Table 12.

Residual spatial autocorrelation of the best performing models for LA. Provided are p-values from global Moran's I tests for spatial autocorrelation, with significant p-values (p < 0.05) in bold. To create the spatial neighbourhood matrix required for computing Moran's I, we use k-nearest-neighbours with k = 50.

	NS1	NS2	S1	S2	S3
$d_i^{\mathrm{day}}$	0.100	0.028	0.369	0.369	1.000
$d_i^{\mathrm{week}}$	0.078	0.056	0.699	0.699	1.000
$d_i^{\mathrm{month}}$	0.002	0.002	0.125	0.125	1.000
log(Δti)	0.001	0.000	0.142	0.979	1.000