7867

• UCREFRP Catalog Number: 7867
• Author: Haines, G. B. and T. Modde
• Title: Evaluation of Marking Techniques to Estimate Population Size and First-Year Survival of Colorado Squawfish
• USFW Year: 1996
• USFW - Doc Type: North American Journal of Fisheries Management
• Copyright Material: YES

MARKING TECHNIQUES FOR COLORADO SQUA WFISH

909

.A.

TABLE 2.-Mark-recapture data for autumn 1992 and
spring 1993 in a 16-1an reach of the Green River near
Jensen, Utah. Abbreviations are j = sample pass, m(j) =
number of marked fish captured on pass j, u(j) = number
of unmarked fish captured on pass j, M(j) = number of
marked fish in the population at beginning of jth sample,
N = number of sites sampled (primarily backwaters but
occasionally low-velocity side channels), CPUE = catch
I1er seine haul.

Dates j m(j) u(j) M(j) N CPUE (SE)
Autumn 1992
22 Sep- 0 383 0 9 NAa
13 Oct
130ct- 2 5 115 297 18 0.720
14 Oct (0.203)
21 Oct- 3 7 150 297 24 1.694
22 Oct (0.717)
Spring 1993
6 Apr- 0 104 0 16 0.483
8 Apr (0.]94)
12 Apr- 2 14 74 104 17 0.394
15 Apr (0.122)
19 Apr- 3 14 247 183 29 0.828
22 Apr (0.345)

a Only selected backwaters were sampled; thus CPUE was not ap-
propriale.

vival averaged 92% (140 of 153 from six back-
waters); on five occasions, fish (N = 98) were held
overnight and had over 95% survival.
Young Colorado squawfish moved only short
distances downstream during autumn and spring
sampling. In autumn, 10 of 13 (77%) recaptures
were from the backwater of origin, as were 25 of
27 (93%) fish marked in the spring. Of those fish
that moved, all moved less than 3.2 km. We es-
timated the probability that a particular fish moved
from one 1.6-km stream section downstream to the
next section per day was 0.044 for the autumn
sampling period. No fish moved from one 1.6-km
section downstream to the next section during
spring sampling.
No marked fish were collected outside the study
reach. On 20 October 1992, the first 11.3 km down-
I
stream of the study reach were sampled and 107
age-O Colorado squawfish were captured; all were
unmarked. On 20-21 April 1993, the first 16.2 km
downstream of the study reach were sampled and
55 yearling Colorado squawfish were captured; all
were unmarked.
Five Colorado squawfish that were marked in
the autumn were caught the following spring. Two
were caught in the backwater of origin, two moved
downstream (0.3 and 4.2 km), and one moved up-
stream (7.4 km).

Parameter Estimates
Two biological results influenced our selection
of an appropriate model for estimating population
size of young Colorado squawfish. Our observa-
tion of the limited movement by recaptured fish
within and outside the study reach as well as that
of Tyus (1991) indicated that the probability of
capture changes markedly with temperature and
therefore among sampling passes. For these rea-
sons we selected model M(t) of Otis et al. (1978)
as the best representation of our situation. This
model assumed population closure but allowed the
probability of capture to vary among sampling
passes.
We estimated an autumn population size of
5,595 (SE, 1,077) with 95% confidence limits of
::':::2,112 and a spring population of 2,523 (SE, 490)
with 95% confidence limits of ::':::961. Estimates of
probability of capture were 0.05, 0.02, and 0.03
for the three sampling passes in the autumn, and
0.04,0.03, and 0.10 for the spring. Winter survival
was estimated as 0.45 (SE, 0.122) with 95% con-
fidence limits of ::':::0.24.
Survival estimates were also calculated from
CPUE data (Table 2). The first pass during autumn
of 1992 consisted of sampling only selected back-
waters; thus, CPUE was not calculated for this
occasion. Survival was calculated from the re-
maining passes; estimates ranged from 0.23 to
1.15, and the coefficients of variation (CV = SEt
mean) from 0.42 to 0.59 (mean, 0.52). The mean
CV was about twice that of the mark-recapture
estimate of 0.27. The survival estimate calculated
from combined catch per seine haul data (Le., all
sampling passes combined), which involved ap-
proximately the same seining effort as that for
mark-recapture, was 0.49 (SE, 0.210). This was
similar to the mark-recapture estimate, but the CV
was 1.6 times greater.

Simulations
Under the study sampling regime (16-km reach,
three sampling passes, probability of capture rang-
ing from 0.02 to 0.05 in the autumn and from 0.03
to 0.10 in the spring), simulated CVs were 0.23
and 0.18 for autumn and spring population esti-
mates, respectively (trials 1, 2; Table 3). As ex-
pected, these values were similar to the empirical
values (CV = 0.19 for both autumn and spring).
The relative biases due to small samples were 5.7%
and 3.2%, and at the most likely rate of immigra-
tion-emigration (probability a fish moves down-
stream from a 1.6-km reach to the next is 0.044
per day), relative bias increased to 14.5% (trial 9).