8021

• UCREFRP Catalog Number: 8021
• Author: Bestgen, K. R. and J. M. Bundy
• Title: Environmental Factors Affect Daily Increment Deposition and Otolith Growth in Young Colorado Squawfish
• USFW Year: 1998
• USFW - Doc Type: Transactions of the American Fisheries Society
• Copyright Material: YES

OTOLITH MICROSTRUCTURE OF COLORADO SQUAWFISH

107

crements of each otolith were counted three times
by the senior author without knowledge of the
treatment. The three counts were averaged. A sec-
ond experienced reader also counted increments
for the 220C treatments (constant and fluctuating
temperatures) to determine the extent of reader
variation or systematic bias.
Each otolith increment consisted of a relatively
wide, light-opaque inner band (L-zone) and a rel-
atively narrow and dark outside band (D-zone; ter-
minology after Kalish et al. 1995). The first in-
crement was one of relatively high contrast (dis-
tinct L-zone) that surrounded the otolith core. The
outer margins of otoliths examined for age vali-
dation were almost always opaque. We interpreted
this as representing the L-zone for the day the fish
was sampled and counted it as the last increment.
We used least-squares regression of increment
count (dependent variable) as a function of known
age (independent variable) to test the hypotheses
that (1) increments were first formed at hatch and
(2) that detectable increments were formed daily.
Regression intercepts not significantly different
than zero (increment count == 0 at hatch) would
support hypothesis (1), and regression slopes not
significantly different from one would support hy-
pothesis (2) (e.g., Rice 1987). Instead of con-
ducting retrospective (a posteriori) power analysis
to detect the likelihood of type II statistical errors
and an arbitrary effect size, we calculated confi-
dence intervals about intercept and slope esti-
mates. Like power analysis, which is mostly a pro-
spective (a priori) experimental design tool (Steidl
et al. 1997), confidence intervals allow assessment
of the magnitude of effect that is detectable with
the data and offer straightforward measures of the
reliability of parameter estimates. Because age of
wild fish will be estimated from counts of daily
increments, prediction intervals were calculated
for known (true) age (independent variable) via
inverse regression (Draper and Smith 1981; Rice
1987) instead of for estimated age (dependent vari-
able).
Analysis of covariance (ANCOV A) was used
both to investigate reader bias (via comparisons of
estimated age with true age) and to detect effects
of temperature on periodicity of increment for-
mation. The ANCOV A interaction of treatment X
class variable (reader or temperature) and the class
variable terms of the analysis tested for homoge-
neity of regression line slopes and intercepts, re-
spectively, among treatments. In addition to hy-
pothesis tests such as ANCOV A that determine if
treatments were statistically significant, least-

squares means (population marginal means; Searle
et al. 1980) were calculated for groups (or treat-
ments) after covariates were accounted for. Dif-
ferences between treatment means and confidence
intervals around those differences give informa-
tion about the biological significance of treat-
ments, regardless of results of statistical tests
(Yoccoz 1991; Johnson 1995).
Preserved specimens were used to assess rela-
tionships between otolith size and fish total length
and to calculate slope and intercept parameters for
use in back-calculation of fish lengths at previous
ages (Carlander 1981). Effects of temperature lev-
el and regime (constant or fluctuating) on otolith
growth were tested by ANCOV A.

Growth Effects

Variables affecting otolith and somatic (body
length) growth were studied in four experiments.
Experiment 1 was to determine if otolith growth
rates were directly proportional to somatic growth
when fish had different somatic growth rates. Fish
from single slow- and fast-growth treatment aquar-
ia (which differed in stocking rate: Table 1) were
sampled at 3, 22, 28, 56, 82, and 122 d of age and
preserved in 100% ethanol. In this experiment and
the ones that follow, individual fish from single
aquaria were treated as statistically independent
experimental units. Three to 10 specimens per pe-
riod and treatment (depending upon availability)
were measured for total length (TL), and the left
lapillus was removed from each fish and measured
as described above. Analysis of covariance and
least-squares means were used to compare differ-
ences in otolith size between treatments.
Experiment 2 was to determine somatic and oto-
lith growth rates before, during, and after a 6-d
starvation period. Healthy and actively feeding lI-
d-old larvae were lightly anaesthetized with tri-
caine (MS-222, 100 mglL), measured (TL) under
a dissecting microscope fitted with an ocular mi-
crometer, allowed to recover, placed in 0.24-L cups
(one per cup), and assigned to a temperature treat-
ment (constant or fluctuating temperature centered
on 220C; Table 1). Each larva was remeasured after
6 d more of feeding, after 6 d of starvation, and
after 6 d of renewed feeding. Two fish from each
treatment regime died during anaesthesia or were
physically damaged, so their complete growth his-
tories were not available. Diameter of the left la-
pillus from each fish was measured at 1,000X mag-
nification.
In experiment 3, groups of 200 larvae 6 d old
(age at first feeding) were randomly assigned to