predicting kernel number in wheat

understanding the variation of kernel number per unit area (kno) is an important factor in respect yield and in identifying opportunities to increase potential yield. some crop models are for dependence on kernel number prediction to yield simulation. in order to quantify the relationship between kernel number per unit area and influencing parameters (environment and non-environment-based factors), compound analysis with randomized complete block design was conducted including four replications. this experiment was done at the gorgan university of agricultural sciences research farm in 2005. the treatments were five different sowing dates (14 december, 20 jan, 20 feb, 20 march, 16 apr) and four wheat variations (kohdasht, shirudi, tajan, zagros). considering results of correlation coefficients, kernel number per unit area appeared to have the best relation with mean temperature, but non-linear models fitting indicated that the highest r2 and the lowest cv belonged to stem dry matter at anthesis. fitted model showed that 95% of kernel numbers are formed when stem dry matter is around 747 g.