Radiation interception and modelling as an alternative to destructive samples in crop growth measurements

Growth analysis presently uses destructive samples to detect temporal variations in biomass. The destructive nature of the measurements, their cost, and statistical considerations limit the application of growth studies in many domains of crop science. In contrast radiation interception data are cheap and easy to obtain without destruction of experimental material. Biomass may be modelled as the product of cumulative radiation intercepted by the crop [ΣI] and a radiation use efficiency coefficient [e]. Therefore, in theory, an alternative to destructive samples is provided by measurement of I at intervals during growth and e. The success of this approach depends on the validity of the value of e and its constancy through time. With measurement of I at intervals the mean radiation use efficiency [̄e] can be estimated from the seasonal ΣI and the final harvest data. The ̄e can then be used with the time series data for ΣI to estimate the biomass for that plot for any date. To test this approach model-derived biomass data were compared with data from destructive samples at seven dates for six groundnut germplasm lines grown in water limiting and fully irrigated conditions. The model-derived data was consistently less than destructively obtained data when the plants were small. This bias was an artifact of the interception measurement technique used not being accurate for small plants. Once plants were tall enough for fractional interception to be measured without substantial error, the nondestructive method effectively described the growth of the well-watered crops. For the drought treatments, it was less effective. However, by dealing with the phases of growth separately, good correlation between the two methods was achieved. An important assumption in the method is that the final harvest biomass is a realistic reflection of the preceding growth, since the model method forces the estimates of growth to that point. In one germplasm line this assumption was not valid and the model-based method did not match the sampled biomass data.