Risk maps for range expansion of the Lyme disease vector, Ixodes scapularis, in Canada now and with climate change

Table 1 Algorithms used to predict the occurrence of Canadian census sub-divisions containing resident I. scapularis populations.

	Algorithm	AUC	SE	95% CI	AIC
1	No. of ticks at model equilibrium (T)	0.921	0.030	0.863 – 0.979	214
2	No. of ticks at model equilibrium categorised (Tc)	0.780	0.052	0.678 – 0.881	206
3	Percent forest area (F)	0.387	0.038	0.313 – 0.461	232
4	Index of larval tick immigration (range 255 km: I_L)	0.816	0.052	0.713 – 0.919	211
5	Index of nymphal tick immigration (range 425 km: I_N)	0.896	0.025	0.848 – 0.949	216
6	T * I_N	0.926	0.029	0.869 – 0.983	180
7	Tc * I_N	0.807	0.055	0.699 – 0.914	183
8	T * I_L	0.845	0.052	0.743 – 0.947	207
9	Tc * I_L	0.723	0.056	0.614 – 0.832	207
10	T * I_N* (0.05* I_L)^†	0.926	0.029	0.869 – 0.983	185¹
11	Tc * I_N* (0.05* I_L)^†	0.807	0.055	0.699 – 0.914	186¹
12	T * I_N* F	0.821	0.050	0.723 – 0.919	185²
13	Tc * I_N* F	0.752	0.054	0.646 – 0.858	186²
14	T * I_N* Log₁₀ F	0.832	0.051	0.732 – 0.933	185²
15	Tc * I_N* Log₁₀ F	0.762	0.056	0.652 – 0.871	187²

^†Larva-to-nymph survival of I. scapularis is approximately one twentieth of nymph-to-adult survival (Ogden et al., 2005). ¹ In all logistic regression models containing variables relating to forest cover, these variables were not significant. ² In neither of these models was the variable (0.05* I_L) significant.
The performance of different risk algorithms in ROC analysis is shown: AUC = area under the ROC curve, SE = standard error, 95% CI = 95% confidence interval for AUC. AIC = Aikeke's Information criterion of a logistic regression model for each algorithm in which the outcome was the occurrence of a known I. scapularis population, and the explanatory variables were the algorithm component(s).

ISSN: 1476-072X