No abstract

The gypsy moth, Lymantria dispar (L.), is a polyphagous defoliator introduced to Medford, Massachusetts in 1869. It has spread to over 860,000 km 2 in North America, but this still only represents of its susceptible host range in the United States. To delay defoliation in the remaining susceptible host range, the government maintains a barrier zone and a quarantine, reflecting a presumption that anthropogenic factors are important in the spread of gypsy moth. We develop a model framework that relates these factors along with biophysical characteristics to a county's susceptibility to gypsy moth invasion. We then compile a dataset for counties within 200 km of the infested area and use trap catch data from 1999 to 2007 to estimate the probability of gypsy moth presence. As expected, gypsy moth is more likely to be found close to the population front and to traps that recorded moths in the previous year. However, when controlling for these factors, our most robust finding is that the use of wood for home heating and energy is consistently positively correlated with the presence of gypsy moth. In contrast, the movement of wood products by industry, which is actively regulated by state and federal governments, is rarely correlated with the presence of gypsy moth. This is consistent with effective regulation of the movement of goods by industry, but not by the public. Our findings provide empirical support for the importance and challenge of firewood as a vector for non-native forest insects.

Ecological forecasting has vast potential to support environmental decision making with repeated, testable predictions across management-relevant timescales and locations. Yet resource managers rarely use co-designed forecasting systems or embed them in decision making. Although prediction of planned management outcomes is particularly important for biological invasions to optimize when and where resources should be allocated, spatial-temporal models of spread typically have not been openly shared, iteratively updated, or interactive to facilitate exploration of management actions. We describe a species-agnostic, open-source framework -called the Pest or Pathogen Spread (PoPS) Forecasting Platform -for co-designing near-term iterative forecasts of biological invasions. Two case studies are presented to demonstrate that iterative calibration yields higher forecast skill than using only the earliest-available data to predict future spread. The PoPS framework is a primary example of an ecological forecasting system that has been both scientifically improved and optimized for real-world decision making through sustained participation and use by management stakeholders.

No abstract