Field pea (Pisum sativum L.) is a cold-season legume belonging to family Leguminosae (2n =2x=14), is a vital protein source, carbohydrates, vitamins and minerals for human as well as animal nutrition. Beyond its nutritional importance, field pea contributes to sustainable agriculture through biological nitrogen fixation, thereby enhancing soil fertility and reduces the need for nitrogen fertilizers (Smýkal et al., 2012). Temperature Abiotic stress is among the key environmental factors that significantly affect field pea productivity. Delayed sowing often exposes the crop to elevated temperatures during flowering and pod development, adversely affecting pollen viability, pod set and seed filling, ultimately reducing yield (Devi et al., 2023). Therefore, developing high-yielding and heat-tolerant cultivars is a most important breeding goal. Crop improvement requires genetic diversity, since it enables breeders to identify and exploit desirable traits (Begna, 2024). Multivariate statistical approaches such as Mahalanobis’ D² statistics and PCA are widely used to measure genetic divergence and determine the characteristics that lead to variability. D² analysis groups genotypes based on genetic distance, while PCA summarizes complex datasets and reveals the major sources of variation within a population (Mahalanobis, 1936; Rao, 1952; Jolliffe, 2002). Therefore, the current research was undertaken to examine genetic diversity among field pea genotypes across TS and LS conditions using Mahalanobis’ D² statistics and PCA, with the objective of identifying diverse and promising genotypes targeted for future crop improvement programs.

This study was carried out across two successive Rabi seasons (2023-24 and 2024-25) at the research farm of Genetics and Plant Breeding, Acharya Narendra Deva University of Agriculture and Technology, Kumarganj, Ayodhya, Uttar Pradesh, India. During Rabi 2023-24, ten diverse field pea genotypes, namely Pant P 545, Pant P 550, RFP 2020-2, KPMR 907, Pant P 554, IPF 22-22, IPF 22-24, HFP 1960, HFP 1961 and VL 42, were subjected to crossing in a half-diallel fashion excluding reciprocals (Griffing, 1956) to generate 45 F₁ hybrids. In Rabi 2024-25, the 45 F₁ hybrids along with their 10 parents were subjected to evaluation under timely-sown and late-sown conditions in three replications using randomized block design. Each genotype was planted in a single row at a spacing of 45 cm × 10 cm and the recommended agronomic practices were followed throughout the crop season. Data collection was carried out on five randomly chosen plants for eleven traits, namely days to 50 per cent flowering, days to maturity, primary branches/plant, secondary branches/plant, plant height, pods/plant, seeds/pod, biological yield/plant, harvest index, seed index and seed yield/plant. The average values were utilized for multivariate analysis using R software. Genetic divergence was estimated using Mahalanobis D² statistics (Mahalanobis, 1936) and Genotypes were classified into clusters using Tocher’s method (Rao, 1952). PCA was performed on standardized data to determine the principal sources of variation among genotypes. components with eigenvalues exceeding unity were retained according to Kaiser’s rule (Kaiser, 1960) and genotype relationships were visualized using principal component scores.

Genetic divergence analysis
 
Genetic diversity among 45 F₁ hybrids and 10 parental lines of field pea was evaluated using Mahalanobis D² statistics. The 55 genotypes were organized into six distinct clusters under both timely and late sown conditions (Table 1 and 2; Fig 1 and 2), indicating substantial genetic variability among the studied material. Under timely sowing, Cluster III represented the largest cluster with 15 genotypes, whereas Cluster IV accounted for the smallest with 5 genotypes. Under late sowing, Clusters I and III each contained 11 genotypes, while Cluster II was the smallest with 7 genotypes. The variation in cluster composition across environments suggests that environmental conditions influenced the expression of quantitative traits and altered genetic relationships among genotypes. Similar findings have been reported by Singh et al., (2021) and Guruprasad et al., (2021).










 
Inter and intra cluster distance
 
Intra-cluster and inter-cluster distances among the six clusters are available in Table 3. In timely sown conditions, cluster V exhibited the high intra-cluster distance (26.76), whereas in late sown conditions, cluster II recorded the high intra-cluster distance (30.28), suggesting greater variability under clusters II and V. The largest inter-cluster value was observed between clusters V and VI (51.12) under timely sowing and between clusters II and V (63.30) under late sowing, reflecting the highest genetic divergence among these cluster combinations. In contrast, the lowest inter-cluster value was recorded between clusters III and VI under both environments, indicating closer genetic relationships among the genotypes. Overall, inter-cluster distances were generally greater than intra-cluster distances, suggesting substantial genetic diversity among clusters. Therefore, crosses involving genotypes from clusters V and VI under timely sowing and clusters II and V under late sowing may be promising for generating broad variability and superior segregants in breeding programmes.


 
Cluster means
 
Substantial variability in cluster means across different traits under timely and late-sown conditions provided further evidence of diversity. Cluster mean values for all eleven characters in TS and LS environments are given in Table 4. As revealed by Table 4, under timely sown conditions, cluster I showed the largest mean value for PH (113.41 cm) and BYPP (80.62 g) and smallest mean values for DTF (67.50 days). Cluster II exhibited the largest mean values for DTM (112.27 days), harvest index (39.83%) and SYPP (31.65 g). Cluster III showed high mean values for PB (2.87), SB (4.86), PPP (29.51), SPP (6.18), SI (19.41), HI (41.70%) and SYPP (35.32 g). Cluster IV recorded the highest value for PH (114.04 cm) but lowest for PB (2.35), SI (17.22), BYPP (66.32 g), HI (35.21%) and SYPP (23.20 g). Cluster V showed highest values for DTF (71.94 days) and DTM (111.11 days), while exhibiting lower values for PH (93.46 cm), SB (3.63), PPP (22.80), BYPP (64.25 g) and SYPP (23.69 g). Cluster VI exhibited the high values for PB (3.13), SB (5.35), PPP (29.96), SI (19.56), BYPP (93.69 g) and SYPP (35.53 g), while lowest values for DTM (106.04 days). Under late sown conditions, cluster I exhibited one of the highest mean values for HI (37.98%). Cluster II showed the highest mean values for DTF (67.76 days) and DTM (104.85 days), but recorded the lowest values for PH (92.14 cm), PB (2.18), SB (2.97), PPP (18.63), SPP (5.35), SI (15.50), BYPP (45.57 g) and SYPP (15.48 g). Cluster III showed the lowest values for DTF (62.87 days) and DTM (99.69 days), while maintaining comparatively higher values for PPP (25.85), SI (16.65), BYPP (69.67 g) and SYPP (23.95 g). Cluster IV was among the superior clusters for SPP (5.95) and BYPP (68.52 g). Cluster V emerged as the most promising cluster under late sowing, exhibiting the largest values for PH (107.76 cm), PB (2.93), SB (4.10), PPP (26.20), SI (17.69), BYPP (74.64 g) and SYPP (27.32 g). Cluster VI showed moderate performance for most traits but recorded the lowest harvest index (30.77%).


       
The results showed that each cluster exhibited strengths for specific characteristics under both TS and LS conditions. However, clusters III and VI under timely sowing and cluster V under late sowing showed better overall performance for seed yield/plant and most of its contributing traits. Therefore, genotypes belonging to these clusters can be considered promising candidates for use as parents in breeding programme. Crossing these genotypes with those from genetically diverse clusters having high inter-cluster distances may help to generate greater variability and increase the chances of obtaining superior recombinants in subsequent generations.
 
Principal component analysis (PCA)
 
Principal Component Analysis (PCA) was conducted to determine the principal sources of variation and traits contributing to genetic diversity among 10 parental lines and 45 F₁ hybrids of field pea under TS and LS conditions. The eigenvalues, factor loadings and percentage of variance explained by the principal components are presented in Table 5, while the scree plots and PCA biplots are shown in Fig 2-5.








       
The initial four principal components exhibited eigenvalues more than one under both timely and late sown conditions and contributed largely to the total variation observed. Under timely sowing, PC1-PC4 noted for 80.51% of the cumulative variation, while under late sowing they explained 78.18% to the overall variation. The scree plots showed the eigenvalues after the first component decline syarply, indicating first four components captured most of the variation present in the dataset. Similar observations have been noted by Anand et al., (2025) and Pratap et al., (2024).
       
The factor loadings revealed that PC1 was predominantly related with seed yield/plant and its governing traits in both environments. Seed yield/plant showed the highest positive loading under timely (0.952) and late sown (0.957) conditions, followed by BYPP, PPP, PB and SB, indicating their major contribution to genetic divergence. In contrast, DTF and DTM exhibited negative loadings. PC2 was mainly influenced by phenological traits and harvest index, while PC3 was largely associated with harvest index under both environments. The variation in trait loadings across principal components reflects the differential contribution of traits to genetic variability under contrasting sowing conditions.
       
The PCA biplot under timely sown (TS) conditions showed that PC1 and PC2 together explained 60.88% of total variability. Traits such as seed yield/plant (SYPP), biological yield/plant (BYPP), pods/plant (PPP), primary branches/plant (PB), secondary branches/plant (SB), seeds/pod (SPP) and seed index (SI) were closely grouped and oriented in the same direction, indicating strong positive associations among these traits. In contrast, DTF and DTM were positioned opposite to the yield-related traits, suggesting a negative association between crop duration and productivity. Parental lines HFP 1961, KPMR 907, Pant P 550, VL 42 and RFP 2020-2 were located away from the origin, indicating greater genetic divergence. Hybrids Pant P 545 × IPF 22-22, Pant P 545 × VL 42, RFP 2020-2 × IPF 22-24 and IPF 22-22 × IPF 22-24 were associated with yield-related vectors, suggesting their potential for yield improvement. Under LS conditions, PC1 and PC2 accounted for 57.24% of the overall variation. A similar clustering of SYPP, BYPP, PPP, PB and SB was observed, indicating strong positive relationships among these traits. The separation between yield-related and phenological traits was more pronounced under LS conditions, highlighting the importance of earliness under terminal heat stress. Parents HFP 1961, KPMR 907, Pant P 550 and RFP 2020-2 occupied extreme positions, while hybrids IPF 22-22 × IPF 22-24, Pant P 545 × IPF 22-22 and RFP 2020-2 × IPF 22-24 were closely associated with yield-enhancing traits. These results revealed that SYPP, BYPP, PPP, PB and SB were the major contributors to variability under both environments. Similar observations have been noted by Patel et al., (2023), Pratap et al., (2024) and Anand et al., (2025), demonstrating the effectiveness of PCA in identifying genetically diverse and high-yielding breeding materials.

This study demonstrated considerable genetic divergence within the evaluated material of field pea under TS and LS conditions. Genotypes were grouped into six clusters based on the D² analysis, while PCA revealed that the first four PC explained greater than 78% of the overall variability under both environments. Clusters III and VI under timely sowing and cluster V under late sowing exhibited superiority for seed yield and related traits. Among the parental lines, IPF 22-22 and IPF 22-24 performed well under late sowing, indicating their potential as heat stress-tolerant parents. The hybrids IPF 22-22 × IPF 22-24 and RFP 2020-2 × IPF 22-24 were identified as promising genotypes due to their strong association with yield and yield-contributing traits. These genotypes may be strategically exploited in breeding programmes focused on developing high-yielding and heat stress-tolerant field pea cultivars.

The present work was carried out with the support of the Department of Genetics and Plant Breeding, Acharya Narendra Deva University of Agriculture and Technology, Kumarganj, Ayodhya, Uttar Pradesh, India.

The authors confirm that no conflicts of interest are associated with the publication of this article. The study was conducted independently, without any external funding or sponsorship influencing its design, data collection, analysis, decision to publish, or manuscript preparation.