Holder and Field: Factors that relate to the occurrence of multiple brooding in Sebastes spp. 
183 
-—■ Multiple brooder -Single brooder 
■-Mean -PacFIN 
— Survey 
100 
0 
Bocaccio 
Canary 
Chilipepper 
Cowcod 
' 
100 
0 ■ 
Flag 
Greenblotched 
Greenspotted 
Greenstriped 
J 
-r 
' o O 
O 
(%) ss 
Halfbanded 
Pink 
Pygmy 
Redstripe 
/ 
r 
r~ 
/ 
(0 
E 
m 100 ■ 
Rosethorn 
Rosy 
Shortbelly 
Silvergray 
/ 
u 
n - 
i/' . . '77 .• 
/ 
a.a™: TT.. 
Speckled 
Squarespot 
Starry 
Stripetail 
100 ■ 
o - 
] 
/ 
r 
s 
100 - 
o - 
Swordspine 
Vermilion and sunset 
Widow 
Yellowtail 
I 
— — ^ 
/j 
33 35 37 39 41 43 45 47 
33 35 37 39 41 43 45 47 33 35 37 39 41 43 45 47 
Latitude (decimal degrees) 
33 35 37 39 41 43 45 47 
Figure 1 
Contextual exploratory analysis of the cumulative distributions of biomass of rockfishes ( Sebastes spp.) of 
the continental shelf in commercial landings based on data from the Pacific Fisheries Information Network 
(PacFIN, dotted line) and in a fishery-independent bottom-trawl survey (Survey, dashed line) along the 
West Coast of the United States in 2004-2015. Each graph shows the mean percentage of biomass from the 
2 data sets (solid line) by latitude (decimal degrees). Black and dark gray lines indicate that the species was 
classified as a multiple-brooding or single-brooding species, respectively. 
Results 
The results of exploratory analysis of the cumulative dis¬ 
tribution of all rockfishes of the continental shelf that were 
present in the trawl survey’s catch data reveal distinct dif¬ 
ferences in latitude between single- and multiple-brooding 
species and similar trends between commercial and survey 
data (Fig. 1). There appears to have been substantial dif¬ 
ferences in both latitude and temperature between single 
and multiple-brooding species and slight differences in 
DO, depth, and all demographic variables between single- 
brooding and multiple-brooding species (Table 2). 
Unsurprisingly, latitude and temperature were highly 
correlated (VIFs: 71 and 67, respectively). To address 
multicollinearity issues, latitude and temperature were 
included in separate models before retesting all VIF scores, 
and they remained separate in the rest of the models. Of 
the 23 models considered, we found substantial support 
for those models that included latitude and found con¬ 
siderably less support for models that included tempera¬ 
ture: (model 1 [Ml]: AAIC c =0.0; model 2 [M2): AAIC c =0.7; 
and model 4 [M4]: AAIC c =3.2; Table 3). There was mar¬ 
ginal support for models that included some demographic 
variables (model 3 [M3]: AAIC c =2.7; Table 3). The poorest 
performing models had essentially no support and were 
those for which latitude and temperature were separate 
(as required by the VIF results) but included all other 
variables (model containing all variables except tempera¬ 
ture: AAIC c =11.6; model containing all variables except 
latitude: AAIC c =14.1; not shown). 
Latitude and temperature were the strongest predic¬ 
tors of whether a shelf rockfish species could be a multiple 
brooder. There was substantial support for Ml, which con¬ 
tained latitude and DO, and for M2, which contained only 
latitude, and there was considerably less support for M4 
and M5, both of which include temperature (Table 3). When 
comparing the coefficients of Ml and M4, which contain DO 
and latitude (Ml) or temperature (M4), the estimate for the 
latitude coefficient indicates that when DO is fixed and lat¬ 
itude increases by 1°, the odds of a shelf rockfish species 
being a multiple brooder is 0.6 times that of being a sin¬ 
gle brooder (for coefficients, see Table 3). Correspondingly, 
