Evaluating spatial variation in the flight period of Lymantria dispar to optimize the timing of slow-the-spread surveillance and treatments

The seasonal patterns of insect abundance are largely determined by temperature, photoperiod, and/or their interaction. Phenological models enable prediction of insect development and are commonly used to guide timing of management interventions. Thus, spatiotemporal variation in abiotic conditions can drive shifts in insect seasonality and potentially inhibit efficacy of time-sensitive surveillance and treatments. The US Forest Service national Slow the Spread (STS) program aims to contain spread of non-native Lymantria dispar (L.) (Lepidoptera: Erebidae) populations via intensive trapping that is used to trigger a variety of treatments including (i) Bacillus thuringiensis kurstaki (Btk) and (ii) mate disruption. Operational timing of trapping and both types of treatments rely on accurate forecasts of L. dispar phenology made using the spongy moth life stage model (GLS), a phenological model of each life stage. The GLS model is an extremely important tool in the management of L dispar, but spatial variation in the accuracy of phenological predictions from the GLS model have not been critically evaluated across a range of climates. We analyzed trap capture in automated traps deployed across the leading edge of the L. dispar invasion to quantify variation in the beginning, peak, and end of male flight, comparing observed vs. predicted flight patterns. The spatial transferability and variation of phenological models for L. dispar and the relevance of such variation for STS treatments will be discussed.