694 
Fishery Bulletin 99(4) 
Figure 1 
Uncorrected (A) and corrected (B) standard length determined by using best 
shrinkage correction model, see Table 1) for laboratory-reared walleye pol- 
lock ( Theragra chalcogramma) larvae subjected to various shrinkage treat- 
ments and preserved in 95% ethanol (n=91). The 95% confidence limits are 
shown for each predicted value, and the diagonal line shows a 1:1 ratio. 
routine data collection, we have shown improved accuracy 
of shrinkage correction models by use of measurements in 
addition to preserved SL: for 5% formalin, 7% more of the 
variance was explained by adding body depth; and for 95% 
ethanol, 8% more of the variance was explained by includ- 
ing body depth and otolith diameter. For 95% ethanol, the 
shrinkage correction model formulated in our study may 
significantly underestimate SL when shrinkage is high, 
but because there were only two larvae in this category, 
results were inconclusive. 
Shrinkage correction models are usually formulated by 
using fed, laboratory-reared larvae (Jennings, 1991; Thei- 
lacker and Porter, 1995; Fox, 1996), but individual larvae 
response to handling and preservation can vary signifi- 
cantly (Pepin et ah, 1998). These correction models were 
developed to allow more accurate calculation of the growth 
rate of fish larvae in the sea and to apply laboratory- 
derived indices to the field. We have shown factors in ad- 
dition to preserved SL that may improve the accuracy of 
live-length estimates for walleye pollock. 
