Surveying the genomic landscape of silage-quality traits in maize(Zea mays L.)

Jtin Shrm, Shuhm Shrm, Krishn Si Krntm, Om Prksh Rigr, Chynik Lhkr,Dinesh Kumr Sini,2, Sushil Kumr, All Singh, Ahijit Kumr Ds, Priti Shrm, Rmesh Kumr,*

a Department of Plant Breeding and Genetics, Punjab Agricultural University, Ludhiana, Punjab 141004, India

b School of Agricultural Biotechnology, Punjab Agricultural University, Ludhiana, Punjab 141004, India

c ICAR-Indian Institute of Maize Research, Ludhiana, Punjab 141004, India

Keywords:Meta-QTL Silage quality Maize GWAS Candidate genes

ABSTRACT Despite the longstanding importance of silage as a critical feed source for ruminants,its quality improvement has been largely overlooked.Although numerous quantitative trait loci (QTL) and genes affecting silage quality in maize have been reported, only a few have been effectively incorporated into breeding programs.Addressing this gap,the present study undertook a comprehensive meta-QTL(MQTL)analysis involving 523 QTL associated with silage-quality traits collected from 14 published studies.Of the 523 QTL,405 were projected onto a consensus map comprising 62,424 genetic markers,resulting in the identification of 60 MQTL and eight singletons.The average confidence interval(CI)of the MQTL was 3.9-fold smaller than that of the source QTL.Nine of the 60 identified MQTL were classified as breeder’s MQTL owing to their small CIs, involvement of more QTL, and large contribution to phenotypic variation.One-third of the MQTL co-localized with DNA marker-trait associations identified in previous genomewide association mapping studies.A set of 78 high-confidence candidate genes influencing silage quality were identified in the MQTL regions.These genes and associated markers may advance marker-assisted breeding for maize silage quality.

1.Introduction

Maize(Zea mays L.)is a staple food and feed crop,with over 70%of global maize production used for livestock feed, particularly ruminants, as high-energy silage [1].Silage maize, cultivated for this purpose, combines high yield potential, palatability, and energy value.Maize silage is an excellent supplement, especially during periods of fodder scarcity, such as dry seasons, owing to its abundant green mass and dry matter production, superior fermentability during storage,and high acceptance by meat and dairy animals.The leading regions in maize silage production include North America (with 40% of the global share), Europe, the Middle East, Latin America, and northern Asia [2].The market value of maize silage is projected to rise from 342 million to 677 million US dollars between 2022 and 2032[2].Over the past two decades,the USA has witnessed a gradual increase of approximately 25%in silage maize production [2].Despite these developments, silage quality has received relatively less attention as compared to other traits.

Silage quality in maize is influenced primarily by its content of cell-wall components and stalk sugar content.The cell wall, comprising much of the plant used for silage, is poorly digested by ruminants[3],limiting the nutritional value of silage.Constituents of the cell wall are cellulose(～45%),hemicellulose(～45%),and lignins(～10%)[4],displaying varying degrees of digestibility.Among ruminants, cellulose and hemicellulose digestibilities range from 50% to 90% and 20% to 80%, respectively [5].Lignin concentration is inversely correlated with cell wall digestibility [3] and varies with genetic background.The presence of lignin in the cell wall matrix hinders its digestion in ruminants owing to its resistance to degradation by rumen microbes.Lignin concentration is the most influential cell-wall component affecting cell-wall digestibility in commercial maize hybrids,accounting for over 50%of variation in the trait [6].Another quality trait, stalk sugar content,increases silage quality.Increased sugar content in the stalk expedites fermentation during ensiling [7] and is strongly correlated with silage quality and palatability [8,9].

Recent advances in genotyping techniques and precision in phenotyping have permitted the discovery of numerous quantitative trait loci (QTL) and genes associated with diverse maize traits[9,10].The identification of QTL and their linked markers associated with silage-quality traits (SQT) is vital for their effective use in marker-assisted maize breeding[11].QTL mapping has revealed several QTL and genes associated with maize SQT[8,9,12,13].However, the practical applicability of these QTL in plant breeding remains a subject of concern owing to differences in genetic backgrounds(parental lines),mapping population sizes and types(e.g.,backcross-BC,F2, F2derived F3–F2:3, and recombinant inbred lines,RILs), and statistical methods employed (single and multiple QTL mapping) across QTL mapping studies.Moreover, many QTL identified in biparental mapping populations exhibit minor effects and limited stability, making them poorly suitable for marker-assisted selection and positional cloning.It is desirable to identify stable QTL(detected across environments)with large effects on trait phenotypes.Another form of QTL mapping, known as the genomewide association study (GWAS), has demonstrated its utility in identifying marker–trait associations (MTAs) in maize [14–16]and other crops [17].In a study [18], the MTAs discovered using GWAS were validated in biparental QTL mapping studies, highlighting the potential benefits of integrating findings from the two approaches to identify genomic regions associated with target traits.

Meta-analysis of QTL allows aggregating information from multiple QTL mapping studies conducted in biparental populations.It identifies consensus and consistent QTL or meta-QTL (MQTL)regions with narrowed confidence intervals (CIs) [19,20].MQTL analysis helps identify QTL hotspots and suggests the presence of pleiotropic effects by forming QTL clusters associated with multiple traits [21].MQTL analysis studies have been conducted for wheat [22,23], rice [24,25], sorghum [26], and maize [27,28].In maize, MQTL analysis has been employed to uncover consensus regions involved in root traits [28], yield traits [29], insect resistance [30], and silage traits [31].

A MQTL analysis for silage quality using data from QTL studies published up to 2010 revealed 68 MQTL[31].Nevertheless,considering the continuous identification and publication of new silagerelated QTL in recent years, it becomes imperative to incorporate the latest QTL data to discover MQTL that exhibit greater stability.Therefore, the current study was conducted to (i) identify consensus genomic regions linked to maize SQT by meta-analysis (based on QTL studies conducted till 2023), (ii) identify potential candidate genes (CGs) in MQTL regions, and (iii) validate the MQTL by reference to previous GWAS and genetic studies.

2.Materials and methods

2.1.Literature review and collection of QTL information

QTL mapping studies for SQT, including cell wall-component traits (CWT), digestibility traits (DT), and stalk sugar traits (SST),were collected from Google Scholar (https://scholar.google.com)and PubMed (https://www.ncbi.nlm.nih.gov/pubmed) using keyword searching.Each mapping study was used as a source for the following information: (i) QTL name (if available) and associated trait, (ii) flanking or closely linked markers associated with individual QTL, (iii) peak genetic position, (iv) CI, (v) LOD (logarithm of odds) scores of individual QTL, (vi) type of markers employed,(vii)mapping population type and size,(viii)chromosomal positions of the QTL, (ix) parental lines used in crosses for mapping-population development, and (x) phenotypic variance explained (PVE or R2) (Table S1).Where a LOD score was not provided,a LOD score of 3.0 was assumed.When the peak positions of individual QTL were missing,the midpoint of the genetic positions of the given flanking markers was substituted.When CI information for a QTL position was absent, the CI (95%) was calculated using the following formulas [32]:

CI (95%, for RIL mapping populations) = 163/(R2× N)

Meanwhile the old woman was standing7 at the mortar8 pounding the rise that was to serve them for the week with a pestle9 that made her arms ache with its weight

CI (95%, for F2and F2-derived populations) = 530/(R2× N)

In the above formulas, N represents the size of the mapping population employed in the QTL mapping study, while R2denotes the phenotypic variance explained by a given QTL.

Information of 523 QTL was collected from 14 QTL mapping studies (Table 1) conducted on SQT spanning the years 2000 to 2023(Tables S1,S2).The mapping populations used in these studies comprised F2, F2:3, F4, and RILs, with population sizes ranging from 131 to 242 (Table S1).A variety of marker types, including restriction fragment length polymorphism (RFLP), amplified fragment length polymorphism (AFLP), simple sequence repeats(SSR), and single-nucleotide polymorphisms (SNPs), were employed for mapping, with the composite interval mapping(CIM) method being the most frequently used approach among the source studies (Table S1).

MQTL analysis, along with candidate genes identification, was conducted for 22 traits (Table S2) grouped into three categories:(i) cell wall component traits (CWT), encompassing 14 traits: 5–5 diFA (DFA), 8-O-4 diFA (ODF), acid detergent fiber (ADF), acid detergent lignin(ADL),cellulose(CEL),etherified ferulic acid(Ether FA), esterified ferulic acid (Ester FA), hemicellulose (Hcell), klason lignin (KL), neutral detergent fiber (NDF), p-coumaric acid (pCA),p-hydroxybenzaldehyde (pHB), syringaldehyde (Sg), and vanillin(Va); (ii) digestibility traits (DT), comprising four traits: in vitro neutral detergent fiber digestibility (IVNDFD), in vitro dry matter digestibility (IVDMD), in vitro digestible organic matter (IVDOM),and in vitro digestibility of non-starch,non-soluble carbohydrates,and non-crude protein (DINAGZ); and (iii) stalk sugar traits (SST),comprising four traits: Brix, three ear leaves area (TELA), plant height (PHT), and days to silking (DTS).All the traits in the CWT and DT categories have shown strong correlations with cell wall digestibility [31], while the traits in the SST category have shown correlations with silage quality and palatability [8,9].

2.2.Development of consensus map and QTL projection

An integrated consensus map was constructed by incorporating various marker types, including SSRs, SNPs, AFLP, random amplified polymorphic DNA (RAPD) and sequence-tagged sites (STSs).It was developed using the LPmerge package [42] of R [43] with the reference map ‘‘ISU Integrated IBM 2009” retrieved from https://www.maizegdb.org/, along with a high-density SNP map involving 54,234 SNPs, which was developed in a multi-parent advanced-generation inter-cross population [44], and markers flanking the QTL derived from various SQT mapping studies.Data retrieved from QTL mapping studies were used for the preparation of QTL and map files, following the Biomercator V4.2.3 software manual [45].The QTL files contained the following information for each QTL: QTL-ID (author name followed by the publication year of the corresponding mapping study and a unique identification number for each QTL), chromosome number, LOD score, peak position, CI, traits associated with the QTL, and the corresponding R2or PVE value.Using the QTLProj commands in Biomercator, the QTL were projected onto the consensus map.This projection was achieved by scaling the positions of the flanking markers of the QTL on the original maps to their respective positions on the consensus map.

Table 1 QTL mapping studies included in the meta-analysis.

2.3.QTL meta-analysis

The QTL projection was followed by meta-analysis using the Veyrieras two-step algorithm [19] for each chromosome in BioMercator.In the first step, the best MQTL model (helps us to ascertain the number of predicted MQTL) was selected based on the lowest value of the selection criteria in at least 3 of the models:average weight of evidence (AWE), Bayesian information criterion(BIC), Akaike information criterion (AIC), AIC model 3 (AIC3), and AIC corrected(AICc).MQTL were then generated based on the chosen best model.Each MQTL was assigned a unique identifier:MQTL1.1,MQTL1.2,MQTL1.3,etc.The number preceding the decimal represents the chromosome number, while the numeral following the decimal denotes the order of the MQTL on its chromosome based on its genetic position.The LOD score and PVE from each QTL comprising a MQTL were averaged to obtain the LOD score and PVE value for the MQTL.The physical positions of flanking markers surrounding each MQTL were obtained with the JBrowse tool(https://jbrowse.maizegdb.org/)(accessed on June 14, 2023) available in the Maize Genetics and Genomics Database(MaizeGDB).When the physical positions of the markers flanking MQTL regions were unavailable in MaizeGDB, the next closest outer flanking marker was used to retrieve the corresponding physical coordinates.Among the MQTL identified in the present study, some of the MQTL were assigned as breeder’s MQTL(bMQTL).These criteria included a CI of less than 2 cM, involvement of a minimum of five projected QTL, and a contribution of at least 10%to total phenotypic variation.MQTL meeting these criteria are considered suitable for incorporation into breeding programs and are consequently referred to as bMQTL.

2.4.Candidate gene mining

Candidate gene identification for identified MQTL was performed with the BioMart tool in the EnsemblPlants database(https://plants.ensembl.org/), using a 1-Mb region on either side of the physical peak position.The physical peak positions of MQTL were calculated using the following formula[23]:peak position in base pairs (bp) = start position in bp + {(end position in bp – start position in bp) / (end position in cM – start position in cM)} × CI(cM; 95%)/2.For each candidate gene, the following information was retrieved: gene ID, gene start position (bp), gene end position(bp) and functional description.Candidate genes were filtered by their functional descriptions and relevance to SQT (that are supposed to be involved in the regulation of lignin,hemicellulose,cellulose,and sugar),considering information found in the literature.Genes that passed these filters and were supported by the literature were assigned as high-confidence candidate genes (hcCGs).

2.5.Expression analysis of hcCGs

In-silico expression analysis (to identify expression pattern) of the hcCGs in MQTL regions was conducted using the qTeller-RNA sequence expression tool https://qteller.maizegdb.org/.The analysis relied on a previously generated transcriptomic dataset [46].From this dataset,transcriptomic data from various tissues at multiple growth stages: leaves (0, 6, 12, 18, and 24 days after pollination (DAP)), internodes (0, 6, 12, 18, and 24 DAP), immature cobs(R1 and V18 stages), and whole seeds (12, 18, and 24 DAP), was retrieved.A heat map was generated with Morpheus software(https://software.broadinstitute.org/morpheus/).Genes with transcript per million (TPM)values of ≥2 among various tissues were selected for further analysis.

2.6.Co-localization of MQTL with GWAS-MTAs and known genes

Data from GWAS-MTAs for SQTs published up to 2023 [14–16,47–49]were collected and used for comparison with the MQTL discovered in the present study.The association mapping panel size, the number of markers used and the environments in which the GWAS studies were conducted were also collected (Table S3).The physical positions of MTAs associated with SQTS were retrieved, permitting a chromosome-wise comparison with the physical coordinates of the MQTL.Any individual GWAS-MTA that fell within the genomic region of a given MQTL was assigned as colocated.The physical coordinates of known genes were retrieved from MaizeGDB and compared with the MQTL regions to determine their co-localizations in the MQTL regions.When the entire gene interval fell within the genomic coordinates of the MQTL, it was considered co-localized with the MQTL.

3.Results

3.1.Features of QTL associated with silage-quality traits

Among the QTL collected from different mapping studies, the LOD score values ranged from 2.0 to 31.2, with an average of 5.1(Fig.1A).The phenotypic variance explained by individual QTL varied from 1.0 to 58.5%, with an average of 9.6%.Among these, 328 QTL showed PVE values below 10%,while 195 QTL showed PVE values of 10%or higher(Fig.1B).The distribution pattern of the number of QTL associated with SQT across categories varied both among and within chromosomes (Fig.1C).

3.2.Consensus map and QTL projection

The consensus map consisted of 62,424 markers, with a total genetic length of 2470 cM.The chromosomes varied widely in genetic length and distribution of the number of markers.(Fig.S1).Of the initial set of 523 QTL, 405 (77.43%) were successfully projected onto the consensus map.The remaining 118(22.57%) could not be projected owing to either (i)a lack of markers common to the consensus map and the genetic maps from the source studies or (ii) the large CI.Among the three categories of SQT, the most projected QTL (268) were contributed by CWT, followed by DT with 75 and SST with 62(Fig.2A).The number of projected QTL varied among chromosomes, with chromosome 7 carrying the fewest(18)and chromosome 1 the most(87)(Fig.2B).

3.3.MQTL associated with maize silage-quality traits

The QTL projection and meta-analysis led to the identification of 60 MQTL (Table S4; Fig.3) and eight singletons (single QTL available from a single initial study).MQTL 2.5 comprised the most projected QTL(24),followed by MQTL 1.2(Table S4).Nine MQTL were associated with all three of the trait groups,41 MQTL with two,and 10 with only one.The average LOD score and PVE of MQTL ranged from 2.6 to 8.5 and 2.7 to 27(Table S4).The CI of MQTL on chromosome 6 was reduced by 6.3-fold, and the CI of MQTL on chromosome 1 was reduced by 5-fold compared to the CIs of the initial QTL present on those chromosomes(Fig.S2).Of the 60 MQTL,nine were assigned as breeder’s MQTL (bMQTL) (Fig.3).Of the nine bMQTL, two co-localized with GWAS-based MTAs.Six of the full set of MQTL (MQTL1.1, 1.2, 1.3, 1.4, 2.5, and 6.1) were associated with 15 or more projected QTL.MQTL2.5 contained the most projected QTL (24), followed by MQTL6.1 and 1.2 with 20 each.Our study revealed the distribution of most of the MQTL in the subtelomeric regions, which are also the regions of high gene density[50]as compared to non-subtelomeric regions of the chromosome.

3.4.Candidate genes associated with maize silage-quality traits

A search for genes in the MQTL regions identified 2069 candidate genes,excluding duplicates and genes with unavailable information (Table S5).MQTL6.4 contained the most candidate genes(102), followed by MQTL6.1 with 84.Among the MQTL regions,14 contained more than 50 genes, while four regions contained fewer than 10 genes (Table S5).Of the 2069 candidate genes, 78 were designated as hcCGs (Table S6).They encoded proteins including laccases, dirigent, cytochrome P450, cellulose synthase,COBRA-proteins, glycosyl transferase 43, epimerase, sucrose synthase, and trehalose phosphatase.

3.5.Expression profiling of hcCGs

Expression profiling revealed that 37 genes showed expression levels of ≥2 TPM(highlighted in green in Table S7)(Fig.4),among which 28 showed expression levels of ≥5 TPM in at least one of the tissues (Table S8).Of the 37 models showing expression ≥2 TPM,10 were constitutively expressed across almost all the tissues studied, whereas the expression of the 27 genes expressed at ≥2 TPM varied among tissues.

3.6.MQTL validation using GWAS-MTAs and known genes

Of the 60 MQTL identified, 20 co-localized with at least one MTA (Fig.3; Table S9).Among them, nine were colocalized with one MTA, two (MQTL1.4 and 3.5) with two MTAs, five (MQTL2.4,3.3, 5.4, 5.6, and 8.3) with three MTAs, three (MQTL3.4, 5.3, and 5.5) with four MTAs, and one (MQTL4.3) with 11 MTAs, for a total of 51 MTAs co-localized with the 20 MQTL (Table S9).Six known maize genes associated with SQT: bm5, bm1, PAL-1, pox3, sus1,and COMT4,were identified in MQTL regions(Table 2).These genes encode various proteins, with five of them involved in lignin biosynthesis and one involved in sucrose biosynthesis.

4.Discussion

The quality of maize silage depends on the composition of its cell wall and sugar content in the stalk.These factors limit the overall fodder value of the crop.To increase biomass quality during the ensiling process,lactic acid bacteria break down complex compounds into simpler forms [57].This breakdown enables ruminants to digest the silage with minimal energy consumption and high bioavailability[57],as illustrated in Fig.5.However,the presence of cell wall components,in particular lignin,hinders this process.Moreover, an increase in sugar content in the maize stalk promotes fermentation during ensiling, providing ruminants with higher energy [8,9].

4.1.Genomic landscape of silage quality and implications for maize breeding

In the present study,we employed a meta-analysis approach to identify MQTL associated with SQT in maize.Although a previous study [31] conducted a meta-analysis for SQT, our study distinguishes itself in several aspects.(i) Truntzler et al.[31] conducted a MQTL analysis and identified 68 MQTL by integrating QTL from studies published up to the year 2010.In spite of that, given the ongoing discovery of novel silage-related QTL in maize from 2010 onwards, we conducted a comprehensive meta-analysis of 523 QTL,finding 60 MQTL regions associated with SQT;(ii)in contrast to the previous study,our investigation expanded the scope to include stalk sugar traits in addition to cell wall and digestibility traits; (iii) the number of projected QTL were almost doubled in our study (405) as compared to the earlier study (209), which led to the clustering of 196 additional projected QTL in the MQTL regions; (iv) the average CI of the MQTL in our study was 3.6 cM,reduced in comparison with the previous MQTL study, which resulted in an average CI length of 17.6 cM.This reduction in CI length may be attributed to the inclusion of a greater number and variety of markers and of genetic maps derived from independent studies to construct the high-resolution consensus map; (v) the higher-density consensus map of 62,424 markers produced more projected QTL; (vi) as many as 20 MQTL were validated with the GWAS-based MTAs in our study, whereas in previous study no validation of MQTL with GWAS was performed, and (vii) in contrast to the earlier study,the expression analysis of hcCGs showing transcript abundance in multiple tissues was performed.Thus,our study represents the most comprehensive analysis to date for identifying maize MQTL associated with SQT.Of the 60 MQTL identified in our study, 44 occupied genetic positions nearly identical to those reported previously.

Fig.1.Features of initial QTL.(A)LOD(logarithm of odds)score-based distribution of initial QTL.(B)PVE(phenotypic variance explained)-based distribution of initial QTL.(C)Chromosomal distribution of QTL associated with silage-quality traits.CWT, cell wall-component traits; DT, digestibility traits; SST, stalk sugar traits.

Fig.2.Distribution of QTL.(A)number and proportion of projected QTL associated with SQT.(B)number of QTL before(white bar)and after(black bar)projection.CWT,Cellwall component traits; DT, digestibility traits; SST, stalk sugar traits.

Fig.3.Chromosome-wise (1 to 10) distribution of MQTL identified in the present study.

Fig.4.Heat map of high-confidence candidate genes expressed at ≥2 TPM.DAP, days after pollination.

Our research also unveiled that the majority of the MQTL were distributed in the sub-telomeric region of the chromosomes, and this trend aligns with previous findings in numerous MQTL analyses conducted on various crops, including maize [23,28].Two observations may explain this distinctive MQTL distribution pattern.First, a greater number of shared genetic markers near subtelomeric regions of chromosomes in individual studies and the consensus map results in an increased occurrence of MQTL in these regions.Second,although some initial QTL were initially identified in proximity to the sub-telomeric ends of the chromosome genetic maps,these QTL were not effectively projected,possibly owing to alack of markers common to the initial genetic maps and the consensus map, resulting in a smaller number of MQTL available in the distal ends of the chromosomes.

Table 2 Co-localization of known genes with MQTL associated with silage-quality traits.

Fig.5.Schematic representation of maize lignocellulose colonization by lactic acid bacteria and associated structural changes in the biomass.

4.2.Candidate genes associated with silage-quality traits

Among the hcCGs, some genes encoded proteins associated with lignin biosynthesis and modification.Laccase proteins function in lignin polymerization[58].Cytochrome P450 proteins function in determining structural characteristics of monomeric lignin units [59].Dirigent-like protein participates in phenylpropanoid biosynthesis and promotes lignification in cotton [60].Multicopper oxidases are associated with lignin degradation processes[61].Cellulose synthase is a key enzyme in cellulose synthesis[62].COBRA-protein is involved in cellulose deposition [63].Endocytosis-associated proteins and UDPGP family proteins are related to cellulose synthesis [64] (Table S6).

Several hcCGs were associated with hemicellulose metabolism.They encoded proteins such as XET-xyloglucan endotransglycosylase, which functions in hemicellulose deposition[65] and cell wall organization [66].Epimerase/dehydratase proteins are involved in the biosynthesis of xylose, which is the main sugar monomer of hemicellulose [67].Glycosyl transferase 43 functions in hemicellulose synthesis [65], and glycosyl transferase 4 is associated with cell-wall organization, according to the functional description.A few hcCGs encode proteins involved in sugar biosynthesis.Sucrose synthase participates in sucrose synthesis[68].Xylose isomerase is involved in xylose biosynthesis[69].Trehalose phosphatase, responsible for trehalose biosynthesis, and glycosyl transferase 20, engaged in the transfer of sugar moieties for trehalose biosynthesis[70],were among the hcCGs.A glycoside hydrolase-type carbohydrate-binding protein functions in glucose metabolic processes [71].Further, 10 genes were constitutively expressed in multiple tissues (≥2 TPM expression), suggesting their use as targets for improving SQT in multiple tissues.They invite cloning and functional characterization and could be used in transgenic and gene editing studies to advance the manipulation of SQT in maize.

4.3.Validation of MQTL with GWAS-based MTAs and their association with known genes in maize

In the present study, one-third of the identified MQTL were found to co-localize with GWAS-MTAs identified in six earlier studies[14–16,47–49],associated with ADF,NDF,IVDMD,IVDOM,LIG, CEL, Hcell and Brix.In the past, several genes associated with SQT in maize have been cloned and functionally characterized[51–56].Among these characterized genes, six genes: bm5 (MQTL 5.4),bm1 (MQTL 5.4), PAL-1 (MQTL 5.5), pox3 (MQTL 6.3), COMT-4(MQTL 10.1), and sus1 (MQTL 9.4) were localized in the MQTL regions.bm5 [51], bm1 [52], PAL-1 [53], pox3 [54], and COMT-4[56]function in lignin biosynthesis,while sus1 is involved in sugar biosynthesis [55] (Table 2), ultimately affecting silage quality.The co-localization of these GWAS-MTAs and the previously characterized genes associated with SQT in MQTL regions supports the potential of the identified MQTL for use in silage-quality-targeted crop improvement programs.These MQTL regions can be searched for other genes for use in marker-assisted and genomic selection programs.Several approaches, including transgenics, CRISPRbased knockdown, and overexpression, could be used to confirm or functionally characterize some of the hcCGs.

5.Conclusions

A meta-analysis of maize silage-quality traits identified 60 MQTL,of which one-third co-localized with GWAS-MTAs from previous studies and hold promise for use in silage-quality breeding.Nine of the MQTL were assigned as breeder’s MQTL.Cloning and functional characterization of the MQTL and hcCGs discovered in the present study could lead to their effective deployment in breeding programs.

CRediT authorship contribution statement

Jatin Sharma:Conceptualization, Formal analysis, Data curation,Investigation,Methodology,Software,Writing–original draft,Writing– review&editing.Shubham Sharma:Conceptualization,Data curation, Investigation, Methodology, Software.Krishna Sai Karnatam:Conceptualization, Methodology, Software.Om Prakash Raigar:Software.Chayanika Lahkar:Formal analysis,Investigation.Dinesh Kumar Saini:Methodology, Writing – review &editing.Sushil Kumar:Investigation.Alla Singh:Supervision.

Abhijit Kumar Das:Supervision.Priti Sharma:Supervision.Ramesh Kumar:Project administration, Supervision, Writing –review & editing.

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.

Acknowledgments

Thanks are due to SERB-CII (Science and Engineering Research Board-A statutory body of the Department of Science and Technology,Government of India and Confederation of Indian Industry)for providing the prestigious‘‘Prime Minister’s Fellowship for Doctoral Research” to Jatin Sharma, Krishna Sai Karnatam, and Om Prakash Raigar.This research received no specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2023.10.007.