636 
Fishery Bulletin 96(3), 1 998 
made by the experienced operator (F 72 38 =4.61, 
P<0.001). 
Initial measurement (mm) 
Initial measurement (mm) 
Experienced operator’s 
measurement (mm) 
Figure 1 
Comparison of repeated measurements of 
the experienced operator on fresh (top 
panel) and preserved (center panel) speci- 
mens as well as the comparison between 
experienced and novice operators (bottom 
panel). Solid lines represent least squares 
regressions (Table 2). Dotted lines show the 
1:1 relationship. Dashed lines in the lower 
panel show the 95% prediction intervals. 
Changes in body length 
A total of 1179 larvae representing 9 different spe- 
cies were used in our analyses. Within each length 
interval, the total number of specimens available 
ranged from 10 to 281 individuals, representing 1 to 
5 species (Table 3). 
Within 1-mm length intervals, preservation re- 
sulted in significant increases in body length for in- 
dividuals 3-6 mm fresh standard length and signifi- 
cant decreases in body length for individuals >7 mm 
fresh standard length (Fig. 2, Table 3). In six in- 
stances there were also significant differences among 
species in preserved body length (Table 3). For fresh 
length intervals between 5 and 9 mm there was con- 
sistency in the order of species; preserved specimens 
of Hippoglossoides platessoides were larger for a 
given length interval than Ulvaria subbifurcata and 
Mallotus villosus. However, above and below the 5- 
9 mm length intervals there was variation in the or- 
der of species. In the larger size classes (>12 mm), 
preserved Stichaeus punctatus and Gadus morhua 
were larger than Clupea harengus and Hippogloss- 
oides platessoides. 
The initial variance of fresh standard lengths 
within each length interval ranged from 0.043 to 
0.088 mm 2 , whereas the variance of preserved stan- 
dard lengths of these same individuals was significantly 
greater (Table 3) and ranged from 0.18 to 3.5 mm 2 
(Fig. 3). Furthermore, the variance of the preserved 
length measurements increased significantly with in- 
creasing fresh length (^=0.70, P<0.01, rc=15) (Fig. 3). 
Kendall’s correlation coefficient (x) revealed that 
individual larvae remained at the same rank about 
1 time in 3 (Fig. 4). This effect increased (i.e. 1 time 
in 4 or 5) with increasing fresh length. 
! 
Discussion 
Changes in body length of larval fish due to han- 
dling and preservation are neither uniform nor con- 
sistent among individual animals: there is substan- 
tial variation in reaction to both handling (no mat- 
ter how gentle) and preservation and variation is 
greatest in absolute terms for larger larvae. 
No previous study (Table 1) has employed sample 
sizes large enough to permit the evaluation of 
changes in the distribution or rank of individual lar- 
vae within small length intervals. We found clear 
evidence that changes in body length can be more 
substantial than previously estimated by means of 
analyses applied to a broad range of sizes. Despite 
significant changes in body length for most length 
intervals, our results show that 1) variation about 
