Wheat is the second most important cereal crop after rice in the context to its antiquity and its use as source of food and energy in India as well as in the world. The multifold increase of wheat production in India has mainly been due to introduction and development of semi dwarf varieties. These new varieties have given a cushioning capacity to the sustainable wheat production under vastly varying environments and have proved their worth against various biotic and abiotic stresses. The success of these modem varieties has been sensational, revolutionary and is seen as a result of close collaboration among the Indian wheat breeders and the international agencies particularly CIMMYT and ICARDA. A wealth of wheat genetic resources that has been accumulated in many parts of the world, is available in gene banks, research institutes, laboratories and long term in-situ storage banks. An enormous amount of passport data on these resources have been gathered through collective efforts across the world. In recent past, CIMMYT initiated a data management system on wheat germplasm that can be used by researchers with a unique identifier. In India, Directorate of Wheat Research, Karnal has been engaged in multi-location evaluation of elite germplasm in the form of various nurseries in order to provide its excess to wheat researchers and thus enable them to utilize these genetic resources as per their need and priorities. We made an attempt to quantify the genetic variability, heritability, yield architecture and clustering pattern of durum wheat germplasm based on multi-location evaluation, so that the information thus generated may be used accordingly. This paper deals with the results obtained from above study along with the possible suggestive breeding methodology to update the utilization and information generation on some important aspects. This appears to be a optimum assumption for improving yields as well as the wheat production in India.

Materials and methods

A set of 62 elite genotypes of durum wheat (Triticum durum) was selected from various national and international sources thereby including material from different origins and agro-ecosystems. Efforts were made to include genotypes from various mega zones considering the characters like duration, height, yield and grain size. The material was planted in a two row plot of 2.5 m length spaced at 23 cm between rows and 10 cm between plants of the row. The experiment was laid out at 12 locations across the country following augmented design wherein the three checks were repeated after every 20th genotype to make statistical analysis precise and useful. At each location similar agronomic practices were followed to raise a good crop. The adjusted means were utilized for estimating genetic variability and other parameters following standard statistical parameters (Johnson et al. 1955).

The observations on five important agronomic traits, namely days to heading, maturity duration,plant height, grain yield per plot and thousand grain weight (TGW) were recorded. Analysis of variance (ANOVA) for all five traits was performed location-wise as well as over pooled data as suggested by Panse and Sukhatme (1967). Analyses then were combined over locations where environment and genotypes were considered and correlations among traits were computed over environments. The pooled data were subjected to cluster analysis as well as to make dendrogram based on the inter-varietal distances (Singh 1994) and accordingly the 62 genotypes were grouped into five dusters considering all the five traits under study. The distances are Euclidean but the pattern is similar to that computed following Mahalanobis distances, however, the dendrogram of which are not shown and discussed here (Mahalanobis 1936; Lee and Kaltsikes 1973). The plotting of linkage distances across steps following Euclidean distances has been made as per the latest available statistical software (Statistica).