506 



Fishery Bulletin 102(3) 



to be determined from partially digested prey hard 

 remains. 



Experimentally derived pollock DCFs were deter- 

 mined from three distinct size ranges of fish (28.5-45 

 cm FL), but the degree of erosion for each structure 

 within each condition category did not show any sig- 

 nificant differences across this range. We also found the 

 relative shape, structure, and proportion of the morpho- 

 logical features used to estimate erosion were consistent 

 for both smaller and larger fish. We therefore believe 

 DCFs can be used for fish outside of the experimental 

 size range of this study. Average size reduction varied 

 between different pollock structures (12.2-24.7%) and 

 also between condition categories, as they did for Atka 

 mackerel (Table 2). We determined that pollock otoliths 

 in fair condition were reduced by 149J in length, close to 

 the 20% value estimated from reference material (Sin- 

 clair et al., 1994). Our criteria for defining a condition 

 category of fair for pollock otoliths equates to a grade 

 between low amounts and medium amounts of diges- 

 tion as defined by Tollit et al. (1997) for Atlantic cod 

 (which has a similar looking otolith). Our value of 14% 

 lies midway between those determined for cod otoliths 

 graded low and medium. 



Jaw bones (DENT) were by far the largest structure 

 used in our study but do not appear to pass through 



the pyloric sphincter without some level of digestion. 

 Usually only the hammerhead and stock (representing 

 less than a third of the whole structure) are recov- 

 ered in scats. The large size accounts for the relatively 

 greater percent mean reduction and hence higher DCF 

 of DENT structures graded either in good or fair con- 

 dition (Table 2). Although quadrates (QUAD) are also 

 relatively large structures with a projecting ridge that 

 is often much reduced when found in scats, we found 

 QUAD structures of Atka mackerel recovered in scats 

 from field studies and captive sea lion studies in rela- 

 tively better condition than those of pollock, leading 

 to differences in grading criteria and resulting DCFs 

 (Tables 1 and 2). Part of the reason may be that the 

 horns on a pollock QUAD project widthwise more than 

 those of Atka mackerel, presenting a greater surface 

 area for digestive erosion of the structural feature that 

 is measured to estimate size (Fig. 1). 



Our overall 95^f confidence intervals around DCFs 

 were generally narrow (Table 2), highlighting the tight 

 fit of the regression formulae used and the benefits of 

 partitioning the data into specific categories. Our boot- 

 strap analysis suggests that resampling errors were 

 the major source of error in calculating DCFs. Future 

 research should concentrate on improving sample sizes 

 for data on percentage size reduction of bones for each 



