Multi-environment QTL mapping of crown root traits in a maize RIL population

Pengcheng Li, Yingying Fn, Shungyi Yin, Yunyun Wng, Houmio Wng,Yng Xu, Zefeng Yng,c,, Chenwu Xu,c,

aJiangsu Key Laboratory of Crop Genetics and Physiology/Key Laboratory of Plant Functional Genomics of the Ministry of Education/Jiangsu Key Laboratory of Crop Genomics and Molecular Breeding, Agricultural College of Yangzhou University, Yangzhou 225009, Jiangsu, China

bJiangsu Co-Innovation Center for Modern Production Technology of Grain Crops, Yangzhou University, Yangzhou 225009, Jiangsu, China

cJoint International Research Laboratory of Agriculture and Agri-Product Safety of Ministry of Education of China, Yangzhou University,Yangzhou 225009, Jiangsu, China

ABSTRACT Crown root traits,including crown root angle(CRA),diameter(CRD),and number(CRN),are major determining factors of root system architecture,which influences crop production.In maize, the genetic mechanisms determining crown root traits in the field are largely unknown. CRA, CRD, and CRN were evaluated in a recombinant inbred line population in three field trials. High phenotypic variation was observed for crown root traits, and all measured traits showed significant genotype-environment interactions. Singleenvironment (SEA) and multi-environment (MEA) quantitative trait locus (QTL) analyses were conducted for CRA, CRD, and CRN. Of 46 QTL detected by SEA, most explained less than 10% of the phenotypic variation, indicating that a large number of minor-effect QTL contributed to the genetic component of these traits.MEA detected 25 QTL associated with CRA, CRD,and CRN,and 2 and 1 QTL were identified with significant QTL-by-environment interaction effects for CRA and CRD, respectively. A total of 26.1% (12/46) of the QTL identified by SEA were also detected by MEA, with many being detected in more than one environment. These findings contribute to our understanding of the phenotypic and genotypic patterns of crown root traits in different environments. The identified environment-specific QTL and stable QTL may be used to improve root traits in maize breeding.

1. Introduction

Water and nutrient deficiencies are primary limitations to plant productivity, severely reducing yields globally and threatening global food security [1]. In the U.S. alone,approximately 67% of crop losses over the last 50 years have resulted from drought [2]. Roots mediate the uptake of water and nutrients and anchor plants. An optimal root system architecture (RSA) supports soil resource use and yield; for example,more roots distributed in deeper soil layers can help crops use the water and nitrate found there [3-6]. Plant breeders are starting to focus on roots in their efforts to produce crops with better yields; thus, roots are the key to a second green revolution [7]. The major challenge for root breeding is poor understanding of the genetic basis of root development.

Maize RSA refers to the shape and spatial arrangement of root tissue in the soil [8,9] and comprises several traits and parameters including root number, length, angle, and diameter.Crown roots form the backbone of the maize root system[10].An optimal crown root number(CRN)in maize promotes deep soil exploration and resource acquisition under drought conditions and suboptimal availability of mobile nutrients[5].Thus,CRN is a trait that can potentially be used for the genetic improvement of nitrogen acquisition from low-nitrogen soils[11].Crown root angle(CRA)influences vertical and horizontal root distributions in the soil and is associated with rooting depth[12].A deeper root system helps crops extract water and nitrogen from deeper soil layers to increase yield [5,6,13].Because of the higher phosphorus availability in topsoil, a shallow basal root growth angle is important for phosphorus acquisition in common bean [14]. Several overlapping quantitative trait loci(QTL)region in sorghum showed correlations between CRA and yield [15]. Root diameter, correlated with anatomical phenes such as cortical thickness and stele diameter, was correlated with root penetration capacity and biomechanical properties [16]. Thicker roots more readily penetrate hard soils to use deeper soil resources [17]. Root biomechanical properties influence root bending strength and adventitious root angle, plant anchorage, and lodging resistance [18]. Large diameters and steep growth angles work together to increase root penetration into hard soils [16].Identifying QTL or genes associated with CRN, CRA, and CRD will be needed for future root breeding aimed at improving crop performance under diverse soil conditions.

Genes/QTL for crown root traits have been identified in biparental populations in cereal crops.Cai et al.[19]identified five QTL for CRN at three maize developmental stages,including one consensus QTL on chromosome bin 10.4. Ku et al.[20]identified two coincident major QTL for total brace root tier number using a set of recombinant inbred lines(RILs)and an immortalized F2population derived from these RILs, and the largest additive effect was 16.4%-17.9%. For root angle,QTL mapping has been conducted in maize, rice, and sorghum. In two maize-teosinte F2populations, 10 rootangle QTL were identified, and the QTL on chromosome 7 was found consistently in all populations [21]. Six major QTL for root angle (DRO1, DRO2, DRO3, DRO4, DRO5, and qSOR1)were confirmed in rice [22], and DRO1 was the first cloned gene associated with root angle that increased drought avoidance in rice[6].Four root-angle QTL have been identified in sorghum, and an individual QTL explained 29.78% of phenotypic variation [15]. In rice, Courtois et al. [23] summarized 675 root QTL detected in 12 populations for 29 root traits and found 123 QTL for root thickness; however, no QTL associated with root diameter have been cloned. A QTL controlling root thickness and root length has been placed in an 11.5-kb region by fine mapping [24]. Overexpressing OsNAC5 and OsNAC9 increased root diameter, drought tolerance, and grain yield under field conditions [25,26]. Few genetic studies of crown root traits in maize, especially in the field,have been reported.

Plant root architecture shows high plasticity in response to environmental factors such as drought and nutrient availability [27]. Water deficit strongly suppresses the development of crown roots, and this response is widely conserved across grass species [28]. CRA becomes steeper to allow roots to capture nitrogen in deep soil layers under low-nitrogen conditions [12]. To improve root traits across different environments, it is necessary to characterize genotype-byenvironment interaction (GEI). Previous studies [29,30] extended QTL mapping to the detection of QTL-by-environment interaction (QEI) of root traits. However, little information is available on multi-environment QTL mapping of crown root traits in maize,especially in the field.

To investigate the genetic basis of crown root traits in maize, we characterized CRA, CRD, and CRN in a maize RIL population in three field trials and performed single- and multi-environment QTL mapping using a high-density linkage map.Our objectives were to evaluate phenotypic variation in crown root traits in several field environments and to identify QTL and QEI for crown root traits.

2. Materials and methods

2.1. Plant materials, field experiments, and root system evaluation

A set of 208 RILs derived from two maize inbred lines,T877 and DH1M,were used as described by Yin et al.[31].Field trials were conducted in three environments: Sanya (N18°23′,E109°44′) in 2015 (15SY) and 2016 (16SY), and Yangzhou(N32°22′, E119°16′) in 2017 (17YZ). In each environment, the population was evaluated in a completely randomized block design of one-row plots with two replications. Each row was 3 m long and 0.5 m wide and contained 11 plants.Shovelomics[32,33] was performed at maturity. Six plants with uniform growth were randomly selected from one plot, and roots of each plant were excavated with shovels by removal of a soil cylinder 40 cm in diameter and 25 cm deep. The excavated root crowns were shaken briefly to remove most of the soil adhering to the crown, and the root layer could be clearly distinguished. Three roots on the first layer of the nodal root situated flush with the soil surface were selected for measurement of CRA,CRN,and CRD[33].CRA was measured with a protractor as degrees from horizontal:horizontal roots were assigned a CRA of 0° and vertical roots a CRA of 90° (Fig. S1).Selected crown roots were cut off at 1 cm from the top of the root and a vernier caliper was used to measure CRD.

2.2. Genotyping and construction of genetic linkage maps

A high-density linkage map has been constructed for the T877 × DH1M population by single-nucleotide polymorphism(SNP) genotyping using the Maize56K SNP Array [31] and contains 56,110 SNPs covering the entire maize genome.Briefly, SNPs polymorphic between two parental lines were allocated to 3227 bin markers,which were then ordered using the ripple function in the qtl package [34]. Genetic distances between bin markers were calculated using the Kosambi function [35]. The numbers of bin markers per chromosome varied from 111 to 503,and the total length of the linkage map was 2450 cM,with a mean genetic distance between adjacent markers of 0.76 cM.

2.3. Data analyses

Phenotypic data analyses, including descriptive statistics(mean, range, standard deviation, skewness, and kurtosis)and Pearson correlations, were calculated using IBM SPSS Statistics 21.0. The “l(fā)me4” package in R was used to estimate genotypic(σ2G),G-E interaction(σ2GE)and error variances(σ2e).The broad-sense heritability (h2) of each measured trait was calculated following Hallauer et al. [36]. Single-environment QTL analysis(SEA)was performed with qtl package in R[34]by composite interval mapping(CIM)with a 1-cM step length and a 10-cM window size. The phenotype in environments 15SY,16SY, and 17YZ, and the simple mean values across environments, were used for the SEA. Colocalized QTL separated in SEA by a distance of less than 10 cM were defined as QTL.Multi-environment QTL mapping (MEA) was performed with the MET (multi-environment trial) functionality in QTL IciMapping 4.0[37,38].In view of the complexity of root traits,a suggestive LOD threshold value of 2.5 was used to avoid ignoring minor-effect loci, following previous reports [19,39].If a QTL identified by SEA shared the same marker with one identified by MEA-QTL, it was assigned as having been detected by both SEA and MEA. Boxplots, correlation diagrams,LOD curves,and QTL profiles in three environments were drawn using corrplot and the ggplot2 package in R. The QTL nomenclature is as follows: environment (subscript character)+trait(capital letter)+chromosome+QTL number.

3. Results

3.1.Crown root phenotypic assessment

The values of the three crown root traits(CRA,CRN,and CRD)differed significantly between the two parental lines except for CRN in 2015SY (Fig. 1). Compared to DH1M, T877 showed larger CRA in 15SY and 16SY but slightly smaller CRA in 17 YZ.DH1M showed larger CRD and CRN than T877 in 15 SY and 17YZ, whereas smaller CRD and CRN in 16SY (Fig. 1; Table 1).On average,DH1M showed a larger CRD,a greater CRN,and a smaller CRA than T877.

Fig.1-Phenotypic differences between the two parental maize lines DH1M and T877 in different environments.Traits differing(* P ＜0.05) between the two lines are marked with asterisks.

In the RIL population, lines in 17YZ showed larger CRD,greater CRN, and smaller CRA than those in other environments, and wide phenotypic variation for the three crown root traits in all environments was observed (Table 1).Coefficients of variation (CV) ranged from 7.38% to 12.09%,with the greatest phenotypic variation occurring in 17YZ,with a CV that ranged from 13.72% to 16.03%.The crown root traits showed approximately normal distributions in all environments(Fig.2;Table S1).The effects of genotype,environment and GEI were significant at P ＜0.01 for all traits (Table 1),indicating that crown roots were highly sensitive to environment. The h2values were moderate, varying from 46.8% to 63.9% (Table 1). Only CRD was positively correlated with CRN(r =0.50,P ＜0.001)(Fig.2).

3.2. Single-environment QTL analysis of maize crown root traits

Table 1-Descriptive statistics and ANOVA for crown root traits in three environments.

Fig.2-Pearson correlations among crown root angle,diameter,and number.Red,green,and blue areas represent phenotypic distributions in the 15SY,16SY,and 17YZ environments,respectively.***,significant at P ＜0.001.

SEA identified 46 putative QTL on all chromosomes,including 12, 17, and 17 QTL associated with CRA, CRD, and CRN,respectively (Table 2; Fig. 3). These QTL explained 1.01%-10.60% of the phenotypic variation. Of the identified QTL, 29 and 17 carried favorable alleles from T877 and DH1M,respectively. For CRA, respectively 3, 4, 2, and 3 QTL were detected in 15SY,16SY,17YZ,and the averaged value over the environments. The proportion of phenotypic variation explained by these individual QTL ranged from 2.03% to 10.60%,and QTL-15CRA1 was the largest QTL for CRA.Respectively 4,4,4,and 5 QTL were identified for CRD 15SY,16SY,17YZ,and the averaged value over the environments and each explained 1.38%-8.61% of the phenotypic variation. For CRN, respectively 5,4,3,and 5 QTL were detected in 15SY,16SY,17YZ,and the average of the three environments. The proportions of phenotypic variation explained by these individual QTL ranged from 1.01% to 8.27% (Table 2; Fig. 3-C). Several QTL were detected in multiple environments: QTL-16CRA1 was colocalized with17CRA1 on chromosome 1, and QTL-15CRA3 was also identified in the averaged value over the environments (avCRA3). Three colocalized QTL,16CRD4,15CRD6, and15CRD9, were detected for CRD, and were detected the averaged value over the environments (avCRD4,avCRD6, andavCRD9, respectively). QTL-15CRN3 was colocalized with17CRN3 on chromosome 3,and QTL-17CRN6 was also identified the averaged value over the environments (avCRN6). CRN was significantly correlated with CRD. Pairs of QTL for CRN and CRD were co-localized on chromosomes 8 (201.0-205.0 cM)and 10(158.0-162.8 cM)(Table 2).

Table 2-QTL for root traits detected in three environments.

3.3. Multi-environment QTL analysis of maize crown root traits

In the MEA,LOD,LODA,and LODAEwere defined as LOD scores for detecting respectively QTL with both average effect of the putative QTL across the environments and QEI effects, QTL with only average effects, and QTL with only QEI effects.Profiles of LOD,LODAand LODAEalong the maize genome are shown in Fig. 4. A total of 25 QTL affecting crown root traits were identified in the three environments (Table 3). These QTL explained 2.24%-10.53% of the phenotypic variation.Eight QTL were detected for CRA and explained 2.24%-10.53% of the phenotypic variation. Two CRA-QTL (MECRA1 andMECRA7) with significant QEI were detected on chromosomes 1 and 7, respectively, and the QEI explained respectively 4.15% and 4.80% of phenotypic variation. Four QTL,MECRA3-2,MECRA4,MECRA8‐1, andMECRA8-2 showed strong QEI with LODAEvalues greater than their LODAvalues,indicating that the QEI effect dominated the phenotypic variation. Nine putative QTL for CRD explained 2.54%-6.25% of phenotypic variation, and onlyMECRD4-1, on chromosome 4, showed a significant QEI. Four QTL,MECRD2,MECRD3,MECRD5, andMECRD10, also showed strong QEI. Eight QTL affecting CRN were identified:four on chromosome 1 and one each on chromosomes 4, 5, 6, and 10, and explained 2.50%-6.52% of phenotypic variation.No QEI were detected for CRNQTL, and only one QTL-MECRN1-1 showed strong QEI. Twelve QTL detected by QEI mapping were also detected by SEA,including 4 QTL each for CRA,CRD,and CRN(Fig.5;Table S2).

Fig.3- Single-environment QTL analysis of crown root traits in the RIL population.(A)LOD curves under different environmental conditions.(B)Additive effects of individual QTL.A positive value indicates that T877 and a negative value,that DH1M carried the allele responsible for an increase in the trait.(C)Phenotypic variation explained by individual QTL.

Fig.4- LOD curves based on multi-environment QTL analysis.LODA indicates average effect of the putative QTL across the environments at the testing position; LODAE indicates a QEI effect.

Table 3-Multi-environment QTL mapping of crown root traits in RIL population.

4. Discussion

Maize embryonic roots, including primary and seminal roots,are essential for the establishment of seedlings after germination.In contrast,a postembryonic component,crown roots,initiated from consecutive underground nodes of the stem,make up most of the maize root system and are primarily responsible for soil resource acquisition later in development[40].Because it is difficult and laborious to evaluate root traits directly in the field, most root genetic studies are performed in laboratory experiments, such as paper roll systems,hydroponics, and pot experiments at the seedling stage[39,41,42]. These systems allow rapid, accurate and highthroughput analysis of root traits at an early growth stage;however, they are not effective for evaluating natural root architecture, especially crown roots, in field experiments.Evaluations of RSA in field-grown plants may reveal the actual root growth pattern in an agriculturally relevant context [43].Currently, new methods are being developed to overcome low-resolution and low-throughput approaches for RSA phenotyping in the field [44]. Shovelomics [32] has allowed researchers to score ten root traits of an adult maize plant in the field in a few minutes. Automatic imaging approaches,such as DIRT (Digital imaging of root traits) and REST (Root Estimator for Shovelomics Traits), have further increased throughput [45]. In the present study, Shovelomics was used to evaluate the CRA,CRD,and CRN in a maize RIL population in three field trials at maturity stage and allowed the observation of a wide range of phenotypic variation. Approximately 1.54-, 1.64-, and 2.04-fold differences were observed for CRA, CRD, and CRN (Table 1; Fig. 1). Trachsel et al. [12]reported that CRA did not vary between different developmental harvest stages, but that significant genetic variation was observed between genotypes. A 1.20-fold difference for CRD was observed between 26 maize genotypes in a temperature-controlled growth chamber [16], and in the IBM RIL population,the difference of CRD was more than 3.80-fold[41].At three developmental stages,CRN showed 1.85-to 2.06-fold variation[19].The relatively high phenotypic variation in the present study indicates that this population is suitable for studying the genetic basis of crown root traits. A significant correlation was observed between CRD and CRN. The two pairs of QTL for CRN and CRD on chromosomes 8 and 10 can be used as target loci to improve CRN and CRD simultaneously.

Fig.5- QTL detected by single-environment(SEA)and multi-environment QTL mapping(MEA). The first column to right of a chromosome bar shows QTL identified by MEA and the second column shows QTL identified by SEA.Red asterisk in the chromosome bar indicates QTL detected by both SEA and MEA.

Crown roots of maize initiate at 10 days after germination,and successive whorls of nodal roots follow an approximately S-shaped growth curve, finally approaching the maximum root number at the end of the grain-filling stage [33,46].Comparisons of root traits of different whorls showed that CRN and CRD were most sensitive to node position and that CRA had the least variation [47]. Different supply levels of nitrogen in the soil had a significant impact on these crown traits [47]. These traits also showed significant variation under different water regimes, and CRA and CRD showed significant GEI effects [30]. Root systems are complex and dynamic and show high plasticity in response to environment; the heritability values were 57.3, 46.8, and 63.9 for CRA, CRD, and CRN,respectively (Table 1). The values were similar to those of other root traits in a field experiment [19]. A moderate heritability close to 0.5 suggests that some phenotypic variation of crown root traits was affected by the environment. Comparisons of crown root traits between the two parental lines in the three environments also showed that the phenotypes of the two lines differed with environment, and significant environment and GEI effects were observed for all the investigated traits. Thus,environmental variation strongly influences crown roots.

The finding that most QTL explained less than 10% of phenotypic variation, indicates [19,30,41] that a large number of minor-effect QTL contribute to the genetic component of these crown root traits in maize. No QTL was identified in all three environments, whereas seven pairs of QTL were detected in multiple environments (Table 2, Fig. 5). Some QTL identified in this study lie near those identified in previous studies (Table S2). For example, QTL-avCRN10 on chromosome 10 (85.3 cM) was located in bin 10.04, a hotspot for root QTL. Three QTL for total CRN (qARN110-1, qARN210-1, and qARN310-1), covering all developmental stages, were located in a similar genomic region[19]. This region also contained QTL for total root length, root dry weight, and vertical root pulling resistance [19,48]. QEI in crops is widespread; to improve complex traits across environmental gradients, it is necessary to explicitly analyze GEI [49]. A total of 25 QTL affecting crown root traits were identified in this study by MEA, including two for CRA and one for CRD that had significant QEI effects. Complex traits such as root traits often show low heritability and large GEI [49]; SEA and MEA can assess both QTL stability and QEI effect. We found that 26.1% (12/46) of the QTL identified by SEA could also be detected by MEA,especially QTL detected in more than one environment (3/7).Furthermore, the QTL effects were similar [37].

Despite the importance of roots, direct selection for optimal RSA in the field is not routine in maize breeding programs [44].However, over the past century, the evolution of maize root phenotypes has been consistent with increasing yield [47]. The CRA of the newest US maize was 7° shallower than the oldest material, and the CRN included at least 1.6 fewer nodal roots than older lines [47]. Although there is no easy method to select an optimal root system in the field, modifying RSA by markerassisted selection (MAS) could contribute to improving of desirable agronomic traits, such as drought tolerance and nutrient efficiency [6,50,51]. The environment-specific QTL and stable QTL identified in the present study may be used to improve root traits in maize breeding.

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2019.12.006.

Declaration of competing interest

Authors declare that they have no conflict of interest.

Acknowledgments

This work was supported by the National Key Research and Development Program of China (2016YFD0100303), the National Natural Science Foundation of China (31972487,31601810, and 31902101), the Natural Science Foundation of Jiangsu Province (BK20180920), and the Priority Academic Program Development of Jiangsu Higher Education Institutions(PAPD).

Author contributions

PCL, ZFY and CWX conceived and designed the study. PCL,YYF,SYY,and YYW conducted experiments.PCL,YYF,and YX analyzed data. PCL, HMW, ZFY, and CWX wrote the manuscript.All authors read and approved the manuscript.