Genetic variability, trait association and divergence analysis of ridge gourd [Luffa acutangula (L.) Roxb.] genotypes for growth, flowering and yield related traits

Introduction

Ridge gourd [Luffa acutangula (L.) Roxb.], a warm season diploid (2n = 2x = 26) native to the Indian subcontinent, is cultivated for its immature fruit as well as for its high nutritional and therapeutic properties (Shendge and Belemkar, 2018) [1] . At the national level, the crop area is approximately 24,800 hectares with an annual production of 3,16,925 metric tonnes (MT). However, the 3,210 hectares of crop land in Karnataka only produced 8.94 MT/ha (Directorate of Horticulture, 2023) [2] , which is significantly lower than yield potential for the crop. India's rich cucurbit genetic diversity (Bose and Som, 1986) [3] offers the potential to close this gap, but fruit yield in ridge gourd is governed by numerous interrelated component traits comprising vegetative growth, reproductive behaviour, fruit and seed development, accordingly germplasm must be screened across a broad trait complex rather than for just yield. Partitioning phenotypic variability into its genotypic and environmental components through genotypic and phenotypic coefficients of variation, heritability and genetic advance helps identify traits amenable to direct selection (Burton, 1952; Johnson et al., 1955) [4,5] and correlation studies reveal patterns of trait association and path coefficient analysis partitions these correlations into direct and indirect effects, permitting more reliable indirect selection (Dewey and Lu, 1959) [6] . Principal component analysis summarizes multi-trait variability along a few orthogonal axes and thereby facilitates identification of divergent, complementary genotypes (Iezzoni and Pritts, 1991 [7] ; Mohammadi and Prasanna, 2003 [8] ), while standardized squared Euclidean distance together with Tocher's clustering provides a complementary, quantitative basis for grouping genotypes according to overall genetic distance and for identifying divergent parental combinations for hybridization (Singh and Chaudhary, 1985) [9] . Combined evaluations of this kind have been reported in ridge gourd by Shah and Kale (2002) [10] , Hanumegowda (2011) [11] , Bhargava et al. (2017) [12] , Shodhan Jain et al. (2023) [13] and Amulya et al. (2025) [14] and in other cucurbits. However, studies comparing released varieties alongside local landraces within a single evaluation panel remain scarce for this crop. Most earlier studies have evaluated either released varieties or local collections but not both together, which limits the scope for identifying complementary parental combinations for breeding programmes. The present study therefore evaluated eight ridge gourd genotypes for growth, flowering and yield related traits, together with their genetic variability, trait association and multivariate diversity, with the objective of identifying superior genotypes and reliable selection criteria for cultivation and breeding.

Material and Methods

Field evaluation was conducted in kharif 2025 at the Agricultural Research Station, Kallolli, University of Agricultural Sciences, Dharwad, Karnataka (16°08'N latitude, 74°49'E longitude, 534 m above sea level). The soil at the experimental site was sandy to medium black, with a pH of 7.5 to 8.0. The climate was semi-arid, with an average annual rainfall of around 600 mm. During the crop season (August-November 2025), the mean maximum and minimum temperatures ranged between 28-30°C and 16-21°C, respectively. Eight genotypes (Arka Sumeet, Arka Prasanna, Pusa Nasdar, Kashi Shivani, Kashi Nanda, Kashi Khushi, Mallapur Local and Gujanatti Local) were evaluated in a randomized complete block design with three replications, at a spacing of 120 cm × 90 cm and plot size of 10.8 m². Vine length and number of primary branches (at 45 and 90 days after sowing), days to first male and female flower, days to 50% flowering, sex ratio (male to female flower ratio), number of fruits per vine, fruit weight, fruit yield (kg per vine and t/ha), number of seeds per fruit, seed yield and test weight (100-seed weight) were recorded on five randomly tagged plants per replication. Seed of the released cultivars was obtained from ICAR-Indian Institute of Horticultural Research, Bengaluru; ICAR-Indian Agricultural Research Institute, New Delhi; and ICAR-Indian Institute of Vegetable Research, Varanasi, while local landraces were collected from farmers in the Northern transition zone of Karnataka. Crop management followed the recommended package of practices for ridge gourd in the region, including basal application of farmyard manure at 20 t/ha and fertilizer at 100:50:50 kg NPK per hectare, with 50% of the nitrogen and the full dose of phosphorus and potassium applied basally and the remaining nitrogen applied in two split doses at 30 and 60 days after sowing. Irrigation was provided as required to maintain adequate soil moisture.

Statistical analysis

Analysis of variance was carried out following Panse and Sukhatme (1985) [15] . Genotypic and phenotypic coefficients of variation were calculated following Burton (1952) [4] as:

Broad-sense heritability was calculated following Lush (1940) [16] as:

Phenotypic variance

Genetic advance was calculated following Johnson et al. (1955) [5] as:

wherein k is the standardized selection differential (k = 2.06 at 5% selection intensity). Genetic advance expressed as percentage of mean was calculated as:

Genotype mean values were used to calculate Pearson's correlation coefficient, applied both to examine trait associations and as the basis for principal component analysis as:

Before path coefficient analysis, the predictor traits were screened for multicollinearity using the variance inflation factor:

wherein R i ² is the coefficient of determination obtained by regressing each predictor trait on the remaining predictor traits and a subset of five traits with variance inflation factor below 10 was retained for path analysis. Path coefficient analysis was carried out following Dewey and Lu (1959) [6] , partitioning the correlation of each trait with fruit yield per vine as:

𝑘 wherein p ij is the direct effect of trait i on fruit yield per vine and the summation term gives the indirect effects operating through all other traits k, with the residual effect calculated as

Genetic divergence among the genotypes was assessed using the standardized squared Euclidean distance:

for each pair of genotypes i and j, wherein Z ik is the standardized value of trait k in genotype i, computed across all 15 traits. This approach removes scale differences among traits while preserving their relative contribution to overall divergence, thereby providing a robust basis for genotype clustering. Genotypes were grouped into clusters following Tocher's method (Rao, 1952; Singh and Chaudhary, 1985) [17,9] , with the cluster-formation threshold defined as the mean of each genotype's minimum squared distance to its nearest neighbour; genotypes were added sequentially to a cluster only while their average distance to the existing cluster members remained below this threshold. All analyses were performed using R software (R Core Team, 2024) [18] . A composite rank-sum multi-trait selection index was additionally computed as:

summed across traits t, by ranking all eight genotypes from 1 (best) to 8 (poorest) for seven yield-determinant traits, namely fruit yield per vine, fruit yield per hectare, seed yield, fruits per vine, fruit weight, test weight and vine length at 45 DAS and summing the ranks across traits for each genotype. A lower rank-sum indicates superior overall performance across this set of traits.

Results and Discussion

Plant growth, flowering and yield performance Analysis of variance revealed significant genotypic differences for all 15 quantitative traits studied (Table 1 [19] , Rathore et al. (2017) [20] and Sushmitha et al. (2019) [21] . Days to first male flower, first female flower and 50% flowering ranged from 35.83 to 42.00, 41.00 to 51.50 and 43.33 to 52.00 days, respectively, while sex ratio ranged from 20.84 in Kashi Khushi to 24.48 in Arka Sumeet. Kashi Khushi with combined earliest flowering and the lowest sex ratio with the lowest fruit yield, indicating that a favourable sex ratio alone cannot compensate for limited vegetative growth, a pattern consistent with the findings of Shah and Kale (2002) [10] and Karuppaiah et al. (2005) [22] . Fruit yield varied from 1.71 t/ha in Kashi Khushi to 26.26 t/ha in Gujanatti Local, followed by Mallapur Local (20.18 t/ha) and Arka Prasanna (17.10 t/ha). The superior genotypes generally combined a higher number of fruits per vine with greater fruit weight, more seeds per fruit and higher seed yield, results are in agreement with Harika et al. (2012) [23] , Varalakshmi et al. (2015) [24] and Sushmitha et al. (2019) [21] . Considerable variation in fruit number, fruit weight, fruit yield and seed-related traits has likewise been reported in ridge gourd by Ananthan and Krishnamoorthy (2017) [25] , Bhargava et al. (2017) [12] and Shodhan Jain et al. (2023) [13] , supporting the usefulness of these traits for genotype discrimination and selection. Similarly wide variability for yield and yield-attributing traits, together with good prospects for genotype selection, has recently been reported for ridge gourd germplasm accessions by Jha et al. (2025) [26] and for other cucurbits such as snap melon by Kumar et al. (2026) [27] .

Genetic variability parameters

For every trait studied, the phenotypic coefficient of variation exceeded the genotypic coefficient of variation (Table 2), indicating some environmental contribution to phenotypic expression, although the difference between the two was relatively small for several yield traits. The narrow gap between genotypic and phenotypic coefficients of variation for fruit yield per hectare (61.29 and 62.11%, respectively) and for seed yield (54.12 and 55.98%, respectively) suggested minimal environmental influence on these traits, reinforcing their reliability for direct selection. In contrast, the wider gap observed for sex ratio (genotypic coefficient of variation 4.45% against phenotypic coefficient of variation 7.13%) indicated substantial environmental modulation, consistent with the known sensitivity of sex expression to environmental factors in cucurbits. Fruit weight, fruit yield per vine, fruit yield per hectare and seed yield showed high heritability (above 85%) together with high genetic advance expressed as percentage of mean (84.1 to 124.58%), indicating substantial genetic variation and favourable prospects for selection. Following the classification of Johnson et al. (1955) [5] , these traits fell into the category of high heritability combined with high genetic advance, suggesting a good response to selection. Similar combinations of high genotypic and phenotypic coefficients of variation, heritability and genetic advance have been reported in ridge gourd by Koppad et al. (2015) [19] , Ananthan and Krishnamoorthy (2017) [25] , Narasannavar et al. (2017) [28] , Shodhan Jain et al. (2023) [13] , Subhash et al. (2024) [29] and Rai et al. (2025) [30] . In contrast, sex ratio combined relatively low heritability (38.92%) with low genetic advance (1.29%), suggesting greater environmental influence and a less predictable response to direct selection. Variation in heritability and genetic advance among flowering and sex expression traits has also been reported previously by Varalakshmi et al. (2015) [24] and Deokar et al. (2024) [31] .

Correlation and path analysis

Phenotypic correlation among the traits studied (Table 3 and Fig. 1) showed that fruit yield per vine was most strongly associated with fruit weight (r = 0.96), seed yield (r = 0.94) and vine length at 45 days after sowing (r = 0.92). Positive associations of fruit yield with fruit weight, fruits per vine, vine length and branching have been reported previously by Varalakshmi et al. (2015) [24] , Ananthan and Krishnamoorthy (2017) [25] , Narasannavar et al. (2017) [28] and Deokar et al. (2024) [31] . Days to flowering traits formed a comparatively distinct group, consistent with earlier reports of weaker or negative associations between earliness traits and yield by Karuppaiah et al. (2005) [22] . Path coefficient analysis using the five-trait subset (Table 4) revealed that sex ratio (0.506), fruits per vine (0.496) and seed yield (0.485) exerted the greatest direct effects on fruit yield per vine. Notably, Varalakshmi et al. (2015) [24] , evaluating ten ridge gourd genotypes, reported fruit weight as the trait with the highest direct effect (0.847) on fruit yield per hectare, whereas in the present study fruit weight was not retained in the final path model owing to screening for multicollinearity and sex ratio, fruits per vine and seed yield emerged instead as the principal direct contributors. This difference may be attributed to the different genotype panels used, the inclusion of local landraces alongside released varieties in the present study and the multicollinearity screening applied, which removed redundant predictor traits. Ananthan and Krishnamoorthy (2017) [25] and Manikandan et al. (2017) [32] reported similar positive direct contributions of fruits per vine and fruit weight, though the extent and relative significance of the direct effects differed among studies. In a larger, companion evaluation of 27 ridge gourd genotypes by Harsha et al. (2026) [33] likewise identified average fruit weight and number of fruits per vine as the traits with the highest direct effects on yield, lending independent support to the central role of these two traits, even though their relative ranking against sex ratio and seed yield differed from that observed in the present panel. Primary branches at 90 days after sowing and days to first female flower acted mainly through indirect pathways. The path model accounted for 99.6% of the variation in fruit yield per vine (residual effect -0.065). .00 VL45/VL90-Vine length 45/90 DAS; B45/B90-Primary branches 45/90 DAS; MF/FF-Male/female flowers; D50F-Days to 50% flowering; SR-Sex ratio; F/V-Fruits per vine; FW-Fruit weight; FY/V-Fruit yield per vine; FY/ha-Fruit yield per hectare; S/F-Seeds per fruit; SY-Seed yield; TW-Test weight. Tabulated values correspond to the heat map shown in Fig. 1. *, ** Significant at 5% and 1% probability levels, respectively (critical r at n=8, df=6: r0.05=0.707, r0.01=0.834).

Principal component analysis

Principal component analysis of all 15 quantitative traits (Table 5 and Fig. 2) accounted for 90.38% of the total variation through the first two components, with the first component (71.22%) representing an overall vigour and yield axis. Gujanatti Local and Mallapur Local were positioned towards the negative side of this axis, associated with higher fruit yield, fruits per vine, fruit weight and seed yield, whereas Kashi Khushi occupied the contrasting region, reflecting its lower yield and vegetative growth. The second component (19.16%) mainly differentiated flowering behaviour, with Arka Sumeet associated with late flowering and a higher sex ratio. Similar multivariate discrimination of ridge gourd genotypes on the basis of vegetative, flowering and yield traits has been reported by Yadav et al. (2017) [34] and Sushmitha et al. (2019) [21] and more recently in the wider Luffa genus by Zeng et al. (2026) [35] and in sponge gourd by Prabakaran et al. (2025) [36] . .0 Note: FY/V-Fruit yield per vine; FY/ha-Fruit yield/h per ctare; SY-Seed yield; F/V-Fruits per vine; FW-Fruit weight; TW-Test weight; VL45-Vine length @ 45 DAS. Genotypes ranked 1 (best) to 8 (poorest) for each trait; rank-sum = sum of ranks across all seven traits (lower = superior overall performance).

The correlation, path coefficient, principal component, divergence and composite rank-sum analyses together revealed broadly consistent patterns of trait association and genotype differentiation across the panel. Given the eight genotype, single environment design of the present study, these estimates should be regarded as indicative of the patterns observed within this panel rather than as precise population parameters and validation across larger, multi-location trials is recommended. Heritability and path coefficient estimates in particular may be influenced by genotype × environment interaction under a single-environment design and should therefore be interpreted with appropriate caution pending confirmation through multi-location trials with larger germplasm panels.

Conclusion

This study identified Gujanatti Local, Mallapur Local and Arka Prasanna as the most promising ridge gourd genotypes for the Northern transition zone of Karnataka, combining high fruit and seed yield with desirable vegetative vigour. Sex ratio, fruits per vine and seed yield emerged as the most important direct contributors to fruit yield and are recommended as primary selection criteria, with vine length and fruit weight as useful correlated traits. Despite its poor mean performance, Kashi Khushi proved to be the most genetically divergent genotype from the high yielding Cluster II, highlighting the value of divergence based characterization in capturing genetic variability that is not reflected in mean performance alone.

Acknowledgement

AcknowledgementThe authors thank the University of Agricultural Sciences, Dharwad and ICAR-Indian Institute of Horticultural Research, Bengaluru; ICAR-Indian Agricultural Research Institute, New Delhi; and ICAR-Indian Institute of Vegetable Research, Varanasi, for supplying seed material of the released cultivars used in this study.

References

1. Shendge PN, Belemkar S. Therapeutic potential of Luffa acutangula: a review on its traditional uses, phytochemistry, pharmacology and toxicological aspects. Frontiers in Pharmacology. 2018;9:1177.
2. Directorate of Horticulture. Horticultural crops statistics of Karnataka state at a glance 2022-23. Bengaluru: Government of Karnataka; 2023. https://horticulturedir.karnataka.gov.in
3. Bose TK, Som MG. Vegetable crops in India. Calcutta: Naya Prokash; 1986.
4. Burton GW. Quantitative inheritance in grasses. In: Proceedings of the 6th International Grassland Congress; Pennsylvania State College, USA; 1952. p. 277-283.
5. Johnson HW, Robinson HF, Comstock RE. Estimates of genetic and environmental variability in soybean. Agronomy Journal. 1955;47(7):314-318.
6. Dewey DR, Lu KH. A correlation and path coefficient analysis of components of crested wheatgrass seed production. Agronomy Journal. 1959;51:515-518.
7. Iezzoni AF, Pritts MP. Applications of principal component analysis to horticultural research. HortScience. 1991;26(4):334-338.
8. Mohammadi SA, Prasanna BM. Analysis of genetic diversity in crop plants: salient statistical tools and considerations. Crop Science. 2003;43(4):1235-1248.
9. Singh RK, Chaudhary BD. Biometrical methods in quantitative genetic analysis. 3rd ed. New Delhi: Kalyani Publishers; 1985.
10. Shah KN, Kale PN. Yield component association analysis in ridge gourd. Journal of Maharashtra Agricultural Universities. 2002;27(2):197-198.
11. Bagalkot, Karnataka: University of Horticultural Sciences; 2011.
12. Bhargava AK, Singh VB, Kumar P, Meena RK. Efficiency of selection based on genetic variability in ridge gourd [Luffa acutangula L. (Roxb.)]. Journal of Pharmacognosy and Phytochemistry. 2017;6(4):1651-1655.
13. Shodhan Jain HS, Vasudevan SN, Hanumantharaya BG, Vishwanath K, Marappa N. Comprehensive assessment of ridge gourd [Luffa acutangula (L.) Roxb.] genotypes: insights into growth, seed yield and quality traits. The Pharma Innovation Journal. 2023;12(12):860-863.
14. Amulya HT, Yashavantakumar KH, Hadimani HP, Chittapur RB, Vijaykumar V. Assessment of genetic variability, heritability and genetic advance in ridge gourd (Luffa acutangula (L.) Roxb.) genotypes. Plant Archives. 2025;25(1):2029-2033.
15. Panse VG, Sukhatme PV. Statistical methods for agricultural workers. 4th ed. New Delhi: Indian Council of Agricultural Research; 1985.
16. Lush JL. Inter-size correlation regression of offspring on dairy as a method of estimating heritability of characters. Proceedings of the American Society of Animal Production. 1940;33:293-301.
17. Rao CR. Advanced statistical methods in biometric research. New York: John Wiley and Sons; 1952.
18. R Core Team. R: a language and environment for statistical computing. Version 4.4. Vienna: R Foundation for Statistical Computing; 2024. https://cran.r-project.org/
19. Koppad SB, Chavan ML, Hallur RH, Rathod V, Shantappa T. Variability and character association studies in ridge gourd (Luffa acutangula Roxb.) with reference to yield attributes. Journal of Global Biosciences. 2015;4(5):2332-2342.
20. Rathore JS, Collis JP, Singh G, Rajawat KS, Jat BL. Studies on genetic variability in ridge gourd [Luffa acutangula L. (Roxb.)] genotypes in Allahabad agro-climate condition. International Journal of Current Microbiology and Applied Sciences. 2017;6(2):317-338.
21. Sushmitha M, Ramana CV, Dorajee Rao AVD, Naram Naidu L, Salomi Suneetha DR. Studies on genetic diversity for yield and quality traits in ridge gourd (Luffa acutangula (L.) Roxb.). International Journal of Chemical Studies. 2019;7(4):2081-2085.
22. Harika M, Gasti VD, Shantappa T, Mulge R, Shirol AM, Mastiholi AB, et al. Evaluation of bottle gourd genotypes [Lagenaria siceraria (Mol.) Standl.] for various horticultural characters. Karnataka Journal of Agricultural Sciences. 2012;25(2):241-244.
23. L.]. Journal of Horticultural Sciences. 2015;10(2):154-158.
24. Ananthan M, Krishnamoorthy V. Genetic variability, correlation and path analysis in ridge gourd (Luffa acutangula (Roxb.) L.). International Journal of Current Microbiology and Applied Sciences. 2017;6(6):3022-3026.
25. Jha M, Pandey CD, Singh PK, Archak S, Verma M. Characterization of ridge gourd (Luffa acutangula) germplasm accessions for qualitative and seed quality traits. The Indian Journal of Agricultural Sciences. 2025;95(9):1041-1047.
26. Kumar S, Yadav LP, Yadav OP, Tunç Y, Khadivi A. Evaluation and selection of superior snap melon (Cucumis melo var. momordica Duth. and Full.) genotypes using morphological, yield, and selected fruit quality traits. Scientific Reports. 2026;16(1):26669.
27. Narasannavar A, Devappa V, Pitchaimuthu M, Fhakrudin B, Lingaiah HB, Anjanappa M, et al. Selection parameters for downy mildew disease, fruit yield and related traits in ridge gourd [Luffa acutangula (Roxb.) L.]. Electronic Journal of Plant Breeding. 2017;8(2):601-608.
28. International Journal of Advanced Biochemistry Research. 2024;8(9):102-106.
29. Rai S, Sarkar RK, Roy SK, Datta S, Rai U, Rai S, et al. Estimates of genetic variability, heritability and genetic advance for yield and yield component traits in ridge gourd [Luffa acutangula (L.) Roxb.]. International Journal of Bioresource and Stress Management. 2025;16(3):1-13.
30. Deokar SP, Dhakare BB, Shinde GC, Anarase SA, Patil MR. Genetic studies in F5 generation of ridge gourd (Luffa acutangula L. Roxb.). International Journal of Advanced Biochemistry Research. 2024;8(2):118-123.
31. Manikandan M, Yassin GM, Kanthaswamy V, Kamaladevi S. Correlation and path coefficient analysis in ash gourd (Benincasa hispida (Thunb) Cogn.) for yield and yield attributing traits. Chemical Science Review and Letters. 2017;6(23):1399-1403.
32. Roxb.] genotypes. International Journal of Economic Plants. 2026;13(5):1-6.
33. Yadav H, Maurya SK, Pooja SK. Genotype screening and character association studies in indigenous genotypes of ridge gourd [Luffa acutangula (Roxb.) L.]. Journal of Pharmacognosy and Phytochemistry. 2017;6(5):223-231.
34. Zeng Z, Zhang L, Guo Y, Li Y, Chen Q, Song H, et al. Genetic diversity analysis of angled luffa germplasm resources based on phenotypic traits and resequencing data. Frontiers in Plant Science. 2026;17:1774008.
35. Prabakaran GP, Praneetha S, Haripriya S, Manikanda BN, Kumaresan D, Chandrakumar K. Assessing per se performance, variability, association and principal component analysis in sponge gourd (Luffa cylindrica [L.] Roem.) genotypes for different horticultural attributes. Plant Science Today. 2025;12(Special Issue 1).