9484

• UCREFRP Catalog Number: 9484
• Author: Osmundson, D. B., V. L. Ryel, V. L. Lamarra and J. Pitlick
• Title: Flow-Sediment-Biota Relations
• USFW Year: 2002
• USFW - Doc Type: Implications for River Regulation Effects on Native Fish Abundance
• Copyright Material: YES

December 2002 FLOW-SEDIMENT EFFECTS ON RIVERINE FISH
proportion of captured individuals ?550 mm TL, and
correction factors based on stratum-specific catch-per-
effort values and kilometers per stratum. Because of
the rarity of this species, all available data (1991-1994
and 1998-1999) were used to characterize population
distribution including years outside of the 1994-1995
study period.
Comparisons of fish body condition among strata
also were used as a means to assess food availability.
Only the three dominant large-bodied native species
were analyzed: Catostomus discobolus Cope (bluehead
sucker), C. latipinnis Baird and Girard (flannelmouth
sucker), and Gila robusta Baird and -Gerard (roundtail
chub). Relative condition (K) is the observed mass of
a given fish divided by the expected mass for a fish of
its length (Le Cren 1951). The expected mass is cal-
culated using constants (slope and intercept coeffi-
cients) derived from regressing log-transformed mass
as a function of length (see Osmundson et al. 1998).
We used length and mass of all captured individuals
from the last three sample periods to develop one
length-mass relationship for each species that could
then be used as a standard for among-strata compari-
sons.
Statistical procedures
Various statistical procedures were used including
analysis of variance (ANOVA), analysis of covariance
(ANCOVA), Pearson correlation, multiple regression,
and principal component analysis (PCA). All statistical
analyses were performed using NCSS (2000). In the
results, the test used for each analysis is stated along
with the pertinent statistics and P values. For statistical
analyses, log transformations (ln[x + 1]) were used to
increase- normality in biological parameters; all anal-
yses of chlorophyll a, macroinvertebrate biomass, and
fish numbers and biomass are from In-transformed val-
ues. For ANOVA tests, the Tukey-Kramer multiple
comparison test was used for post hoc testing among
treatments (P < 0.05).
PCA was used to explore relationships among phys-
ical attributes of the riverbed. The PCA attempts to
maintain the information of several variables with few-
er independent variables (often one to three). Values
(scores) of this reduced'set of variables are calculated
from transformed original data using loading coeffi-
cients (maximum absolute value = 1.0). Ranges in the
scores of the resulting variables are useful in charac-
terizing the original data set and relating to other var-
iables. We used PCA to characterize sites relative to
bed-sediment characteristics. The scores from the first
principal component from seven physical variables
were regressed against chlorophyll a and macroinver-
tebrate biomass to assess relationships between the
physical environment and the lowest trophic levels. The
seven physical variables were: fraction of the substrate
<2 mm, median particle size of the surface layer, vol-
ume of interstitial void space, fraction of the surface
1725
area consisting of fines, total DFS, midcolumn water
velocity, and CPOM dry mass.
RESULTS
Longitudinal variation
Physical characteristics.-River gradient increased
almost exponentially in an upstream direction, varying
from 0.028% in stratum 1 to 0.196% in stratum 11 (Fig.
1, lower). The overall mean gradient of 0.100% was
equaled or exceeded in strata 7-11 and stratum 4. Bank-
full depth of the river channel increased systematically
downstream (Fig. 2a), whereas bankfull width varied
greatly among strata but with no longitudinal trend.
Mean bankfull depth doubled between strata 11 and 5
and reached a maximum in canyon-bound stratum 3;
it then decreased through strata 2 and 1. Actual water
depth at base flow was not measured. Habitat mapping
indicated a steady, nearly linear, downstream decrease
in surface area of riffles, varying from 12.9% of total
surface area in stratum 11 to <0.1 % in stratum 1 (Fig.
2b).
Estimates of discharges necessary to initiate motion
of coarse bed materials (characterized by sporadic mo-
tion of a few particles somewhere on the bed) at 50%
of the cross sections increased with distance down-
stream, but were similar among strata bounded by the
same major tributaries (Table 1; Fig. 1); in strata 2-
11, these discharges averaged 40% of the bankfull dis-
charge (Pitlick and Cress 2000). Discharges necessary
to cause widespread bed mobilization at 50% of the
cross sections also increased downstream (-Table 1); in
strata 6-11, these discharges were similar among strata
bounded by the same major tributaries and correspond-
ed to the bankfull discharge. Downstream of the Do-
lores River inflow (strata 1-5), some very high dis-
charges are required to reach bankfull owing to large
increases in either channel width or depth, and thresh-
olds for widespread bed mobilization were assumed to
occur at discharges less than bankfull.
Other physical habitat characteristics were estimated
separately for riffle and run habitats. However, char-
acterization of riffles was precluded in stratum 1 be-
cause of the near absence of this habitat type. For the
whole study area, the mean Dso (median grain size) of
riffle substrates was greater than that of run substrates
(ANOVA, F,,gg = 21.12, P = 0.00001). Also, there was
a fairly continuous and significant (ANCOVA, F,,,, =
52.19, P < 0.00001) downstream decrease in the Dso
of runs, though not of riffles (ANCOVA, F1,72 = 0.96,
P = 0.33). In riffles, the DSO increased in the three
strata immediately downstream of stratum 6 (Fig. 2c).
In contrast, when fine sediments were excluded from
the analysis and only coarse substrate particles (>_2 mm
diameter) were considered (Fig. 2d), mean grain sizes
of riffle and run substrates were not significantly dif-
ferent (ANOVA, F,,,, = 0.50, P = 0.48), and mean
grain sizes in runs did not decrease with distance down-