Analysis of seed growth by linear increase in harvest index

A convenient and stable method for expressing crop seed growth is important both in experimental and simulation studies. The conventional approach has been to measure the seed growth rate (SGR) of individual seeds and multiply this by the number of seeds per unit ground area. An alternative to measurements of SGR has emerged from the observation that during seed growth, harvest index (HI) increases linearly over time. The objective of this study was to examine and compare SGR and dHI/dt as the appropriate description of crop seed growth on the basis of 22 data sets representing four crop species for which there were intensive measures of seed growth and HI. Seed growth was analyzed by linear, quadratic, cubic, and logistic models, in regression analysis. These analyses showed that SGR and dHI/dt were not constant for the entire seed growth period. However, constant values for these two measures of seed growth were statistically equivalent to a cubic function of seed growth for much of the seed growth period. A linear increase in SGR and dHI/dt was found to be statistically appropriate when seed weight and harvest index were increasing from 10 to 90% of their maximum value. It was found in comparisons of the coefficients of variability that the stability of dHI/ dt was greater than for SGR for individual cultivars grown under a range of conditions. Overall, this analysis indicated that a constant dHI/dt is an attractive approach for expressing crop seed growth.