NOTE Pepin et a I.: Changes in the distribution of larval fish body length following preservation 
637 
Table 2 
Results of the comparison of initial ( I ) and repeated (R) measurements on fresh and preserved larvae by the experienced ( E ) and 
novice (N) operators. Numbers in square brackets represent the standard error of the intercept (a) and slope ((3). The sixth and 
seventh columns show the test of the hypotheses that the intercept is not significantly different from 0 and that the slope is not 
significantly different from 1. 
Treatment 
Initial 
Repeat 
Regression (y = a + fix) 
F - Value 
f-value 
H u : a = 0 
t-value 
tf 0 :p=l 
Residual 
variance 
Fresh 
E 
E 
R = 0.074 [0.128] + 0.983 [0.0229] / 
F 1 40 =1848,P< 0.001 
0.58, P> 0.05 
0.74, ns 
0.020 
Preserved 
E 
E 
R = 0.113 [0.126] + 0.977 [0.0228] / 
F l ,38=183 3. P < 0.001 
0.90, P> 0.05 
1.01, ns 
0.018 
Preserved 
E 
N 
R = -0.271 [0.149] + 0.959 [0.0263]/ 
F 1 72 =1414,P< 0.001 
1.81,0.1 >P> 0.05 
1.56, ns 
0.083 
4 -i 
o 
o 
o 
-4 -| — i — 1 — i — | — i — | — i — | — i — |-2 h — | — i — | — i — |- 
0 2 4 6 8 10 12 14 16 
Fresh length (mm) 
Figure 2 
Box plot of the 25th, 50th, and 75th percentile of the dif- 
ference in mean individual length of larval fish within each 
millimeter fresh length interval. Capped whiskers show 
the 10th and 90th percentiles and the open circles show 
the distribution of observations outside those confines. The 
dotted lines represent the 95% confidence intervals of the 
residual population variance from the comparison of mea- 
surement error between operators. 
the mean can be substantial and that 2) the relative 
position of an individual larva within a length inter- 
val may shift substantially. 
Our findings are consistent with those of Hay 
(1982), Theilacker (1986), and Fox (1996) who all 
found that postpreservation changes in body length 
were greatest for large larvae. However, the marked 
increase in variance of preserved larval lengths in 
relation to the initial measurements was not caused 
by operator error. The residual variance of repeated 
1 r 
hico^iniostooio'-ort 
^oic l )4u l )(f!)Nc6j7V T I’ 
<T> O ■*“ CN 
Length interval (mm) 
Figure 3 
Variance in body length of fresh (circles) and pre- 
served (squares) specimens within each millimeter 
interval. 
measurements among operators was of the same or- 
der as the variance in fresh lengths within individual 
length intervals and was comparable to that obtained 
by Jennings ( 1991), Hjorleifsson and Klein-Macphee 
(1992), and Fox (1996). However, Fox (1996) found 
some evidence of increased variance with increased 
length, in contrast with our results. We conclude that 
remeasurement of larvae per se is unlikely to be a 
major contributor to our observation that the order 
of individual larvae, in relation to population esti- 
mates, is not maintained following preservation. 
Our finding of postpreservation increases in body 
length, for individuals <6 mm SL, contrasts with most 
studies of the effects of formaldehyde preservation 
on larval fish. Fox (1996) and studies of other meth- 
