Scale‐appropriate spatial modelling to support area‐wide management of a polyphagous fruit fly (Diptera: Tephritidae)

Abstract

AbstractControl of mobile pests frequently requires area‐wide management (AWM) that spans commercial and non‐commercial habitats. Spatial modelling of habitat suitability can guide investment and implementation of AWM, but current approaches rarely capture population drivers, including local foraging, at an appropriate spatial resolution. To support the development of AWM for the fruit fly pest, Bactrocera tryoni (Tephritidae), we developed a habitat suitability model for the three premier fruit‐growing regions in south‐eastern Australia (~34,780 km2). Expert elicitation and published literature was used to develop a Bayesian network to model the drivers of B. tryoni habitat suitability, as determined by the ability of populations to persist and increase. The effect of uncertainty was tested through sensitivity analysis. The model was then linked with spatially explicit data (at 10 m resolution) to generate risk maps, using moving windows to capture local foraging movement. Habitat suitability was most strongly influenced by host availability within a distance of 200 m. Climate stress, and soil moisture for pupation, was also limiting. Experts were uncertain regarding the relative importance of drivers of host availability (host preference, host density, fruit seasonality), but this did not greatly affect model outputs. Independent trapping data supported model predictions, but their value was limited as traps were placed almost exclusively in optimal or suitable habitat. Amenability to AWM, when assessed as the ratio of suitable or optimal habitat that was under non‐horticultural versus horticultural land‐uses, differed by region (0.15–1.17). However, risk‐mapping did identify where ratios were locally most favourable (lowest). Also, predominantly local dispersal by B. tryoni suggests AWM for pest suppression could be applied at a landscape‐scale. Results show that a relatively simple model could capture the multi‐scale drivers of population dynamics and the complexity of landscapes sufficiently to guide AWM of a mobile pest.