cultivated peanut is allotetraploid(AABB,2n=4x=40)with a large genome(about 2.7 GB)and complex genetic background.At present,the marker density of the molecular genetic linkage map is low.Varshney et al.[13]constructed the first genetic linkage map based on SSR markers using a RIL population developed from a cross of ICGV86031 and TAG 24.It contained 135 SSR markers in 22 linkage groups and covered a genetic distance of 1270.5 cM.The map was enriched to 1785.4 cM by Ravi et al.[14]and included 191 SSR markers with an average distance of 9.34 cM between markers.In addition,Shirasawa et al.[15]later constructed a high-density genetic linkage map using 1114 markers distributed in 21 linkage groups,and the total distance of this map was 2166.4 cM with the average distance of 1.9 cM between markers.

Several major QTL for pod size and weight in cultivated peanut were identified recently[16,17].Notably,5 QTL showed major and stable expressions in four environments.qHPWA07.1,qPLA07,and qPWA07 explaining 17.93%-43.63%of the phenotypic variation in three traits(hundred pod weight,pod length,and pod width)were co-localized in a 5 cM interval(1.48 Mb in the physical map)on chromosome A07 in which 147 candidate genes related to catalytic activity and metabolic process were located.Moreover,qHPWA05.2 and qPLA05.1 were co-localized with minor QTL qPWA05.2 to a 1.3 cM genetic interval(280 kb in the physical map)on chromosome A05 containing 12 candidate genes.Because the pod is one of the most important organs economically,the above results are important in regard to improvement in pod yield.In addition,genes/QTL for many other important yieldrelated traits in peanut,such as plant height[18,19],branch length[20],total branch number[21,22],plant type[23,24],seed trait[25,26]and shelling percentage[27,28]were mapped.However,QTL for the multi-seed pod trait have not been reported.

Here,we studied the genetic basis of multi-seed pod in peanut using a RIL population derived from a cross between Silihong(var.fastigiata)and Jinonghei 3(var.hypogaea).We mapped QTL for the ratio of multi-seed pods per plant(RMSP),explored QTL/gene related to multi-seed pod,and then the QTL detected under biparental populations(BIP)and MET module were analyzed.QTL×environment interaction effects were estimated.The study provides a theoretical basis for genetic improvement of the multi-seed pod trait in peanut.

2.Materials and methods

2.1.Experimental materials

A RIL population of 248 lines was developed from a cross between female parent Silihong(A.fastigiata)and male parent Jinonghei 3(A.hypogaea).Silihong,a local cultivar in northeastern China is an early maturing cultivar and most of its pods are multi-seeded.Jinonghei 3,an improved black peanut,is a mid-late maturing cultivar and most of its pods are two-seeded.These parents also differ in pod shape and size,seed shape,size,and seedcoat color and flowering time(Fig.1).

2.2.Field trials and trait measurement

Field trials were carried out at Baoding and Handan in Hebei province from 2012 to 2017.The specific trial environments were as follows:E1 at Baoding 2012,E2 at Handan 2012,E3 at Baoding 2013,E4 at Handan 2013,E5 at Baoding 2014,E6 at Baoding 2015,E7 at Baoding 2016,and E8 at Baoding 2017.Materials were planted in complete random block design with two repetitions in each environment.Each RIL and parent was planted in a 1 m row with an interplant spacing of 10 cm within rows that were 0.35 m apart.Protection lines were set on every side and crop management followed local practices.Four plants from the middle of each line were sampled at maturity,and the pods of each plant were collected and dried.Total pod number(TPN)and number of pods with≥3 seeds on each plant were recorded.In addition,the number of 1-seed pod(1-SPN)and 2-seed pod(2-SPN),pod yield per plant(PY)were also investigated in 2017.Pods with≥3 seeds were named as multi-seed pods.The ratio of multi-seed pods per plant(RMSP)was a ratio of number of multi-seed pod to total pods per plant.

Fig.1-Seed pods,size and color of parents Silihong and Jinonghei 3.

2.3.Data statistics and analysis

Tests of normality of phenotypic data and correlation analyses were analyzed with SPSS 17.0 software,and analysis of variance(ANOVA)and broad-sense heritability(H2)were calculated by SAS 9.2 software.RMSP was calculated by:

where,each value was the mean of 4 plants.

Broad-sense heritability[29]was calculated as:

where,H2=broad-sense heritability;=genotypic variance;=variance of genotype×environment interaction;=error variance;e=environment number;and r=number of replications.

A mixed linear model was constructed with the lmer function in lme4 package of R software[30](http://www.R-project.org/)to calculate BLUP values of RMSP in eight environments.

BLUP values were calculated by:

where,yiwas the phenotypic value of i-th line;μ was the mean phenotypic value of the RIL in all environments;fiwas the genotypic effect;eiwas the environmental effect;and εiwas the random error.The mean phenotypic value was a fixed effect,and the genotypic and environmental effects were random effect.

2.4.QTL mapping analysis

Based on studies by Zhou[31]and Li[32]in our lab,we increased the density of molecular markers,optimized grouping parameters and constructed a new genetic linkage map with JoinMap 4.0 software[33].The linkage map contained 226 SSR markers in 30 linkage groups,and was 1511.32 cM in length with an average distance between adjacent markers of 6.69 cM.

The inclusive composite interval mapping(ICIM)method in QTL IciMapping V4.1 software[34,35]was used for scanning the phenotypic value and BLUP value of RMSP in 8 environments.In order to judge the existence of an authentic QTL and get more information of QTL,the two threshold methods,LOD≥2.5 and 1000 permutation test(PT),were each calculated.The specific parameters were set as:for detecting additive QTL,two threshold methods were used,Step was 1.0 cM,PIN was 0.001,Type I error was 0.05;for detecting epistatic QTL,the PT method was used,Times was 1000,Step was 5.0 cM,PIN was 0.0001,Type I error was 0.05.In individual environment analyses,the joint mapping analysis under multi-environment and epistasis analysis by the BIP and MET functional module of QTL IciMapping V4.1 software[34,36],the LOD threshold values of permutation tests were 3.13,6.61,and 9.25,respectively.These were used to estimate the interaction effects between additive/epistatic QTL and environments.QTL were named as:q+the abbreviated English name for the trait+linkage group no.,or named as q+the abbreviated English name+linkage group no.+an ordered number designating one of multiple QTL in a single linkage group.For example,qRMSP6.2 indicates the second QTL related to RMSP in the sixth linkage group.

3.Results

3.1.RMSP in parents and RIL population

The descriptive statistical results indicated that there were significant differences(P＜0.01)for RMSP between the parents,with that of Silihong being higher than Jinonghei 3(Table 1).Phenotypic variation in RMSP among RILs was very large,and transgressive segregation indicated that the multiseed pod trait was quantitatively inherited.The absolute values of skewness and kurtosis of RMSP in each environment were＜1(Fig.2).

ANOVA showed the variation in RMSP between lines,environments,and genotype×environment interaction were all significantly different(P＜0.01)(Table 2).Broadsense heritability was 88.3%,indicating that genotypic variance accounted for a high proportion of the phenotypic variation in the multi-seed pod trait,and the trait should be selectable in the early generation.However,the trait was also affected by environment.

Correlation analysis indicated that pod yield per plant was significantly and positively correlated with 2-SPN,≥3-SPN,TPN,and RMSP,with correlation coefficients ranging from 0.113 to 0.189.RMSP was negatively correlation with 1-SPN,2-SPN,and TPN,with correlation coefficients ranging from-0.089 to-0.340,and RMSP was positively correlated with≥3-SPN,with a correlation coefficient of 0.858.TPN was highly correlation with 1-SPN,2-SPN,and≥3-SPN,with correlation coefficients of 0.244-0.815(Table 3).This showed that yield factor had a close association with pod-seed trait,and these quantitative traits are mutually restrictive and influenced.

3.2.QTL mapping analysis of phenotypic data in individual environments and BLUP values

Using the ICIM-ADD mapping method in the BIP module,the phenotypic value of RMSP in each environment was used for QTL mapping.Seven QTL related to the multi-seed pod trait and distributed on the linkage groups 1,6,9,14,19(2),and 21,respectively.And they explained 49.51% of the overall phenotypic variation,with each one explaining 4.42%-11.51%(Table 4,Fig.3).

QTL qRMSP19.1 located in the linkage group 19 and detected in environments E1 and E4,explained 7.21%-11.51%of the phenotypic variation with additive effects ranging from 3.04 to 3.60.QTL qRMSP19.1 flanked by GM1988 and SATE0405 was detected by BLUP and individual environment analyses,and explained 4.93% of the total phenotypic variation.Another QTL,qRMSP19.2,in linkage group 19 detected in environments E6 and E7 was mapped in the interval seq9A7_2-SATE0620 with LOD values of 2.64 and 2.73,respectively.It explained 5.15%-6.38%of the phenotypic variation with additive effects of 3.22 and 4.25.QTL qRMSP6.1 in interval AHGS1354-AhTE0628 in linkage group 6 was detected in E1 with a LOD score of 5.50;it explained 10.61%of the phenotypic variation and had an additive effect of-3.43.qRMSP1.1 in interval AHGS2130-Ai120H20 in linkage group 1,was detected in E7 with a LOD score of 2.51;it explained 5.70%of the phenotypic variation,and had an additive effect of 4.22.In addition,three QTL,qRMSP9,qRMSP14.1,and qRMSP21.1,were detected on E1(Fig.3).Using the RIL population,we detected 7 QTL related to the multi-seed pod trait,among which the favorable alleles of qRMSP1.1,qRMSP9,qRMSP19.1,qRMSP19.2,and qRMSP21.1,were from Silihong.

3.3.Joint analysis for the multi-seed pod trait in multiple environments

Using the MET module in QTL IciMapping V4.1 software,we analyzed the phenotypic values of RMSP for the RIL population grown in 8 environments.Seventeen additive QTL associated with multi-seeded pods were identified with LOD values ranging from 3.73 to 9.76(Table 5),and these QTL explain 22.99%of the phenotypic variation in the trait.Three of them,qRMSP1.2,qRMSP19.1,and qRMSP20.1,were significant QTL comparing with LOD threshold value(6.61)from permutation tests,and they explained 6.61%of the phenotypic variation.The contribution rates of interaction among 17 additive QTL and the environment ranged from 0.11%to 1.17%.

More QTL were detected by the joint mapping method compared with those based on individual environments.The contribution of additive QTL effects in the joint analysis was lower than that in individual environments.The QTL detected in individual environment analyses(qRMSP1.1,qRMSP6.1,qRMSP9,qRMSP19.1,qRMSP19.2,and qRMSP21.1)were all mapped to the same or adjacent marker intervals to those detected by the joint analysis.However,the additive contribution rates of qRMSP9 and qRMSP21.1 were lower than those between the additive effect and environment interaction,indicating that the additive effects of qRMSP9 and qRMSP21.1 were not the main contributors to phenotypic variation.QTL qRMSP1.1 was detected in only one environment(E7)with a lower LOD value,further indicating that these QTL environmentally specific and may not provide stable control of the multi-seed pod trait.

Table 1-Phenotypic variation of RMSP in parents and RIL population across eight environments.

Fig.2-Frequency distributions of RMSP in the RIL population in eight environments.

Table 2-Analysis of variance(ANOVA)for RMSP in the RIL population grown in multiple environments.

3.4.Epistatic QTL by environment interactions for multienvironment trials

Using the MET module of QTL IciMapping V4.1 software,the interaction effects among EPI QTL and environment in each of the eight environments were analyzed by the ICIM-EPI mapping method.As shown in Table 6,43 epistatic QTL were detected by permutation tests of 1000 replications and scanning steps of 5.0 cM,covering 24 of the 30 linkage groups except for linkage groups 5,9,12,22,25,and 28.A total of 57 loci were involved,and among them,38 loci had only one interacting QTL,others interacted with more than one QTL(Fig.4).Moreover,there were 7,5,4,and 4 loci involved in QTL interaction on linkage groups 6,8,17,and 18.The epistatic effects(AA)of 43 QTL ranged from-2.78 to 2.08,and they collectively explained 66.53%of the phenotypic variation in the multi-seed pod trait,with each one explaining 0.56%-2.70%.

Thirty nine QTL showed epistatic interaction between different linkage groups,whereas another 4 QTL showed epistatic interaction loci within a single linkage group.Three of the 39 QTL mapped within the AHGS1296-AHGS1127 interval in linkage group 18,AhTE0536-AHGS1487 in linkage group 20,and TC4F10-TC1G04 in linkage group 23 interacted with the ARS702-seq19D09 interval in linkage group 30,but their epistatic effects(AA)were different,thus indicating that the epistatic effects of one locus on others were different.The interaction effects between these QTL and environment were also quite different.

Table 3-Correlation between yield and pod-seed trait in the peanut RIL population.

4.Discussion

4.1.The phenotypic variation and genetic characteristics of RMSP in peanut

Many factors affect the accuracy of QTL mapping,especially genetic background and environment,although QTL mapping has become one of the common methods of genetic study of quantitative traits.The present 248 RIL population was F12.It was planted at 8 environments from 2012 to 2017.The parents of the RIL population had significant differences in RMSP,seed number per pod and seed size,except for the advantage of maturity and environmental adaptability.The analysis of phenotypic variation showed that the RMSP of female parent Silihong was higher than that of the male parent Jinonghei 3.Meanwhile,the phenotypic variation of multi-seed pod in the RIL population was extremely rich,and the transgressive segregations were observed in some lines,their RMSP was higher than that of parent Silihong.And ANOVA indicated that the multi-seed pod existed abundant genetic variation in population and had strong interaction with environment,which indicating multi-seed pod was influenced by environment easily.The result was supported by the studies of Yang[37]and Wang et al.[38].As the two parents of RILs had totally different phenotypic traits and the variation of population was very wide,it was suitable materials for multi-seed pod research in cultivated peanut.

4.2.The contribution of multi-seed pod to yield per plant in peanut

It is generally agreed that seed number per pod is an important breeding target trait for pod-bearing oil crops.Various studies indicated that seed number per pod is a complex quantitative trait and the separation of which was continuously changing in genetic populations[37].Although many studies on the pod-seed trait in oil crops,such as soybeans[39,40]and rapeseed[41,42]have led to valuable knowledge,fewer studies have been undertaken with cultivated peanut.

In this study,we analyzed the correlation between yield per plant and pod-seed traits in a RIL population.The results demonstrated that pod yield per plant(PY)was positively correlated with numbers of pods with≥3-SPN,TPN and RMSP.Our results were consistent with those of Liang et al.[39].This means that yield per plant can be effectively improved by increasing the number of multiseed pods,TPN and RMSP,The results will have important theoretical guiding for the selection and determination of high-yield breeding indexes in peanut.Moreover,the number of multi-seed pods(≥3-SPN)had an extreme significant negative correlation with 1-SPN and 2-SPN,this indicated that the development of seeds was complicatedly regulated by both genetic and environment factors in peanut,and the formation mechanism of 1-seed and 2-seed pods was different from that of multi-seed pod.They might be controlled by different QTL/gene.

Table 4-QTL associated with the multi-seed pod trait detected in individual environments.

4.3.Application of BLUP in QTL mapping for multi-seed pod of peanut

The BLUP method of analysis first proposed for animal breeding by Henderson[43]combines the selection index and least square methods. It considers fixed environmental and random genetic effects at the same time.Thus,the accuracy of the prediction of BLUP value is improved with prediction of BLUP values in different years,different test locations and individuals of different generations[44,45].This method is now widely used in genome wide association studies(GWAS),genome selection(GS),and QTL mapping in maize and rice[46-48]as well as other crops.

Fig.3-QTL of RMSP detected with RIL population derived from Silihong×Jinonghei 3.

Table 5-Additive QTL×the environment interaction effects for the multi-seed pod in RIL population.

In this study,the BLUP values of RMSP in peanut across 6 years and 8 environments were used for mapping QTL.It effectively reduced the impact of environmental and locational differences.QTL qRMSP19.1 mapped in the marker interval GM1988-SATE0405 in linkage group 19 was detected using BLUP data,explaining 4.93%of the phenotypic variation.However,the same QTL with high PVE values(＞10%)was detected in two other environments indicating that this QTL is authentic and should be a focus for further in-depth study of the multi-seed pod trait in peanut.

4.4.The impact of QTL×environment interaction on QTL mapping

It is generally accepted that a QTL with a phenotypic contribution of＞10%of trait value and repeatedly detected in individual environments can be designated as the main effect QTL.Large LOD values are also indicative of higher accuracy.In the present,qRMSP19.1 and qRMSP19.2 were detected in more than one environment.qRMSP19.1 had a PVE value of＞10%and was also detected using the BLUP value for multiple environments,so be considered a major QTL for multi-seed pod trait.

Both environment and the epistasis interaction among QTL have effects on gene expression[49].In our study,17 additive QTL were detected in joint multi-environmental analyses.By comparing phenotypic contribution rates it is clear that the contributions of QTL identified by multi-environmental analysis were much less than those contributed by individual environmental analyses.It was due to both QTL additive and QTL×environment interaction effects were considered in the multi-environmental joint analysis,whereas the individual environmental analysis method only estimates the additive effects of QTL.Moreover,some QTL detected by individual environmental analysis were also identified by multi-environmental joint analysis and they had certain interaction effect with environment.However,the effects of four QTL×environment interactions made lower phenotypic contributions than the additive effects.This showed that the additive effects of these QTL had a significant influence on phenotypic variation of the multi-seed pod trait.We also detected two environmentally specific QTL with low LOD values.Boer et al.[50]suggested that interaction between QTL and environment can be considered as a specific expression of QTL caused by a range of factors such as year,location and temperature.Therefore,joint analysis over multiple years and locations can be used to establish the stability of QTL and to estimate interaction between additive QTL and environment.

Richey[51]proposed that epistasis was the interaction between multiple gene loci.Previous studies confirmed that epistatic effect have an important influence on the inheritance of crop yield-related traits and the occurrence of heterosis[52,53].In this study,43 EPI QTL were detected and 57 loci were involved in the interaction.Most of these QTL belonged to epistatic interaction effect loci in different linkage groups and several QTL were involved in a complex interaction network.In addition,marker intervals close to QTL qRMSP1.1,qRMSP6.1,qRMSP19.2,and qRMSP21.1 were involved in epistatic interaction.Moreover,interaction effects between epistatic QTL and environment were also found,but the phenotypic contribution rates of QTL interaction(AA)were higher than the interaction(AAE)between epistasis and environment.Our results indicated that the additive and epistatic effects were both important factors influencing the genetic variation in the multi-seed pod trait in peanut.However,three significant additive QTL and 43 epistatic QTL only explained 73.14%of the phenotypic variation in the multi-seed pod trait,a value lower than the broad-sense heritability(88.3%),thus showing that the environment also had significant effects on the phenotypic variation.

Table 6-Epistatic QTL byenvironment interaction effects for the multi-seedpod trait inthe RIL population.

Fig.4-A cyclic representation of significant epistatic QTL.

5.Conclusions

The broad-sense heritability of the multi-seed pod trait was 88.3%,indicating that it was controlled mainly by genetic factors,and it was suitable for selection at an early segregating stage.RMSP was highly correlated with pod yield per plant.Using phenotypic and BLUP values of RMSP for QTL mapping,we detected 7 QTL associated with the multi-seed pod trait in individual environments.Among them,qRMSP19.1 and qRMSP19.2 were detected in multiple environments.In addition,17 additive QTL were detected by the multiple environmental joint analysis,and 43 epistatic QTL were detected by ICIM-EPI mapping in the MET module.The epistatic QTL had two modes of action:one was epistatic interaction between two QTL in the same linkage group,and the other was epistatic interaction between two QTL in different linkage groups.The multi-seed pod trait was mainly controlled by minor effect genes,and the epistatic effects were more important than additive effects.The QTL detected in multiple environments can be considered as candidate loci for further studies on the multi-seed pod trait in peanut.

Acknowledgments

This study was supported by the China Agriculture Research System(CARS-13),the National Natural Science Foundation of China(31771833),the Hebei Province Science and Technology Support Program(16226301D),and Key Projects of Science and Technology Research in Higher Education Institution of Hebei province(ZD2015056).