86 differences stemmed from the number of communities attacked over the 30-year time horizon under the fast (634 communities attacked) and slow (386 communities attacked) spread scenarios. In fact, based on the slow spread rate, the infestation had not reached western Canada by the end of the simulation period (Table 5). This is evident in Figure 2, where there is an obvi- ous rise in costs mid-to-late in the simulation under the me- McKenney et al.: Cost of EAB in Canadian Municipalities dium and fast spread rates due to the arrival of EAB at cities in western Canada (particularly Winnipeg, Manitoba); a simi- lar pattern does not appear under the slow dispersal rate. This result is largely driven by the high ash abundance in western communities. There is, of course, an inverse relationship be- tween the present value of the cost estimates and the discount rate. For example, a 2% discount rate resulted in costs rang- ing from $413 million to $870 million, while a 10% discount rate produced costs ranging from $265 million to $422 mil- lion (Table 5). Higher discount rates effectively reduce the present value of future costs. The influence of discount rate was also apparent in Figure 2 where the low discount rate was associated with higher costs, particularly under the fast spread rate and 50% treatment rate; conversely, the high dis- count rate resulted in considerably lower costs and relatively little difference in cost projections between scenarios. Note however that the equivalent annual cost estimates in Table 5 increase as the discount rate increases. While this may seem counterintuitive, it is a standard result because present val- ues of annuities decrease as interest rates increase and in- crease when interest rates decline (see Boardman et al. 2001). As might be expected, increased treatment rates had higher overall costs for the 0%, 2%, and 4% discount rates; however, this pattern was reversed under the 10% discount rate (Table 5). This result is particularly sensitive to the time horizon of the simulation and the spatiotemporal pattern of the spread. Many large urban centers in eastern Canada (e.g., Toronto, Ontario; Montreal, Quebec) were attacked very early in the simulation, thus a large pool of trees accumulated substantial treatment costs by the end of the 30-year period. Since treatment costs are ac- cumulated through time, they are also strongly influenced by the discount rate. For example, under a medium spread rate and 0% discount rate, treating 50% of trees resulted in a total cost of $914 million; for the same spread and treatment rates, this value dropped to $318 million under a 10% discount rate (Table 5). Figure 2. Mean economic impact of EAB over time, based on three scenarios: a) slow spread rate and 0% of ash trees treated, b) medium spread rate and 10% of ash trees treated, and c) fast spread rate and 50% of ash trees treated. ©2012 International Society of Arboriculture Sensitivity Analysis Table 5 also presents standard deviations of the cost distribu- tions for each scenario based on the Monte Carlo simulations. Standard deviation values were generally within 40% of the mean, indicating that the impact estimates are relatively robust to plausible changes in the input parameter values. In eastern Canada, estimated costs were strongly affected by the proportion of ash and the number of trees per unit of road length (Figure 3a). Costs associated with removal, replacement, and treatment of large and/or medium trees made up most of the remaining sig- nificant input parameters. Detection lag had a relatively minor, negative impact on cost estimates in eastern Canada. Higher lag values meant that EAB attacks were detected later, resulting in lower discounted costs or, for grid cells attacked very late in the simulation, costs being pushed outside the 30-year time horizon. In western Canada, the most influential parameter on final cost estimates was the number of large trees/km (Figure 3b). Detection lag had a much stronger influence in the west; since many western communities were attacked very late in the simu- lation, any increase in the detection lag resulted in a significant number of grid cells being excluded from the 30-year analysis. Costs associated with removal, replacement, and treatment of large and/or medium trees made up most of the remaining sig-
May 2012
| Title Name |
Pages |
Delete |
Url |
| Empty |
Search Text Block
Page #page_num
#doc_title
Hi $receivername|$receiveremail,
$sendername|$senderemail wrote these comments for you:
$message
$sendername|$senderemail would like for you to view the following digital edition.
Please click on the page below to be directed to the digital edition:
$thumbnail$pagenum
$link$pagenum
Your form submission was a success. You will be contacted by Washington Gas with follow-up information regarding your request.
This process might take longer please wait
