REGULATION OF AN INSECT POPULATION UNDER BIOLOGICAL-CONTROL

California red scale is suppressed to very low densities by the parasi toid Aphytis melinus. The system also appears stable. We report on an experimental test of the hypothesis that stability is caused by a refu ge for scale. In a grapefruit grove in southern California in 1984-198 5, the bark in the interior part of the tree provided a partial refuge from parasitism. Scale were approximate to 100 times denser there tha n in the exterior of trees. In a field experiment, we removed Argentin e ants from some blocks of trees to test whether (1) ants caused the r efuge by interfering with Aphytis and (2) the expected reduction in sc ale density in the refuge would lead to an unstable interaction in the exterior. We also tested for density-dependent parasitism, host mutil ation, and predation by analyzing data from samples and from scale pla ced in the field. The temporal variability of the scale was at the low end of the range recorded in field populations. The experiment provid ed some evidence in support of the refuge hypothesis. The population i n the refuge fluctuated much less than that in the exterior. Ant exclu sion led to increased parasitism and lower scale density in the interi or, and to increased fluctuations in abundance in the refuge and exter ior. However, these changes were relatively small and perhaps temporar y, suggesting that (1) ants are not the main cause of the refuge and t hat (2) we did not reduce the refuge density enough to determine wheth er the system would go unstable in the absence of the refuge populatio n. Parasitism, host mutilation, and predation rates on scale showed no temporal density dependence, either direct or delayed, though detecti on of such patterns is difficult. Possible alternative stabilizing mec hanisms include size-dependent interactions between red scale and Aphy tis.