For both of the important cacao Mirids, Sahlbergella singularis Hagl. and Distantiella theobroma (Dist.), there is a well-marked annual population cycle. Numbers are at their lowest in March and, although showing fluctuations from month to month, remain low until July. After this period, there is an increase in numbers to a peak level which may be bimodal, an earlier peak corresponding with the maximum numbers on pods and a later peak on stems alone in January or February. This is followed by an abrupt drop in the dry season to the minimum values of March.

The period of increasing numbers is, at first, associated with the presence of more cacao pods but evidence is given to prove that Mirid increase is nor determined by pod increase but that both reflect the action of some common cause. The final peak population is associated with the symptoms of seasonal blast, but although Mirid numbers beneath the canopy are at their maximum at this time, there is no corresponding increase in the resident canopy population, the blast symptoms being produced by a transient population of adults from the sub-canopy levels.

Even at their highest the numbers of Mirids are small compared with most crop pests, the normal maximum being of the order of 1,000 per acre in mature stands of cacao.

Such aspects of mortality as could be studied from field observations are described. There is a steady loss during nymphal life, more dying during the later stages because these last longer. Parasitism, food supply and predation appear to be the main controlling factors, but, of these, only predation is examined quantitatively. Mantids, ants and Reduviids kill between 16 per cent. and 21 per cent. of the nymphs with which they are associated and the proportion of Mirids killed by such predators increases as the number of Mirids increases. Predation plays an important part in the natural controla of Mirid populations but, although optimum conditions for predation exist in Mirid pockets, economic control is not effected.

The use of ants as a Mirid control measure is examined and rejected. Infestation of healthy trees by ants does not protect nearby attacked trees and the association between ants and healthy trees is due to the colonisation of such trees as the only ones suitable in an area subject to Mirid damage. A single instance is quoted where Oecophylla longinoda (Latr.) appeared to exert a control but only at a level where the ants were so numerous as to constitute a hazard to humans in the plot.

A final discussion of the three papers of the series is included to focus attention upon those aspects of Mirid biology which need immediate investigation. These include an evaluation of the loss caused by Mirids and the need for experimental work to determine the effect of physical factors on Mirids. Suggestions are made for sampling which could be employed in any future field observations.