Genetic Dissection of Yield-Related Traits in a Set of Maize Recombinant Inbred Lines Under Multiple Environments

Abstract

Agronomic advancements have led to significant increases in maize yield per hectare in Northeast China, primarily through improved density tolerance. However, the genetic mechanism underlying grain yield responses to density stress remains poorly understood. Here, a population of 193 recombinant inbred lines (RILs) derived from the cross between ZM058 and PH1219 was employed to identify quantitative trait loci (QTLs) under two planting densities across three locations over two years. Six yield-related traits were investigated: ear tip-barrenness length (BEL), cob diameter (CD), ear diameter (ED), ear length (EL), kernel number per row (KNR), and kernel row number (KRN). These traits exhibited distinct and divergent responses to density stress, with the values of CD, ED, EL, KNR and KRN decreasing as planting density increased, except for BEL. A total of 81 QTLs were identified for these traits: 39 were unique to low planting density, 22 to high planting density, and 20 were shared across both conditions. Additionally, nine QTL clusters implicated in the development of multiple traits were detected. The results indicate that planting density significantly affects yield traits, primarily through the interaction of numerous minor QTLs with multiple effects. This insight enhances our understanding of the genetic basis of yield-related traits and provides valuable guidance for breeding high-density-tolerant varieties.

1. Introduction

Maize (Zea mays L.) is a cornerstone of global food security, playing a pivotal role in human nutrition, livestock feed production, and industrial applications [1]. To address the challenges posed by population growth and the rising demand from agro-industries, enhancing maize yield has become a critical agricultural imperative [2,3]. Yield gains in recent decades have primarily resulted from the synergistic effects of improved crop management practices and genetic advancements, with increased planting density identified as a particularly effective strategy for productivity enhancement [4,5,6]. This strategy continues to be a significant target in current breeding programs.

Maize inflorescence and seed development, critical for yield formation, are influenced by multiple factors including hormones and nitrogen fertilization [7,8]. In recent years, several key yield-related genes have been cloned in maize. KNOTTED1 (KN1), specifically expressed in meristematic cells, negatively regulates gibberellin (GA) accumulation and directly binds to and represses VANISHING TASSEL2 (VT2), a co-ortholog of TRYPTOPHAN AMINOTRANSFERASE OF ARABIDOPSIS1 (TAA1) functioning in the tryptophan-dependent auxin biosynthesis pathway [9,10,11,12]. ZmGln1-3 plays a major role in kernel yield and is highly accumulated in leaf mesophyll cells, pivotal for nitrogen supply and partitioning. The overexpression of Gln1-3 in leaves increases the kernel number by 30% [13]. Dissecting the molecular mechanisms of yield-related traits in maize is useful for improving grain yield.

The northeast region of China, a typical high-latitude rainfed area, is characterized by a limited crop growth duration due to low temperatures, although it benefits from adequate solar radiation. This unique climatic condition suggests that the impact of dense planting on yield in northeast China may differ significantly from that observed in mid- and low-latitude regions, such as the North China Plain, where higher temperatures and relatively lower solar radiation are prevalent [14,15]. According to national statistics from 2010, maize grain yield and plant density in the region were approximately 5,255 kg ha^–1 and 52,500 plants ha^–1, respectively, accounting for about two-thirds of the plant density and one-half of the grain yield observed in the U.S. Corn Belt [16,17,18]. Over the past decades, plant density in the northeast region of China has increased to 70,900 plants ha⁻¹ [19], and this trend is continuing. However, the high-density planting of maize introduces potential risks, particularly during ear development, such as poor kernel set and increased tip barrenness, which ultimately threaten yield stability. Numerous studies have been conducted in recent years to evaluate the effects of plant population on maize grain yield in northeast China [20,21,22]. Understanding the genetic architecture of ear-related traits in maize under high-density conditions is critical for improving grain yield.

Maize yield is a complex trait influenced by various factors, including single-ear performance and the number of ears per hectare, which is affected by plant density [18,23]. Single-ear performance can be further categorized into several components, such as ear length, ear diameter, cob diameter, kernel row number, and kernel number per row. Numerous quantitative trait loci (QTLs) associated with maize yield have been identified through linkage mapping [24,25,26] and association mapping [27,28,29]. Notably, density-responsive QTLs have been characterized across diverse genetic materials. Gonzalo et al. identified a total of 19 loci associated with ear number per plant (7), barrenness (7), and yield per unit area (5) under two distinct planting density regimes, utilizing a recombinant inbred line (RIL) population derived from the cross between B73 and Mo17 [30]. Guo et al. detected 27 QTLs for yield-related traits under both low and high planting densities, employing 231 F_2:3 families generated from Zhengdan 958, a widely cultivated Chinese maize hybrid known for its density tolerance [31]. Yi et al. observed 110 QTLs for 16 yield-related traits using 301 RILs derived from the cross YE478 × 08–641 under two planting densities [32]. The maize short ear mutant ead1 was identified through map-based cloning, revealing that its encoded ALMT protein regulates ear development by exporting malate. Notably, the overexpression of EAD1 can increase both ear length and the number of kernel rows [33]. The kernel row number 2 (KRN2) encodes WD40 proteins and interacts with the gene DUF1644 to negatively regulate grain number. Its knockout has been shown to increase grain yield by approximately 10% to 8% without apparent trade-offs in other agronomic traits in maize [34]. Additionally, the flint kernel architecture 1 (fka1) encodes the ARFTF17 transcription factor, which regulates PIN1 expression and flavonoid biosynthesis by inhibiting the function of MYB40. This regulation subsequently affects IAA levels in the seed coat and determines maize kernel shape [35]. However, there has been limited research evaluating yield-related traits at different planting densities using early-maturing maize germplasm.

The primary objectives of this study were (1) to identify and quantify the effects of genetic factors influencing maize yield-related traits under contrasting planting densities using a RIL population, and (2) to provide actionable insights for marker-assisted selection aimed at optimizing grain yield in high-density agroecosystems.

2. Materials and Methods

2.1. Plant Materials and Field Experiments

In this study, a population consisting of 193 RILs was constructed from the cross ZM058×PH1219. ZM058, an elite maize inbred line derived from the Flint heterotic group, serves as a pivotal male parent in early-maturing maize breeding programs due to its exceptional tolerance to high density and broad adaptability to short-season environments. PH1219, belonging to the Iodent heterotic group, is derived from the renowned early-maturing hybrid XY1219, which is prominent in the Northeast China Early-Maturing Maize Zone, and exhibits relatively low tolerance to high planting densities. The heterotic pattern Iodent × Flint is one of the predominant patterns in northeast China. A set of 193 RILs (F₈) from the cross ZM058×PH1219 was produced in 2022 using the single-seed-descent method. The RIL population and its two parents were planted across six environments under two distinct plant densities. The six environments were established as follows: two at the maize breeding base of the Grass Science Institute, Heilongjiang Academy of Agricultural Sciences, in Harbin, Heilongjiang Province (HEB, 44°04’ N, 125°42’ E; mean temperatures recorded from the 5 months in which the corn plants are in the field: 14.8 to 25.6°C; average annual rainfall: 400–800 mm; elevation: ~150 m; annual average sunshine hours: 2500–3000 h; annual average accumulated temperature above 2700 °C) on 17 May 2023 and 20 May 2024 (2023HEB, 2024HEB); two at the maize breeding base of the Suihua Branch of the Heilongjiang Academy of Agricultural Sciences in Suihua, Heilongjiang Province (SH, 45°10’ N, 124°53’ E; mean temperatures recorded from the 5 months in which the corn plants are in the field: 13.8 to 25.2 °C; average annual rainfall: 483–550 mm; elevation: ~135 m; annual average sunshine hours: 2600–2900 h; annual average accumulated temperature: 2500–2700 °C) on 9 May 2023 and 25 April 2024 (2023SH, 2024SH); and two at the maize breeding base of the Heihe Branch of the Heilongjiang Academy of Agricultural Sciences in Heihe, Heilongjiang Province (HH, 47°42’ N, 124°45’ E; mean temperatures recorded from the 5 months in which the corn plants are in the field: 11.2 to 24.0 °C; average annual rainfall: 491–540 mm; elevation: ~131 m; annual average sunshine hours: 2562–2800 h; annual average accumulated temperature: 2100–2300 °C) on 11 May 2023 and 10 May 2024 (2023HH, 2024HH). The two planting densities at each location were designed as follows: low plant density (LPD, 75,000 plants/ha with a spacing of 20.5 cm between plants within a row and 65 cm between rows; 16 plants per row) and high plant density (HPD, 105,000 plants/ha with a spacing of 14.5 cm between plants within a row and 65 cm between rows; 24 plants per row). A total of 32 and 48 kernels were planted in each row of LPD and HPD, respectively. Subsequently, young seedlings at the V3 stage (three-leaf with collars visible) were thinned in the field to ensure the maintenance of the two designed planting densities.

E1, E2, E3, E4, E5, and E6 represent 2023 Harbin, 2023 Suihua, 2023 Heihe, 2024 Harbin, 2024 Suihua, and 2024 Heihe under low plant density (LPD, 75,000 plants ha⁻¹), respectively. Conversely, E7, E8, E9, E10, E11, and E12 denote the same locations under high plant density (HPD, 105,000 plants ha⁻¹) for the years 2023 and 2024. Each trial was performed using a randomized complete block design with two repetitions, and each plot consisted of a single row. Trials for both planting densities were conducted within a square and uniform parcel in each environment. Field management adhered to standard agricultural practices.

2.2. Phenotypic Measurements and Analysis

Seven ears from the center of each row were harvested for further evaluation. A total of six yield-related traits were measured: ear tip-barrenness length (BEL, cm), cob diameter (CD, cm), ear diameter (ED, cm), ear length (EL, cm), kernel number per row (KNR), and kernel row number (KRN) (Table S1). The means of two replications in each environment were used as measurements. The means of the phenotypic traits in the RIL population across all environments under each planting density were used for phenotypic distribution analyses. Analysis of variance (ANOVA), correlation analysis and broad-sense heritability (H²) calculation were performed using SPSS version 22. To minimize the influence of external factors on phenotypic variation, the best linear unbiased estimate (BLUE) value for each recombinant inbred line across replicates at each planting density was calculated using R. The H² value was estimated as follows:

$$H^2={\displaystyle \frac{{\sigma }_G^2}{{\sigma }_G^2+\frac{1}{e}{\sigma }_{GE}^2+\frac{1}{re}{\sigma }_E^2}}$$

Here, ${\sigma }_G^2$ represents the genetic variance of the measured traits, ${\sigma }_{GE}^2$ represents the interaction variance between genotype and environment, and ${\sigma }_E^2$ represents the residual error variance of the measured traits. Additionally, e indicates the number of environments, while r refers to the number of replications within each environment.

2.3. Genetic Map and QTL Mapping

Genomic DNA was extracted from V10 stage leaves (ten leaves with visible collars) of the 193 RILs and the parental lines using a modified CTAB method [36]. The extracted DNA was subsequently sent to Shijiazhuang MolBreeding Biotechnology Co., Ltd. (Shijiazhuang, China) for genotyping with the maize 3K SNP array, which comprises 3617 SNPs. The genotyping process employed an in-solution probe capture method, ensuring controlled sequencing depth to capture multiple surrounding SNPs for each target SNP. Following filtering criteria of minor allele frequency (MAF) > 0.05 and a missing rate < 10%, 1815 SNPs were found to be polymorphic between the two parents and were employed for mapping. The mapping module was used to construct a genetic map, following the basic steps of grouping, ordering, ripping and outputting. The Kosambi function was employed to calculate the distance between markers, with the genetic map distance measured in centi-Morgan (cM). The scanning step and PIN value were set to 1 cM and 0.001, respectively. The total length of the map is 9394.32 cM, with an average genetic distance of 5.36 cM between adjacent markers. QTLs were identified using the Inclusive Composite Interval Mapping (ICIM) method in QTL IciMapping 4.2 software, applying an LOD threshold of 2.5 [37]. The confidence interval for a single QTL was defined according to a 2-LOD support interval. The QTLs were named using the following format: q + trait abbreviation + chromosome number + serial number of QTL. For single-environment QTL mapping analysis, phenotype data consisted of BLUE values across replicates in a single environment. Notably, QTLs identified at two plant densities were considered stable QTLs, while location-specific QTLs were defined as those identified in only one location, and density-specific QTLs as those identified under a single plant density. BIN genetic regions were annotated based on information provided by the MaizeGDB bin viewer (https://maizegdb.org/bin_viewer, accessed on 3 March 2025). Overlapping QTLs for different traits were regarded as a QTL cluster.

2.4. QTL Epistasis Effect Analysis

Considering the potential epistatic genetic effects among the linkage groups of the RILs population, we utilized QTLNetwork v2.1 software (Website: http://ibi.zju.edu.cn/index.html/bcl/software/qtlnetwork.html, accessed on 3 March 2025) to perform an epistatic analysis of QTL. This software employs Mixed-Model-Based Composite Interval Mapping (MCIM) [38,39], and incorporates a Bayesian test to identify the loci exhibiting significant interactions, with a threshold set at p ≤ 0.005. QTLNetwork v2.1 is used to detect Epistatic Interaction (EPI; additive by additive interaction, namely the AA effect) for yield-related traits by integrating data across all environments.

3. Results

3.1. Phenotypic Evaluation of Yield-Related Traits Across Two Plant Densities

Six yield-related traits were examined in both parental lines across two planting densities. For ZM058, the length of CD was approximately 2.45 cm in LPD and 2.35 cm in HPD, both of which were higher than those of PH1219, which measured 1.81 cm and 1.71 cm in LPD and HPD, respectively. Similarly, the ED, EL, KNR, and KRN also exhibited significantly higher values in ZM058 than those in PH1219 under both planting densities. However, the BEL of ZM058 was recorded at 0.23 cm in LPD and 0.37 cm in HPD, both lower than the corresponding values for PH1219, which were 1.79 cm and 2.25 cm in LPD and HPD, respectively (

Table 1. Descriptive statistics for six yield-related traits of the parental lines and RIL population under two plant densities.

Trait	Plant Density	PH1219	ZM058	RIL Population
				Mean ± SD	Range	Skewness	Kurtosis	CV (%)	H2 (%)
BEL	LPD	1.79 ± 0.06	0.23 ± 0.01	1.11 ± 0.72	0–4.48	0.96	1.18	65.45	81.54
	HPD	2.25 ± 0.15	0.37 ± 0.05	1.16 ± 0.71	0–5.62	0.82	1.46	61.21	80.15
CD	LPD	1.81 ± 0.07	2.45 ± 0.11	2.19 ± 0.22	1.54–2.98	0.41	0.38	10.05	93.01
	HPD	1.71 ± 0.04	2.35 ± 0.07	2.17 ± 0.22	1.61–3.01	0.53	0.48	10.14	92.70
ED	LPD	3.18 ± 0.11	4.03 ± 0.13	3.99 ± 0.31	2.67–4.91	−0.19	0.29	7.77	90.44
	HPD	3.14 ± 0.07	3.99 ± 0.09	3.95 ± 0.31	2.94–4.81	0	−0.14	7.59	90.45
EL	LPD	9.45 ± 0.49	14.53 ± 0.59	14.26 ± 1.62	7.99–19.95	−0.15	0.77	11.36	87.27
	HPD	9.33 ± 0.55	14.39 ± 0.46	13.93 ± 1.59	7.81–19.28	−0.15	0.4	11.41	87.54
KNR	LPD	13.51 ± 0.81	25.71 ± 0.68	25.99 ± 4.73	8.26–41.98	−0.13	0.24	18.20	87.89
	HPD	12.59 ± 0.76	25.17 ± 0.36	25.23 ± 4.64	10.44–42.33	0.01	0.09	18.39	86.71
KRN	LPD	13.63 ± 0.18	14.95 ± 0.41	15.61 ± 1.91	10.32–21.36	0.23	−0.41	12.24	94.23
	HPD	13.48 ± 0.35	14.72 ± 0.29	15.55 ± 1.88	11.06–21.69	0.31	−0.27	12.09	92.60

BEL, length of barren ear tip, cm; CD, cob diameter, cm; ED, ear diameter, cm; EL, ear length, cm; KNR, kernel number per row; KRN, kernel row number; LPD, low plant density (75,000 plants ha⁻¹); HPD, high plant density (105,000 plants ha⁻¹); BLUE, best linear unbiased estimate; SD, standard deviation; CV (%), coefficient of variation; H²(%), broad-sense heritability.

). Additionally, all traits except BEL for both parents and the RIL population decreased with increasing planting density. The average length of CD in RILs was approximately 2.19 cm in LPD, decreasing to 2.17cm in HPD. Similarly, the ED, EL, KNR, and KRN were recorded as 3.99 cm to 3.95 cm, 14.26 cm to 13.93 cm, 25.99 cm to 25.23 cm, and 15.61 cm to 15.55 cm, respectively. In contrast, the mean length of BEL increased from 1.11 cm in LPD to 1.16 cm in HPD (Table 1; Figure 1 and Figure S2). The marked transgressive segregation and quantitative characteristics of these traits indicate strong potential for phenotypic improvement, revealing substantial genetic divergence between the parental lines concerning yield-related characteristics. All six traits demonstrated high broad-sense heritability (H²) across all environments, with values ranging from 80.15% (BEL in HPD) to 94.23% (KRN in LPD).

The analysis of the same traits across three locations over two years revealed high correlations among the six traits. Additionally, phenotype data collected at two different densities demonstrated similar consistency (Figure 2 and Figure S1, Table S3). Among the correlation coefficients for these traits, significant positive correlations were observed between CD and ED, with values of 0.82 and 0.83; CD and KRN, with values of 0.45 and 0.41; ED and KRN, with values of 0.55 and 0.51; and EL and KNR, with values of 0.62 and 0.66 in both densities. However, significant negative correlations were noted between BEL and KNR in both densities. Additionally, CD and KNR, as well as EL and KRN, exhibited significant negative correlations only in HPD. Genotype (G) had a significant influence on all traits, particularly KRN, which accounted for 55.7% of the phenotypic variance. The Sum of Squares Error/Sum of Squares Total (SSE/SST) values for BEL, CD, ED, EL, and KNR were 35.3%, 50.6%, 50.4%, 41.5%, and 34.5%, respectively (Table S4). Furthermore, the interactions of genotype-by-environment (G × E) and genotype-by-environment-by-density (G × E × D) had critical influences on all traits, while the influence of genotype-by-density (G × D) was weaker than that of G × E and G × E × D.

3.2. QTL Mapping of Yield-Related Traits in the RIL Population

A total of 81 QTLs for the six traits were detected across all environments (three locations, two densities and two years) (Figure 3,

Table 2. QTLs identified for yield-related traits using high-resolution bin map.

Trait	Name	Chr.	Interval (Mb)	LOD	PVE Range (%)	ADD Range	Environment
BEL	qBEL1-1	1	26.91–35.00	3.92	7.88	−0.26	E10
	qBEL1-2	1	233.517–233.518	2.50	2.23	0.15	E5
	qBEL2-1	2	25.93–37.39	3.09	2.92–7.73	0.18–0.22	E1/E4/E5
	qBEL2-2	2	213.75–216.33	3.63	6.90	−0.22	E6
	qBEL2-3	2	223.09–224.39	2.63	6.68	0.18	E9
	qBEL4	4	176.00–177.07	3.04	6.87	−0.20	E11
	qBEL5	5	9.83–68.68	2.93	14.06	0.42	E5
	qBEL6	6	36.48–40.05	3.26	5.84	0.20	E6
	qBEL7-1	7	13.97–16.20	4.49	3.97	−0.22	E5
	qBEL7-2	7	162.09–174.14	3.54	5.48–10.72	0.22−0.25	E1/E6/E10/E12
	qBEL8	8	142.20–149.42	3.88	6.61	0.24	E6
	qBEL9	9	103.15–110.48	2.59	6.04	−0.17	E7
	qBEL10	10	26.14–65.43	4.91	8.35	0.28	E10
CD	qCD1-1	1	7.24–10.82	3.93	6.54–8.17	−0.06 to −0.07	E3/E7
	qCD1-2	1	40.23–43.50	3.07	6.06	0.04	E8
	qCD1-3	1	83.44–85.03	3.5	5.15–7.79	0.05−0.07	E3/E7
	qCD4-1	4	47.23–48.44	2.79	5.30	0.04	E3
	qCD4-2	4	163.42–166.94	2.95	4.52–7.28	−0.05 to −0.06	E1/E8
	qCD4-3	4	230.03–231.88	3.94	7.27	−0.07	E12
	qCD5	5	189.86–198.87	4.69	7.48	0.07	E1
	qCD6	6	162.88–163.20	2.52	5.83	−0.05	E2
	qCD7	7	138.41–164.83	3.30	4.69–5.28	0.05–0.07	E1/E7
	qCD8-1	8	72.69–99.70	4.04	6.57–6.89	−0.06 to −0.07	E1/E6
	qCD8-2	8	139.81–141.13	2.74	5.08	−0.06	E12
	qCD10	10	142.93–144.36	2.86	5.19	0.05	E7
ED	qED1-1	1	40.23–43.50	2.69	5.28	0.08	E8
	qED1-2	1	79.45–85.03	3.20	7.36	0.08	E7
	qED1-3	1	286.60–288.77	3.22	7.70	−0.08	E9
	qED2	2	49.03–56.07	3.33	9.01	0.08	E1
	qED3-1	3	3.50–5.32	3.10	8.13	0.09	E3
	qED3-2	3	88.33–104.92	3.24	7.28	0.08	E7
	qED3-3	3	197.47–198.60	2.73	4.50	0.08	E2
	qED3-4	3	214.20–214.45	3.82	7.72	−0.08	E5
	qED4-1	4	149.08–151.11	2.58	5.18	0.08	E5
	qED4-2	4	163.42–166.94	2.96	7.38	−0.10	E8
	qED5	5	40.26–132.69	3.31	10.59	0.14	E6
	qED8	8	13.79–15.99	2.68	2.86	0.08	E6
	qED10	10	138.80–140.93	2.79	6.66	–0.10	E2
EL	qEL1	1	15.35–23.61	2.94	4.89–6.17	−0.45 to −0.49	E3/E6
	qEL2	2	45.01–46.79	3.27	5.63	0.47	E1
	qEL3-1	3	47.83–57.99	3.33	6.46–7.21	−0.38 to −0.45	E3/E7
	qEL3-2	3	101.20–106.14	3.68	6.67	−0.45	E5
	qEL3-3	3	158.55–180.31	2.88	5.37–8.34	−0.46 to −0.48	E10/E11
	qEL3-4	3	196.78–201.76	3.50	7.36–8.73	−0.41 to −0.52	E4/E5/E12
	qEL4-1	4	37.26–38.32	3.68	8.06	−0.43	E7
	qEL4-2	4	85.70–126.63	4.38	7.77	−0.53	E1
	qEL5	5	14.77–16.45	2.63	5.40	0.45	E3
	qEL6-1	6	37.90–74.20	4.77	8.81	−0.53	E5
	qEL6-2	6	107.27–122.11	3.50	6.45–10.26	−0.26 to −0.58	E4/E8/E10/E11
	qEL6-3	6	165.00–167.6	2.65	5.68	−0.36	E7
	qEL7	7	137.55–138.41	3.16	5.49	0.43	E6
	qEL8	8	119.58–144.13	3.33	6.68–8.17	−0.41 to −0.49	E6/E12
	qEL9	9	40.45–154.11	2.86	8.41	0.74	E1
KNR	qKNR2-1	2	0.14–1.28	2.69	4.89	−0.95	E6
	qKNR2-2	2	9.98–10.22	2.76	6.03	−1.15	E7
	qKNR3	3	165.49–218.41	3.82	4.92–9.52	−0.96 to −1.34	E4/E5/E6/E7/E10/E11/E12
	qKNR4-1	4	11.09–12.22	3.29	7.87	1.29	E2
	qKNR4-2	4	157.64–158.53	3.11	5.19	−1.04	E5
	qKNR6	6	65.43–96.10	3.10	4.56–9.45	−0.95 to −1.28	E5/E10/E11
	qKNR7-1	7	133.08–139.17	4.54	7.73	1.27	E5
	qKNR7-2	7	163.00–164.83	4.87	9.13	−1.34	E5
	qKNR8-1	8	2.76–4.17	3.04	7.56	−1.05	E5
	qKNR8-2	8	119.58–149.42	3.87	6.64–12.22	−1.13 to −1.54	E4/E6/E10/E12
	qKNR8-3	8	163.29–164.96	3.10	8.40	−1.13	E1
KRN	qKRN1-1	1	37.85–40.23	4.10	9.03–10.32	0.57–0.58	E1/E7
	qKRN1-2	1	81.98–82.03	3.40	4.17	0.43	E11
	qKRN1-3	1	260.21–263.27	2.67	3.15	0.37	E5
	qKRN2-1	2	46.79–52.38	2.59	4.06	0.51	E12
	qKRN2-2	2	199.05–207.81	3.62	4.57–9.53	0.51–0.61	E1/E2/E6/E7
	qKRN3-1	3	21.76–27.77	8.77	8.46–14.94	0.56–0.94	E3/E5/E6/E11
	qKRN3-2	3	33.78–44.22	15.60	13.18	1.06	E4
	qKRN3-3	3	59.05–114.81	6.62	5.88–8.30	−0.50 to −0.80	E4/E5/E6/E11
	qKRN4-1	4	24.49–30.96	5.86	11.04–12.19	0.65–0.72	E2/E8
	qKRN4-2	4	84.05–87.83	2.80	2.27	0.47	E4
	qKRN4-3	4	143.12–163.42	5.34	5.00–9.90	0.51–0.72	E5/E10/E11/E12
	qKRN5-1	5	155.03–214.16	3.36	8.57–10.91	−0.70 to −0.74	E8/E10
	qKRN5-2	5	177.17–189.86	4.51	10.63	0.62	E3
	qKRN8-1	8	15.99–16.35	4.09	4.14	−0.50	E6
	qKRN8-2	8	135.25–165.28	3.66	3.64–9.28	−0.39 to −0.69	E5/E11/E12
	qKRN9	9	5.67–7.46	5.06	7.89	0.63	E10
	qKRN10	10	147.95–148.36	3.51	5.16	0.49	E8

Chr., chromosome; Interval, the confidence interval between two adjacent bin markers on the B73 reference genome version RefGen_v3; LOD, logarithm of odds score; PVE, phenotypic variance explained by individual QTL; ADD, additive effects value; BEL, length of barren ear tip; CD, cob diameter; ED, ear diameter; EL, ear length; KNR, kernel number per row; KRN, kernel row number. A positive ADD value means that the ZM058 allele increased the phenotypic value of the trait, and a negative value indicates that the PH1219 allele increased it.

and Table S5). The number and distribution of QTLs varied significantly across environments.

For BEL, 13 QTLs were identified through single-environment mapping, with the contribution to phenotypic variance ranging from 2.23% (qBEL1-2 in E5) to 14.06% (qBEL5 in E5). Among these, qBEL7-2 was consistently detected across both plant densities and four environments (E1, E6, E10, and E12), contributing between 5.48% and 10.72% to phenotypic variance. Additionally, qBEL2-1 was stably observed in three environments (E1, E4, and E5), contributing between 2.92% and 7.73%. These results indicate that the alleles carried by qBEL7-2 and qBEL2-1 from the parent PH1219 significantly contribute to the development of barren ear tips.

In our study on CD, we identified a total of 12 QTLs, and qCD1-1, qCD1-3, and qCD4-2 were detected in different locations in the same year. Notably, only qCD7 was detected across both planting densities in the same year and location, exhibiting phenotypic contribution rates of 4.69% to 5.28%. qCD1-3 showed phenotypic contribution rates ranging from 5.15% to 7.79%. Conversely, qCD1-1, qCD4-2, and qCD8-1 exhibited negative phenotypic contributions derived from PH1219 alleles, with contribution rates ranging from 6.54% to 8.17%, 4.52% to 7.28%, and 6.57% to 6.89%, respectively. For ED, 13 QTLs were detected across all chromosomes except for chr. 6, 7 and 9. Among these QTLs, qED5 exhibited the highest phenotypic contribution rate at 10.59%, while qED8 exhibited the lowest at 2.86%.

A total of 15 QTLs affecting EL were identified, distributed across all chromosomes except for chr. 10. Among these QTLs, qEL6-2 was consistently detected across various planting densities and four environments (E4, E8, E10 and E11), contributing between 6.45% and 10.26% to the phenotypic variance. Additionally, qEL3-4 was stably detected in three environments (E4, E5 and E12), contributing 7.36% to 8.73%. Furthermore, qEL1, qEL3-1, qEL3-3, and qEL8 were consistently detected in two environments each, with phenotypic contribution rates ranging from 4.89% to 8.17%.

For KNR, 11 QTLs were identified across all chromosomes except for Chr.1, Chr.5, Chr.9, and Chr.10. Among these, qKNR3, qKNR6 and qKNR8-2 were detected across both planting densities. Notably, qKNR3 was consistently observed in seven environments (E4, E5, E6, E7, E10, E11, and E12), contributing between 4.92% and 9.52% to the phenotypic variance. qKNR8-2 was stably identified in four environments (E4, E6, E10 and E12), contributing 6.64% to 12.22%. qKNR6 was detected in three environments (E5, E10 and E11).

For KRN, 17 QTLs were identified across all chromosomes except for chr. 6 and chr. 7. Among these QTLs, seven (qKRN1-1, qKRN2-2, qKRN3-1, qKRN3-3, qKRN4-1, qKRN4-3 and qKRN8-2) were consistently detected across both planting densities. Specifically, qKRN2-2, qKRN3-1, qKRN3-3, and qKRN4-3 were observed in four environments. qKRN8-2 was stably identified in three environments (E5, E11 and E12), contributing between 3.64% and 9.28% to the phenotypic variance. qKRN1-1, qKRN4-1 and qKRN5-1 were consistently detected in two environments, contributing 9.03% to 10.32%, 11.04% to 12.19% and 8.57% to 10.91%, respectively.

3.3. Epistatic Interactions Under Contrasting Planting Densities

A total of 27 digenic epistatic locus pairs were identified for six yield-related traits across both planting density regimes (Table S6). The distribution of these interactions revealed distinct density-dependent patterns. Under low plant-density treatments, four pairs of loci exhibited epistatic interactions, with two associated with BEL and two with KNR. Under high plant-density treatments, 23 pairs of loci displayed epistatic interactions, encompassing all six traits: four pairs for BEL, three pairs for CD, five pairs for ED, four pairs for EL, three pairs for KNR, and four pairs for KRN. For each trait, the number of loci involved in epistatic interactions ranged from three to six. Additionally, the total phenotypic variation explained by loci with EPI ranged from 0.20% for the EL condition to 0.11% for ED under high-density treatment. However, no locus exhibited EPI in both planting densities.

3.4. Analysis of QTL Clusters Associated with Yield-Related Traits

A total of nine QTL clusters (QCs) were detected, distributed across chromosomes 1, 2, 4, 7, 8, and 10 (see

Table 3. QTL clusters identified for six yield-related traits.

QTL Cluster	Chr.	Interval (Mb)	Physical Length (Mb)	Bin (B73 RefGen_v3)	No. of QTLs	Integrated QTLs	Associated Gene
QC1	1	37.85–43.50	5.65	1.03	3	qCD1-2, qED1-1, qKRN1-1	TS2 [40]
QC2	1	79.45–85.03	5.58	1.05	3	qCD1-3, qED1-2, qKRN1-2	TALE5 [41]
QC3	2	45.01–56.07	11.06	2.07	3	qED2, qEL2, qKRN2-1
QC4	2	199.05–224.39	25.34	2.09–2.10	3	qBEL2-2, qBEL2-3, qKRN2-2	BAD1 [42], WAB1 [43]
QC5	4	143.12–177.07	33.95	4.06–4.08	6	qBEL4, qCD4-2, qED4-1, qED4-2, qKNR4-2, qKRN4-3	FEA2 [44], NATL1 [45], XTH32 [46], TU1 [47]
QC6	7	133.08–139.76	6.68	7.03	3	qCD7, qEL7, qKNR7-1	RA1 [48], TSH4 [49]
QC7	7	162.08–174.14	12.06	7.04	3	qBEL7-2, qCD7, qKNR7-2	RA3 [50]
QC8	8	119.58–149.42	29.84	8.04–8.06	5	qBEL8, qCD8-2, qEL8, qKNR8-2, qKRN8-2	TALE33 [51], ATX5 [52]
QC9	10	138.80–148.36	9.56	10.06	3	qCD10, qED10, qKRN10	ZFL1 [53]

BEL, length of barren ear tip; CD, cob diameter; ED, ear diameter; EL, ear length; KNR, kernel number per row; KRN, kernel row number.

). These QCs collectively encompassed 32 QTLs (refer to Table 2 and Table 3). Among these, QC5 on chromosome 4 contained the highest number of QTLs, totaling six, followed by QC8 on chromosome 8, which harbored five QTLs. The remaining QCs each contained three QTLs, located on chromosomes 1, 2, 7 and 10, respectively. Notably, two hotspots were identified on chromosomes 4 (bins 4.06–4.08) and 8 (bins 8.04–8.06), each containing QTLs for five traits. No individual cluster harbored QTLs for all six traits. A total of eight QCs contained QTLs for at least three traits. Among these, QC5 harbored QTLs for several traits, including BEL, CD, ED, KNR and KRN. QC8 controlled QTLs for BEL, CD, EL, KNR and KRN. Additionally, QC1 and QC2 each contained QTLs for CD, ED and KRN, and were detected under high planting-density conditions. Other QCs were associated with distinct sets of traits: QC3 harbored QTLs for ED, EL and KRN, while QC6 contained QTLs for CD, EL and KNR.

4. Discussion

Over the past few decades, high planting density has been established as a powerful and reliable method for enhancing maize yield [54]. However, yield-related traits have been observed to vary significantly in response to planting density stress [31,32]. In this study, all values of CD, ED, EL, KNR, and KRN in both the male parent, female parent, and RIL population decreased with increasing planting density, while the values of BEL increased with higher planting density (Figure 1; Table 1 and Table S2). Similar results were corroborated by a previous study [32]. Additionally, the architecture of maize plants changed with variations in planting density [55]. These results suggest that plants may systematically alter their morphology and ear architecture to adapt to density stress. The observed phenomenon of increased plant height and reduced leaf size indicates that maize exhibits a shade-avoidance response to higher density, reallocating energy resources into stems. This adaptation may enhance the plants’ ability to compete more effectively for sunlight, warmth, carbon dioxide, and water necessary for seed development.

In the northeast region of China, maize faces challenges related to limited crop growth duration due to low air temperatures and prolonged frost periods. Here, optimizing planting density becomes critical for maximizing maize yield. High planting density, as demonstrated in this study and supported by previous research, can enhance yield per unit area; however, it may also produce negative effects and adverse changes in yield-related traits and plant architecture. Understanding and managing these responses through genetic improvement and optimized agronomic practices could be essential for maximizing maize yield in this region.

Yield-related traits in maize are regulated by multiple genes that are interconnected within complex gene networks and are highly susceptible to variation due to the interaction between genotype and environment [25,26,55,56]. In this study, a total of 81 QTLs were identified for six traits, showing significant variation in the number of QTLs among these traits (Table 2). For instance, KRN had as many as 17 QTLs, whereas KNR had only 11. Among the detected QTLs, 13 loci were identified across two planting densities, including 7 for KRN (qKRN1-1, qKRN2-2, qKRN3-1, qKRN3-3, qKRN4-1, qKRN4-3 and qKRN8-2), 3 for KNR (qKNR3, qKNR6 and qKNR8-2), 1 for BEL (qBEL7-2), and 1 for CD (qCD7). KRN exhibited greater potential stability as a yield component under diverse planting conditions, while other traits appeared to be more sensitive to changes in density. Additionally, approximately 25.3% of QTLs (22/81) were specific to the high-density condition, and 24.7% (20/81) were common to both planting densities. Similarly, Yi et al. [32] reported that 110 QTLs were observed for 16 yield-related traits under two planting densities, with about 30% of QTLs uniquely identified under high-density conditions, while 24.5% of QTLs were detected across both planting densities in RILs derived from the cross between Ye478 and 08-641. These results indicate the complexity and diversity of the genetic control of yield-related traits in maize under different planting densities, highlighting the intricate interaction between genotype and environment in shaping maize yield. Genetic variation and phenotypic plasticity present significant challenges for breeders aiming to design effective breeding tools (such as genomic selection) and strategies, and are critical for developing new maize varieties with improved adaptation to diverse planting densities and environments, ultimately enhancing yield stability and overall productivity.

Despite the complexity of yield-related traits, several consistent QTLs were identified under varying planting densities, which aligns with findings from previous studies. For insistence, qED1-1, which is specific to high density on Chr1 (40.23–43.50 Mb) in our study, overlaps with qED1 identified across both densities by Yi et al. [32]. Additionally, qKRN3-3, detected in four environments across both densities on Chr3, was co-located with qRN3-2 under two different densities [32]. However, the discrepancies observed across studies highlight the intricate nature of yield, which is influenced not only by planting density but also by various other factors, including genetic background and environmental interactions.

The phenomenon of quantitative trait loci pleiotropy has been extensively documented in numerous studies [31,32]. In this study, nine QCs integrated with thirty-two QTLs related to yield traits were detected (Table 3, Figure 3). Among these, two QCs (QC1 and QC2) were specifically observed in high-density or both-density treatments, as reported in previous studies. The remaining seven genomic regions also contained overlapping QTLs for different traits under low and high planting densities (Table 3). Three QTL clusters (QC1, QC3 and QC7) co-located with the QTL hotspot region (QC1-1, QC2-1 and QC7-6) were reported by Yi et al. [32], and these QTLs were associated with BEL, CD, ED, KNR and KRN. Additionally, QC5 was identified as one of the hotspot regions in this study, associated with up to five distinct traits. Notably, it overlaps with Fasciated Ear 2 (FEA2), a critical gene regulating floral meristem size and organ number by modulating the ear inflorescence meristem [44,57]. Interestingly, another gene located in the same region, Tunicate1 (TU1), has been shown to enhance maize yield by increasing ear weight, grain weight per ear, and hundred-grain weight. This finding is consistent with recent research demonstrating that TU1 positively controls maize yield and serves as a promising target for maize breeding [47]. Furthermore, QC6 demonstrated a significant effect on the development of CD, EL and KNR. Consistently, the genes Ramosa1 (RA1) and Tasselsheath4 (TSH4), which govern branching architecture in the maize tassel and ear, were co-located with QC6 [48,49]. The UB2/UB3/TSH4 complex directly regulates the expression of BIF2 and ZmTCP30, thereby influencing tassel branching and ear development. QC9 affects ED, CD, and KRN. FLORICAULA1 (ZFL1), a key maize gene that plays a conserved role in floral organ development and the transition from vegetative to reproductive stages, was co-located within the region of qED10 [53]. In addition to the consistent QTL clusters, our study identified several novel QTL clusters that have been less reported in the literature. These new loci may provide valuable insights into the genetic mechanisms underlying density stress tolerance in maize. In conclusion, these findings suggest that there are likely causal relationships, specifically pleiotropy, among yield-related traits that respond differently to density stress. These pleiotropic regions, as potentially valuable loci, may offer deeper insights into the genetic basis of high-density tolerance in maize.