Sagarese et al.: Spatiotemporal overlap of Squalus acanthias and commercial fisheries within the northeast U.S. marine system 
119 
Appendix Table 
Model selection criteria used to identify optimal semivariogram models for spiny dogfish ( Squalus acanthias ) catch 
per unit effort in the northeast U.S. shelf large marine ecosystem between 1989 and 2009 for the sink gillnet (SGN) 
fishery, otter trawl (OT) fishery, and Northeast Fisheries Science Center bottom trawl survey (Survey). Model types 
include nugget (Nug), exponential (Exp), Gaussian (Gau), and spherical (Sph). AIC=Akaike’s information criterion; 
AAIC=difference in AIC with respect to the AIC of the best candidate model; wAIC=model weights. 
Autumn Spring 
Nug 
Exp 
Gau 
Sph 
Nug 
Exp 
Gau 
Sph 
SGN 
AIC 
283.3 
255.1 
304.1 
257.0 
284.6 
245.0 
231.7 
242.5 
A AIC 
28.3 
0.0 
49.0 
1.9 
52.9 
13.3 
0.0 
10.8 
wAIC 
0.0 
72.3 
0.0 
27.7 
0.0 
0.1 
99.4 
0.5 
OT 
AIC 
315.0 
265.1 
263.5 
263.8 
283.3 
254.7 
262.9 
254.8 
AAIC 
51.5 
1.6 
0.0 
0.4 
28.6 
0.0 
8.3 
0.1 
wAIC 
0.0 
19.8 
43.8 
36.4 
0.0 
50.9 
0.8 
48.3 
Survey 
AIC 
263.8 
181.1 
215.1 
205.8 
273.6 
209.7 
289.2 
208.2 
AAIC 
82.6 
0.0 
34.0 
24.7 
65.4 
1.5 
81.0 
0.0 
wAIC 
0.0 
100.0 
0.0 
0.0 
0.0 
32.2 
0.0 
67.8 
Nugget: - y(h) = C 0 = C s ; (2) 
Gaussian: ~t(h) = C 0 + (C S -C 0 )x 
i ( h 2 ' 
1-exp — j 
(3) 
Exponential: 'y(h) = C 0 + (C S -C 0 )x 1-exp ; (4) 
l CL J 
Spherical: 
7 (h) — C 0 + (Cg — Cq) 
1.5 x - -0.5x - 
(5) 
Optimal semivariogram models were selected on the 
basis of the lowest Akaike’s information criterion (AIC) 
calculated with the following equation: 
AIC — n\n(R) + 2p, (6) 
where n = the number of experimental points on the 
semivariogram; 
R = the residual sum of squares; and 
p = the number of parameters in the model 
(Webster and McBratney, 1989). 
AIC values also were reported in terms of delta (A) AIC, 
which represents the difference in AIC with respect to 
the AIC of the best candidate model (AAICi=AICi-mini- 
mum AIC), and weights to determine conditional prob- 
abilities of each model configuration (Wagenmakers 
and Farrell, 2004). 
Results 
Optimal semivariogram models for each of 2 major 
commercial fisheries, the sink gillnet (SGN) fishery 
and the otter trawl (OT) fishery, and the Northeast 
Fisheries Science Center bottom trawl survey incor- 
porated anisotropy and varied seasonally in structure 
(see Table 2 in the main article). For the SGN fishery, 
the spatial correlation of spiny dogfish CPUE was best 
fitted by an exponential model during autumn and a 
Gaussian model during spring (Appdx. Table 1). For the 
OT fishery, Gaussian and exponential models were se- 
lected for autumn and spring, respectively. The spatial 
correlation of spiny dogfish CPUE for the survey was 
best fitted by exponential and spherical models during 
autumn and spring, respectively. Overall, optimal semi- 
variogram models were at least adequate in capturing 
the overall trend indicated by the sample semivario- 
grams (Appdx. Fig. 1). 
Literature cited 
Matheron, G. 
1971. The theory of regionalized variables and its appli- 
cations, 211 p. Ecole Nationale Superieure des Mines 
de Paris, Paris. 
Pebesma E., D. Cornford, G. Dubois, G. B. M. Heuvelink, D. 
Hristopulos, J. Pilz, U. Stohlker, G. Morin, and J. O. Skpien. 
2011. INTAMAP: the design and implementation of an 
interoperable automated interpolation web service. 
Comput. Geosci. 37:343-352. Article 
Wagenmakers, E. J., and S. Farrell. 
2004. AIC model selection using Akaike weights. Psy- 
chon. Bull. Rev. 11:192-196. Article 
Webster, R., and A. B. McBratney. 
1989. On the Akaike Information Criterion for choosing 
models for variograms of soil properties. J. Soil Sci. 
40:493-496. Article 
