COS 14-9 - Predicting risks of invasion impact: Interpreting the probability of extreme events

Monday, August 7, 2017: 4:20 PM
D135, Oregon Convention Center


K. Cuddington, University of Waterloo; Stephi Sobek-swant, rare Charitable Research Reserve; Jill C. Crosthwaite, University of Western Ontario; Barry Lyons, Canadian Forest Service; Brent J. Sinclair, University of Western Ontario


Predictions regarding the impact of climate variation on the potential establishment of non-native species are often limited to the use of factors such as average temperature in correlative models. While such climate summary statistics may be related to the physiological or ecological events that limit distribution, in some cases, models based on these statistics can give us a distorted picture of potential risks. For example, key mechanisms that can limit distribution (e.g., overwinter mortality) are often determined by the variance and autocorrelation about such climate means. However, even when we do have a relatively robust mechanistic model that directly incorporates the probability of extreme events (e.g., minimum winter temperature below a mortality threshold), there are still potential pitfalls of interpretation. We illustrate this point with work predicting the probability of emerald ash borer winter mortality.


Using a mechanistic model of both overwintering mortality of emerald ash borer and underbark temperatures of ash. We illustrate that different estimates of risk arise when we consider the simple occurrence of extreme winter minimum temperatures over decades versus the return time of such temperatures. For this species, since tree mortality is thought to be complete after 6 years, the most relevant estimate of the risk of impact is the probability that winter temperatures which cause complete mortality of emerald ash borer occur more frequently than this interval. Simply finding those locations where temperatures that could cause complete mortality have occurred in the past will underestimate this risk. Further, we note that, in general, the average of the coldest temperatures (a commonly reported climate metric), does not help with the task of determining the probably of passing a given winter minimum temperature threshold when this summary statistic separated from the related values of variance and autocorrelation. With this more nuanced definition of risk, our predictions differ from previous estimates of potential range. We suggest that only the most northern of Canadian cities are likely to escape the costs of managing emerald ash borer, and that there will be little or no northern refuge for ash in its native range.