The dispersal of the Western cherry fruit fly, Rhagoletis indifferens (Diptera: Tephritidae), in structured environments

Understanding the changes in animal distribution over time at a landscape scale is critical to many research priorities including the management of agricultural insect pests. Yet little is known about the dispersal capabilities, movement behaviours or landscape-scale habitat selection patterns of pest insects that contribute to population spread and host crop damage. Using the invasion of domestic cherry trees (Prunus spp.) in Lillooet, BC, by the host-specific pest Rhagoletis indifferens Curran (Diptera: Tephritidae) as a case study, I employed spatial analysis, tethered flight mill laboratory experiments, mark-recapture and modelling techniques to examine the dispersal behaviour and movement patterns of these flies. Over a period of seven years, 91% of sampled trees became infested with flies. A spatially explicit kernel analysis approach revealed that the fly population did not expand continuously as in a ripple effect, but expanded through localized spread and spot infestations. Flight mill analysis demonstrated that 87% of flies tested made short flights of less than 500 m, although both sexes are capable of maximum ca. 3 km flights. Of the factors tested for influence on flight behaviour, conspecific crowding most strongly influenced the flight response of individuals. Mark-recapture field experiments demonstrated that 96% of released flies were recaptured within 100 m of their release point, again emphasizing the short distance flight response of the flies. Further field investigations revealed that flies responded strongly to both tree structure and fruit load over short distances. A first principles model using differential attraction to tree structure and fruit load demonstrated that the perceptual range of the fly influenced the distribution patterns obtained in the field. Together, these experimental results were used to develop a spatially explicit, probabilistic population models of fly spread that were then tested against the original fly invasion dataset from Lillooet. The models demonstrated that unrestricted random movement was a poor description of fly distribution on the landscape and that localized movement or combinations of localized and random movement better predicted the risk of infestation. This research advances landscape ecology theory by demonstrating how the spatial distribution of an insect can be visualized and quantified on the landscape.