Hydrobiologia ***: *** ***, ****.
**** ****** ******** **********. ******* in the Netherlands.
Response of unionid mussels to dam removal in Koshkonong Creek, Wisconsin
(USA)
Suresh A. Sethi, Andrew R. Selle, Martin W. Doyle, Emily H. Stanley* & Helen E. Kitchel
Center for Limnology, University of Wisconsin, 680 N. Park St., Madison, WI 53706, U.S.A.
(*Author for correspondence: Tel.: +1-608-***-****, Fax: +1-608-***-****, E-mail: abqk2j@r.postjobfree.com)
Received 25 September 2003; in revised form 31 January 2004; accepted 9 February 2004
Key words: mussels, reservoir, restoration, sediments, siltation, small dams
Abstract
Dam removal is a potentially powerful tool for restoring riverine habitats and communities. However, the
e ectiveness of this tool is unknown because published data on the e ects of dam removal on in-stream
biota are lacking. We investigated the e ects of a small dam removal on unionid mussels in Koshkonong
Creek, Wisconsin (USA). Removal of the dam led to mortality both within the former impoundment and in
downstream reaches. Within the former reservoir, mortality rates were extremely high (95%) due to des-
iccation and exposure. Mussel densities in a bed 0.5 km downstream from the dam declined from
3.80 0.56 mussels m)2 in fall 2000 immediately after dam removal to 2.60 0.48 mussels m)2 by
summer 2003. One rare species, Quadrula pustulosa, was lost from community. Mortality of mussels buried
in deposited silt was also observed at a site 1.7 km below the dam. Silt and sand increased from 16.8 and
1.1% of total area sampled in fall 2000 to 30.4 and 15.9%, respectively, in summer 2003. Total suspended
sediment concentrations in the water column were always higher downstream from the reservoir than
upstream, suggesting that transport and deposition of reservoir sediments likely contributed to downstream
mussel mortality. Thus, while bene ts of the dam removal included sh passage and restoration of lotic
habitats in the former millpond, these changes were brought about at some cost to the local mussel
community. Pre-removal assessments of potential ecological impacts of dam removal and appropriate
mitigation e orts should be included in the dam removal process to reduce short-term negative ecological
e ects of this restoration action.
Introduction
North America is home to the richest assemblage Dams have a range of negative e ects on
of freshwater mussels in the world (Watters, 2001). freshwater mussels, with altered sediment cycles
In recent times, this diverse fauna has come under and restrictions in host sh distributions the most
considerable pressure from human activity. Over dominant e ects (Watters, 1992; Box & Mossa,
40% of the North American freshwater mussel 1999). Dams alter the ow regimes of rivers and
species are in need of management, 62 are listed as streams, arresting natural geomorphic processes
federally endangered, and 19 are presumed extinct and causing sediments to collect in both upstream
(USFWS, 1999; Vaughn & Taylor, 1999). Broad- and downstream reaches (Galay, 1983; Po et al.,
scale damming of riverine habitats has been 1997; Stanley & Doyle, 2003). This shift from lotic
implicated as a major factor in the decline of to lentic habitat can smother sediment-intolerant
freshwater mussels (Williams et al., 1992; Bogan, mussel species and reduce available mussel habitat
1993; Vaughn & Taylor, 1999; Bednarek, 2001). (Houp, 1993; Box & Mossa, 1999). High levels of
158
suspended solids immediately below some dams are released downstream (Doyle et al., 2003). The
can reduce mussels planktonic food supply and rate at which these sediments move through the
clog ltering apparatuses (Kat, 1982). Unionid system depends on the amount and particle size of
mussels employ a parasitic glochidial larval stage stored sediment, the ow regime of the system,
where larvae attach themselves to host sh gills stabilization e orts within the impoundment, and
and ns (see Wachtler et al., 2001 for a compre- local precipitation (Pizzuto, 2002). Mussels
hensive description of the unionid larval biology). downstream from a removal site may be able to
Restrictions in host sh distributions can limit the cope with short bursts of sediment if it is quickly
distribution of adult mussels (Watters, 1992, 1996; transported through a system, however, persistent
Hornbach, 2001). Furthermore, isolation of small siltation can cause mortality (Vannote & Minshall,
populations of mussels between dammed river 1982; Houp, 1993; Box & Mossa, 1999). To study
segments may reduce numbers below minimum the e ects of dam removal on freshwater mussels,
viable populations leading to increased likelihood we monitored communities in Koshkonong Creek,
of local extinctions (Bauer & Wachtler, 2001). south central Wisconsin (USA) for 3 years fol-
Due to the deleterious e ects dams have on lowing removal of the Rockdale Dam to track
mussels and other lotic communities (Petts, 1984; changes in species composition and density.
Vaughn & Taylor, 1999), it is reasonable to
assume that dam removal is a viable tool for
restoring riverine communities and ecosystems. Materials and methods
However, little research exists on the response of
in-stream biota to dam removal and the restora- Study site
tion e ectiveness of dam removal is largely
untested. We investigated the e ects of dam re- Koshkonong Creek in south central Wisconsin is
a fth-order (average width 10 m) warmwater
moval on freshwater mussels, a previously
stream draining a 360-km2 catchment. The basin
unconsidered component of lotic systems in dam
removal studies. Dam removals in mid-sized lies in a low-relief glacial outwash plain dominated
streams in southern Wisconsin have shown a rel- by agricultural land use. The Rockdale dam was
atively quick recovery to a lotic habitat, with rst constructed as a rock and timber crib in 1848
desirable sh species numbers increasing (i.e., to run a grain mill, and then was converted to a
smallmouth bass; Kanehl et al., 1997) and lotic concrete structure in 1887. The most recent ver-
assemblages of benthic invertebrates returning sion, a 3.3-m high concrete run-of-river dam with a
within 1 year of dam removal (Stanley et al., xed crest spillway, was constructed ca. 1925 and
2002). These reports yield a positive outlook for created a 42 45-ha impoundment in 2000. Average
biotic response to dam removal. However, the impoundment depth decreased from 2 3 m in the
organisms sampled in these studies were either 1960s to 1.5 m at the time of removal (Doyle et al.,
highly mobile ( sh) or had short life cycles 2003). The Wisconsin Department of Natural
(macroinvertebrates) and thus have a high capac- Resources breached the Rockdale dam on Sep-
ity to recolonize new areas following habitat tember 12, 2000. Dewatering was rapid and the
modi cation. It is likely mussels will show a dif- reservoir emptied in approximately 36 h, exposing
ferent response to dam removal due to their slow extensive amounts of sediment upstream from the
growth and sedentary nature. dam. Physical changes to the reservoir and
In the long run, mussels may be expected to downstream channel caused by the removal are
bene t from dam removal as increases in the described in detail by Doyle et al. (2003).
numbers of sh and continuity of sh populations Mussels were surveyed at three sites within the
aid mussel distribution, and additional mussel vicinity of the dam: upstream from the dam in the
habitat is created by converting upstream lentic silt-dominated former reservoir, 0.5 km down-
habitats to lotic habitats through the draining of stream from the dam in a ri e area characterized
the impoundment. In the short run, the picture is by a mix of cobble, gravel, and silt substrate, and
less clear. During a dam removal, considerable 1.7 km downstream from the dam in a low-gra-
amounts of sediment stored in the impoundment dient run with predominantly silty substrate.
159
stretch of the creek for mussel density and species
Data collection and analysis
composition. As suggested by Obermeyer (1998),
Following dam removal, we observed large num- quadrats were used over a timed search e ort to
bers of stranded mussels within the former reser- gain higher precision in density measurements.
Quadrats of 0.25 m2 were sampled on 20 transects
voir. Unfortunately, attempts to quantify this
mortality immediately following removal were not that were spaced at 5-m intervals. The study reach
possible because the instability of the deep, satu- was sampled on October 22, 2000 using two
quadrats per transect (n 40), on April 27, 2001
rated reservoir sediments made sampling unsafe.
When the reservoir was rst accessible in May using three quadrats per transect for transects
2001, we surveyed 11 transects spaced at 100-m spanning the widest part of the channel and two
intervals perpendicular to the channel ow for those at narrower parts of the channel
(n 49), and again on June 16, 2003, using three
(Fig. 1), identifying any living or dead mussels
quadrats per transect (n 60). Quadrats were
within 1.0 m on either side of the transect. No
attempts were made to nd mussels buried beneath randomized along transects, and rerandomizing
the sediment surface as sampling could not be each sampling session. Each quadrat was exca-
consistent. Shells of dead mussels were brought vated until aggregated substrate, which would
back to the lab to verify eld identi cations. All prevent mussel burrowing was reached, usually a
mussel identi cation was conducted using minimum depth of 10 cm. Specimens were identi-
nomenclature in Stern (1990). We used transect ed to species, and only living specimens were
data to estimate the total mortality observed from recorded. Cumulative plots of species richness
dewatering by calculating mussel abundance over versus number of samples reached asymptotes for
the sampled area then extrapolating to the total all three sampling dates, indicating that our sam-
area of the former impoundment. pling e ort was su cient to capture most species
Downstream e ects of dam removal were in the study reach. In fall 2000 and summer 2003
assessed by monitoring the status of a mussel bed samplings, we recorded the percent area of silt and
0.5 km below the dam. We sampled a 100-m sand within each quadrat. Changes in substrate
Figure 1. Aerial view of the Rockdale Millpond in May 2001, 8 months after dam removal. Flow is from left to right. Black lines
perpendicular to the channel indicate transect locations for the upstream mussel survey. Section A of the former reservoir is char-
acterized by unconsolidated silts in a wide, shallow channel (width 20 50 m, depth 0.3 m). Area B is separated from A by a
knickpoint, which delimits the upstream extent of channel formation. The channel below the knickpoint is single, narrow, discrete,
channel (width 8.0 m, depth 1.0 m).
160
composition were estimated by summing the total ve species were collected in the former reservoir
percentage of sampled area in silt or sand for each (Fig. 2). Taxa found in the reservoir were sedi-
of the two respective sampling dates. Di erences ment-tolerant species, demonstrated by the most
among dates were also assessed by considering abundant species collected, Anodonta grandis
changes in the number of plots with 0, 5 20, 21 (Table 1). Only two living specimens were col-
40, 41 60, 61 80, and 80 100% silt or sand via a lected, and these individuals were found in a small
v2-test. channel near the head of the reservoir that con-
Relative species abundances, Simpson s diver- tinued to hold water after removal. Using the
sity index (Simpson, 1949), and Pielou s evenness degree of shell weathering and the presence of
index (Pielou, 1966) were calculated to character-
ize and track the population composition through
30
time. Potential di erences in mussel densities
Number of specimens collected
among sample dates were examined using a 25
repeated measures ANOVA to allow for possible
non-independence from repeat measurements on 20
individuals across sampling dates.
15
In addition to quantitative sampling in the
former reservoir and 0.5 km downstream of the
10
dam, we conducted a qualitative survey in May
2001 to observe possible e ects of conspicuous silt 5
accumulations in a reach 1.7 km downstream of
the dam site. We conducted a 40-min search in a 0
A.grandis L.complanata L.siliquoidea Toxolasma Leptodea
30-m reach of river, making sweeps perpendicular parvus fragilis
to the channel and feeling for mussels by hand.
Figure 2. Mussels collected in the former reservoir of Kosh-
This survey was in conjunction with a separate konong Creek, WI, May 2001.
study monitoring downstream sediment deposition
(see Doyle et al., 2003).
Table 1. Freshwater mussels observed in upstream and down-
Total suspended solids (TSS) were monitored
stream reaches surrounding the former Rockdale dam on the
monthly or bimonthly and opportunistically
Koshkonong Creek, WI and their reported tolerance to
between regular sampling dates throughout the 3- sedimentation*
year period upstream of the reservoir and 0.4 km
Species Location Sediment Sediment
downstream from the dam site. Three replicate
tolerant intolerant
grab samples were collected 10 25 cm below the
water surface at points distributed on quarter, one-
half, and three-quarters of the way across the Both B, S, PH,
A. grandis
channel. TSS concentrations were determined Ha, T, BS
gravimetrically following ltration through a pre- D/S BS H, MB, S
Fusconaia ava
weighed, pre-ashed 1.0 lm im glass ber lter and D/S BS
L. ovata
drying at 60 C for 48 h. Di erences in TSS Both S H
L. siliquoidea
entering and leaving the reservoir were evaluated Both PH BS
L. complanata
using a Wilcoxon signed rank test (data were non- D/S B, PH BS
L. fragilis
normal even after transformation). D/S H, BS
L. recta
Strophitus undulatus D/S S H
U/S B, PH, BS, S
T. parvus
Results
*U/S = upstream, within the former reservoir, D/S = 0.5 km
downstream of the former impoundment, B = Bates (1962),
Within the former reservoir
H = Houp (1993), MB = Marking & Bills (1980), S = Stein
(1972), Ha = Harman (1971), PH = Parmalee & Hughes
In May 2001, when sampling on the unstable (1993), T = Tucker (1998), BS = Blalock and Sickel (1996).
sediments was rst possible, 39 individuals from Table adapted from Box &Mossa (1999).
161
remnant tissue as indicators of time since mortal- Table 2. Diversity indices across three sampling dates 0.5 km
downstream the former Rockdale dam on Koshkonong Creek
ity, we ruled out that shells were relics of pre-re-
moval mortality. No live mussels were found Community measure Fall 2000 Spring Summer
during additional searches of the impoundment 2001 2003
beyond the established transects. These numbers
Pielou s evenness index 0.89 0.87 0.87
are an absolute minimum estimate of density
Simpson s index 0.18 0.19 0.21
immediately following the breach as mussels may
have burrowed into the reservoir sediments or
have been removed by predators (e.g., raccoons) in
the weeks following dam removal. The total pustulosa in August 2003 and no specimens were
transect area sampled was 3500 m2. Projecting the located, con rming that this was an extirpation
sampled mussel density throughout the reservoir, a and not a sampling e ect. Simpson s diversity
minimum estimate of total abundance and thus index and Pielou s evenness index changed little
total mortality in this case would be near 4700 across the sampling dates suggesting that any
individuals within the former impoundment. changes to the population were distributed equally
across all species (Table 2).
Downstream e ects of dam removal The results from the repeated measures ANO-
VA tests indicated a signi cant change in mussel
Eight species of mussels were identi ed in this densities through time (F 3.718, p 0.034), with
study reach (Fig. 3). All eight species are consid- mussel density decreasing from an average of 3.80
mussels m)2 in fall 2000 sampling immediately
ered generalist, and all are listed as widespread and
after dam removal, to 2.60 mussels m)2 by summer
common throughout the Midwest except Ligumia
recta, which is listed as widespread but uncommon 2003, approximately 3 years later (Fig. 4). Post hoc
(INHS, 2003). Species composition changed little least-squares di erences testing con rms that fall
among sample dates (Fig. 3), with two species, 2000 and spring 2001 mussel densities were not
Lampsilis ovata and Lasmigona complanata, dom- signi cantly di erent, but summer 2003 density
inating the population throughout the study. One was signi cantly less than for the fall 2000
locally rare species within the study population, (p 0.017) and spring 2001 (p 0.029) samplings.
Quadrula pustulosa, was not found during the Repeated measures ANOVAs were also run to test
summer 2003 sampling. An additional qualitative changes in density across sampling dates for each
search was conducted in the study reach for Q.
5
A A B
Fall 2000
1.5
4
Spring 2001
Species density (#/m2)
Total density (#/m2)
Summer 2003
3
1.0
2
0.5
1
0
*
0.0
Fall 2000 Spring 2001 Summer 2003
Figure 4. Total mussel density (+1 SE) during three sampling
dates 0.5 km downstream from the former dam site in Kosh-
Figure 3. Mussel densities (+1 SE) by species across sampling
konong Creek. Letters above the bars denote the presence/ab-
dates 0.5 km below the dam in Koshkonong Creek. *No
sence of signi cant di erences among dates (dates with the
Q. pustulosa individuals were found in the study reach for the
same letters are not signi cantly di erent).
summer 2003 sampling.
162
35
1
30
Percent of total area sampled
TSS (g/L)
Fall 2000
0.1
25
Summer 2003
20
0.01
15
0.001
30-60-901**-*** 360-***-***-*** 840 960
10
Days since breach
Figure 6. TSS concentration in streamwater entering (up-
5
stream: closed circles) and leaving (downstream: open triangles)
the former reservoir site in Koshkonong Creek following dam
0
removal in September 2000. Values are means 1 SE. Note
Silt Sand log scale of the y-axis. Arrows indicate dates of downstream
mussel surveys.
Figure 5. Change in the percentage ( 1 SE) of total area
sampled with silt and sand substrates, respectively.
L. siliquoidea, L. complanata, and Q. pustulosa.
The lack of shell weathering and the presence of
remnant tissue again suggested recent mortality.
of the respective species; however, no signi cant
results were found and power estimates were
extremely low.
Silt and sand cover increased dramatically Discussion
between the initial post dam removal sampling
We observed widespread mussel mortality in both
in fall 2000 and summer 2003 (Fig. 5), signifying
upstream and downstream reaches of the Rockdale
a movement of sediment from the former
dam following removal. Mortality within the res-
impoundment into the downstream study reach.
ervoir resulted from stranding, desiccation, and
Silt increased from 16.8% of total area sampled in
predation due to rapid dewatering. Below the dam,
fall 2000 to 30.4% in summer 2003, and sand in-
we observed more subtle and delayed e ects of the
creased from 1.1 to 15.9%. The percent of quadrats
removal; however, the data suggest that increased
with no silt or sand decreased over time, and the
reservoir-born sediment inputs were a contributing
fraction of plots with high silt and sand cover in-
factor to mussel declines. The assemblage located
creased signi cantly for but sediment types
(v2 33.03, p
p 0.028 for sand). uniform reduction in density across all species and
loss of the rarest species within the study reach,
TSS concentrations downstream from the dam
Q. pustulosa. This suggests a general reduction in
site were higher than concentrations entering the
habitat quality and increased stress on the mussel
reservoir on every sampling date (Fig. 6), thus
community that is likely due to the elevated con-
upstream and downstream sites were signi cantly
di erent throughout the study (W 1275, centrations of suspended sediments and/or sedi-
ment deposition that has persisted in the 3 years
p