9598

• UCREFRP Catalog Number: 9598
• Author: Lubinski, K., J. Wiener and N. Bhowmik.
• Title: Regulated Rivers Research and Management.
• USFW Year: 1995.
• USFW - Doc Type: \
• Copyright Material: NO

FISH MONITORING 269
level of the individual fish. Twenty runs were sampled with replacement until a sample of 10 runs was
selected. The RFAI scores were calculated for 1000 of these samples and the sampling error was estimated
from the distribution of RFAI scores.
RESULTS
Variation in reservoir rankings, 1989-1992
Although some mainstem reservoirs were consistently ranked among the three sample years, the scores
for most reservoirs fluctuated substantially. We limit the discussion to the general statistical properties of the
data set rather than specifics of individual sites. Annual rankings of two metrics (PTOL, TLIT) were
consistent among years in all three reservoir zones (Table II). Most other metrics ranked reservoirs consis-
tently across years in at least one reservoir zone; however, neither FHAI nor TAXA ranked reservoirs
consistently in any of the three zones. Ranks of overall reservoir index scores differed among years
(Table II).
This overall instability in rankings raises questions about those reservoirs and regions that did rank con-
sistently among years. On closer examination we find that reservoirs with consistent RFAI scores exhibit
considerable variation in metric scores. That is, apparent RFAI stability conceals changes in metrics scores
that cancelled each other out, a situation that is rarely encountered in the behaviour of stream IBI scores for
fish or insects (J. R. Karr, personal observation).
Bootstrap estimation of variance of RFAI scores
Four major sources of variability influence RFAI scores: sampling error, measurement error and temporal
and spatial variability in resident fish assemblages resulting from natural events or anthropogenic influences.
Bootstrap sampling provides an estimate of RFAI's measurement error, or precision. Patterns within and
among reservoirs are not strong nor consistent (Table III). The average measurement error for a particular
reservoir is not consistently high or low across inflow, transition and forebay; neither are reservoir regions
consistently higher or lower across reservoirs. We conclude that measurement error is independent of
location and reservoir.
1992 supplemental data set
The RFAI scores calculated for samples taken during 1992 failed to consistently rank the four reservoirs
from September to December (Figure 2). Differences in scores were not associated with any obvious seasonal
pattern; the direction of monthly changes in scores differed among reservoirs. The rank orders of the four
0
0
0
LO
m
`o
rn
0
v
Q
u_
a:
co
0
N
.,
. .........................
i
?? - ?pnt
Ick
Sep Oct Nov Dec
Figure 2. Reservoir fish assemblage index (RFAI) scores for monthly supplemental collections from Chickamauga, Kentucky, Nicka-
jack and Watts Bar reservoirs for September to December 1992. The RFAI scores were calculated with 10 runs closest to the dam
(upstream). All samples for a reservoir used the same fish health assessment index (FHAI) score