9727

• UCREFRP Catalog Number: 9727
• Author: Marsh, P.C., B.R. Kesner and C.A. Pacey.
• Title: Repatriation as a management strategy to conserve a critically imperilled fish species.
• USFW Year: 2005.
• USFW - Doc Type: North American Journal of Fisheries Management
• Copyright Material: YES

d during annual March
release year includes
002 Combined
4 30
4 23
3 33
15 115
21 133
4 37
20 89
22 68
17 45
1 1
0
III 574
in objective of the
termine the number
ach adult size after
d utilizing an alter-
fish (i.e., per unique
tse and at just one
Liltiple recaptures of
not result in addi-
Actual age at re-
)n mixing at rearing
age at release into
the number of years
it VBG curve is
0-01,
(asymptotic length
wth coefficient de-
iptotic length is ap-
ge at length 0) were
ne in Microsoft Ex-
A-squared residuals
le growth rates and
Iinckley 1983; Mc-
A thus, gender-spe-
ivenile or unknown
cker database con-
mpture or recapture
lril 2002. After the
)er recorded twice)
57,778 release re-
(Table 1). Also, 9
REPATRIATION TO CONSERVE IMPERILED FISH
551
TABLE 2.-Annual, single-census population estimates (N*, Chapman modification of the Peterson method; Seber
1973) and 95% confidence interval (CI) of repatriated razorback suckers from collection data acquired during annual
March censuses in Lake Mohave, Arizona-Nevada, 1999-2002. Abbreviations are as follows: M = number of repatriated
fish captured the first year, C = number of repatriated fish captured the following year, and R = total number of
repatriated fish captured in both the first and following year. Also shown are poststocking survival estimates for repa-
triated razorback suckers based on N* and the number of fish released.
Number of Survival
Year N* 95% Cl M C R fish releaseda (%)
2002 1,335 449-4,804 105 62b 4 57,778 2
2001 1,813 835-4,113 153 105 8 46,999 4
2000 2,494 845-9,033 80 153 4 39,709 6
1999 1,017 418-2,700 87 80 6 23,479 4
a Each number is a cumulative number of fish released based on data from repatriates released
before March 1 of the population estimate (N*) year (e.g., in 2002, the number 57,778 represents
fish released from 1992 to February 28, 2002).
b The 2003 data were reported by definition of C
captures were removed because no corresponding
release records were found, leaving 813 capture or
recaptures-specifically, 132 recaptures and 681
captures (i.e., unique records of fish capture).
Thus, nearly 99% (N = 57,097) of the released
fish had unknown fates.
Analyses of population abundance and survival
estimates used only capture and recapture records
from March censuses (N = 579). For all analyses,
we removed 25 records of recapture from the same
census and 5 capture records without release data.
The remaining 549 records represented 497 unique
captures and 52 recaptures. Two capture records
could not be used in Program MARK because re-
lease and capture year were the same.
Annual single-census population estimates were
calculated for only 4 years (1999-2002; Table 2)
because the probability of a biased population es-
timate significantly increases as the number of re-
captures decreases. In our study, only one of the
four estimates was based on more than seven re-
captures whereas two estimates were based on four
(a recommended minimum; Ricker 1975). Seber
(1973) concluded there is a 95% likelihood that
bias is negligible on estimates based on at least
seven recaptures. The 2002 estimate of 1,335 fish
is approximately 53% less than the estimate gen-
erated in 2000, both estimates based on only four
recaptures (Table 2).
Fish that met the size-at-release target of 301-
350 mm TL represented only 19-25% of the total
number of repatriates released, whereas 71-77%
of the total was 300 mm or less (Table 3). We
found that size-class significantly (log-linear anal-
ysis: AX2 = 76.35, df = 8, P < 0.01) interacted
with capture status, demonstrating the effect of
size-class on survivorship, assuming that capture
rates for all size-classes would be similar 4 years
after release (Table 4). In addition, a significant
interaction between census year and capture status
(log-linear analysis: AX2 = 12.63, df = 9, P <
0.01) indicated a lack of agreement between years,
and a significant interaction between size-class and
year (log-linear analysis: 0X2 = 210.92, df = 12,
P « 0.01) indicated that the three size-classes
were not released in equal proportions over the
study period.
Program MARK estimated first-year survivor-
ship of a 350-mm fish to be 2.5 times greater than
for a 300-mm fish (Figure 1). Low recapture and
survival rates resulted in some parameters being
impossible to estimate for models with as few as
nine parameters. However, all models that includ-
ed length at release as a covariate had a better fit
than those without length (Table 4). In addition,
among various models that incorporated length at
release as a covariate, parameter estimates of first-
year survival were similar, which produced a ro-
bust parameter estimate that linked size at release
and first-year survivorship.
The VBG curve analysis included all of the
unique capture data (N = 681) minus fish with
missing length data or unspecified gender, for a
total of 545 fish. Average total length at release
was 287 mm (SD = 46, range 96-555) for all
repatriated fish (N = 57,689) and 320 mm (SD =
51, range = 169-563) for captured repatriates. Of
the captured fish, females at release average 321
mm (SD = 84, range = 169-500) and males were
313 mm (SD = 49, range = 175-563), a 3% size
difference, but their average size at capture dif-
fered by 13%, females averaging 557 rum (SD =
56, range = 356-670) and males 487 mm (SD =
77, range = 298-672). Overall growth generally
was rapid for both females and males after initial
release, and growth was asymptotic within ap-