Zulqarnain Haider et al./ Elixir Appl. Botany 79 (2015) 30729-30737 30729 Introduction Drought, generally accepted by scientists, is: “a shortfall of water availability sufficient to cause loss in yield”, or “a period of no rainfall or irrigation that affects crop growth” (Price, 2002; Fukai and Cooper, 1995). Drought stress is a major constraint to rice (Oryza sativa L.) production and yield stability in rainfed ecosystems (Dey and Upadhyaya, 1996). The global reduction in rice production due to drought averages 18 M t annually. In Asia alone, it is estimated that a total of 23 M ha of rice fields (10 M ha in upland and 13 M ha in lowland) are drought-prone (Pandey et al., 2000). Drought is also a particularly important production constraint in Pakistan, with more than 10 M ha of drought-prone fields, where yield losses due to droughts are reported to cost an average of 250 million US$ per year. So developing drought resistant cultivars especially with good performance under late season drought stress is one of the major objectives in rice breeding programs (Boonjuing and Fukai, 1996; Pantuwan et al., 2002). However, the progress in breeding for drought resistance is rather slow in rice due to complexity of the traits. So, there is a need to develop rice varieties that are better yielders under both drought and favorable environments, and are least affected by drought. Many traits are known to contribute in improving yield under drought, but their actual contribution towards yield reduction in not understood. Also comparing the responses of rice genotypes, in terms of change, increase or decrease in characters due to stress is better criteria than comparing their values only in drought stress. Therefore, increase and decrease in characters were measured and compared with one another as well as with reduction in yield as resultant variable in order to find out actual parameters that majorly contribute in yield reductions under drought stress. Simple correlation coefficient analysis is generally used to measure the relationship between two variables (kown and torrie). But it cannot give enough information about the relationships between different variables. Therefore, other statistical methods such as cluster analysis, factor analysis and principal component analysis were also included to interpret these results. These multivariate statistical analyses can provide more insights on the deep structure of data and traits’ Relationship between Yield and Yield Components in Rice (Oryzae sativa L.) under Drought using Multivariate Statistics Zulqarnain Haider 1,2,* , Abdul Razaq 2 , Asrar Mehboob 2 , Muneeb Munawar 1 , Aftab Iqbal 3 , Attiq ur Rehman 1 , Samta Zia 1 and Khalid Mehmood 2 1 University of Agriculture, Faisalabad, Pakistan. 2 Maize and Millets Research Institute, Yusafwala, Sahiwal, Pakistan. 3 Potato Research Institute, Sahiwal, Pakistan. ABSTRACT The aim of study was to screen out rice (Oryzae sativa L.) genotypes with better drought tolerance, to explore observable responses of rice plants towards drought stress and multivariate statistics were used to compare the relationships among the responses as well as their effects on yield and yield reduction under drought stress. Genotypes responded differently under same level of stress. Results revealed that as a response to drought, all the genotypes showed 42 % average increase in seedling root length; 39 % average decrease in seedling shoot length; 21 % average reduction in 1000 grain weight; delay in heading time of 13 days in average; average yield per plant reduction of 47 %; reduction in grains per panicle of 52% in average, when subjected to simulated drought stress. It was concluded that in addition to correlation and path analysis of yield components, cluster analysis, factor analysis and principal component analysis are also very important methods for screening drought tolerant cultivars of rice in large populations and give better understandings of associations and helps better to choose important variables and genotypes. Factor analysis and Principal component analysis revealed the importance of drought response index, as a best measurement for yield under stress condition. Seedling root length to shoot length ratio, followed by seedling root length ratio contrubute majorly in yield per plant under drought stress. Reduction in plant height and delay in heading time were main causes of yield reduction under stress. It was further concluded that selecting the genotypes showing higher drought response index, higher increase in s ling root length, seedling root to shoot length ratio and less reduction in plant height and delay in heading time, are important criteria in screening for drought tolerance. Also these characters showed high heritability and genetic advance, signifying their values in selection breeding, and higher phenotypic correlation values indicating their polygenic behavior and high influence of environment. © 2015 Elixir All rights reserved ARTICLE INFO Article history: Received: 6 May 2013; Received in revised form: 15 February 2015; Accepted: 25 February 2015; Keywords Oryzae sativa L., Drought stress, Yield reduction, Yield stability, Drought related responses, Correlation analysis, Path analysis, Factor analysis, Principal component analysis, Cluster analysis. Elixir Appl. Botany 79 (2015) 30729-30737 Applied Botany Available online at www.elixirpublishers.com (Elixir International Journal) Tele: E-mail addresses: z.haider.breeder@gmail.com © 2015 Elixir All rights reserved