The feeding of the cacao thrips, Selenothrips rubrocinctus (Giard), on cashew, Anacardium occidentale, one of its host plants in Trinidad, West Indies, is considered in relation to the annual period of maximum population increase on this host and to the choice of feeding sites on individual leaves. On trees observed for three years, populations regularly increased during the dry season, from a low level in December and January to a peak in April or May, and then rapidly declined during the wet season. Even when thrips were most abundant, some trees were free from attack, and this could not be attributed to protective morphological features, to specific repellent substances in the leaf, or to chance. S. rubrocinctus was found to feed on leaves that were subjected to water-stress and to breed only on debilitated trees: the evidence suggested that the adequacy of its supply of nutrients depends on the induction of suitable metabolic conditions within the leaf by water-stress.

Both nymphs and adults normally feed on the lower, stomata-bearing surface of the leaf, but in a very humid atmosphere only a weak preference is shown for this surface and if, under natural conditions, it is exposed to insolation by inversion of the leaf, the insects migrate to the other surface. Since the thrips were shown to be indifferent to bodily posture, the observation suggests that their behaviour is governed primarily by avoidance of exposure to undue heat or dryness and only secondarily by the attractiveness of the stomata-bearing surface.

Leaves of cashew tend not to become infested while still immature, and become most heavily infested, if at all, soon after they have hardened. Breeding does not occur on senescent leaves. The positions of feeding thrips are almost random on leaves under abnormal water-stress, but otherwise conform to certain patterns that mainly develop in fixed sequence. On reversal of an undetached leaf and consequent transfer of thrips from one surface to the other, there is no appreciable change in their distribution pattern or the apparent acceptability of the substrate. Changes of pattern were readily induced by injury to the plant during a period of water-stress and less easily, or not at all, when water-stress was low. Injury of areas of the leaf by heat was followed by their colonisation by thrips, and partial severance of branches by increased attack on their leaves.

Leaves detached from uninfested trees invariably became acceptable for feeding within four hours. During this period, leaf water-content declined and the ratios of soluble-carbohydrate content and α-amino acids to fresh-leaf weight fell slightly and rose considerably, respectively. In the field, the latter ratio was invariably higher for infested than for uninfested leaf tissue, even on portions of the same leaf. If the nutrient value of leaf tissue is determined by the rate at which α-amino acids are extractable through a stylet puncture, the observed change in acceptability for feeding following plucking may be accounted for by the increase in α-amino-acid concentration. Feeding that is restricted on any one tree to the margins of local leaf injuries during prolonged high water-stress and totally absent when stress is low can be correlated with an α-amino-acid content in the living marginal tissue that is high or low, respectively. The ability of thrips to establish themselves and breed on leaves of a particular tree in the dry season and their failure to do so on leaves of the same tree in the wet season conforms with the greater or less amino-acid concentration occurring in the leaf at these respective times.