Table 1. Summary statistics for 68^th and 95^th percentiles of spatial errors (and their latitudinal and longitudinal components) associated with each Argos Location Class (LC) for Kalman-filtered Argos telemetry data, collected from bearded seals (N = 6) and ringed seals (N = 10) between 19^th July 2011 and 18^th April 2013 along the west coast of Svalbard.

Argos location class		68th percentile			95th percentile			Fitted T Distribution parameters
	N	Distance (km)	Latitude	Longitude	Distance (km)	Latitude	Longitude	Distance		Latitude			Longitude
								Theta	DF	Theta	DF	Theta	DF
Bearded seals
3	10	0.58	0.25	0.40	0.82	0.53	0.79	-	-	-	-	-	-
2	37	0.61	0.44	0.34	1.33	1.09	0.84	0.30	4.63	0.27	2.18	0.21	1.73
1	40	1.14	0.86	0.62	2.19	2.01	1.27	0.45	2.59	0.55	2.18	0.62	9.13
0	37	2.42	1.89	1.47	4.63	3.64	2.96	1.05	10.01	1.78	50.21	1.47	1.56
A	222	1.38	0.84	0.89	5.61	2.77	3.63	0.46	1.27	0.69	1.91	0.64	1.75
B	1938	2.00	1.31	1.17	4.61	3.58	3.36	0.90	2.63	1.05	2.55	0.94	2.64
Ringed Seals
3	16	0.47	0.27	0.31	0.92	0.69	0.77	0.10	0.57	0.15	0.66	0.16	0.68
2	29	0.74	0.49	0.47	1.47	1.35	0.88	-	-	0.44	3.18	-	-
1	28	1.56	1.24	0.77	2.67	2.15	1.58	0.80	9.28	1.30	3.93	0.83	7.07
0	14	2.58	1.83	1.70	26.43	26.13	5.65	0.85	1.00	0.78	0.81	0.76	1.15
A	174	1.65	1.12	0.99	4.90	2.78	3.71	0.69	1.74	0.77	1.85	0.69	1.72
B	2014	2.04	1.27	1.28	5.75	4.16	4.13	0.78	1.42	0.88	1.51	0.90	1.66
Combined
3	26	0.57 (0.51, 0.49)	0.26	0.36	0.96	0.82	0.91	0.15	0.87	0.18	0.92	0.17	0.81
2	66	0.74(0.67, 1.01)	0.45	0.43	1.47	1.31	0.89	0.36	6.73	0.34	2.53	0.39	8.35
1	68	1.32 (1.02, 1.20)	1.00	0.64	2.67	2.10	1.48	0.61	4.34	0.79	3.56	0.71	14.19
0	51	2.42 (4.15, 4.18)	1.88	1.61	4.67	3.73	3.07	0.85	1.84	1.35	2.01	1.26	3.44
A	396	1.51 (10.19, 6.19)	1.01	0.95	5.07	2.82	3.71	0.58	1.52	0.72	1.87	0.66	1.72
B	3952	2.02 (9.24, 10.28)	1.30	1.22	5.29	3.96	3.66	0.82	1.75	0.95	1.82	0.90	1.95

Values in parentheses in the combined 68^th percentile estimates represent data derived from Least Squares processed Argos telemetry data taken from [9] and [12], respectively. Dashed lines represent a failure to converge on a distribution parameter by maximum likelihood. The 68^th Percentile error estimates from LC 3,2 and 1 were comparable to those described elsewhere, however the estimates for the lower classes were considerably smaller.