304 Insecticide Resistance Ecological Influences Operational Influences Several aspects of pest ecology, including the dynamics, phenology, and dispersal capabilities of pest organisms, act as primary determinants of resistance development However, their influence on selection rates can be unpredictable without a sound knowledge of how they interact with patterns of insecticide use As an example, movement of pests between untreated and treated parts of their range may delay the evolution of resistance, due to the diluting effect of susceptible immigrants Conversely, large-scale movement can also accelerate the spread of resistance by transferring resistance alleles between localities For highly polyphagous crop pests, interactions between pest ecology and insecticide treatments play a particularly critical role in determining selection pressures Key factors to be considered are the seasonality and relative abundance of treated and untreated plant hosts and patterns of migration between hosts at different times of the year A good example relates to the two major bollworm species (Lepidoptera: Noctuidae) attacking cotton in Australia Only H armigera (Figure 3(a)) has developed a strong resistance; H punctigera, despite being an equally important cotton pest, has remained susceptible to all insecticide classes The most likely explanation is that H punctigera occurs in greater abundance on a larger range of unsprayed hosts than H armigera, thereby precluding a significant increase in resistance on treated crops However, polyphagy can sometimes be deceptive In the cotton/vegetable/melon production systems of the southwestern USA, the highly polyphagous whitefly B tabaci has become a devastating pest and a primary target of insecticide sprays The consequences for resistance have varied substantially on a regional basis Resistance problems have proved much more severe and persistent on cotton in south-central Arizona than in the extreme southwest of the state and the adjacent Imperial Valley of California This appears attributable to the higher proportion of unsprayed hosts, especially alfalfa, acting as a buffer to resistance in the latter areas and preventing any directional increase in the severity of resistance over successive seasons In south-central Arizona, these untreated refuges are much less abundant during the time that cotton is treated with insecticides Thus, a large proportion of the local whitefly population is forced through a selection ‘‘bottleneck’’ on cotton and exposed to intense selection for resistance, despite an abundance of alternative hosts at other times of the year Enclosed environments, such as greenhouses and glasshouses, which restrict migration and escape from insecticide exposure under climatic regimes favoring rapid and continuous population growth, provide ideal ecological conditions for selecting resistance genes Very low or zero damage tolerance thresholds for high-value ornamental or vegetable produce accentuate the problem by promoting overfrequent spraying and hence intensify selection for resistance Over the years, these environments have proved potent sources of novel resistance mechanisms for a diverse range of control agents and have presented a particular challenge to attempts at resistance management Although closely linked to the aspects of pest genetics and ecology, operational factors are best distinguished as ones which, in principle at least, are at human’s discretion and can be manipulated to influence selection rates Factors exerting a major influence in this respect include the rate, method, and frequency of applications, their biological persistence, and whether insecticides are used singly or as mixtures of active ingredients Equating operational factors with selection is often difficult, because without a detailed knowledge of the resistance mechanisms present it is impossible to test many of the assumptions on which genetic models of resistance are based Anticipating the selection pressure imposed by a particular application dose of insecticide is a case in point If resistance alleles are present, the only entirely nonselecting doses will be ones sufficiently high to overpower all individuals, regardless of their genetic composition, or ones sufficiently low to kill no insects at all The latter is obviously a trivial option Prospects of achieving the former depend critically on the potency and dominance of resistance genes present A pragmatic solution to this dilemma is to set application doses as far above the tolerance range of SS individuals as economic and environmental constraints permit, in the hope that at least RS genotypes will be effectively controlled Even this approach can backfire badly if resistance turns out to be more common than suspected (resulting in the presence of RR homozygotes) or resistance alleles exhibit an unexpectedly high degree of dominance Unless a high proportion of insects escape exposure altogether, the consequence could then be to select very rapidly and effectively for homozygous resistant populations In practice, concerns over optimizing dose rates to avoid resistance are secondary to ones regarding the application process itself Delivery systems or habitats promoting uneven or inadequate coverage will generally be more prone to selection for resistance, because pests are more likely to encounter exposure conditions under which selection is most intense This was elegantly demonstrated through experiments assessing the relative survival of endosulfan-susceptible and resistant phenotypes of the coffee berry borer (Hypothemus hampei) in coffee plantations treated with this chemical in New Caledonia The practice of spraying plantations from roadsides with vehicle-mounted mistblowers generated gradients in the concentration of endosulfan that resulted in different selection pressures in different parts of each field Similarly, underdosing with the fumigant phosphine in inadequately sealed grainstores has been implicated as a primary cause of resistance to this chemical in a range of stored product pests The timing of insecticide applications relative to the life cycle of a pest can also be an important determinant of resistance A good example relates to the selection of pyrethroid resistance in the cotton bollworm, H armigera, in Australia On cotton foliage freshly treated with the recommended field dose, pyrethroids killed larvae up to 3–4 days old irrespective of whether they were resistant or not by laboratory criteria As the sensitivity of larvae of all genotypes to pyrethroids was found to decline with increasing larval size, the greatest