Correlation and Path Coefficient Analyses in Sweet Melon (Cucumis melo Var. Aegyptiacus L.) Under Irrigated and Drought Conditions

MATERIALS AND METHODS

Thirteen sweet melon genotypes (10 local open-pollinated cultivars and 3 commercial cultivars) were used as genetic material for this study (Table 1). Local open-pollinated cultivars were collected from different places in Egypt and single plants from each cultivar were selfed for one generation during the growing season of 2008. The soil texture at the experimental site is clay-loam.

All genotypes were sown and evaluated under irrigated and drought conditions at Qaha vegetable research station, Qalubia governorate on 6 April of 2009 season. Each of two experiments was designed in a randomized complete blocks with three replicates. Each experimental unit area was consisted of four ridges each of 5 m length and 1.5 m in width and one plant per hill with 50 cm apart. The culture practices were done according to the general program of sweet melon cultivation. Drought conditions were started after first irrigation and created by reducing the frequency of irrigation watering by one half to that of irrigated crop, i.e., missing alternate irrigation.

Ten plants were randomly selected from each plot and observations were recorded on days to first female flower appearance. At 60 days after planting, a random sample of six plants was taken from the two inner ridges of each experimental unit to determine the growth parameters (stem length (cm) and number of primary branches per plant). At the harvesting time, a random sample of 12 plants was taken from each experimental unit and data were recorded for number of fruits per plant, fruit weight (g), total yield per plant (g), fruit length (cm), fruit width (cm), flesh fruit thickness (cm) and Total Soluble Solids (TSS). Total soluble solids were determined using a refractometer.

Components of variances and covariances were estimated for individual environment (irrigation and drought stress conditions) data according to method of Steel and Torrie (1980). Analysis of variance was computed among thirteen genotypes according to Snedcor and Cochran (1982).

Table 1:	The names and sources of sweet melon genotypes used in the study

The genotypic variance was estimated by analysis of variance as described by Dospekhov (1984) as follows:

where, σ²g is genotypic variance, σ²e is environmental variance, M₂ is mean square due to genotypes, M₁ is mean square due to error, r is number of replications.

The estimates of genotypic correlation coefficient was worked out by using the following equations as outlined by Singh and Choudhary (1979):

where, Cov g₁g₂ is the genotypic correlation between any pairs of traits. σ²g₁ and σ²g₂ are the genotypic variance of the first and second trait, respectively. The significant of the r_g was tested with “t” test as described by Cochran and Cox (1957).

Genotypic correlations were partitioned into path coefficient using the technique outlined by Dewey and Lu (1959) for assessing the direct and indirect effects of each trait on yield per plant.

RESULTS AND DISCUSSION

The mean squares due to local genotypes were highly significant (p<0.01) for all studied traits under both normal and drought conditions (Table 2). These results were expected due to the great variability among the genotypes. This can be utilized through appropriate crossing and suitable selection indices for better genotypes in the segregating populations.

Genotypic correlation
Under irrigated conditions: Total yield per plant showed the highest positive correlation with fruit weight (0.973), followed by flesh fruit thickness (0.831) and fruit length (0.756). Significant and positive correlation of total yield per plant with number of primary branches per plant (0.331) was also observed (Table 3). These results are in agreement with those obtained by Choudhary et al. (2003), Taha et al. (2003), Singh and Lal (2005), Zalapa et al. (2006), Reddy et al. (2007) and Kumar et al. (2013). However, these results are not consistent with Feyzian et al. (2009) who reported a non-significant correlation between total fruit weight per plant and number of primary branches.

Moreover, total yield per plant showed the highest negative correlation with TSS (-0.655), number of fruits per plant (-0.615) and fruit width (-0.576) (Table 3). In contrast, Somkuwar et al. (1997) reported that improvement in muskmelon is possible by selecting genotypes for number of fruits per plant and TSS. Also, Reddy et al. (2007) reported that fruit diameter exhibited a strong positive correlation with yield per plant, while number of fruits per plant and TSS exhibited a non-significant correlation with yield per plant. Moreover, Abd-El-Salam and Marie (2002) and Choudhary et al. (2003) and Iathet and Piluek (2006) found that fruits number per plant had highly positive correlation with yield per plant. The differences may be due to breeding material used and different environmental conditions.

Table 2:	Mean squares of 13 genotypes grown under irrigated and drought conditions
**: Significant at 0.01% probability level

Table 3:	Genotypic correlation coefficients for different pairs of traits in sweet melon under irrigated (above diagonal) and drought stress (below diagonal)
SL: Stem length, No.B: No. of primary branches per plant, DF: Days to first female flower appearance, FW: Fruit weight, No.F: No. of fruits per plant, TY: Total yield per plant, FL: Fruit length, FD: Fruit width, FFT: Flesh fruit thickness, TSS: Total soluble solids, *,**Significant at 0.05 and 0.01 probability level, respectively

Fruit weight, fruit length and flesh fruit thickness showed the highest positive correlation with each other and showed the highest negative correlation with number of fruits per plant, fruit width and TSS (Table 3). This confirmed the results obtained by Lal and Singh (1997) in muskmelon who found that number of fruits per plant showed a highly significant negative correlation with fruit length, fruit weight, fruit diameter and flesh thickness. Yadav and Ram (2002) found that muskmelon fruit weight had a positive and significant correlation with fruit polar diameter, fruit equatorial diameter and flesh thickness that were positively correlated with each other and the total soluble solid content was not significantly associated with any of the traits. Singh and Ram (2003) in muskmelon reported that fruit weight was positively correlated with flesh thickness and was negatively correlated with total soluble solids content. Taha et al. (2003), Zalapa et al. (2006) and Feyzian et al. (2009) reported a negative association between fruits number and average weight per fruit. In contrast, Abd-El-Salam and Marie (2002) found high significant and positive correlations between fruit weight and total soluble solids, fruit length and fruit flesh thickness. Also, Eduardo et al. (2007) found high significant and positive correlations between fruit weight and fruit diameter.

Significant and negative correlation of fruit length with fruit width suggested that any increase in fruit length will induce reduction in fruit diameter. While a decrease in this character will inversely increase fruit diameter. Negative correlation between fruit length and fruit width has also been reported by Iathet and Piluek (2006).

The stem length showed a highly significant positive correlation with days to first female flower appearance (0.345) and a highly significant negative correlation with TSS (-0.425) and number of primary branches per plant (-0.345) (Table 3). Taha et al. (2003) and Reddy et al. (2007) reported similar type of results. Whereas, Reddy et al. (2007) found that stem length had a non-significant positive correlation with TSS.

Days to first female flower appearance showed a highly significant positive correlation with stem length (0.345) and had highly significant negative correlation with flesh fruit thickness (-0.331) while other traits had non significant correlation with days to first female flower appearance (Table 3). Similar results were reported by Reddy et al. (2007).

Under drought stress conditions: Total yield per plant had the highest positive and significant correlation with fruit weight (0.964), followed by flesh fruit thickness (0.822), fruit length (0.707) and stem length (0.511) (Table 3). Fruit weight showed the highest positive correlation with total yield per plant. Increase in fruit length, flesh fruit thickness and stem length resulted in increase in fruit weight, which ultimately increased the yield per plant (Table 3). These results are in agreement with those obtained by Choudhary et al. (2003), Taha et al. (2003), Singh and Lal (2005), Reddy et al. (2007) and Zalapa et al. (2006).

Other characters viz fruit width, TSS and number of fruits per plant showed the highest negative correlation with total yield per plant (Table 3), these results contradicted with the findings of Somkuwar et al. (1997), Abd-El-Salam and Marie (2002), Iathet and Piluek (2006) and Reddy et al. (2007), while number of branches per plant and days to first female flower appearance showed no correlation with total yield per plant (Table 3). These results corroborate the findings of Reddy et al. (2007) and Feyzian et al. (2009).

The stem length showed a highly significant positive correlation with fruit weight (0.494), fruit length (0.383) and flesh fruit thickness (0.530) and a highly significant negative correlation with TSS( -0.429) (Table 3). Similar results have also been reported by Taha et al. (2003) and Reddy et al. (2007).

Fruit weight, fruit length and flesh fruit thickness showed the highest positive correlation with each other and had the highest negative correlation with number of fruits per plant, fruit width and TSS (Table 3). It is quite possible that with the increase in the number of fruits, the fruit weight starts decreasing. This is in good agreement with the findings of Lal and Singh (1997), Yadav and Ram (2002), Singh and Ram (2003), Taha et al. (2003), Zalapa et al. (2006) and Feyzian et al. (2009).

The correlations of all characters with days to first female flower appearance were non significant (Table 3), similar results were reported by Reddy et al. (2007). Similarly, number of primary branches per plant revealed non significant correlation with all characters under discussion, except fruit width which showed a highly significant negative correlation (Table 3). Similar results have been determined by Zalapa et al. (2006).

The negative associations of characters like fruit weight and TSS will become problem in combining these important traits in a single genotype for high yield per plant. Some suitable recombination might be obtained through bi-parental mating, mutation breeding or diallel selective mating by breaking undesirable linkage.

Path coefficient analysis
Under irrigated conditions: Path coefficient analysis provides an effective way of finding out direct and indirect sources of correlations, using genotypic correlations of different plant attributes.

Table 4:	Estimates of direct (bold- diagonal values) and indirect effects of studied traits on total yield per plant at genotypic levels under irrigated conditions
SL: Stem length, No.B: No. of primary branches per plant, DF: Days to first female flower appearance, FW: Fruit weight, No.F: No. of fruits per plant, FL: Fruit length, FD: Fruit width, FFT: Flesh fruit thickness, TSS: Total soluble solids, rg: Genotypic correlation coefficients

Table 5:	Estimates of direct (bold-diagonal values) and indirect effects of studied traits on total yield per plant at genotypic levels under drought conditions
SL: Stem length, No.B: No of primary branches per plant, DF: Days to first female flower appearance, FW: Fruit weight, No.F: No. of fruits per plant, FL: Fruit length, FD: Fruit width, FFT: Flesh fruit thickness, TSS: Total soluble solids, rg: Genotypic correlation coefficients

Total yield per plant was selected as resultant variable, whereas, the remaining nine traits were considered as casual variables. The results of path coefficient analysis (Table 4) revealed that fruit weight exerted the highest direct positive effect (+1.263) on total yield per plant followed by number of fruits per plant (+0.198) and stem length (+0.070), while TSS had low positive effect (+0.056). It means a slight increase in any one of the above traits may directly contribute towards total yield per plant. Similar findings were reported by Pandita et al. (1990), Choudhary et al. (2003), Reddy et al. (2007), Tomar et al. (2008) and Kumar et al. (2013). Whereas, Reddy et al. (2007) reported low positive direct effect of fruit weight on total yield per plant.

The negative direct effects over total yield per plant were exhibited by days to first female flower (-0.168), flesh fruit thickness (-0.143), TSS (-0.056), fruit width (-0.066), fruit length (-0.049) and number of primary branches per plant (-0.025) (Table 4). The results were confirmed by Reddy et al. (2007), Tomar et al. (2008). However, these results did not agree with the report by Singh and Lal (2005) who reported that flesh thickness exhibited high, positive and direct effects on yield, Reddy et al. (2007) who found that positive direct effect of days to first female flower on total yield per plant and Tomar et al. (2008) who found that high and positive direct effect of flesh fruit thickness on total yield per plant.

Maximum positive indirect effects on total yield per plant were exhibited by fruit length (1.073) and flesh fruit thickness (1.070) through fruit weight (Table 4). Similar findings were also reported by Tomar et al. (2008). Similarly, the indirect effects of stem length and number of primary branches per plant through fruit weight were positive and fairly high. While maximum negative indirect effects were observed by number of fruits per plant (-0.963), TSS (-0.849) and fruit width (-0.745) through fruit weight (Table 4).

The indirect effects of all traits under discussion on total yield per plant through number of primary branches per plant were found at lower values between-0.007 and 0.008 (Table 4).

Under drought stress conditions: Path coefficient analysis permits a through understanding of contribution of various characters by partitioning the correlation coefficient into components of direct and indirect effects. The direct and indirect effects of investigated characters on total yield per plant under drought stress conditions are presented on (Table 5).

The results indicated that fruit weight which is positively and significantly correlated with total yield per plant (0.964) had the greatest positive direct effect (1.477) on total yield per plant. This was followed by number of fruits per plant, fruit length and TSS with 0.634, 0.281 and 0.045, respectively (Table 5). Similar findings were reported by Choudhary et al. (2003), Reddy et al. (2007) and Tomar et al. (2008). However, the indirect effects of number of fruits per plant and TSS via fruit weight were negative and high in magnitude (-1.000 and -0.978, respectively) (Table 5). Significant negative correlations between number of fruits per plant and total yield per plant (-0.473) and between TSS and total yield per plant (-0.477) were mainly due to their effects through fruit weight, which were high and negative (Tables 3 and 5). This showed that selection for the characters number of fruits per plant and TSS would not be realized in increased total yield per plant.

Fruit length indicates the positive direct effect and maximum positive indirect effect was noted through fruit weight (Table 5).

Stem length had negligible negative direct effect on total yield per plant (-0.014) but due its positive indirect effect via fruit weight (0.731) and fruit length (0.108) it had significant positive association with total yield per plant (0.513). The indirect effects of the other eight traits on stem length remained at lower values between -0.008 and 0.008 (Table 5).

The direct effects of other traits were negative direction towards reducing the total yield per plant. Especially flesh fruit thickness had a prominent effect on total yield per plant in negative direction with -0.371 (Table 5). The results were confirmed by Reddy et al. (2007) and Tomar et al. (2008). On the other hand, comparing total yield per plant and flesh fruit thickness the correlation coefficient was 0.834, it seemed that indirect effect appeared to be the cause of correlation (Table 5). Hence, these traits may be used simultaneously with the other characters. Flesh fruit thickness had a high positive indirect effect on total yield per plant via fruit weight (Table 5). Therefore, indirect selection for flesh fruit thickness through fruit weight would be effective for total yield per plant improvement.

The correlation of fruit width with total yield per plant was negative. The direct effect of fruit width on total yield per plant was also negative (-0.141), this result was confirmed by Reddy et al. (2007). The indirect effect via most traits was positive but negligible. The fruit width has its major influence via negative effect of fruit weight (-0.929) (Table 5). Therefore, selection through this trait will affect the total yield per plant negatively under genetic material to be studied.

Number of primary branches per plant exhibited negative direct effect on total yield per plant (-0.227) but this was compensated by other traits except TSS (Table 5).

CONCLUSION

From the above results, it may be concluded that the genetic variability in the local genotypes of sweet melon can be utilized through appropriate crossing and suitable selection indices for better genotypes in the segregating populations. While improving the total yield per plant of sweet melon varieties can be obtained by selecting plants for more fruit weight under irrigated conditions and selection for fruit weight and fruit length under drought conditions.