Content uploaded by Kevin Bestgen
Author content
All content in this area was uploaded by Kevin Bestgen on Dec 15, 2015
Content may be subject to copyright.
Swimming Performance and Fishway Model Passage Success of
Rio Grande Silvery Minnow
KEVIN R. BESTGEN*
Larval Fish Laboratory, Department of Fish, Wildlife, and Conservation Biology,
Colorado State University, Fort Collins, Colorado 80523, USA
BRENT MEFFORD
Water Resources Research Laboratory, U.S. Bureau of Reclamation,
Denver Federal Center, Denver, Colorado 80225-0007, USA
JAY M. BUNDY,CAMERON D. WALFORD, AND ROBERT I. COMPTON
Larval Fish Laboratory, Department of Fish, Wildlife, and Conservation Biology,
Colorado State University, Fort Collins, Colorado 80523, USA
Abstract.—We used a swim chamber, flume, and large-scale fishway models to assess the swimming
performance, behavior, and passage success of endangered Rio Grande silvery minnow Hybognathus amarus.
Field-captured silvery minnow (53–88 mm total length) swam 114–118 cm/s (i.e., up to 20.9 body lengths/s)
in a swim chamber in water temperatures of 15, 19, and 238C. The relationship between time to fatigue and
water velocity showed that endurance declined sharply at velocities above 60 cm/s, a threshold that is
consistent with critical swimming speed estimates and may represent a transition from aerobic to anaerobic
metabolism. Water temperature and fish length were positively correlated with swimming performance. At
water velocities of 60 cm/s and less, silvery minnow routinely swam the equivalent of 50 km (125 km
maximum) in a swim chamber in less than 72 h. The proportions of silvery minnow that successfully ascended
a flume over sand, gravel, or cobble substrate declined as water velocity increased to 53 cm/s. Passage times
increased at higher water velocities and at a faster rate over sand substrate because fish were stationary for
longer periods over the lower-velocity boundary layers created by gravel and cobble at all velocities. Dual-
vertical-slot fishway passage was 52% in a less turbulent flow of 78 cm/s; passage was 8% at a faster, more
turbulent flow of 87 cm/s. Conversely, 75% of silvery minnow ascended a rock channel fishway with a
holding pool present and a 1% hydraulic gradient in low (58-cm/s) and high (83-cm/s) mean flow velocities.
Differences in willingness to swim, longer test duration, and the mosaic of water velocities created by the bed
roughness elements may explain the higher silvery minnow passage success in the rock channel. Predictive
swimming fatigue relationships, together with fish length and water temperature, may guide decisions
regarding fishway lengths and velocities so as to permit passage of Rio Grande silvery minnow.
Locomotion and passage capability of fishes is
thought to be largely a function of body size, but
swimming performance and passage studies of rela-
tively small-bodied fishes have been, until recently,
underrepresented in the literature (Beamish 1978 ;
Nelson et al. 2003; Ward and Hilwig 2004; Leavy
and Bonner 2009). Swimming performance of com-
mercially important or relatively large-bodied species,
such as salmonids, has been reasonably well docu-
mented because of negative effects of dams on passage.
Only recently has it been recognized that low-head
diversion dams may negatively affect distribution and
abundance of small-bodied fishes, particularly those
that require relatively long and uninterrupted reaches to
complete their life history in arid-land streams of
western North America (Cross et al. 1985; Bestgen and
Platania 19 91; Wi nston et al. 1991; Dudl ey and
Platania 2007). Unobstructed dispersal routes are
particularly important for several Great Plains stream
fishes that produce semibuoyant eggs and larvae that
drift long distances downstream (Moore 1944; Cross et
al. 1985; Fausch and Bestgen 1997; Platania and
Altenbach 1998). Downstream dispersal of early life
stages requires juveniles or adults to r ecolonize
upstream reaches, but swimming performance and
passage capability of those fishes in streams modified
by diversion dams are poorly understood.
Upstream dispersal may be an important life history
attribute for the Rio Grande silvery minnow Hybog-
nathus amarus , an endangered species endemic to (and
formerly widespread and abundant in) the warmwater
main-stem reaches of the Rio Grande and its larger
tributaries. This minnow now occurs only in the middle
* Corresponding author: kbestgen@colostate.edu
Received May 7, 2009; accepted October 20, 2009
Published online January 14, 2010
433
Transactions of the American Fisheries Society 139:433–448, 2010
Ó Copyright by the American Fisheries Society 2010
DOI: 10.1577/T09-085.1
[Article]
Rio Grande, New Mexico, an area representing less
than 10% of its former range (Bestgen et al. 1989;
Bestgen and Platania 1990, 1991; Cook et al. 1992;
Bestgen and Propst 1996). Because of reductions in
historical range and threats to remaining populations,
the species is presently listed as endangered by the U.S.
Fish and Wildlife Service (U.S. Department of the
Interior 1994). Factors implicated in the demise of Rio
Grande silvery minnow include habitat modifications,
hybridization, altered flow regimes, negative effects of
nonindi genous fishes, and disruption of dispersal
routes by diversion dams, including several in the
middle Rio Grande (Bestgen et al. 1989; Bestgen and
Platania 1990; Platania 1991; Luttrell et al. 1999;
Dudley and Platania 2007). Reestablishment of up-
stream dispersal pathways for Rio Grande silvery
minnow and other plains stream fishes may require
construction of fishways for passage over low-head
diversion dams. Traditional fishways are engineered
primarily for relatively large-bodied fishes that are
powerful swimmers. However, few fishways have been
designed for relatively small-bodied (10-cm total
length [TL] or less) species whose swimming perfor-
mance is poorly understood.
We assessed swimming performance and behavior
of Rio Grande silvery minnow in laboratory experi-
ments. First, we measured the critical swimming speed
and swimming endurance (time to fatigue) of silvery
minnow in a swimming chamber at water velocities up
to 118 cm/s in water temperatures of 15, 19, and 238C.
Second, we measured their long-distance swimming
ability at moderate water velocities and the same water
temperatures. Third, we observed their behavior and
estimated the proportion and passage rate for silvery
minnow that swam the length of a flume with differing
water velocities and substrate particle sizes. Finally, we
determined the proportion and passage rate of silvery
minnow introduced into large-scale laboratory models
of dual-vertical-slot and rock channel fishways. We
then used time-to-fatigue relationships to evaluate
tradeoffs between silvery minnow swimming endur-
ance and different combinations of fishway channel
length and water velocity. These swimming perfor-
mance experiments, perhaps the most comprehensive
available for a small-bodied fish, may assist managers
tasked with design of fishways capable of passing Rio
Grande silvery minnow and other small-bodied fishes
in rivers.
Methods
Fish culture and handling .—Age-1 or older Rio
Grande silvery minnow adults (mean TL ¼ 71 mm,
range ¼ 53–88 mm; mean weight ¼ 2.65 g, range ¼
1.16–5.40 g) were obtained by seining (4.6 3 1.8 m;
4.7 mm mesh) in the main-stem Rio Grande in central
New Mexico and were collected in all seasons except
summer (to avoid heat stress) in 2001 and 2002. These
field-captured fish were preferred over cultured stocks
because of potential differences in swimming ability
(Ward and Hilwig 2004). Hatchery-cultured fish (80–
105 m m TL) we re used to fi nish long-dista nce
swimming tests at 158C(n¼ 8) and to complete
testing in dual vertical-slot fishways (five low-velocity
and four high-velocity tests; N ¼ 10 fish/test) and in
rock channel fishways (eight tests, N ¼ 10 fish/test).
Fish were held and velocity chamber experiments
conducted at the Aquatic Research Laboratory, Colo-
rado State University, Fort Collins. All other testing
was conduct ed at the Water Resources Research
Laboratory, U. S. Bureau of Reclamation, Denver,
Colorado. The fish were transported in insulated live
wells (76 L) supplied with aerated water and allowed to
acclimate to test conditions for 2 h. Batches of fish
were held long-term in 900-L circular tanks containing
aerated well water at 198C. Water velocity was
maintained at 10 cm/s to exercise fish and photoperiod
was 14 h light and 10 h dark. Fish were fed frozen
brine shrimp and dry flake food several times each day.
Pretest stress reduction was effected by reducing visual
disturbance, minimizing handling before tests and
during tank cleaning, frequent feedings (3–5 times/d),
high water exchange rates (2 times/h), and maintaining
moderate densities (less than 100 fish per 900-L tank)
of fish in circular tanks (not troughs) that were allowed
to accumulate biofilms. Healthy fish (minimal bruising
and fin erosion, good body condition) were tempera-
ture-acclimated for 4 d before testing and fed as
described above; food was withheld for 24 h before
tests (Beamish 1978).
Swim chamber description and testing.—Illustra-
tions of the swim chamber and detailed functional
descriptions and images of fishway types are available
in Bestgen et al. (2003). The swim chamber was
constructed with 7.6-cm (inside diameter), schedule 40,
clear polyvinyl chloride (PVC) pipe. Pipe diameter was
sized to meet the recommendation that fish body area
be less than 10% of swim chamber area (45.6 cm
2
)to
minimize blocking effects (Bell and Terhune 1970;
Smit et al. 1971). This criterion was met because the
mean cross-sectional area of the Rio Grande silvery
minnow of average length (71 mm TL) used in all
velocity chamber studies was 3.0 cm
2
(6.6% of the
chamber area) and the range of the cross-sectional area
was 2.16–4.55 cm
2
(4.7–10% of the chamber area).
Clear PVC was used to enable observations of fish
position and swimming behavior. The open-ended
upper section (1.25-m long) received inflow tubes from
each of three water pumps that were used singly or in
434 BESTGEN ET AL.
combination, and pump speed was controlled with a
rheostat. Pumps were submerged with the swim
chamber in a 3-m-long trough filled with well water
and delivered flow into the upstream open end of the
swim chamber. This created a venturi effect and
achieved higher velocities with less turbulence than if
pumps had been linked to the chamber in a closed
system. The downstream 1.25-m-long section, which
served as the swimming chamber, was confined with
plastic-coated large-mesh screens on each end. Dye and
neutrally buoyant gelatinous beads flowed with
minimal vertical dispersion when pumped through the
system, suggesting flow turbule nce in the swim
chamber was minimal and an even velocity distribution
was achieved. A flowmeter and passage rate of five
beads through the swim chamber were used to calibrate
estimates of water velocity before testing; each fish was
introduced at the downstream end of the swimming
chamber. A 30-cm section of the middle of the
swimming chamber was enclosed with black plastic
that provided cover and was attractive to fish; this
prevented most individuals from moving to the
upstream screen, where velocity may have been
slightly reduced. When fish did move to the upstream
screen, they were encouraged to retreat to the enclosed
area by flashing a light or using short bursts of higher
velocity water. A small, diffuse light source was
provided when fish swam during nighttime hours to aid
orientation. Swim chamber water temperature was
regulated by exchanging water in the trough with that
from an aerated and temperature-controlled external
reservoir.
Three different types of swimming chamber perfor-
mance tests were completed with Rio Grande silvery
minnow (Table 1). Critical swimming speed experi-
ments (Brett 1964) used randomly selected fish (mean
TL ¼ 65.6 mm, SD ¼6.0, range ¼ 53–75 mm) that were
allowed either a relatively long (mean 14.3 h, SD ¼
4.58; N ¼ 6) or short (mean ¼ 2.2 h, SD ¼ 0.75, N ¼ 6)
acclimation period in the swim chamber. To orient fish,
velocity was set at 10 cm/s (similar to that in the
holding tanks), or 1.4 body lengths (BLs) per second,
based on the average length fish we used, and aerated
water was set at 198C. The actual measured mean
velocity value was 9.41 cm/s (SD, 0.69), and the mean
temperature was 18.98C (SD, 0.31; N ¼ 12). After
acclimation, water velocity was increased by 10 cm/s
every 20 min until the fish was swept to the
downstream screen. Failure time was noted, and fish
were weighed (g), measured (TL; mm), and placed in a
recovery tank. Fish were not used more than once in
these tes ts. Critical s wimming speed ( U
crit
)was
determined following Brett (1964), that is,
U
crit
¼ u
i
þðt
i
=t
ii
Þðu
ii
Þ; ð1Þ
where u
i
is the highest velocity (cm/s) maintained for
the last full 20-min period, u
ii
is the 10-cm/s velocity
increment, t
i
is the time (min) that the fish swam before
it fatigued in the final interval, and t
ii
is the duration of
the interval (20 min).
We then estimated Rio Grande silvery minnow
(mean TL ¼ 70.7 mm, SD ¼ 6.8, range ¼ 55–88 mm)
swimming speed at a range of velocities and water
temperatures to describe endurance (time to fatigue) as
a function of water velocity (Table 1). We conducted
both short-term and longer-term swim chamber tests at
158C (mean of 36 measurements before every fish was
tested ¼ 14.98C, SD ¼ 0.47), 198C (mean of 59
measurements ¼ 18.38C, SD ¼ 0.30), and 238C (mean
of 51 measurements ¼ 23.08C, SD ¼ 0.16) to estimate
temperature effects on swimming performance; only
nominal temperatures were reported or used in analyses
because they were similar to measured values. This
range of temperatures approximates that of the Rio
Grande when fish may be dispersing, and was
representative for conditions in spring and autumn
and portions of the summer and winter (KRB, personal
observation). The range of flow velocities tested also
mimicked those that may be encountered by Rio
Grande silvery minnow in a natural setting, including
relatively high velocities present during spring runoff
or in riffle habitat.
Comparison of U
crit
data where fish had short (mean
velocity ¼ 50.2 cm/s, SE ¼ 4.81, N ¼ 6) or long (mean
TABLE 1.—Test conditions and data gathered for swimming performance experiments with Rio Grande silvery minnow. Wild
fish (W) were used in most experiments; a few hatchery fish (H) were used to finish swimming distance tests at water
temperatures of 158C as well as in the high-velocity tests of the rock channel fishway after a downstream holding pool was
installed.
Experiment Equipment Fish origin Variable measured Velocity (cm/s) Temperature, substrate
U
crit
Swim chamber W Critical swim speed 36–69 198C
Endurance swimming Swim chamber W Fatigue time 30–118 15, 19, 238C
Distance swimming Swim chamber W, H Distance 30–57 15, 19, 238C
Flume Adjustable flume W Percent passage, passage time 8, 23, 38, 53 16–188C; sand, gravel, cobble
Fishways Dual vertical slot W, H Percent passage, passage time 78, 87 16–188C
Rock channe l W, H Percent passage, passage time 58, 83 16–188C; cobble, holding pool or not
RIO GRANDE SILVERY MINNOW PASSAGE 435
velocity ¼ 52.8 cm/s, SE ¼ 5.28, N ¼ 6) acclimation
times suggested no difference in performance (t-test; P
¼ 0.73), so all fish were allowed a 1-h acclimation
period. We chose test velocities within the interval 30–
118 cm/s; we assumed that fish would be able to swim
indefinitely at velocities less than 30 cm/s. We tested
45 fish at a water temperature of 158C, 52 at 198C, and
45 at 238C. After the acclimation period, during which
fish were oriented to the swimming chamber at 10-cm/s
water velocity (measured mean value ¼ 10.04 cm/s, SD
¼ 0.68), we increased the water velocity to the desired
test speed over as much as 2 min, the adjustment time
increasing in proportion to the test velocity. These
initial tests were terminated at a maximum time of 220
min (Beamish 1966; Brett 1967).
Finally, exploratory tests were conducted to deter-
mine the upper limit of endurance and swimming
distance for Rio Grande silvery minnow (fish were of
the same length as in other tests). Our initial tests of up
to 220 min at 198C suggested that fish could probably
swim long durations at water velocities of 35–50 cm/s.
Therefore, we progressively increased the duration of
testing for individuals from a minimum of 245 min up
to 4,320 min (72 h); tests were terminated at a preset
time within that range or when the fish fatigued,
whichever came first. Fish were tested in water
temperatures of 158C(N ¼ 15 fish ¼ 7 field captured,
8 hatchery), 198C(N ¼ 10 fish; field captured), and
238C(N ¼ 14; field captured). Mean water velocities
used at 158C were 43 cm/s (SD ¼ 0.05) for field-
captured fish and 36 cm/s (SD ¼ 0.02) for hatchery
fish. Mean water velocity was 38 cm/s (SD ¼ 0.07) at
198C and 40 cm/s (SD ¼ 0.08) at 238C. We calculated
swimming distance for each fish by multiplying water
velocity by test duration.
For all swim chamber tests, a fish failed when it
became impinged on the downstream screen and could
not be encouraged to swim by gently probing it with a
glass rod, flashing a light, or subjecting it to short
bursts of high-velocity water. A fish was also
considered to fail when the same stimuli could not
prevent it from occupying the area adjacent to the
upstream screen of the swimming chamber. During
high velocity swimming tests, some fish would
impinge, perhaps due to a fright or behavioral response
(Peake and Farrell 2006), apparently before fatigue was
reached. Water velocity was then reduced, and after the
fish repositioned itself (Smit et al. 1971), the test
velocity was resumed. If the fish failed immediately,
the first failure was deemed genuine, and swimming
duration was recorded. If the fish continued swimming,
the time of the first and second intervals were summed
to obtain an estimate of swimming endurance. Five fish
at 158C, four fish at 198C, and no fish at 238C were
classified as nonswimmers in the velocity chamber
swimming endura nce tests ; those fish were not
considered further.
Flume studies.—Volitional swimming performance
of Rio Grande silvery minnow over different substrate
particle sizes and flow velocities was assessed in a
Plexiglas flume (18.3 m long, 0.91 m wide). Substrate
on the flume floor consisted of coarse-sand (about 1
mm diameter), gravel (25 mm), or cobble (152 mm).
Flume slope was set horizontal (zero gradient), and
water was 30.5 cm deep. Target water velocities of 8,
23, 38, and 53 cm/s were measured at 15 cm above the
bed before each test. Water velocities were also
measured along a vertical transect over each substrate
type to determine velocity distributions (Figure 1). A
total of 15 fish were tested in each substrate ( N ¼
3) and
water velocity (N ¼ 4) combinations (N ¼ 12 3 15 ¼
180 total fish); water temperature was 16–188C.
FIGURE 1.—Vertical water velocity profiles in flumes with
sand, gravel, and cobble substrate flowing at 8, 23, 38, and 53
cm/s during volitional swimming tests with Rio Grande
silvery minnow.
436
BESTGEN ET AL.
The ends of the flume were screened with perforated
steel plate upstream and nylon netting downstream so
that fish could not escape. Fish were held in a
downstream screened enclosure and allowed to orient
to the current for 1–10 min before testing. One or two
fish were released at a time, depending on the number
of observers available, and the upstream progress of
each individual was measured at intervals of 1, 2, 3, 4,
5, 10, and 15 min. Observations through clear Plexiglas
walls were made at a distance so fish were not
disturbed (a behavior characterized by random, high-
velocity swimming). Time to swim the flume length
was recorded if a fish finished. For fish that did not
finish after 15 min, maximum upstream swimming
distance was recorded. We also observed fish swim-
ming behavior, vertical position, and use of substrate
particles during all tests.
Fishway studies.—Four different fishways were
constructed in the laboratory to evaluate Rio Grande
silvery minnow passage performance: (1) prototype
size-baffled (Denil), (2) single vertical slot, (3) dual
vertical slot, and (4) rock channel with boulder weirs
(Clay 1995; see Bestgen et al. 2003 for detailed images
of structures and results not reported here). Water
temperatures in all fishway studies ranged from 16–
188C. Because of low passage success and small
sample sizes, the results for the Denil and single-
vertica l-slot fishways were not robust enough to
warrant reporting here. The dual-vertical-slot fishway,
based on a design of the U.S. Bureau of Reclamation
(unpublished data, B. Mefford), had V-shaped 308
baffles that were placed with the leading edge pointed
into the current. The dual slots, one in each side of the
baffle, produced a less pronounced eddy pattern and
greater downstream flow alignment between baffles,
but this fishway was otherwise similar in functional
characteristics to the single-vertical-slot fishway. Tests
were conducted with a fishway flow of 140 L/s. A
tailwater depth of 67 cm (1.7 % average hydraulic
gradient) produced a low mean slot velocity of 78 cm/s,
and a 55-cm depth (2.1% average hydraulic gradient)
produced a high mean slot velocities of 87 cm/s. Test
duration was 1–2 h; longer tests were not warranted
because preliminary trials showed most fish that
finished did so in a relatively short time.
The rock channel fishway was designed to simulate
a channel that could either bypass a portion of river
flow around a dam or act as an in-river ramp to provide
a low-gradient path over a dam. The experimental rock
channel fishway was 30-m long and straight, had a
trapezoidal cross-section 2.4-m wide at the bottom, and
608 from horizontal side slopes that were 76-cm deep
(Figure 2). The upstream 23-m functional channel had
a gradient of about 2% and a horizontal (0% gradient)
downstream section of reduced flow velocity. The rock
bed lining the channel was angular cobble 15–25 cm in
diameter. Weirs were formed by placing three 60–90-
cm (diameter) boulders set in an upstream-pointing V-
shape, the largest placed in the center of the channel
and smaller ones set downstream about 30 cm at 308 to
the perpendicular. The five boulder weirs were spaced
about 3.1 m apart in the center of the fishway. Fishway
experiments were conducted with a flow of 240 L/s.
Low-velocity conditions had average hydraulic gradi-
ent of about 0.5%, pool depths of 26 cm, and average
water velocity was 58 cm/s (37–75). High-water
velocity conditions had average hydraulic gradient of
1.0%, pool depths of about 22 cm, and average water
velocity between boulders in the weirs was 83 cm/s
(67–103). Water velocity in the rock channel fishway
was adjusted with tailboards at the downstream end. In
low-flow tests, fish were released into the downstream
pool created by the tailboards, which allowed for
orientation. In the first few high-velocity tests (no pool
condition), fish were released directly into the current
because t here was no catchment pool. Lack of
opportunity to orient resulted in immediate impinge-
ment of a few individuals so the tailrace was modified
to create a holding pool for additional high-velocity
tests.
All fishway tests were based on volitional passage of
groups of 10–20 fish released into the tailwater
downstream of the fishway. Owing to the limited
availability of fish, we focused our testing on the most
promising fishways and hydraulic conditions. The
dual-vertical-slot fishway was tested at low (5 tests,
10 fish/test) and high (4 tests, 10 fish/test) water
velocities. Rock channel fishway t ests used four
conditions: (1) low velocity, pool, field-captured fish
(four tests, 15 fish/test), (2) high velocity, no pool,
field-captured fish (four tests: one with 10 fish, two
with 15 fish/test, and one with 20 fish), (3) high
velocity, no pool, hatchery fish (two tests, 10 fish/test),
and (4) high velocity, pool, hatchery fish (six tests, 10
fish/test). Not all fish (usually 0 or 1) were recaptured
from the rock channel fishway after tests. Therefore, all
fish not accounted for from a prior test were added to
the number released to achieve a potential number that
could be captured in the subsequent test. This accounts
for some proportions that were not possible based on
the number of fish released.
The numbers of silvery minnow that passed through
all fishways were recorded, as was swimming passage
time and notes on behavior. We were not able to
immediately remove fish that ascended the fishways
without disturbing the progress of other fish in the
model. In a few instances (most of these in rock
channel fishway), fish that had passed were observed
RIO GRANDE SILVERY MINNOW PASSAGE 437
retreating from the head box back to the fishway.
Therefore, for the dual-vertical-slot fishway we
recorded the total number of finishes, recognizing that
a single fish may have sometimes finished more than
once. Because direct observations of small-bodied
silvery minnow were not possible in the rock channel
fishway, we constantly monitored the number of fish
that ascended the model into the head box and
remained there for the first 2–4 h after fish release.
Passage times were recorded as fish entered the head
box. Thereafter, observations were at more irregular
intervals, usually spanning a 24-h period, but some-
times were for as long as 120 h. Observations were also
made at night (about 2300 hours) the day of release to
determine the extent of nocturnal fish movement. Fish
were first removed from the head box only after the
initial 2–4-h observation period to reduce disturbance.
Fish observed swimming back into the model during
that interval were not counted as finishers. Total
number of fish that finished was calculated as the
number of fish removed plus the maximum number of
fish observed in the head box after those removals.
Thus, the total number of fish estimated to finish the
rock channel length was conservative because fish that
finished but left the head box and never reentered were
not counted.
Data analysis.—Swim chamber data were plotted as
swimming endurance (min) in short-term and long-
term tests as a function of water velocity; the number of
body lengths per second that a fish swam was also
calculated for critical swimming velocity tests. Rela-
tionships for swimming endurance as a function of time
for each temperature were fit with nonlinear regression
after endurance data were log
e
transformed (i.e.,
log
e
[log
e
fminutes swimming endurance þ 1g] þ 1).
The double-log relationship was used to achieve a
better fit to the data which exhibited a strong threshold
for swimming performance as a function of time. We
subtracted a velocity constant of 25 cm/s in the
nonlinear analysis to achieve a better fit for the
intercept. This was justified because fish could swim
for long periods at low water velocities, such that an
intercept may never be achieved; subtracting the
constant had no effect of slope estimates. We also fit
a response surface regression to swimming endur-
ance—again as log
e
(log
e
[min swimming endurance þ
1]) þ 1—as a joint function of water velocity,
temperature, their squared and interaction terms, and
fish total length. We also used velocity as a weight in
nonlinear and response surface regressions to enhance
model fit at higher velocities.
The flume data analyzed were the proportion of
silvery minnow that swam the entire flume length in
the test period and duration of successful swimming
trials. We used analysis of covariance (ANCOVA) to
assess differences in the slopes of the relationships of
the proportion of fish that finished over the range of
velocities tested among sand, gravel, and c obble
substrate types (Proc GENMOD in SAS, logit link).
This allowed us to make inferences about effects of
FIGURE 2.—Plan and cross-sectional views of the 30-m-long rock channel fishway used for volitional passage tests of Rio
Grande silvery minnow. The numbers between individual boulders are the flow velocities (m/s) under high-velocity test
conditions.
438
BESTGEN ET AL.
substrate type on the proportion of fish that finished
flume tests as velocity increased. Each fish was treated
as a binomial response (finished or not). We conducted
a similar analysis of slopes of relationships of times to
finish as a function of water velocity over sand, gravel,
and cobble substrate types (time as a continuous
variable; Proc GLM, SAS). This allowed us to make
inferences about effects of substrate type on finish
times of fish as velocity increased. Each fish in those
analyses was assumed to be an independent observa-
tion because interactions that might affect swimming
progress of individuals, such as schooling behavior or
fright response caused by other fish, were not observed.
Performance data for dual-vertical-slot and rock
channel fishways were the proportion of silvery
minnow that successfully ascended each fishway under
each set of experimental conditions. Mean finish times,
proportion of fish that finished in day versus night,
effects of low and high flows, raceways with and
without a holding pool, and effect of field-captured
versus hatchery fish on those responses were also
reported for the rock channel fishway.
Fishway applications.—Swimming performance cri-
teria for Rio Grande silvery minnow were used to
parameterize an equation from Peake et al. (1997) to
predict fishway water velocities that permitted passage
for a given fish length, water temperature, and fishway
length. The equation is of the form
V
fishway
¼ V
swimming
ðD=E
s
Þ; ð2Þ
where V
fishway
is the maximum allowable fishway
water velocity (cm/s), V
swimming
is the swimming speed
of the fish (cm/s), D is fishway length (cm), and E
s
is
endurance of silvery minnow (s) swimming at V
s
. The
E
s
was determined from the response surface equation
parameters instead of nonlinear regression relationships
because effects of water temperature and fish body
length could be jointly estimated as well. To illustrate
the use of this relationship to assess appropriate
fishway water velocities, we assumed a D of 50 m
and an intermediate water temperature of 198C and
solved for the maximum water velocity that allowed
fish of various lengths to pass before exhaustion
occurred. Similarly, we estimated maximum flow
velocities that allowed an intermediate-sized (7 cm
TL) Rio Grande silvery minnow to ascend a 50-m-long
fishway in water temperatures of 15, 19, and 238C.
We then integrated passage distance, fishway water
velocity, and velocity-dependent fatigue relationships
to determine the maximum water velocity that would
allow passage in a fishway of a given length. We
accomplished this by solving equation (2) for a variety
of fishway lengths (D ¼ 0.5–100 m). At each fishway
length, we derived a maximum swimming and fishway
velocity to allow passage by iteratively refitting the
equation.
Results
Critical Swimming Speed
Mean U
crit
for 12 fish (53–75 mm TL) with short and
long acclimation times was 51.5 cm/s (SE ¼ 3.43,
range ¼ 35.6–69.4 cm/s) or a relative swimming rate of
7.8 BLs/s (5.8–9.7 BLs/s). The regression relationship
that described U
crit
(cm/s) as a function of total length
(mm) for the 12 fish was positive (U
crit
¼41.77 þ
1.42
TL; F ¼ 10.93, P ¼ 0.008, r
2
¼ 0.52), which
suggested faster sustained swimming speeds of larger
than smaller fish.
Endurance Swimming
Nonlinear regression relationships of swim ming
endurance as a function of water velocity suggested
endurance increased as velocity declined from 118 to
about 60 cm/s, reached a threshold at about 50–60 cm/
s, and then increased exponentially as water velocity
declined further (Figure 3). At all water temperatures,
Rio Grande silvery minnow were capable of swimming
for 5–15 s at water velocities of 100–118 cm/s (up to
20.9 BLs/s).
Rio Grande silvery minnow were capable of
swimming longer durations at warmer water tempera-
tures. This general relationship was evidenced by the
smaller slopes and intercepts for nonlinear regression
relationships as water temperature increased. Nonover-
lapping confidence intervals about the slope for the
158C relationship, compared with those for the 198C
and 238C relationships, suggested a statistically
significant difference. Broad overlap of confidence
intervals for the slopes for the 198Cand238C
relationships suggested that a statistically significant
effect was not detectable. B iologically important
effects were noted, particularly at water velocities
between 45 and 100 cm/s. For example, at 50 cm/s
water velocity, nonlinear regression predictions sug-
gested that silvery minnow could swim for approxi-
mately 9 min at 158C, 15 min at 198C, and 22 min at
238C. At a higher water velocity of 80 cm/s, nonlinear
regression predictions suggested that silvery minnow
could swim for approximately 20 s at 158C, 55 s at
198C,and71sat238C. Swimming endurance
predictions for fish at water velocities less than 40
cm/s was high at all temperatures because few fish
failed in those tests.
A response surface regression relationship fitted to
endurance data to predict swimming time under the
range of water temperatures and fish lengths tested
showed that silvery minnow endurance was a function
of significant effects of water velocity, water temper-
RIO GRANDE SILVERY MINNOW PASSAGE 439
ature, and fish total length (Table 2). The statistically
significant V
2
term indicated that swimming endurance
changed in a nonlinear fashion as water velocity
increased; inclusion of a squared term for water
temperature was not warran ted (P ¼ 0.84), and
suggested only a linear increase in endurance as water
temperature increases.
Long-Distance Swimming
Field-captured Rio Grande silvery minnow were
capable of swimming for long durations and distances.
At 158C, 4 of 7 silvery minnow were still swimming at
the end of distance tests of 265–4,375 min, but all 10
fish finished tests of 320–2,883 min at 198C, and all 14
fish finished tests of 313–4,187 min at 238C. The range
of swimming distances for those fish were 7.1–125.4
km at 158C, 4.9–62.7 km at 198C, and 6.0–119 km at
238C.
Seven of eight hatchery Rio Grande silvery minnow
in 158C distance swimming tests failed before 8 h
(246–468); the eighth swam 1,547 min before failing.
The average swimming distance of hatchery fish (9.8
km; range, 2.9–34.2 km) was significantly lower (one-
tailed t-test with unequal variances: percent ¼ 0.05, P ¼
0.026) than that of field-captured fish at 158C (53.6
km; r ange, 7.1–125.4 km); the differences were
probably conservative because four of seven field-
captured fish were still swimming when the trials were
terminated.
Flume Tests
Rio Grande silvery minnow held in the flume
orientation area were attracted to incoming flows and
in less than 60 s, most fish were nudging the upstream
barrier screen with their snouts. Upon release, most fish
moved upstream immediately; 98% attempted up-
stream swimming at test velocities of 8 cm/s, 96% at
23 cm/s, 91% at 38 cm/s, and 84% at 53 cm/s. Released
fish showed no behavioral aversion to swimming over
the various-sized substrate particles including cobble,
which was an uncommon substrate type in the sand-
bedded Rio Grande.
FIGURE 3.—Nonlinear regression relationships between the
swimming endurance of Rio Grande silvery minnow and
water velocity at 15, 19, and 238C. To enhance the fits, the
swimming endurance times were double–log
e
transformed
(see text). The estimated equations as shown are as follows:
158C: y ¼ 4.257
exp(0.0514x) (95% confidence intervals ¼
2.956 to 5.558 and 0.0386 to 0.0641 for the intercept and
slope, respectively; df ¼ 35, r
2
¼ 0.78, P , 0.0001); 198C: y ¼
2.98
exp(0.0324x) (95% confidence intervals ¼ 2.610 to
3.349 and 0.0272 to 0.0376; df ¼ 52, r
2
¼ 0.84, P ,
0.0001); 238C: y ¼ 2.987
exp(0.0299x) (95% confidence
intervals ¼ 2.617 to 3.357 and 0.0250 to 0.0348; df ¼ 45, r
2
¼ 0.86, P , 0.0001).
TABLE 2.—Least-squares general linear model coefficient estimates, significance probabilities, and Type III sums of squares
(SS) for a response surface regression relating Rio Grande silvery minnow swimming endurance to water temperature (8C), water
velocity (cm/s), and fish total length (cm). Model statistics are as follows: model F
5,122
¼ 184.3, R
2
¼ 0.86, P , 0.0001.
Endurance times were transformed as log
e
[log
e
(minimum swimming endurance þ 1)] þ 1.
Variable df Coefficient estimate SE Type III SS P
Intercept 1 3.46 0.4351 ,0.0001
Velocity 1 0.0856 0.0067 763.8 ,0.0001
Velocity
2
1 0.00040 0.00004 383.4 ,0.0001
Temperature 1 0.0303 0.0080 68.3 0.0002
Length 1 0.0883 0.0362 27.9 0.0163
440 BESTGEN ET AL.
In general, the proportion of silvery minnow that
successfully ascended the flume declined as water
velocity increased, but those relationships varied
among substrate types (Figure 4). The ANCOVA
showed that the relationship of proportion (prop) of
fish that finished as a function of water velocity (V; cm/
s) over sand substrate was negative and statistically
significant (logit prop ¼ 4.046 0.1097
V; P ,
0.0001), and had a steeper slope than did relationships
for gravel (P , 0.0001) and cobble (P , 0.0001)
substrates. Slopes of relationships of proportion of fish
that finished over gravel (logit prop ¼ 1.481
0.0344
V; P ¼ 0.060) and cobble (logit prop ¼ 1.107
0.0366
V; P ¼ 0.044) were less steep; those
relationships were not statistically significant from
each other in the ANCOVA slope difference test (P ¼
0.93) and the intercept difference test (P ¼ 0.32).
Silvery minnow ascent time (t; s) to successfully
swim the entire flume length over sand substrate
increased significantly (t ¼ 8.63 þ 8.50
V; r
2
¼ 0.40, P
, 0.0001) as water velocity increased (Figure 5).
Ascent times were shorter and increased at a lower rate
over gravel substrate (t ¼ 64.90 þ 2.94
V; r
2
¼ 0.12,P
¼ 0.040). Ascent times over cobble substrate were not
significantly related to water velocity (t ¼ 300.1
143
V; r
2
¼ 0.01, P ¼ 0.61). The ANCOVAs showed
that the slope for the relationship of ascent time, as a
function of water velocity over sand substrate, was
significantly different than the slope of the relationship
for gravel (P ¼ 0.015) or cobble (P ¼ 0.003) substrate.
The slopes of relationships of ascent time as a function
of water velocity for gravel and cobble substrate were
FIGURE 4.—Percentages of Rio Grande silvery minnow that
successfully swam the length of the 18.3-m-long flume as a
function of water velocity over different substrate types. The
error bars represent SEs.
FIGURE 5.—Relationships between the time required by Rio
Grande silvery minnow to swim the length of the 18.3-m-long
flume and water velocity over different substrate types.
RIO GRANDE SILVERY MINNOW PASSAGE 441
not significantly different (P ¼ 0.14), but mean ascent
time over gravel (136 s) was significantly less (P ¼
0.004) than for cobble (265 s). Mean swimming speed
for fish that passed the flume over all substrate types
was 59 cm/s at a flume velocity of 53 cm/s, which was
similar to threshold swimming performance values
noted in U
crit
and swimming endurance tests.
For silvery minnow classified as swimmers that did
not completely ascend the flume, the maximum
distan ce swum averaged 7.8 m over all velocity
conditions over the 15-min test, or about 50% of
flume length. The difference i n mean maximum
distance swam varied little among the different water
velocities tested (7.0–8.8 m) or among the different
substrate particle sizes (7.1–8.4 m). Those fish and
ones that successfully swam the entire flume length
generally remained within 10 cm or less of the
substrate, where velocities were reduced in the
boundary area, particularly for gravel and cobble
substrate (Figure 1). Nonfinishing fish often rested on
the bottom.
Fishway Studies
For the five low-velocity tests (78 cm/s), an average
of 52% (10–100%) of the Rio Grande silvery minnow
(10 fish/test) successfully ascended the dual-vertical-
slot fishway. Although turbulence made it difficult to
track progress of individual fish, upstream swimming
must have been immediate and rapid because passage
times in the 8-m-long structure averaged 8.2 min (0.9–
18.7 min). In the four higher-velocity tests (87 cm/s; 10
fish/test), only 8% (0–10%) successfully ascended the
dual-vertical-slot fishway; all three fish that finished
moved through the structure in 5 min or less (mean ¼
2.6 min, range ¼ 0.6–4.6 min).
In the rock channel fishway, both field-captured and
hatchery silvery minnow successfully negotiated the
structure at both low and high water velocities (Table
3). In most cases, fish were attracted to the current after
release and began swimming upstream immediately. In
the low-velocity tests, an average of 74% of individuals
ascended the fishway when there was a holding pool,
compared with only 42% in high-velocity tests when
there was not. The addition of a downstream
orientation pool in subsequent high-velocity tests
reduced impingement and increased the proportion of
hatchery fish that finished to 76%.
Finish times in all rock channel tests ranged from
less than 1 min up to about 192 h. In all treatment
combinations, 26–50% of fish that finished were fast
finishers, which was similar to flume test results at the
highest velocity. In all treatment combinations, 50–
74% of fish were slow finishers ( . 60 min) and
ascended the rock channel in 922–2,296 min. About
half (56%) of the fish in the low-velocity treatment
finished during the day, whereas 65–80% of the fish
tested at high velocity finished during the day.
Observations with underwater cameras showed that
silvery minnow moved easily up and down the rock
channel. Fish passed upstream through higher velocity
gaps in weirs using areas adjacent to boulders and also
swam in the slower velocity boundary layer that existed
TABLE 3.—Mean proportions of Rio Grande silvery minnow that ascended the functional length of the 23-m-long rock channel
fishway under low-velocity (mean ¼ 58 cm/s, minima and maxima ¼ 37 and 75 cm/s) and high-velocity (mean ¼ 83 cm/s;
minima and maxima ¼ 67 and 103 cm/s) flow conditions. In the low-velocity tests, wild fish were released in a downstream pool
created by tailboards; in the high-velocity tests without a pool, fish were released into the current at the downstream end of the
fishway. The rock channel fishway was modified to create a downstream holding pool so that other high-velocity test fish
(hatchery fish were the only ones available) were allowed more opportunity for orientation before ascending the fishway. Either
10, 15, or 20 fish were released in each test. The mean proportions of fish that finished in the short- (1 h) and long-term
intervals (.1 h) and their average finish times are for all individuals in all tests. The proportion of daytime finishes was the total
number of fish observed in the head box during daytime (0600 to 1800 hours) relative to the total number observed; the
complement would be the proportion of nighttime finishes. Parenthetical values are ranges.
Wild fish Hatchery fish
Variable
Low velocity
with pool
(N ¼ 4)
High velocity
without pool
(N ¼ 4)
High velocity
without pool
(N ¼ 2)
High velocity
with pool
(N ¼ 6)
Proportion finishing 0.74 0.42 0.50 0.76
(0.65–0.87) (0.29–0.62) (0.40–0.60) (0.50–0.90)
Proportion short-term finishers 0.38 0.50 0.27 0.26
Proportion long-term finishers 0.62 0.50 0.73 0.74
Average time (min), short-term finishers 7 21 23 27
(1–41) (3–57) (2–60) (1–60)
Average time (min), long-term finishers 2,223 2,213 922 1,870
(95–7,200) (63–11,520) (109–2,080) (96–7,200)
Proportion daytime finishes 0.56 0.80 0.65 0.72
442 BESTGEN ET AL.
a few centimeters above the cobble substrate (e.g.,
Figure 1).
Fishway Applications
Based on time to fatigue, the maximum velocities at
which Rio Grande silvery minnow were able to ascend
a hypothetical 50-m-long fishway were 43.6, 45.5, and
47.6 cm/s for 6-, 7-, and 8-cm-long fish, respectively
the swimming speeds that the fish had to maintain to
pass the length of the fishway before becoming
fatigued were 48.9, 51.1, and 53.4 cm/s, respectively.
Similarly, the maximum flow velocities at which an
intermediate-sized (7-cm) silvery minnow was able to
ascend a 50-m-long fishway were 42.9 cm/s (main-
tained speed, 49.1 cm/s) at 158C, 45.5 cm/s (51.1) at
198C, and 48.3 cm/s (54.3) at 238C.
Velocity maxima for each hypothetical fishway
length for silvery minnow 6, 7, and 8 cm long at
198C declined rapidly as fishway length increased
(Figure 6), which was consistent with the time-to-
fatigue relationships that underlie those relationships
(Figures 2, 3). For example, the maximum fishway
flow velocities that would allow 6–8-cm silvery
minnow to pass a distance of just 5 m were 62–70
cm/s, because under those conditions fish would
probably employ anaerobic sprinting and tire quickly.
In even moderate-length fishways (e.g., 20 m), flow
velocities would need to be 49–54 cm/s for 6–8-cm
silvery minnow to pass the fishway, assuming constant
swimming.
Discussion
Swimming performance in fishes is dictated by an
array of intrinsic species-specific factors, including
body size and morphology, condition, growth rates,
flow conditioning, natal stream characteristics, and
environmental factors (e.g., water temperature, water
quality, and stream discharge regimes, which may be
affected by landscape characteristics such as urbaniza-
tion; Taylor and McPhail 1985; Nelson 1989; Hammer
1995; Billerbeck et al. 2001; Nelson et al. 2008).
Although affected by environmental factors, Rio
Grande silvery minnow were capable of relatively
high-speed and long-distance swimming over a variety
of substrate types; a relatively high proportion were
able to ascend two different fishway types.
Critical Swimming Speed Estimates
The mean U
crit
values of 51.5 cm/s and 7.8 BLs/s for
Rio Grande silvery minnow at 198C were within the
range of, or higher than, estimates for many other
species (e.g., Hocutt 1973; Jones et al. 1974) including
salmonids of comparable size. For example, four
groups of rainbow trout Oncorhynchus mykiss (89–95
mm TL) had relatively low U
crit
velocities of 36.1–41.9
cm/s, or a relative swimming rate of 3.7–4.4 BLs/s
(Beamish 1978); those metrics were similar to the
mean absolute and relative U
crit
values (41.3 cm/s and
4.25 BLs/s, respectively) measured for similar-sized
rainbow trout fed to satiation (Gregory and Wood
1998). Large Rio Grande silvery minnow had a higher
U
crit
than smaller ones, even within the relatively small
size range we used (53–75-mm TL), which supported
the generality that larger-bodied fish were faster
swimmers than smaller ones (Adams et al. 1999;
Billerbeck et al. 2001; Nelson et al. 2008; but see
Nelson et al. 2002).
Endurance Swimming
The relatively fast short-term swimming speeds of
Rio Grande silvery minnow (118 cm/s, 20.9 BLs/s)
exceeded those for large-bodied salmonids as well as
other diverse taxa, including many other cyprinids
(Beamish 1978; Scott and Magoulick 2008; Leavy and
Bonner 2009). However, maximum swimming speeds
of 100 cm/s or more were sustained for only 15 s or
less. Endurance increased steadily as water velocity
declined to about 60 cm/s and then increased rapidly
such that Rio Grande silvery minnow rarely fatigued at
water velocities in the range of 35–50 cm/s, even in
long-durati on tests. Even though predicted silvery
minnow swimming endurance was different among
test temperatures at water velocities exceeding 100 cm/
s, such differences should be interpreted cautiously
FIGURE 6.—Relationships between maximum swimming
(passage) distances for 6-, 7-, and 8-cm Rio Grande silvery
minnow at various maximum flow velocities in hypothetical
fishways at 198C. The relationships were iteratively estimated
by maximizing the predictions of velocity-dependent time-to-
fatigue swimming relationships for silvery minnow of various
lengths with different combinations of swimming distance and
water velocity. Thus, any combination of swimming distance
and maximum water velocity that lies above the relationship
for a silvery minnow of a specific size would result in passage
failure.
RIO GRANDE SILVERY MINNOW PASSAGE 443
because of slight lack of fit of regressions to high-
velocity data in the 158C and 238C relationships.
The swimming endurance threshold observed for
Rio Grande silvery minnow at 50–60 cm/s may reflect
a shift from mostly aerobic swimming powered by red
muscle fibers to anaerobically fueled swimming, which
recruits white muscle fibers (Drucker 1996; Plaut 2001;
Peake and Farrell 2004). Metabolism mode shifts
typically occur during the transition from swimming at
a prolonged pace to sprinting (Beamish 1978), but
metrics such as U
crit
may not necessarily reflect
optimum swimming velocities because fish may
choose to swim at higher, anaerobically fueled
velocities (e.g., Peake 2004). This was apparently not
the case for Rio Grande silvery minnow because U
crit
values and threshold velocities from endurance swim-
ming relationships were similar.
In contrast to large-bodied fish (Peake 2004; Haro et
al. 2004; Castro-Santos 2005; Peake 2008), flume
swimming velocities for Rio Grande silvery minnow
were greater than those estimated by U
crit
for only 27%
of fish tested, when averaged over all substrate
conditions. Similarly, only 20% (i.e., 0.26 short-term
fish 3 0.76 of the total number that finished; Table 3)
of the fish that finished after release in a rock channel
fishway under low-velocity conditions and 28% of
those released under high-velocity conditions with a
holding pool present finished in a relatively short time
(,1 h), even though an average of 75% of all fish
finished over a longer period. The short-term swim-
ming performance of small-bodied Rio Grande silvery
minnow may be lower than that of large-bodied fish of
other taxa owing to their lower absolute swimming
speeds (Peake 2004, 2008), smaller glycogen stores
available for anaerobic swimming, or lower short-term
motivation to swim in the flume or rock channel. The
roles that swimming behavior, willingness to swim,
and body size differences play in passage success
deserve more study, especially for understudied small-
bodied fishes.
Traditional fatigue–water velocity relationships for
fishes sometimes show distinct break points that
correspond to burst, prolonged, and sustained swim-
ming modes (Beamish 1978 and the references therein;
Peake et al. 1997; Adams et al. 1999; Plaut 2001).
Burst swimming is typically defined as occurring over
a time interval less than 20 s, prolonged swimming as
occurring over an interval from 20 s to 200 min, and
sustained swimming as occurring over a longer, often
indefinite period (Beamish 1978). However, we
observed only a single but relatively broad velocity
threshold (50–60 cm/s) for Rio Grande silvery minnow
that did not correspond to traditional fatigue intervals; a
time-based breakpoint between burst and sustained
swimming may be inappropriate for Rio Grande silvery
minnow.
As in most other swimming performance studies,
which show strong effects of water temperature
(Billerbeck et al. 2001; Lee et al. 2003; Haro et al.
2004; review of MacNutt et al. 2004; but see Fangue et
al. 2008), the performance of Rio Grande silvery
minnow improved a t higher water temperatures
because endurance at a given water velocity was twice
as long (or longer) at 238C than at 158C. This was
supported by regression predictions and empirical data;
that is, a substantial number of silvery minnow
endurance times were less than 5 min at 158C and
water velocities of 45–55 cm/s, whereas fewer silvery
minnow had endurance times of less than 5 min in
similar water velocities at 198C and none at 238C. The
effects of water temperature on the swimming
performance of Rio Grande silvery minnow must be
accounted for to design effective passage structures
because the seasonal timing of movements and passage
of naturally occurring populations are unknown in this
river, in which water temperatures can vary widely.
Long-Distance Swimming
Rio Grande silvery minnow were able to swim
several kilometers in just a few hours and up to 125 km
over longer durations at water velocities of 35–50 cm/s.
These distances may be conservative because many
fish were swimming when tests were terminated.
Although these data were collected according to a
nonexperimental protocol (sometimes the tests were
terminated), they nonetheless support the notion that
some plains stream fishes are capable of moving long
distances upstream as part of their life history, and Rio
Grande silvery minnow may be able to access distant
fish passage structures in the Rio Grande.
In general, field-captured fish in good condition and
exercise-conditioned fish perform better in swimming
performance tests than fish that are acclimated to
laboratory conditions (ponds or tanks in which the
water velocity is typically lower than that in a natural
setting; Vincent 1960; Taylor and McPhail 1985; Ward
and Hilwig 2004). The lower performance of hatchery
silvery minnow in our long-distance swimming tests
may be due to insufficient exercise training in their
rearing environment.
Flume Tests
We expected better performance of Rio Grande
silvery minnow at the highest flume velocities based on
swim chamber performance. For example, at a
hypothetical flume velocity of 50 cm/s, a silvery
minnow swimming at 60 cm/s would require only
about 2.5 min to ascend the length of the flume.
444 BESTGEN ET AL.
Because we allowed tests to continue for up to 15 min
we expected most fish to finish at the highest flume
water velocity of 53 cm/s. Because the highest flume
water velocity was similar to the mean U
crit
value of
51.5 cm/s, fish may have been avoiding the anaerobic
exercise required to ascend the flume during those
volitional swimming tests.
Observations during the flume tests suggested that
fish released over sand substrate moved immediately
upstream and, because velocity break s were no t
available, may have motivated the fish to swim nearly
constantly. Thus, constant swimming may have tired
fish more quickly in tests with sand substrate and
reduced the proportion of finishers at higher velocities
relative to those with gravel or cobble (the regression
slope for sand substrate was negative and steeper). The
reduced boundary layer velocity over sand substrate
may also explain why the swimming time to pass the
flume increased faster over sand than other substrate
types.
Conversely, when Rio Grande silvery minnow were
released into the flume with gravel (and especially
cobble) substrate, they were much more likely to swim
for a short time, sometimes via an upstream sprint, and
then find a velocity break behind or just above a substrate
element and rest. Unless they were disturbed, fish would
maintain position well past the posttest period. The
minimal proportion of silvery minnow that finished in
tests over gravel and cobble substrate at velocities of 8
and 23 cm/s and the lack of relationship of finish time of
silvery minnow as a function of water velocity over
cobble substrate suggested fish were not motivated to
finish swimming the flume length and instead simply
held positions longer. The results from the rock channel
fishway support the idea that some individuals take a
longer time to finish, so longer test times (e.g., .15 min)
may have resulted in a greater proportion of fish
swimming the entire distance of the flume.
Fishway Studies
Few comparative studies have been completed on
passage success in technical structures, particularly for
small-bodied fishes (Schwalme et al. 1985; Katopodis
et al. 1991; Katopodis 1992; Webber et al. 2007). The
dual-vertical-slot fishway facilitated moderate passage
success of Rio Grande silvery minnow at the relatively
low 78 cm/s water velocity. Fish oriented quickly and
immediately s wam upstream when passage was
observed. Moderate passage success at the lower
velocity was probably due to the absence of large
eddies and high-velocity turbulent flows that were
present in higher velocity Denil and single-vertical-slot
fishways (Bestgen et al. 2003), and the dual-vertical-
slot fishway at the high velocity, where passage
success was negligible. High turbulence has been
demonstrated to reduce passage success even in
relatively large-bodied white sturgeon Acipenser trans-
montanus (Webber et al. 2007), so the negative effect
of turbulence on small-bodied Rio Grande silvery
minnow was not surprising.
In the rock channel fishway, 75% of the Rio Grande
silvery minnow successfully ascended in low and high
velocities when a holding pool was available. Fast
finish times (1–27 min) suggested some silvery
minnow sprinted through the structure upon release.
For example, in the rock channel fishway with mean
velocity of 58 cm/s and a mean finish time of 7 min
(over the 23-m-long functional length of the channel),
the average fish would be required to swim at a
velocity of 63.5 cm/s to ascend the fishway. In the rock
channel fishway with mean velocity of 83 cm/s and
mean finish time of 24 min, the average fish would be
required to swim at a velocity of 84.6 cm/s to ascend
the fishway. These calculations assume fish swim in
the average water velocity while traversing the rock
channel fishway.
To mimic a natural stream, the rock channel fishway
flow velocity was controlled by design features
including gradient, bed roughness, channel hydraulic
radius, and large-scale flow obstructions. The high
passage success of Rio Grande silvery minnow in the
rock channel may be due to design considerations such
as low fishway gradient and the availability of a mosaic
of high and low water velocities in lateral and vertical
dimensions that fish could choose. This is in contrast
with the dual-vertical-slot fishway (and most other
technical fishways), where silvery minnow had to
traverse high-velocity baffle slots to ascend the
structure. Those differences may explain the higher
passage success in the rock channel fishway, even
though the average water velocities in the rock channel
and dual-vertical-slot fishways were similar (83 and 87
cm/s, respecti vely) in the high-flow test s. Video
observations in the rock channel fishway verified the
use of boundary areas as well as the fastest water
between boulder weirs to pass upstream, apparently
with minimal energy expenditure.
We were not able to test all possible combinations of
pool configuration, velocity, and source of fish because
our field-captured fish supply was exhausted. Howev-
er, our assumption that field -captured fish would
perform at least as well as hatchery fish was supported
by comparable results in the raceways with a holding
pool at low velocity, where 42% of field-captured and
50% of hatchery fish finished. The comparable or
higher performance of field-captured fish was also
supported by long-distance swimming tests at 158C.
Thus, even though we did not test it specifically, we
RIO GRANDE SILVERY MINNOW PASSAGE 445
feel that field-captured fish would perform similarly to
hatchery fish in the rock channel under conditions of
high velocity with a pool.
Fishway Passage and Applications
Understanding hypothetical water velocity–passage
distance relationships based on time-to-fatigue rela-
tionships was of interest because lengths of laboratory
fishway models are typically shorter than fish passage
structures in streams. Integrating passage distance,
fishway water velocity, and velocity-dependent fatigue
relationships allowed us to determine the maximum
water velocity that would allow Rio Grande silvery
minnow passage in a fishway of a given length. Water
velocities for successful passage through a hypothe tical
50-m-long fishway were all in the range of the 45–60
cm/s threshold noted where swimming endurance
changed rapidly with swim chamber water velocity.
Velocity-dependent fatigue relationships suggested that
even in relatively short structures, Rio Grande silvery
minnow passage via sustained sprint swimming was
unlikely. Predicted water velocity to achieve a
relatively short upstream swimming distance of 0.5 m
was 83–103 cm/s, depending on fish size. Therefore,
we recommend that maximum water velocities en-
countered by fish, even for very short periods (e.g.,
10 s), not exceed about 100 cm/s at all temperatures.
A rock channel gradient of 1% or less is recommended
to produce such passage velocities if turbulence effects
are minimized. Predictions of shorter-distance sprint
swimming may also be useful in optimizing the spatial
distribution of low-velocity resting areas for small-
bodied Rio Grande silvery minnow (as was recom-
mended by Webber et al. 2007 for the passage of white
sturgeon) because passage may not occur via sustained
swimming. A mix of flow velocities in the fishway,
including resti ng areas or a boundary layer of large
particle size, would allow Rio Grande silvery minnow
to choose the most energetically efficient passage route.
Acknowledgments
Major funding for the study was provided by the
Middle Rio Grande Endangered Species Act Collabo-
rati ve Progr am; additional funding was from the
Bureau of Reclamation, Albuquerque Area Office;
the Bureau of Reclamation Science and Technology
Program, Denver, Colorado; and U.S. Fish a nd
Wildlife Service, Albuquerque, New Mexico. We also
thank the Rio Grande Conservan cy District for
assistance. S. Platania and staff supplied field-captured
Rio Grande silvery minnow for the study, and J. E.
Brooks and J. Landye provided hatchery fish. P.
Chapman and J. Zumbrunnen, Statistics Department,
Colorado State University, provided advice on nonlin-
ear statistical modeling. C. Gorbach and M. Porter
(Bureau of Reclamation) provided administrative and
technical study assistance and manuscript revi ew;
additional reviews of M. Bowen, S. Peake, T. Wesche,
and anonymous reviewers are appreciated.
References
Adams, S. R., J. J. Hoover, and K. J. Killgore. 1999.
Swimming enduranc e of juvenile pallid sturgeon,
Scaphirhynchus albus. Copeia 1999:802–807.
Beamish, F. W. H. 1966. Swimming endurance of some
northwest Atlantic fishes. Journal of the Fisheries
Research Board of Canada 23:341–347.
Beamish, F. W. H. 1978. Swimming capacity. Pages 101–172
in W. S. Hoar and D. J. Randall, editors. Fish physiology,
volume VII. Locomotion. Academic Press, New York.
Bell, W. H., and L. D. B. Terhune. 1970. Water tunnel design
for fisheries research. Canada Fisheries and Marine
Service Technical Report 195.
Bestgen, K. R., B. Mefford, J. Bundy, C. Walford, and B.
Compton. 2003. Swimming performance of Rio Grande
silvery minnow. Final report to the U.S. Bureau of
Reclamation, Albuquerque, New Mexico. Colorado State
University, Larval Fish Laboratory Contribution 132,
Fort Collins.
Bestgen, K. R., and S. P. Platania. 1990. Extirpation of
Notropis simus simus (Cope) and Notropis orca Wool-
man (Pisces: Cyprinidae) from the Rio Grande in New
Mexico, with notes on their life history. Occasional
Papers of t he Museum of Southwestern Biology,
University of New Mexico 6:1–8.
Bestgen, K. R., and S. P. Platania. 1991. Status and
conserva ti on of the Rio Grande silve ry minnow,
Hybognathus amarus. Southwestern Naturalist 36:225–
232.
Bestgen, K. R., S. P. Platania, J. E. Brooks, and D. L. Propst.
1989. Dispersal and life history traits of Arkansas River
shiner Notropis girardi introduced into the Pecos River,
New Mexico. American Midland Naturalist 122:228–
235.
Bestgen, K. R., and D. L. Propst. 1996. Redescription,
geographic variation, and taxonomic status of the Rio
Grande silvery minnow, Hybognathus amarus (Girard
1856). Copeia 1996:41–55.
Billerbeck, J. M., T. E. Lankford Jr., and D. O. Conover.
2001. Evolution of intrinsic growth and energy acquisi-
tion rates, I. Trade-offs with swimming performance in
Menidia menidia. Evolution 55:1863–1872.
Brett, J. R. 1964. The respiratory metabolism and swimming
performance of young sockeye salmon. Journal of the
Fisheries Research Board of Canada 21:1183–1226.
Brett, J. R. 1967. Swimming performance of sockeye salmon
(Oncorhynchus nerka) in relation to fatigue time and
temperature. Journal of the Fisheries Research Board of
Canada 24:1731–1741.
Castro-Santos, T. 2005. Optimal swim speeds for traversing
velocity barriers: an analysis of volitional high-speed
swimming behavior of migratory fishes. Journal of
Experimental Biology 208:421–432.
Clay, H. C. 1995. Design of fishways and other fish facilities,
2nd edition, CRC Press, Boca Raton, Florida.
446
BESTGEN ET AL.
Cook, J. A., K. R. Bestgen, D. L. Propst, and T. L. Yates.
1992. Allozymic divergence and systematics of the Rio
Grande silvery minnow, Hybognathus amarus (Teleostei:
Cyprinidae). Copeia 1992:36–44.
Cross, F. B., R. E. Moss, and J. C. Collins. 1985. Assessment
of dewatering impacts on stream fisheries in the Arkansas
and Cimarron rivers. Museum of Natural History,
University of Kansas, Lawrence.
Drucker, E. G. 1996. The use of gait transition speed in
comparative studies of fish locomotion. American
Zoologist 36:555–566.
Dudley, R. K., and S. P. Platania. 2007. Flow regulation and
fragmentation imperiled pelagic-spawning riverine fish-
es. Ecological Applications 17:2074–2086.
Fangue, N. A., M. Mandic, J. G. Richards, and P. M. Schulte.
2008. Swimm ing performance and energetics as a
function of temperature in killifish Fundulus heteroclitus.
Physiological and Biochemical Zoology 81:389–401.
Fausch, K. D., and K. R. Bestgen. 1997. Ecology of stream
fishes indigenous to the central and southern Great
Plains. Pages 131–166 in F. L. Knopf and F. B. Samson,
editors. Ecology and conserva tion of Great Plains
vertebrates. Springer-Verlag, New York.
Gregory, T. R., and C. M. Wood. 1998. Individual variation
and interrelationships between swimming performance,
growth rate, and feeding in juvenile rainbow trout
(Oncorhynchus mykiss). Canadian Journal of Fisheries
and Aquatic Sciences 55:1583–1590.
Hammer, C. 1995. Fatigue and exercise tests with fish.
Comparative Biochemistry and Physiology 112A:1–20.
Haro, A., T. Castro-Santos, J. Noreika, and M. Odeh. 2004.
Swimming performance of upstream migrant fishes in
open-channel flow: a new approach to predicting passage
through velocity barriers. Canadian Journal of Fisheries
and Aquatic Sciences 61:1590–1601.
Hocutt, C. H. 1973. Swimming performance of three warm-
water fishes exposed to a rapid temperature change.
Chesapeake Science 14:11–16.
Jones, D. R., J. W. Kiceniuk, and O. S. Bamford. 1974.
Evaluation of the swimming performance of several fish
species from the Mackenzie River. Journal of the
Fisheries Research Board of Canada 31:1641–1647.
Katopodis, C. 1992. Introduction to fishway design. U.S. Fish
and Wildlife Service, National Conservation Training
Center, Shepherdstown, West Virginia.
Katopodis, C., A. J. Derksen, and B. L. Christensen. 1991.
Assessment of two Denil fishways for passa ge of
freshwater species. Pages 306–324 in J. Colt and R. J.
White, editors. Fisheries bioengineering. American
Fisheries Society, Symposium 10, Bethesda, Maryland.
Leavy, T. R., and T. H. Bonner. 2009. Relationships among
swimming ability, current velocity association, and
morphology for freshwater lotic fishes. North American
Journal of Fisheries Management 29:72–83.
Lee, C. G., A. P. Farrell, A. Lotto, M. J. MacNutt, S. G.
Hinch, and M. C. Healey. 2003. The effect of
temperature on swimming performanc e and oxygen
consumption in adult sockeye (Oncorhynchus nerka)
and coho ( O. kisutch) sal mon stocks. Journal of
Experimental Biology 206:3239–3251.
Luttrell, G. R., A. A. Echelle, W. L. Fischer, and D. J.
Eisenhour. 1999. Declining status of two species of the
Macrhybopsis aestivalis complex (Teleostei: Cyprinidae)
in the Arkansas River basin and related effects of
reservoirs as barriers to dispersal. Copeia 1999:981–989.
MacNutt, M. J., S. G. Hinch, A. P. Farrell, and S. Topp. 2004.
The effect of temperature and acclimation period on
repeat swimming performance in cutthroat trout. Journal
of Fish Biology 65:342–353.
Moore, G. A. 1944. Notes on the early life history of Notropis
girardi. Copeia 1944:209–214.
Nelson, J. A. 1989. Critical swimming speeds of yellow perch
Perca flavescens: comparison of populations from a
naturally acidic lake and a circumneutral lake in acid and
neutral water. Journal of Experimental Biology 145:239–
254.
Nelson, J. A., P. S. Gotwalt, S. P. Reidy, and D. M. Webber.
2002. Beyond U
crit
: matching swimming performance
tests to the physiological ecology of the animal, including
a new fish ‘‘ drag strip.’’ Comparative Biochemistry and
Physiology 133A:289–302.
Nelson, J. A., P. S. Gotwalt, and J. W. Snodgrass. 2003.
Swimming performance of blacknose dace (Rhinichthys
atratulus) mirrors home-stream current velocity. Cana-
dian Journal of Fisheries and Aquatic Sciences 60:301–
308.
Nelson, J. A., P. S. Gotwalt, C. A. Simonetti, and J. W.
Snodgrass. 2008. Environmental correlates, plasticity,
and repeatability of differences in performance among
blacknose dace (Rhinichthys atratulus) populati ons
across a gradient of urbanization. Physiological and
Biochemical Zoology 81:25–42.
Peake, S. J. 2004. An evaluation of the use of critical
swimming speed for determination of culvert water
velocity criteria for smallmouth bass. Transactions of the
American Fisheries Society 133:1472–1479.
Peake, S. J. 2008. Behavior and passage performance of three
warmwater nonsalmonid fish species in an experimental
raceway. North American Journal of Fish Management
28:321–327.
Peake, S. J., F. W. H. Beamish, R. S. McKinley, D. A.
Scruton, and C. Katopodis. 1997. Relating swimming
performance of lake sturgeon, Acipenser fulvescens,to
fishway design. Canadian Journal of Fisheries and
Aquatic Sciences 54:1361–1366.
Peake, S. J., and A. P. Farrell. 2004. Locomotory behaviour
and postexercise physiology in relation to swimming
speed, gait transition, and metabolism in free-swimming
smallmouth bass (Micropterus dolomieu). Journal of
Experimental Biology 207:1563–1575.
Peake, S. J., and A. P. Farrell. 2006. Fatigue is a behavioural
response in respirometer-confined smallmouth bass.
Journal of Fish Biology 68:1742–1755.
Platania, S. P. 1991. Fishes of the Rio Chama and upper Rio
Grande, New Mexico, with preliminary comments on
their longitudinal distribution. Southwestern Naturalist
36:186–193.
Platania, S. P., and C. S. Altenbach. 1998. Reproductive
strategies and egg types of seven Rio Grande basin
cyprinids. Copiea 1998:559–569.
Plaut, I. 2001. Criti cal swimming speed: its ecological
relevance. Comparative Biochemistry and Physiology
131A:41–50.
Schwalme, K., W. C. Mackay, and D. Lindner. 1985.
RIO GRANDE SILVERY MINNOW PASSAGE 447
Suitability of vertical slot and Denil fishways for passing
north-temperate nonsalmonid fish. Canadian Journal of
Fisheries and Aquatic Sciences 42:1815–1822.
Scott, M. K., and D. D. Magoulick. 2008. Swimming
performance of five warmwater stream fish species.
Transactions of the American Fisheries Society 137:209–
215.
Smit, H., J. M. Amelink-Koutstaal, J. Vijverberg, and J. C.
von Vaupel-Klein. 1971. Oxygen consumption and
efficiency of swimming goldfish. Comparative Biochem-
istry and Physiology 39A:1–28.
Taylor, E. B., and J. D. McPhail. 1985. Variation in burst and
prolonged swimming performance among British Co-
lumbia populations of coho salmon, Oncorhynchus
kisutch. Canadian Journal of Fisheries and Aquatic
Sciences 42:2029–2033.
U.S. Department of the Interior. 1994. Endangered and
threatened wildlife and plants: final rule to list Rio
Grande silvery minnow as an endangered species.
Federal Register 59:(20 July 1994):36988–36995.
Vincent, R. E. 1960. Some influences of domestication upon
three stocks of brook trout (Salvelinus fontinalis Mit-
chill). Transactions of the American Fisheries Society
89:35–52.
Ward, D. L., and K. D. Hilwig. 2004. Effects of holding
environment and exercise conditioning on swimming
performance of southwestern native fishes. North
American Journal of Fisheries Management 24:1083–
1087.
Webber, J. D., S. N. Chun, T. R. Maccoll, L. T. Mirise, A.
Kawabata, E. K. Anderson, T. Sung Cheong, L. Kavvas,
M. Mcgee Rotondo, K. L. Hochgraf, R. Churchwell, and
J. J. Cech Jr. 2007. Upstream swimming performance of
adult white sturgeon: effects of partial baffles and a ramp.
Transactions of the American Fisheries Society 136:402–
408.
Winston, M. R., C. M. Taylor, and J. Pigg. 1991. Upstream
extirpation of four minnow species due to damming of a
prairie stream. Transactions of the American Fisheries
Society 120:98–105.
448
BESTGEN ET AL.