Unwanted spatial bias in predicting establishment of an invasive insect based on simulated demographics

A strategy to estimate the probability of successful establishment of the invasive gypsy moth (given an introduction) is growing in popularity. The strategy calls for an examination of the demographic output of a phenology model of the complete life-cycle to estimate the generational success under the climate of the location under consideration. The probability is maximal where the climate satisfies the life-cycle requirements of all life-stages of 100% of the population every year. The probability decreases where a smaller proportion of the population has its requirements satisfied every year, or where the frequency of unsatisfactory years increases. The strategy can give an unbiased and objective estimate of the probability. However, implementation of the strategy has most often forced unnatural and overly simplistic modifications onto the demographic structure that is simulated by the phenology model, and used an inappropriate and arbitrary calendar date to estimate demographic changes from winter mortality. This produces pronounced spatial bias in the estimates of generational success, and therefore in the estimates of climate-mediated establishment probability. In an examination of the strategy, as implemented in New Zealand, one demographic simplification caused an overestimate of 21% in a southern location; a second simplification caused an overestimate of 17% in a northern location. One hundred percent of the generations were incorrectly considered to have failed in a northern location because of the arbitrary calendar date that was used; and 78% of the generations were incorrectly considered successful in a southern location because of the arbitrary date.