Zooplankton composition and abundances were quantified in the lower Roanoke River and Albemarle Sound, North Carolina. The spatial and temporal overlap between larval alosines, including American shad Alosa sapidissima, river herring (alewife A. pseudoharengus and blueback herring A. aestivalis), hickory shad A. mediocris, and Zooplankton were examined to determine whether larval alosines in this system are food limited. Samples were collected weekly at 19 stations from March through June 2008–2009 in three habitats: River, Delta, and Sound. Spatial differences in Zooplankton were observed, with the abundance in the Sound (16,546 ± 14,678 [number/m3 ± SD]) being significantly higher than those in the River (4,934 ± 3,806) and Delta areas (4,647 ± 2,846). Zooplankton composition was dominated by Daphniidae, Bosminidae, calanoid and cyclopoid copepods, copepod nauplii, and rotifers. The spatial patterns in alosine abundance were the opposite of those for Zooplankton, being highest in the River (21.0 ± 127.6) and lower in the Delta (7.5 ± 35.5) and Sound (4.6 ± 24.8). Mouth gape models for each alosine species showed that copepod nauplii and rotifers are the most suitable-sized prey for the first feeding after yolk sac absorption. There was a high degree of spatial and temporal overlap between larval alosines and size-appropriate prey items, suggesting that the larval alosines are not food limited in Albemarle Sound.
Year-class strength and recruitment of fish is strongly correlated with survival during egg and larval development. During the past century, a number of hypotheses were developed to explain fluctuations in recruitment. Temporal and spatial overlap between fish and their prey is considered one of the most influential factors affecting survival and growth during the larval phase (Cushing 1990; Leggett and DeBlois 1994; Gotceitas et al. 1996). Hjort (1914) developed the critical-period hypothesis, which directly links larval survival with feeding success. Under this hypothesis, there is a critical period, defined as the period after yolk sac absorption when a larva is transitioning from endogenous to exogenous feeding. Cushing's match/mismatch hypothesis (1972) builds on Hjort's critical-period hypothesis. The match/mismatch hypothesis extends the importance of prey availability and feeding success to include the entire larval period, not just that of the first feeding. Laboratory experiments and field studies that test critical-period hypotheses have received mixed reviews (De Lafontaine and Leggett 1987; Paradis et al. 1996). These studies are often confounded by attempts to balance the proportions of predators and prey. Despite this critique, substantial evidence suggests that larval nutrition and starvation are leading causes for the variability in early life history (Leggett and DeBlois 1994; Fortier et al. 1995; Houde 2008). Smaller larvae, such as alosines, are believed to highly susceptible food limitation and starvation, resulting in high mortality rates (Miller et al. 1988; Houde 1994).
In addition to having spatial and temporal overlap with potential Zooplankton prey, the available prey items need to be a size capable of being consumed. In larval fish, mouth gape is a limiting factor at the onset of exogenous feeding, restricting the prey size that can be consumed (DeVries et al. 1998; Yufera and Darias 2007). Mouth gape becomes less limiting as fish grow. There is a positive relationship between fish length and mouth gape (DeVries et al. 1998; Puvanendran et al. 2004). Studies have suggested that the maximum functional mouth gape is attained when the mouth is open at a 90° angle. Optimal prey sizes are generally within 30–50% of the mouth gape (Bremigan and Stein 1994; Turingan et al. 2005; Riley et al. 2009).
The early life history of most species is intertwined with variability in biotic and abiotic factors that influence feeding and behavior in larval fish. The distribution and abundance of suitable prey is a key factor in feeding success. Prey density thresholds have been established for many species of larval fish. Baseline densities ranging from 9,000 to 100,000 zooplankton/m3 are critically important for the growth and survival of most estuarine and marine fishes (Hunter 1981). In laboratory studies, Leach and Houde (1999) observed that Zooplankton densities of 500,000 individuals/m3 or more were necessary for the production of larval American shad Alosa sapidissima. Similarly, Riley (2012) showed that Zooplankton densities of 100,000 individuals/m3 or more were necessary to support larval growth after first feeding. Food deprivation for as little as 2 d can significantly affect survival and predator avoidance (Johnson and Dropkin 1995, 1996). These studies highlight the importance of food availability at varying spatial and temporal scales and underscore a multitude of factors that contribute to recruitment and year-class strength.
In most aquatic systems, Zooplankton abundance is highly variable and fluctuates seasonally with primary production, water temperature, and hydrographic conditions. Low densities and patchy distributions of Zooplankton are often observed in coastal rivers and estuaries that serve as primary nursery grounds for many economically important species of fish and shellfish (Hynes 1970; Chick and Van Den Avyle 1999). The coastal ecosystems of North Carolina support a high level of productivity and form the second largest estuarine complex in the United States, with over 7,500 km2 of open water. More than 90% of North Carolina's commercial fisheries landings and over 60% of the recreational harvest are comprised of estuarine-dependent species (Street et al. 2005). These species depend on an abundance of food within coastal rivers and sounds to complete their life cycle.
Roanoke River supports numerous populations of anadromous fishes, many threatened or endangered. Anadromous alosines such as American shad, river herring (blueback herring A. aestivalis and alewife A. pseudoharengus), and hickory shad A. mediocris are in sharp decline and some stocks are bordering on collapse (ASMFC 2007; Greene et al. 2009; NCDMF 2010). Additionally, North Carolina and Virginia once accounted for 90% of the total commercial river herring landings in the United States. (Crecco and Gibson 1990). Shads and river herring spawn in the Roanoke River. Alewives, blueback herring, and American shad return to their natal rivers to spawn, but there is no evidence to support this with hickory shad (Green et al. 2009). Blueback herring spawn in both lotic and lentic waters over hard substrates, avoiding areas with standing water (Walsh et al. 2005; Greene et al. 2009). Alewives begin spawning in lentic waters over a variety of substrates, including gravel, sand, detritus, and submerged vegetation (O'Connell and Angermeier 1997; Able and Fahay 1998; Walsh et al. 2005). In areas where alewives and blueback herring have overlapping distributions, alewives begin spawning in late February, 3–4 weeks earlier than blueback herring, which begin spawning in late March (O'Connell and Angermeier 1997; Able and Fahay 1998). In the sympatric range, blueback herring and alewives utilize different spawning habitats. Blueback herring do not migrate as far upstream as alewives. Blueback herring spawn predominately in the main-stream flow, while alewives select shorebank eddies and deep pools for spawning (Able and Fahay 1998; Greene et al. 2009). American shad spawn when water temperatures are between 12°C and 20°C, typically between March and early June (Able and Fahay 1998; Bilkovic et al. 2002). In the Roanoke River, hickory shad were observed spawning in water less than 1 m deep with moderate to high velocity, over substrates containing cobble, gravel, and sand (Greene et al. 2009; Harris and Hightower 2010).
Alosine larvae use nursery habitats within the lower reaches of the Roanoke River and Albemarle Sound (Greene et al. 2009; Harris and Hightower 2010). One possible explanation for the failure of these stocks is high levels of larval mortality caused by food limitation. The abundance and distribution of Zooplankton in the lower Roanoke River is the lowest among coastal rivers in North Carolina (Table 1). A long-term study conducted from 1984 to 1991 by Rulifson et al. (1993) and a similar study by Coggins (2005) in 2003 documented Zooplankton abundances between 1 and 2 orders of magnitude lower than in neighboring systems, although the large mesh size (250 µ) used by Rulifson et al. (1993) may have in part accounted for the low numbers reported. There are no known published studies of Zooplankton composition and abundances in this system prior to Rulifson et al. (1993), so it is unclear whether these low numbers are indicative of the long-term history or capture periods of low abundance.
Average Zooplankton abundance in several North Carolina coastal river systems as determined by various studies.
Numerous studies suggest that stock recovery begins with habitat restoration (i.e., dam removal) and the need to improve the availability of habitat for spawning adults (Beasley and Hightower 2000; Moser et al. 2000). While we agree, we propose an alternative hypothesis. We hypothesize that the precipitous decline in anadromous alosines during the past 30 years reflects changes in the quality and availability of suitable nursery habitat for early life stages. Variability in abiotic conditions and fluctuations in food abundance could structure nursery habitat and severely restrict recruitment, thereby limiting the recovery of these important fisheries.
The goal of this study was to investigate the ecological processes influencing the recruitment of larval alosines to nursery habitats within the lower Roanoke River and Albemarle Sound. The results were compared with those of previous studies to identify long-term fluctuations in recruitment dynamics, food availability, and water quality. Larval feeding ability was assessed through development of mouth gape models for each alosine species. Spatial and temporal overlap between larval alosines and size-appropriate Zooplankton prey were analyzed to determine whether match/mismatch regulation could exist and result in food limitation or starvation.
Roanoke River and Albemarle Sound.—The Roanoke River basin is the largest basin of any North Carolina estuary, encompassing 25,035 km2 (Konrad 1998; NCDENR 2000). The Roanoke River originates in the Blue Ridge Mountains of Virginia and flows southeast approximately 660 km to where it empties into Albemarle Sound, North Carolina (Konrad 1998; Pearsall et al. 2005). The Roanoke and Chowan rivers are the two main tributaries emptying into Albemarle Sound. The Roanoke River accounts for over 50% of the freshwater input into Albemarle Sound (Gray and Copeland 1989). It is one of the largest alluvial rivers on the East Coast. The lower region below the fall line is surrounded by bottomland hardwood floodplain forest and is the largest and least fragmented ecosystem of this type in the mid-Atlantic region (NCDENR 2000; Pearsall et al. 2005).
Albemarle Sound is a shallow estuary with mean depths less than 5 m that is part of the Albemarle—Pamlico estuarine system. This system is made up of broad, shallow, drowned river valleys. It is the second largest estuary and the largest lagoonal estuary in the United States. Pamlico Sound and Albemarle Sound are the two main basins in this system. Albemarle Sound is the northernmost basin and is separated from Pamlico Sound by Croatan and Roanoke sounds (Gray and Copeland 1989). The Outer Banks form a barrier separating Albemarle Sound from the Atlantic Ocean. Oregon Inlet is located south of Albemarle Sound and acts as the only source of saltwater intrusion. This limited saltwater intrusion, combined with high freshwater input from several rivers, results in Albemarle Sound's having salinity values less than 5‰ (Copeland et al. 1983; Pearsall et al. 2005). The Outer Banks also protect Albemarle Sound from gravitational tides, with water circulation being primarily wind driven (Copeland et al. 1983).
Field collection.—Larval fish and Zooplankton samples were collected concurrently, beginning at sunset, from March through June in 2008 and 2009. Samples were collected from 19 stations located in three areas within the lower Roanoke River and Albemarle Sound classified as River, Delta, and Sound (Figure 1). River is the area furthest upstream with seven stations scattered throughout the main stem of the river and its tributaries and distributaries between river kilometers (rkm) 9.5 and 22 as measured from the mouth of the Roanoke River (rkm 0). Delta is the transitional region between River and Sound, where the Roanoke, Middle, and Cashie rivers converge at the Highway 45 bridge before diverging and flowing into Albemarle Sound. There are two stations each in the Roanoke, Middle, and Cashie rivers, for a total of six stations. The Delta station furthest upstream is located at rkm 5 in the Roanoke River. Sound has six stations in Batchelor Bay, the western portion of Albemarle Sound. The stations extend 2–4 km from the mouths of the Roanoke and Middle rivers.
Zooplankton samples were collected using a 3:1 conical net with a 0.5-m opening and 90-µm nitex mesh. The plankton net was deployed using a vertical net haul technique whereby the net was lowered to the bottom and then pulled vertically through the water column. A preliminary study comparing the catch efficiency of vertical hauls, surface tows, and the use of a bilge pump to filter water through the net showed no significant differences in species abundance or composition when using vertical hauls or the pumping method, while abundances were significantly lower (F2,45 = 21.49, n = 48, P < 0.001) using surface tows (authors' unpublished data). The contents of the net were washed down and condensed into a sample jar and preserved with 5% buffered formalin.
Ichthyoplankton were collected using paired surface push nets mounted on the bow of the boat. Each net was housed in an aluminum frame with a 0.5-m-square opening. Each larval fish net had a 5:1 ratio and was constructed from 505-µ nitex mesh. A Sea-Gear Model MF315 flowmeter was mounted in the center of each net to estimate the amount of water filtered during each tow. The nets were pushed into the current for 2 min at a speed of 1.03 ± 0.11 m/s (Overton and Rulifson 2007). The contents of each net were condensed into a 1-L plastic collection jar. The contents of the left net were preserved with 95% ethanol, while the contents of the right net were preserved with 5% buffered formalin. The amount of ethanol used for preservation changed between the two sampling years. In 2008, the ratio of sample water to ethanol was approximately 70% sample water to 30% ethanol and there was a high amount of deterioration. In 2009, the amount of ethanol used for preservation was increased to account for 95% of the sample volume. To determine larval abundance, the catches between the two nets were averaged together. In 2009, samples were not collected during calendar week 20 in the Delta and during weeks 12, 16, and 20 in the Sound because of mechanical issues with the boat and severe weather.
Environmental parameters were recorded at each station during each sampling event. Air temperature (°C), wind speed (m/s), and direction were measured using a Skymate Model Sm-18. Surface and bottom water temperatures (°C), salinity, conductivity (µ8), and dissolved oxygen concentration (mg/L) were measured using a YSI Model 85 Multiparameter Water Quality Meter. A Hanna Model HI 98128 pH meter was used to measure surface pH. Current velocity (m/s) and direction were measured 1 m below the surface using a Marsh-McBirney FLO-MATE Portable Velocity Flow Meter, Model 2000.
Data processing.—For each sampling date, three Zooplankton samples from each area were randomly selected for processing. Subsamples were taken using a Hensen-Stempel pipette. Individuals were counted and identified using an Olympus Model SZX-ILLD100 stereomicroscope. Zooplankton were identified using taxonomic keys found in Thorp and Covich (2001) and Balcer et al. (1984). Zooplankton were lumped together into the following groups to allow for direct comparison with previous work by Rulifson et al. (1993) and Coggins (2005): calanoid copepods, cyclopoid copepods, copepod nauplii, rotifers, and cladocerans. Body length and widths were measured to the nearest 0.1 mm using Image-Pro Discovery 4.5.
Larval alosines were identified to species and enumerated, and their notochord lengths were measured to the nearest 0.1 mm. For the mouth gape analysis, only individuals preserved in formalin were used. Few American shad larvae were collected, so mouth gape analysis was conducted on larvae obtained from U.S. Fish and Wildlife Service's Edenton National Fish Hatchery in Edenton, North Carolina. Larvae were separated into 1-mm size bins based on notochord length, and up to 20 individuals were analyzed per size bin. All measurements were made using Image-Pro Discovery 4.5. For each fish, the upper and lower jaws were measured to the nearest 0.1 mm. The upper jaw was measured across the premaxilla and maxilla to the point of articulation with the dorsal process of the dentary. The lower jaw was measured along the length of the dentary to the point of articulation with the angular and maxillae. Mouth gape was calculated using the law of cosines for a mouth open at a 90° angle (Riley et al. 2009). This angle is considered the maximum functional degree of opening for feeding in most larval species (Turingan et al. 2005; Riley et al. 2009).
Statistical analyses.—An independent-samples t-test was used to evaluate whether Zooplankton abundances and environmental parameters differed between sampling years. Differences in abundance between this study and previous studies (Rulifson et al. 1993; Coggins 2005) were also assessed using this test. Spatial and monthly differences between abundances and abiotic factors were evaluated using a one-way analysis of variance (ANOVA). If the ANOVA was significant, the differences were further examined using the Ryan—Einot—Gabriel—Welch post hoc test, which holds the family wise alpha at 0.05.
Primer-E version 6 (Primer-E Ltd., Plymouth, UK) was used to evaluate the spatial differences in community structure. Prior to analysis, Zooplankton data were fourth-root transformed, with rare species being down-weighted (Clarke and Warwick 2001). Nonmetric multidimensional scaling was used as a visual representation, showing similarities within areas based on differences in Bray—Curtis dissimilarity values. The closer two points are located on the ordination plot, the more similar those two points are. One-way analysis of similarity (ANOSIM) was used to test whether there were significant differences in community structure among the three areas. The test statistic for ANOSIM, R, usually ranges from 0 to 1, with 0 indicating similarity among groups. Similarity percentages (SIMPER) were used to analyze which taxa were driving the dissimilarity among areas. The SIMPER procedure decomposes Bray—Curtis dissimilarity values and transforms them into percentage contributions from each taxon. SIMPER also allows for the identification of discriminating taxa, those that consistently contribute to the dissimilarity between two areas (Clarke and Warwick 2001; Clarke and Gorley 2006).
Prey and American shad stocking densities from Johnson and Dropkin (1995) and Riley et al. (2012). The stocking densities given on a per-liter basis represent the values used in the experiments. Those converted to a per-cubic-meter basis allow comparison with the abundances in this study. To determine the ratio of prey to larvae, the prey density was divided by that of American shad.
For comparisons of Zooplankton abundance and composition between this study and Rulifson et al. (1993), rotifers and copepod nauplii were excluded from the analysis. Rulifson et al. (1993) collected Zooplankton using a 250-µm-mesh net. Rotifers and copepod nauplii are typically less than 200 µm in size and are not efficiently collected in a 250-µm-mesh net (Thorp and Covich 2001).
The alosine abundance data did not meet normality assumptions (even with data transformation), so nonparametric tests were used for comparisons. Yearly differences in alosine abundance were evaluated using a Wilcoxon rank—sum test. A Kruskal—Wallis ANOVA was used to evaluate the differences in abundance among species and sampling areas. If the ANOVA was significant, post hoc comparisons were conducted using Wilcoxon rank—sum tests with the familywise alpha being controlled by a Dunn—Sidak adjustment.
Mouth gape models for each species were created using linear regression analysis. A one-way analysis of covariance (ANCOVA) was used to test whether mouth gapes differed significantly among species. Prey size estimates were based on the length when the yolk sac is absorbed and larvae begin to feed exogenously. Yolk sac absorption occurs at 6 mm in alewives and blueback herring, at 7 mm in hickory shad, and between 9 and 12 mm in American shad larvae (Lippson and Moran 1974). Unless otherwise noted, all statistical analyses were conducted using SAS 9.2 (SAS Institute, Cary, North Carolina).
To evaluate the spatial and temporal overlap between larval alosines and Zooplankton, weekly abundances for each group were plotted against each other for each sampling area. Only Zooplankton that were estimated to be size-appropriate prey at first feeding for all taxa were included, and the four alosine species were grouped together. Zooplankton abundances from American shad feeding studies were used to establish a baseline for high zooplankton—alosine overlap that supports optimal growth and survival (Johnson and Dropkin 1995; Riley et al. 2012, this special section). The stocking densities of larval fish and Zooplankton in laboratory feeding trials are often inflated and do not correspond to the values observed in field studies (De Lafontaine and Leggett 1987; Paradis et al. 1996). In Johnson and Dropkin (1995) and Riley et al. (2012) both American shad and Zooplankton were stocked at concentrations higher than those observed in this study. To make the concentrations of American shad and Zooplankton comparable to those observed in the Roanoke River and Albemarle Sound, the concentration of Zooplankton was divided by that of American shad. The baseline for high overlap, which supported optimal growth and survival in the two studies, was 125 zooplankton/L of American shad larvae (Table 2).
The environmental data (n = 97) are summarized in Table 3. Monthly patterns were observed for dissolved oxygen and air and water temperature. Dissolved oxygen followed the expected seasonal pattern of being highest in March and decreasing throughout the sampling season. Hypoxic conditions (<2.0 mg/L) occurred infrequently in late May and June, accounting for 1% of bottom dissolved oxygen readings. Dissolved oxygen also varied between years and was significantly higher in 2009 (7.9 ± 2.0 mg/L [mean ± SD]) than in 2008 (6.5 ± 1.5 mg/L). Air and water temperatures followed the opposite pattern, being lowest in March and increasing through June. Significant yearly differences were observed for pH and wind speed (Table 4). In 2008 (7.5 ± 0.2), pH was significantly higher than in 2009 (6.7 ± 0.4). Wind speed was also higher in 2008 (5.2 ± 5.1 m/s) than in 2009 (1.7 ± 1.1 m/s).
Salinity, current velocity, and wind speed were significantly higher in the Sound (Table 5). Mean salinity was 0.1 ± 0.0‰ in both the River and Delta, significantly lower (F2,94 = 35.69, n = 97, P < 0.001) than in the Sound, where it ranged from 0.1 to 2.8‰ with a mean ± SD value of 0.5 ± 0.6‰. The majority of salinity values did not exceed 1.0‰. Current velocity was similar in the River and Delta, ranging from 0.0 to 0.83 m/s, with a mean ± SD velocity in both areas of 0.1 ± 0.1. Current velocity was significantly higher (F2,94 = 3.98, n = 97, P = 0.022) in the Sound (0.2 ± 0.1 m/s), ranging from 0.0 to 1.2 m/s, and currents were most frequently from the west. Generally, wind speeds were less than 10.0 m/s and were typically from the south or southwest. The wind speed in the Sound (6.4 ± 5.5 m/s) was significantly higher (F2,94 = 13.69, n = 97, P < 0.001) than in the River (1.7 ± 2.2 m/s) and Delta (2.8 ± 2.8 m/s) areas. The depth at the River (4.8 ± 0.7 m) stations was significantly higher (F2,94 = 85.44, n = 97, P < 0.001) than at the Delta (3.2 ± 0.6 m) and Sound (3.3 ± 0.3 m) stations (Table 5).
Average monthly values (means ± SDs) for environmental variables collected in March–June 2008 and 2009 in the lower Roanoke River and Albemarle Sound.
Comparison between the values (means ± SDs) of environmental parameters in 2008 and 2009 in the lower Roanoke River and Albemarle Sound.
Zooplankton Abundance and Taxonomic Composition
Zooplankton abundances were highly variable. Mean ± SD Zooplankton abundances (number/m3) were not significantly different (t = -1.47, df = 95, n = 97, P = 0.144) between 2008 (7,214 ± 8,048) and 2009 (9,774 ± 11,967). Area had a significant (F2,94 = 12.98, n = 97, P < 0.001) effect on Zooplankton abundance, the Sound area having a significantly higher abundance (16,547 ± 14,678) than the River (4,934 ± 3,806) and Delta (4,647 ± 2,846) areas. There were no clear temporal patterns in any area, so the years were evaluated separately. The abundances in the Sound were the most variable, with the widest range. Even though overall abundances were significantly higher in the Sound, the lowest observed abundance in any area was in the Sound at week 20 in 2008 (935 ± 496). The highest abundance occurred in week 12 (33,384 ± 47,621). In 2009, the lowest abundances were once again in the later part of the sampling season, occurring at week 21 (2,710 ± 466). The peak abundances in the Sound occurred the following week and were the largest observed during the study (51,816 ± 52,092). Temporal patterns were more consistent in the River. The highest abundances were observed in week 13 in 2008 (13,562 ± 10,797) and 2009 (19,751 ± 25,719). In both years, the lowest values were observed in early summer. The lowest abundances occurred in week 22 in 2008 (1,380 ± 20) and in week 23 in 2009 (1,261 ± 1,430). In the Delta, the 2 years exhibited opposite patterns. In 2008, Zooplankton peaked in late June (week 26) with an abundance of 10,672 ± 7,901 and was lowest in March (week 13) with an abundance of 997 ± 598. In 2009, Zooplankton abundance was highest in week 11 (12,727 ± 4,235) and lowest in week 24 (1,802 ± 940) (Figure 2).
Mean values of environmental variables in the three sampling areas in the lower Roanoke River and Albemarle Sound. Means with common letters are not significantly different at the 0.5 level according to the Ryan—Einot— Gabriel—Welch procedure.
Zooplankton communities were dominated by five taxa: calanoid copepods, cyclopoid copepods, copepod nauplii, rotifers, and cladocerans. Calanoid and cyclopoid taxa include both copepodite and adult life stages. Several families of cladocerans were identified in this study, including Daphniidae, Bosminidae, Sididae, Chydoridae, and Leptodoridae. These five taxa account for a minimum of 96% of the composition of each area across both years. Some of the less common taxa included ostracods, gammarid amphipods, and harpacticoid copepods.
Zooplankton community structure varied both temporally and spatially. Temporal differences occurred on both monthly and yearly scales. In 2008, monthly changes in composition were observed. In the River, rotifers were dominant in March, representing over 60% of Zooplankton. Rotifers were less abundant in April, and cladocerans were the dominant taxa, representing 37% of Zooplankton. In May (47%) and June (36%) rotifers were again dominant. From March through May calanoid copepods were not common in the River, but in June there was an increase in abundance (32%) and they were almost as abundant as rotifers. In the Delta, calanoid copepods (65%) were dominant in March but had low abundances April–June. In April and May, copepod nauplii and rotifers had similar abundances and were the most common taxa. Rotifers accounted for 32% of the composition in April and 33% in May. Copepod nauplii represented 33% in April and 32% in May. Rotifers were the dominant taxa in June (48%). In the Sound, copepod nauplii were the dominant taxa for all of 2008. In 2009, Zooplankton communities did not follow the temporal patterns observed in 2008. One of the biggest differences was the increased dominance of rotifers. Rotifers were the dominant taxa, except during March in the River and Sound, and always accounted for at least 35% of the Zooplankton (Figure 3).
The Zooplankton communities in this study were similar to those in Rulifson et al. (1993) in the River and Delta but differed in the Sound. In the River and Delta, Bosminidae, Daphniidae, and cyclopoid copepods were the most abundant taxa in both studies (when copepod nauplii and rotifers are excluded). In the Sound, the Zooplankton community was dominated by Bosminidae, Daphniidae, and unknown cladoceran species in Rulifson et al. (1993). Throughout the multiyear study, cladocerans always accounted for at least 40% of Zooplankton in the Sound. In contrast, in this study the Zooplankton composition was dominated by calanoid copepods, which made up 76% of the Sound community.
There were also major differences in the River community structure between this study and Coggins (2005). In this study, the main River taxa were rotifers (41%), copepod nauplii (18%), Daphniidae (11%), Bosminidae (11%), and cyclopoid copepods (10%). The Zooplankton community in Coggins (2005) was dramatically different. The main difference is the lack of rotifers, which accounted for 0% of the composition in Coggins (2005). The main River Zooplankton taxa in Coggins (2005) were cyclopoid copepods (35%), Daphniidae (19%), copepod nauplii (13%), Bosminidae (12%), and Diptera (10%). Daphniidae, Bosminidae, and copepod nauplii had similar prevalences in the Zooplankton communities in the two studies. Few insects were collected in this study, composing less than 1% of the community versus 10% in Coggins (2005).
A one-way ANOSIM indicated weak (global R = 0.298) but significant (P = 0.001) differences in Zooplankton composition among the areas (Figure 4). Post hoc comparisons revealed that the River and Delta were not significantly different (R = 0.054, P = 0.13), while the Sound was significantly different from both the River (R = 0.527, P = 0.001) and Delta (R = 0.357, P = 0.001). SIMPER analysis comparing the Sound with both the River and Delta showed the level of dissimilarity among comparisons. The Bray—Curtis average dissimilarity was 36.9 in the Sound—River comparisons and 34.0 in the Sound—Delta comparisons. Calanoid copepods contributed the most to the differences between the Sound and other areas. Copepod nauplii and rotifers were also important to the dissimilarity among the areas, ranking as the second- and third-highest contributors. Their order differed between the Sound—River and Sound—Delta comparisons. In the Sound—River comparisons, Ostracoda and Chydoridae were identified as discriminating taxa. Harpacticoid copepods were the discriminating taxa in the Sound—Delta comparisons (Table 6).
SIMPER analysis evaluating the dissimilarity between areas identified as significantly different using ANOSIM. Abundances are fourth-root transformed. The average Bray—Curtis dissimilarity scores are listed as average dissimilarity. Diss/SD identifies how consistently taxa contribute to the dissimilarity. Asterisks identify discriminating taxa. The contribution percentage is the amount of dissimilarity that can be attributed to a taxon.
ALOSINE ABUNDANCE AND DISTRIBUTION
Differences in alosine abundance (number/100 m3) and composition were observed between sampling years. Abundances were significantly higher in 2009 (30.8 ± 149.8; P ≤ 0.001) than in 2008 (4.1 ± 20.9). Alewives (28.5%) were the most abundant alosine in 2008, followed closely by blueback herring (26.7%) and hickory shad (24.4%). In 2009, blueback herring were dominant, accounting for 64.8% of the larval alosines collected. Hickory shad (23.8%) were the second most abundant species. American shad were the least common alosine in both years.
Weekly patterns were different between the two years. Larval alosines were collected during all weeks of sampling in 2008 but were not present the first week of sampling in 2009. Blueback herring were the first species collected in both years. Alewives were observed earlier in 2008 and were the second species collected. In 2009, hickory shad were the second species collected. American shad larvae were not common (Figure 5). Mean alosine abundances were significantly different (P ≤ 0.001) among the three sampling areas. The highest abundances were observed in the River (21.0 ± 127.6), with successively lower ones in the Delta (7.4 ± 35.5) and Sound (4.6 ± 24.8). This trend was consistent across both years and for all species.
There are no clear patterns when species abundance is analyzed on temporal and spatial scales. Alewife abundances were highest in the River during March of both years. In 2008 alewife abundances peaked in March, whereas in 2009 peak abundances were observed in April. In the Sound, alewives were present in all months in 2008 but were absent in March of 2009. Blueback herring abundances were consistent in 2008 across all months and within each area. There was an increase in blueback herring abundances in 2009 and abundances varied across the months. In all areas, abundances peaked in April and were at least twice as large as those observed in March. After peaks in April, there was a sharp decrease in abundances for the remaining months. In 2009, the abundance patterns for hickory shad were similar to those for blueback herring. Hickory shad abundance also increased in 2009 with peaks in April. In 2008, monthly hickory shad trends were similar in the River and Delta, with highest abundances in April (Figure 5).
ALOSINE MOUTH GAPE ANALYSIS AND PREY SIZE ESTIMATES
Larval alosine notochord lengths ranged from 3 to 14 mm, with 97% of larvae having lengths of 7 mm or less. Over 90% of the alewives, blueback herring, and American shad were between 4 and 7 mm, with lengths exceeding 8 mm rarely being collected. American shad larvae had the narrowest length distribution and were typically larger, with all but one larva being 7 mm or more (Figure 6). The mean ± SD length was similar between alewives (4.7 ± 1.2 mm) and blueback herring (4.6 ± 1.0 mm). The mean length of hickory shad (6.5 ± 1.0 mm) was larger than those of alewives and blueback herring. American shad larvae had the largest mean length (8.7 ± 2.1 mm)
For all species, there was a strong linear relationship between mouth gape and notochord length (Figure 7). A one-way ANCOVA indicated that there were significant differences (F4,459 = 2,115.0, P < 0.001) in mouth gape size among species. American shad larvae had the largest mean mouth gape (mean, 0.67 mm), followed by alewives (0.57 mm), blueback herring (0.56 mm), and hickory shad (0.53 mm). The alewife and blueback herring mouth gapes did not significantly differ from each other, while all other comparisons were significantly different.
For mouth gapes calculated at a 90° opening, the estimated prey size at yolk sac absorption was similar among alewives, blueback herring, and hickory shad larvae but larger for American shad larvae. At 6 mm, alewives and blueback herring have a 400-µ mouth gape with an estimated maximum prey size of 200 µm. Hickory shad at 7 mm have a 430-µm mouth gape and an estimated prey size of 215 µm. At 9 mm, American shad have a mouth gape that is twice as wide at first feeding, 820 µm with a maximum prey size of 410 µm (Figure 7).
The Zooplankton size distribution was estimated for the most abundant taxa. These taxa included calanoid and cyclopoid copepods, copepod nauplii, rotifers, Daphniidae, and Bosminidae, which accounted for 98% of total abundance. Mean Zooplankton body lengths and widths are plotted in Figure 8. Based on these measurements, Zooplankton taxa were separated into two size-classes. The smaller taxa included rotifers and copepod nauplii, with remaining taxa comprising the larger sizeclass. Along the length axis the two groups begin to separate at 200 µm, and across the width axis they begin to separate at approximately 120 µm. Based on mouth gape estimates, copepod nauplii and rotifers are size appropriate for alewives, blueback herring, and hickory shad larvae at first feeding. Bosminidae and cyclopoid copepods at the low end of their size distributions could also serve as prey for these species. American shad larvae have a wider potential prey choice, including rotifers, copepod nauplii, cyclopoid copepods, and Bosminidae. Smaller Daphniidae and calanoid copepods are also within the estimated prey size range of American shad.
LARVAL ALOSINE AND ZOOPLANKTON SPATIAL AND TEMPORAL OVERLAP
There was a high degree of spatial and temporal overlap between larval alosine and Zooplankton abundances in all three sampling areas. Weekly mean alosine abundances were generally less than 1/m3. Mean weekly Zooplankton abundances, including all taxa, ranged from 934 to 51,815/m3. When the most suitable prey (copepod nauplii and rotifers) were evaluated separately, weekly abundances ranged from 283 to 51,034/m3. There was always overlap between larval alosines and size-appropriate prey. Spatially, the highest overlap occurred in the Sound, where Zooplankton abundances were the highest and larval alosine abundances were the lowest. Higher weekly peaks occurred in the River than in the Delta, but there was still overlap in Zooplankton and larval alosines in all areas (Figure 9).
The Zooplankton abundances in all three areas were significantly higher than those reported in previous work and more similar to those observed in other North Carolina coastal river systems (Table 1). Many factors could be attributing to this increase in Zooplankton abundance. Rulifson et al. (1993) measured phytoplankton concentrations in the Roanoke River and Albemarle Sound and found that they were large enough to support much higher Zooplankton abundances than were observed. The authors hypothesized that various environmental factors were responsible for these abundances. The values for dissolved oxygen, salinity, temperature, turbidity, and pH were similar between this study and Rulifson et al. (1993). The main difference between the two studies was the mesh size used to collect Zooplankton. Rulifson et al. (1993) used a 250-µm-mesh net and we used a 90-µm-mesh net. We attempted to correct for the differences in the sizes of plankton by removing rotifers and copepod nauplii from our analysis. These two groups are typically less than 200 µm and are not efficiently collected in a 250-µm-mesh net (Thorp and Covich 2001). Even after removing the smaller plankton from the analysis the Zooplankton concentrations and composition were still different between the two studies (Binion 2011). These differences are most likely simply due to the difference in mesh size.
Significant differences in Zooplankton composition and abundance were observed between the Sound and the other areas sampled. SIMPER analysis showed that calanoid copepods were the taxa most responsible for these differences. Calanoid copepods were more abundant in the Sound and comprised a higher percentage of the Zooplankton. Higher abundances of copepod nauplii and rotifers also contributed to the Sound's being significantly different. In the River and Delta, higher abundances of cyclopoid copepods, cladocerans, and ostracods also contributed to the significant differences from the Sound (Table 6). Salinity was significantly higher in the Sound and these differences in composition may be explained by the salinity preferences of the different taxa.
The results of the one-way MANOVA suggest that the majority of the variance (η;2 = 89%) in Zooplankton trends and the environment can be explained by the effect of area. Significant differences in Zooplankton abundance and composition were not observed between the River and Delta, despite the strongly significant MANOVA. Differences between the two areas were more pronounced in 2008, but in 2009 rotifers were consistently dominant and the composition was similar between the areas (Figure 3). It is possible that the surge of rotifers is masking the differences between the River and Delta. In the Rulifson et al. (1993) study, Delta Zooplankton abundances were significantly higher in the River.
Larval Prey Size and Mouth Gape
Differences in mouth gape in coexisting larvae allow for a reduction in interspecific competition (Crecco and Blake 1983; Bremigan and Stein 1994; Makrakis et al. 2008). In this study, alewife and blueback herring mouth gapes were not significantly different from each other but all comparisons with American shad and hickory shad showed significant differences. Our mouth gape calculations for blueback herring and hickory shad are comparable to those calculated by Crecco and Blake (1983). When mouth gape size at yolk sac absorption was calculated, alewife, blueback herring, and hickory shad had similar mouth gapes and American shad larger ones. Within this system, there appears to be interspecific prey competition among those first three alosine species.
Zooplankton abundances were higher in this study than in those previously conducted and there was a high ratio of zooplankton to larval alosines. This high overlap was observed even when only taxa that could be consumed by all alosine species, copepod nauplii and rotifers, were included in the analysis. Competition between larval alewives, blueback herring, and hickory shad may be reduced by the large volume of available prey. Studies evaluating diet niche overlap found dissimilarities among larval fish with similar mouth gapes (Gaughan and Potter 1997; DeVries et al. 1998; Makrakis et al. 2008). It has been hypothesized that when Zooplankton abundances are high, competition is reduced and larval fish with similar mouth gapes exhibit different feeding behaviors and selectivity that is afforded by high prey availability (Gaughan and Potter 1997).
The maximum larval fish mouth gape is typically estimated for 90°, but some larval fish are capable of opening their mouths to 120° (Riley et al. 2009). The mouth morphology of hickory shad is different from that of the other alosine species. The lower jaw of hickory shad slopes at an angle more than 40°, while the lower jaws of the other species slope at angles less than 40° (Walsh et al. 2005). This difference in morphology may allow hickory shad to open their mouths at larger angles, enabling them to consume larger prey at smaller sizes. This could be one of the reasons that hickory shad populations have remained more stable than those of the other alosines. In the Roanoke River and Albemarle Sound, previous Zooplankton abundances have been significantly lower (Rulifson et al. 1993; Coggins 2005). During this time, American shad and river herring stocks were below historical levels despite increased regulations and management. Hickory shad stocks have declined slightly but are not considered depleted (ASMFC 2007; Greene et al. 2009; NCDMF 2010). If hickory shad are capable of feeding at a larger mouth gape opening, this may reduce competition between them and river herring. At first feeding, hickory shad may be capable of consuming larger prey items than predicted in this study, allowing them to feed on prey resources not available to river herring. Feeding studies need to be conducted with hickory shad to determine whether they are capable of opening their mouths at larger angles.
The hypothesized mouth gape and maximum prey size estimates for alewives and blueback herring are supported by diet analyses of larvae collected in the Roanoke River and Albemarle Sound (Figure 7). For larvae of all sizes, copepod nauplii and rotifers accounted for 85% of stomach contents. First-feeding larvae feed primarily on copepod nauplii, rotifers, small Bosminidae, and small cyclopoid copepods, as predicted by the mouth gape models. Throughout ontogeny, there was little change in the diet composition of larval alewives and blueback herring, but as larval size increased prey size increased as well. American and hickory shad larvae were not included in the stomach content analysis (Riley 2012).
LARVAL ALOSINE AND ZOOPLANKTON SPATIAL AND TEMPORAL OVERLAP
Across both years in all areas, there was a high amount of overlap between Zooplankton and larval alosines. Laboratory studies examining growth and survival are typically conducted with larval alosine and Zooplankton abundances higher than those observed in this study. In Riley et al. (2012), larval American shad growth and survival were evaluated at three different prey abundances equivalent to 1,000, 50,000, and 500,000/m3. These abundances reflect all Zooplankton and not just size-appropriate prey. Larval American shad were stocked at abundances equivalent to 4,000 individuals/m3. Growth was significantly higher in treatments with the two highest abundances, but survival rates were similar in all three. Except for one observation in the Sound, Zooplankton abundances never exceeded 50,000, but alosine abundances were typically less than 1/m3. In Johnson and Dropkin (1995), larval American shad were stocked at the equivalent of 8,000/m3 and Artemia nauplii were stocked at 500,000 and 1,000,000/m3. Growth was not different between the two treatments, but survival was higher in the treatment with larger nauplii abundances. If Zooplankton abundance is divided by larval shad abundance, the ratio of Zooplankton to fish is 125:1 for both studies (Table 2). Ratios this low were never observed during this study, even when zooplankton abundances were calculated with rotifers and copepod nauplii only (Figure 9). This suggests that the larval alosines in this system are not food limited and that all three areas are suitable as larval alosine nursery habitat based on prey availability.
The reasons for the declining recruitment of river herring in Albemarle Sound remain unclear. The variability in the annual recruitment of alewives in the Great Lakes has been linked to spawning stock size, salmonid predation, and winter severity (Henderson and Brown 1985; Madenjian et al. 2005). River herring are included in the diet of large estuarine predators, including striped bass Morone saxatilis. However, predation by striped bass is not the cause of poor recruitment in Albemarle Sound (Tuomikoski et al. 2008). The failure of alosine stocks to recover in this system is not the result of food limitation during early life history. Recruitment failure may be related to the demographics of the spawning population or other factors operating on multiple scales.
This research was supported by the North Carolina Sea Grant College Program (project R/MRD-55). We thank the following people for assistance in the field and laboratory: J. Kim, N. Myers, B. Deehr, R. Spidel, J. Edwards, J. Osbourne, B. Davis, J. Robinson, A. Angula, N. Duquette, B. Cooper, and D. Mayer. We also thank S. Jackson and the staff of the U.S. Fish and Wildlife Service's Edenton National Fish Hatchery for providing American shad larvae for mouth gape analysis. Gratitude is extended to R. Rulifson, D. Kimmel, and J. Rock and two anonymous reviewers for their helpful comments on this work.