GENOTYPE X SITE INTERACTIONS IN SEED PRODUCTION, HARD SEED BREAKDOWN AND REGENERATION OF ANNUAL MEDICS (MEDICAGO SPP) IN WEST ASIA

Seed production of annual medics was measured at four sites in Syria a nd Jordan between 1986/87 and 1988/89. Plots sown to medic in the firs t year were sown to wheat in the second year, and medic regeneration w as measured in the third year. At three of the sites, adjacent plots w ere established in the second and third years and herbage production, and regeneration where appropriate, were measured in the third year. I n this way, genotype x site interactions were used to assess medic ada ptation to wheat/pasture rotations, continuous pasture and newly sown pasture. Seed production was correlated with rainfall, the mean yield of 14 genotypes ranging from zero at 160 mm to 1 tonne/ha at 510 mm. M edicago polymorpha produced the highest yields, though not at the lowe st rainfalls. Yields of M. rotata were also above average. There was, however, wide variation between species in seed weight, pod weight and seeds/pod. Regeneration was better after wheat than after pasture. He rbage production was slightly better after pasture than after wheat, a nd both were greatly superior to that of newly sown pasture. After 2 y ears' germination, most genotypes retained more than half (mean 60 %) and up to 99 % of the seed produced in the first year. Genotype x envi ronment interactions were tested using analysis of variance (ANOVA), r egression (Finlay & Wilkinson 1963) and principal component analysis ( PCA). Because adaptation in pasture plants depends on many variables, PCA proved to be the most useful technique and the simplest to apply. ANOVA and regression were limited by their inability to process more t han one variable at a time. Regression analysis is further limited by requiring a greater number of sites. Based on seed production, seed su rvival, seedling regeneration and herbage production, certain locally collected genotypes of M. rigidula, M, noeana, M, polymorpha and possi bly M. rotata, were adapted to conditions in Syria and Jordan. A simil ar methodology could be used to test pasture legumes throughout west A sia and north Africa.