Inbreeding was thought to be responsible for the loss in the second filial generation (F₂) of Amalafrida leithella Dyar, Cactoblastis mundelli Heinrich, Nanaia sp. Heinrich, Sigelgaita sp. Heinrich and Sigelgaita transilis Heinrich in the laboratory. This pre-empted the investigation of the effects of inbreeding on another cactophagous moth, Mimorista pulchellalis Dyar, an established biological control agent of jointed cactus in South Africa. Initially three populations were set up. A randomly-mating control (OUT 1) population, and a sibmating experimental (IN 1) population, consisted of laboratory-reared stock . A second experimental population (KR 1) comprised a small number of field-collected randomly-mating individuals which recreated the conditions under which the five abovementioned species were lost. The inbreeding depression of fifteen fitness components was assessed. The mean values of each component in each generation of treatments IN 1 and KR 1 were compared with those of OUT 1. In addition the mean values of IN 1 were regressed against the coefficients of inbreeding since inbreeding depression is linear with respect to the probability of two genes at any locus being homozygous through ancestry. The component egg viability was important because a reduction in OUT 1, IN 1 and KR 1 in the F₂ resulted from mated females producing no viable eggs. Duplicate treatments OUT 2, IN 2 and KR 2 were set up to confirm whether this was a general F₂ phenomenon. Assessment of the fitness components prevented a direct evaluation of the numbers of offspring produced. However a hypothetical estimate of population size and growth rate was made using the percentage survival calculated from life-table analysis. Although not statistically demonstrable in the component analysis, life table analysis indicated that egg viability suffered an inbreeding depression and affected population fitness. It was also evident that treatments OUT 1 and 2 were fitter than treatments IN 1 and 2 and KR 1 and 2 with respect to population fitness. Thus, inbreeding, resulting from sibmating and introduction of a small number of individuals to a laboratorYJ caused a decrease in numbers of offspring produced and population growth rate. This is important in relation to the demise of the cactophagous Lepidoptera and to future biological control programmes.

From the introduction: The work on Mimorista is divided into two sections: a laboratory orientated study which describes the biology and rearing techniques and a field orientated study where the establishment of the insect and its impact on jointed cactus populations are examined. In the final chapters the integration of this insect in the current control program are discussed and recommendations regarding its future in South Africa are given.

Pereskia aculeata Miller (Cactaceae) is an environmental weed that is damaging to natural ecosystems in South Africa. The plant is native to Central and South America and was first recorded in South Africa in a botanical garden in 1858. In this thesis, research into the biological control of P. aculeata was conducted with the intention of improving the control of the weed. A pre-release study of the relationship between P. aculeata density and native plant biodiversity indicated that P. aculeata has a negative impact on native biodiversity. The native plant biodiversity associated with different P. aculeata densities was used to determine threshold values and goals for the control of the weed. A threshold value of 50% P. aculeata density was calculated, indicating that P. aculeata density must be maintained below 50% in order to conserve native plant biodiversity. The ultimate goal of the control programme should be to maintain P. aculeata densities below 30%. At these densities there was no significant difference in native plant biodiversity from if the weed were absent from the ecosystem. The biological control agent, Phenrica guérini Bechyne (Chrysomelidae), has been released in South Africa but the potential of the agent to impact P. aculeata is not known and no post release evaluation has been conducted. Impact assessment studies indicate that P. guérini does not impact P. aculeata, even at high densities, but the results of greenhouse experiments should be interpreted with caution because of problems with extrapolation into the field. Although observations in the field suggest that P. guérini has reduced P. aculeata densities at one site, it is clear that new biological control agents are needed to reduce the weed to acceptable levels. Identifying the origin of the South African P. aculeata population was believed to be important to the biological control programme due to the disjunct native distribution and intraspecific variation of the species. Natural enemies associated with plant genotypes in different parts of the native distribution may have developed specialised relationships with certain intraspecific variants of the plant, resulting in differences in agent efficacy on certain host plant genotypes. A molecular study indicated that the closest relatives to the South African weed population found in the native distribution were in Rio de Janeiro Province, Brazil. A bioassay experiment in which fitness related traits of the biological control agent, P. guérini, were measured on various P. aculeata genotypes was conducted to determine the importance of host plant intraspecific variation. There was little variation in fitness traits between genotypes and no evidence of intraspecific host plant specialization. Although intraspecific variation had no effect on agent efficacy in the case of P. guérini, it is possible that other natural enemies may be more specialized. Genotype matching is expected to be more important when natural enemies likely to be specialised to individual genotypes are considered for biological control. Potential biological control agents were prioritized from data collected on surveys in the native distribution. The most promising of these, based on the presence of feeding, incidence, predicted host range, climatic matching, genotype matching and mode of damage, are two species of Curculionidae, the current biological control agent P. guérini and the stem boring moth, Maracayia chiorisalis Walker (Crambidae). The two curculionid species and M. chlorisalis should be considered priorities for host specificity studies. Releases of P. guérini and any new biological control agents should be made at sites where the pre-release study was conducted so that post-release evaluation data can be compared with the pre-release data and the impact of biological control can be evaluated. Retrospective analyses of biological control programmes provide important ways of improving aspects of biological control programmes, such as methods of agent selection. The evaluation of success in biological control programmes is essential for retrospective analyses because factors that have lead to successes or failures can be analysed. Retrospective analyses of biological control programmes, such as this thesis, may improve weed management, thereby contributing to the conservation of natural resources.