Harris et al.: Otolith morphometric analysis for species discrimination of Sebastes melanostictus and S. aleut/anus 
239 
Table 3 
Regression statistics for each of 6 morphometric parameters and otolith weight. Each set of 2 rows 
provides the parameter versus log 10 -transformed age in blackspotted rockfish (Sebastes melanostictus) 
and rougheye rockfish ( S. aleutianus), as well as the ^-statistics and P-values used to examine whether 
the species have different slopes or intercepts. Because multiple tests were conducted, a Bonferroni 
correction was applied such that differences are considered significant only if P<0.003. Data used in 
the regression model are from specimens collected in the Gulf of Alaska in 2009 and 2013. 
Otolith variable 
Parameter 
Blackspotted 
rockfish 
Rougheye 
rockfish 
Difference 
t 
P 
Area 
Intercept 
-65.1 
-49.2 
15.9 
3.61 
<0.001 
Slope 
108.5 
123.0 
14.4 
4.09 
<0.001 
Perimeter 
Intercept 
-7.11 
3.67 
10.78 
8.30 
<0.001 
Slope 
34.3 
31.4 
-2.9 
-2.85 
0.004 
Major axis length 
Intercept 
-0.66 
2.13 
2.79 
6.46 
<0.001 
Slope 
10.8 
10.6 
-0.2 
-0.67 
0.506 
Minor axis length 
Intercept 
-0.22 
1.37 
1.59 
6.91 
<0.001 
Slope 
5.55 
5.55 
-0.005 
-0.004 
0.979 
Otolith length 
Intercept 
-0.71 
2.41 
3.12 
6.68 
<0.001 
Slope 
11.56 
11.08 
-0.48 
-1.28 
0.202 
Otolith width 
Intercept 
-0.50 
1.30 
1.80 
7.11 
<0.001 
Slope 
6.20 
6.08 
-0.117 
-0.58 
0.563 
Otolith weight 
Intercept 
-0.357 
-0.456 
-0.099 
-4.49 
<0.001 
Slope 
0.463 
0.695 
0.232 
13.15 
<0.001 
Genetic identification 
• Blackspotted 
° Rougheye 
— Blackspotted 
-- Rougheye 
150 
Age (years) 
Figure 3 
Estimated von Bertalanffy growth curves for rougheye rockfish (Sebastes 
aleutianus ) (open circles, dashed line) and blackspotted rockfish (S. mela¬ 
nostictus) (black circles, solid line). Data used in the growth model come 
from specimens collected in the Gulf of Alaska in 2009 and 2013. Similar 
to the pattern predicted for otolith parameters, rougheye rockfish grow at 
a substantially faster rate and achieve a greater length at age than black¬ 
spotted rockfish. 
Simulation 
The classification process was found to 
be robust with respect to moderate 
errors in the age assigned to each spec¬ 
imen through otolith analysis (Fig. 5). 
Classification accuracy averaged across 
both species remained above 90% when 
bias and CV were each less than 10%. 
Random errors in the aging process 
(represented by CV) generally were not 
a severe concern unless they were par¬ 
ticularly strong with overall accuracy 
exceeding 80% even with a CV of 0.2. In 
contrast, classification accuracy indi¬ 
cates a strong response to directional 
bias and particularly suffers if the 
assigned ages are too low. 
Discussion 
In this study, otolith data, fork length, 
and age were sufficient to correctly 
identify 97% of blackspotted rockfish 
and 86% of rougheye rockfish (Table 6). 
This model’s accuracy provides a 
slight improvement compared with 
the 92-94% of blackspotted rockfish 
