378 
Fishery Bulletin 106(4) 
Figure 2 
Category-2 otolith (specimen 225) from Dover sole ( Microstomus pacificus) treated by the break-and-burn method. The “T” 
arrow points to the transition region from fast early growth to slower later growth. In the enlarged section, the two series 
of dots represent different options of interpreting the translucent zones; the final age of 20 years and the maximum age 
of 32 years are shown. The “S” points to a translucent zone that splits to form two translucent zones. The sulcus region 
of the otolith is indicated as a reference point. 
process. Each core was tailored to the size and shape 
of the apparent third year. A secondary guide for the 
coring process was the weight and size of otoliths from 
3-year-old fish in the 2005 survey collection, and an 
additional survey collection (not used for 14 C measure- 
ments) from 2003. These otoliths from 3-year-old fish 
had an average weight of 9.5 mg (±0.4 mg standard er- 
ror) and an average size of 2.41 x 3.84 x 0.68 mm (?i=21). 
The core would occasionally break into several pieces as 
material was removed, but all salvageable pieces were 
retrieved and used. The core size was recorded for all 
intact cores and compared to otoliths from young fish. 
Finally, the cores were cleaned in an ultrasonic cleaner, 
dried, weighed, and then stored in acid-washed vials 
before 14 C analysis. 
14 C analysis 
The 14 C and 13 C of Dover sole otolith cores were mea- 
sured at the National Ocean Sciences Accelerator Mass 
Spectrometry Facility at the Woods Hole Oceanographic 
Institution, Woods Hole, MA. Samples were treated at 
the Woods Hole Oceanographic Institution with a routine 
acid hydrolysis procedure to produce a graphite target, 
and analyzed with accelerator mass spectrometry. We 
report results as 4 14 C, which is defined in Stuiver and 
Polach (1977) as the relative difference between an inter- 
national standard (base year 1950) and sample activity. 
The zi 14 C is normalized to 1950 and corrected for isotopic 
fractionation with the d 13 C measurement and normalized 
to a d 13 C VPDB value of -25 %c. 
To evaluate the Dover sole ages, we compared the oto- 
lith 4 14 C results from the otolith cores with the Pacific 
halibut reference chronology using several procedures. 
Initially, the 4 14 C from the Dover sole otolith cores in 
both categories was plotted along with a loess (locally 
weighted least squares) smoothed curve of the 4 14 C in 
Pacific halibut. All loess-smoothed curves in this study 
were fitted by using Splus (Insightful Corp., Seattle, 
WA) (Chambers and Hastie, 1992) with a span of 2/3 
and a degree of 2. To further analyze the 4 14 C results in 
comparison with the reference chronology, we used three 
procedures first introduced by Kastelle et al. (2008). 
The first procedure was to purposely bias the category-1 
ages by 0, ±1, ±2, and ±4 years, generating seven sets 
of ages. For each of these seven sets of biased ages, pos- 
ited birth years were calculated and a sum of squared 
residuals ( SSR ) between the A 14 C in Dover sole otoliths 
and the loess smoothed data in the Pacific halibut ref- 
erence 4 14 C chronology was calculated. The smallest of 
the seven SSR s indicated which purposely biased set of 
ages represented the best fit, and thereby indicated if 
an overall aging error in the Dover sole ages existed. 
We call this procedure the “unstandardized” analysis. 
In the second procedure used to investigate the ac- 
curacy of Dover sole ages in category 1 we performed 
a “standardization” of the measured 4 14 C values (Kas- 
telle et ah, 2008). This is a linear transformation of 
the 4 14 C values in the Dover sole otolith cores which 
removes any difference in scale between the valida- 
tion sample 4 14 C measurements and the reference 
chronology. It does not change the timing of the val- 
idation sample 4 14 C values or their relative magni- 
tude. For this standardization, let {v ^1 be the series 
of j- 1, ..., n validation observations of 4 14 C, where 
y\j] refers to the year core j was formed. We defined 
the standardized series for {v^} as {v^^j=(v y[j j + p)/a|, 
where p and o can be estimated by a least squares fit 
(i.e., by minimizing the SSR) to the loess-smoothed 
curve of the reference chronology data set {Z ^j}: 
