2006 SOUTHEASTERN NATURALIST 5(1):31–52
Unionid Habitat and Assemblage Composition in
Coastal Plain Tributaries of Flint River (Georgia)
PAULA GAGNON1,5,*, WILLIAM MICHENER2, MARY FREEMAN3, AND JAYNE BRIM BOX4
Abstract - Effective conservation of mussels in streams of the lower Flint River
basin, southwest Georgia, requires more rigorous understanding of mussel-habitat
associations and factors shaping assemblage composition in stream reaches. We
surveyed mussels and habitat conditions at 46 locations, and used regression, correlation
and multivariate direct gradient analysis (Canonical Correspondence Analyses)
to identify species-habitat relationships and characteristic species-assemblage types
in Flint basin streams. Riparian wetland and catchment forest cover, average midchannel
depth, and drainage network position accounted for 49% of the variability in
mussel species richness, 36% of the variability in unionid abundance, and 32% of the
variability observed in Shannon-Wiener diversity across survey sites. Species were
grouped into four assemblage types based on their habitat associations: large-riverriffle
associates, slackwater associates, habitat generalists, and stream-run associates.
Results are broadly concordant with anecdotal reports of mussel-habitat relationships
and provide insight into the habitat conservation needs of mussels.
Introduction
Freshwater mussels are among the most imperiled of North American
fauna. About 70% of almost 300 native species are currently identified as
endangered, threatened, or sensitive (Neves et al. 1997, Williams et al.
1993), with most species experiencing habitat loss across their ranges.
Developing effective conservation and recovery strategies for unionids requires
knowledge of habitat needs for mussels and environmental factors
controlling distribution patterns. Prior studies indicate that unionid distribution
and abundance is related to physical, chemical, and biotic factors across
multiple scales (Bauer et al. 1991; Brim Box 1999; DiMaio and Corkum
1995; Haag and Warren 1998; Mamilton et al. 1997; Howard 1997; Morris
and Corkum 1996; Strayer 1983, 1993; Strayer and Ralley 1993; Strayer et
al. 1994; Vannote and Minshall 1982; Watters 1992, 1993). However, despite
substantial anecdotal information about mussel habitat preferences,
little empirical evidence currently exists to demonstrate links between habitat
and particular mussel species and assemblages.
The objective of this study was twofold. First, we sought to identify
patterns in species-assemblage composition across stream reaches in the
1Jones Ecological Research Center, Route 2, Box 2324, Newton, GA 31770. 2LTER
Network Office, Department of Biology, University of New Mexico, Albuquerque,
NM. 3US Geological Survey, Patuxent Wildlife Research Center, University of
Georgia, Athens, GA 30602 4Utah State University, UMC 5210, Logan, UT 84322.
5Current address - 6923 31st Street, Berwyn, IL 60402. *Corresponding address -
jmaluap@yahoo.com.
32 Southeastern Naturalist Vol. 5, No. 1
tributary streams of the Coastal Plain portion of the Flint River Basin (lower
FRB), and, if possible, characterize specific assemblage types based on the
habitat associations of mussels. Second, we aimed to identify multiple-scale
habitat conditions associated with the occurrence of freshwater mussel assemblages,
and mussel community richness and diversity within 100-m
stream segments in the lower FRB. We assessed habitat features at two
spatial scales. Mesoscale (reach-scale) habitat conditions included composite
measures of stream gradient, channel depth, hydrologic variability,
stream ion content, and substrate composition. Macrohabitat (catchmentscale)
variables included measures of stream size and drainage position
(catchment area, d-link magnitude), predominant catchment landcover, riparian
land use, and physiographic province type.
Methods
Study area and mussel surveys
Most of the streams in the lower FRB originate in the Fall Line Hills
physiographic province, and are characterized by sandy mud bottoms and
high turbidity. At the lower end of the drainage, streams flow into the
Dougherty Plain physiographic province, a Coastal Plain region underlain
by a shallow carbonate aquifer with direct links to surface water. Streams in
the Dougherty Plain frequently dissect carbonate rocks and are relatively
high in conductivity and alkalinity. Coastal Plain streams are generally
naturally high in fine sand substrates and are low gradient throughout the
entire system; large portions of the stream continuum, from headwaters to
confluences, may anastomose and flow through marshy, slackwater,
swampy areas.
In the lower FRB, agriculture and plantation forest are the predominant
land-cover types, but stream-drainage networks are buffered by extensive
forested floodplains (Houhoulis and Michener 2000, Lowrance et al. 1984),
and marked by low levels of urban impacts, high water quality (Golladay et
al. 2000), and relatively intact biotic communities. Most purported impacts
likely result from agricultural practices and are not associated with point
sources, except near urban areas.
Unionid communities derive from eastern Atlantic and western Gulf
drainages (Johnson 1970). The FRB once supported 29 unionid species,
seven of which were endemic to the Apalachicola drainage (Clench and
Turner 1956). Currently, four federally endangered or threatened species
persist in the lower FRB. Mussel communities within lower FRB streams
represent the most diverse assemblages in the Apalachicola-Chatahoochee-
Flint River Basin (Brim Box and Williams 2000). Although non-native
Dreissena spp. have not been found in the basin, exotic Corbicula spp. are
widespread and abundant.
During mid-June–late August, 1999, we conducted mussel surveys at 46
sites on 12 tributary streams in the lower FRB (Fig. 1). Survey sites were
2006 P. Gagnon, W. Michener, M. Freeman, and J. Brim Box 33
located near bridge crossings at regular intervals along the longitudinal
progression from headwaters to the Flint River confluence on each stream.
At each site, we sampled a 100-m segment of stream bed beginning at 100 m
and ending at 200 m upstream from the bridge. We sampled for mussels by
visually searching the substrate and by sieving surface sediment with our
fingers to a depth of about 5 cm. To standardize the amount of streambed
sampled among various-sized streams, the following methodology was em-
Figure 1. Locations (represented by circles and triangles) of mussel survey sites in
the lower Flint River Basin (Coastal Plain portion), southwest Georgia. Circles
indicate sites where endangered species were found.
34 Southeastern Naturalist Vol. 5, No. 1
ployed. In small streams (< 12 m wide), we surveyed the entire bed surface
within the 100-m survey reach. In large streams (> 12 m wide), we conducted
mussel searches along six transects placed parallel to stream flow
along the length of the stream reach. Transects were two meters wide and
evenly spaced across the width of the stream, with one transect in each bank.
In both stream-size classes, we conducted surveys only in the main
channel(s) of the stream; minor channels and ponded areas in floodplains
were not searched.
We removed live mussels from the substrate, identified them, and then
immediately returned them to the stream bottom. If more than 1000 individuals
of any species were found before reaching the end of the sample
area, we discontinued counting that species. Unionids were identified to
species level, except Elliptio complanata (Lightfoot) and Elliptio icterina
(Conrad), which were grouped together because of the difficulty in distinguishing
the two species in the field. Dead shells found during the
searches were collected, identified, and deposited at the Georgia Museum
of Natural History.
Mesohabitat data
Habitat factors selected for this study were demonstrated to influence
mussel distribution patterns elsewhere in North America (Bauer et al. 1991;
Brim Box 1999; DiMaio and Corkum 1995; Haag and Warren 1998; Howard
1997; Morris and Corkum 1996; Strayer 1981, 1983, 1993; Strayer et al.
1994; Vannote and Minshall 1982; Watters 1992, 1993; Way et al. 1989), or
were thought to be major habitat drivers in the Flint River basin (Brim Box
1999). We collected all mesohabitat data (except water samples) during
baseflow conditions between July and August 1999 (Table 1).
We measured canopy openness (canop), coarse woody debris density
(cwdebris) and average mid-channel depth (avgdepth) along a mid-channel
transect extending the length of each 100-m survey reach. At 20-m intervals
along the transects, we determined canopy openness by taking four spherical
densiometer readings (upstream, downstream, left bank, right bank). At 10-
m intervals we made water depth measurements. We estimated coarse
woody debris density in the survey area by recording the size and number of
logs > 10 cm in diameter intercepting the mid-channel transect.
At each site, we determined sediment bulk density (bulkdens), a measure
of substrate porosity, from three substrate core samples collected using a
4.7-cm diameter PVC pipe inserted to a depth of 8.5 cm in the substrate at
mid-channel at the beginning of each survey reach. Sediments captured in
the core were removed, and samples were brought to the lab, dried for two
months under ambient conditions, then weighed. Bulk density was calculated
as sediment dry weight per unit wet volume.
We collected water samples from each survey site between August 31
and September 8 following several weeks of no rain. Samples were placed
in plastic storage bottles and refrigerated until processed, 5–7 days after
2006 P. Gagnon, W. Michener, M. Freeman, and J. Brim Box 35
Table 1. Physical habitat parameters collected within each survey reach (mesohabitat parameters), watershed and riparian zone landscape data (macrohabitat
factors) determined for each study site, and mussel-community metrics employed in analyses.
Metric Description Collection method Transformation In reduced data set
Mesohabitat variables
avgdepth Average baseflow water depth at midchannel (m) Longitudinal line transect log10 Yes
slope Elevation change over 100 m survey reach (m) Longitudinal profile survey sqrt Yes
flowstab Base flow water level: bankful water level x-section profile survey Yes
incision Bankful width: bankful depth x-section profile survey log10 Yes
mannn Manning’s n parameters Visual log10 Yes
habdiv Shannon index of microhabitat diversity x-section line transect log10 No
pool Frequency of pool microhabitat (%) x-section line transect - No
riffle Frequency of riffle microhabitat (%) x-section line transect - No
fines Frequency of fine sediment cover (%) x-section line transect arcsin-sqrt Yes
bulkdens Dry mass of sediment/mL of wet volume (g/mL) Drying and weighing Yes
detritus Frequency of detritus on substrate surface (%) x-section line transect Yes
cwdebris Frequency of logs > 10 cm diameter (#/10m) Longitudinal line transect arcsin-sqrt Yes
cond Water conductivity (mhmos) YSI meter - Yes
canop Canopy openness at midchannel (%) Longitudinal line transect sqrt Yes
Macrohabitat variables
physio Physiographic province type 1:2000 K Physiography map No
catarea Catchment area 1: 100 K DEMs log10 Yes
netpos d-link magnitude 1:100 K DLGs log10 Yes
catrddens Catchment road density (km/km2) 1:100 K DLGs - No
catforest Proportion of catchment in forest cover (%) 1998–90 Landsat TM Yes
catwet Proportion of catchment in wetland cover (%) 1998–90 Landsat TM - Yes
catag Proportion of catchment in agriculture cover (%) 1998–90 Landsat TM No
caturban Proportion of catchment in urban cover (%) 1998–90 Landsat TM - No
riprddens Riparian road density (km/km2) 1:100 K DLGs - No
ripforest Proportion of riparian area in forest cover (%) 1998–90 Landsat TM arcsin-sqrt Yes
ripwet Proportion of riparian area in wetland cover (%) 1998–90 Landsat TM Yes
ripag Proportion of riparian area in agriculture cover (%) 1998–90 Landsat TM arcsin-sqrt No
ripurban Proportion of riparian area in urban cover (%) 1998-–90 Landsat TM - No
Mussel metric
richness Number of species at survey site 1999 survey NA
abund Number of live individuals at survey site 1999 survey log10 NA
swdiv Shannon-Weiner diversity
36 Southeastern Naturalist Vol. 5, No. 1
collection. Alkalinity (alk) was determined using a Mettler DL12 titrator
(Mettler Electronics Corporation, Miami, FL). Conductivity (cond) measurements
were made using a YSI® Meter #33 (YSI Incorporated, Yellow
Springs, OH).
We measured the remaining mesohabitat variables along two crosssection
transects (at 10-m and 75-m points along each 100-m reach)
mapped following Harrelson et al. (1994). We calculated stream incision
(incision) as the ratio of bankfull (determined from visual estimates of
bankfull level) width to bankfull depth measurements (therefore, a smaller
number indicates a higher degree of streambed incision). Flow stability
(flowstab), an approximation of the degree of hydrologic variability, was
the ratio of average base-flow water depth to average bankfull water depth
(a value of 1 indicates high flow stability; values less than 1 indicate flow
instability). The proportion of survey area covered by detritus (detritus),
fine sediments (fines), and slackwater/pool (pool) and riffle (riffle) habitat
we estimated by recording the type of substrate and flow conditions at 1-m
intervals along each cross-section transect. We classified substrate by visual
identification according to a modified Wentworth scale (Cummins
1962). Slackwater/pool habitat was defined as areas of slow-moving to
stagnant water greater than 10 cm deep with substrate consisting of mud,
clay, or detritus. Riffle habitat was defined as areas of fast-moving water
less than 10 cm in depth, with substrate consisting of gravel and cobbleand
boulder-sized limestone. We determined the Shannon-Wiener index of
habitat diversity (habdiv) from the number of unique combinations of
substrate (7 classes: clay, sand, gravel, cobble, boulder, bedrock, detritus)
and water depth (4 classes: 0–0.1m, 0.1–0.5m, 0.5–1.5m, > 1.5m) occurring
along the cross-section transects.
Finally, we assessed channel roughness (mannn), a measure of streambed
irregularity and frequency of flow obstructions and refuges, using
Manning’s n roughness coefficients. We determined the composite
Manning’s n value from scores we assigned to each of six stream roughness
factors: substrate composition, channel irregularity, variation in
channel cross section, obstructions, vegetation, and degree of meandering
(Cowan 1956).
Macrohabitat data
Using a Geographical Information System (Environmental Systems Research
Institute, Inc., 1999), we compiled all of the macrohabitat data
(Table 1). We determined the physiographic province (physio) into which
each site fell based on the 1:2,000,000 Georgia Geologic Survey Physiographic
Province Map.
We characterized drainage network position (netpos) of each survey site
using the 1:100,000 Digital Line Graphs (DLGs, US Geological Survey,
Reston, VA) to calculate d-link magnitude. The d-link magnitude is the
number of first-order streams draining into the point immediately below the
2006 P. Gagnon, W. Michener, M. Freeman, and J. Brim Box 37
first confluence downstream from each survey reach (Osborne and Wiley
1992). Unlike stream order or link magnitude metrics, d-link magnitude
approximates the functional size of a stream by adjusting for the effect of
adjacent waterways and the position of the stream in the drainage network.
For example, the d-link magnitude value for streams draining directly into
large rivers is the sum of the link magnitude of the subject stream plus the
link magnitude of the adjacent large river.
We identified catchment areas (catarea) for each survey site by using
1:100,000 Digital Elevation Models (US Geological Survey, Reston, VA)
and the ArcView Hydrologic Modeling Extension (v1.0) to create polygons
around the drainage area for each site. We delineated riparian zones adjacent
to and upstream from each site by digitally creating a 250-m buffer polygon
around 1 km of stream length above each survey location. To determine road
density (riprddens; catrddens) upstream from each survey site we overlaid
each catchment and riparian polygon on 1:100,000 DLGs. We determined
the proportion of riparian zone and catchment area in forest (ripforest;
catforest), agriculture, wetland, and urban landcover in the same manner as
the roads using a landcover map prepared from Landsat TM imagery taken
during the winters of 1988–1990 (Georgia Department of Natural Resources,
Atlanta, GA).
Analysis methods
To quantify mussel communities in each survey reach, we calculated
three metrics: unionid richness (richness; number of species), abundance
(abund; number of individuals), and Shannon-Wiener diversity (swdiv).
Where necessary, we transformed habitat measurements to normalize data
(Table 1) and removed highly correlated variables and habitat factors that
did not exhibit a wide range of variability across the survey sites. We also
identified sources of multicollinearity by calculating Pearson correlations
between all variables. We omitted variables if they were highly correlated
with others (r > 0.5) and duplicated information provided by other variables
(e.g., we omitted catag because it was highly negatively correlated
to catforest).
We identified the strongest meso- and macrohabitat predictors of diversity,
richness, and abundance using multiple regression with the maximum
adjusted r2 selection method (PROC REG, SAS v 8.1). We tested for patterns
of mussel species and habitat associations using Canonical Correspondence
Analysis (CCA; ter Braak 1986; PCord v 2.0) of species abundances and
meso and macrohabitat variables across all survey sites.
To test the strength of habitat variables associated with each assemblage
type, habitat variables at sites supporting each mussel assemblage type were
compared to variables of non-supporting sites using the Mann-Whitney U
test (PROC NPAR1WAY, SAS v 8.1). Sites were classified as supporting a
particular mussel assemblage type if 10 or more individuals of assemblagetype
species were found at that site during our survey.
38 Southeastern Naturalist Vol. 5, No. 1
Results
General mussel survey data
Across all survey sites, we found 14,873 unionids, including 19 of the 29
species historically inhabiting the lower FRB. Four species comprised 85%
Table 2. Relative abundance of each species across all survey sites. Total abundance of species
in all survey sites was 14,786. Current conservation status (Brim Box and Williams 2000) is
indicated next to the species name: * = currently stable; # = special concern; ## = federally
endangered
Species Relative abundance (%)
Elliptio complanata/icterina (eastern elliptio and variable spike) * 57.70
Villosa lienosa (little spectaclecase) * 14.92
Elliptio crassidens (elephantear) * 13.51
Villosa vibex (southern rainbow) * 3.20
Toxolasma paulus (iridescent lilliput) * 3.18
Quincuncina infucata (sculptured pigtoe) # 3.01
Lampsilis subangulata (shiny-rayed pocketbook) ## 1.16
Elliptio purpurella (inflated spike) # 1.03
Uniomerus carolinianus (Florida pondhorn) * 0.79
Pleurobema pyriforme (oval pigtoe) ## 0.34
Utterbackia imbecillis (paper pondshell) * 0.28
Villosa villosa (downy rainbow) # 0.21
Lampsilis straminea claibornensis (southern fatmucket) # 0.20
Elliptio arctata (delicate spike) # 0.18
Medionidus pencillatus (gulf moccasinshell) ## 0.07
Pyganodon grandis (giant floater) * 0.07
Strophitus subvexus (southern creekmussel) # 0.07
Megalonaias nervosa (washboard) * 0.04
Utterbackia peggyae (Florida floater) * 0.03
Table 3. Models of mussel community diversity (Shannon-Wiener), species richness and
unionid abundance selected through stepwise regression analysis using meso and macrohabitat
predictor variables. (Mussel abundance, avgdepth, and netpos were log10 transformed prior to
analyses.)
Variables
Dependent Independent R2 F Prob > F
Macrohabitat variables only
Diversity 0.12 + 1.78(ripwet)◊◊ + 0.27(netpos)◊ 0.16 3.98 0.02
Richness -1.71 + 16.89(ripwet)◊ + 2.76(netpos)◊ 0.37 12.57 0.00
Abundance 0.34 + 5.88(ripwet)◊ + 0.45(netpos)◊ 0.25 5.36 0.00
Mesohabitat variables only
Diversity 1.97-0.63(flowstab)◊◊ -3.21(bulkdens)◊ 0.27 4.79 0.01
+ 1.91(avgdepth)◊
Richness 1.76 + 21.86(mannn)◊ + 12.95(avgdepth)◊ 0.28 7.74 0.00
Abundance No significant model
Meso- and Macrohabitat variables combined
Diversity 0.88 + 0.70(avgdepth)◊ + 0.25(netpos*ripwet)◊◊ 0.32 9.94 0.00
Richness 3.43 + 3.67(avgdepth)◊ + 2.17(netpos*ripwet)◊ 0.49 13.59 0.00
+ 2.14(netpos*catforest)◊
Abundance 0.18 + 0.80(avgdepth)◊ + 1.70(netpos)◊ 0.36 5.72 0.00
+ 4.31(ripwet)◊ - 1.36(netpos*ripwet)◊
◊p < 0.05; ◊◊p < 0.10
2006 P. Gagnon, W. Michener, M. Freeman, and J. Brim Box 39
of the individuals encountered, and nine species accounted for 96% of the
mussels surveyed (Table 2). Three federally endangered species, Lampsilis
subangulata (Lea), Pleurobema pyriforme (Lea), and Medionidus
pencillatus (Lea) were found in isolated locations throughout the basin,
usually in very low numbers (Fig. 1). Species richness ranged from 0 to 11
species per site, averaging 6 per site, and unionid abundance ranged from 0
to 1723 individuals per site, averaging 323 native mussels per site. Corbicula
numbers ranged from 0 to > 1000 per site.
Mussel assemblage-habitat associations
The reduced habitat data set included 11 mesohabitat and 7 macrohabitat
variables (Table 1). Average mid-channel depth, drainage network position,
riparian wetland cover, and catchment forest cover were the best predictors
of mussel richness, abundance, and diversity (Table 3). These habitat variables
accounted for 32–49% of the variance observed in mussel community
metrics. When only macrohabitat variables were included in the regression
analysis, riparian wetland cover and drainage network position were consistently
selected as the best predictors, explaining 16–37% of the variance in
the three mussel metrics. In a regression analysis consisting only of
mesohabitat variables, flow stability, bulk density, average mid-channel
depth, and channel roughness (mannn) were selected as the variables that
best explained diversity and richness.
Table 4. Canonical correspondence analysis variance scores, species-habitat Pearson correlations,
Monte Carlo test results (for 99 runs of species-habitat correlations), and correlation
scores (bottom portion of table) for habitat variables and CCA axes.
Meso- +
macrohabitat CCA Macrohabitat CCA Mesohabitat CCA
% variance explained 16.3 12.6 13.2 8.1 14.6 7.3
Species-habitat correlation 0.829 0.838 0.762 0.698 0.792 0.705
Monte-Carlo test p < 0.07 p < 0.07 p < 0.01 p < 0.01 p < 0.03 p < 0.07
Variable Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2
avgdepth -0.485 -0.132 - - -0.538 0.533
slope -0.039 -0.133 - - -0.016 -0.017
flowstab -0.131 -0.310 - - -0.129 0.529
incision -0.057 -0.158 - - -0.028 0.130
mannn 0.054 -0.189 - - 0.079 0.058
fines 0.708 -0.252 - - 0.754 0.374
bulkdens -0.293 -0.137 - - -0.295 0.043
detritus 0.475 -0.216 - - 0.515 0.068
cwdebris 0.345 -0.438 - - 0.380 0.598
cond -0.110 0.201 - - -0.103 -0.501
canop -0.405 -0.047 - - -0.453 0.168
catarea -0.637 -0.403 -0.796 -0.401 - -
netpos -0.377 0.394 -0.374 0.501 - -
catforest 0.150 0.135 0.157 0.030 - -
catwet -0.184 0.089 -0.207 0.237
ripforest 0.247 -0.041 0.265 -0.129 - -
ripwet 0.421 -0.515 0.358 -0.755 - -
40 Southeastern Naturalist Vol. 5, No. 1
Overall, macro- and mesohabitat variables accounted for up to 28.9% of
the variation observed in mussel community composition across the sites
(Table 4). The macrohabitat-only CCA had a total explained variance of
21.3% (Table 4, Fig. 2). Two axes were significant. The first axis was
correlated with catchment area and the second was most closely correlated
with riparian wetland cover. The CCA considering only mesohabitat
Figure 2. Biplot of species scores and macrohabitat variable vectors for CCA axes
one and two from the CCA ordination of the 1999 survey data. Site scores have been
omitted for clarity. Species names are abbreviated by indicating the first name of the
genus and the first four letters of the epithet. Species groups have been manually
encircled on the biplot. Arrows represent the habitat variables most highly correlated
with each ordination axis. The length and direction of each arrow corresponds to the
magnitude and direction of positive change for the habitat variable. Variable names
correspond to abbreviations in Table 1. Table 4 details the correlation scores between
habitat variables and CCA axes. Group 1 = slackwater associates; Group 2 = large
river riffle associates; Group 3 = generalists; Group 4 = stream run associates.
2006 P. Gagnon, W. Michener, M. Freeman, and J. Brim Box 41
variables resulted in 21.9% of the observed variance explained by the habitat
variables (Table 4, Fig. 3). Only Axis 1 was significant, with multiple
habitat variables, including average mid-channel depth, percent fines, and
detritus cover demonstrating similar levels of correlation with the axis. In
the combined macrohabitat plus mesohabitat CCA, no axis was significant
(Table 4, Fig. 4).
In all of the CCAs, species were arrayed in similar clusters in habitat
space. Accordingly, four assemblage types were identified from CCAs
(Figs. 2–4): slackwater associates, large-river riffle associates, habitat generalists,
and stream-run associates. Mann-Whitney U tests largely confirmed
habitat differences among sites supporting different assemblage types
(Table 5).
The slackwater associates included Utterbackia peggyae (Johnson),
Utterbackia imbecillis (Say), Villosa villosa (Wright) and Pyganodon
grandis (Say). CCA indicated these species were associated with small,
shallow streams high in conductivity, fine sediment cover, and detritus
(Table 4, Figs. 2–4). These physical conditions are indicative of pool
Figure 3. Biplot of species scores and mesohabitat variable vectors for CCA axes one
and two. See Figure 2 legend for explanation.
42 Southeastern Naturalist Vol. 5, No. 1
environs, stream margins and backwater areas, and small headwater creeks
in marshes. Sites supporting slackwater associates had significantly more
pool habitat and were lower in riparian wetland cover than sites not supporting
slackwater species (Table 5).
Figure 4. Biplot of species scores and meso- and macrohabitat variable vectors for
CCA axes one and two. See Figure 2 legend for explanation.
2006 P. Gagnon, W. Michener, M. Freeman, and J. Brim Box 43
Elliptio arctata (Conrad), Megalonaias nervosa (Rafinesque), and Elliptio
crassidens (Lamarck) comprised the group of large-river riffle associates.
These taxa were associated with streams having large catchments, deeper
flow, high levels of coarse wood debris density and flow stability, and low
levels of fine sediments and detritus (Figs. 2–4; Table 4). Sites inhabited by
large-river riffle species were significantly larger and deeper, and had more
riffle habitat and lower amounts of detritus, fines, and coarse wood debris than
non-supporting sites. Large-river riffle type species were also found in areas
with lower riparian forest and riparian wetland cover (Table 5).
The most abundant and widespread species in the basin, Elliptio
complanata/ icterina, Villosa lienosa (Conrad), Villosa vibex (Conrad),
Toxolasma paulus (Lea), Uniomerus carolinianus (Bosc), were clustered
into the generalist group. Sites supporting generalist species had few distinguishing
features: significantly more pool habitat and riparian wetland
cover, but lower sediment bulk density than non-supporting sites (Table 5).
Table 5. Mann-Whitney U (Wilcoxon rank-sum) test results comparing habitat differences
between sites supporting and not supporting each assemblage type.
Supporting sites Non-supporting sites
Mean StdError Mean Std Error Z p > Z
Slackwater (n = 2) (n = 44)
pool 0.35 0.31 0.06 0.01 1.39 0.08
netpos* 2.62 0.59 1.79 0.09 1.43 0.07
catarea* 3.82 0.21 4.37 0.08 -1.43 0.08
ripwet* -0.83 0.25 -0.52 0.05 -1.54 0.06
Large river/riffle (n = 5) (n = 41)
detritus 0.15 0.04 0.39 0.04 -2.29 0.01
cwdens** 0.27 0.08 0.45 0.03 -1.71 0.04
fines** 0.84 0.20 1.41 0.04 -2.77 0.01
pool 0.00 0.00 0.09 0.02 -1.79 0.04
riffle 0.35 0.15 0.03 0.01 2.37 0.01
avgdepth* -0.01 0.05 -0.30 0.05 2.29 0.01
netpos* 2.43 0.20 1.75 0.09 2.33 0.01
catarea* 4.98 0.16 4.27 0.08 2.57 0.01
ripwet* -0.76 0.19 -0.50 0.05 -1.55 0.06
ripforest** 0.90 0.11 1.11 0.03 -1.87 0.03
Generalist (n = 36) (n = 10)
bulkdens 1.44 0.04 1.64 0.05 2.63 0.01
pool 0.09 0.02 0.02 0.02 -2.21 0.01
ripwet* -0.44 0.04 -0.85 0.14 -3.02 0.00
Stream/run (n = 16) (n = 30)
cond 87.7 18.56 162.7 17.44 -2.71 0.01
flowstab 0.45 0.03 0.36 0.04 1.93 0.03
avgdepth* -0.15 0.04 -0.34 0.07 2.07 0.02
netpos* 1.98 0.12 1.73 0.13 1.46 0.07
catarea* 4.59 0.10 4.22 0.12 2.32 0.01
ripwet* -0.36 0.05 -0.62 0.07 2.33 0.01
catforest 0.49 0.03 0.41 0.02 2.25 0.01
*Values not back-transformed from log10 transformation.
**Values not back-transformed from arcsine squareroot transformation.
44 Southeastern Naturalist Vol. 5, No. 1
Table 6. Comparison of historic accounts of species-habitat relationships and findings of this study. Numbered references given in parentheses represent the
following sources: 1 = Headlee 1906, 2 = Clench and Turner 1956, 3 = Johnson 1965, 4 = Johnson 1970, 5 = Jenkinson 1973, 6 = Heard 1975, 7 = Heard 1979,
8 = Huehner 1987, 9 = Butler 1989, 10 = Counts et al. 1991, 11 = Cummings and Mayer 1992, 12 = Williams and Butler 1994, 13 = Brim Box and Williams 2000,
14 = www.inhs.uiuc.edu/cbd/ilspecies/mollusks,list.htm.
Habitat preference
Species Reported mesohabitat Reported substrate identified in this study
Slackwater Pool mesohabitat; fine
Pyganodon grandis Ponds, lakes, impoundments, mud-bottomed creeks and Sand, mud (13,1,8) sediments; low flow stability,
rivers (13,1,14) stream incision, detritus cover,
Utterbackia imbecillis Slackwater, banks of large rivers, ponds, reservoirs (4,13,2) Sand, mud (13,2) and sediment bulk density;
Utterbackia peggyae Sluggish streams and ponds (3) Sand, mud (13,3) closed canopy; smaller streams
Villosa villosa Slackwater, banks of large rivers, ponds, reservoirs, Mud, silt, clay, sand, vegetation,
spring-fed streams and clear rivers (2,4,7,9) limestone (7,9,13)
Large-river riffle Riffle mesohabitat; low levels of
Elliptio arctata Shore; swift current (4,13) Sand, gravel, limestone, fine sediments, detritus, and
vegetation (7,13) coarse woody debris; high
Elliptio crassidens Moderate to strong current; sandbars (7,4) Sand, limestone, rock, muddy incision ratio; large streams
sand (7,13) and rivers
Megalonaias nervosa Large rivers; main channel (13,7,2) Sand, limestone, rock, muddy
sand (7,13,2)
Generalist General stream habitat and pool
Elliptio complanata All types (5,7,10,13) All types (5,7,10,13) habitat; low sediment bulk
Elliptio icterina Streams, lakes, reservoirs, ponds, large rivers; swift to Mud,clay,sand,gravel, limestone density; riparian wetland cover
moderate current (4,7,13) (13,5)
2006 P. Gagnon, W. Michener, M. Freeman, and J. Brim Box 45
Table 6, continued.
Habitat preference
Species Reported mesohabitat Reported substrate identified in this study
Generalist, cont.
Toxolasma paulus Small streams with slight current; lakes; banks (7,2) Mud, clay, sand, vegetation,
rock (5,7,13)
Uniomerus carolinianus Slight current; lakes (4,13) Sand, mud, clay, limestone (7,13)
Villosa lienosa Slight to moderate current; small to large streams (5,7,13,2) Mud, silt, clay, sand, limestone
(5,7,13,2)
Villosa vibex Small rivers, creeks and lakes; slight to moderate current Mud, silt, clay, sand, limestone
(4,7,12) (4,7,12,13)
Stream-run Intermediate stream habitat; low
Lampsilis straminea Main channel and banks of large creeks and rivers, slow to Sand, sandy mud bottom, conductivity; high flow
claibornensis moderate current (13,2,7) limestone (7,2,13) stability; larger streams and
Elliptio purpurella Sand and limestone (14) rivers; riparian wetland cover
Quincuncina infucata Sand-bottomed pools, rocky areas with swift current, small Sand, mud, fine gravel,
to large streams (5,2,7,6) limestone, detritus (6,7,13)
Medionidus pencillatus Streams and rivers with moderate current, sandy areas with Mud, clay, sand, gravel, cobble,
slight current (5,2,6,7,13) limestone (5,2,6,7,13)
Pleurobema pyriforme Small to large streams with moderate current, clean Mud, sand, clay, cobble
substrate; midchannel (6,712,13) (5,6,7,12,13)
Strophitus subvexus Backwater, slow to moderate currents, large creeks and Sand, mud (12,13,7)
rivers (12,2,7,13)
Lampsilis subangulata Small creeks to large rivers; slow to moderate current Sand, clay, rock (6,7,13)
(12,2,7)
46 Southeastern Naturalist Vol. 5, No. 1
The final group, the stream-run associates, included the federally endangered
Lampsilis subangulata, Medionidus pencillatus, and Pleurobema
pyriforme, and species of special concern in the FRB: Elliptio purpurella
(Lea), Lampsilis straminea claibornensis (Lea), Quincuncina infucata
(Conrad), and Strophitus subvexus (Conrad) (Brim Box and Williams 2000).
Although the CCA biplot placed these species close to generalist species in
habitat space, Mann-Whitney U-tests demonstrated that these taxa were
found in larger streams (greater drainage network position, catchment area,
and average mid-channel depth) that had significantly lower levels of conductivity,
greater flow stability, and catchment forest and riparian wetland
cover in comparison to non-supporting (Figs. 2–4, Tables 4 and 5). Because
these taxa occupied “intermediate” habitat space on the CCA biplot (i.e.,
they fell between riffle habitat and slackwater habitat—presumably in a
habitat space akin to stream runs), we labeled them as stream-run associates.
Comparisons of the findings of this study and historic anecdotal and
experimental evidence of mussel-habitat associations for Flint basin mussels
are provided in Table 6. Our results conform well to the habitats identified
with slackwater and large-river riffle associates. In our study, generalist
species showed no strong patterns of habitat associations, a situation which
is also suggested in the reported habitat conditions tied to these taxa in other
locations. Similarities between the habitats identified for stream-run associates
in our study and those reported from other studies were not strong.
Discussion
Freshwater mussels of the FRB appear to be significantly influenced and
structured by both meso- and macroscale habitat conditions. Primary
macrohabitat drivers include drainage network position, riparian wetland
and catchment forest cover, and mesohabitat complexes such as riffles and
slackwater areas. Multiple species demonstrated similar habitat preferences,
and could be grouped into assemblage types based on patterns of co-occurrence
and habitat use.
Among the habitat variables we examined, two emerge as important
indicators of FRB mussel community richness, diversity, and abundance
metrics. The first of these is stream drainage network position, which has
also been positively correlated with mussel richness across North
America: in central Alabama, the Ohio basin, southeastern Michigan, and
northern Atlantic Slope streams (Haag and Warren 1998; Strayer 1983,
1993; Watters 1992, 1993). Possible mechanisms driving this pattern are
greater flow stability, lower environmental stochasticity, as well as
higher habitat heterogeneity and host fish abundance in larger streams
(Haag and Warren 1998, Strayer 1993). In the lower FRB, the persistence
of perennial flows may be one of the greatest factors contributing to the
increased diversity of larger stream habitats, as extensive amounts of
headwater and low-order stream reaches stagnate and dry up during
2006 P. Gagnon, W. Michener, M. Freeman, and J. Brim Box 47
drought conditions, and most basin species do not tolerate hypoxic and
desiccating conditions (Johnson 2001).
The second habitat variable important in predicting mussel richness,
abundance, and diversity in the FRB is riparian wetland and catchment
forest cover. Again, this pattern of mussel association with forested areas
has been observed elsewhere in North America (Howard 1997, Morris and
Corkum 1996). The influence of catchment forest and riparian wetland cover
on unionid assemblages is likely due to the capacity of these land cover
types to retain sediments from overland flow, and chemically filter water
before entering streams in Coastal Plain ecosystems (Lowrance et al. 1984,
1986), particularly since mussels are thought to be very sensitive to pollution
and sedimentation (for summaries, see Brim Box and Mossa 1999,
Fuller 1974).
While patterns of diversity and richness are related more to catchmentscale
features, site-level assemblage compositions in the lower FRB seem to
be constrained by a combination of mesohabitat parameters. In particular,
factors such as water depth, coarse wood debris, detritus cover, sediment
bulk density and riparian wetland cover combine to create mesohabitat
complexes (e.g., riffles, slackwater areas) that support predictable combinations
of basin unionid species.
Two groups of species occupy extreme ends of the mesohabitat continuum
from riffle and slackwater environs, while two other species groups,
the generalists and stream-run associates, occupy intermediate conditions
between these two extremes. Despite their close proximity in the CCA
diagram, the two intermediary species groups display different patterns of
habitat associations. Based on the results of the Mann-Whitney U tests, we
conclude that the intermediary placement of generalist species represents the
widespread, non-specific habitats of component species. In contrast, the
placement of stream-run species on the CCA plot seems to be an indication
that these species primarily occupy habitats that are truly intermediate
between riffle and slackwater environs. Although limited due to the paucity
of locations where stream-run species were found, our data suggest that the
habitats of stream-run species include areas of relatively low conductivity,
and high flow stability, water depth, coarse woody debris density, and
canopy openness.
Features of unionid shell morphology and respiratory physiology have
been hypothesized as factors governing species-habitat relationships and may
prove to be the mechanistic link between mussels and their preferred habitats
(Chen et al. 1997, Sheldon and Walker 1989, Watters 1994). For example,
heavy valve structure is thought to confer “anchorage” against shear forces
(Sheldon and Walker 1989), resistance to lethal abrasion and desiccation, and
protection against predators (McMahon 1991, Watters 1994). Consequently,
thick-shelled species may inhabit more abrasive riffle environments and large
river systems. Thin shells provide fewer defenses against predators, abrasion
48 Southeastern Naturalist Vol. 5, No. 1
and desiccation, but allow for faster growth, earlier maturation (McMahon
1991), and buoyancy in soft sediments. Thin-shelled species are thought to be
adapted to pool environments, and soft-bottom, slow-moving backwater areas
of large rivers and streams.
Respiratory response under hypoxia may also be a key physiological
determinant of unionid distribution potential. Some mussels (particularly
pool and lake dwellers, which experience seasonal hypoxic conditions)
demonstrate the ability to regulate oxygen uptake in response to declining
levels of dissolved oxygen. Other species (particularly riffle and riverine
species) are unable to regulate oxygen uptake, and cannot tolerate hypoxic
conditions (Chen et al. 1997, Sheldon and Walker 1989).
Available data on the hypoxia-related mortality and shell characteristics
of FRB mussels align with the results that hypotheses described in the
previous paragraphs would predict (Table 7). Among mussels that were
measured and monitored (Johnson 2001), slackwater species had the thinnest
shells, while large river riffle species had the thickest. Stream-run and
generalist species had intermediate shell thickness. Mortality under hypoxia
also varied with habitat association: generalist species were relatively tolerant
of hypoxia; stream-run species had higher mortality under hypoxia; and
the highest mortality under hypoxic conditions was observed with one largeriver
riffle species.
Table 7. Comparison of hypoxia tolerance, shell thickness, and conservation status of species in
each assemblage type. “NA” means data were not available. Hypoxia data were originally
reported in Johnson 2001.
Average mortality
under hypoxia Average shell
Species (DO < 5 mg/L) thickness:length
Slackwater
Pyganodon grandis na 1.01
Utterbackia imbecillis na 0.64
Large-river riffle
Elliptio arctata na na
Elliptio crassidens 82% ± 9% 5.94
Megalonaias nervosa na 6.74
Generalist
Elliptio complanata/icterina 9% ± 2% 2.23
Toxolasma paulus 23% ± 12% 3.83
Uniomerus carolinianus 0% ± 0% 2.82
Villosa lienosa 9% ± 5% 2.64
Villosa vibex 3% ± 1% 1.62
Stream-run
Lampsilis straminea claibornensis na 5.17
Elliptio purpurella na 3.01
Quincuncina infucata na 5.18
Medionidus pencillatus 50% ± 29% 3.25
Pleurobema pyriforme 15% ± 7% 3.76
Lampsilis subangulata 28% ± 10% 2.62
2006 P. Gagnon, W. Michener, M. Freeman, and J. Brim Box 49
The results of this study provide further insight into the habitat associations
of FRB mussels, but do not offer a complete picture of mussel-habitat
requirements. We were able to identify several habitat and landscape variables
that accounted for a moderate amount of variation in mussel richness,
abundance, and diversity in the basin. Furthermore, we elucidated distinct
mussel assemblage groups and habitat associations that are broadly concordant
with previous qualitative descriptions for these species. However,
numerous habitat variables remain to be examined and much still needs to be
done to identify the habitat needs and conservation requirements of FRB
mussels, in general, and rare species (the stream-run associates), in particular.
Such research would be critical to successful conservation of mussel
diversity in this region.
Acknowledgments
We are grateful to Annie Hill, Ryan Johnson, Kim Lellis, Walter Cotton, Derek
Fussell, Roger Birkhead, Steve Gagnon, and Carson Stringfellow for providing field
and technical help with this research. Liz Cox was helpful in obtaining reference
materials and literature. Critical reviews on an earlier draft of this manuscript were
provided by Stephen Golladay and Cathy Pringle. Funding was provided by the Jones
Ecological Research Center, the Georgia Department of Natural Resources, and the
Josh Laerm Memorial Award.
Literature Cited
Bauer G., B. Hochwald, and W. Silkenat. 1991. Spatial distribution of freshwater
mussels: The role of host fish and metabolic rate. Freshwater Biology 26:377–386.
Brim Box, J. 1999. Community structure of freshwater mussels (Bivalvia:
Unionidae) in Coastal Plain streams of the southeastern US. Ph.D. Dissertation.
University of Florida. Gainesville, FL.
Brim Box, J., and J. Mossa. 1999. Sediment, land use, and freshwater mussels:
Prospects and problems. Journal of the North American Benthological Society
18(1):99–117.
Brim Box, J., and J.D. Williams. 2000. Unionid mollusks of the Apalachicola basin
in Alabama, Florida, and Georgia. Alabama Museum of Natural History Bulletin
21. Tuscaloosa, AL. 143 pp.
Butler, R.S. 1989. Distributional records for freshwater mussels (Bivalvia:
Unionidae in Florida and south Alabama, with zoogeographic and taxonomic
notes. Walkerana 3(10):239–261.
Chen, L., A.G. Heath, and R.J. Neves. 1997. Oxygen consumption and anaerobic
metabolite changes of freshwater mussels (Unionidae) from different habitats
during declining dissolved oxygen exposure. Abstracts of the 1997 National
Shellfisheries Association Annual Meeting, Fort Walton Beach, FL.
Clench, W.J., and R.D. Turner. 1956. Freshwater mollusks of Alabama, Georgia, and
Florida from the Escambia to the Suwannee River. Bulletin of the Florida State
Museum 1(3):97–239.
Counts III, C.L., T.S. Handwerker, and R.V. Jesien. 1991. The naiades (Bivalvia:
Unionoidea) of the Delmarva Peninsula. American Malacological Bulletin
9(1):27–37.
50 Southeastern Naturalist Vol. 5, No. 1
Cowan, W.L. 1956. Estimating hydraulic roughness coefficients. Agricultural Engineering
37:83–85.
Cummings, K.S., and C.A. Mayer. 1992. Field Guide to Freshwater Mussels of the
Midwest. Illinois Natural History Survey Manual 5. 194 pp.
Cummins, K.E. 1962. An evaluation of some techniques for the collection and
analysis of benthic samples with special emphasis on lotic waters. American
Midland Naturalist 67:477–504.
DiMaio, J., and L.D. Corkum. 1995. Relationship between the spatial distribution of
freshwater mussels (Bivalvia: Unionidae) and the hydrological variability of
rivers. Canadian Journal of Zoology 73:663–671.
Fuller, S.L.H. 1974. Clams and Mussels (Mollusca: Bivalvia). Pollution ecology of
freshwater invertebrates. Pp. 215–273, In C.W. Hart, Jr. and S.L.H. Fuller (Eds.).
Pollution Ecology of Freshwater Invertebrates. Academic Press, New York, NY.
389 pp.
Golladay, S.W., K.Watt, S. Entrekin, and J. Battle. 2000. Hydrologic and geomorphic
controls on suspended particulate organic matter concentration and transport
in Ichawaynochaway Creek, Georgia, USA. Archiv fur Hydrobiologie
149(4):655–678.
Haag W.R., and M.L.Warren. 1998. Role of ecological factors and reproductive
strategies in structuring freshwater mussel communities. Canadian Journal of
Fisheries and Aquatic Science 55:297–306.
Hamilton H., J. Brim Box, and R.M. Dorazio. 1997. Effects of habitat suitability on
the survival of relocated freshwater mussels. Regulated Rivers Research and
Management 13:537–541.
Harrelson, C.C., C.L.Rawlins, and P. Potyondy. 1994. Stream channel reference
sites: An illustrated guide to field technique. US Forest Service General Technical
Report RM-245. Rocky Mountain Forest and Range Experiment Station, Fort
Collins, CO. 67 pp.
Headlee, T.J. 1906. Ecological notes on the mussels of Winona, Pike, and Center
lakes of Kosciusko County, Indiana. Biological Bulletin 11(6):305–319.
Heard, W.H. 1975. Determination of the endangered status of freshwater clams of
the Gulf and Southeastern states. Terminal Report for the Office of Endangered
Species, Bureau of Sport Fishereis and Wildlife, US Department of the Interior,
Washington, DC. Contract 14-16-000-8905. 31 pp.
Heard, W.H. 1979. Identification manual of the freshwater clams of Florida. Florida
Department of Environmental Regulation Technical Series 4(2):1–82.
Houhoulis, P.F., and W.K. Michener. 2000. Detecting wetland change: A rule-based
approach using NWI and SPOT-XS data. Photogrammetric Engineering and
Remote Sensing 66:205–211.
Howard, J. 1997. Land use effects on freshwater mussels in three watersheds in east
central Alabama: A geographic information systems analysis. M.Sc. Thesis.
University of Florida, Gainsville, FL. 188 pp.
Huehner, M.K. 1987. Field and laboratory determination of substrate preferences of
unionid mussels. Ohio Journal of Science 87(1):29–32.
Jenkinson, J.J. 1973. Distribution and zoogeography of the Unionidae
(Mollusca:Bivalvia) in four creek systems in east-central Alabama. M.Sc.Thesis.
Auburn University, Auburn, AL. 96 pp.
Johnson, P.M. 2001. Habitat associations and drought responses of mussels in the
lower Flint River basin, southwest GA. M.Sc. Thesis. University of Georgia,
Athens, GA. 114 pp.
2006 P. Gagnon, W. Michener, M. Freeman, and J. Brim Box 51
Johnson, R.I. 1965. A hitherto overlooked Anodonta (Mollusca: Unionidae) from the
Gulf drainage of Florida. Breviora 213:1–7 + 2 plates.
Johnson, R.I. 1970. Systematics and zoogeography of the Unionidae (Mollusca:
Bivalvia) of the southern Atlantic Slope region. Bulletin of the Museum of
Comparative Zoology 140(6):263–449.
Lowrance, R.R., R.L. Todd, J. Fail, Jr., O. Hendrickson, Jr., R. Leonard, and L.
Asmussen. 1984. Riparian forests as nutrient filters in agricultural watersheds.
Bioscience 34:374–377.
Lowrance, R., J.K. Sharpe, and J.M. Sheridan. 1986. Long-term sediment deposition
in the riparian zone of a Coastal Plain watershed. Journal of Soil and Water
Conservation 41:266–271.
McMahon, R.F. 1991. Mollusca: Bivalvia. Pp. 315–399, In J.H. Thorpe and A.P.
Covich. Ecology and Classification of North American Freshwater Invertebrates.
Academic Press, New York, NY. 911 pp.
Morris, T.J., and L.D. Corkum. 1996. Assemblage structure of freshwater mussels
(Bivalvia: Unionidae) in rivers with grassy and forested riparian zones. Journal
of the North American Benthological Society 15:576–586.
Neves, R.J., A.E. Bogan, J.D. Williams, S.A. Ahlstedt, and P.W. Hartfield. 1997. Pp.
43–86, In G.W. Benz (Ed.). Status of aquatic mollusks in the southeastern United
States: A downward spiral of diversity. Aquatic Fauna in Peril: The Southeastern
Perspective. Lenz Design and Communications, Decatur, GA. 554 pp.
Osborne, L.L., and M.J. Wiley. 1992. Influence of tributary spatial position on the
structure of warmwater fish communities. Canadian Journal of Fisheries and
Aquatic Sciences 49:671–681.
Sheldon, R., and K.F. Walker. 1989. Effects of hypoxia on oxygen consumption by
two species of freshwater mussel (Unionacea: Hyriidae) from the River Murray.
Australian Journal of Marine and Freshwater Research 40:491–499.
Strayer, D.L. 1981. Notes on the microhabitats of unionid mussels in some Michigan
streams. American Midland Naturalist 106(2):411–415.
Strayer, D.L. 1983. Effects of surface geology and stream size on freshwater mussel
(Bivalvia, Unionidae) distribution in southeastern Michigan, USA. Freshwater
Biology 13:253–264.
Strayer, D.L. 1993. Macrohabitats of freshwater mussels (Bivalvia: Unionacea) in
streams of the northern Atlantic Slope. Journal of the North American
Benthological Society 12(3):236–246.
Strayer, D.L., and J. Ralley. 1993. Microhabitat use by an assemblage of streamdwelling
unionaceans (Bivalvia), including two rare species of Alasmidonta. Journal
of the North American Benthological Society 12(3): 247–258.
Strayer, D.L., D.C. Hunter, L.C. Smith, and C.K. Borg. 1994. Distribution, abundance,
and roles of freshwater clams (Bivalvia, Unionidae) in the freshwater tidal
Hudson River. Freshwater Biology 31:239–248.
ter Braak, C.J.F. 1986. Canonical correspondence analysis: A new eigenvector
technique for multivariate direct gradient analysis. Ecology 67:1167–1179.
Vannote, R.L., and G.W. Minshall. 1982. Fluvial processes and local lithology
controlling abundance, structure, and composition of mussel beds. Proceedings
of the National Academy of Sciences 79:4103–4107.
Watters, G.T. 1992. Unionids, fishes, and the species-area curve. Journal of Biogeography
19:481–490.
52 Southeastern Naturalist Vol. 5, No. 1
Watters, G.T. 1993. Mussel diversity as a function of drainage area and fish diversity:
Management implications. Pp. 113–116, In K.S. Cummings, A.C.
Buchanan, and L.M. Koch (Eds.). Conservation and management of freshwater
mussels. Proceedings of an Upper Mississippi River Conservation Committee
symposium, 12–14 October 1992, St. Louis, MO. Upper River Conservation
Committee, Rock Island, IL. 189 pp.
Watters, G.T. 1994. Form and function of unionoidean shell sculpture and shape
(Bivalvia). American Malacological Bulletin 11(1):1–20.
Way, C.M., A.C. Miller, and B.S. Payne. 1989. Influence of physical factors on the
distribution and abundance of freshwater mussels (Bivalvia: Unionidae) in the
lower Tennessee River. Nautilus 103(3):96–98.
Williams, J.D., and R.S. Butler. 1994. Class Bivalvia, Order Unionoida, freshwater
bivalves. Pp. 53–128, In M. Deyrup and R.Franz (Eds.). Rare and Endangered
Biota of Florida, Vol. II: Invertebrates. University of Florida Press, Gainesville,
FL. 798 pp.
Williams, J., M. Warren, K.Cummings, J. Harris, and R. Neves. 1993. Conservation
status of freshwater mussels of the United States and Canada. Fisheries
18(9):6–22.