Arthropod species that have the potential to damage crops are food resources for communities of predators and parasitoids. From an agronomic perspective these species are pests and biocontrol agents respectively, and the relationships between them can be important determinants of crop yield and quality. The impact of biocontrol agents on pest populations may depend on the availability of other food resources in the agroecosystem. A scarcity of such resources may limit biological control and altering agroecosystem management to alleviate this limitation could contribute to pest management. This is a tactic of ‘conservation biological control’ and includes the provision of flowers for species that consume prey as larvae but require floral resources in their adult stage. The use of flowers for pest management requires an understanding of the interactions between the flowers, pests, biocontrol agents and non-target species. Without this, attempts to enhance biological control might be ineffective or detrimental. This thesis develops our understanding in two areas which have received relatively little attention: the role of flowers in biological control by true omnivores, and the implications of flower use by fourth-trophic-level life-history omnivores. The species studied were the lacewing Micromus tasmaniae and its parasitoid Anacharis zealandica. Buckwheat flowers Fagopyrum esculentum provided floral resources and aphids Acyrthosiphon pisum served as prey. Laboratory experiments with M. tasmaniae demonstrated that although prey were required for reproduction, providing flowers increased survival and oviposition when prey abundance was low. Flowers also decreased prey consumption by the adult lacewings. These experiments therefore revealed the potential for flowers to either enhance or disrupt biological control by M. tasmaniae. Adult M. tasmaniae were collected from a crop containing a strip of flowers. Analyses to determine the presence of prey and pollen in their digestive tracts suggested that predation was more frequent than foraging in flowers. It was concluded that the flower strip probably did not affect biological control by lacewings in that field, but flowers could be significant in other situations. The lifetime fecundity of A. zealandica was greatly increased by the presence of flowers in the laboratory. Providing flowers therefore has the potential to increase parasitism of M. tasmaniae and so disrupt biological control. A. zealandica was also studied in a crop containing a flower strip. Rubidium-marking was used to investigate nectar-feeding and dispersal from the flowers. In addition, the parasitoids’ sugar compositions were determined by HPLC and used to infer feeding histories. Although further work is required to develop the use of these techniques in this system, the results suggested that A. zealandica did not exploit the flower strip. The sugar profiles suggested that honeydew had been consumed by many of the parasitoids. A simulation model was developed to explore the dynamics of aphid, lacewing and parasitoid populations with and without flowers. This suggested that if M. tasmaniae and A. zealandica responded to flowers as in the laboratory, flowers would only have a small effect on biological control within a single period of a lucerne cutting cycle. When parasitoids were present, the direct beneficial effect of flowers on the lacewing population was outweighed by increased parasitism, reducing the potential for biological control in future crops. The results presented in this thesis exemplify the complex interactions that may occur as a consequence of providing floral resources in agroecosystems and re-affirm the need for agroecology to inform the development of sustainable pest management techniques.

The sap beetle genus Carpophilus Stephens (Coleoptera: Nitidulidae) is a large genus consisting of over 200 species and are found worldwide. Several species are important pests of crops and stored products, and are frequently intercepted as part of biosecurity operations. The genus is poorly known taxonomically, and there are several species groups that are challenging to identify by morphological methods. In particular, two species found across the Pacific, C. maculatus Murray and C. oculatus Murray are frequently confused with each other. These two species are similar in size and colour, but differ primarily by the shape of the colour pattern on their elytra. However, this colour pattern is highly variable within both species, leading to ambiguity in the indentification of these species. Within C. oculatus, three subspecies have been described based on differences in the male genitalia and pronotal punctation: C. o. oculatus and C. o. gilloglyi Dobson are distributed widely across the Pacific, while C. o. cheesmani Dobson is known only from Vanuatu. A search of literature records and specimen collections revealed 32 species of Carpophilus recorded from the Pacific region. In addition there remain several unidentified specimens representing at least four species, two of which will be described subsequent to this research. A number of species recorded in the literature may have been misidentified, and these require further field collections and inspection of museum specimens to confirm their presence in the Pacific. To test the validity of the subspecies of C. oculatus, and its distinctiveness from C. maculatus, a phylogeny of available specimens of Carpophilus was inferred from one mitochondrial gene (cytochrome c oxidase subunit I (COI)), and two nuclear genes (28S ribsomal RNA (28S) and the internal transcribed spacer 2 (ITS2)). These data show large genetic distances between the three subspecies of C. oculatus of 7-12%. Given these distances are similar to those between other species in the genus, this indicates these subspecies may be elevated to full species. The data also consistently support a monophyletic relationship between C. o. oculatus and C. o. gilloglyi. Nuclear genes also support C. o. cheesmani as part of a clade with the other subspecies, but these relationships are unresolved in COI. Carpophilus maculatus was not supported as being the sister taxon of the C. o. oculatus and C. o. gilloglyi clade. Other relationships within Carpophilus were unresolved, possibly due to a combination of incomplete taxon sampling, and saturation of substitutions within the COI gene. Phylogeographic analysis of specimens collected from several localities within the range of C. oculatus showed that, with only one exception, there were no shared haplotypes between archipelagoes. This result suggests it may be possible to determine the provenence of intercepted specimens, providing further information regarding potential invasion pathways. A degree of geographic structuring was also present within C. o. gilloglyi, being separated into a western clade found in Fiji and Rotuma and an eastern clade distributed from the Kermadec Islands and Tonga to French Polynesia. This separation was most profound in COI data, with a mean pairwise distance between the clades of 7%. ITS2 data also demonstrates a degree of differentiation between the two clades, based on differences in the insertions and deletions between the clades. The variability in the shape and colour of the elytral pattern of C. oculatus was also investigated. Colour was quantified using a method based on Red-Green-Blue (RGB) colour values derived from digital photographs, while an outline analysis of the elytral pattern was conducted using elliptic Fourier analysis (EFA). Principal Components Analysis of the RGB values and EFA coefficients showed no clear separation between subspecies, nor were any trends correlated with host fruit or collection localities. Variation at all levels and all measures studied in this thesis show that this geographic region and this genus of beetles offer intruiging insights into speciation, biogeography and biological invasions. There is much scope for further research on the causes and consequences of this variation and the lives of these interesting insects.