9697

• UCREFRP Catalog Number: 9697
• Author: Bestgen, K.R., D.W. Beyers, J.A. Rice and G.B. Haines.
• Title: Factors affecting recruitment of young Colorado pikeminnow
• USFW Year: 2006.
• USFW - Doc Type: synthesis of predation experiments, field studies, and individual-based modeling.
• Copyright Material: YES

YOUNG COLORADO PIKEMINNOW RECRUITMENT

The response variable measured for each mesocosm
trial was number of larvae eaten. To estimate the
number of attacks that occurred in a trial, the number of
larvae eaten was divided by its respective size-
dependent capture probability (equations 1 and 2;
e.g., if 10 larvae were eaten and probability of capture
was 0.25, then, attacks = 10/0.25 = 40). Total attacks
was then divided by the number of predators (usually
3). A global general linear model was used to estimate
an equation for number of attacks per trial per
individual larva (distributed as Poisson, number of
prey was used as an offset) as a function of predator
TL, prey TL, predator TL z, prey TL z, predator feeding
strategy, water clarity, alternative prey, and frrst-order
interactions. Predator feeding strategy was included
because preliminary observations suggested that
groups of red shiners exhibited different levels of
predation that were not related to their size. We
assumed that the different behaviors represented the
range of feeding strategies employed by groups of red
shiners in the natural environment. A maximum
likelihood model was fit, and model selection was by
AlCc' Model selection suggested that the best predic-
tors of attack rate were prey TL, predator feeding
strategy, water clarity, and alternative prey. The form
of the attack-rate equation was

y = -4.75 + 0.147(xJ) + 0.91O(xz) + l.157(X3)
- 0.439(fsl) + 0.1036(fsz) - 4.45(fs3)
- 1.88(fs4); (3)

y = loge (number of attacks on an individual larva
in a mesocosm per 6 h per predator);
Xl = prey TL (mm);
Xz = 1 if water is clear, 0 if water is turbid;
x3 = 1 if alternative prey is absent, 0 if present;
fS1 = 1 if simulating the feeding strategy of predator
group 1, else = 0;
fsz = 1 if simulating the feeding strategy of predator
group 2, else = 0;
fS3 = 1 if simulating the feeding strategy of predator
group 3, else = 0;
fs 4 = 1 if simulating the feeding strategy of predator
group 4, else = O.

Note that when all fs coefficients = 0, the model
simulates the feeding strategy of predator group 5.
Standard errors for coefficients are as follows: intercept
= 0.557, Xl = 0.0294, Xz = 0.266, x3 = 0.331, fSl =
0.272, fsz = 0.239, fS3 = 1.224, and fS4 = 0.360. A
goodness-of-fit test did not detect significant lack of fit
of the data to the model (P = 0.25). Model fit was
further assessed by plotting deviance and chi-square

1727

residuals as functions of predictor variables and
predicted values. No trends in residuals were observed.

Field Studies
Drift net samples.-Colorado pikemim10w larvae
were sampled with drift nets at two locations in 1991
and 1992. The lower Yampa River site was located
about 1 lcm upstream of the confluence of the Yampa
and Green rivers. The second site was located in the
Green River about 15 lcm upstream of Green River,
Utah (Figure 2). Sampling started in mid to late June 4-
6 weeks after peak spring discharge when daytime
water temperature exceeded 160C (Nesler et al. 1988;
Tyus and Haines 1991; Bestgen et al. 1998). Sampling
continued for 4-6 weeks after the first pikeminnow was
captured and ended when no larvae were captured for
3-5 consecutive days (usually mid-August; Bestgen et
al. 1998). Colorado pikeminnow larvae were sampled
daily at dawn with conical drift nets (0.15-mz mouth
diameter, 4-m long, 560-/illl mesh) set near shore in
water 30-40 em deep. Three nets were set on each
sampling occasion for up to 2 h, but sampling stopped
if debris load exceeded 3.8 L/sample. Flowmeters
(General Oceanics, Miami, Florida, Model 2030) were
suspended in each net mouth to continuously measure
through-net water velocity during sampling periods.
Additional diel and cross-charrnel sampling in 1992 in
the Yampa River showed that daily nearshore samples
collected at dawn adequately represented the spatial
and temporal distributions of drift of Colorado pike-
minnow in the Yampa River (Bestgen 1997; Bestgen et
al. 1998). Therefore, dawn samples were the only ones
used to describe patterns of reproduction and drift
abundance of Colorado pikeminnow in the Green River
system in this study. We were confident the sampling
regime encompassed the Colorado pikemim10w repro-
ductive season because there were few or no larvae
captured the first several days of sampling, larvae were
absent in collections before conclusion of sampling,
and only a few autumn juveniles had estimated
hatching dates earlier than drift-net sampling dates.
Samples were fixed and preserved in 100% ethanol. A
standardized protocol was used to remove fish from
sample debris within 4 h of collection. Tests with five
samples showed that the protocol recovered 100% of a
known number of larvae.
Seine samples.- Two 20-lcm reaches in the Green
River representative of habitat for young-of-year
Colorado pikeminnow were sampled in 1991 and
1992. One reach (middle Green River nursery area)
was located 407-427 river kilometers (RK) upstream
of the confluence of the Green and Colorado rivers.
The other reach (lower Green River nursery area) was
located 69-89 RK upstream of the confluence.