7786

• UCREFRP Catalog Number: 7786
• Author: Emlen, J. M., T. A. Strekal and C. C. Buchanan
• Title: Probabalistic Projections for Recovery of the Endangered Cui-ui
• USFW Year: 1993
• USFW - Doc Type: North American Journal of Fisheries Management
• Copyright Material: YES

470

EM LEN ET AL.

that the chance of extinction under this particular
regime was 20%.
Using a computer algorithm developed by Lane
and Frevert (1985), we generated a multivariate
conditional density function based on an 89-year
(1901-1989) water runoff data set covering 1510-
cations in the Truckee watershed. The variance-
covariance structure among locations was pre-
served as well as 1- and 2-year serial correlations.
In the absence of reasons to believe otherwise, we
assumed that the temporal and spatial structure
of the runoff for the 89 years would continue to
characterize the Truckee system for the foreseea-
ble future. Accordingly, a representative sampling
of possible future scenarios was constructed by
using this multivariate function to generate 200
"stochastic traces" (equally likely sequences) de-
scribing flow at several key locations.
A hydrologic accounting model of the Truckee
River (Buchanan and Strekal 1988) was used to
convert runoff data from these 15 key locations in
the river basin into lower Truckee River flow and
Pyramid Lake level. Eleven parameters derived
from these data were input to the model for every
month of every year of record. Releases from
Stampede Reservoir, reflecting U.S. Army Corps
of Engineers flood control regulations and legal
mandates for threatened and endangered fish spe-
cies, were incorporated into this model, as well as
diversions for agriculture. Reservoir storage vol-
ume limits, seasonal distribution of irrigation de-
mand, and reservoir evaporation were assumed
constant. Municipal and industrial demand was
set at a projected future level.
To project cui-ui numbers into the future we
made use of an existing single-species, age-projec-
tion (Leslie) matrix (cui-ui model); reproduction
rates were written as functions of hydrological in-
put and affected by density feedback. Reproduc-
tion rates were characterized, on an age-specific
basis, as the product of a female's likelihood of
joining the prespawning aggregate (related to at-
traction flow from the Truckee River), the likeli-
hood of her passage into the river (dependent on
lake level), and the survival of her eggs (a function
of river flow). Specific relations are provided in
Buchanan and Strekal (1988), amended as de-
scribed in the Appendix.
To simulate the effects of supplemental water
on future population dynamics, a total supple-
mental water budget was decided upon, and in-
crements of water were added to the zero-supple-
ment water levels, month by month, in amounts
proportional to the historic mean monthly flows.

Simulations were initialized to 1990 conditions by
using a population density structure based on es-
timated existing adult females and back-calculat-
ing larvae numbers from these estimates along with
the presumed mortality rates. For a summary of
demographic parameter values see Buchanan and
Strekal (1988) and the Appendix.
To identify the amount of supplemental water
needed for recovery, two sets of simulations were
run. In the first, 200 stochastic traces were used
to drive the cui-ui model; lower (0.001), upper
(0.003), and intermediate "best guess" (0.002) es-
timates of early survival (see Appendix) were em-
ployed. Supplemental water was added into the
system beginning in the first year of simulation
(1991). The second set of simulations was designed
to illustrate the consequences of procrastination
in recovery efforts. These runs differed from those
of the first set in that, in keeping with political and
economic realities, supplemental water was not
added until 1994, and then was increased in in-
crements, in equal steps of 3,000 or I 0,000 ac~e-
feet, until the full yearly allotment was reached.
Because each year of less than the estimated op-
timal supplement generally results in population
decline, the more realistic incremental schemes
are likely to require higher final supplemental wa-
ter flows than those in which water becomes avail-
able immediately. Only the best guess estimate
(0.002) of early mortality was used in these sim-
ulations.

Results

Results from the first set of runs are shown in
Figure I. The uncertainty in the value of larval
survival is reflected in the area between the upper
(survival = 0.003) and lower (survival = 0.001)
curves. When we examined these curves and used
a decision maker's prerogative to set the satisfac-
tory level of certainty of persistence at 95%, we
found that between 45,000 and more than 120,000
acre-feet of water will be required before the pop-
ulation can be said to be capable of recovery (95%
of equally likely future scenarios lead to survival
for cui-ui over 200 years or more). The gulf be-
tween these estimates of required water represents
the extent of our uncertainty (vis-a-vis model pre-
dictions). This region is currently very wide, but
it can be narrowed with improved data on de-
mographic parameter values, thus resolving po-
tential conflict between cui-ui advocates and op-
posing political and economic interests. Research
continues, and as new, improved estimates be-
come available, Figure 1 will continue to be re-