In this thesis, experiments were conducted in the laboratory and the field to determine whether the provision of floral resources to hoverflies could enhance the biological control of aphids. The overall aim was to clarify hoverfly behaviour and ecology in an agroecosystem in order to understand the potential of these insects for biocontrol under a conservation biological control (CBC) regime. A preliminary experiment in New Zealand compared the effect of different coloured water-traps on catches of the hoverflies Melanostoma fasciatum (Macquart) and Melangyna novaezelandiae (Macquart). Significantly more individuals were caught in completely yellow traps than in traps with green outer walls and yellow inner walls or in completely green traps. This suggested that if a measure of hoverfly numbers relating to a particular distance along a transect is required, consideration should be given to the ability of hoverflies to detect yellow traps from a distance. The use of traps that are green outside would more accurately reflect the local abundance of hoverflies, as the insect would be likely to see the yellow stimulus only when above or close to the trap. Also, the addition of rose water significantly increased the number of M. fasciatum caught. From a suite of flowering plants chosen for their ability in other studies to increase hoverfly visit frequencies, laboratory experiments were conducted in France to determine the plant’s effectiveness at enhancing Episyrphus balteatus (De Geer) ‘fitness’, and to evaluate whether adult feeding on flowers was related to performance. Phacelia (Phacelia tanacetifolia Bentham cv. Balo), followed by buckwheat (Fagopyrum esculentum Moench cv. Katowase) and coriander (Coriandrum sativum L.) gave the optimal reproductive potential of female E. balteatus. There was no correlation between pollen and nectar consumption, and there was no discernible positive correlation between the quantity of pollen ingested and the resulting female performance. Phacelia and buckwheat were then studied as resource subsidies in the field in New Zealand. The effect of incorporating phacelia or buckwheat in the margins of 5 m x 5 m broccoli plots was tested for hoverfly activity and floral ‘preferences’. Hoverflies which had fed on phacelia and buckwheat pollen were found up to 17.5 m from the floral strips and females of M. fasciatum and M. novaezelandiae consumed more phacelia pollen than that of buckwheat in the field. These results support the choice of phacelia as an ideal floral resource subsidy in crops for enhanced biological control by these New Zealand species. The need for studying hoverfly movement in a large-scale field experiment was apparent from the field studies, so the next experiment was carried out in a field 450 × 270 m and flies were marked via their ingestion of the pollen of phacelia. The focus was on the proportion of flies having consumed the pollen. Although large quantities of pollen were found in some hoverfly guts, most did not contain phacelia pollen and very few were captured at 50 m from phacelia, compared with numbers at the border of the floral strip. A possible explanation was that hoverflies feed on a large variety of pollen species, reducing the relative attraction of phacelia flowers. Another possibility was that hoverflies dispersed from the phacelia away from the crop. Also, pollen digestion rates are likely to be a factor. Finally, a series of experiments was conducted in the field and laboratory to study hoverfly efficacy through oviposition and larval behaviour. In field experiments, female M. fasciatum and M. novaezelandiae laid more eggs where buckwheat patches were larger; however higher oviposition rates did not lead to improved aphid population suppression. In greenhouse experiments, larvae of E. balteatus could initiate a decline in aphid numbers at the predator: prey ratio 1: 8.3, however this control did not persist. Experiments in the laboratory showed that hoverfly larvae became more active and left the system while aphid numbers declined or numbers of larvae increased. This behaviour was caused by two factors: hunger and avoidance of conspecific larvae. Further experiments showed that the avoidance of conspecifics was caused by mutual interference rather than cannibalism. The results of this work highlight the importance of hoverfly dispersal ability. Given the observations of foraging behaviour of females and mutual interference observed between larvae, and the lack of success in CBC by hoverflies in experiments at the crop scale, it is essential to assess the impact of insect predators and parasitoids at a landscape scale.