SAMPLE-SIZE DETERMINATION FOR THE STALK-EYED FLY DIOPSIS-LONGICORNIS MACQUART (DIPTERA, DIOPSIDAE) DAMAGE ON RICE UNDER NATURAL FIELD CONDITIONS

Proportions of rice tillers damaged by the stalk-eyed fly (SEF), Diops is longicornis Macquart, were statistically analysed for nine sampling periods (rice growth stages) to determine the optimal sample size for use in damage assessment in the future. The major variation (>90%) in the proportion of damaged tillers was attributable to differences bet ween hills, with <10% attributable to differences between stations. Th e effects of stations, although relatively small, were found to be sta tistically significant in seven of nine sampling periods. Graphs of co efficient of variation (yBAR) against the number of stations (t) and n umber of sample hills (n) showed that using t = 2 to t = 4 and n = 20 to n = 25 would give a precision of <2.0%. Damage levels remained almo st constant during the nine sampling periods, but variability was obse rved to be greater at the initial and final sampling periods, and smal ler in the intermediate sampling periods. The optimal combinations of the number of stations (t(opt)) and sample size (n(opt)) were obtained for a given ratio of the cost due to station (C(S)) to the cost due t o hill (C(H)).