Advanced correlation analysis of the performance of pea varieties

Аннотация

Aim. The question of the identity or difference of both the matrices of characteristics of different accessions and the matrices of their correlation coefficients and path coefficients is important. We carried out a comparative analysis of correlation matrices and matrices of path coefficients between productivity and its components for pea varieties bred by the Institute of Plant Production na V.Ya. Yuryev National Academy of Sciences (Ukraine) in 2002-2018.

Materials and methods. The field trials were carried out in the years 2014-2016 in the scientific crop rotation of the Institute of Plant Production na V.Ya. Yuryev (PPI) in the Kharkiv region. 14 varieties of peas own breeding of PPI and included in the State Register of Plant Varieties which are suitable for distribution in the Ukraine were used as research material.

The path correlation coefficients were calculated in accordance with the description of Dewey and Lu. The Spearman's rank coefficient sensu Rao G. Nageswara was used for the assessment of the series identity of both genotypic and path coefficients. The table of critical values of Spearman's coefficient was used for the assessment of significance.

Result and discussion. Highly reliable genotypic correlation coefficients with strong association were recorded for the trait "number of seeds per plant": 10 out of 14 varieties. To determine the consistency of the correlation matrix for the researched varieties, Spearman's correlation coefficient was calculated. Reliable values of Spearman's coefficients were found for 66 pairs of comparisons out of a total of 182. All the Spearman's correlation coefficients for the matrix of varieties in Hajduk were low and not significant.

There was no unidirectional influence of any trait on productivity in the analysis of direct and indirect effect indicators for the varieties in the sample. Only the trait "number of seeds per plant" had the highest positive values of direct effect. In other cases, differences in the productivity of the varieties studied are confirmed by both positive and negative high values of direct effects.

Spearman's correlation coefficients, calculated to determine agreement between path coefficient correlation matrices for varieties in the sample, were significant in only 16 pairwise comparisons (for direct and indirect effects) of the total calculated. Spearman's correlation coefficients for direct effects ranged from r_s = 0.92 to r_s = -0.78 and for indirect effects from r_s = 0.82 to r_s = -0.70 in pairwise comparisons.

Thus, differences in the strength and direction of the relationships between the characteristic matrices of the samples can be assessed using Spearman's correlation coefficient.

The differences in the characteristics determining the direct effects between our results and those of the cited publications can be explained both by the different genetic nature of the research objects and by the influence of meteorological conditions during the growth period and plant cultivation technologies. Note that most researchers tend to correlate morphological traits in aggregated samples. This leads to a generalised description of the relationship without the need for individualisation.

Conclusions. Correlation coefficients, path analysis and Spearman's rank correlation coefficients allow determination not only of the general trend of interaction between traits in the population, but also of the significance of differences between genotypes.