Abstract: Yield is the outcome of many component characters, thus assessing the extent of contribution of each factor is important in designing an effective plant breeding programmes. Phenotypic correlations between bolls per plant with yield was r = 0.998, boll weight with yield was r = -0.768 and lint% with yield was r = 0.932. These results suggested that improvement in yield could be achieved by selecting cotton plant for higher bolls and more lint%. On the contrary, selection for boll weight will decrease cotton yield. These phenotypic correlations were further partitioned in to their direct and indirect effects on seed cotton yield by path coefficient analysis. This analysis revealed that bolls per plant had significant direct and indirect effects on yield, whereas boll weight and lint% had negligible effects. Multiple correlation coefficient revealed that about 91.8% of the total variation in yield is dependent on variables, bolls per plant, boll weight and lint%.
Introduction
The information on the extent of association between the yield and other factors is important to bring the simultaneous improvement in correlated traits. Although, knowledge of phenotypic correlation of agronomic characters with yield in cotton (Shaver and Dilday, 1982) is indispensable in the characterization of component influences on manifesting the characters, these associations yet do not provide explicit information on the relative importance of direct and indirect effects of each componential character on seed cotton yield. As the number of variables increase, it becomes imperative to measure the contribution of each variable towards the observed correlation. Therefore, partitioning the observed correlation coefficients in to components of direct and indirect influences provide perceptions in the characterizations of more complex traits, like yield (Baloch et al., 1992). In that situation, path coefficient analysis provides precise information on the direct and indirect effects in order to perceive the most influencing characters to be utilized as selection criteria in cotton breeding programmes.
Materials and Methods
The variety CRIS-121 (Gossypium hirsutum L.) evolved by cytogenetics section of Central Cotton Research Institute at Sakrand was grown in randomized complete block design with four replications during, 1999. The plot size was 12.5’ x 45’ and row to row and plant to plant distances were kept at 2.5’ and 9.0’’ respectively. Therefore, there were five rows per plot each 45 feet long. Ten plants at random from only three central rows, thus 40 plants in total were tagged as index plants for taking the observations. The data on four quantitative characters such as bolls per plant, average boll weight weighed in grams, lint% calculated as the ratio of lint against cotton seed from seed cotton sample and seed cotton yield per plant weighed in grams were recorded.
Phenotypic correlation coefficients (r), multiple correlation (R2) and regressions were determined by the methods outlined by Gomez and Gomez (1984). Phenotypic correlation coefficients were further partitioned in to the components of direct and indirect effects by path coefficient analysis by Dewey and Lu (1959) used by Singh and Choudhry (1979).
Results and Discussion
Phenotypic correlation (r), multiple correlation coefficient (R2) and path coefficient analysis were worked-out for yield and its important components. Phenotypic correlation coefficients presented in Table 1 reveal significant and positive correlations of r = 0.998 between bolls per plant and yield per plant, and r = 0.894 between lint% and yield and significant but negative correlation of r = -0.814 between boll weight and yield per plant. Based on the strong associations of other characters with yield, it is expected that improvement in number of bolls and lint% will bring simultaneous or say indirect improvement in cotton yield. Other researchers (Khan et al., 1980; Baloch et al., 1992) concluded from their studies that bolls per plant best be actuated for developing high yielding cotton varieties. Whereas negative association between boll weight and yield indicated that one should not expect positive correlated response in yield while selecting for higher boll weight. Baloch et al. (1992) and Khan et al. (1980) also reported similar results in their studies.
Yield is influenced by many of its components, thus cumulative effect of all the factors contributing to yield is important to determine. Multiple correlations were therefore exercised to determine the total effect of independent variables such as bolls per plant, boll weight and lint% on dependent variable (yield per plant). The mean square from multiple regressions was significant (Table 1) which suggested that independent variables under present study are important in exploring the yield. Multiple correlation coefficient, R2 = 0.957 and coefficient of determination, r2 = 91.8 further revealed that about 91.8% of total variability in yield is accounted due to three independent variables.
Path coefficient analysis was also carried-out so as to determine direct and indirect effects of each independent variable on dependent variable (Table 2). The direct effect of bolls per plant on yield was very high, 0.969 whereas indirect effects via boll weight and lint% were -0.104 and 0.123 respectively, however total effect (direct+indirect) on yield was 0.998 which was mainly due to direct effect of bolls per plant. Baloch et al. (1992) reported that out of 0.978 total phenotypic correlation, 63.1% was attributable to only direct effect of bolls per plant on seed cotton yield. Rajper et al. (1987) also noted that about 118.3% of total phenotypic correlation in yield was due to direct effect of bolls per plant.
| Table 1: | Phenotypic correlation coefficients (r) and multiple regression analysis for yield and its components in Gossypium hirsutum L. |
| **Significant at 1% probability levels | |
| Table 2: | Direct and indirect pathway effects of bolls per plants, boll weight and lint% on seed cotton yield in Gossypium hirsutum L. |
The direct effect of boll weight on yield was very low (0.128), nonetheless indirect effects via boll number and lint% were -0.789 and -0.106 respectively. The negative phenotypic correlation coefficient, r = -0.768 between boll weight and yield suggest that as we increase the boll size, it gives corresponding decrease in yield. The large indirect effect of -0.789 via boll weight further connoted that decrease in yield is actually due to fewer number of bolls formed. In fact, it clearly indicated that as the boll weight increases above certain limit, the yield decreases due to proportional decrease in total number of bolls set by the cotton plant. Negligible direct effect of boll weight on yield in present study is in conformity with that of Baloch et al. (1992). The direct effect of lint% on yield was only 0.148 from 0.932 total phenotypic correlation whereas indirect effects via bolls per plant was 0.866 which is much higher as compared to boll weight, that in fact was negative, -0.093. Thus, 92.9% of total phenotypic correlation was again due to in direct effect via bolls per plant. Rajper et al. (1987) also reported similar findings.
In brief, it was observed that number of bolls per plant take toll of major effect on seed cotton yield. The large direct and indirect effects of bolls per plant on yield were also supported with their high correlation coefficient values, r = 0.998. It is therefore concluded from present studies that significant improvement in cotton yield could be made due to varieties that set higher bolls rather than big bolls.