Common bean, also known as dry bean, is the most important legume crop in the world and the main source of dietary protein in the developing countries [1] [2] . It is grown on subsistence farms by small holders in Africa with acreage exceeding 4 million hectares and serves more than 70 million livelihoods in sub-Saharan African continent [3] -[5] . Highlands (>1500 m) cover about two-thirds of Ethiopia and therefore are crucial to any effort to expand agriculture in the country. Common bean is cultivated both in highlands and lowlands, but its production is constrained by low temperature prevalence in the Ethiopian highlands with varying amounts of rainfall. Ethiopia receives on an average rainfall of 800 to 2200 mm in the highlands [6] . Rainfall is divided in two seasons—the big rains (kirmet) from mid-June to mid-September and small rains (belg) from February to May. Common bean cultivation is widely practiced during the main growing season (meher) coinciding with the big rains. However, there exists the potential of increased acreage if the crop can also be grown during secondary rainy period when other major crops fail. The suboptimal temperatures (0˚C - 10˚C) in the Ethiopian highlands constitute a critical problem to the production of common bean during the secondary rainy period in the region. Lowlands have very conducive temperature for common bean cultivation during secondary rainy period but also have varying rainfall patterns. Screening for genotypes with superior cold tolerance is, thus, needed in order to achieve the goal of two crops of common bean in a year. A few reports of genotypic variation for cold tolerance in common bean exist suggesting the probable success of such efforts [7] -[9] .

This experiment was carried out with the objective to evaluate common bean germplasm accessions (original source-CIAT) for adaptability to the lowland (Dire Dawa), highland (Haramaya) or common to both elevation ranges of Ethiopia which differ by 900 m altitude. These accessions were evaluated for agronomic performance for yield and other morphological traits in the two locations. Since many variables account for the growth and ultimate seed yield of a plant but not all of them contribute equally, and the amount of total variance represented contains a huge redundancy due to correlated traits measuring the same trait. The principal component analysis (PCA) is a multivariate analysis based on the factorization of variance matrix to produce a reduced set of variables [10] -[12] which may be used effectively for germplasm screening. This study was carried out to identify genotypes with cold tolerance potential using principal component analysis to reduce a huge redundancy due to correlated traits measuring the same trait. The results of the study were thus subjected to PCA for the identification of the most suited genotypes for the two elevations.

A large germplasm collection of common bean is available [14] . However, only few usually outperform at a given location due to inherent genetic potential to adapt to prevailing environmental conditions. Therefore, assessment of genetic variation present in a set of germplasm core collection is important for evaluating their agronomic performance in a given set of environment [15] . This further helps in making decisions related to the selection of sites and germplasms for in-situ conservation [16] .

A narrow genetic base exists for common bean cultivars within the market class in Ethiopia which can be improved by introgression of favorable alleles for resistance to abiotic stresses [17] [18] . The information on cold tolerant common bean genotypes in Ethiopia is inadequate due to lack of research effort in this area and the existing varieties are susceptible to temperatures below 15˚C [19] -[23] . Therefore, screening of common bean germplasm for genetic variation for cold tolerance is needed.

Summary of the descriptive statistics are presented (Table 2). Descriptive statistics showed higher mean values for most traits measured at Dire Dawa (chill-free location) compared to Haramaya (Chill-prone location). Of these traits, plant height and nodule number showed significant difference between the two locations which is useful information for dry bean researchers in this region due to increased emphasis on water-use efficiency and nitrogen-use efficiency in common bean breeding programs. The observed variation is attributed to genotypic response of 100 common bean genotypes to low temperature prevailing in Haramaya location during secondary rainy period. Correlation matrix revealed the significant correlations between the traits affecting yield (Table 3). The correlation coefficient values indicated a significant correlation between seed yield and yield related traits at Haramaya and Dire Dawa locations. This study revealed ample genetic variation among yield and other morphological traits due to effect of differential response of genotypes to both the chill prone and chill free conditions. Common bean yields were higher on average for Dire Dawa location compared to Haramaya location indicating that lower temperature imposed a limit on common bean productivity in Haramaya location. Variance estimates for genotype x location interactions are significantly different for nine traits (Table 4). The results revealed that yield values were significantly different between two locations arising from genotype X location effect. Genetic variation was significantly different for yield, number of pods/plant, number of seed/pod, seed weight/pod. This indicates the genotype performance for Haramaya and Dire Dawa is different for all the traits.

PCA analysis was done to assess the multiple traits to screen genotypes that exhibit frost and low temperature tolerance. The evaluation of germplasm through principal component analysis is a robust tool to reduce the large number of correlated variables and germplasms into smaller components (factors) of uncorrelated variables.