Indian J. Genet., 74(4) Suppl., 558-563 (2014) DOI: 10.5958/0975-6906.2014.00889.X Abstract Nature and complexity of genotype × environment interaction (GEI) was studied among eight rabi grain sorghum cultivars across 11 locations during rabi 2011-12 and 2012-13 using GGE biplot analysis. Location (L) contributed for 89.9% of variation for grain yield, while genotypes (G) and G × L interactions accounted for 1% and 9% of variation only. The first two principal components (PCs) of GGE biplot accounted for 50% of variation in data for grain yield, which not ideally explained overall variation in the data. However, the biplot clearly demonstrated that across environments, SPH 1721 was the highest yielding stable genotype followed by CSH 15R. High crossover GEI was recorded among the testing locations and close correlation among these locations was not detected. ‘Which-won-where’ analysis detected three mega-environments (ME) among the testing locations, with ME1 represented by 5 locations, ME2 with 4 locations and ME3 with 2 locations. The study indicated the possibility to reduce the number of testing locations. Key words : GGE biplot, GxE interaction, sorghum, mega-environment Introduction Sorghum [Sorghum bicolor (L.) Moench] is the fifth most important cereal grown extensively in the arid and semi- arid tropics under low fertile soil conditions. India ranks first in area and second to the USA in production across the world [1]. Sorghum of two adaptive types, viz., kharif and rabi sorghum, are cultivated in India, of which rabi sorghum is predominantly consumed as food grain [2, 3]. Productivity of rabi sorghum is much lower than the kharif sorghum [4]. Concentrated breeding efforts are being made under All India Coordinated Sorghum Improvement Project (AICSIP) to release better yielding rabi cultivars. Multi-location trials (MLT) play a crucial role in the process of identification and release of improved and stable cultivars. However, often during the analysis of MLT data genotype evaluation is limited on genotype main effects (G), while genotype × environment interactions (GEI) are ignored as noise, which is otherwise equally important [5]. Various statistical models like analysis of variance (ANOVA), principal component analysis (PCA), and linear regression (LR) have been suggested over time to understand the complex GEI [6, 7]. Each procedure has its own advantages and disadvantages [6, 8-10]. Genotype (G) main effect plus GE interaction (GGE) biplot analysis [7] is a robust method to visualize and interpret MLT data graphically. Utility of GGE biplot in understanding GEI has been demonstrated in many crops including sorghum [10, 11]. To have an insight into the nature and complexity of GEI in the rabi grain sorghum MLT data, performances of eight rabi sorghum cultivars across 11 locations for two years (rabi seasons of 2011-12 and 2012-13) were studied using GGE biplot analysis. Efforts were also made to identify mega- environments within the testing locations. *Corresponding author’s e-mail: rakshit@sorghum.res.in Published by Indian Society of Genetics & Plant Breeding, F2, First Floor, NASC Complex, PB#11312, IARI, New Delhi 110 012 Online management by indianjournals.com GGE biplot analysis of genotype × environment interaction in rabi grain sorghum [Sorghum bicolor (L.) Moench] Sujay Rakshit*, K. N. Ganapathy, S. S. Gomashe, M. Swapna, A. More 1 , S. R. Gadakh 2 , R. B. Ghorade 3 , S. T. Kajjidoni 4 , B. G. Solanki 5 , B. D. Biradar 6 and Prabhakar Directorate of Sorghum Research, Rajendranagar, Hyderabad 500 030, Telangana; 1 Marathwada Agricultural University, Parbhani 431 402, Maharashtra; 2 Mahatma Phule Krishi Vidyapeeth, Rahuri 413722, Distt. Ahmednagar 413722, Maharashtra; 3 Dr. Panjabrao Deshmukh Krishi Vidyapeeth, Akola 444104, Maharashtra; 4 Main Sorghum Research Station, Univ. of Agricultural Sciences, Dharwad 580005, Karnataka; 5 Navsari Agricultural University, Athwa Farm, Surat 395007, Gujarat; 6 Agricultural Research Station, University of agricultural Sciences, Bijapur 586 101, Karnataka (Received : September 2014; Revised : october 2014; Accepted: November 2014)