Productivity of Functional Guilds of Fishes in Managed Wetlands in Coastal South Carolina Free

Fish production represents the total amount of new tissue created by a fish population in a particular time frame (Chapman 1978). Production estimates are excellent for determining the potential yield of a particular population or guild (Randall and Minns 2000; Cowley and Whitfield 2002) and assessing the health of a population within a given habitat (Randall and Minns 2000). Production estimates also can be used to model the flow of biomass and energy through an ecosystem (Iverson 1990).

Estuaries are highly productive habitats and intertidal wetlands within estuarine systems have been characterized as net exporters of secondary production, which includes production from fishes (Kneib 2003). Fish move production from the intertidal wetlands to the open estuary through the “trophic relay,” which is a series of predator–prey interactions between small resident species and larger transient species (Kneib 1997, 2003). The transfer of production from the intertidal marsh to the open estuary via the trophic relay can be broken through anthropogenic activities such as impounding or otherwise disconnecting portions of intertidal marsh habitat from the open estuary (Kneib 2003).

Throughout the world, intertidal marshes have been impounded or fragmented for a variety of purposes. In the United States, marshes in Florida and New England were impounded for mosquito control (Harrington and Harrington 1982; Raposa and Roman 2001). Coastal Louisiana contains numerous impoundments, most of which presently are used for waterfowl and furbearer management as well as for attempts at erosion control (Herke et al. 1992). In Europe, managed realignment of land–sea borders (i.e., seawalls and dikes) has resulted in the inundation of many areas that were originally estuarine marshes. Some of these newly flooded marshes have been managed through techniques such as “controlled reduced tide,” in which a system of dikes and valves reduces tidal amplitude to control timing and volume of flooding (Jacobs et al. 2009). These areas are similar to mosquito impoundments in Florida because they have reduced tidal amplitude and flooding (Jacobs et al. 2009). Along the southeastern coast of the United States, thousands of acres of estuarine marsh were ditched, drained, and impounded in the late 18th century for rice cultivation (Milgarese and Sandifer 1982). Although a large portion of these impoundments have been abandoned and are subject to tidal inundation, 11% of oligo-mesohaline marshes of the Atlantic Coast of the United States were impounded as of the 1980s (Montague et al. 1987; Portnoy 1999). In all, about 28,000 ha of impoundments remain intact in South Carolina (Tiner 1977; DeVoe et al. 1987; Kelley 1999).

Most impoundments in South Carolina are managed by manipulation of water levels to promote the growth of food plants for migratory waterfowl (Wenner et al. 1986; McGovern and Wenner 1990). Ricefield trunks (wooden frames through which water can flow) are buried in impoundment dikes, connecting tidal creeks and the perimeter ditches of the impoundments, to allow for water movement through the dikes (see McGovern and Wenner 1990 for illustration). Flap gates attached to these trunks permit water-level manipulation through tidal forcing. These impounded marshes are closed to fish migration for most of the year. When water levels are manipulated, only limited immigration and emigration occurs. The reduced emigration disrupts the trophic relay of predator–prey interactions that moves nutrients and energy from the marsh to the estuary (Kneib 1997, 2003; Stevens et al. 2006a). Avian, reptilian, and mammalian predators consume fishes inside the impoundments and transport nutrients to the open marsh in the form of feces, allowing for the retention of fish production in the terrestrial ecosystem (Stolen et al. 2009); however, most of this production no longer is exported to coastal marine waters through the trophic relay.

In intertidal marshes, in contrast to impounded wetlands, nekton production accumulates in small resident fishes and detritivorous juvenile transients and is transferred from the marsh to the estuary via predation by piscivorous juvenile transients and larger resident fishes (Kneib 1997). By preventing or limiting the transfer of energy that otherwise would be moved to the estuary, impoundments act as a sink for matter and energy (Milgarese and Sandifer 1982; Talbot et al. 1986; Kneib 2003). An exhaustive literature review of estuarine impoundments revealed very few studies that have assessed fish production within these structures. Therefore, the extent to which impoundments affect fish production and energy transfer is largely unknown and probably region-specific because of regional differences in secondary production in the marsh.

In addition to breaking the trophic relay, impoundments throughout the Atlantic and Gulf coasts of the United States share some unique abiotic characteristics that differ from the adjacent marsh and tidal creek habitats and that may create unfavorable environments for some fish species. Generally, managed impoundments are shallower (Rozas and Minello 1999), have increased sedimentation rates (Wenner and Beatty 1988; McGovern and Wenner 1990), and have more organic and less mineral sediments (Bryant and Chabreck 1998) compared with adjacent marshes. Managed impoundments also have greater fluctuations in temperatures (McGovern and Wenner 1990; Rozas and Minello 1999), lower dissolved oxygen (McGovern and Wenner 1990; Raposa and Roman 2001), and lower salinities than the adjacent marsh (Talbot et al. 1986; Raposa and Roman 2001), though all of these characteristics may not be present in an impoundment year-round. These habitat characteristics, especially the low dissolved oxygen levels often present in the early morning hours of the summer (McGovern and Wenner 1990), may reduce habitat quality for certain fish guilds that inhabit these structures.

Our objectives in this 3-y study were to understand how summer conditions in impounded wetlands in coastal South Carolina affect the fish that inhabit these structures and to quantify summertime productivity lost from the trophic relay as a result of impounding intertidal marsh. To meet these objectives, we calculated summer production for estuarine-use functional guilds within three study impoundments along the Combahee River, Beaufort County, South Carolina. Water-level management in our study impoundments called for “spring drawdown,” in which the water levels are drawn down slowly in the spring so that water is confined to the perimeter canals and the system is essentially closed during the summer. The trunks remained open slightly and riser boards were placed to allow some surface transfer of water, but fish egress was minimal (Wenner et al. 1986). We chose to focus on summertime production because the system was closed during this time period, which allowed us to assess production levels in three study impoundments and generate estimates of production that were influenced minimally by migration. We predicted that marine migrant production levels would be greatest in the most saline impoundment, where the salinity is similar to natural summer habitat. Likewise, we hypothesized that the freshwater fishes would have increased production in the least saline impoundment. Others (Rogers et al. 1994; Hoese and Konikoff 1995; Rozas and Minello 1999) have hypothesized that estuarine residents may thrive in impounded wetland habitat; as such, we expected that these species would be very productive in all impoundments.

The study areas were managed by Nemours Wildlife Foundation and the U.S. Fish and Wildlife Service, which partnered for this project. The study impoundments were located on the banks of the Combahee River in Beaufort County, South Carolina, from approximately 48 to 60 river kilometers from the coast (Figure 1). The Combahee River is part of the ACE Basin watershed, so named because it includes the coastal areas of the Ashepoo, Combahee, and Edisto rivers. Much of the land around the ACE Basin is protected through private and public land stewardships; as such, this study system had fewer anthropogenic influences than other estuarine areas in coastal South Carolina.

The three study impoundments were monitored from 2008 to 2010. These impoundments were designed similarly and included an elevated central marsh surrounded by a continuous canal (“perimeter canal”). Desired water levels were achieved via a system of dikes and trunks. Each impoundment had two trunks buried in the dikes, connecting tidal creeks and perimeter canals. Water levels were manipulated through tidal forcing by the use of flap gates on either side of the trunk. For example, to move water out of the impoundment, the trunk was opened on the impoundment side of the dike. During ebb tide, water flowed out of the impoundment, pushing open the flap gate on the tidal creek side of the dike. During flood tide, water in the tidal creek pushed the gate closed, preventing water from entering the impoundment (see McGovern and Wenner 1990 for illustration). The central marsh area of all three impoundments consisted of marsh edges and small open water areas. The three impoundments, ACE Basin East (ACE; 32.669°N, 80.712°W), Nieuport (32.668°N, 80.683°W), and Big Rice Field (BRF; 32.639°N, 80.663°W), varied in both salinity regime (ACE was freshest and BRF most saline) and size. Nieuport was the largest at approximately 118 ha, and ACE and BRF were substantially smaller at approximately 43 and approximately 48 ha, respectively.

These impoundments were managed through water-level manipulation to enhance the growth of food plants for migratory waterfowl, but this manipulation varied among the study impoundments. Regardless of the salinity of the Combahee River, the water in Nieuport and BRF was drained partially in the spring to expose the central marsh and allow for sediment oxidation and the germination of food plants. The impoundments then were flooded slowly throughout the late summer to reach a depth of approximately 0.5 m above the interior marsh in the fall; this depth provided the proper habitat for food plant growth and waterfowl foraging. The flap gates remained open slightly throughout the year and riser boards maintained the desired water level. These riser boards were placed directly in front of the impoundment side of the trunk and if the water level exceeded the height of the boards, it flowed out of the impoundment. In this way, the water could be kept at a constant level. This procedure allowed for some water exchange to avoid hypoxic conditions. The ACE Basin-East impoundment was maintained as an oligohaline impoundment and the water levels were not manipulated if the river salinity was higher than about 5 practical salinity units (psu). This approach further decreased fish migration into and out of this impoundment and was typical during drought years, when the salt wedge (the dense lower layer of salt water in the tidal portion of the river that moves up- and downriver with the tides and changes in freshwater inflow) moved farther upriver. Water levels in ACE were manipulated in the same manner as the other two impoundments in 2008. Because of drought conditions throughout the region, ACE water levels were not manipulated after the 2009 spring drawdown. Additionally, unlike BRF and Nieuport, the trunks at ACE were not kept open slightly throughout the year; the resultant lack of water exchange and circulation could cause hypoxic events within the impoundment. For all three impoundments, additional fish immigration occurred when the river water topped the dikes, either during extreme high tides or during storm events (E. Mills, Nemours Wildlife Foundation, personal communication).

Within each impoundment, three sections of the canal around the central marsh were chosen as study sites. Regardless of water level, these three canal sections were connected. Selection of canal sections was constrained primarily by our ability to sample by boat and these sections were as equidistant from one another as possible. The sections ranged in size from 0.09 to 0.27 ha and had an average depth of 0.80–2.08 m during initial site delineation (Table 1).

We sampled each impoundment with rotenone twice per year, from 2008 to 2010, when water was confined to the canals. The sampling events occurred in May (“early summer”) after the spring drawdown and in September and October (“late summer”) just before fall flooding. For each sampling effort, we blocked off each of the nine canal sections with two 6.4-mm mesh block nets (4.3 m deep × 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 parts per billion [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 three specimens rose to the surface within a 20-min time period. We then 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 <1.2% of the total impoundment area in each of the three impoundments.

All specimens collected were grouped by species, placed into 2-cm size classes, counted, and weighed. We placed any fish that could not be identified in the field in 95% ethanol and transported it back to the laboratory for identification. Rotenone sampling usually involves collecting all fish in the study area the day that the rotenone was applied, as well as a second-day collection to retrieve fish that surfaced overnight. We confined sampling efforts to a first-day pick-up because of concerns about alligators Alligator mississippiensis and water birds consuming fish overnight, and thus biasing our samples. During each sampling effort, alligators were seen consuming fish within each sample site, which may have introduced some bias into our sampling. Because this disturbance occurred in all impoundments and not more than one alligator was observed in a canal section, this effect probably was small.

We measured the water quality of each canal section within each impoundment during each sampling event. Temperature, salinity, and dissolved oxygen (DO) were measured with an YSI-85® (Yellow Springs Instruments, Inc., Yellow Springs, Ohio) handheld dissolved oxygen and conductivity meter. We did not measure dissolved oxygen in ACE in early summer 2008 because of equipment malfunction. We took all measurements just below the water surface and we reported the averages of the water-quality measurements taken in each of the three canal sections within each impoundment. We performed a nested analysis of variance (ANOVA) on each water quality variable to analyze differences among impoundments nested by sampling period. Significant differences (α  =  0.05) were further analyzed with Tukey's Honestly Significant Differences test. These analyses were performed in R (R Development Core Team 2010).

We assigned all fishes to one of five estuarine-use functional guilds as defined by Elliott et al. (2007): catadromous (migrate from freshwater to the ocean to spawn), estuarine (complete their life cycle within the estuary), freshwater migrant (reside in estuaries but are found in freshwater), freshwater straggler (reside in freshwater but occasionally move to the upper estuary), and marine migrant (spawn offshore and use the estuary as nursery habitat) species (Table 2). We chose to use the estuarine-use functional guild designations so that we could understand how fishes that make use of the estuary in different ways are affected by residing in impounded wetlands. We made one change to the estuarine-use functional guild system: we grouped amphidromous species, such as bay anchovy Anchoa mitchilli, with the estuarine species. Some species we collected were listed in Elliott et al. (2007) as examples of the estuarine-use functional guilds (e.g., menhaden Brevoortia tyrannus and American eel Anguilla rostrata), but this was uncommon. We used the information available on Fishbase (www.fishbase.org), literature related to the biology of collected species, and our own observations to assign each species to the appropriate guild. The size-frequency (Hynes) method of production estimation was used to calculate seasonal production of each guild in each sampling year (Garman and Waters 1983). For each year, we used the data collected from the early and late summer samples from that year to calculate production estimates. The size-frequency method requires that the species groupings have similar life-history strategies (Hayes et al. 2007). This method uses size classes rather than cohorts and takes into account reproduction, growth, and mortality throughout the time interval being analyzed. For each guild, we pooled all individuals and placed them into 2-cm length groups for each sampling date. For each impoundment, we used the maximum-length group captured for each guild over the 3 y of sampling as the maximum guild length group for all production estimates in that impoundment. These length groups contained the estimated absolute densities of fishes in numbers per hectare. The estimate of variance was calculated as a measure of sampling variability for each length group (Garman and Waters 1983). To calculate production with the size-frequency method, the seasonal mean number of individuals within each of k  =  1 to c length groups (⁠⁠) and the seasonal mean weight of each length group (⁠⁠), as well as the associated variances were calculated as

and

with i  =  1 to a sampling dates, where was the density, was the mean weight for the ith date in the kth length group, and D_i was the weighting factor in days for the interval between the first sampling date and the ith sampling date. These values were used to calculate the summer production for each guild via the size-frequency method by

The variance of the production estimate (Garman and Waters 1983) was calculated as

where the variance of the mean annual density of each length group was estimated as

The variance associated with the mean weight of each length group was estimated as

The 95% lognormal confidence intervals (Burnham et al. 1987) also were calculated as

and

where

All calculations were performed using the Microsoft Excel (Microsoft Corp., Seattle, WA) spreadsheet provided by Hayes et al. (2007).

In annual production estimates, the production value is divided by the mean cohort production interval, which is calculated as the average maximum age of the fish population in question (Garman and Waters 1983). Because we only calculated seasonal production for each guild, we omitted the cohort production interval from the production estimates (Benke 1984). If the estimated production value for a given guild was negative, we assumed production to be zero and did not calculate 95% confidence intervals. We reported guild production and associated 95% confidence intervals in g⋅m⁻²·summer⁻¹. Using these estimates, we calculated production-to-mean standing stock biomass (B̄) ratios (P/B̄) for each guild (Garman and Waters 1983), where

Climatic conditions varied widely during the 3 y of sampling. In summer 2008, the coast of South Carolina experienced a moderate drought (SCDNR 2010), which caused below-average discharge in the Combahee River (Figure 2). Average summer (May–September) precipitation in Yemassee, Beaufort County, South Carolina, is 70.1 cm (SERCC 2011). In summer 2008, precipitation was 70.6 cm, but all months were below the monthly average except August, which received 15.2 cm more rainfall than average (SERCC 2011). In 2009, drought conditions subsided (SCDNR 2010) but coastal precipitation patterns remained below average (summer precipitation  =  62.4 cm; SERCC 2011) and discharge was lower than average in the Combahee River during fall 2009 (Figure 2). Sampling during early summer 2010 occurred during normal precipitation conditions, but late summer 2010 sampling occurred during incipient drought conditions (SCDNR 2010). Summer 2010 precipitation (66.1 cm) was still below average (SERCC 2011). Because of drought conditions, water levels in ACE were not manipulated as frequently as in Nieuport and BRF. Water levels were drawn down in February 2008, before the first sampling period, and the impoundment was flooded only once (i.e., winter 2008, after our first year of sampling) during our 3 y of sampling.

Mean salinity for each sample period in the study impoundments ranged from 1.6 ± 0.1 psu (SD) to 20.4 ± 0.6 psu (Figure 3; Table S1, Supplemental Material). The results of the nested ANOVA for salinity indicated that there was a significant effect of sample period on salinity among impoundments (F  =  29.249, df  =  15, P < 0.001) and salinity among impoundments was significantly different (F  =  6.232, df  =  2, P  =  0.011). The Tukey's Honestly Significant Differences multiple-comparison tests showed that all three impoundments differed significantly in salinity (ACE  =  3.6 ± 1.6 psu; Nieuport  =  8.2 ± 4.6 psu; BRF  =  11.8 ± 4.8 psu; P < 0.001 for all comparisons). Mean water temperatures for each sample period in the study impoundments ranged from 20.4 ± 0.3°C to 31.0 ± 1.3°C (Figure 3; Table S1, Supplemental Material). There was an effect of sample period on temperature (F  =  13.537, df  =  15, P < 0.001), but the temperature did not vary among the three impoundments (F  =  0.376, df  =  2, P  =  0.692). Mean DO for each sample period in the impoundments ranged from 0.9 ± 0.1 mg/L to 13.3 ± 0.3 mg/L (Figure 3; Table S1, Supplemental Material). There was a significant effect of sample period on DO (F  =  10.357, df  =  15, P < 0.001), but DO did not vary among impoundments (F  =  0.957, df  =  2, P  =  0.408).

Over the 3-y sampling program, 44 fish species from 23 families were collected from the three impoundments (Table 3). In each sampling year, the abundance of fishes in each guild was greater in the early summer than in the late summer (Figure 4; Table S2, Supplemental Material). Freshwater migrants, freshwater stragglers, and marine migrants were the three most abundant guilds in each sample period. In ACE, freshwater stragglers were consistently the most abundant guild. In Nieuport, freshwater migrants were most abundant in early summer 2009; this guild was composed almost entirely of sailfin mollies Poecilia latipinna (n  =  3,365). In early summer 2010, three guilds were almost equally abundant in Nieuport: estuarine species (composed mostly of rainwater killifish Lucania parva; n  =  1,115); freshwater migrants (composed mostly of sailfin mollies; n  =  1,062); and marine migrants (composed mostly of spot Leiostomus xanthurus; n  =  1,032). Marine migrants were most abundant in all other Nieuport samples. The species composition of the marine migrant guild in Nieuport varied among sample dates such that the bulk of the species collected in the late summer were species able to withstand hypoxic conditions through aquatic surface respiration (e.g., ladyfish Elops saurus, tarpon Megalops atlanticus, and striped mullet Mugil cephalus). Marine migrants were most abundant in BRF in all samples except early summer 2009, when freshwater migrants were most abundant and composed mostly of sailfin mollies (n  =  501). Abundance of all guilds in ACE in late summer 2010 was very low, with a 23 total fishes collected in two of the three canal sections. The third section was not sampled because it was completely covered in duckweed Lemna sp. for the duration of the late-summer sampling period. Dissolved oxygen levels in ACE in late summer 2010 ranged from 0.75 to 1.04 mg/L (Table S1, Supplemental Material), which suggests that the third canal section would have yielded an equally low abundance of fishes. Almost all fish that were captured in ACE in late summer 2010 were air breathers such as bowfin Amia calva and longnose gar Lepisosteus osseus that can withstand hypoxic conditions.

Summer production estimates mirrored abundances, with the greatest production values across the 3 y of estimates in the freshwater stragglers guild in ACE (15.56–117.38 g⋅m⁻²·summer⁻¹; Table 4). Freshwater migrant production was greatest in ACE in 2008 (16.64 g⋅m⁻²·summer⁻¹) but greatest in Nieuport in 2009 (7.03 g⋅m⁻²·summer⁻¹) and 2010 (5.02 g⋅m⁻²·summer⁻¹). Marine migrant production was greatest in BRF in 2008 (27.23 g⋅m⁻²·summer⁻¹) and 2009 (22.92 g⋅m⁻²·summer⁻¹), but marine migrants had slightly greater production in Nieuport (30.22 g⋅m⁻²·summer⁻¹) than in BRF (26.15 g⋅m⁻²·summer⁻¹) in 2010. Production of estuarine species was greatest in BRF in 2008 (1.51 g⋅m⁻²·summer⁻¹) and greatest in Nieuport in 2009 (0.46 g⋅m⁻²·summer⁻¹) and 2010 (0.91 g⋅m⁻²·summer⁻¹). Production of catadromous species was low in all years in Nieuport and BRF, which was expected, given that we collected one catadromous species Anguilla rostrata in low numbers. Production was not calculated for estuarine species in 2010 or for catadromous species in all years in ACE because insufficient numbers of individuals were collected. Production was assumed to be zero for catadromous species in 2008 and 2009 in Nieuport and in 2010 in BRF, freshwater migrants in 2008 in Nieuport, freshwater stragglers in 2010 in Nieuport and BRF, estuarine species in 2008 in ACE, and marine migrants in all years in ACE, because calculated production values were negative for these guilds (Table 4). Production-to-biomass ratios in the three impoundments ranged from 0.37 to 20.86 (Table 5).

In our 3-year study, we observed a large decrease in fish abundance each summer. This decline in abundance likely was caused by large mortality events in each of the study impoundments. High mortality rates within the structures contributed to the low overall productivity of all fish guilds that we observed throughout the study. Although the production levels of all guilds within our study impoundments was generally low, we observed that guild-specific production varied both among years and among impoundments. Abiotic factors such as salinity and dissolved oxygen contributed to the variability in our estimates of production, as well as to the high mortality rates that likely occurred.

The large decrease in fish abundance between the early and late summer collecting periods indicated that the habitat quality in these structures over the summer was low. Guild-specific abundances declined by up to three orders of magnitude (e.g., freshwater migrants in Nieuport in 2009; Figure 4), and we believe that this pattern was caused by mortality in the impoundments each summer. Emigration provides one alternative explanation for these decreases in abundance, although this explanation seems highly unlikely. Previous research has shown that once fish immigrate into impounded wetlands in South Carolina, most do not leave (Wenner et al. 1986; McGovern and Wenner 1990). Additionally, the preliminary results from a pilot investigation in the study impoundments showed that very few fish emigrated during spring drawdown in ACE and Nieuport (K.F. Robinson and C.A. Jennings, unpublished data). Some transient fish may emigrate from Nieuport and BRF during stochastic events such as large storms that flood the impoundment past the height of the riser boards (Taylor et al. 1998). However, emigration achieved in this fashion, especially from large impoundments (43–118 ha), is not sufficient to explain the large decrease in abundances we observed each summer. In addition, the presence of riser boards at the trunks can inhibit the emigration of many demersal species from the impoundments. McGovern and Wenner (1990) suggested that removing these riser boards during drawdown would allow for increased emigration rates of all fishes. Finally, we noted a decline in fish abundance in ACE each summer. This impoundment was closed off from the Combahee River for the last 2 y of our collections, and the closure eliminated the potential for emigration from this structure. Based on this information, we concluded that the observed decrease in abundance of all guilds in this study was mainly the result of mortality throughout the summer.

Differences in the habitat characteristics of the individual impoundments led to variability in guild-specific production estimates. The three impoundments followed a regular pattern of salinity, which reflected their distance from the estuary. Big Rice Field was the most saline, Nieuport was intermediate, and ACE was freshest. We hypothesized that the production of many of the guilds would follow the salinity gradient among the three impoundments. Our expectations generally were confirmed, though the contrast was not as great as expected for many of the guilds. We predicted that marine migrants would be the most productive in BRF. We observed that marine migrant production in BRF ranged from nearly equal to production in Nieuport to 24.5 times greater than marine migrant production in Nieuport. In the 3 y of our study, marine migrant production was zero in ACE. We observed a similar pattern in the productivity of estuarine species. We expected that the productivity of this guild would be relatively high in all three impoundments, but we found that estuarine species had similarly low ranges of productivity in Nieuport (0.31–0.91 g⋅m⁻²·summer⁻¹) and BRF (0.08–1.51 g⋅m⁻²·summer⁻¹), and they were unproductive in ACE throughout the study. This pattern for marine migrants and estuarine species may reflect a threshold effect of salinity on production, in which the environment in Nieuport and BRF is sufficient to support positive production for these generally euryhaline species.

Unlike marine migrants and estuarine species, production of freshwater stragglers was greatest in ACE each year, as expected. Freshwater migrant production was also greatest in ACE for the first year of the study, but in 2009 and 2010, production levels in Nieuport were 4.2 and 3.5 times greater, respectively, than in ACE. Within ACE, freshwater straggler and freshwater migrant production levels declined throughout the study. Estimates of production for freshwater stragglers were 7.5 times greater in 2008 (117.38 g⋅m⁻²·summer⁻¹) than in 2010 (15.56 g⋅m⁻²·summer⁻¹), while those for freshwater migrants were 11.5 times greater in 2008 (16.64 g⋅m⁻²·summer⁻¹) than in 2010 (1.44 g⋅m⁻²·summer⁻¹; Table 4). Dissolved oxygen levels throughout the study contributed to the decline in the seasonal production estimates observed in the freshwater guilds in ACE. Because of drought conditions, water-level manipulation in ACE did not occur after the spring drawdown in 2009. Also, unlike in Nieuport and BRF, the trunks were closed throughout the year and prevented water circulation. These management actions created a situation in which hypoxia developed sometime over the winter of 2009–2010 and worsened over the summer of 2010. The low average DO values in ACE in 2010 (2.57 ± 0.03 mg/L (SD) in early summer and 0.91 ± 0.15 mg/L in late summer; Figure 3) coincided with low abundances of all fishes in this impoundment in the early summer and a very low abundance of even the most hypoxia-tolerant species (e.g., bowfin and longnose gar) in the late summer. The process of drawing water into the impoundments served to keep the DO from dropping to hypoxic levels and probably provided a pool of immigrants that could repopulate the fish community after the summer mortality events.

In all impoundments, we believe that summer mortality events were driven, at least in part, by dissolved oxygen levels. For example, much of the production amassed over the summer in the marine migrant guild was probably the result of a few species that are tolerant of low DO and high temperature. In Nieuport and BRF in all 3 y, striped mullet, tarpon, and ladyfish were the only marine migrant species that did not decline in abundance over the summer. These species are capable of aquatic surface respiration and therefore can tolerate low DO levels. Other, less hypoxia-tolerant marine migrant species, such as spot, either were collected in reduced numbers or were absent in the late summer. Previous studies have indicated that habitat quality inside impoundments degrades over the summer, when water temperatures can reach high levels and the water can become anoxic in the early morning hours (Wenner et al. 1986; Portnoy 1991). This habitat degradation can lead to fish mortality either directly through suffocation or indirectly through consumption of fish by wading birds as these fish rise to the oxygen-rich surface waters (Wenner et al. 1986; McGovern and Wenner 1990). Overall, dissolved oxygen likely drove both the low production values that we observed in all impoundments throughout the study and the among-year variation in freshwater guild productivity in ACE.

Although production of all guilds within our study impoundments was low each summer, the high abundances of many species in our early summer collections suggested that some guilds may have been much more productive within impoundments in other seasons. The water-level manipulation strategies employed in these structures probably served to reduce guild-specific production for the freshwater migrant and estuarine species guilds. For example, ACE and Nieuport most likely provided suitable habitat for freshwater migrant species in the winter and spring, but the bulk of the individuals large enough to be caught by our nets did not survive the summer. These species can reproduce when the central marsh is flooded (Stevens et al. 2006b), and there may have been some influx of freshwater migrants when water was drawn into the impoundments. These events served to sustain the populations within the impoundments. Others have reported that the abundance of small-bodied resident fishes can be quite high in fall and winter (Rey et al. 1990; Poulakis et al. 2002; Stevens et al. 2006b). Productivity of freshwater migrants probably declined in ACE throughout our study because water-level manipulation was halted and water was confined to the canals in 2009 and 2010. These species did not have access to the shallow water over the marsh for reproduction, and the lack of water exchange blocked the immigration of additional individuals into the impoundment. Like the freshwater migrants, reproduction rates of estuarine species are much greater when these fish have access to the central marsh area of the impoundment (Stevens et al. 2006b). Therefore, productivity of estuarine species probably was at its lowest in the summer in our study impoundments and was greater when these species could inhabit the central marshes. Because productivity of many guilds likely was lowest in the summer, our estimates provide a minimum value of production that is lost from the trophic relay and provide quantitative evidence for the poor habitat quality that these structures provided during the summer.

Although published production estimates within impoundments are rare, our guild-specific estimates fall on the low end of previously reported values for members of these guilds in natural estuarine systems around the world (Table 6). For example, single-species production estimates for estuarine species in South Africa ranged from 0.002 g⋅m⁻²⋅y⁻¹ to 41.35 g⋅m⁻²⋅y⁻¹ (Cowley and Whitfield 2002). Single-species production estimates for estuarine species ranged from 0.004 g⋅m⁻²⋅y⁻¹ to 31.855 g⋅m⁻²⋅y⁻¹ in California (Allen 1982), and from 0.188 g⋅m⁻²⋅y⁻¹ to 9.328 g⋅m⁻²⋅y⁻¹ in a Mexico lagoon (Warburton 1979). Mummichog Fundulus heteroclitus summer production in a Massachusetts salt marsh was 9.1 g⋅m⁻²·summer⁻¹ (Valiela et al. 1977). The comparison of summer production of mummichogs in Massachusetts to our estimates of estuarine-species summer production is particularly apt. In natural marsh habitat in the ACE Basin, the mummichog is the dominant member of the estuarine species guild (Upchurch and Wenner 2008; Goldman et al. 2010). Our greatest estimates of estuarine species production during our study (0.91 g⋅m⁻²·summer⁻¹ in Nieuport and 1.51 g⋅m⁻²·summer⁻¹ in BRF) are much less than these estimates of summer mummichog production. We expected that our guild-wide estimates, which included production of multiple species, would have been greater than single-species estimates, but this was not the case. Interestingly, mummichogs were noticeably absent from our study impoundments (Table 3); this result may indicate that the population dynamics of the estuarine species guild differed in impounded wetlands and natural marsh habitats (Robinson and Jennings, in press). Mummichogs also were found in reduced numbers in impoundments near North Inlet, South Carolina, where they comprised 7% of the fish collected with seines in impoundments, but 40% of the fish collected with rotenone in the tidal creek adjacent to the study impoundments (Wenner et al. 1986).

Our production estimates for marine migrants also fell on the low end of the ranges of previously reported single-species estimates from other estuarine systems (Table 6). Marine migrant production estimates from North Carolina and Virginia ranged from 0.38 g⋅m⁻²·summer⁻¹ (Weinstein and Brooks 1983) to 10,429 g⋅m⁻²⋅y⁻¹ (Adams 1976). Single-species estimates ranged from 1.19 g⋅m⁻²⋅y⁻¹ to 9.36 g⋅m⁻²·y⁻¹ in Mexico (Warburton 1979) and from 0.01 g⋅m⁻²⋅y⁻¹ to 2.26 g⋅m⁻²⋅y⁻¹ in South Africa (Cowley and Whitfield 2002). Of these previously reported numbers, productivity estimates from species found in coastal North Carolina and Virginia are the best comparison for estimates from our study impoundments, because the species composition of these areas is very similar to that of South Carolina estuarine systems. Based on these published estimates for individual species, we believe that summer production of the marine migrant guild in our study impoundments (1.10–30.22 g⋅m⁻²·summer⁻¹) was inhibited. Estimates of production from previous studies of estuarine systems for the other guilds that we sampled were quite rare; and therefore, we cannot make inferences about differential levels of fish production in impounded and natural marsh habitat for these guilds (Table 6). The published estimates we report were calculated with cohort-based production metrics, which differ from the size-frequency method of estimating production. The use of different methods of calculating production could introduce some differences in the values obtained (Minello et al. 2008).

Fish production is an indicator of habitat quality (Randall and Minns 2000; Cowley and Whitfield 2002). The relatively low summer production values obtained for the guilds residing in the study impoundments indicated that the habitat for fishes provided by these managed wetlands in the summer was inferior to natural marsh habitat. These estimates also provide a low-end estimate of the amount of biomass that was lost from the trophic relay (Iverson 1990; Kneib 2003). Productivity for some guilds, such as freshwater migrants and estuarine species, is probably higher in other seasons when habitat conditions are more favorable (Rey et al. 1990; Poulakis et al. 2002; Stevens et al. 2006b). Additionally, some summertime production may have been moved out of the impoundments and into the terrestrial ecosystem through predation by water birds and alligators. Though we are unsure of the rate at which such transfer might be occurring in our study system, high rates (8–13%) of consumption of fish productivity by water birds were found in a Florida impoundment (Stevens et al. 2006a).

Overall, our study indicates that impounded wetlands managed for waterfowl did not provide high-quality summertime habitat for fish. Further, based on the data from the ACE impoundment, we conclude that the water quality within an impoundment can degrade if a connection with the adjacent river is not maintained for at least part of the year; decreased water quality further reduces year-round habitat quality in the impoundment. Although impoundments are providing poor-quality fish habitat during the summer months and removing biomass from the trophic relay, the current proportion of wetlands in the Combahee River that is impounded (approx. 10%; Kelley 1999) probably is not great enough to negatively affect the estuarine fish community. For example, the population dynamics of spot (a dominant marine migrant in South Carolina estuaries) were unaffected by the presence of impounded wetlands in the Combahee River estuary (Robinson and Jennings 2012). Impounded wetlands managed for waterfowl and other water birds provide high-quality habitat for avian fauna in South Carolina, including some threatened and endangered species (US Endangered Species Act [ESA 1973, as amended]; Boettcher et al. 1995; DeSanto et al. 1997; Gordon et al. 1998; Harrigal and Cely 2004; Post 2004). In our study system, the benefits to avian wildlife may outweigh the negative aspects of impoundments for fish that reside within them—the North American Waterfowl Management Plan named the ACE Basin as a flagship project because of its importance for overwintering waterfowl (Tufford 2005). However, the ACE Basin is a relatively pristine system with minimal anthropogenic influences. In areas where impoundments represent a much larger proportion of the estuarine landscape, these structures may exert a substantial negative influence on estuarine fish communities.

This research represents a step in the process of understanding how impounded wetlands managed for waterfowl affect coastal ecosystems. Our estimates of production for the most productive guilds can be used to model ecosystem-level energy exchange in impounded wetlands, which is relevant given that about 11% of oligo-mesohaline marshes of the Atlantic coast of the United States are impounded in some way (Montague et al. 1987; Portnoy 1999). In addition to impoundments in the United States, managed realignment efforts in Europe use techniques such as reduced tidal exchange or controlled reduced tides to reclaim tidal marshes in systems such as the Schelde estuary, Belgium (Cox et al. 2006; Maris et al. 2007; Jacobs et al. 2009). These restored marshes may function similarly to waterfowl impoundments for fish productivity because of the reduced tidal exchange. To further our knowledge about the fish communities in impounded wetlands, future studies could incorporate habitat characteristics such as the amount of submerged aquatic vegetation, which is often a predictor of resident fish species abundance (Kneib 1997; Rozas and Minello 2010). Although much remains to be learned about impounded wetlands and their role in coastal ecosystems, our research suggests that the influence of these structures on fish depends not only on the way in which they are managed (e.g., salinity regime and water level manipulation) but also on the component members of the fish community inhabiting the impoundments.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Table S1. Temperature, salinity, and dissolved oxygen measurements for each canal section of the ACE-Basin East (ACE), Nieuport, and Big Rice Field (BRF) impoundments on the Combahee River, South Carolina, sampled in early and late summer, 2008–2010.

Found at DOI: 10.3996/112012-JFWM-099.S1 (46 KB XLSX).

Table S2. Guild-specific data for fishes collected in the early summer (“spring”) and late summer (“fall”) in the ACE-Basin East (ACE), Nieuport, and Big Rice Field (BRF) impoundments on the Combahee River, South Carolina, from 2008 to 2010. Guilds were catadromous (CA), freshwater migrants (FM), freshwater stragglers (FS), estuarine species (ES), and marine migrants (MM). Guilds were grouped by 2-cm length group. N per ha  =  season-specific density of fish, Var N  =  season-specific variance in density, Var W  =  season-specific variance in weight, mean N  =  average density of fish across seasons, mean w  =  average weight across seasons, V(N)  =  variance of density across seasons, V(w)  =  variance of weight across seasons.

Found at DOI: 10.3996/112012-JFWM-099.S2 (95 KB XLSX).

Data S1. Supplemental data metafile.

Found at DOI: 10.3996/112012-JFWM-099.S3 (2 KB TXT).

We thank M. Alber, J. Peterson, and S. Schweitzer for input into experimental design and comments on this manuscript; and 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. We also thank M. Collins, J. Robinson, S. Upchurch, and one anonymous reviewer for helpful comments in reviewing this manuscript. This research was supported with a grant from the National Fish and Wildlife Foundation. The Georgia Cooperative Fish and Wildlife Cooperative Research Unit is sponsored jointly by Georgia Department of Natural Resources, the University of Georgia, the U.S. Fish and Wildlife Service, the U.S. Geological Survey, and the Wildlife Management Institute. This study was performed under the auspices of the University of Georgia Animal Use Protocol 2009-3-060.

Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.