Causes of variation in the rate of increase of wheat harvest index

Several crop simulation models calculate grain yield by assuming that the rate of change of harvest index (δHI/δt) is constant (at rate k) during grain growth. Such behaviour has been identified in many crops, although the literature contains many examples of variations in k. The concept is useful if it approximates the truth in most circumstances, or if departures from both linearity and constancy are predictable from either the environment or the state of the crop. In this paper we examine the hypothesis that much of the variation in k is related to both crop biomass at the start of grain filling (BGF) and the crop growth rate during grain growth (CGF). Calculations using simple partitioning rules indicated that both factors are important. We showed that k increases rapidly as BGF decreases below about 9.0 Mg DM/ha, but decreases only slowly with increases of BGF above 9.0 Mg DM/ha. The analysis also showed that the increase in HI with time is quadratic rather than linear. We analysed data from 68 field grown wheat crops with variation in cultivar, location, irrigation, ambient CO2 concentration and sowing dates. These showed an almost three-fold variation in k (0.0058–0.0164 day−1). Across all data sets, there was a negative linear relationship (y = 0.02 − 0.0006x, R2 = 0.41, p < 0.001) between k and BGF. Overall, these results suggest that some of the variation in k can be accounted for by differences in BGF. We suggest that any model that uses harvest index change as a basis for calculating yield should account, at least, for variations in BGF.