Content uploaded by Jeremy S. Tiemann
Author content
All content in this area was uploaded by Jeremy S. Tiemann on Feb 03, 2014
Content may be subject to copyright.
q
2005 by the American Society of Ichthyologists and Herpetologists
Copeia, 2005(3), pp. 539–549
Spatiotemporal Patterns of Fish Assemblage Structure in a River
Impounded by Low-Head Dams
D
AVID
P. G
ILLETTE
,J
EREMY
S. T
IEMANN
,D
AVID
R. E
DDS
,
AND
M
ARK
L. W
ILDHABER
We studied spatiotemporal patterns of fish assemblage structure in the Neosho
River, Kansas, a system impounded by low-head dams. Spatial variation in the fish
assemblage was related to the location of dams that created alternating lotic and
lentic stream reaches with differing fish assemblages. At upstream sites close to
dams, assemblages were characterized by species associated with deeper, slower-
flowing habitat. Assemblages at sites immediately downstream from dams had higher
abundance of species common to shallow, swift-flowing habitat. Temporal variation
in assemblage structure was stronger than spatial variation, and was associated with
fish life history events such as spawning and recruitment, as well as seasonal changes
in environmental conditions. Our results suggest that low-head dams can influence
spatial patterns of fish assemblage structure in systems such as the Neosho River
and that such assemblages also vary seasonally.
R
IVERINE fish assemblage structure often
varies along environmental gradients from
headwaters to lower mainstem (Schlosser, 1982;
Gelwick, 1990; Edds, 1993). When these gradi-
ents are interrupted, however, alternate pat-
terns can result. Such interruptions can be nat-
ural, as is the case with waterfalls (Balon and
Stewart, 1983; Maret et al., 1997) or rapids (Bal-
on and Stewart, 1983), or anthropogenic, as oc-
curs with river fragmentation from damming
(reviewed by Baxter, 1977; Dynesius and Nils-
son, 1994; and Richter et al., 1997).
Dams affect lotic systems in many ways, and
their impacts are often reflected in the spatial
patterns of fish assemblages. Dam construction
can fragment watersheds (Dynesius and Nilsson,
1994), affecting fish assemblages directly by
eliminating or reducing movement of fishes,
leading to reduced upstream species richness,
especially for migratory species (ReyesGavilan
et al., 1996; Holmquist et al., 1998; March et al.,
2003). Alteration of the natural flow regime also
influences fish assemblage structure, reducing
in abundance species reliant upon seasonal flow
variation to complete their life cycle (Bonner
and Wilde, 2000; Minckley et al., 2003). Hypo-
limnial-release dams can also sharply decrease
downstream water temperature, resulting in de-
creased growth and increased time to maturity
for native fishes (Clarkson and Childs, 2000)
and often replacing native warmwater assem-
blages with non-native coldwater assemblages
over time (Quinn and Kwak, 2003). Impound-
ment-induced changes in current velocity also
cause habitat alteration, with decreased flow ve-
locities and high siltation rates upstream of
dams (Kondolf, 1997; Bennett et al., 2002) and
increased flow velocities leading to substrate
scouring downstream (Kondolf, 1997; Camargo
and Voelz, 1998). Many of these habitat alter-
ations create conditions favorable for non-na-
tive species (Marchetti and Moyle, 2001), which
can then further alter fish assemblage compo-
sition via predation and competition (Richter et
al., 1997; Godinho and Ferreira, 1998; Eby et
al., 2003).
Although the number of free-flowing riverine
ecosystems in the world continues to decline
(Poff et al., 1997), more research is required to
understand the spatial pattern of fish assem-
blage structure in impounded rivers. Compli-
cating the situation is the fact that not all types
of dams affect lotic ecosystems similarly. For ex-
ample, deep hypolimnial release dams, with
their associated cooling effect downstream, af-
fect rivers in different ways than smaller flood-
control dams that do not alter river temperature
(Petts, 1984).
Low-head dams (
,
4 m in height) are a type
of impoundment common to many North
American rivers. For example, Kansas has the
second-highest number of dams in the United
States (5,699; Shuman, 1995), and the vast ma-
jority of these are low-head dams. The Neosho
River in eastern Kansas alone is impounded by
15 such structures. Given the well-documented
effects of other types of impoundments on lotic
systems, low-head dams, in spite of their small
size, appear likely to affect riverine fish assem-
blages. Indeed, Porto et al. (1999) showed that
low-head dams with heights less than 1.5 m can
alter fish assemblage composition, with species
richness declining immediately upstream. How-
ever, Dodd et al. (2003) and Raborn and
Schramm (2003) documented habitat alteration
in the presence of low-head dams, but little
540 COPEIA, 2005, NO. 3
Fig. 1. Map of study area in Lyon County, KS,
showing eight study sites and two low-head dams
along the Neosho River from Americus to Emporia.
overall change to the fish assemblage. Although
they are common throughout North America,
little is know about the spatiotemporal patterns
of fish assemblages within river systems im-
pounded by low-head dams. Our objective was
to quantify patterns of spatiotemporal variation
in fish assemblages in the Neosho River, Kansas,
and to evaluate the extent to which low-head
dam impoundments affected these patterns.
M
ATERIALS AND
M
ETHODS
Study area.—The Neosho River lies within the
Prairie Parkland ecosystem province (Bailey,
1983), and is part of the Arkansas River drain-
age. It is fifth-order in our study reach, draining
mostly mixed-grass prairie and cropland, with
mature riparian vegetation along some sections.
We sampled eight sites along a 34-km stretch of
the river from Americus to Emporia in Lyon
County, Kansas (Fig. 1). Sites were selected to
be representative of the Neosho River based on
location relative to two low-head dams (see be-
low) and appropriate for our sampling meth-
odology (i.e., relatively shallow, with consistent
current velocity). Overall stream gradient is low
(0.44 m/km), although variation in gradient be-
tween sites varied due to impoundments. Gra-
dient was lowest between sites 5 and 6 (negli-
gible) and highest between sites 6 and 7 (1.31
m/km). There are no permanent tributaries
along this stretch of the Neosho River. At each
site, we fixed five permanent transects perpen-
dicular to shore, spaced equally every 5 to 10
m, depending on site length.
Sampling.—We sampled each site monthly from
November 2000 to October 2001. Samples were
taken between the 9
th
and 22
nd
of each month
during daylight hours, and sampling order of
sites was randomized each month. We were un-
able to adequately sample Site 2 in Januar y and
February and Sites 5 and 6 from December
through February due to ice cover, and Site 7
in August and December because flow velocities
were too low for our sampling methodology.
Sampling at each site proceeded from down-
stream to upstream transects, and from near
shore to far shore points along each transect.
We sampled up to five points along each tran-
sect, depending on river width and depth and
landowner permission. At Site 8, landowner per-
mission was only obtained for one side of the
river. However, we were able to obtain represen-
tative samples of the fish assemblage at this site
in spite of this limitation. All sampling points
along each transect were spaced at least 0.5 m
apart to minimize disturbing adjacent points. At
each point we sampled fishes by kick-seining,
using a 1.5-m length by 1.8-m height section of
3-mm mesh seine. Upon fixing the seine at a
sampling point, we disturbed substrate begin-
ning 3 m upstream. In this manner, fishes with-
in a 4.5-m
2
area were carried downstream into
the seine. This methodology effectively captures
both water column and benthic fish species
(Matthews, 1990; Wildhaber et al., 1999). We
counted and identified fishes as juvenile or
adult, using a 30-mm total length (TL) maxi-
mum juvenile length for minnows (Campostoma,
Phenacobius,Pimephales,Cyprinella,Notropis, and
Lythrurus spp.) and darters (Etheostoma and Per-
cina spp.), and a 50-mm TL maximum juvenile
length for madtoms (Noturus spp.) and sunfish-
es (Lepomis spp.), following Gelwick (1990).
Other fishes were measured individually and
classified as juvenile or adult based on pub-
lished accounts. We used a 305-mm maximum
juvenile cutoff length for Channel Catfish (Ic-
talurus punctatus), 380 mm for Flathead Catfish
(Pylodictis olivaris), and 280 mm for redhorses
(Moxostoma spp.) based on work by Deacon
(1961) in the Neosho River, 220 mm for Spot-
ted Bass (Micropterus punctulatus) and White
Bass (Morone chrysops), and 240 mm for carp-
suckers (Carlander, 1969, 1977, 1997). We did
not distinguish between juvenile and adult Gam-
busia affinis. All fishes were held until sampling
of the site was completed, then returned to the
river. Juvenile redhorses and carpsuckers were
difficult to identify in the field due to small size
and were recorded as Moxostoma sp. and Carpi-
odes sp., respectively.
541GILLETTE ET AL.—RIVERINE FISH ASSEMBLAGES
Habitat measurement.—We measured water
depth, current velocity at 60% depth, substrate
composition, and substrate compaction at all
points along each cross-stream transect. Velocity
was measured using a Global Flow Probe (Glob-
al Water Company, Gold River, CA). We visually
estimated substrate at each point as percentage
composition of clay/silt, sand, gravel, pebble,
cobble, boulder, and bedrock (Mullner et al.,
2000). Definition of substrate categories and
sampling methodology followed Bain (1999).
Compaction, a surrogate of the amount of fine
sediment surrounding larger substrate types,
was quantified by tactile evaluation; each point
was assigned a compaction index value from 1
to 4, with 1 representing loose substrate, 2 sub-
strate lightly packed with clay/silt, 3 substrate
tightly packed with clay/silt, and 4 bedrock (Fu-
selier and Edds, 1996).
After fish were collected, water quality was
measured immediately upstream of transects.
We measured water temperature with a labora-
tory thermometer, and dissolved oxygen and
pH with a Hach kit model AL-36B. We then
took a 1-L sample of surface water for further
analysis. From this sample we measured alkalin-
ity and hardness with a Hach kit model AL-36B;
nitrate, ammonia, carbon dioxide, total acidity,
and orthophosphate with a Hach Surface Wa-
ters kit; chloride and sulfate with a Hach kit
model DREL/1C; and turbidity with a Hach
2100P turbidimeter. Two 100-ml portions of the
1-L sample were vacuum filtered using Pall-Gel-
man Type A/E round 47-mm glass fiber filters,
and the filtrate frozen at
2
10 C for future de-
termination of chlorophyll aand particulate or-
ganic carbon (POC). We measured chlorophyll
ausing a model 10-AU-005 Field Flourometer
(Turner Designs, Sunnyvale, CA) and POC us-
ing a Coulometrics Carbon Model 5014 Analyz-
er (UIC, Inc., Joliet, IL).
Data analysis.—Analyses were performed using
SAS v.8 and SPSS v.7.5.1. Ordinations were con-
ducted using PC-Ord v.4. We included only
those taxa occurring in
.
5% of collections for
ordination (Gauch, 1982).
To examine spatiotemporal patterns of fish
assemblage structure, we used correspondence
analysis (CA) to ordinate collections from each
site during each month. We included 84 collec-
tions in the CA, composed of 17 juvenile and
16 adult taxa in addition to G. affinis (Table 1).
We excluded samples taken in December at Site
8 and January at Site 7 because these collections
consisted of only one fish. We analyzed conspe-
cific juveniles and adults as separate taxa be-
cause spatial and trophic resource use varies on-
togenetically for many stream fishes (Schlosser,
1982; Gelwick, 1990; Gido and Propst, 1999). To
test for effects of low-head dam impoundments
on fish assemblages, we grouped sites into one
of three levels of an ‘‘impoundment treat-
ment,’’ based on proximity to these dams. Sites
less than 2 km downstream from dams (Sites 3,
4, and 7) comprised a ‘‘downstream’’ level of
treatment, sites less than 5 km upstream (Sites
2, 5, and 6) an ‘‘upstream’’ level, and sites great-
er than 5 km from dams (Sites 1 and 8) a ‘‘dis-
tant’’ level of treatment.
To test for impoundment and temporal ef-
fects on CA 1 and 2 scores, we used the mixed
linear model (SAS Proc Mixed). Scores from
both axes were modeled separately as depen-
dent variables, with month, impoundment, and
the interaction between the two as fixed effects
(Agresti and Finlay, 1997). Sites were modeled
as repeated subjects nested within levels of im-
poundment treatment. The mixed linear model
is a generalization of the general linear model
that allows data to exhibit correlation and non-
constant variability; fixed effect parameters are
associated with known explanatory variables, as
in the general linear model. Where appropri-
ate, we used a Tukey-Kramer multiple compar-
ison test on least-square means to distinguish
significant differences among treatment levels.
To examine relationships between environ-
mental gradients and CA axes, we calculated
Pearson’s correlation coefficient between axis
scores and environmental variables for each col-
lection. To assess potential effects of hydrologi-
cal variation on assemblage structure, we in-
cluded in the correlation matrix river discharge
for the day of each collection, measured at the
U.S. Geological Survey gauging station on the
Neosho River at Americus, KS.
Because of the large number of environmen-
tal variables measured, we used Principal Com-
ponents Analysis (PCA) on a collections-by-en-
vironmental variables matrix to eliminate re-
dundant variables. In the case of suites of vari-
ables loading similarly on the first three PCA
axes, we selected the one variable most biolog-
ically meaningful to represent the group. In this
analysis, water depth was selected from a group
of variables including percent substrate com-
position of clay/silt, gravel and pebble, POC
and ammonia. Dissolved oxygen was selected
from a suite of water chemistry variables includ-
ing sulfate, alkalinity, dissolved carbon dioxide,
hardness, pH, and nitrate. Current velocity was
selected from a group including substrate em-
beddedness and water turbidity, and percent
cobble substrate composition from a group in-
cluding percent boulder substrate composition.
542 COPEIA, 2005, NO. 3
T
ABLE
1. S
PECIES
C
OLLECTED FROM
E
IGHT
S
ITES ON THE
N
EOSHO
R
IVER
,L
YON
C
O
., KS, N
OVEMBER
2000
TO
O
CTOBER
2001, S
HOWING
P
ERCENT
C
OMPOSITION BY
S
PECIES
,I
NCLUSION OF
A
DULT
(A)
AND
J
UVENILE
(J) T
AXA
IN
CA O
RDINATION
,
AND
C
ORRELATIONS
(P
EARSON
’
S
R
)
OF
T
AXON
A
BUNDANCE WITH
CA 1
AND
2. Criterion for
inclusion was occurrence in at least 5% of collections. (*Juvenile and adults were not distinguished for Gam-
busia affinis).
Species % Assemblage
composition Taxa included
in CA CA 1 rCA 2 r
Cyprinella lutrensis 45.04 A
2
0.15
2
0.51
J 0.41
2
0.10
Notropis buchanani 11.13 A
2
0.64 0.19
J
2
0.07
2
0.02
Pimephales notatus 10.88 A
2
0.28 0.05
J
2
0.01 0.36
Lepomis humilis 7.67 A
2
0.22
2
0.17
J 0.01 0.13
Percina phoxocephala 7.56 A
2
0.26
2
0.45
J
2
0.30 0.06
Pimephales vigilax 4.94 A 0.04 0.02
J 0.25 0.10
Etheostoma spectabile 3.04 A 0.40
2
0.15
J
2
0.17
2
0.12
Pimephales tenellus 2.78 A
2
0.34 0.03
J
2
0.06 0.04
Phenacobius mirabilis 1.29 A
2
0.11
2
0.53
Campostoma anomalum 1.20 A
2
0.15
2
0.24
J
2
0.37 0.06
Notropis stramineus 0.85 A
2
0.09
,
0.01
Ictalurus punctatus 0.75 J 0.05 0.03
Percina caprodes 0.56 A
2
0.24
2
0.34
J
2
0.30 0.02
Noturus placidus 0.44 A
2
0.03 0.03
J 0.01 0.13
Gambusia affinis 0.42 * 0.23 0.09
Percina copelandi 0.42 A
2
0.28
2
0.02
Lepomis cyanellus 0.36 A
2
0.14
2
0.15
J 0.09 0.17
Lepomis macrochirus 0.22 J 0.25 0.24
Noturus flavus 0.16 A 0.08
2
0.16
J
2
0.06
2
0.10
Cyprinella camura 0.06
Aplodinotus grunniens 0.06 J
2
0.19
2
0.05
Moxostoma sp. 0.04 J
2
0.14
2
0.12
Lepomis megalotis 0.03
Fundulus notatus 0.02
Pylodictis olivaris 0.02
Lythrurus umbratilis 0.02
Micropterus punctulatus 0.01
Morone chrysops 0.01
Dorosoma cepedianum 0.01
Etheostoma flabellare 0.01
Carpiodes sp. 0.01
This resulted in the following 10 variables re-
tained for correlation analysis with CA axes: per-
cent substrate composition of sand, cobble, and
bedrock, water depth and current velocity, water
temperature, dissolved oxygen, chloride, chlo-
rophyll aand river discharge.
R
ESULTS
We collected 15,215 fishes of 31 species, ac-
counting for 44 taxa inclusive of juveniles and
adults (Table 1). Ten families from five orders
were represented; Cyprinidae had the greatest
543GILLETTE ET AL.—RIVERINE FISH ASSEMBLAGES
T
ABLE
2. M
EAN AND
S
TANDARD
D
EVIATION OF
E
NVIRONMENTAL
V
ARIABLES FOR
E
IGHTY
-F
OUR
C
OLLECTIONS
M
ADE
O
VER
T
WELVE
M
ONTHS FROM
E
IGHT
S
ITES
. Eight collections were omitted from upstream sites (two from De-
cember and three from both January and February), three from downstream sites (one in December, one in
January, and one in August), and one from distant sites (December).
Upstream
Mean S. D.
Downstream
Mean S. D.
Distant
Mean S. D.
Substrate composition (%)
Bedrock 0.0 0.0 27.7 39.5 0.0 0.0
Boulder 1.6 3.4 0.2 0.5 0.0 0.0
Cobble 4.7 7.0 2.1 2.9 0.3 1.0
Pebble 33.4 8.7 26.1 17.1 36.4 7.1
Gravel 38.1 8.1 30.7 18.8 41.8 6.2
Sand 3.4 1.4 3.9 3.9 6.0 2.9
Clay/Silt 18.9 16.5 8.3 8.7 15.7 7.9
Other microhabitat variables
Substrate compaction 2.2 0.3 2.4 1.0 1.8 0.3
Water depth (cm) 58.2 8.7 26.7 13.5 38.1 17.3
Flow velocity (m/s) 0.1 0.1 0.4 0.2 0.4 0.2
Water properties
Dissolved oxygen (mg/L) 8.3 2.2 10.0 2.4 8.5 2.9
pH 8.0 0.1 8.0 0.2 8.0 0.2
Alkalinity (mg/L) 171.2 54.8 183.2 60.0 171.2 49.6
Hardness (mg/L) 232.8 42.8 239.7 51.4 237.9 47.9
Turbidity (NTU) 33.4 27.6 38.9 44.9 43.4 40.7
Dissolved carbon dioxide (mg/L) 10.2 3.5 10.4 4.0 10.7 4.2
Ammonia (mg/L) 0.0 0.0 0.0 0.0 0.0 0.0
Nitrate (mg/L) 0.0 0.0 0.0 0.0 0.0 0.0
Chloride (mg/L) 9.1 4.6 9.2 5.5 8.7 3.5
Sulfate (mg/L) 28.1 9.7 27.5 7.1 28.4 8.2
Particulate organic carbon (
m
g/L) 1555.8 931.7 1556.8 925.9 1793.7 1027.8
Chlorophyll a(
m
g/L) 704.0 604.7 510.7 620.8 506.5 487.1
number of species (10), followed by Percidae
(5), Centrarchidae (5), and Ictaluridae (4).
Habitat varied among upstream, downstream,
and distant sites (Table 2). Upstream sites were
deeper and slower-flowing than downstream
and distant sites. Downstream sites were shallow-
est, had the highest percentage substrate com-
position of bedrock, and had the lowest per-
centage composition of clay/silt.
Relative abundance of fishes varied with im-
poundment treatment and season (Table 3).
Cyprinella lutrensis was the most abundant taxa
at downstream and distant sites, and P. notatus
or N. buchanani most abundant at upstream
sites, depending on season. Percina phoxocephala
was more abundant at downstream and distant
sites than at upstream sites, and Lepomis humilis
more abundant at upstream sites than at down-
stream or distant sites.
Axis 1 of the CA showed a temporal pattern
of fish assemblage structure, covering a gradient
of 5.6 standard deviations with an eigenvalue of
0.399, and explaining 16.8% of the variance.
Month significantly affected Axis 1 scores (Ta-
ble 4), with winter collections scoring highest
and early summer collections lowest (Figs. 2, 3).
There were no significant impoundment or in-
teraction effects (Table 4). Axis 1 was positively
correlated with dissolved oxygen and negatively
correlated with water temperature, chlorophyll
a, water depth and current velocity, percent sub-
strate composition of sand, and river discharge
(Table 5). Axis 1 thus represented a pattern of
fish assemblage structure along a gradient from
cold, shallow, and slow-flowing winter condi-
tions to warm, deeper, and swifter-flowing sum-
mer conditions.
Taxa characteristic of winter collections were
positively correlated with Axis 1, and taxa char-
acterizing summer collections were negatively
correlated (Tables 1, 3). Strong positive corre-
lates included G. affinis, which was only collect-
ed in November and December, adult E. specta-
bile, and juvenile C. lutrensis,P. vigilax, and L.
macrochirus (Table 1). Strong negative correlates
of CA 1 included juvenile P. caprodes,C. anom-
544 COPEIA, 2005, NO. 3
T
ABLE
3. R
ELATIVE
A
BUNDANCE OF
T
AXA
C
OLLECTED FROM
U
PSTREAM
,D
OWNSTREAM
,
AND
D
ISTANT
S
ITES
D
URING
W
INTER AND
S
PRING
(A)
AND
S
UMMER AND
F
ALL
(B). Abundance of some species differed by less than 0.1%;
others were equal in abundance. (A
5
adult, J
5
juvenile).
Upstream
Rank Taxon %
Downstream
Rank Taxon %
Distant
Rank Taxon %
A. Winter–Spring
1P. notatus A 21.5 1 C. lutrensis A 43.9 1 C. lutrensis A 65.8
2N. buchanani A 20.6 2 P. phoxocephala A 11.7 2 P. notatus A 5.2
3L. humilis J 14.7 3 N. buchanani A 10.7 3 P. phoxocephala A 4.8
4P. tenellus A 7.4 4 P. notatus A 7.4 4 C. lutrensis J 4.7
5C. lutrensis A 6.8 5 C. lutrensis J 6.0 5 P. tenellus A 3.8
6P. vigilax A 5.7 6 L. humilis J 4.9 6 L. humilis J 3.4
7P. notatus J 5.6 7 P. vigilax A 4.0 7 P. vigilax A 3.0
8P. phoxocephala A 4.9 8 E. spectabile A 3.5 8 N. buchanani A 2.5
9C. lutrensis J 3.7 9 P. mirabilis A 1.9 9 E. spectabile A 2.0
10 E. spectabile A 3.3 10 N. stramineus A 1.2 10 P. copelandi A 0.9
B. Summer–Fall
1N. buchanani A 18.6 1 C. lutrensis A 38.2 1 C. lutrensis A 20.5
2C. lutrensis A 17.9 2 C. lutrensis J 23.3 2 N. buchanani A 16.3
3L. humilis J 14.6 3 P. phoxocephala A 7.3 3 C. lutrensis J 15.4
4C. lutrensis J 13.6 4 P. notatus A 5.4 4 P. notatus A 8.9
5P. notatus A 8.9 5 E. spectabile A 4.3 5 P. phoxocephala A 7.4
6P. notatus J 4.5 6 N. buchanani A 4.1 6 P. vigilax A 4.5
7P. tenellus A 3.8 7 P. vigilax A 3.0 7 P. notatus J 3.6
7P. phoxocephala A 3.8 8 L. humilis J 2.0 8 L. humilis J 3.1
9P. vigilax A 3.3 9 P. mirabilis A 2.0 9 E. spectabile A 3.0
10 E. spectabile A 1.2 10 P. vigilax J 1.9 10 P. mirabilis A 2.9
T
ABLE
4. R
ESULTS OF
T
WO
-W
AY
R
EPEATED
-M
EASURES
A
NALYSIS OF
CA 1
AND
CA 2 S
CORES FOR
C
OLLECTIONS
FROM THE
N
EOSHO
R
IVER
, KS, 2000–2001,
WITH
M
ONTH AND
I
MPOUNDMENT AS
T
REATMENTS
.
Axis Effect Numerator d.f. Denominator d.f. FP
CA 1 Impoundment 2 5 0.66 0.558
Month 11 45 31.34
,
0.0001
Interaction 20 45 1.21 0.291
CA 2 Impoundment 2 5 24.51 0.003
Month 11 45 3.78 0.001
Interaction 20 45 2.59 0.004
alum, and P. phoxocephala; these age-0 fishes were
present only from June through early Fall. Low-
scoring adult taxa included N. buchanani and P.
tenellus; these two taxa were collected most fre-
quently from April through August, with N.
buchanani almost completely absent in other
months.
Axis 2 of the CA showed a spatial pattern of
fish assemblage structure related to low-head
dams, with a slight temporal component, cov-
ering a gradient of 3.5 standard deviations with
an eigenvalue of 0.315, and explaining 13.3%
of the variance. Impoundment, month, and im-
poundment-by-month interaction significantly
affected Axis 2 scores (Table 4). Upstream sites
scored higher than downstream sites (Fig. 2).
Relative position of distant site scores varied
temporally, grouping with downstream sites in
winter and spring, and upstream sites in sum-
mer and fall (Fig. 2). Axis 2 was negatively cor-
related with current velocity and dissolved oxy-
gen and positively correlated with water depth
and chloride (Table 5). Axis 2 thus represented
a pattern of fish assemblage structure along a
gradient from lotic habitat downstream from
dams to lentic habitat upstream from dams.
Correlations of fish taxa with Axis 2 differed
between riffle species, predominant at down-
stream sites, and pool species, prevalent at up-
stream sites (Tables 1, 3). With the exception of
545GILLETTE ET AL.—RIVERINE FISH ASSEMBLAGES
Fig. 2. Plot of CA 1 vs. CA 2 scores by season for
collections from the Neosho River, KS, 2000–01,
grouped by impoundment treatment. Seasons were
defined monthly as Winter (December–Februar y),
Spring (March–May), Summer (June–August), and
Fall (September–November).
Fig. 3. Plots of monthly mean and standard devi-
ation collection CA 1 scores. Months not differing sig-
nificantly, as determined by Tukey-Kramer tests on CA
1 scores, share lowercase letters. Means for which sam-
ple sizes were too small to calculate least square
means used in Tukey-Kramer tests are denoted by an
asterisk.
T
ABLE
5. P
EARSON
’
S
C
ORRELATION
C
OEFFICIENT OF
S
ELECTED
E
NVIRONMENTAL
V
ARIABLES WITH
CA 1
AND
CA 2 C
OLLECTION
S
CORES
. Correlations significant at
a5
0.05 are denoted by an asterisk, and those sig-
nificant at
a5
0.01 by two asterisks.
CA 1 CA 2
Percent substrate composition
Sand
2
0.333** 0.006
Cobble
2
0.030
2
0.041
Bedrock 0.197
2
0.209
Other microhabitat variables
Water depth
2
0.331** 0.344**
Water flow velocity
2
0.213*
2
0.558**
Water chemistry variables
Dissolved oxygen 0.540**
2
0.252*
Chloride 0.021 0.249*
Other variables
Chlorophyll a
2
0.223* 0.182
Water temperature
2
0.638** 0.086
River kilometer
2
0.182
2
0.214
River discharge
2
0.471**
2
0.172
C. lutrensis, most strong negative correlates of
Axis 2 were benthic riffle fishes, such as P. phox-
ocephala,P. mirabilis,C. anomalum, and P. capro-
des. Strong positive correlates included midwa-
ter species most abundant in slow waters of up-
stream sites, such as P. notatus juveniles, Lepomis
spp., and N. buchanani adults (Table 1).
Differences in seasonal abundance patterns
among species led to the significant temporal
and interaction effects on Axis 2 scores (Table
3). Several positively-correlated taxa (N. buch-
anani adults, P. notatus juveniles, and Lepomis hu-
milis juveniles) were more abundant, and neg-
atively-correlated taxa (C. lutrensis adults, P.
phoxocephala adults, and P. mirabilis adults) less
abundant, at upstream sites than at downstream
sites in all seasons (Table 3). Abundance at dis-
tant sites, however, varied seasonally (Fig. 2). As
a consequence of these temporal changes in
fish species composition, Axis 2 scores for dis-
tant sites were relatively higher in summer and
fall than in winter and spring, leading to the
significant interaction. Several taxa, including
N. buchanani adults, P. vigilax juveniles, and G.
affinis, loaded strongly on both Axis 1 and Axis
2, and thus were important in defining both
temporal and impoundment gradients. The fact
that these taxa were strongly associated with len-
tic habitat upstream from dams and also varied
temporally in abundance likely contributed to
the significant month effect on Axis 2 scores.
D
ISCUSSION
This study indicated that low-head dams can
influence structure of small-bodied fish assem-
546 COPEIA, 2005, NO. 3
blages in shallow waters of rivers via habitat al-
teration. Sites upstream from dams were deep-
est with slow-flow velocities and high siltation
levels, with fish assemblages characterized by a
high abundance of lentic habitat fishes. Sites
downstream from low-head dams were shallow-
est, with scoured substrata including bedrock,
and low levels of silt accumulation. Fish assem-
blages at these sites showed a higher abundance
of riffle species commonly found in shallow,
high-current velocity habitats. Fish assemblages
intermediate to these two extremes occurred at
sites distant from low-head dams. This pattern
of upstream and downstream habitat alteration
is similar to that shown for larger dams (e.g.,
Kondolf, 1997; Camargo and Voelz, 1998; Ben-
nett et al., 2002), but ours is one of the first
studies to document these patterns of habitat
alteration and associated fish assemblage differ-
ences in a river impounded by low-head dams.
Our results differed slightly from those of
previous investigators studying fish assemblages
in systems with low-head dams. Raborn and
Schramm (2003) and Dodd et al. (2003)
showed differences in habitat, but not fish as-
semblages, between dammed and free-flowing
streams and stream segments. The discrepancy
between our results and theirs is likely due to
spatial scale; we compared small sites within a
river, as opposed to stream reaches or entire
streams. Our sites were shorter than those of
the above investigators and spaced adjacent to
multiple impoundments, allowing detection of
these smaller-scale alterations of the fish assem-
blage. In rivers impounded by large dams, fish
assemblages can be influenced by impound-
ments for many kilometers downstream (Kin-
solving and Bain, 1993). Effects of low-head
dams, however, appear to be more localized, re-
stricted to habitat alteration immediately up-
stream and downstream. In addition, low-head
dams such as those in our study do not severely
alter river temperature and discharge as hydro-
electric dams do (Kinsolving and Bain, 1993;
Clarkson and Childs, 2000). This may explain
the high abundance of benthic fishes we ob-
served downstream from dams, as compared to
the low abundance of these fishes shown by
Travnichek and Maceina (1994) downstream of
a hydroelectric dam. Likewise, we did not find
the pattern of decreased upstream species rich-
ness shown by Porto et al. (1999); rather, fish
assemblage structure followed repeated gradi-
ents of lentic habitat upstream from dams to
lotic habitat downstream. This result is not sur-
prising, given that the species we collected in
the Neosho River are not migratory, thus elim-
inating the need to cross these barriers for spe-
cies to persist. In addition, the Neosho River is
a much larger system than the Great Lakes trib-
utaries studied by Porto et al. (1999), perhaps
limiting downstream transport of fishes and
providing sufficient habitat to maintain fish
populations upstream. Despite these differences
with previous studies, our results do show that
low-head dams can produce noticeable changes
in the spatial pattern of lotic fish assemblages.
Study sites also exhibited a great deal of tem-
poral variation in assemblage structure, as
shown by CA Axis 1 scores. Because CA calcu-
lates axes of decreasing ecological significance
(Gauch, 1982), it may be inferred that assem-
blage patterns associated with Axis 1 were stron-
ger than those associated with Axis 2. Complete
faunal turnover typically occurs across an axis
length of 4 standard deviations (Gauch, 1982);
thus, Axis 1’s length of 5.6 standard deviations
represented a strong temporal pattern. The
temporal nature of Axis 1 is confirmed by
strong correlations of axis scores with environ-
mental variables that vary seasonally, such as wa-
ter temperature (Table 5). A high degree of
overlap among multiple comparison groupings
indicates that seasonal fish assemblages were
not mutually exclusive, but rather components
of a gradual assemblage shift over the study
year. These results are consistent with Gelwick’s
(1990) conclusion that lotic fish assemblages in
shallow water show a great deal of temporal var-
iation. Separation of species into juvenile and
adult taxa could inherently bias our study to-
wards temporal variation because of natural
processes such as recruitment. However, a par-
allel analysis on species only also showed assem-
blage variation to be greater temporally than
spatially (D. P. Gillette, 2002, unpubl. data).
This supports the conclusion that, at least at the
spatial scale of the present study, temporal pat-
terns of shallow-water fish assemblages in the
Neosho River, as measured by position in mul-
tivariate space, are stronger than spatial pat-
terns.
Temporal assemblage variation came from
two sources: fish life history processes and as-
semblage responses to changing abiotic condi-
tions. As an example of the former, N. buchan-
ani was absent from our study sites until early
summer, when it occurred in great numbers.
This was likely a spawning migration from near-
by pools; Pflieger (1997) stated that this pool
species spawns over riffles from late April
through August, dates corroborated in Kansas
by Cross and Collins (1995). Reproduction of E.
spectabile,P. phoxocephala,P. caprodes,P. mirabilis,
and C. anomalum also changed assemblage com-
position through an influx of juveniles persist-
547GILLETTE ET AL.—RIVERINE FISH ASSEMBLAGES
ing from June through September. In addition
to these patterns, many species also declined
greatly in abundance or were absent during win-
ter. This pattern appears to be unrelated to life
history events, because all of these species
spawn from late spring through summer in Kan-
sas (Cross and Collins, 1995) and were present
at our study sites during both early spring and
late fall. Rather, this pattern is likely due to a
sharp drop in water temperature from Novem-
ber to December that, coupled with shallower
river depths in winter, caused these species to
vacate gravel bars and retreat to nearby pools.
With the exception of N. flavus, all of these spe-
cies have been shown to inhabit pools at various
times. Noturus flavus spawns in pools with mod-
erate current in Kansas (Cross and Collins,
1995), so it may also be able to use pool habitat
when water temperatures on shallow gravel bars
become too cold. After water temperatures rose
sharply from March to April, these species re-
turned to gravel bars. The fact that adults of
many species survived the winter to spawn, but
were not collected on gravel bars during winter,
suggests that deeper water may play a major role
in providing winter refugia for species that fre-
quent gravel bars in warmer months. This con-
clusion supports recent conceptual models of
stream fish ecology emphasizing the spatial ar-
rangement of habitat patches used by fishes un-
der varying abiotic conditions and during dif-
ferent life history stages (Schlosser, 1991, 1995;
Fausch et al., 2002).
Given the high degree of habitat variability,
variable sampling efficiency among our study
sites cannot be ruled out. Although there are
no published accounts of sampling efficiency
for the kick-set methodology we employed, Pe-
terson et al. (2004) showed that estimation of
salmonid abundance by multipass electrofishing
varied with stream area and substrate composi-
tion. As mentioned in Materials and Methods,
we eliminated two collections because flow ve-
locity was insufficient to allow effective sam-
pling. However, water depth, flow velocity, and
substrate composition varied among sites and
months during our study, possibly resulting in
variable sampling efficiency.
Few studies have examined spatial patterns of
fish assemblage structure on scales large
enough to assess assemblage response to mul-
tiple impoundments. Reyjol et al. (2001)
showed that flow alteration by hydroelectric im-
poundments along a salmoniform-cypriniform
transitional gradient caused an oscillation in
dominant taxa corresponding to alterations in
current velocity. With few free-flowing river sys-
tems remaining in the world, other situations
similar to that in the Neosho River likely exist
where multiple impoundments affect the spatial
pattern of riverine fish assemblages via localized
habitat alteration. Effective conservation of
these lotic systems and their biota requires
knowledge of the spatiotemporal structure of
fish assemblages in response to such alterations.
A
CKNOWLEDGMENTS
We thank B. Chance, J. Dean, L. Freeman, B.
Harkins, J. Howard, S. Sherraden, and I. Singh
for assistance in the field. River access was gen-
erously provided by Mr. and Mrs. W. Leffler, Mr.
and Mrs. P. Matile, Mrs. L. Schlessener, Mr. G.
Gulde, the City of Emporia, and Emporia State
University (ESU) Natural Areas. ArcView soft-
ware assistance was provided by R. Sleezer.
Throughout the course of this study, D. Zelmer,
L. Scott, and D. Moore provided valuable com-
ments, and J. Mendoza assisted with statistical
analysis. Laboratory assistance was provided by
S. Olson, B. Lakish, J. Albers, J. Fairchild, C.
Witte, and A. Allert; S. Gillette helped with data
entry and management. We thank W. Matthews
for critically reviewing an earlier version of the
manuscript. Funding for this study was provided
by a Faculty Research and Creativity Grant and
a Graduate Student Research Grant from ESU,
and by the U.S. Geological Survey, Department
of the Interior, under USGS Cooperative Agree-
ment No. 00CRAG0025. All fishes were collect-
ed under Kansas Department of Wildlife and
Parks Scientific Collector’s Permits SC-065–2000
(2000) and SC-033–2001 (2001), issued to DRE.
This study was conducted in accordance with
ESU Animal Care and Use Committee guide-
lines.
L
ITERATURE
C
ITED
A
GRESTI
,A.
AND
B. F
INLAY
. 1997. Statistical Methods
for the Social Sciences. Prentice Hall, Upper Sad-
dle River, New Jersey.
B
AILEY
, R. G. 1983. Delineation of ecosystem regions.
Environ. Manage. 7:365–373.
B
AIN
, M. B. 1999. Substrate, p. 95–100. In: Aquatic
Habitat Assessment: Common Methods. M. B. Bain
and N. J. Stevenson (eds.). American Fisheries So-
ciety, Bethesda, Maryland.
B
ALON
,E.K.,
AND
D. J. S
TEWART
. 1983. Fish assem-
blages in a river with unusual gradient (Luongo,
Africa–Zaire system), reflections on river zonation,
and description of another new species. Environ.
Biol. Fish 9:225–252.
B
AXTER
, R. M. 1977. Environmental effects of dams
and impoundments. Annu. Rev. Ecol. Syst. 8:255–
283.
B
ENNETT
, S. J., C. M. C
OOPER
,J.C.R
ITCHIE
,J.A.D
UN
-
BAR
,P.M.A
LLEN
,L.W.C
ALDWELL
,
AND
T. M.
548 COPEIA, 2005, NO. 3
M
CGEE
. 2002. Assessing sedimentation issues within
aging flood control reservoirs in Oklahoma. J. Am.
Water Resour. As. 38:1307–1322.
B
ONNER
,T.H.,
AND
G. R. W
ILDE
. 2000. Changes in
the Canadian River fish assemblage associated with
reservoir construction. J. Freshwater Ecol. 15:189–
198.
C
AMARGO
, J. A.,
AND
N. J. V
OELZ
. 1998. Biotic and abi-
otic changes along the recovery gradient of two im-
pounded rivers with different impoundment use.
Environ. Monit. Assess. 50:143–158.
C
ARLANDER
, K. D. 1969. Handbook of Freshwater Fish-
ery Biology. Volume 1. Iowa State University Press,
Ames, Iowa.
———. 1977. Handbook of Freshwater Fishery Biol-
ogy. Volume 2. Iowa State University Press, Ames,
Iowa.
———. 1997. Handbook of Freshwater Fishery Biol-
ogy. Volume 3. Iowa State University Press, Ames,
Iowa.
C
LARKSON
,R.W.,
AND
M. R. C
HILDS
. 2000. Tempera-
ture effects of hypolimnial-release dams on early
life stages of Colorado River Basin big-river fishes.
Copeia 2000:402–412.
C
ROSS
, F. B.,
AND
J. T. C
OLLINS
. 1995. Fishes in Kansas.
2
nd
Edition, Revised. University Press of Kansas,
Lawrence, Kansas.
D
EACON
, J. E. 1961. Fish populations, following a
drought, in the Neosho and Marais de Cygnes riv-
ers of Kansas. Univ. KS. Pub., Mus. Nat. Hist. 13:
359–427.
D
ODD
, H. R., D. B. H
AYES
,J.R.B
AYLISS
,L.M.C
ARL
,
J. D. G
OLDSTEIN
,R.L.M
CLAUGHLIN
,D.L.G.N
OAKES
,
L. M. P
ORTO
,
AND
M. L. J
ONES
. 2003. Low-head sea
lamprey barrier effects on stream habitat and fish
communities in the Great Lakes basin. J. Great
Lakes Res. 29:386–402 Suppl. 1.
D
YNESIUS
, M.,
AND
C. N
ILSSON
. 1994. Fragmentation
and flow regulation of river systems in the northern
third of the world. Science 266:753–762.
E
BY
, L. A., W. F. F
AGAN
,
AND
W. L. M
INCKLEY
. 2003.
Variability and dynamics of a desert stream com-
munity. Ecol. Appl. 13:1566–1579.
E
DDS
, D. R. 1993. Fish assemblage structure and en-
vironmental correlates in Nepal’s Gandaki River.
Copeia 1993:48–60.
F
AUSCH
, K. D., C. E. T
ORGERSON
,C.V.B
AXTER
,
AND
H. W. L
I
. 2002. Landscapes to riverscapes: bridging
the gap between research and conservation of
stream fishes. Bioscience 52:483–498.
F
USELIER
, L.,
AND
D. R. E
DDS
. 1996. Seasonal variation
of riffle and pool fish assemblages in a short miti-
gated stream reach. Southwest. Nat. 41:299–306.
G
AUCH
, H. G., J
R
. 1982. Multivariate analysis in com-
munity ecology. Cambridge University Press, Cam-
bridge.
G
ELWICK
, F. P. 1990. Longitudinal and temporal com-
parison of riffle and pool fish assemblages in a
northeastern Oklahoma Ozark stream. Copeia
1990:1072–1082.
G
IDO
,K.B.,
AND
D. L. P
ROPST
. 1999. Habitat use and
association of native and nonnative fishes in the
San Juan River, New Mexico and Utah. Ibid. 1999:
321–332.
G
ODINHO
, F. N.,
AND
M. T. F
ERREIRA
. 1998. The rela-
tive influences of exotic species and environmental
factors on an Iberian native fish community. Envi-
ron. Biol. Fish. 51:41–51.
H
OLMQUIST
, J. G., J. M. S
CHMIDT
-G
ENGENBACH
,
AND
B.
B. Y
OSHIOKA
. 1998. High dams and marine-fresh-
water linkages: effects on native and introduced
fauna in the Caribbean. Conserv. Biol. 12:621–630.
K
INSOLVING
,A.D.,
AND
M. B. B
AIN
. 1993. Fish assem-
blage recovery along a riverine disturbance gradi-
ent. Ecol. Appl. 3:531–544.
K
ONDOLF
, G. M. 1997. Hungry water: effects of dams
and gravel mining on river channels. Environ. Man-
age. 21:533–551.
M
ARCH
, J. G., J. P. B
ENSTEAD
,C.M.P
RINGLE
,
AND
F. N.
S
CATENA
. 2003. Damming tropical island streams:
problems, solutions, and alternatives. Bioscience
53:1069–1078.
M
ARCHETTI
,M.P.,
AND
P. B. M
OYLE
. 2001. Effects of
flow regime on fish assemblages in a regulated Cal-
ifornia stream. Ecol. Appl. 11:530–539.
M
ARET
, T. R., C. T. R
OBINSON
,
AND
G. W. M
INSHALL
.
1997. Fish assemblages and environmental corre-
lates in least-disturbed streams of the upper Snake
River basin. T. Am Fish. Soc. 126:200–216.
M
ATTHEWS
, W. J. 1990. Spatial and temporal variation
in fishes of riffle habitats: a comparison of analyti-
cal approaches for the Roanoke River. Am. Midl.
Nat. 124:31–45.
M
INCKLEY
, W. L., P. C. M
ARSH
,J.E.D
EACON
,T.E.
D
OWLING
,P.W.H
ENDRICK
,W.J.M
ATTHEWS
,
AND
G.
M
UELLER
. 2003. A conservation plan for native fish-
es of the lower Colorado River. Bioscience 53:219–
234.
M
ULLNER
, S. A., W. A. H
UBERT
,
AND
T. A. W
ESCHE
.
2000. Visually estimating substrate composition at
potential spawning sites for trout in mountain
streams. J. Freshwater Ecol. 15:199–207.
P
ETERSON
, J. T., R. F. T
HUROW
,
AND
J. W. G
UZEVITCH
.
2004. An evaluation of multipass electrofishing for
estimating the abundance of stream-dwelling sal-
monids. T. Am. Fish. Soc. 133:462–475.
P
ETTS
, G. E. 1984. Impounded Rivers. John Wiley,
Chichester, U.K.
P
FLIEGER
, W. L. 1997. The Fishes of Missouri. Revised
Edition. Missouri Department of Conservation, Jef-
ferson City, Missouri.
P
OFF
, N. L., J. D. A
LLAN
,M.B.B
AIN
,J.R.K
ARR
,K.L.
P
RESTEGAARD
,B.D.R
ICHTER
,R.E.S
PARKS
,
AND
J. C.
S
TROMBERG
. 1997. The natural flow regime: a par-
adigm for river conservation and restoration. Bio-
science 47:769–784.
P
ORTO
, L. M., R. L. M
CLAUGHLIN
,
AND
D. L. G. N
OAKES
.
1999. Low-head barrier dams restrict the move-
ments of fishes in two Lake Ontario streams. N.
Am. J. Fish. Manage. 19:1028–1036.
Q
UINN
,J.W.,
AND
T. J. K
WAK
. 2003. Fish assemblage
changes in an Ozark river after impoundment: a
long-term perspective. T. Am. Fish. Soc. 132:110–
119.
R
ABORN
,S.W.,
AND
H. L. S
CHRAMM
. 2003. Fish assem-
blage response to recent mitigation of a channel-
ized warmwater stream. River Res. Appl. 19:289–
301.
549GILLETTE ET AL.—RIVERINE FISH ASSEMBLAGES
R
EYES
G
AVILAN
, F. G., R. G
ARRIDO
,A.G.N
ICIENZA
,M.
M. T
OLEDO
,
AND
F. B
RANA
. 1996. Fish community
variation along physical gradients in short streams
of northern Spain and the disruptive effect of
dams. Hydrobiologia 321:155–163.
R
EYJOL
, Y., P. L
IM
,F.D
AUBA
,P.B
ARAN
,
AND
A. B
ELAUD
.
2001. Role of temperature and flow regulation on
the Salmoniform-Cypriniform transition. Arch. Hy-
drobiol. 152:567–582.
R
ICHTER
, B. D., D. P. B
RAUN
,M.A.M
ENDELSON
,
AND
L. L. M
ASTER
. 1997. Threats to imperiled freshwater
fauna. Conserv. Biol. 11:1081–1093.
S
CHLOSSER
, I. J. 1982. Fish community structure and
function along two habitat gradients in a headwater
stream. Ecol. Monogr. 52:395–414.
———. 1991. Stream fish ecology: a landscape per-
spective. Bioscience 41:704–712.
———. 1995. Critical landscape processes that influ-
ence fish population dynamics in headwater
streams. Hydrobiologia 303:71–81.
S
HUMAN
, J. R. 1995. Environmental considerations for
assessing dam removal alternatives for river resto-
ration. Regul. River. 11:249–261.
T
RAVNICHEK
, V. H.,
AND
M. J. M
ACEINA
. 1994. Com-
parison of flow regulation effects on fish assem-
blages in shallow and deep-water habitats in the Tal-
lapoosa River, Alabama. J. Freshwater Ecol. 9:207–
216.
W
ILDHABER
, M. L., A. L. A
LLERT
,
AND
C. J. S
CHMITT
.
1999. Potential effects of interspecific competition
on Neosho madtom (Noturus placidus) populations.
Ibid. 14:19–30.
(DPG, JST, DRE) D
EPARTMENT OF
B
IOLOGICAL
S
CIENCES
,E
MPORIA
S
TATE
U
NIVERSITY
,E
MPO
-
RIA
,K
ANSAS
, 66801; (MLW) C
OLUMBIA
E
NVI
-
RONMENTAL
R
ESEARCH
C
ENTER
,C
OLUMBIA
,
M
ISSOURI
65201. P
RESENT
A
DDRESSES
: (DPG)
D
EPARTMENT OF
Z
OOLOGY
,U
NIVERSITY OF
O
KLAHOMA
,N
ORMAN
,O
KLAHOMA
73019;
( JST) C
ENTER FOR
B
IODIVERSITY
,I
LLINOIS
N
ATURAL
H
ISTORY
S
URVEY
,C
HAMPAIGN
,I
LLI
-
NOIS
61820. E-mail: (DPG) dgillette@ou.edu.
Send reprint requests to DPG. Submitted: 30
April 2004. Accepted: 6 April 2005. Section
editor: C. M. Taylor.