Southeastern Naturalist
237
K.F. Robinson and C.A. Jennings
22001144 SOUTHEASTERN NATURALIST 1V3o(2l.) :1233,7 N–2o6. 02
A Comparison of Resident Fish Assemblages in Managed
and Unmanaged Coastal Wetlands in North Carolina and
South Carolina
Kelly F. Robinson1,2 and Cecil A. Jennings3,*
Abstract - The dominant fish species within impounded coastal wetlands in the southeastern
US may be different from the species that dominate natural marshes. We tested the
hypothesis that resident fish assemblages inhabiting impounded coastal wetlands in South
Carolina would differ from resident assemblages in natural marshes of the southeastern
United States. We used rarefied species richness, Shannon’s H' diversity, J' evenness,
Morisita’s index of similarity, and the percent similarity index to compare resident fish
assemblages from two impoundments to 12 open-marsh resident fish assemblages from
previously published studies in North and South Carolina. We used rotenone to sample fish
assemblages in impoundments. The assemblages in natural marsh habitat had been sampled
with rotenone and seines. We classified comparisons yielding a similarity index ≥0.50 as
moderately similar and those with an index ≥0.75 as very similar. Fifty-three percent of
the among-impoundment comparisons (Morisita’s index) were at least moderately similar,
whereas 7% of impoundment–natural marsh comparisons were moderately similar. A difference
in tidal influence was the only parameter in the best-fitting model describing the
observed Morisita’s indices. The index of similarity decreased by 63% when tidal influence
differed between compared assemblages. Species richness and diversity were greater in
impoundments than natural marshes, but evenness was similar between habitat types. Our
results support the hypothesis that resident fish assemblages in impounded wetlands and
natural marshes are different, and suggest that a degree of tidal influence is the most important
factor behind the difference.
Introduction
In the late 18th century, thousands of hectares of estuarine marsh along the
southeastern coast of the United States were ditched, drained, and impounded for
rice cultivation and mosquito control. Although a large number of these impoundments
have been abandoned and now are subject to tidal inundation, about 28,000
ha of coastal impoundments are still intact in South Carolina (DeVoe et al. 1987,
Kelley 1999, Tiner 1977). Water-level manipulation strategies in South Carolina
impoundments, which are designed to benefit migratory waterfowl, greatly reduce
1Georgia Cooperative Fish and Wildlife Research Unit, Warnell School of Forestry and
Natural Resources, 180 East Green Street, The University of Georgia, Athens, GA 30602-
2152. 2Current address - New York Cooperative Fish and Wildlife Research Unit, Department
of Natural Resources, B02 Bruckner Hall, Cornell University, Ithaca, NY 14853. 3US
Geological Survey, Georgia Cooperative Fish and Wildlife Research Unit, Warnell School
of Forestry and Natural Resources, 180 East Green Street, The University of Georgia, Athens,
GA 30602-2152. *Corresponding author - jennings@uga.edu.
Manuscript Editor: Andrew Rypel
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
238
tidal exchange within these structures and can affect the fish and invertebrate assemblages
within the managed area. For example, in a comparison of former rice
fields on the Cooper River, SC, the amount of water export from an intact impoundment
was between 0.56% and 1.01% of the amount of water export from breached
former rice fields (Joyner 2007).
Reduced tidal exchange decreases fish and invertebrate access to managed
wetlands, creates unfavorable abiotic conditions such as salinity extremes and
low dissolved oxygen, and reduces population exchange with adjacent unaltered
environments (Wenner et al. 1986). Additionally, the manipulation of water levels
causes the plant community to shift from emergent halophytic macrophytes
to submerged aquatic vegetation (SAV) found more commonly in freshwater and
brackish environments (Portnoy 1999, Rogers et al. 1994, Rozas and Minello
1999). Water-level manipulations in impoundments encourage vegetative communities
that are beneficial for waterfowl (Tiner 1977) and that probably favor
certain species of crustaceans and fish (Rogers et al. 1994). Although many studies
(McGovern and Wenner 1990, Robinson and Jennings 2012, Rogers et al.
1992, Rozas and Minello 1999, Wenner et al. 1986) have documented the negative
effects of impoundments on marine-transient fishes (i.e., those species that
use the estuary for a portion of their life cycle), fewer studies have focused on the
resident fishes within these structures.
Impoundments may provide some potential benefits to resident fishes. These
benefits include greater abundances of prey associated with SAV, reduced competition
for food resources, and decreased predation risk because of the lower overall
abundance of fish in the impoundments (Hoese and Konikoff 1995, Rozas and Minello
1999). For example, phytoplankton and zooplankton are often more readily
available in impoundments and may therefore represent more stable environments
for certain types of fishes (Herke et al. 1992, Milgarese and Sandifer 1982, Talbot
et al. 1986). Therefore,environmental conditions in impounded marshes, whether
beneficial or detrimental to resident fish species, have the potential to alter assemblage
structure.
Reduced immigration into impounded wetlands (McGovern and Wenner 1990,
Rozas and Minello 1999, Wenner et al. 1986), as well as impoundment-specific
environmental characteristics, may alter the resident fish assemblages within these
structures (Rogers et al. 1994, Rozas and Minello 1999). For example, previous
researchers have suggested that resident fish assemblages within estuarine impoundments
that lack diel tidal cycles may become dominated by poeciliid fishes
at the expense of killifishes (Kneib 1997). Poeciliids are often dominant or codominant
in lower-latitude, low-energy marshes, such as those on the coasts of the
Gulf of Mexico and Florida, but are not as abundant in the marshes of the Carolinas
(see Nordlie 2003 for review).
Considering these documented effects of impoundments on marsh-fish assemblages,
our objectives for this study were to: 1) compare the resident fish assemblages
in two mesohaline waterfowl impoundments along the Combahee River, SC, with
resident assemblages in natural marsh habitats along the southeastern coast of the
Southeastern Naturalist
239
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
US, and 2) determine the factors that influence any observed differences in assemblage
structure. We predicted that the resident fish assemblages within our study
impoundments would differ from assemblages collected in natural marsh habitats.
For example, we expected that stress-tolerant species would be more abundant in
the impoundments than in the natural marsh habitats, where they do not have a
competitive advantage. We also expected that the differences in tidal influence and
salinity would drive any observed dissimilarities in fish assemb lages.
Field-site Description
The two mesohaline study impoundments, Nieuport and Big Rice Field (BRF),
were located on the banks of the Combahee River in Beaufort County, SC, ≈48–53
river-km from the coast (Fig. 1). Nieuport was much larger (≈118 ha) than BRF
(≈48 ha). The Combahee River is part of the ACE Basin watershed, so named because
it drains the Ashepoo, Combahee, and Edisto rivers. Much of the land around
the ACE Basin is protected through efforts of private and public entities, and as
such, this study system has experienced fewer recent anthropogenic influences than
other estuarine areas in coastal South Carolina.
Both study impoundments were managed to enhance the growth of forage, including
Ruppia maritima L. (Widgeongrass), for migratory waterfowl and other
shore birds and water birds. These impoundments were of the same basic design: an
Figure 1. Map of the study impoundments (Nieuport and Big Rice Field) along the Combahee
River, SC, sampled 2008–2010.
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
240
elevated central marsh composed of open-water areas and marsh edge, surrounded
by canals that were maintained at desired water levels via a system of dikes and
trunks. In each impoundment, two rice-field trunks (McGovern and Wenner 1990)
spanned the dikes and were used to lower water levels in the spring to expose
the central marsh, which allowed seed germination and oxidation to occur. The
impoundments were reflooded throughout the summer to reach a depth of ≈0.5 m
above the interior marsh in the fall. Additionally, the trunks remained slightly open
throughout the year, and riser boards placed on the impoundment side of each
trunk maintained the desired water level. This procedure allowed for some water
exchange to limit hypoxia.
Within each impoundment, we chose three sections of the perimeter canal as
our study sites. Site selection was constrained primarily by our ability to sample
by boat, and the sites were as close to equidistant from one another as possible.
During initial site delineation, when water was contained completely in the canals,
Nieuport sites were 0.09, 0.10, and 0.27 ha, respectively, with average
depths of 0.80 (0.11 SD), 1.48 (0.05), and 1.64 m (0.09), respectively. Each site
at BRF was 0.11 ha, and the three sites had average depths of 0.82 (0.12), 1.00
(0.60), and 2.08 m (0.11).
Methods
Data collection
We sampled each impoundment with rotenone in May from 2008 to 20 10 when
water was confined to the canals. For each sampling effort, we blocked off each of
the 6 canal sections with two 6.4-mm mesh block-nets (4.3 m deep x 30.5 or 45.7 m
wide). We used a battery-operated pesticide sprayer to apply liquid rotenone (Prenfish
Toxicant®; Prentiss Incorporated, Floral Springs, NY; 1 ppb) under the surface
within each canal section. We used dip nets (6.4-mm mesh) to collect all fish as they
rose to the surface. Fish collection continued until less than 3 specimens rose to the
surface within a 20-min time period. We applied potassium permanganate (Argent
Chemical Laboratories, Redmond, WA; 3 ppb) within the sampled areas to neutralize
the rotenone. The total area sampled with rotenone represented less than 1.2%
of the total impoundment area in each of the impoundments.
We grouped all collections by species and counted the number of fish per species.
We placed any fish that we could not identify in the field in 95% ethanol and
transported them back to the laboratory for identification. Generally during rotenone
sampling, researchers retrieve fish from the water both the day the rotenone
is applied and again the next day. However, we observed Alligator mississipiensis
(Daudin) (American Alligator) and many species of water birds consuming the
limited fish kill that surfaced outside the block nets and we became concerned
that many of the fish that otherwise would have been collected on the second day
would be consumed. Thus, for this study, we confined our sampling efforts to a first
day pick-up to avoid the sample bias resulting from fish consumption by wildlife.
During each sampling effort, we saw American Alligators consuming fish at all
sample sites, which may have introduced some bias into our sampling. Because
Southeastern Naturalist
241
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
this disturbance occurred in all impoundments and we did not see more than one
American Alligator in any canal section, this effect was probably equal across sites.
We measured the water quality of each canal section within each impoundment
during each sampling event. We measured temperature, salinity, and dissolved
oxygen (DO) with a YSI-85® (Yellow Springs Instruments, Inc., Yellow Springs,
OH) handheld meter. We took all measurements just below the water surface and
reported the averages of the water-quality measurements taken in each of the 3
canal sections within each impoundment. We performed a nested analysis of variance
(ANOVA) on each water-quality variable to analyze differences among impoundments
while taking into account differences among sampling years. When
ANOVAs produced significant differences (α = 0.05), we further analyzed the results
with Tukey’s HSD. We used R (R Development Core Team, 2010) to perform
these analyses.
We used data from a total of 12 fish assemblages sampled from North Carolina
and South Carolina for comparison to our impoundment resident assemblages.
Weinstein et al. (1980) used block nets and either 1.0-mm mesh seines or rotenone
to collect fish-assemblage data from 11 tidal-creek stations in the Cape Fear River
estuary in North Carolina monthly for 20 months. Additionally, Wenner et al.
(1986) performed a study of tidal creek and impoundment fish communities in the
North Inlet estuary of South Carolina. We used the fish assemblage data collected
from their tidal-creek rotenone survey. Wenner et al. (1986) used 0.8-mm mesh
block-nets and rotenone to conduct monthly surveys (n = 17) in three small subtidal
creeks that drained Chainey Creek.
We chose these assemblages for comparison to our impoundment samples for
many reasons. These studies represented two of the few studies available that have
quantified resident fish abundances in North and South Carolina estuaries. We focused
solely on tidal creeks in North and South Carolina because the tidal ranges
in these regions were similar. Florida’s topography, which also contained many
impounded wetlands, created a very different tidal hydrology. The studies we chose
provided the most recent estimations of fish-assemblage abundances in the Carolinas
(Nordlie 2003). Rotenone was used to collect a portion of the fish samples in
these studies, which allowed us to reduce gear bias in our comparisons. Rotenone
sampling has largely been replaced by other methods (e.g., gill nets, electrofishing,
and seines) in recent studies. Additionally, all three of the locations used in
our analyses were located within the same biogeographic region, and salinities and
temperatures recorded in the three river systems were similar (Table 1). Salinities
in the estuarine portion of the Combahee River ranged from 2.0 to 20.7 psu (K.F.
Robinson and C.A. Jennings, unpubl. data; Upchurch and Wenner 2008), whereas
salinities in the North Carolina study ranged from 1.7 to 22.4 psu (Weinstein et
al. 1980) and salinities in the South Carolina study ranged from 0.0 to 31.9 psu
(Wenner et al. 1986). The vegetative communities of the tidal marshes in the three
study locations were dominated by Spartina alterniflora Loisel (Smooth Cordgrass)
and Juncus roemerianus Scheele (Needlegrass Rush) (McGovern and Wenner
1990, Upchurch and Wenner 2008, Weinstein 1979), and the sample stations were
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
242
located in first-order tidal creeks. The study in South Carolina provided data on fish
assemblages sampled adjacent to waterfowl impoundments in a wildlife preserve,
which provided a similarly low level of anthropogenic influence as was observed
in our study sites in the ACE Basin. The estuarine portion of the Cape Fear River
was ≈45 river km in length (Weinstein 1979), which was similar to the estuarine
portion of the Combahee River (≈61 river km; Upchurch and Wenner 2008), and
the Cape Fear River estuary contained the largest tidal salt marsh in North Carolina.
Although these two studies provided pooled abundances of resident fishes from 17
(Wenner et al. 1986) to 20 months (Weinstein et al. 1980) of sampling, the resident
assemblages sampled in these published studies did not vary greatly in species proportions
over the course of sampling.
Assemblage comparisons and model selection
We retained only those species described as permanent resident or freshwater
transient in Nordlie (2003) for all analyses. We calculated a suite of community
indices to describe the 18 fish assemblages from impounded and tidal creek
Table 1. Sampling method, mean salinity, temperature, and DO (1 SD) for each of the natural and
impounded marsh locations sampled for fish-assemblage comparisons. Impoundment samples were
from spring sampling in Nieuport and Big Rice Field (BRF), on the Combahee River, SC, 2008–2010.
For each impoundment, values are the average of point estimates from three sites. The Chainey Creek
values are the range from 24 months of sampling in a tidal creek in North Inlet Estuary, SC. Data
are from the study of Wenner et al. (1986). Averages for Baldhead Creek, Barnard’s Creek, Cape
Creek, Dutchman Creek, Governor’s Creek, Hechtic Creek, Town Creek, and Walden Creek are from
monthly measurements over 20 months from samples in tidal creeks in the Cape Fear River Estuary,
NC. Data are from the study of Weinstein et al. (1980). NA = data not given in original paper.
Sampling
Location Habitat type method Salinity (psu) Temp.(°C) DO (mg l-1)
Nieuport 2008 Impounded Rotenone 5.0 (0.7) 26.7 (1.4) 3.2 (1.5)
Nieuport 2009 Impounded Rotenone 7.5 (0.5) 25.8 (0.5) 9.5 (2.2)
Nieuport 2010 Impounded Rotenone 3.0 (1.3) 31.0 (1.3) 5.8 (4.3)
BRF 2008 Impounded Rotenone 13.0 (1.8) 26.4 (1.7) 8.5 (1.4)
BRF 2009 Impounded Rotenone 5.4 (1.1) 25.1 (1.7) 9.0 (0.9)
BRF 2010 Impounded Rotenone 10.8 (0.2) 27.5 (2.2) 7.4 (1.2)
Chainey Creek Natural Rotenone 0.0–31.9A 4.0–32.4A NA
Baldhead Creek Natural Rotenone 22.0 (6.4) 19.9 (8.4) NA
Barnard’s Creek Natural Rotenone 1.7 (2.3) 17.4 (7.3) NA
Cape Creek Natural Rotenone 22.4 (5.5) 21.1 (8.6) NA
Dutchman Creek Natural Rotenone 14.2 (6.8) 18.8 (7.8) NA
Dutchman Creek Natural Seine 19.3 (5.5) 18.4 (8.1) NA
Governor’s Creek Natural Seine 2.2 (2.3) 19.7 (8.4) NA
Hechtic Creek Natural Seine 2.5 (3.2) 19.0 (7.9) NA
Town Creek Natural Rotenone 2.2 (3.3) 18.1 (9.3) NA
Walden Creek Natural Rotenone 4.8 (3.0) 20.1 (8.4) NA
Walden Creek 1 Natural Seine 5.0 (3.2) 19.9 (8.0) NA
WaldenCreek 2 Natural Seine 4.4 (2.7) 19.3 (8.2) NA
AMean salinity and temperature were not given, only the range of measurements for the entire study
were reported.
Southeastern Naturalist
243
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
habitats. We calculated rarefied species richness curves for each assemblage and
standardized the species richness for each assemblage to 100 individuals because
the smallest number of fish collected at an individual site was 112. Additionally, we
used Shannon’s index (H') to assess assemblage diversity,
s
H' = -Σ (pi)(loge pi),
i = 1
where S = total number of species and pi = the proportion of the total sample represented
by the ith species (Kwak and Peterson 2007). We calculated H' and rarefied
species richness curves with the vegan package (Oksanen et al. 2010) for R (R
Development Core Team 2010). We calculated the evenness of each assemblage,
based on the Shannon index:
H'
J ' =
logeS
We compared rarefied species richness, H', and J' between habitat types (impoundments
versus tidal creeks) with Welch’s two-sample t-tests (α = 0.05).
We used Morisita’s index of similarity and percent similarity to compare the
resident assemblages of our study impoundments with those of the previous studies
of natural marshes. Similarity indices have been shown to diverge from one
another (Bloom 1981); we calculated two separate indices of similarity to account
for this potential divergence. Morisita’s index ranges from 0 (assemblages are
completely dissimilar) to 1 (assemblages are identical). Percent similarity ranges
from 0% (zero species in common) to 100% (assemblages are identical). We chose
these two indices because they have been described as the most robust and best
measures of similarity (Krebs 1998, Kwak and Peterson 2007). Our assemblages
contained size displacements and many species collected in small numbers; unlike
the more widely used Bray-Curtis coefficient, neither of the indices is sensitive to
these characteristics of the data (Kwak and Peterson 2007, Wolda 1981). Additionally,
Morisita’s index provides a measure of similarity in terms of a probability: the
ratio of the probability that an individual drawn from assemblage 1 and an individual
drawn from assemblage 2 will be the same species to the probability that two
individuals drawn from either assemblage 1 or 2 will be the same species (Krebs
1998, Kwak and Peterson 2007). Based on this ratio, we chose similarity cutoffs
for both indices of 0.25 (somewhat similar), 0.50 (moderately similar), and 0.75
(very similar) to describe the similarities of the compared assemblages. Morisita’s
index uses abundance data (total number of individuals) while percent similarity
uses relative species abundances (Kwak and Peterson 2007, Wolda 1981). We
included rare species in the analysis because indices of similarity did not differ
between calculations in which rare species were included or excluded. We used
the vegan package (Oksanen et al. 2010) for R (R Development Core Team 2010)
to calculate Morisita’s index of similarity. We conducted pairwise comparisons of
the resident assemblages from each of three years (2008–2010) of May samples
from Nieuport and BRF, and the 12 assemblages from North Carolina (Weinstein
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
244
et al. 1980) and South Carolina (Wenner et al. 1986). The output from R for the
calculation of Morisita’s index provided the dissimilarity index, so we subtracted
all calculated indices from one to obtain the similarity index. For brevity, we refer
to each impoundment collection by the first letter of the impoundment (B or N) and
the last two digits of the collection year (08, 09, or 10).
We created a set of hypotheses and a corresponding candidate set of linear
models that might explain the observed similarity indices, and chose to use
Morisita’s index for this analysis because it provides similarities in terms of a
probability. The models used linear regression analysis (Neter et al. 1990) to
describe the relationships among the observed Morisita’s indices and the following
variables: the absolute difference in average salinity between compared sites
(maximum recorded salinity was used for the data from Wenner et al. 1986), a
categorical variable to describe differences in fish-collection methods (0 = same
methods between compared sites, 1 = different methods between compared sites),
and a categorical variable to describe differences in tidal influence between sites
(0 = both sites impounded or natural, 1 = impounded to natural site comparison).
We used an information theoretic approach to assess the relative fit of each candidate
model. The global model, which contained all of the variables thought to
influence assemblage similarity, was used to create the candidate set of models,
which included all combinations of the above variables. We assessed this global
model for goodness-of-fit by examination of the residuals and normal probability
plots. The pairwise comparisons of sites might not be independent because all
pairs that include a particular site may have similar Morisita values, which would
require the use of hierarchical linear models. An analysis of variance (ANOVA)
of the residuals of the global linear model by site was not significant (P = 0.61),
which indicated that the use of a linear model was appropriate in our case. We
evaluated the fit of each model in the candidate set by calculating Akaike’s information
criterion (AIC; Akaike 1973) with the small-sample bias adjustment
(AICc; Hurvich and Tsai 1989). We assessed Pearson correlations for all pairs of
predictor variables to avoid multicollinearity, and used only uncorrelated predictor
variables (r < 0.40) in the candidate models (all predictor variables were uncorrelated
in this case; Moore and McCabe 1993). We calculated Akaike weights
(wi, range = 0–1; Burnham and Anderson 2002) to assess the relative fit of each of
the candidate models; the best-fitting model had the highest Akaike weight.
We incorporated model-selection uncertainty by calculating the modelaveraged
estimates of the linear regression model coefficients and standard
errors (Burnham and Anderson 2002). We created a confidence set of models
that contained all models whose weights were within 10% of the highest weight
(Thompson and Lee 2000). The coefficients of the estimates and their standard
errors were weighted by their Akaike weights (wi) and summed across the confidence
set of models to obtain the composite model. Our interpretations were
based on this composite model. We calculated 95% confidence intervals based
on the t-statistic with n - 1 degrees of freedom to assess the precision of our
model-averaged coefficient estimates. Any confidence intervals that spanned zero
Southeastern Naturalist
245
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
indicated that the associated parameter’s relationship with similarity indices (positive
or negative) could not be inferred. All statistical analyses were performed in
R (R Development Core Team 2010).
Results
Data collection
The average spring salinity in our study impoundments ranged from 3.0 psu
(1.3 SD) to 13.0 psu (1.8) throughout the three years of fish collections (Table 1).
Salinity ranged from 1.7 psu (2.3) to 19.3 psu (5.5) in the North Carolina study
areas (Weinstein et al. 1980) and from 0.0 psu to 31.9 psu in the South Carolina
study areas (Wenner et al. 1986). Temperature ranged from 25.1 °C (1.7) to 31.0 °C
(1.3) in our study impoundments, from 17.4 °C (7.3) to 21.1 °C (8.6) in the North
Carolina tidal creeks (Weinstein et al. 1980), and from 4.0 °C to 32.4 °C in the
South Carolina tidal creek (Wenner et al. 1986). DO ranged from 3.2 mg/L (1.5) to
9.5 mg/L (1.5) in our study impoundments, but DO was not reported for the other
study areas. Nieuport and Big Rice Field were statistically similar with respect to
all water-quality variables, as indicated by the results of the nested ANOVAs for
each variable (salinity [F = 3.02, df = 1, P = 0.157], temperature [F = 0.71, df = 1,
P = 0.447], DO [F = 1.30, df = 1, P = 0.318]). Salinity varied among sampling years
(F = 26.43, df = 4, P < 0.001), as did temperature (F = 5.88, df = 4, P = 0.007). DO
did not vary among years (F = 3.20, df = 4, P = 0.053).
We collected 7946 resident fishes representing 12 species in Nieuport and
1258 resident fishes representing 15 species in BRF (Table 2). The most abundant
resident species varied among years in each impoundment, though four species—
Table 2. Numbers of resident species captured during spring rotenone sampling in Nieuport and Big
Rice Field impoundments in the Combahee River, SC, 2008–2010.
Big Rice Field Nieuport
Species Common name 2008 2009 2010 2008 2009 2010
Lepisosteus osseus (L.) Longnose Gar 2 4 7 2 18 10
Myrophis punctatus Lütken Speckled Worm-eel 0 0 2 0 0 0
Dorosoma cepedianum (Lesueur) Gizzard Shad 1 3 0 4 7 14
Carassius auratus (L.) Goldfish 0 154 0 0 0 0
Cyprinus carpio L. Common Carp 0 2 0 0 1 0
Notemigonus crysoleucas (Mitchill) Golden Shiner 0 0 0 0 0 1
Ameiurus catus (L.) White Catfish 0 0 1 1 13 24
Ameiurus nebulosus (Lesueur) Brown Bullhead 0 0 3 0 0 0
Menidia beryllina (Cope) Inland Silverside 54 56 42 461 683 224
Fundulus heteroclitus L. Mummichog 0 0 1 0 0 1
Lucania parva (Baird and Girard) Rainwater Killifish 4 0 15 3 0 1115
Gambusia holbrooki Girard Eastern Mosquitofish 16 203 35 272 117 346
Poecilia latipinna (Lesueur) Sailfin Molly 132 501 5 89 3365 1062
Cyprinodon variegatus (Lacepède) Sheepshead Minnow 1 1 0 0 40 72
Syngnathus scovelli (Evermann and Gulf Pipefish 10 1 0 0 1 0
Kendall)
Gobiosoma bosc (Lacepède) Naked Goby 0 1 1 0 0 0
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
246
Gambusia holbrooki (Eastern Mosquitofish), Lucania parva (Rainwater Killifish),
Poecilia latipinna (Sailfin Molly), and Menidia beryllina (Inland Silverside)—
tended to dominate in our samples. In Nieuport, Inland Silverside had the greatest
density (0.08/m2) in 2008, though overall, all species had low densities that year.
Sailfin Molly had the greatest density (0.99/m2) in Nieuport in 2009, and Rainwater
Killifish (0.38/m2) and Sailfin Molly (0.32/m2) dominated samples of resident
species in 2010 (Fig. 2a). In BRF, Sailfin Molly had the greatest density in 2008
(0.04/m2) and 2009 (0.16/m2), whereas Eastern Mosquitofish (0.01/m2) and Inland
Silverside (0.01/m2) had the greatest densities in 2010 (Fig. 2b).
Figure 2. Densities (±1 SE) of the four most abundant resident fishes—Gambusia holbrooki
(Eastern Mosquitofish), Lucania parva (Rainwater Killifish), Poecilia latipinna (Sailfin
Molly), and Menidia beryllina (Inland Silverside)— collected in spring rotenone samples in
a) Nieuport and b) Big Rice Field impoundments, Combahee River, SC, 2008 – 2010. Note
the difference in y-axis ranges in the two plots.
Southeastern Naturalist
247
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
Assemblage comparisons
Rarefied species richness (n = 100) ranged from 1.0 to 9.7 species (Fig. 3a).
Mean rarefied species richness was significantly greater in impoundments (6.0
[2.0 SD]), than in tidal creeks (3.4 [1.9]; t = 2.6, df = 9.58, P = 0.026). H' diversity
ranged from 0.00 to 1.60 (Fig. 3b), and J' evenness ranged from 0.16 to 0.93
(Fig. 3c). Mean diversity was significantly greater in impoundments (1.17 [0.31])
Figure 3. a) Rarefied species richness (n = 100), b) Shannon (H') diversity, c) J' evenness
calculated for fish assemblages captured during spring rotenone sampling in Big Rice
Field (B) and Nieuport (N) impoundments, Combahee River, SC, 2008–2010, and fish assemblages
sampled with seine (S) and rotenone (R) for previous studies of tidal creeks in
Cape Fear River estuary, NC, (Hechtic, Governor’s, Walden, Dutchman, Barnard’s, Town,
Cape, and Baldhead creeks) and North Inlet estuary, SC, (Chainey Creek). Evenness was
not calculated for Dutchman (R) or Walden (R) because only one species was collected in
these sites. Data from North Carolina are from the study of Weinstein et al. (1980) and data
from South Carolina are from the study of Wenner et al. (1986).
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
248
than in tidal creeks (0.68 [0.41]; t = 2.8, df = 12.82, P = 0.016), but evenness did
not differ between habitat types (impoundments = 0.53 [0.13], tidal creeks = 0.55
[0.32]; t = 0.1, df = 12.84, P = 0.899).
Similarity indices for comparisons of resident fish assemblages among impoundments
ranged from 0.23 to 0.96 for Morisita’s index and from 24 to 80% for
percent similarity (Appendices A, B). Overall, 40.0% (percent similarity) to 53.3%
(Morisita’s index) of the among-impoundment comparisons were at least moderately
similar (Table 3). Comparisons among all tidal creek assemblages, regardless
of sampling method, ranged from 0.39 to 1.00 for Morisita’s index and from 32 to
100% for percent similarity (Appencies C, D). Overall, 63.6% (percent similarity)
to 85.7% (Morisita’s index) of the comparisons among all tidal creek assemblages
were at least moderately similar (Table 3).
Resident fish assemblages sampled with rotenone in impoundments and natural
marshes were dissimilar; only 2.4% (percent similarity) to 4.8% (Morisita’s
index) of these comparisons were at least moderately similar (Table 3). The exceptions
were the comparisons between N08 and Town Creek (Morisita index = 0.75,
percent similarity = 55%), and B10 and Town Creek (Morisita index = 0.62) (Appendices
E, F). In contrast, 52.4% (percent similarity) to 76.2% (Morisita index) of
the comparisons among rotenone-sampled natural marsh assemblages were at least
moderately similar (Table 3). When comparing the rotenone-sampled resident assemblages
in impoundments to natural marsh assemblages sampled by seine, 6.7%
of the comparisons (Morisita’s index and percent similarity) were moderately similar.
N08 (Morisita index = 0.83, percent similarity = 68%) and B10 (Morisita index
= 0.79, percent similarity = 61%) were moderately to very similar to the Governor’s
Creek resident assemblage (Apendices E, F). Conversely, 70.0% (percent similarity)
to 90.0% (Morisita’s index) of the comparisons among tidal creek assemblages
Table 3. Percent of fish-assemblage comparisons using Mosita’s index and percent similarity index:
Imp = among our study impoundments (Big Rice Field and Nieuport, Combahee River, SC); Imp–
NM = between our study impoundments and natural marsh habitat from previously published studies
(Cape Fear River estuary, NC, and North Inlet estuary, SC); and NM = among natural marshes from
previously published studies, whose similarity indices fall in the categories of ≥0.75, 0.50–0.75,
0.25–0.50, and 0.00–0.25 for samples collected by rotenone and seine. All NM = the comparisons of
marshes sampled by both rotenone and seine.
Rotenone Seine
Index Imp Imp–NM NM Imp–NM NM All NM
Morisita index
≥0.75 26.7 2.4 47.6 6.7 60.0 56.1
0.50–0.75 26.7 2.4 28.6 0.0 30.0 27.3
0.25–0.50 40.0 2.4 23.8 3.3 10.0 16.7
0.00–0.25 6.7 92.9 0.0 90.0 0.0 0.0
Percent similarity index
≥75% 6.7 0.0 4.8 0.0 30.0 18.2
50–75% 33.3 2.4 47.6 6.7 40.0 45.5
25–50% 53.3 2.4 47.6 6.7 30.0 36.4
0–25% 6.7 95.2 0.0 86.7 0.0 0.0
Southeastern Naturalist
249
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
that were sampled with seines were moderately similar (Table 3). Among all comparisons,
the observed percent similarities were slightly lower than the Morisita’s
indices, but the patterns of similarity were the same between the two methods (Appendices
A–F).
Model selection
The model containing only a difference in tidal influence was the most plausible
linear regression model explaining the observed Morisita’s indices (wi = 0.765; Table
4). The model of a difference in tidal influence was 6.1 times more likely to provide
the best explanation for similarity indices than was the global model, which was the
next most plausible model. These two models and the model that considered a difference
in tidal influence and salinity made up the confidence set of models; the other
models did not have sufficient evidence to be considered as plausible explanations
for differences among resident fish assemblages. Because the global model was a part
of the confidence set of models, all parameters were included in the composite model
(intercept = 0.72, 95% CI = 0.67–0.77; coefficients: tidal influence = -0.63, 95% CI =
-0.69 to -0.57; method = 0.07, 95% CI = -0.08–0.21; salinity = 0.00, 95% CI = 0.00–
0.01). The observed Morisita’s index of similarity values were most affected by tidal
influence. The estimated coefficient for tidal influence indicated that Morisita’s index
of similarity decreased by 0.63 when comparisons were made between impounded
and natural marshes. Sampling method had a very weak positive influence on the
index of similarity (0.07), but the confidence intervals spanned zero, which indicates
that comparisons between our samples collected by rotenone and natural marsh assemblages
collected by seine were valid. Salinity appeared to have very little to no
influence on assemblage similarity.
Discussion
The results of our study indicated that, as expected, the resident fish assemblages
in our study impoundments were dissimilar to resident assemblages in natural
tidal creek systems along the southeastern coast of the United States. In addition
to the habitat-specific differences indicated by the estimates of two indices of
similarity, rarefied species richness and H' diversity were significantly greater in
the impoundments than the tidal creeks. Poeciliids were the most abundant species
in the study impoundments in all samples except N08 and B10, whereas Fundulus
Table 4. Candidate models of parameter influences on Morisita’s index of similarity values obtained
for comparisons of resident fish assemblages in impounded and natural wetlands, number of parameters
in each model (K), deviations from best-fitting model (Δ i), and Akaike weights (wi).
Model K Δi wi
Difference in tidal influence 3 0.000 0.765
Differences in tidal influence, salinity and method 5 3.629 0.125
Differences in tidal influence and salinity 4 3.864 0.111
Difference in method 3 194.885 0.000
Difference in salinity 3 194.567 0.000
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
250
heteroclitus (L.) (Mummichogs) were most abundant in all but two (Governor’s and
Town creeks) of the natural marsh assemblages (Weinstein et al. 1980, Wenner et
al. 1986). Inland Silverside was most abundant in N08, B10, and Governor’s and
Town creeks, and these four assemblages ranged in similarity (Morisita’s index)
from 0.62 to 0.93.
Results of our model-selection analysis showed that the index of similarity between
two resident fish assemblages decreased by more than half of the range of the
index when there was a difference in tidal influence. We hypothesized that salinity
would also play a role in the observed differences among fish assemblages, but the
results of our model-selection analysis did not support this expectation. Our results
corroborated the assertion of Weinstein et al. (1980) that gear bias was negligible
for seine- and rotenone-sampled fish assemblages collected in Cape Fear River tidal
creeks because 85.7% of the comparisons among tidal creeks had Morisita’s index
values ≥0.50.
The results of our model-selection analysis indicated that tidal influence was
the factor driving differences in fish-assemblage structure between our study
impoundments and natural marsh habitats in North and South Carolina. Tidal influence
was drastically reduced in the impoundments because managers held the
interior marsh of these structures in the early stages of the drought-recovery cycle
that is characteristic of temporary wetlands (Batzer et al. 2006). This reduction in
tidal influence can alter estuarine-community structure by changing colonization/
extinction dynamics and by shifting the dominant vegetation in the impoundments
from emergent-marsh vegetation to submerged aquatic vegetation (SAV; Rozas and
Minello 1999, Weaver and Holloway 1974).
The difference in tidal influence manifests partially as an issue of connectivity.
The lack of connectivity of our impoundments to the salt marsh for most of the year
means that colonization might be an important factor influencing impoundment
resident-fish assemblage structure. Species had a brief window of time to enter impoundments
during fall flooding, and species that were most abundant during this
time immigrated into the impoundments in greater numbers than those that were less
abundant (Rogers et al. 1992, Wenner et al. 1986). Connectivity, through reduced
tidal influence, is a large-scale process that helps determine the local species available
for establishment in impoundments (Baber et al. 2002, Snodgrass et al. 1996,
Taylor 1997). This process provides a potential explanation for why salinity was
not driving the difference in assemblage structure between residents in impounded
and natural marsh habitats: it likely was overshadowed by tidal influence. For example,
the bulk of the tidal creek assemblages were dominated by Mummichogs
(Weinstein et al. 1980, Wenner et al. 1986), which were almost completely absent
from the impoundment resident assemblages. Mummichogs are a dominant member
in the tidal creek communities of the ACE Basin. In a study in the Ashepoo River
(part of the ACE Basin) concurrent with our impoundment sampling, Goldman et
al. (2010) used minnow traps set in tidal creeks to capture >140,000 Mummichogs
during a 9-mo period. Their catch per unit effort was lowest in September, which
is typically when impoundments are flooded. In addition, we captured very few
Southeastern Naturalist
251
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
Mummichogs in baited minnow traps set in tidal creeks across the river from Nieuport
and Big Rice Field in September, whereas this species dominated springtime
trap-sets in these low salinity tidal creeks (2–6 psu; K.F. Robinson and C.A. Jennings,
unpubl. data). If this seasonal pattern of Mummichog abundance is typical
of the ACE Basin, the lack of Mummichogs in our study impoundments could be
a direct result of the lack of diel tidal exchange. The current water-management
regime could be selecting against the most dominant member of the natural marsh
resident-fish community.
In conjunction with colonization, extinction-like processes also might have influenced
the structure of the resident fish assemblages in our study impoundments.
In the summer, the majority of the fish in these structures perish (Robinson 2011).
This large-scale mortality can be attributed to poor-quality habitat (e.g., declining
DO levels) and predation by birds and other aquatic animals (Robinson 2011). Conditions
in natural marsh habitats can be stressful for most fish species, but unlike
fish in natural marshes, fish in impoundments are unable to emigrate out of harsh
conditions (Rogers et al. 1992, Wenner et al. 1986). This yearly disturbance created
an assemblage structure with characteristics of a community living in a frequently
disturbed habitat (Menge and Sutherland 1976, Poff and Ward 1989, Snodgrass et
al. 1996, Wiens 1984). In frequently disturbed communities, biotic interactions are
overshadowed by the disturbance (Snodgrass et al. 1996) and colonization–extinction
dynamics become important.
In addition to encouraging colonization–extinction dynamics, the water-level
management scheme practiced in the study impoundments used reduced tidal
influence to replace the emergent-marsh vegetation found in natural salt marshes
with SAV, mostly Widgeongrass. SAV is one of the most important variables in
predicting relative abundances of small resident fish species such as Sailfin Molly
and Eastern Mosquitofish in natural marsh and impounded habitats (Castellanos
and Rozas 2001, Harrington and Harrington 1982, Kanouse et al. 2006, Rozas and
Minello 1999). In the study impoundments, four species were most abundant each
year: Eastern Mosquitofish, Sailfin Molly, Rainwater Killifish, and Inland Silverside.
These species have been described as rapid colonizers of habitat because of
their omnivorous feeding habits and ability to withstand stressful conditions such
as hypoxia and extremes in salinity and temperature (Felley and Daniels 1992,
Martin et al. 2009, Meffe and Snelson 1989, Ruetz et al. 2005). In numerous studies
in impounded wetlands as well as in low-energy Gulf of Mexico marshes, Eastern
Mosquitofish (Kanouse et al. 2006, Rozas and Minello 1999, Weaver and Holloway
1974), Sailfin Molly (Kanouse et al. 2006, Rogers et al. 1994, Rozas and Minello
1999, Weaver and Holloway 1974), and Rainwater Killifish (Castellanos and Rozas
2001, Rogers et al. 1994, Rozas and Minello 1999) were much more abundant in
SAV than in emergent marsh or open water habitats.
Although we did not quantify SAV in the study impoundments, some interesting
trends in the fish assemblages among years follow our qualitative observations
of SAV (Widgeongrass) cover. The resident fish assemblages in Nieuport in 2008
and Big Rice Field in 2010 were more similar to each other (0.93) than they were
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
252
to any other impoundment sample (0.23–0.53). These two assemblages had lower
resident-fish densities and were uniquely dominated by Inland Silverside. We observed
much lower cover of SAV in these two impoundments in 2008 and 2010.
Inland Silverside prefers emergent-marsh vegetation to SAV (Castellanos and Rozas
2001) and has been classified as an edge-marsh user because it rarely moves
into interior marshes (Peterson and Turner 1994). The proportion of interior and
edge marsh, as well as the amount of SAV in an impoundment, probably influenced
which of the suite of stress-tolerant species was most dominant in a given year and
helped structure the resident assemblages in impounded wetlands .
Each year, we observed gravid females of each of the four species most common
in our impoundments— Eastern Mosquitofish, Sailfin Molly, Rainwater Killifish,
and Inland Silverside—which suggests that relative abundances of these species
may change because of differences in reproduction and not solely through stresstolerance
and immigration. To reproduce, Mummichogs require access to marsh
pools that are accessible only on very high tides (Taylor et al. 1979), and, therefore,
any Mummichogs that entered the impoundments likely could not reproduce, further
reducing the relative abundance of the species there. In the future, examining
a larger suite of environmental characteristics in conjunction with fish-assemblage
sampling may lead to a better understanding of what factors drive the change in
resident-assemblage structure that we observed in impounded wetlands as compared
to natural marsh habitat.
Although the natural marsh studies that we used for our comparisons were
older than we would have preferred, the temporal difference likely did not have
a major influence on our results. The high degree of similarity observed for
comparisons between North Carolina resident assemblages and the assemblage
sampled from Chainey Creek, SC, (Morisita index = 0.65–0.94) indicated that
resident assemblages from North and South Carolina were comparable, despite
the studies being separated by six years and over 100 miles of coastline. Additionally,
these two studies represented the most recent assessments of estuarine fish
assemblage structure in the Carolinas. Unlike most studies of fish in these habitats,
the authors of our comparison studies provided complete descriptions of the
fish assemblages; these same papers were recently used to characterize estuarine
fish assemblages in the eastern United States (Nordlie 2003). These studies also
provided descriptions of southeastern estuarine fish assemblages from relatively
pristine habitats. Although the assemblages in specific locations may have been
subject to anthropogenic influences in recent years, when these studies were performed,
the habitat structure was quite similar to the current state of estuarine
habitat in the relatively pristine ACE Basin.
The effects of changes in hydrologic connections on estuarine fishes in the
southeastern US are understudied. Our results indicate that reduced connectivity
and tidal influence can greatly change the composition of resident fish assemblages.
Reduced connectivity probably changes the assemblage structure through a combination
of reduced migration and alterations in the environment that favor certain
species over others. We found that four small resident species comprised the bulk of
Southeastern Naturalist
253
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
the fish within these impoundments and that different species dominated in different
years. As habitat fragmentation in estuarine systems along the Atlantic and Gulf
coasts of the United States intensifies, the habitat characteristics that create these
assemblage-level changes within managed wetlands may become more prevalent in
other fragmented marshes. Fragmentation through reduced hydrologic connectivity
can affect fish assemblages beyond changes in resident-species composition. For
example, these systems act as sinks for many species that use estuaries as nursery
grounds (Robinson and Jennings 2012, Wenner et al. 1986). These differences between
natural and altered estuarine habitats can result in reduced productivity and
reduced export of nutrients from estuaries to coastal waters (Kneib 1997, Robinson
2011). For these reasons, future research seeking to understand the dynamics of
species assemblages within impounded wetlands may provide us with valuable
information about fish-community dynamics in other fragmented sy stems.
Acknowledgments
We thank M. Alber, J. Peterson, and S. Schweitzer for input into experimental design
and comments on this manuscript. We also thank B. Carswell, G. Crouch, P. Dimmick, J.
Dycus, P. Ely, M. Homer, J. Kirsch, E. Mills, M. Mundy, R. Peterson, J. Robinson, J. Ruiz,
E. Wiggers, and S. Zimpfer for assistance in field collections and logistical support. Further,
we thank A. Overton, J. Robinson, and A. Rypel, and two anonymous reviewers for helpful
comments on earlier drafts of this manuscript. This research was supported with a grant
from the National Fish and Wildlife Foundation. Any use of trade, product, or firm names is
for descriptive purposes only and does not imply endorsement by the US Government. The
Georgia Cooperative Fish and Wildlife Cooperative Research Unit is sponsored jointly by
Georgia Department of Natural Resources, the US Fish and Wildlife Service, the US Geological
Survey, and the Wildlife Management Institute. This study was performed under the
auspices of the University of Georgia protocol #2009-3-060.
Literature Cited
Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle.
Pp. 267–281, In B.N. Petrov and F. Csaki (Eds.). Second international symposium on
information theory. Akademai Kiado, Budapest, Hungary. 451 pp.
Baber, M.J., D.L. Childers, K.J. Babbitt, and D.H. Anderson. 2002. Controls on fish distribution
and abundance in temporary wetlands. Canadian Journal of Fisheries and Aquatic
Science 59:1441–1450.
Batzer, D.P., R. Cooper, and S.A. Wissinger. 2006. Wetland animal ecology. Pp. 242–284,
In D.P. Batzer and R.R. Sharitz (Eds.). Ecology of Freshwater and Estuarine Wetlands.
University of California Press, Berkeley, CA. 568 pp.
Bloom, S.A. 1981. Similarity indices in community studies: Potential pitfalls. Marine Ecology
Progress Series 5:125–128.
Burnham, K.P., and D.R. Anderson. 2002. Model Selection and Inference: A Practical Information
Theoretic Approach. Springer-Verlag, New York, NY. 488 pp.
Castellanos, D.L., and L.P. Rozas. 2001. Nekton use of submerged aquatic vegetation,
marsh, and shallow unvegetated bottom in the Atchafalaya River Delta, a Louisiana tidal
freshwater ecosystem. Estuaries 24:184–197.
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
254
DeVoe, M.R., D.S. Baughman, and J.M. Dean. 1987. South Carolina’s wetland impoundments:
A summary of research and policy issues. Pp. 487–497, In W.R. Whitman and
W.H. Meredith (Eds.). Proceedings of a symposium on waterfowl and wetlands management
in the coastal zone of the Atlantic flyway. Delaware Coastal Management
Program, Delaware Department of Natural Resources and Environmental Control,
Dover, DE. 522 pp.
Felley, J.D., and G.L. Daniels. 1992. Life history of the Sailfin Molly (Poecilia latipinna) in
two degraded waterways of southwestern Louisiana. Southwestern Naturalist 37:16–21.
Goldman, S.F., A.R. Rourk, M.R. Collins, and M.J.M. Reichert. 2010. Effects of simulated
commercial harvest on Mummichogs in tidal creeks in South Carolina, USA. North
American Journal of Fisheries Management 30:921–927.
Harrington, R.W., Jr., and E.S. Harrington. 1982. Effects on fishes and their forage organisms
of impounding a Florida salt marsh to prevent breeding by salt marsh mosquitos.
Bulletin of Marine Science 32:523–531.
Herke, W.H., E.E. Knudsen, P.A. Knudsen, and B.D. Rogers. 1992. Effects of semi-impoundment
of Louisiana marsh on fish and crustacean-nursery use and export. North
American Journal of Fisheries Management 12:151–160.
Hoese, H.D., and M. Konikoff. 1995. Effects of marsh management on fisheries organisms:
The compensatory adjustment hypothesis. Estuaries 18(1A):180–18 7.
Hurvich, C.M., and C. Tsai. 1989. Regression and time-series–model selection in small
samples. Biometrika 76:297–307.
Joyner, D.F. 2007. Characterizing the hydrography and hydrodynamics of breached former
tidal rice field wetlands on the Cooper River, South Carolina. M.Sc. Thesis. College of
Charleston, Charleston, SC. 241 pp.
Kanouse, S., M.K. La Peyre, and J.A. Nyman. 2006. Nekton use of Ruppia maritima and
non-vegetated bottom habitat types within brackish marsh ponds. Marine Ecology Progress
Series 327:61–69.
Kelley, B.J. 1999. South Carolina rice fields. The Citadel, Charleston, SC. Available online
at http://www.citadel.edu/computing/mm/test/cb/survey/index.html. Accessed 1 April
2010.
Kneib, R.T. 1997. The role of tidal marshes in the ecology of estuarine nekton. Oceanography
and Marine Biology: Annual Review 35:163–220.
Krebs, C.J. 1998. Ecological Methodology. Benjamin-Cummings, Menlo Park, CA. 620 pp.
Kwak, T.J., and J.T. Peterson. 2007. Community indices, parameters, and comparisons. Pp.
677–763, In C.S. Guy and M.L. Brown (Eds.). Analysis and Interpretation of Freshwater
Fisheries Data. American Fisheries Society, Bethesda, MD. 961 pp.
Martin, S.B., A.T. Hitch, K.M. Purcell, P.L. Klerks, and P.L. Leberg. 2009. Life-history
variation along a salinity gradient in coastal marshes. Aquatic Biology 8:15–28.
McGovern, J.C., and C.A. Wenner. 1990. Seasonal recruitment of larval and juvenile fishes
into impounded and non-impounded marshes. Wetlands 10:203–221.
Meffe, G.K., and F.F. Snelson. 1989. An ecological overview of poeciliid fishes Pp. 13–31,
In G.K. Meffe and F.F. Snelson (Eds.). Ecology and Evolution of Livebearing Fishes
(Poeciliidae). Prentice Hall, Englewood Cliffs, NJ. 453 pp.
Menge, B.A., and J.P. Sutherland. 1976. Species diversity gradients: Synthesis of the role of
predation, competition, and temporal heterogeneity. American Naturalist. 110:351–369.
Milgarese, J.V., and P.A. Sandifer. 1982. An ecological characterization of South Carolina
wetland impoundments. Technical report number 51. Marine Resources Research
Institute, South Carolina Wildlife and Marine Resources Department, Charleston, SC.
132 pp.
Southeastern Naturalist
255
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
Moore, D.R., and G.P. McCabe. 1993. Introduction to the Practice of Statistics. W.H. Freeman,
New York, NY. 854 pp.
Neter, J., W. Wasserman, and M.H. Kutner. 1990. Applied Linear Statistical Models. R.D.
Irwin, Homewood, IL. 1152 pp.
Nordlie, F.G. 2003. Fish communities of estuarine salt marshes of eastern North America,
and comparison with temperate estuaries of other continents. Reviews in Fish Biology
and Fisheries 13:281–325.
Oksanen, J., F.G. Blanchet, R. Kindt, P. Legendre, R.B. O'Hara, G.L. Simpson, P. Solymos,
M. Henry, H. Stevens, and H. Wagner. 2010. Vegan: Community Ecology Package. R
package version 1.17-2. Available online at http://CRAN.R-project.org/package=vegan.
Accessed 1 May 2010.
Peterson, G.W., and R.E. Turner. 1994. The value of salt marsh edge vs. interior as a habitat
for fish and decapod crustaceans in a Louisiana tidal marsh. Est uaries 17:235–262.
Poff, N.L., and J.V. Ward. 1989. Implications of streamflow variability and predictability
for lotic community structure: A regional analysis of streamflow patterns. Canadian
Journal of Fisheries and Aquatic Science 46:1805–1818.
Portnoy, J.W. 1999. Salt marsh diking and restoration: Biogeochemical implications of
altered wetland hydrology. Environmental Management 24:111–120.
R Development Core Team. 2010. R: A language and environment for statistical computing.
R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0. Available
online at http://www.R-project.org. Accessed 1 January 2010.
Robinson, K.F. 2011. A comparison of the fish communities in managed and unmanaged
wetlands in coastal South Carolina. Ph.D. Dissertation, University of Georgia, Athens,
GA. 159 pp.
Robinson, K.F., and C.A. Jennings. 2012. Maximizing age-0 Spot export from a South
Carolina estuary: An evaluation of coastal impoundment management alternatives via
structured decision-making. Marine and Coastal Fisheries: Dynamics, Management, and
Ecosystem Science 4:156–172.
Rogers, B.D., W.M. Herke, and E.E. Knudsen. 1992. Effects of three different water-control
structures on the movements and standing stocks of coastal fishes and macrocrustaceans.
Wetlands 12:106–120.
Rogers, D.R., B.D. Rogers, and W.H. Herke. 1994. Structural marsh management effects on
coastal fishes and crustaceans. Environmental Management 18:351– 369.
Rozas, L.P., and T.J. Minello. 1999. Effects of structural marsh management on fishery
species and other nekton before and during a spring drawdown. Wetlands Ecology and
Management 7:121–139.
Ruetz, C.R., III, J.C. Trexler, F. Jordan, W.F. Loftus, and S.A. Perry. 2005. Population
dynamics of wetland fishes: Spatio-temporal patterns synchronized by hydrological
disturbance? Journal of Animal Ecology 74:322–332.
Snodgrass, J.W., A.L. Bryan, R.F. Lide, and G.M. Smith. 1996. Factors affecting the occurrence
and structure of fish assemblages in isolated wetlands of the upper coastal plain,
USA. Canadian Journal of Fisheries and Aquatic Science 53:443–454.
Talbot, C.W., K.W. Able, and J.K. Shisler. 1986. Fish species composition in New Jersey
salt marshes: Effects of marsh alterations for mosquito control. Transactions of the
American Fisheries Society 115:269–278.
Taylor, C.M. 1997. Fish species richness and incidence patterns in isolated and connected
stream pools: Effect of pool volume and spatial position. Oecologia 1 10:560–566.
Taylor, M.H., G.J. Leach, L. DiMichele, W.M. Levitan, and W.F. Jacob. 1979. Lunar spawnSoutheastern
Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
256
ing cycle in the Mummichog, Fundulus heteroclitus, (Pisces: Cyprinodontidae). Copeia
2:291–297.
Thompson, W.L., and D.C. Lee. 2000. Modeling relationships between landscape-level attributes
and snorkel counts of Chinook Salmon and Steelhead parr in Idaho. Canadian
Journal of Fisheries and Aquatic Science 57:1834–1842.
Tiner, R.W, Jr. 1977. An inventory of South Carolina’s coastal marshes. Office of Conservation
and Management, South Carolina Wildlife and Marine Resources Department,
Charleston, SC. 33 pp.
Upchurch, S., and E. Wenner. 2008. Fish and decapod crustacean assemblages from the
Ashepoo-Combahee-Edisto Basin, South Carolina (1993–1999). Journal of Coastal
Research S1(55):200–213.
Weaver, J.E., and L.F. Holloway. 1974. Community structure of fishes and macrocrustaceans
in ponds of a Louisiana tidal marsh influenced by weirs. Contributions in Marine
Science 18:58–69.
Weinstein, M.P. 1979. Shallow marsh habitats as primary nurseries for fishes and shellfishes,
Cape Fear River, North Carolina. Fishery Bulletin 77:339–357.
Weinstein, M.P., S.L. Weiss, and M.F. Walters. 1980. Multiple determinants of community
structure in shallow marsh habitats, Cape Fear River Estuary, North Carolina, USA.
Marine Biology 58:227–243.
Wenner, C.A., J.C. McGovern, R. Martore, H.R. Beatty, and W.A. Roumillat. 1986. Ichthyofauna.
Pp. 415–526, In M.R. DeVoe and D.S. Baughman (Eds.). South Carolina
Coastal Wetland Impoundments: Ecological Characterization, Management, Status, and
Use. Volume II: Technical synthesis. Publication No. SC-SG-TR-86-2. South Carolina
Sea Grant Consortium, Charleston, SC. 611 pp.
Wiens, J.A. 1984. On understanding a non-equilibrium world: Myth and reality in community
patterns and processes. Pp. 439–457, In D.R. Strong, D. Simberloff, L.G. Abele,
and A.B. Thistle (Eds.). Ecological Communities: Conceptual Issues and Evidence.
Princeton University Press, Princeton, NJ. 613 pp.
Wolda, H. 1981. Similarity indices, sample size, and diversity . Oecologia 50:296–302.
Southeastern Naturalist
257
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
Appendix A. Morisita’s index of similarity calculated for comparisons of resident fish assemblages
captured during spring rotenone sampling in Big Rice Field (BRF) and Nieuport
impoundments, Combahee River, SC, 2008–2010. Values followed by a = similarity ≥0.75;
b = similarity is 0.50–0.75.
Nieuport Nieuport
BRF 2008 BRF 2009 BRF 2010 2008 2009
BRF 2009 0.89a
BRF 2010 0.42 0.37
Nieuport 2008 0.53b 0.41 0.93a
Nieuport 2009 0.96a 0.87a 0.23 0.34
Nieuport 2010 0.70b 0.68b 0.48 0.34 0.64b
Appendix B. Percent similarity index calculated for comparisons of resident fish assemblages
captured during spring rotenone sampling in Big Rice Field (BRF) and Nieuport
impoundments, Combahee River, SC, 2008–2010. Values followed by a = similarity ≥0.75;
b = similarity is 0.50–0.75.
Nieuport Nieuport
BRF 2008 BRF 2009 BRF 2010 2008 2009
BRF 2009 68.4
BRF 2010 39.0 33.0
Nieuport 2008 43.6 39.2 73.9b
Nieuport 2009 79.9a 63.7b 24.0 30.1
Nieuport 2010 55.2b 55.9b 38.9 31.8 49.3
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
258
Appendix C. Morisita’s index of similarity calculated for comparisons of fish assemblages sampled in previous studies of tidal creeks in
the Cape Fear River estuary, NC, and the North Inlet estuary, SC. Five North Carolina sites (Dutchman, Walden 1 and 2, Governor’s, and
Hechtic creeks) were sampled with seines (S) and six North Carolina sites (Baldhead, Cape, Walden, Dutchman, Town, and Barnard’s
creeks) were sampled with rotenone (R). The South Carolina site (Chainey Creek) was sampled with rotenone. Values followed by a =
similarity ≥0.75; b = similarity is 0.50–0.75. Data from the Cape Fear River estuary are from Weinstein et al. (1980) and data from the
North Inlet estuary are from Wenner et al. (1986).
Sampled with seines Sampled with rotenone
Dutchman Governor’s Hechtic Walden 1 Walden 2 Chainey Baldhead Cape Walden Dutchman Town
Governor’s (S) 0.49
Hechtic (S) 0.93a 0.67b
Walden 1 (S) 0.93a 0.52b 0.97a
Walden 2 (S) 0.93a 0.57b 0.98a 1.00a
Chainey (R) 0.90a 0.68b 0.97a 0.92a 0.94a
Baldhead (R) 0.89a 0.45 0.80a 0.78a 0.79a 0.87a
Cape (R) 0.87a 0.50b 0.90a 0.88a 0.90a 0.94a 0.89a
Walden (R) 0.92a 0.47 0.95a 1.00a 0.99a 0.89a 0.76a 0.86a
Dutchman (R) 0.92a 0.47 0.95a 1.00a 0.99a 0.89a 0.76a 0.86a 1.00a
Town (R) 0.45 0.90a 0.65b 0.48 0.52b 0.65b 0.39 0.42 0.45 0.45
Barnard’s (R) 0.66b 0.51b 0.71b 0.66b 0.68b 0.75b 0.59b 0.66b 0.64b 0.64b 0.48
Southeastern Naturalist
259
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
Appendix D. Percent similarity index calculated for comparisons of fish assemblages sampled in previous studies of tidal creeks in the Cape
Fear River estuary, NC, and the North Inlet estuary, SC. Five North Carolina sites (Dutchman, Walden 1 and 2, Governor’s, and Hechtic
creeks) were sampled with seines (S) and six North Carolina sites (Baldhead, Cape, Walden, Dutchman, Town, and Barnard’s creeks) were
sampled with rotenone (R). The South Carolina site (Chainey Creek) was sampled with rotenone. Values followed by a = similarity ≥0.75;
b = similarity is 0.50–0.75. Data from the Cape Fear River estuary are from Weinstein et al. (1980) and data from the North Inlet estuary
are from Wenner et al. (1986).
Sampled with seines Sampled with rotenone
Dutchman Governor’s Hechtic Walden 1 Walden 2 Chainey Baldhead Cape Walden Dutchman Town
Governor’s (S) 32.4
Hechtic (S) 75.4b 50.6b
Walden 1 (S) 75.3b 37.8 82.1a
Walden 2 (S) 73.9b 44.1 84.8a 93.6a
Chainey (R) 74.2b 32.0 54.2b 53.9b 52.5b
Baldhead (R) 61.7b 32.0 62.0b 61.6b 61.6b 63.5b
Cape (R) 73.5b 32.0 77.4a 92.6a 87.8a 72.6b 68.1b
Walden (R) 73.5b 32.0 77.4a 92.6a 87.8a 65.1b 52.5b 61.6b
Dutchman (R) 34.9 77.6a 52.4b 37.1 41.2 65.1b 52.5b 61.6b 100.0a
Town (R) 45.4 43.1 56.6b 47.5 51.6b 49.8 34.7 34.7 34.7 34.7
Barnard’s (R) 65.8b 47.7 81.9a 68.3b 72.3b 59.1b 45.1 45.1 45.1 45.1 45.7
Southeastern Naturalist
K.F. Robinson and C.A. Jennings
2014 Vol. 13, No. 2
260
Appendix E. Morisita’s index of similarity calculated for comparisons of fish assemblages
captured during spring rotenone sampling in Big Rice Field (BFR) and Nieuport impoundments,
Combahee River, SC, 2008–2010, and fish assemblages sampled with seine (S) and
rotenone (R) for previous studies of tidal creeks in Cape Fear River estuary, NC, (Hechtic,
Governor’s, Walden, Dutchman, Barnard’s, Town, Cape, and Baldhead creeks) and North
Inlet estuary, SC, (Chainey Creek). Values followed by a = similarity ≥0.75; b = similarity
is 0.50–0.75. Data from North Carolina are from the study of Weinstein et al. (1980) and
data from South Carolina are from the study of Wenner et al. (1986).
Nieuport Nieuport Nieuport
BRF 2008 BRF 2009 BRF 2010 2008 2009 2010
Hechtic Creek (S) 0.08 0.02 0.17 0.19 0.05 0.03
Governor’s Creek (S) 0.33 0.21 0.79a 0.83a 0.16 0.19
Walden Creek 1 (S) 0.01 0.01 0.05 0.04 0.01 0.01
Walden Creek 2 (S) 0.03 0.03 0.10 0.10 0.02 0.02
Dutchman Creek (S) 0.00 0.00 0.02 0.00 0.00 0.00
Barnard’s Creek (R) 0.07 0.02 0.14 0.15 0.03 0.02
Town Creek (R) 0.33 0.09 0.62b 0.75a 0.18 0.12
Dutchman Creek (R) 0.00 0.00 0.01 0.00 0.00 0.00
Walden Creek (R) 0.00 0.00 0.01 0.00 0.00 0.00
Cape Creek (R) 0.00 0.00 0.02 0.00 0.00 0.00
Baldhead Creek (R) 0.00 0.00 0.02 0.00 0.00 0.00
Chainey Creek (R) 0.09 0.03 0.18 0.19 0.05 0.04
Appendix F. Percent similarity index calculated for comparisons of fish assemblages
captured during spring rotenone sampling in Big Rice Field (BFR) and Nieuport impoundments,
Combahee River, SC, 2008–2010, and fish assemblages sampled with seine (S) and
rotenone (R) for previous studies of tidal creeks in Cape Fear River estuary, NC, (Hechtic,
Governor’s, Walden, Dutchman, Barnard’s, Town, Cape, and Baldhead creeks) and North
Inlet estuary, SC, (Chainey Creek). Values followed by a = similarity ≥0.75; b = similarity
is 0.50–0.75. Data from North Carolina are from the study of Weinstein et al. (1980) and
data from South Carolina are from the study of Wenner et al. (1986).
Nieuport Nieuport Nieuport
BRF 2008 BRF 2009 BRF 2010 2008 2009 2010
Hechtic Creek (S) 18.8 7.2 19.7 18.8 16.9 9.0
Governor’s Creek (S) 31.9 28.1 60.7b 67.9b 18.9 20.0
Walden Creek 1 (S) 5.8 5.8 6.6 5.8 5.4 5.8
Walden Creek 2 (S) 12.1 11.7 13.0 12.1 9.3 12.1
Dutchman Creek (S) 0.4 0.5 1.3 0.3 0.4 0.4
Barnard’s Creek (R) 11.0 6.0 11.9 11.0 11.0 7.8
Town Creek (R) 24.5 6.0 38.4 55.4b 16.1 7.8
Dutchman Creek (R) 0.0 0.0 0.9 0.0 0.0 0.0
Walden Creek (R) 0.0 0.0 0.9 0.0 0.0 0.0
Cape Creek (R) 0.0 0.1 1.8 0.0 0.0 0.0
Baldhead Creek (R) 0.0 0.1 1.8 0.0 0.0 0.0
Chainey Creek (R) 16.8 7.1 18.9 16.2 16.5 9.6