Statistical models for stability analysis in watermelon R.Venugopalan and M. Pitchaimuthu1 Section of Economics and Statistics Indian Institute of Horticultural Research Hessearghatta Lake PO, Bangalore-560 089, India E-mail: venur@iihr.ernet.in ABSTRACT Fourteen promising F 1 hybrids of watermelon namely IIHR-188 X IIHR-118, IIHR 114 X IIHR 118 , IIHR 119 X IIHR- 20-1, Arka Manik X IIHR 46, IIHR 43 X IIHR 46, Arka Manik X IIHR-188, Arka Jyothi, NS-295, Kushboo, Madhubala, Apoorva, CWH-7 and Riya were evaluated in experimental plots of Division of Vegetable Crops, Indian Institute of Horticultural Research, Bangalore during 2002-04. Information about biometrical characters such as fruit length (cm), fruit girth (cm), days to first male flower opening & female flower opening, rind thickness(cm) and TSS (%) along with yield (t ha-1), were used to develop stability models to identify stable hybrid(s) for a wide range for cultivation. Stability models thus developed indicated that two hybrids, viz., Arka Jyothi (with yield potential of 75.91 t ha-1) across the years and NS-295 (64.25 t ha-1) were stable for a wide range for cultivation. Statistical measures of stability, viz., regression coefficient, deviation from regression co-efficient and ecovalence measures, were worked out and utilized for grouping of hybrids into different categories based on their cumulative performance over the years. Key words: Ecovalence measure, GE interaction, stability models, watermelon J. Hortl. Sci. Vol. 4 (2): 153-157, 2009 1Division of Vegetable Crops, IIHR, Banglaore-560 089. INTRODUCTION In any crop improvement research, plant breeders before recommending release of a particular variety/hybrid for commercial exploitation in farmer’s field, ensure the stability of varieties, by testing it across different environments/periods. In such studies, the breeders’ main interest will be to estimate the average response of the varieties and also to test the consistency of the yield response from region to region/environment to environment. The presence or absence of the so-called Genotype x Environment (GE) interaction, coupled with high yield, will largely dictate the good performance of the genotypes. However, in practice, genotypes responsible for showing higher yield are less stable and vice versa. The presence of such a GE interaction also alters the relative ranking of different varieties in addition to reducing the correlation between phenotype and genotype, thus making it difficult for a breeder to judge the true genetic potential of variety. Hence, the main aim was to strike a balance between these two extremes by evolving appropriate statistical methods to reduce the component of GE interaction for identifying stable genotypes that interact less with the environment. MATERIAL AND METHODS Fourteen promising F 1 hybrids of watermelon namely, IIHR-188 X IIHR-118, IIHR 114 X IIHR 118 , IIHR 119 X IIHR-20-1, Arka Manik X IIHR 46, IIHR 43 X IIHR 46, Arka Manik X IIHR-188, Arka Jyothi, NS-295, Kushboo, Madhubala, Apoorva, CWH-7 and Riya evaluated in the experimental plots of Division of Vegetable Crops, Indian Institute of Horticultural Research, Bangalore during 2002- 04, were utilized to develop stability models with a view to identify best variety(s) for commercial exploitation based on Yield (t ha-1), fruit length (cm), fruit girth (cm), days to first male flower opening & female flower opening, Rind thickness(cm) and TSS (%). Two different approaches based on Eberhart and Russell (ER) model (Eberhart and Russell, (1966); Bhargava et al., 2008) and Freeman and Perkins (FP) model (Dehghani et al., 2008, Freeman and Perkins, 1973) were utilized for carrying out stability analysis research. The details of these methods are well elucidated in Prabhakaran and Jain (1992) and more recently from application of point of view by Venugopalan and Gowda (2005). 154 Measures of stability Different measures of stability viz., mean performance (X i ), regression coefficient (b i ), deviation from regression coefficient (S2d i ) and Wricke’s ecovalence (W i ) measures (Wricke ,1962) were computed using standard formulae, as given below : (i) Regression coefficient (b i ): - - - - b i = Σ (Y ij -Y i .)(Y .j -Y..) / S (Y .j -Y..)2 (ii) Deviation from regression (S2d i ): - - - S2d i = [ [ Σ (Y ij -Y i .) 2 – bi2 Σ (Y .j -Y..)2] / (s-2)]- S2 e (iii) Wricke’s ecovalence (Wi): - - - W i = Σ (Y ij -Y i. -Y .j -Y..).2 Based on the these measures, hybrids were classified into any one of the following three groups. Group I: Ideal genotype (suitable for wide range of environment) b i =1 and S2d i =0 Group II: Average stability genotype (Suitable for poor environment) b i <1 and S2d i =0 Group III: Above average stability (suitable for favorable environment) b i >1 and S2d i =0. In general, a hybrid showing high yield potential under favorable environment and having great phenotypic stability is considered to be stable. Moreover, the lower the value of W i smaller will be the fluctuations from the predictable response in different environments. Accordingly, as an index of ranking in their order of stability/adaptability characteristics, the genotype with least ecovoalence is considered to be the most stable. RESULTS AND DISCUSSION Results of analysis of variance indicated for a differential behavior of all the 14 hybrids across three years (2002-04). Results of stability analysis presented in Table 1 confirmed the presence of (Genotype X Environment) G X E interaction as the mean sums of squares for all the characters across the genotypes were significantly differing from each other (p<0.05). This shows that the hybrids had divergent linear response to environmental changes. Four measures of stability values, viz., X i , b i , S2d i and W i were also worked out and are presented in Table 2. Based on these measures, genotypes were grouped into three groups (specific to their adaptability to a given environment) and the results are presented separately for ER and FP methods in Table 4. Further, as an in depth study of the results achieved under ER and FP methods pertaining to target group of the breeders, viz., ideal hybrids group, based on their W i values, 14 hybrids were ranked and are presented in Table 4. Perusal of the results presented in Table 2 to Table 4 brings out the following salient findings: Yield (t ha-1) : Under the Freeman-Perkins model three F 1 hybrids (Arka Jyothi (c), Riya, and NS 295) were identified as ideal, suitable for wide range of cultivation. Looking into the values of mean performance (Xi) of these ideal lines (Table 2), Arka Jyothi performed better (75.91 t ha-1), across the years, followed by NS-295 (64.25 t ha-1), than all the other lines. Accordingly, ecovalence values (W i ) worked out (Table 4) for the ideal lines showed that Arka Jyothi followed by NS 295 were stable for wide range of cultivation for yield t/ha, as they possesses least ecovalence values as compared to other lines. Further, IIHR-178 x Arka Manik (with yield potential of 51.93 t ha-1) is classified as an above Table 1. Sability analysis for different characters in Watermelon (Mean sum of squares) Source / Character Yield (t/ha) Fruit length Fruit girth Days to first male Days to first Rind (cm) (cm) flower opening opening thickness female flower (cm) Eberhart-Russell (ER) Method Genotype 190.15 29.66 29.14 2.337 2.67 0.04 V x Env (Linear) 85.29 16.61 0.311 1.073 1.82 0.08 Pooled Deviations 45.46 2.54 1.70 0.914 0.71 0.06 Average Error 0.52 1.16 0.23 0.239 0.34 0.003 Freeman-Perkins (FP) Method Genotypes 191.63 28.98 29.81 2.442 2.44 0.045 Environments 329.09 38.65 0.14 2.851 4.17 0.007 Combined reg. 653.52 73.61 0.03 5.707 8.34 0.009 Residual 4.65 3.69 0.24 0.001 0.001 0.004 Hetero of reg. 85.81 17.29 0.50 1.142 2.33 0.070 Residual 55.59 4.08 1.10 0.826 0.49 0.083 Average Error 0.77 2.20 0.27 0.345 0.45 0.005 J. Hortl. Sci. Vol. 4 (2): 153-157, 2009 Venugopalan and Pitchaimuthu 155 average genotype, which will respond well to a poor environment. Similarly for the other biometrical characters, results presented in Table 3 and 4, revealed a marked difference among the number of hybrids grouped separately under two methods. Results indicated clearly about the change in cluster membership while adopting Freeman-Perkins model. In addition to this analysis, based on additional two year yield data (2005-06), optimum number of years required for inferring the stability of the above hybrids was made by testing the W i values of subsequent years with the preceding value. It was observed four years (yield data) was sufficient (in addition to stable performance of Arka Jyothi and NS-295) to reach the stability of the evaluated genotypes as the measure of ecovalence till fourth year was significantly different from the earlier period and were on par from 5th year onwards. To summarize, stability models (with R2 = 81.4%-99.4%.) Table 2. Stability parameters of six quantitative traits for 14 watermelon lines under Freeman-Perkins model Name of the hybrid Yield (t/ha) Fruit length (cm) Fruit girth (cm) X i b i S2d i W i X i b i S2d i W i X i b i S2d i W i IIHR-178 x Arka Manik 51.93 0.63 3.12 7.93 22.58 1.46 -2.19 2.5 21.00 -0.98 1.8 2.06 IIHR-114 x IIHR-118 63.41 1.56 15.77 27.59 21.75 -0.62 -1.71 16.81 18.83 -0.45 0.71 0.82 IIHR-119 x IIHR-20-1 51.91 2.08 67.65 115.55 27 2.08 -1.16 12.16 21.88 0.05 -0.23 0.0 IIHR-118 x IIHR-20-1 62.16 2.58 129.72 237.81 20.92 -0.5 -2.08 12.91 18.83 -0.71 2.23 2.31 Arka Manik x IIHR-46 61.2 2.49 23.06 136.9 21.25 -0.3 1.68 7.69 19.98 1.75 -0.26 0.9 IIHR-43 x IIHR-46 52.43 -1.47 134.09 427.74 22.58 -0.18 -2.14 7.79 17.9 -0.19 -0.24 0.1 Arka Manik X IIHR-188 52.58 3.0 36.44 221.7 24.25 2.6 -2.18 20.44 21.08 -0.35 -0.26 0.06 Arka Jyothi (C) 75.91 0.07 2.0 44.83 26.58 0.28 -2.18 2.62 19.46 -1.1 -0.23 0.53 NS-295 (C) 64.25 0.88 7.44 14.22 28.63 3.43 -0.77 47.41 21.41 0.7 0.11 0.36 Kushboo 58.08 -0.35 13.74 104.54 22.5 -0.54 -1.07 14.04 14.16 -0.53 -0.06 0.22 Madhubala 56.25 0.81 97.64 111.69 27.72 3.87 -2.12 61.62 18.75 0.0 -0.27 0.01 Apoorva 51.01 1.54 124.77 127.25 25.97 1.5 -1.86 3.77 28.41 -1.15 -0.13 0.52 CWH-7 45.76 1.04 51.94 61.56 19.67 -0.55 45.9 55.08 17.66 0.62 0.15 0.69 Riya 48.08 -1.00 -0.76 188.85 18.92 -0.5 -2.08 12.91 18.41 3.56 7.39 12.09 Name of the hybrid Days to first male flower opening Days to first female flower opening Rind thickness(cm) X i b i S2d i W i X i b i S2d i W i X i b i S2d i W i IIHR-188 x Arka Manik 32.42 1.35 0.95 1.57 36.6 0.1 -0.39 0.57 1.26 1.81 0.14 0.14 IIHR-114 x IIHR-118 34.25 2.95 -0.11 3.45 37.42 -0.08 -0.41 0.71 1.3 -4.26 0.07 0.16 IIHR-119 x IIHR-20-1 33.58 -0.23 -0.34 0.69 37.42 2.06 -0.42 0.15 1.46 5.26 0.01 0.08 IIHR-118 x IIHR-20-1 34.16 -1.3 3.17 6.36 36.92 3.7 0.02 2.01 1.48 -0.21 -0.0 0.0 Arka Manik x IIHR-46 32.16 1.36 0.18 0.74 39.72 7.05 0.35 10.25 1.33 -3.88 0.07 0.14 IIHR-43 x IIHR-46 33.66 0.37 -0.27 0.17 38.37 3.54 -0.45 1.34 1.21 -4.17 0.08 0.17 Arka Manik X IIHR-188 32.75 0.14 1.26 1.8 37.67 0.99 -0.45 0.05 1.52 3.95 0.05 0.11 Arka Jyothi (C) 33.25 0.23 0.21 4.67 37.42 2.18 0.8 1.41 1.24 -0.24 0.0 0.0 NS-295 (C) 34.75 0.13 0.01 0.67 37.13 -2.01 -0.45 3.49 1.53 -0.98 0.1 0.12 Kushboo 33.41 1.07 0.41 0.86 37.22 1.78 1.13 1.6 1.43 5.97 0.04 0.13 Madhubala 34.58 -0.23 0.4 1.48 37.33 -2.18 0.8 5.11 1.51 -7.59 0.34 0.53 Apoorva 34.41 -0.6 0.19 1.8 39.08 -3.26 0.44 7.39 1.6 1.45 0.0 0.01 CWH-7 34.8 1.85 -0.13 1.02 36.92 1.62 -0.44 0.02 1.31 9.31 0.09 0.34 Riya 34.83 0.75 -0.06 0.26 38.58 4.32 -0.45 2.5 1.42 -0.2 0.0 0.0 developed for yield and yield attributing biometrical characters of 14 watermelon F 1 hybrids indicated that Arka Jyothi followed by NS-295 were stable for wide range of cultivation for yield t ha-1, as they possesses least ecovalence values as compared to other lines. Results further indicated that IIHR-178 x Arka Manik is suitable for poor environment. Thus, Arka Jyothi performed better (75.91 t ha-1), across the years, followed by NS-295 (64.25 t ha-1), than all the other hybrids. These two hybrids are widely used in crop breeding research and also cultivated for higher productivity across years and seasons. Hence, the message arising out from this present study is that breeders may exploit the use of Freeman-Perkins approach for performing stability research while analyzing multi-location/year/season trails, with a view to cluster the breeding materials/ genotypes based on their stability/adaptability to a specific situation and also to select promising lines for further hybridization programme and for commercial exploitation. Statistical models for stability analysis in watermelon J. Hortl. Sci. Vol. 4 (2): 153-157, 2009 156 T ab le 3 . G ro u p in g of h yb ri d s b as ed o n r es u lt s of s ta b il it y an al ys is ( u n d er E R a n d F P m od el ) fo r 14 h yb ri d s of w at er m el on S l. N o . C h ar ac te r Id ea l g en o ty p e1 b i = 1 a n d S 2 d i = 0 A b o v e A v er ag e g en o ty p e2 B el o w A v er ag e g en o ty p e3 b i < 1 a n d S 2 d i = 0 b i > 1 a n d S 2 d i = 0 1 . Y ie ld ( t/ h a) E R m o d el R iy a -- -- F P m o d el A rk a Jy o th i (c ) , R iy a, N S 2 9 5 II H R -1 7 8 x A rk a M an ik -- 2 . F ru it l en g th ( cm ) II H R -1 7 8 x A rk a M an ik , A rk a M an ik x I IH R - II H R -1 1 4 x I IH R -1 1 8 ,I IH R -1 1 8 x I IH R -2 0 - II H R -1 1 9 x I IH R -2 0 -1 , A rk a M an ik E R m o d el 4 6 ,A p o o rv a 1 , II H R -4 3 x I IH R -4 6 , A rk a Jy o th i X I IH R - (C ), K u sh b o o , R iy a II H R -1 1 4 x I IH R -1 1 8 , 1 8 8 , M ad h u b al a F P m o d el II H R -1 7 8 x A rk a M an ik , A p o o rv a II H R -1 1 8 x I IH R -2 0 -1 , A rk a M an ik x I IH R - 4 6 , II H R -4 3 x I IH R -4 6 , A rk a Jy o th i (C ), K u sh b o o , R iy a II H R -1 1 9 x I IH R -2 0 -1 3 . F ru it g ir th ( cm ) II H R -1 1 9 x I IH R -2 0 -1 , II H R -4 3 x I IH R -4 6 , A rk a -- II H R -1 1 9 x I IH R -2 0 - E R m o d el M an ik X I IH R -1 8 8 , N S -2 9 5 ( C ), M ad h u b al a, C W H -7 1 , K u sh b o o , C W H -7 II H R -1 1 4 x I IH R -1 1 8 , II H R -1 1 9 x I IH R -2 0 -1 , A rk a -- -- M an ik x I IH R -4 6 , II H R -4 3 x I IH R -4 6 , A rk a M an ik X F P m o d el II H R -1 8 8 , A rk a Jy o th i ( C ), N S -2 9 5 ( C ), K u sh b o o , - M ad h u b al a, A p o o rv a, C W H -7 4 . D ay s to f ir st m al e II H R -1 1 9 x I IH R -2 0 -1 , A rk a M an ik X I IH R -1 8 8 , N S - II H R -4 3 x I IH R -4 6 , A p o o rv a II H R -1 1 4 x I IH R -1 1 8 , C W H -7 fl o w er o p en in g 2 9 5 ( C ), K u sh b o o , M ad h u b al a, R iy a E R m o d el F P m o d el A rk a M an ik x II H R -4 6 , II H R -4 3 x I IH R -4 6 , II H R -1 1 9 x I IH R -2 0 -1 , N S -2 9 5 C W H -7 K u sh b o o , R iy a (C ), M ad h u b al a, A p o o rv a 5 D ay s to f ir st f em al e II H R -1 7 8 x A rk a M an ik , II H R -1 1 8 x I IH R -2 0 -1 , II H R - II H R -1 1 4 x I IH R -1 1 8 , A rk a M an ik X I IH R - fl o w er o p en in g 4 3 x I IH R -4 6 , A rk a Jy o th i ( C ), K u sh b o o , C W H -7 1 8 8 , N S -2 9 5 ( C ) E R m o d el II H R -1 7 8 x A rk a M an ik , II H R -1 1 4 x I IH R - II H R -1 1 9 x I IH R -2 0 -1 , A rk a M an ik x F P m o d el A rk a M an ik X I IH R -1 8 8 1 1 8 A rk a Jy o th i ( C ), N S -2 9 5 II H R -4 6 , R iy a (C ), K u sh b o o , M ad h u b al a, C W H -7 6 . R in d t h ic k n es s II H R -1 1 8 x I IH R -2 0 -1 , A rk a Jy o th i ( C ), A p o o rv a, R iy a -- II H R -1 1 9 x I IH R -2 0 - E R m o d el 1 , K u sh b o o , C W H -7 F P m o d el II H R -1 1 8 x I IH R -2 0 -1 , A rk a Jy o th i ( C ), A p o o rv a, R iy a -- - 7 . T S S ( % ) II H R -1 7 8 x A rk a M an ik , II H R -1 1 8 x I IH R -2 0 -1 , R iy a -- -- E R m o d el F P m o d el II H R -1 7 8 x A rk a M an ik , II H R -1 1 8 x I IH R -2 0 -1 , R iy a -- -- 1 S u it ab le f o r a w id e ra n g e o f en v ir o n m en ts 2 S u it ab le f o r p o o r en v ir o n m en t 3 S u it ab le f o r fa v o ra b le e n v ir o n m en t J. Hortl. Sci. Vol. 4 (2): 153-157, 2009 Venugopalan and Pitchaimuthu 157 Table 4. Ranking among ideal watermelon hybrids under ER and FP models based on measure of ecovalence (W i ) Based on Eberhart-Russell (ER) Procedure Based on Freeman-Perkins (FP) Procedure Character Ideal Genotype Ranked Wi values Ideal Genotype Ranked Wi values 1. Yield (t/ha) NS-295 56.686 Arka Jyothi (c ) 13.24 Kushboo 105.441 NS 295 22.58 Riya 189.434 Riya 122.54 2 Fruit length (cm) IIHR-188 X IIHR-118 2.228 IIHR-188 X IIHR-118 2.503 Apoorva 2.765 Apporva 2.621 IIHR 43 X IIHR 46 4.432 3. Fruit girth (cm) Madhubala 0.069 IIHR 119 X IIHR-20-1 0.002 IIHR 43 X IIHR 46 0.144 Madhubala 0.019 IIHR 119 X IIHR-20-1 0.310 Arka Manik X IIHR-188 0.065 4. Days to first male NS-295 0.527 IIHR 43 X IIHR 46 0.167 flower opening Kushboo 0.603 Arka Manik X IIHR 46 0.735 5. Days to first female Kushboo 0.384 Arka Manik X IIHR-188 0.054 flower opening IIHR-188 X IIHR-118 0.538 IIHR-118 x IIHR-20-1 0.859 6. Rind thickness(cm) Arka Manik X IIHR 46 0.003 Arka Manik X IIHR 46 0.002 Riya 0.003 NS-295 0.003 NS-295 0.008 Riya 0.004 Apoorva 0.011 Apoorva 0.009 7. T.S.S. (%) Riya 0.122 Arka Manik X IIHR 46 0.061 Arka Manik X IIHR 46 0.204 Riya 0.371 IIHR-188 X IIHR-118 0.335 IIHR-188 X IIHR-118 0.781 REFERENCES Bhargava, A., Shukla, S. and Ohri, D. 2008. Genotype x environment interaction studies in Chenopodium album L.: an underutilized crop with promising potential. Comm. in Biom and Crop Sci., 3:3–15 Dehghani, H., Sabaghpour, S.H and Sabaghnia, N. 2008. Genotype × environment interaction for grain yield of some lentil genotypes and relationship among univariate stability statistics. Spanish J. Agril. Res., 6:385-394 Eberhart S A. and Russell W.A. 1966. Stability parameters for comparing varieties. Crop Sci., 6:36-40 Freeman G.H. and Perkins J.M. 1971. Environmental and genotype-environmental components of variability. VIII. Relations between genotypes grown in different environments and measures of these environments. Heredity, 27:15-23 Prabhakaran V.T. and Jain J P, 1992. Statistical techniques for studying Genotype-Environment interactions. SAP Pvt. Ltd. New Delhi Wricke, G. 1962. Uber eine Methods Sur Erafassung der okologischen streubreite in Foldversuchan. Z. Pflanzonzuchig, 47:92-96 Venugopalan, R and Veere Gowda, R 2005. Stability Analysis In Onion: A Statistical Look. J. Ind. Soc. Coastal Agric. Res., 23:123-29 (MS Received 2 December 2008, Revised 10 July 2009) J. Hortl. Sci. Vol. 4 (2): 153-157, 2009 Statistical models for stability analysis in watermelon