Study on Comprehensive Evaluation of Agronomic Traits and High-Yield Breeding Selection Strategy of Brassica napus L.

Abstract

In order to elucidate the trait structure of yield formation and optimize the selection strategy for breeding high-yield spring rapeseed, this study systematically evaluated the genetic variation, interrelationship, and contribution to yield of 10 key agronomic traits. A comprehensive assessment of 26 varieties across five test environments was conducted using the coefficient of variation, phenotypic correlation, path analysis, principal component analysis, and grey relational analysis. The results showed that the variations in plant height, branch position, and the number of primary effective branches were the most abundant (CV > 0.20), indicating high genetic improvement potential. Among the yield components, a significant positive correlation was observed between the number of effective pods per plant and the number of seeds per pod. The direct positive effect of pod length on yield per plant was the strongest (path coefficient = 0.467), indicating that yield formation was more dependent on pod structure and grain filling ability. Principal component analysis showed that PC1 had a contribution rate of 94.2%, driven mainly by the effective pod number of the whole plant. This could be used as a comprehensive index to distinguish between different ecological groups and evaluate the overall growth potential. Grey correlation analysis further clarified that the effective length of the main inflorescence was most closely related to yield per plant (correlation degree = 0.847). In summary, this study proposes a high-yield breeding strategy of ‘quality first, collaborative improvement’, whereby pod length, 1000-grain weight, and effective length of the main inflorescence are used as core selection traits. This novel study involves coordinating and optimizing the number of effective branches and inflorescence structure, as well as screening stable genotypes through multi-environment identification, in order to achieve the efficient integration of yield components.

1. Introduction

Rapeseed (Brassica napus L.) is a crucial oil crop in China, and enhancing its yield is a key objective for securing edible oil supply [1,2]. Over the past 40 years, the global average rapeseed yield has increased at an annual rate of 27 kg/ha; however, yield stability has not improved correspondingly, and significant disparities exist in growth trends across different countries [3]. As a complex quantitative trait, yield is affected by multiple agronomic traits, and there are complex genetic correlations and interaction effects among traits [4]. Adhi Shankar et al. [5] also clarified that yield-related traits are controlled by both additive and non-additive gene effects. Zhang et al. [6] suggested that increasing pod number per plant, seed number per pod, and branch number was beneficial for increasing yield. Researchers such as Yang and Peng [7], Zhang et al. [8], Ni et al. [9], and Bai et al. [10] have shown that it is an effective way to increase yield by selecting varieties with great potential in main inflorescence length, branch number, effective pod number, and 1000-grain weight. In recent years, quantitative genetic analysis methods, such as phenotypic correlation analysis, path analysis, principal component analysis, and grey correlation analysis, have provided powerful tools for analyzing complex trait structures, quantifying trait contributions, and guiding comprehensive selection. International research has deepened the understanding of rapeseed yield formation from multiple perspectives. At the genetic breeding level, the study not only confirmed that the number of pods per plant, the number of seeds per pod, and the 1000-grain weight were the key yield components but also identified a large number of related genetic loci through genome-wide association analysis and developed efficient prediction models such as genome selection to accelerate the breeding process [11]. At the physiological level of cultivation, the study emphasizes the interactive effects of genotype, environment, and management measures [12].

Furthermore, studies addressing climate change have highlighted the detrimental impact of stressors, such as high temperatures during flowering, on yield [13], and the integration of remote sensing and machine learning offers novel solutions for the accurate field-scale yield prediction and management [14]. Zheng et al. [4] employed correlation analysis, path analysis, multiple regression analysis, and principal component analysis to determine the key traits affecting yield, identifying plant height, seed number per pod, primary effective branch height, and primary effective branch number as the most influential factors. Guan et al. [15] carried out grey correlation degree, path analysis, and factor analysis on yield per plant and agronomic traits. They found that the strongest associations were exhibited by effective pods per plant, seeds per pod, and branch number, with significant correlations being shown by plant height, effective pods per plant, and seeds per pod with yield per plant. Wang et al. [16] found that the number of pods and secondary effective branches of the whole plant should be considered in the high-yield breeding of Brassica juncea using correlation and path analyses.

Although a lot of research has been carried out on the yield and agronomic traits of Brassica napus by predecessors, most of this research has focused on a specific region or local main cultivars. A multi-environment test is a key means to evaluate the stability and adaptability of varieties, and systematic research on multiple varieties in different ecological regions by using correlation analysis, path analysis, principal component analysis, and grey correlation analysis has rarely been reported. Therefore, in this study, 26 spring rapeseed varieties (lines) were used as materials to carry out systematic planting and phenotypic identification in five different ecological test areas, and 10 key agronomic traits, including plant height, branch, and silique traits, were determined. Based on the phenotypic data of multiple environments, a variety of quantitative genetic methods such as variation analysis, correlation analysis, and path analysis were used to quantify the genetic variation characteristics of key agronomic traits in multiple environments and to analyze the phenotypic correlation and causality between traits. The operable high-yield breeding selection strategy for spring rapeseed provides a theoretical basis and practical guidance for parent selection and offspring screening.

2. Materials and Methods

2.1. Experimental Design and Method

A total of 26 varieties (lines) were tested, which were derived from new and excellent varieties bred by 9 key breeding units in China in recent years. The genetic background of the varieties was clear, and the purity was reliable. They were not improved by transgenic or gene editing technology. Qingza No.5 was used as the control, and the specific name and source are shown in

Table 1. Information table of tested varieties.

Variety Origin	Variety Name
Shanxi Hybrid Rapeseed Research Center	Qinyou1618, Qinzayou109
Oil Crops Research Institute, Chinese Academy of Agricultural Sciences (OCRI CAAS)	22Zhongyou10, 21BP191
Sichuan Academy of Agricultural Sciences	Chuanyou117, Chuanyou228
Qinghai University	QH3365, QH33, QH403, Chunyou267, Chunyou254
commercially purchased	Qingza5(CK)
Hubei Kangnong Seed Co., Ltd.	KR2401, KR2402
Hunan Agricultural University	Xiangzayou20, Xiangyou078
Huazhong Agricultural University	Y43024, Y43039, Y90466, 23Min1, 23Min4, Huaruiyou704, Huaruiyou706
Gansu Agricultural University	Yunyouza15, 2019qw-1, L737

.

2.1.1. Experimental Site

In this experiment, rapeseed was sown in March 2024 and harvested in August. In order to clarify the application potential of the research results in this region and similar ecological regions, a total of 5 sites were involved in the experiment, and all of them were the main rapeseed producing areas in Gansu Province, including Xijiazhuang Village (L1), Yeliguan Town, Lintan County, Gannan Prefecture, Gansu Province, with an altitude of 2280 m, and longitude and latitude of (34.404581, 103.661957); Qianjin Village (L2), Xinzhuang Township, Hezheng County, Linxia Hui Autonomous Prefecture, Gansu Province, with an altitude of 2300 m, and longitude and latitude of (35.342069, 103.261545); Maying Village (L3), Damaying Town, Shandan County, Zhangye City, Gansu Province, with an altitude of 2456 m, and longitude and latitude of (38.080157, 101.352855); Yongquan Village (L4), Fengle Town, Minle County, Zhangye City, Gansu Province, with an altitude of 2103 m, and longitude and latitude of (38.404888, 100.922104); and Gaji Village (L5), Liugou Township, Jishishan County, Linxia Hui Autonomous Prefecture, Gansu Province, with an altitude of 2320 m, and longitude and latitude of (35.659607, 102.877307).

2.1.2. Experiment Design

A randomized block design with 3 repetitions was implemented. Each plot covered an area of 20 m² and was separated from the next by a walkway. There were also more than four rows of protection around the test site. The fertility level was medium, the land was leveled, and irrigation and drainage were convenient. Field management was carried out according to the local field production and cultivation management level. During the whole growth period of rape, irrigation was performed 1–2 times, and weeding was performed 1–2 times according to the actual local situation. The disease was mainly controlled by Sclerotinia sclerotiorum, and the pest was mainly controlled by Phyllotreta striolata and Aphid. According to the local situation, base fertilizer borax was applied at a rate of approximately 15 kg/ha. Seeds without any treatment, artificial sowing, timely sowing, timely seedling after emergence, or seedlings had a density of 300,000 plants/ha.

2.2. Traits Investigation and Methods

At the maturity stage, 10 representative plants were randomly selected from each plot, and the following 10 traits were measured [17]:

(1)

Plant height (PH): the length from the cotyledon node to the top of the main inflorescence (cm).

(2)

Branches position (BP): the height from the cotyledon node to the lowest effective branch (cm).

(3)

Number of primary effective branches (FBN): The number of primary effective branches on the main inflorescence.

(4)

Effective length of main inflorescence (MRL): the length of pods on the main inflorescence (cm).

(5)

Effective pod number of main inflorescence (MRS): The total number of effective pods on the main inflorescence.

(6)

Effective pod number of whole plant (TPS): The total number of effective pods per plant.

(7)

Number of seeds per pod (SPS): 20 pods were randomly selected to calculate the average number of seeds per pod.

(8)

Silique length (SL): the same as the average length (cm) of the above 20 siliques.

(9)

Thousand grain weight (TSW): Thousand grain weight (TSW) was measured by thousand-grain method, and the average value (g) was repeated three times.

(10)

Yield per plant (YPP): the total weight of all grains per plant (g).

2.3. Data Analysis Methods

The coefficient of variation, grey correlation degree, and path coefficient were calculated using Microsoft Excel 2021 (version 16, Microsoft Corporation, Rodmond, WA, USA) software. The correlation and principal components were analyzed using IBM SPSS Statistics 20.0 (IBM Corporation, Armonk, NY, USA) and MetaboAnalyst 6.0 online platform (www.metaboanalyst.ca, accessed on 18 November2025, Version 2024). The path diagram was drawn using draw.io (version 29.0.3, JGraph Ltd., Northampton, Northamptonshire, UK).

3. Results

3.1. Variation Analysis of Agronomic Traits and Yield Traits

In order to quantify the degree of phenotypic variation of each trait in the test population, the coefficients of variation (CVs) of the agronomic traits of 26 varieties (lines) in five test sites were calculated. The results showed that (

Table 2. Variation analysis of the agronomic traits of different rapeseed varieties (lines).

Items	PH	BP	FBN	MRL	MRS	TPS	SPS	SL	TSW	YPP
Minimum value	190.70	80.00	17.00	79.90	78.00	935.00	30.00	9.70	6.60	24.60
Maximum value	12.00	16.00	1.00	33.00	21.00	100.00	14.00	4.40	2.68	14.80
Mean value	138.42	41.71	5.04	50.19	43.63	259.42	21.82	7.11	3.93	19.77
standard deviation	39.10	9.19	1.20	0.71	3.54	43.42	3.54	0.21	0.28	0.85
Coefficient of variation (%)	28.25	22.04	23.87	1.41	8.10	16.74	16.20	2.98	7.20	4.29

Note: PH represents plant height; BP represents branches position; FBN represents number of primary effective branches; MRL represents effective length of main inflorescence; MRS represents effective pod number of main inflorescence; TPS represents effective pod number of whole plant; SPS represents number of seeds per pod; SL represents silique length; TSW represents thousand grain weight; YPP represents yield per plant. The same as below.

, Figure 1) the coefficients of variation of the agronomic traits of different B.napus varieties in different test sites were different, which were 1.41–28.25%, and the coefficient of variation of PH was the highest (28.25%). Followed by FBN (23.87%), BP (22.04%), TPS (16.74%), SPS (16.20%), MRS (8.10%), TSW (7.20%), YPP (4.29%), SL (2.98%), and MRL (1.41%).

PH, BP, and FBN have high coefficients of variation, which indicate rich phenotypic diversity within the population and sensitivity to environmental or genetic regulation. This also suggests substantial potential for genetic improvement and selection. TPS and SPS had moderate coefficients of variation, reflecting that they have a considerable variation basis and maintain certain stability in the population, which is an important selection target for both yield potential and trait stability. In contrast, MRL, MRS, SL, TSW, and YPP display low coefficients of variation, suggesting high genetic stability and minimal variability within the population. Notably, YPP, as the final target trait, exhibits the least variability, and this may be attributed to its regulation by multiple relatively stable traits or strict control under experimental conditions.

3.2. Correlation Analysis Between Agronomic Traits and Yield

In order to clarify the correlation model between agronomic traits and their correlation with yield, the correlation analysis between agronomic traits and yield was carried out in this study. The results are presented in a correlation coefficient matrix (

Table 3. Correlation analysis of the agronomic traits of different rapeseed varieties (lines).

	PH	BP	FBN	MRL	MRS	TPS	SPS	SL	TSW	YPP
PH	1
BP	0.078	1
FBN	0.116	0.238 *	1
MRL	0.166	0.287 **	0.383 **	1
MRS	0.225 *	0.398 **	0.550 **	0.523 **	1
TPS	0.197 *	0.296 **	0.792 **	0.466 **	0.641 **	1
SPS	0.237 *	0.381 **	0.544 **	0.492 **	0.579 **	0.576 **	1
SL	0.084	0.078	0.029	0.071	0.088	0.027	0.146	1
TSW	−0.352 **	0.029	0.022	0.121	−0.035	−0.129	−0.14	0.044	1
YPP	0.101	0.186	0.057	0.075	0.074	0.021	0.024	0.249 *	0.17	1

Note: The table shows the Pearson correlation coefficients. * p < 0.05, ** p < 0.01.

) and visualized using a heat map (Figure 2).

The analysis revealed that there was a significant positive correlation between multiple traits of different varieties in different test areas. TPS was significantly positively correlated with BP, FBN, MRL, MRS, and SPS; MRS was also significantly positively correlated with BP, FBN, MRL, and SPS; YPP was only significantly positively correlated with SL, and positively correlated with other traits but did not reach a significant level; FBN was significantly positively correlated with MRL and SPS, significantly positively correlated with BP, and positively correlated with SL, TSW, and YPP; PH was significantly positively correlated with MRS, TPS, and SPS, and significantly negatively correlated with TSW, but had no significant correlation with other traits; BP was significantly positively correlated with MRL and SPS, and significantly correlated with FBN number.

3.3. Path Analysis of the Effect of Agronomic Traits on Yield

In order to further analyze the effect of the agronomic traits on YPP, the path analysis was carried out. As shown in

Table 4. Path coefficients of 9 main traits to YPP.

Factor	Correlation Coefficient	Direct Path Coefficient	Indirect Path Coefficient									Comprehensive Effect
			PH	BP	FBN	MRL	MRS	TPS	SPS	SL	TSW
PH	0.101	0.182	-	0.014	0.021	0.030	0.041	0.036	0.043	0.015	−0.064	0.137
BP	0.186	0.176	0.014	-	0.042	0.051	0.070	0.052	0.067	0.014	0.005	0.314
FBN	0.057	−0.214	−0.025	−0.051	-	−0.082	−0.118	−0.169	−0.116	−0.006	−0.005	−0.572
MRL	0.075	0.375	0.062	0.108	0.144	-	0.196	0.175	0.185	0.027	0.045	0.941
MRS	0.074	−0.103	−0.023	−0.041	−0.057	−0.054	-	−0.066	−0.060	−0.009	0.004	−0.306
TPS	0.021	0.043	0.008	0.013	0.034	0.020	0.028	-	0.025	0.001	−0.006	0.123
SPS	0.024	−0.051	−0.012	−0.019	−0.028	−0.025	−0.030	−0.029	-	−0.007	0.007	−0.144
SL	0.249 *	0.467	0.039	0.036	0.014	0.033	0.041	0.013	0.068	-	0.021	0.265
TSW	0.17	0.148	−0.052	0.004	0.003	0.018	−0.005	−0.019	−0.021	0.007	-	−0.065

Note: The * shows the indicates the strongest correlation.

, the direct contributions of the nine agronomic traits across 26 varieties at different test sites to YPP ranked as follows: SL (0.467) > MRL (0.375) > FBN (|−0.214|) > PH (0.182) > BP (0.176) > TSW (0.148) > MRS (|−0.103|) > SPS (|−0.051|) > TPS (0.043). SL and MRL had a significant positive effect on YPP, and SL had a significant positive correlation with YPP, indicating that they had the strongest direct positive effect on the target traits. FBN and MRS had negative direct path coefficients, indicating that they had a direct inhibitory effect on the target traits.

Among the indirect comprehensive effects of the nine agronomic traits on YPP through other traits, FBN, MRS, SPS, and TSW demonstrated negative effects, and the other traits demonstrated positive effects, followed by MRL, BP, SL, PH, and TPS.

To illustrate the causal relationships and directional effects among traits more intuitively, a path diagram (Figure 3) was developed based on path analysis. Besides the direct and indirect comprehensive effects described above, it was found that BP indirectly and positively influences yield traits via PH (0.014). This suggests that BP can promote plant growth to some extent. In addition to its direct effect, MRL also contributed positively via indirect pathways involving PH (0.066) and FBN (0.144), indicating that it enhances yield potential in conjunction with plant architecture traits.

3.4. Principal Component Analysis of Agronomic Traits Structure

Based on the principal component analysis of the agronomic traits of 26 varieties in different test sites, the agronomic traits were affected by multiple factors such as genotype, environment, and interaction, and the source of variation was complex. Therefore, it can be seen from

Table 5. Principal component analysis table.

Index	PC1	PC2	PC3
PH	0.295	0.44	−0.597
BP	0.390	0.207	0.113
FBN	0.815	0.073	−0.036
MRL	0.733	0.158	0.090
MRS	0.826	0.126	−0.081
TPS	0.838	-0.084	−0.139
SPS	0.692	-0.066	−0.151
SL	0.046	0.752	−0.113
TSW	0.049	0.227	0.859
YPP	0.036	0.715	0.252
c-value	3.453	1.417	1.103
Contribution rate (%)	34.535	14.168	11.026
Cumulative contribution rate (%)	34.535	48.702	59.728

that the cumulative contribution rate of the first three principal components was 59.73%, of which the contribution rate of principal component 1 was 34.53%, and its high-load traits included TPS, MRS, and FBN, which can be defined as ‘yield and pod-setting ability factor’. Principal component 2 is mainly related to SL and YPP, which is defined as ‘pod structure factor’; principal component 3 was mainly determined by TSW, which was defined as ‘grain weight factor’.

To clarify the overall differences in phenotypic traits among different locations, a multivariate analysis of variance (MANOVA) was first conducted using SPSS. The results indicated that the location effect had a highly significant impact on the combination of 10 traits (Pillai’s Trace = 2.371, F(40, 476) = 17.315, p < 0.001). To visualize the specific pattern of this significant difference, a principal component analysis (PCA) was performed based on the average values of each location, and a scatter plot of the experimental sites was drawn according to the principal component analysis scores of the average values of the 10 agronomic traits at each experimental site (Figure 4). The biplot (Figure 5) and the scatter plot (Figure 4) showed that the first two principal components cumulatively explained 98.5% of the total variation, with PC1 contributing 94.2% and PC2 contributing 4.3%. L1 was located at the farthest end in the positive direction of PC1, L2 was located near the positive end, and L3, L4, and L5 were all located in the negative direction of PC1, with L5 being at the farthest left. The direction and length of the arrows representing each agronomic trait indicated that the trait vectors pointing in the positive direction of PC1 were longer, which was the absolute dominant dimension for distinguishing locations. This dimension was almost entirely defined by the total number of effective pods per plant (load > 0.99), indicating that the total number of effective pods per plant was the most dominant and sensitive single trait driving the trait differentiation among locations. This result was consistent with the conclusion from the previous principal component analysis based on the entire data set that the “yield and pod-setting ability factor” was the most important dimension.

3.5. Grey Correlation Analysis of Agronomic Traits and Yield

To further quantify the correlation degree and stability of each agronomic trait with YPP according to the grey system theory, the grey system theory was used to analyze the correlation degree of nine agronomic traits and the yield of 26 different rapeseed varieties. The phenotypic stability parameters of each trait were calculated. The results are shown in

Table 6. Grey Correlation Analysis between Agronomic Traits and Crop Yield.

Trait	Grey Correlation Degree	Sorting	Standard Deviation	Coefficient of Variation (%)	Extreme Difference	Stability Level	Prominence
PH	0.834	2	0.101	12.104	0.320	low	ab
BP	0.764	7	0.135	17.698	0.418	low	cd
FBN	0.718	8	0.168	23.371	0.660	low	d
MRL	0.847	1	0.099	11.732	0.312	low	a
MRS	0.793	6	0.126	15.871	0.486	low	c
TPS	0.676	9	0.164	24.314	0.607	low	d
SPS	0.817	5	0.136	16.585	0.495	low	bc
SL	0.825	4	0.120	14.582	0.396	low	b
TSW	0.833	3	0.107	12.856	0.360	low	ab

Note: The closer the grey correlation degree is to 1, the closer the correlation between the trait and the target trait is. Different letters in the same column indicated that there were significant differences in the correlation between traits at the p < 0.05 level.

and Figure 6.

The grey correlation analysis revealed that MRL exhibited the highest correlation degree (0.847), significantly exceeding that of most other traits. This indicates that MRL showed the strongest synchrony with the dynamic changes in YPP and is a key index influencing the target traits. The correlation between PH (0.834) and TSW (0.833) followed, and the difference between the two was not significant, which together constituted a high correlation trait group. The correlation degree of TPS was the lowest (0.676), and it was significantly different from the traits of the high association group, indicating that the degree of synergistic change with the target traits was weak.

The coefficients of variation of all the traits were greater than 0.10, and the stability level was ‘low’, indicating that all the traits showed high phenotypic plasticity in the environment. Among them, the coefficients of variation of the number of FBN and TPS were the highest (23.371%, 24.314%), their ranges were the largest, their stabilities were the worst, and they were the most sensitive to environmental changes. In contrast, MRL had the lowest coefficient of variation (0.117) and the best stability, though it still fell within the “low” stability category.

4. Discussion

In this study, we systematically analyzed the structural characteristics of agronomic traits in spring rapeseed and their contributions to yield formation using multiple analytical approaches. PH, BP, and FBN showed high phenotypic variation, indicating that these traits were greatly affected by genetic and environmental interactions and had high improvement potential [18]; therefore, the yield potential can be further explored through ideal plant type design. The moderate variation of TPS, which has both selection space and certain stability, is an important goal to coordinate yield and adaptability. YPP has the smallest variation, suggesting that it is regulated by multiple relatively stable traits, or there is a compensation effect between traits [19], which may be an important reason for the current population yield performance to be balanced. Therefore, in high-yield breeding programs for rapeseed, prioritizing improvements in TPS may represent a more effective strategy than focusing solely on thousand-seed weight (TSW).

Correlation analysis showed that there was a significant positive correlation between FBN and TPS, MRS, and SPS, which was consistent with the previous research conclusions on rapeseed [4,20,21,22]. In breeding practice, using FBN as a key selection index is expected to achieve synergistic improvement of effective pod number and SPS, break the constraints between traits, and then improve YPP. There was a significant positive correlation among TPS, MRS, and SPS, indicating that the yield components were mainly synergistic rather than trade-off, which means that it is feasible to increase the number of pods and SPS simultaneously through genetic improvement or cultivation management. High-yield breeding can be achieved by taking into account such traits and the synergy between traits [23,24,25], rather than seeking extreme breakthroughs in a single trait. It is worth noting that YPP is only significantly positively correlated with SL, which is consistent with previous research results [6,15,26,27]. This indicates that direct selection for yield alone may be inefficient, and greater emphasis should be placed on its key component traits. There was a significant negative correlation between TSW and PH, indicating that the ideal plant type should maintain a reasonable height, ensure the ventilation and light transmittance of the population and lodging resistance, and optimize the transport efficiency of assimilates to grains.

The path analysis further revealed the direct and indirect contribution of each trait to yield, and the pod length had the greatest direct positive contribution to yield per plant, which should be used as a key selection index in breeding. The MRL also showed a significant positive effect, while the direct effect of the FBN was negative, suggesting that the increase in the number of branches may lead to competition within the plant, thereby limiting per-plant productivity. Therefore, high-yield breeding should focus on the optimization of branch quality, especially the development of main inflorescence branches. Wang et al. [20] showed that the direct effects of plant height, yield per plant, pod length, pod grain number, and 1000-grain weight on yield were positively correlated, and the direct effect of plant height on plot yield was the largest, which was consistent with the conclusion of this study. The difference in the order of trait importance was mainly due to the specificity between different ecological types and genetic populations.

The multi-level principal component analysis showed that the cumulative contribution rate of the first three principal components based on variety × location data was 59.73%, and the cumulative contribution rate of the first two principal components of PCA based on location average was as high as 98.5%, which was not contradictory to the above results but provided complementary information. There were significant environmental gradients in the average performance of agronomic traits in the five experimental sites selected in this study, among which PC1 (contribution rate of 94.2%) was the dominant dimension driving site differentiation. The scatter plot clearly divided the test sites into high-value groups (L1 and L2) and low-value groups (L3, L4, and L5), and the number of effective pods per plant was the dominant trait driving phenotypic differentiation between different test sites (the contribution rate of PC1 was 94.2%), indicating that environmental factors mainly regulated the final yield performance by affecting pod-setting ability, which provided a clear direction for ecological adaptive breeding: when breeding wide-adaptability varieties, attention should be paid to the stability and response ability of the number of effective pods per plant in different environments.

Grey correlation analysis revealed that MRL had the closest association with YPP and exhibited relatively high phenotypic stability, making it a suitable key index for synergistic high-yield selection. This is different from the research results of Wang et al. [28] and Zhang et al. [29]. This is caused by the large differences in environmental effects, genetic background, compensation between traits, and other factors. In the varieties with high pod numbers, the contribution rate of the main inflorescence may be relatively reduced, and in this experimental group, the direct contribution of the pod-setting ability of the main inflorescence to the overall yield is more prominent. Plant height, 1000-grain weight, and silique length constitute a highly correlated trait group, suggesting that the synergistic improvement of these traits should be paid attention to in breeding, so as to avoid the over-selection of a single trait from destroying the overall balance.

5. Conclusions

PH, BP, and FBN exhibit high potential for genetic variation and can serve as key targets for plant architecture improvement; SL contributes the most directly to YPP and represents a core selection trait for high-yield breeding in rapeseed; TPS is a key trait that drives phenotypic differences between environments and should be used as a key response index in breeding in different ecological regions; and MRL is most closely related to yield and has good stability, which is suitable as a key reference trait for high-yield collaborative selection. It is recommended to adopt a breeding strategy of “quality first, coordinated improvement,” focusing on SL, TSW, and MRL, coordinating the optimization of branching structure and pod-setting ability, and achieving efficient integration of yield-related traits through multi-environment evaluation.