108 
Fishery Bulletin 113(2) 
Table 1 
Permutational multivariate analysis of variance on Bray-Curtis distances for the center of 
spiny dogfish ( Squalus acanthias ) abundance based on 3 modes of fishing, sink gillnet (SGN), 
otter trawl (OT), and the Northeast Fisheries Science Center (NEFSC) bottom trawl survey 
(Surv), during autumn and spring between 1995 and 2009 in the northeast U.S. shelf large 
marine ecosystem. Overall significance (*) is based on an a priori a=0.05, and comparison 
significance is based on an adjusted a of 0.0167 (a=0.05 corrected for 3 comparisons between 
modes of fishing). Pairwise a posteriori comparisons were executed with t, a multivariate ver- 
sion of the f-statistic based on distances. Note that the period from 1989 through 1994 was 
excluded from analysis because of a lack of coverage of the Mid-Atlantic Bight by the NEF- 
SC Northeast Fishery Observer Program for the sink gillnet fishery. df=degrees of freedom; 
SS=sums of squares; MS=mean square; F=pseudo-F ratio test statistic; P (perm)=permutated 
P- value; MC=Monte Carlo asymptotic P-value. 
Source 
df 
SS 
MS 
F 
P (perm) 
P (MC) 
Autumn 
Modes of fishing 
2 
81.48 
40.74 
10.04 
0.0002* 
0.0002* 
Residual 
42 
170.50 
4.06 
Total 
44 
251.97 
Spring 
Modes of fishing 
2 
260.88 
130.44 
18.16 
0.0001* 
0.0001* 
Residual 
42 
301.59 
7.18 
Total 
44 
562.47 
Comparison 
t 
P (perm) 
P (MC) 
Autumn 
SGN vs OT 
3.366 
0.0005* 
0.0019* 
SGN vs Surv 
2.313 
0.0148 
0.0202 
OT vs Surv 
4.918 
0.0001* 
0.0001* 
Spring 
SGN vs OT 
7.692 
0.0001* 
0.0001* 
SGN vs Surv 
3.853 
0.0006* 
0.0004* 
OT vs Surv 
1.421 
0.1659 
0.1682 
gated across gears and seasons, model structure and 
range estimates were similar to those obtained across 
all years. Ranges obtained from optimal semivariogram 
models varied from 0.10 km to 20.19 km and rarely 
exceeded 2 km (Table 2) when examined annually. The 
large range identified for the OT fishery during spring 
was driven by trends in 1995 and 2003. Details on 
semivariogram fits and model selection are provided in 
Appendix Figure 1 and Appendix Table 1, respectively, 
in the Appendix. 
Spatial overlap of spiny dogfish distribution with fishery 
effort For the SGN fishery, SOe during autumn gener- 
ally exceeded estimates during spring, indicating that 
a greater portion of the spiny dogfish stock was avail- 
able to that fishery during autumn (Fig. 3A). During 
spring, SOe gradually increased throughout the 1990s 
and remained relatively high for the remainder of the 
time series (Fig. 3A). For the OT fishery before 1997, 
SOe during spring consistently exceeded SOe during 
autumn (Fig. 3B). SOe for the OT fishery showed a 
trend of slightly increasing values over time during au- 
tumn but no noticeable pattern during spring (Fig. 3B). 
Spatial overlap of spiny dogfish distribution with fishery 
catch For the SGN, lower SOq during spring, compared 
with values during autumn, indicated less overlap be- 
tween spiny dogfish distribution and the areas where 
the SGN encountered them (Fig. 4, A and B). During 
spring, SOq values ranged from 0% in 1990 and 1991 to 
peak estimates in the late 2000s (overall mean 47% [SD 
24]) (Fig. 4B). Predicted spatial distributions of spiny 
dogfish estimated with interpolation methods were ac- 
companied by reasonable 100-fold cross-validation er- 
rors for the SGN fishery and the bottom trawl survey, 
whereas RMSE estimates from mean predictions were 
large for the survey (Table 3). SO\ yielded trends simi- 
lar to those observed for SOq and often exceeded SOq 
estimates during both seasons (Fig. 4, A and B). Dur- 
ing spring, SOi was relatively high with the exception 
of values in the mid-2000s (overall mean 48% [SD 34]) 
(Fig. 4B). SOi estimated for the period 2003-2006 was 
0 because of small predicted catches (<0.5) that were 
converted into either presence (>0.5) or absence (<0.5) 
for analyses. Saturated values (i.e., 100%) of SOi were 
often artifacts of very low sample sizes (< 10 grid cells) 
used during interpolation. 
