Sampling schemes for estimating spikelet incidence of Fusarium head blight

Fusarium head blight (FHB) is an important disease of wheat crops. The incidence of diseased spikelets in the population is most naturally estimated by sampling ears of wheat and counting the number of infected spikelets, X, and the total number of spikelets, Y, on each ear. However, there is usually much more variation from ear to ear in the number of infected spikelets than in the total number of spikelets and in practice it is common to obtain a larger sample of X values than Y values. In this paper, we investigate the relative efficiencies of these alternative sampling procedures. Based on historical data, we show that for the levels of disease incidence typically encountered, the latter sampling scheme is indeed more efficient. The results are relevant to estimating disease incidence in other crops.