9711

• UCREFRP Catalog Number: 9711
• Author: Coggins, L.G.
• Title: Abundance trends and status of the Little Colorado River population of humpback chub
• USFW Year: 2008.
• USFW - Doc Type: an update considering 1989-2006 data.
• Copyright Material: NO

that the USFWS growth curve tends to predict somewhat
smaller sizes at young ages and larger sizes at older ages than
is implied by the mark-recapture data. The TIGM and TDGM
predict very similar length-at-age with the exception of
10-25-year-old fish. Two features are apparent in the TDGM:
(I) temperature-dependent periodic change in growth rate at
ages younger than about age-5, and (2) an apparent "bend"
in relationship at approximately age-4. This age con'esponds
to the length-at-transition (Lf) where HBC are rapidly shift-
ing from primarily LCR occupancy to primarily mainstem
Colorado River occupancy. A L, length of 236 mm TL is most
strongly supported by the data and the TDGM (table 3).
Finally, it is informative to use the TDGM to predict
monthly growth increments as a function of TL. These predic-
tions, based solely on field data, can then be compared to
laboratory observations of the same or similar species. Growth
rate predictions from the LCR population are much higher
than a population experiencing constant IOoC temperature
(fig. 24). This latter growth rate is presented as a prediction of
monthly growth rates that would be observed in the mainstem
Colorado River.

Incorporation of Ageing Error in ASMR
Assessments

The TDGM and the procedures identified in the methods
sections were used to construct seasonal PCp , I) matrices. The
resulting probability distributions were subsequently plotted as
surfaces to allow examination of the uncertainty in predicting
age given length (fig. 25). The most obvious feature of these
probability surfaces is the increasing uncertainty in age assign-
ment with increasing length. For instance and considering the
April-June Pea II) surface (fig. 27, top left panel), one can see
that a 150-mm-TL fish is age-2 with highest probability, but
there is some chance that it could be any age between age-l
and age-4. In contrast, a 300 mm fish is approximately age-7
with highest probability, but could be as young as age-4 or as
old as age-18. It is such uncertainty that this assessment was
intended to incorporate.
As described in the methods section, a stochastic assign-
ment of age to each fish was made using the appropriate
Pea II) matrix, depending on the time of year the fish was first
captured. Using this procedure. a total of 1,000 input datasets
were generated and the ASMR 3 model was fit to each. For
each model fit, the estimated annual adult abundance and
95% profile likelihood confidence bOlmds were retained. The
estimated brood year recmitment and 95% profile likelihood
confidence bounds were also rctained. Notc that becausc of
the uncertainty in assigning age to even the smallest fish in
the dataset. newly tagged fish now have a possibility of being
age-I. As a result, it was necessary to expand the age range of
the model such that recruitment estimates were for age-l fish.
Estimated adult abundance (age-4+) ranged from 9,322
(95% CI 8,867-9,799) in 1989 to 6,017 (95% CI 5,369-6,747)
in 2006 (fig. 26). The coefficient of variation for these

13

estimates ranges from approximately 1%-7%, in contrast
to 0.5%-3% if uncertainty in assignment of age is ignored
(figs. 13 and 27). The trend in recruitment considering the
new growth function and ageing error contains much greater
uncertainty than when ageing error is ignored (figs. 12 and
28). Although the point estimates from the two models are in
agreement that recmitment has been increasing since the mid-
1990s, the uncertainty in the recruitment estimates from the
latter assessment makes statements about differences among
years tenuous.

Discussion

2006 Humpback Chub Assessment Update with
Refinements

The overall result of the mark-recapture-based open
population model assessment is that the adult portion of the
LCR HBC population appears to lJave increased in abundance
since 200 I. The assessment model best supported by the data
is ASMR 3 with ageing elTor. This model produces a 2006
adult abundance estimate of approximately 6,000 fish. In addi-
tion, this analysis suggests that there has been an increase of
approximately 20%-25% in adult abundance since 2001. This
increase is likely related to an increasing recruitment trend
beginning perhaps as early as 1996, but likely no later than
1999. Recruitment of juvenile HBC since 2000 appears stable,
but the precision of these estimates is low when ageing error is
included in the assessment.
The LCR hoop-net abundance index suggests a modest
increase in the abundance of juvenile fish and stability in the
abundance of adult fish. The LCR inflow reach trammel-net
abundance index indicates stability with a slight indication of
increased abundance in 2005 and 2006. Although confidence
in the mark-recapture-based open population model results
might be higher if the catch-rate metrics indicated similar
trends, it is not surprising that the catch-rate metrics are not
able to detect a 25% increase in abundance. The basic, and
frequently violated, assumption that must be made when
evaluating a catch-rate time series is that capture probability
must remain constant for the metric to be well correlated with
abundance (MacKenzie and KendalL 2002). There is good
reason to suspect that this assumption is violated for the index
data series presented in this update because of the influence
of abiotic factors on catchability (Arreguin-Sanchez, 1996).
As an example, a likely significant driver of catch ability in
the LCR is turbidity (Dennis Stone, U.S. Fish and Wildlife
Service, oral commun., 2007; U.S. Fish and Wildlife Service,
unpub. data, 2007), and turbidity varies greatly in the main-
stem Colorado River and the LCR as a function of tributary
freshets and dam operations.
A more significant concern is the lack of correlation
between ASMR 3 results and the closed population model