The sorghum [Sorghum bicolor (L.) Moench] germplasm collection at the ICRISAT gene bank exceeds 37,000 accessions. A core collection of 2247 accessions was developed in 2001 to enable researchers to have access to a smaller set of germplasm. However, this core collection was found to be too large. To overcome this, a sorghum mini core (10% accessions of the core or 1% of the entire collection) was developed from the existing core collection. The core collection was evaluated for 11 qualitative and 10 quantitative traits in an augmented design using three control cultivars in the 2004–2005 post‐rainy season. The hierarchical cluster analysis of data using phenotypic distances resulted in 21 clusters. From each cluster, about 10% or a minimum of one accession was selected to form a mini core that comprised 242 accessions. The data in the mini core and core collections were compared using statistical parameters such as homogeneity of distribution for geographical origin, biological races, qualitative traits, means, variances, phenotypic diversity indices, and phenotypic correlations. These tests revealed that the mini core collection represented the core collection, which can be evaluated extensively for agronomic traits including resistance to biotic and abiotic stresses to identify accessions with desirable characteristics for use in crop improvement research and genomic studies.

Finger millet [Eleusine coracana (L.) Gaertn.] is an important cereal food crop in Africa and South Asia. It is a hardy crop that can be grown in very diverse environments from almost at sea level to about 2400 m.a.s.l. Finger millet has an excellent food value as its seeds contain protein ranging from 7 to 14% and are particularly rich in methionine amino acid, iron, and calcium. Despite all these merits, this crop has been neglected from the main stream of crop improvement research. One of the means to boost its production and productivity is to enhance utilization of finger millet germplasm to breed superior varieties. Keeping this objective in view, a core subset of finger millet germplasm (622 accessions) based on origin and data on 14 quantitative traits was developed from the entire global collection of 5940 accessions held in the genebank at ICRISAT, Patancheru, India. The comparison of means, variances, frequency distribution, Shannon-Weaver diversity index (H¢) and phenotypic correlations indicated that the core subset represents the entire collection. These tests indicated that sampling was optimal and the diversity has been captured very well in the core subset. The correlation analysis indicated that panicle exsertion and longest finger length could be given lower priority in the future germplasm evaluation work of finger millet.

Foxtail millet (Setaria italica (L.) Beauv.) is one of the ten small millets and is cultivated in 23 countries. The foxtail millet is valued as a crop of short duration, which is good as food, feed and fodder. In general, grain yield levels of foxtail millet are low in comparison with other staple cereals. The greater use of diverse germplasm in breeding is suggested as a means to improve the productivity of this crop. The International Crops Research Institute for the Semi-Arid Tropics genebank is presently holding 1474 cultivated germplasm accessions from 23 countries. To utilize this diversity in research, a core collection (10% of the entire collection) was established using the taxonomic and qualitative traits data. The germplasm accessions were stratified into three taxonomic races (Indica, Maxima and Moharia). Principal coordinate analysis was performed on 12 qualitative traits for each of the biological races, separately that resulted in the formation of 29 clusters. From each cluster, 10% of the accessions were selected to constitute a core collection of 155 accessions. The composition and diversity of the core collection was validated by the x 2 -tests of the frequencies of origin, races, subraces and data on qualitative traits. The analysis of the quantitative traits for mean, range, variance, Shannon -Weaver diversity index and phenotypic associations indicated that the diversity from the entire collection was optimally represented in the core collection. The core subset will be evaluated in replicated trials to make a more precise assessment of diversity and further efforts to identify the sources of agronomic and grain nutritional traits for utilization in breeding programmes.

We analysed the patterns of variation for 14 qualitative and 12 quantitative traits in 11,402 pigeonpea germplasm accessions from 54 countries, which were grouped into 11 regions. Semi-spreading growth habit, green stem colour, indeterminate flowering pattern and yellow flower colour were predominant among qualitative traits. Primary seed colour had maximum variability and orange colour followed by cream were the two most frequent seed colours in the collection. Variances for all the traits were heterogeneous among regions. The germplasm accessions from Oceania were conspicuous by short growth duration, short height, fewer branches, pods with fewer seeds, smaller seed size and lower seed yields. The accessions from Africa were of longer duration, taller, with multi-seeded pods and larger seeds. The germplasm diversity indicated by Shannon -Weaver diversity index (H 0 ) pooled over all traits, was highest for Africa (0.464^0.039) and lowest for Oceania (0.337^0.037). The cluster analysis based on three principal component scores using 12 quantitative traits revealed formation of three clusters: cluster 1 includes accessions from Oceania; cluster 2 from India and adjacent countries; and cluster 3 from Indonesia, Thailand, the Philippines, Europe, Africa, America and the Caribbean countries. Pigeonpea-rich countries such as Myanmar, Uganda, and others like Bahamas, Burundi, Comoros, Haiti and Panama are not adequately represented in the collection, and need priority attention for germplasm exploration.

The chickpea and pigeonpea are protein-rich grain legumes used for human consumption in many countries. Grain yield of these crops is low to moderate in the semi-arid tropics with large variation due to high GxE interaction. In the Indian subcontinent chickpea is grown in the post-rainy winter season on receding soil moisture, and in other countries during the cool and dry post winter or spring seasons. The pigeonpea is sown during rainy season which flowers and matures in post-rainy season. The rainy months are hot and humid with diurnal temperature varying between 25 and 35°C (maximum) and 20 and 25°C (minimum) with an erratic rainfall. The available soil water during post-rainy season is about 200–250 mm which is bare minimum to meet the normal evapotranspiration. Thus occurrence of drought is frequent and at varying degrees. To enhance productivity of these crops cultivars tolerant to drought need to be developed. ICRISAT conserves a large number of accessions of chickpea (>20,000) and pigeonpea (>15,000). However only a small proportion (<1%) has been used in crop improvement programs mainly due to non-availability of reliable information on traits of economic importance. To overcome this, core and mini core collections (10% of core, 1% of entire collection) have been developed. Using the mini core approach, trait-specific donor lines were identified for agronomic, quality, and stress related traits in both crops. Composite collections were developed both in chickpea (3000 accessions) and pigeonpea (1000 accessions), genotyped using SSR markers and genotype based reference sets of 300 accessions selected for each crop. Screening methods for different drought-tolerant traits such as early maturity (drought escape), large and deep root system, high water-use efficiency, smaller leaflets, reduced canopy temperature, carbon isotope discrimination, high leaf chlorophyll content (drought avoidance), and breeding strategies for improving drought tolerance have been discussed.