Sagarese et al.: Spatiotemporal overlap of Squalus acanthias and commercial fisheries within the northeast U.S. marine system 
107 
SGN fall SGN spring OT fall OT spring Survey fall Survey spring 
Figure 1 
Seasonal clustering of spiny dogfish (Squalus acanthias) catch per unit 
of effort (CPUE) in the northeast U.S. shelf large marine ecosystem 
between 1989 and 2009 during autumn and spring for the sink gillnet 
(SGN) fishery, otter trawl (OT) fishery, and Northeast Fisheries Science 
Center bottom trawl survey (Survey). Moran’s I statistic ranges from +1.0 
(clustered) to -1.0 (dispersed) with values of 0.0 indicative of a random 
spatial association. The thick horizontal line indicates the median, the 
notched box represents the interquartile range and the 25 th (bottom) and 
75 th (top) percentiles, and the whiskers show either the maximum value 
or 1.5 times the interquartile range. Nonoverlapping notches provide a 
rough impression of the significance of the differences between medians. 
trawl (1996-2008: 1.4% [SD 1.2]) (Waring et al., 2011). 
Mean rates of spiny dogfish bycatch for the northeast 
(94.8% [SD 5.2]) and mid-Atlantic (89.7% [SD 14.0]) 
OT fisheries were similar but differed by a factor of 
2 between the northeast (42.0% [SD 29.1]) and mid- 
Atlantic (22.2% [SD 24.0]) SGN fisheries. Fishery be- 
havior was assumed similar between the northeast 
and mid-Atlantic regions for each commercial fishery 
and, therefore, regional trends were combined for all 
analyses. 
Distribution of seasonal catch 
Both commercial fisheries operated year-round in all 4 
regions that are covered by the NEFSC bottom trawl 
survey. Although spiny dogfish were encountered in 
each region during all months, they were less common- 
ly encountered in the Mid-Atlantic Bight during sum- 
mer months. During cooler months (November through 
April), high catches of spiny dogfish occurred off Cape 
Hatteras, North Carolina, in both fisheries. In contrast, 
during warmer months (May through October), spiny 
dogfish were more common throughout the GM and 
on GB. The SGN fishery operated throughout the NES 
LME with the exception of GB, and 
the OT fishery operated both on the 
NES LME and along its edge. 
Spatial distribution 
Seasonal spatial patterns of spiny 
dogfish CPUE were significantly more 
clustered in the SGN fishery (Moran’s 
I median: -0.31) than in the OT fish- 
ery (Moran’s I median: 0.07-0.11) and 
the bottom trawl survey (Moran’s I 
median: 0.13-0.17 (Fig. 1; signifi- 
cance determined by lack of overlap 
in notches). Although seasonal me- 
dians in Moran’s I were higher dur- 
ing autumn than during spring for 
the OT fishery and the survey, these 
differences were not statistically sig- 
nificant. CPUE estimates from both 
the OT fishery and the bottom trawl 
survey were characterized by low 
Moran’s I (<0.2) throughout the time 
series, indicating a fairly random 
spatial association of spiny dogfish 
CPUE (Fig. 1). Overall, the paucity 
of negative Moran’s I values indicates 
that spiny dogfish CPUE was never 
dispersed. 
Centers of abundance 
Annual centers of spiny dogfish catch- 
es differed significantly during both 
seasons for each fishery and for the 
bottom trawl survey ( P <0.05; Table 
1), matching expected trends in seasonal availability. 
In addition, the locations of these centers differed sig- 
nificantly among modes of fishing (P a( jj <0.0167; Table 
1), with the exceptions of the bottom trawl survey ver- 
sus the SGN fishery during autumn and the survey 
versus the OT fishery during spring. During autumn, 
both the survey and SGN fishery frequently encoun- 
tered spiny dogfish in the southwestern GM (Fig. 2A). 
Centers of abundance from the survey were located far- 
thest offshore in the central GM between the early to 
mid-1990s (Fig. 2A). During spring, spiny dogfish were 
encountered along the edge of the continental shelf in 
SNE during both survey and OT fishery trips (Fig. 2B). 
Centers of spiny dogfish abundance determined from 
SGN fishery data revealed a northward shift during 
much of the 2000s (Fig. 2B). 
Spatial analyses 
Grid size A grid cell size of 0.25° latitude x 0.25° lon- 
gitude was selected and resulted in 468 grid cells, with 
grid areas ranging from 532 km 2 to 644 km 2 in the 
northernmost and southernmost grids, respectively. 
Although annual semivariogram trends were investi- 
