Growth Rates of Juvenile Winter Flounder under Varying Environmental Conditions

Introduction

Since Hjort (1914) developed his critical-period concept for larval fishes, fisheries scientists have been trying to understand the effects of environmental factors and their timing on the growth and recruitment of fish. In an update of his match–mismatch hypothesis, Cushing (1990) suggested that the influence of environmental factors on growth and recruitment could continue through and beyond metamorphosis. It is widely believed that, with rapid growth, juvenile fishes have an increased probability of surviving the first winter (Post and Evans 1989b), and there is less time spent in smaller, more predation vulnerable size ranges (Parker 1971; Post and Evans 1989a). However, higher growth rates also demand higher food intake, thereby increasing search time and vulnerability to predators (Pepin 1991).

Winter flounder Pleuronectes americanus have a unique life history (Pearcy 1962) that offers the opportunity to study distinct cohorts of juveniles with limited movement (Saucerman 1990). Growth rates of individual winter flounder during settlement season have been compared between estuaries at specific sites and periods during the settlement season (Sogard and Able 1992) and over many years (Sogard et al. 2001). Interannual variability in growth rates appears to be controlled by local environmental factors or genetic differences among populations (Sogard et al. 2001). Caging studies indicate that location within an estuary and seasonal timing at that location are important in determining growth of juvenile flounder (Meng et al. 2000; Phelan et al. 2000, Manderson et al. 2002).

Generally, little information exists on the individual growth rates of free-living postmetamorphic individuals in the nursery habitat during the entire settlement season. Sogard and Able (1992) suggest that growth, particularly during the first few weeks of juvenile life, may be crucial in determining an individual's probability of survival. Otolith increments provide a means of obtaining individual growth rates and dates of metamorphosis from fish living under natural conditions (Sogard and Able 1992). These individual measurements along with abundance surveys and environmental monitoring would produce a time series of growth estimates throughout the settlement season, thus, indicating which times and general locations offer the best growth conditions.

Although caging studies can provide precise measurements of growth during discrete periods and locations, they may not reflect growth rates of unrestrained fish. In this study we examine otolith-derived growth rates of 258 free-living winter flounder caught at 18 fixed stations within the river and bay of the Navesink–Sandy Hook Bay estuarine system in New Jersey during the 2000 field season (April–October). We examined growth in relation to concurrently measured environmental variables during the entire settlement season (April–June). To examine localized genetic differences in growth between the two regions of the estuary and over time, we also conducted laboratory experiments on similarly sized fish collected from river and bay habitats throughout the settlement season, growing them at a constant temperature. Our field work was compared with previously conducted studies using cages in the field (Manderson et al. 2002). The fall sampling allowed us to determine which dates of metamorphosis would be prevalent in fall-captured age-0 fish and whether these dates would be related to growth rate estimates for winter flounder during those periods.

Methods

Laboratory experiments

Postmetamorphic winter flounder were captured on four different sampling dates between May 10 and June 21, 2000, from Plum Island Cove in Sandy Hook Bay and the Navesink River (Figure 1) by using a 7.8-m beach seine with 3.1-mm mesh. Fish 20–35 mm total length (TL) from each site were collected and brought back to the laboratory. Once in the laboratory, fish were acclimated to a constant temperature of 15°C in 25‰ seawater, measured to the nearest 0.01 mm TL and standard length (SL) and once blotted dry, weighed to the nearest milligram. Ten fish were randomly selected, placed individually into opaque 6-L buckets containing 1 L of 15°C seawater (25‰) and 50 mL of fine-grained sand. Containers were placed in a room with constant temperature and exposed to fixed light cycle: 14 h light and 10 h dark. Each fish was fed natural copepods (750/d) plus laboratory-reared brine shrimp Artemia sp. (150/d). Water and food were replaced every other day for the duration of each growth experiment (12 d). At the end of each growth experiment, fish were again measured and weighed, as previously described, and preserved in 95% ethanol.

Field samples

For the field survey, the study area was divided into two regions (river and bay), each with nine randomly assigned fixed stations (Figure 1). Juvenile winter flounder were surveyed during eight separate biweekly collections from March 7 to June 27, 2000 (Table 1) and captured with a 1-m beam trawl (3-mm mesh liner, 2 tickler chains). During each survey, triplicate trawl tows (speed = 1 m/s, duration = 30 s) were made at the 18 stations. An additional beam trawl survey was conducted on July 5. Winter flounder that were collected with the beach seine and not used in laboratory experiments and those that were collected in the beam trawls were preserved in 95% ethanol and brought back to the laboratory for otolith analysis. Additional sampling to collect otoliths from surveyed fish continued into the fall at our seine sampling sites on September 7, 12, and 29, 2000. We also collected fish throughout the estuary in the fall with beam and otter trawls on October 20–24, 2000 (see Stoner et al. 2001 for trawl survey methods).

Table 1. Date of capture and number of age-0 estuarine winter flounder used in otolith analysis, by gear type

Otolith analysis

Sagittae from age-0 winter flounder were examined to determine the time of occurrence of accessory primordia. The formation of accessory primordia marks a period in development during metamorphosis when eye migration nears completion (Sogard 1991). When possible, only the left sagitta was used due to the asymmetric growth of the right (Sogard 1991); the right sagitta served as a reserve when needed. Counts between left and right sagittae are not significantly different (Sogard 1991).

Allowing adjustment for varying focal planes, three increment counts were made under a compound microscope at 200–400× power. Counts were made along the axis of maximum growth from the accessory primordia to the rostrum. The mean value of the counts was used as an estimate of the number of daily increments (DI). For each fish we back-calculated date of metamorphosis (DOM) by counting daily increments and subtracting them from date of capture (DOC).

Assuming an average size at metamorphosis (SM) of 10 mm (Rose et al. 1996), we then randomly assigned SM from a normal distribution (mean = 10 mm, SD = 1) to each fish and calculated growth rates (mm/d) as follows:

Physical and biological variables

During the spring surveys, we monitored physical and biological variables. Bottom water temperatures (°C) were monitored hourly with electronic recorders (Onset Corp.) deployed on the substrate at each station. Salinity was measured with a probe (Yellow Springs Instruments) four times (twice/week) at each station during each run of the survey. For each fish within a region of capture (between its DOM and DOC), we calculated a regional (river or bay) temperature average and range from hourly temperature recorders. We also calculated salinity averages from DOM to DOC for each fish by using the same method, although these data were collected on a coarser temporal scale.

We used planktonic food in our laboratory work; however, benthic food items become a larger part of the diet of winter flounder as they increase in size and as the settlement season progresses (Stehlik and Meise 2000). We therefore measured benthic food abundance for our field work. On the third week of each month, triplicate benthic samples were taken using Van Veen grabs (400 cm²) at each station. A core sample (4.8 cm diameter, 3 cm depth) was taken from each grab for analysis. The benthic samples were sorted to a lowest possible taxon; an average (number/m²) of each taxon for the samples was taken and summed for the total number of available organisms to be eaten by juvenile winter flounder. We used the following taxa for food items: softshell Mya arenaria, Capitellidae, Nereididae, Spionidae, Ampeliscidae, and Mysidaceae (Stehlik and Meise 2000). Analysis of variance (ANOVA) and Tukey's studentized range (Sokal and Rohlf 1981) were used to test differences of the summed total of the above species between the four benthic sampling dates over the total estuary.

We used analysis of covariance (ANCOVA) to test for regional and temporal differences in growth rate in our laboratory experiments, using initial SL as the covariate. In the field, because we were concerned with growth rates from newly settled fish, only fish caught through July 5 were used in statistical analyses. Fish collected after July 5 were used to see which settlement dates were successful. Winter flounder growth rates were divided into regions (river and bay) and ANOVA was used to test the null hypothesis that there were no regional differences. We used stepwise general additive models (GAM) to examine environmental effects on the growth rate of field-captured juvenile winter flounder. Stepwise GAMs are nonparametric models and allow the use of both linear and nonlinear (in the form of scatter plot smoothers) terms. Akaike Information Criterion (AIC) was used to measure the goodness of fit of the model. From the AIC values for each fitted model we selected a final model that makes the most biological sense (Venables and Ripley 1997). Growth rate was modeled against linear and nonlinear terms of regional temperature averages, regional temperature range, regional salinity averages, food abundance, and date of capture. Date of capture was used as a surrogate variable for environmental parameters we may have missed and that vary consistently through time. Residuals were examined for independence, and variables were considered significant at P = 0.05. Independent effects, F-test value, and probability are derived for the two best models from ANOVA tests of the full models against stepwise removal of each independent term (Venables and Ripley 1997).

Results

Growth

Fish originating from either Sandy Hook Bay or the Navesink River and reared in water temperatures of 15°C showed no difference in growth due to capture location (ANCOVA: P > 0.05, initial SL as covariate), but the growth of fish in both regions decreased through runs of the experiment (P < 0.05). Growth calculated from standard lengths for individual fish ranged from 0 to 0.47 mm/d and the overall average was 0.15 mm/d (coefficient of variation [CV, defined as SD/mean] = 0.64 for the entire study). The highest growth rates occurred in May for fish from both regions. Over time, growth of bay fish was reduced to approximately one-fifth of the original May rate; the river fish showed a marked increase in growth during July, making it the second highest rate for river fish in the series (Figure 2a).

There was no significant relationship between growth rate and juvenile winter flounder numbers at a station during any of the surveys (Pearson correlation R values range from −0.18 to 0.42, P > 0.05). Standard length was positively correlated with DOC (R = 0.50, P < 0.01). Widths between otolith increments did not appear irregular, suggesting that none of the fish experienced poor growth or starvation (Sogard 1991). Age-0 winter flounder were collected in beam trawls starting April 11 and in seine hauls starting May 10. Fish appeared in the river approximately 2 weeks before those caught in the bay (Figure 2b). Growth was highly variable between stations (overall mean = 0.93 mm/d, CV = 0.44, N = 258; Figure 2b) and, although not significant (ANOVA, P > 0.05), appeared to be slightly higher in the Navesink River (overall mean 1.08 mm/d, CV = 0.40, N = 125) than in Sandy Hook Bay (overall mean 0.79 mm/d, CV = 0.42, N = 133). For comparison purposes we applied Sogard's et al. (2001) “interim growth equation,” where growth is broken down into growth just after metamorphosis, interim growth, and growth just before capture. We found our calculations to be comparable to theirs when averaged over the postmetamorphic period (Sogard et al.: mean growth = 0.58 mm/d, CV = 0.28; this study: mean growth = 0.74mm/d, CV = 0.28). Our growth and CV estimates (range 15–40%) from all sampling dates and stations in 2000 were also comparable to those reported in Sogard et al. (2001) for a single sampling date and site over 10 years (1.1 mm/d, CV range = 6.4–30.8%) in Sandy Hook Bay.

The ANOVA and Tukey's tests revealed significant differences (P < 0.05) in food availability between benthic sampling dates. June was significantly higher than the three previous months (Figure 3).

A summary of the physical variables used in the model building by survey date is shown in Figure 4. Average temperature and salinity were highest during the last two surveys. The temperature range experienced by individual fish was impossible to show; however, we do show minimum and maximum temperatures during each sampling period in each region.

Table 2 lists both the lowest AIC model (model 2) and the preferred model (model 1). The AIC criterion is different by two-tenths. The deviance explained differs by 2 percentage age points. The difference between these two models was the nonlinear model used to fit salinity in model 2. Whereas model 2 explained a small amount of additional variance (2%), the complexity of the model (an additional 4 df) did not appear necessary, when examined against the data, nor justified by the variance explained. In model 1, growth was significantly and linearly related to average regional temperature (positive), temperature range (negative), and salinity (negative). Food abundance was nonlinearly related to growth, indicating that growth rate declined as food levels dropped below 50 individuals/m² (Table 2; Figure 5). The ANOVA test between the full model (all terms) and a stepwise (with replacement) removal of single terms revealed all terms were significant in both models, average regional temperature and food abundance explaining the most variance in model 1 and food abundance explaining the most variance in model 2.

Table 2. General additive model (GAM) results for the growth rate of winter flounder with temperature average (°C), food abundance (number/m²), temperature range (°C), and salinity (‰). The sign of all linear effects is indicated in parentheses after each variable; nonlinear terms are preceded by asterisks. There were 259 observations for each model. The Akaike information criterion generated in the stepwise GAM algorithm was 32.87 for model 1 and 32.68 for model 2. The independent-effects column lists the percentage of the total variance explained by each variable. The F-value and probability are derived from analysis of variance tests of the full model against stepwise removal of each independent term (Venables and Ripley 1997)

Discussion

In our study, temperature proved to be the most pivotal and influential element in the growth rate of juvenile winter flounder. In an earlier study conducted with caged fish in the Navesink–Sandy Hook Bay estuarine system, Manderson et al. (2002) found growth to be highest at cooler temperatures (16°C) and to decline at temperatures above 22°C. Delong et al. (2001) found growth rates to be weakly and negatively associated with temperature averages in Narragansett Bay, Rhode Island. Here, highest growth occurred at temperature averages around 18°C. We found increasing growth within the regional temperature averages that our fish were exposed to during the study (9.85–22.75°C). Our relationship was not curvilinear, and growth did not decline after 16°C, as suggested by Manderson et al. (2002). However, our fish were not confined by cages and probably experienced integrated regional temperatures during their lifetime. The negative association between regional temperature range and growth occurred largely because higher values of growth (>1.0 mm/d) were not reached when the ranges exceeded about 11°C. When examined individually, the majority of fish with high ranges were caught in early May (Figure 6). The large temperature range occurred when temperature increased significantly during an early spring warming period in the first 2 weeks of May; average daily temperatures in the river ranged from 10°C to 21°C (median = 16°C) and from 8°C to 17°C (median = 12°C) in the bay. Some of these low-growth and high-range observations occurred in areas of low food abundance and may have been additionally influenced by food limitation.

Food availability is another key variable in winter flounder growth. Studies on plaice Pleuronectes platessa over coarser time and space scales indicated growth differences that may be attributable to food quality and quantity (Berghahn et al. 1995; Van der Veer and Witte 1993). These studies looked at growth differences at the individual and population level between two very different feeding grounds in the Dutch Wadden Sea. We observed similar patterns over finer scales of time and space. When average benthic food abundance was fewer than 50 individuals/m², food levels were positively related to growth; at higher food densities, winter flounder growth plateaued. Many of the low growth-low food observations occurred early in the season when average benthic food abundance was low (Figure 3). Hence, the colder waters and poorer diets experienced by winter flounder juveniles early in the season corresponded to slower growth rates. There was a decline in growth at food levels between 150 and 200 individuals/m²; however all these fish were seined from the same station over a 3-week period in early June. It is possible their growth was depressed by some localized stress we did not measure. In our temperature-controlled laboratory experiments the differences between runs may have been explained by planktonic food quality. An increase in sampling effort was needed to collect the necessary number of copepods for experiments as time progressed. It may have been that plankton quality, as a food item for winter flounder, declined as the season progressed (this is to be examined in future work).

Although winter flounder are euryhaline (Bigelow and Schroeder 1953; Pearcy 1962), our work shows a weak but significant negative relationship of growth rate with salinity. Frame (1973) showed that increased respiration rates for fish at salinities above 20‰ indicate higher osmoregulatory energy expenditures at those levels (see also Meng et al. 2000). This expenditure in energy has the potential to negatively affect growth, and as discussed previously, higher salinity values only occurred in the bay where growth, on average, was slightly lower.

We did not collect many juvenile winter flounder in our fall surveys; juvenile flounder are thought to migrate out of shallow estuaries to overwinter (Pearcy 1962). It is possible that the flounder had already started migration out of the estuary by September; however, age-0 flounder have been shown to migrate little during their first summer, remaining in the estuary until late fall (Saucerman 1990; Saucerman and Deegan 1991). We collected a total of 32 fish in our September and October surveys and did not expect to catch fish as small as 44–56 mm TL (Figure 7). Most of these fish had calculated metamorphic dates of July 12 to August 13, 2000. There is a possibility that they were all subjected to substandard growth conditions and did not deposit daily increments. Sogard (1991) found some negative somatic growth and zero deposition along the sagittae of caged winter flounder larger than 50 mm TL, but only in certain habitats. Our fish were caught throughout the estuary, both in inshore habitats by seine and offshore by beam and otter trawl. Although possible, it is unlikely that all 32 fish had the same poor growth environments. There are other observations suggesting an extended winter flounder settlement season. Winter flounder larvae have been occasionally reported in fall larval fish cruises (Morse et al. 1987), and some very small juvenile winter flounder were reported in a fall seasonal field survey in Little Egg Harbor, New Jersy, and Narragansett, Bay Rhode Island (Able and Fahay 1998; Delong et al. 2001). However, there have been no reports of otolith work done on these fish to estimate DOM. Winter flounder are known to spawn over several weeks to months (Stoner et al. 1999). This study suggests an extended spawning period for some adults that could range through June or July. By spreading the duration of a cohort's early life history through time, the likelihood of survival would be spread over a variety of physical and predator–prey environments, perhaps improving survival (Den Boer 1968; Lambert and Ware 1984). We were hopeful the fall fish would be indicative of fish that survived summer predatory and environmental conditions. However, they did not provide us with a set of dates or environmental conditions that would suggest a successful time of metamorphosis; instead, the fish we caught appeared to have been late settlers. The growth rates of the fish caught early in the season were highly variable throughout the capture period, with no one set of weeks having significantly higher rates than any other, which suggests that it is estuarine placement, not timing, that determines recruitment success. This is consistent with conclusions of Sogard et al. (2001) that recruit supply and growth are controlled by local conditions of nursery grounds that are relatively consistent in the long term.

Field growth rates of winter flounder have been measured by using a variety of methods, such as length frequency analysis for population-level estimates (Pearcy 1962; Howell 1993; Meise et al. 1999; Delong et al. 2001), cage studies (Phelan et al. 2000; Meng et al. 2000; Manderson et al. 2002), and otolith analysis (Sogard and Able 1992; Sogard et al. 2001). Length frequency analysis is a useful tool for comparing population-level growth rates between estuaries and through time; however, variance estimates of growth rates associated with this method are usually generated through bootstrapping methods and miss the information available at the individual level. Estimating growth of individual free-living fish in a well-monitored system is advantageous because it avoids the artifacts associated with cages and may give a better estimate of overall growth conditions at any one time within the estuary. Its disadvantages are that it is more labor intensive than length frequency analysis; environmental conditions experienced by free-living fish are less easily monitored than those confined to cages, and estimating growth rate with our method necessitates an assumption about the size at metamorphosis. Although the relationships we detected between growth and average temperature and salinity confirmed those found in other studies, we were able to define more precisely the effects of varying temperatures and changing food conditions on growth throughout the life of juvenile winter flounder. More importantly, we documented how variable growth rates are within a single estuary and throughout the settlement season, both in the field and under laboratory conditions.

Both DOC and growth rates were positively related to standard length, indicating that larger, faster-growing fish were more prevalent at the end of the settlement season (Sogard and Able 1992; this study). What are some of the potential causes for this observation and given this, why does growth rate variability remain high?

In the field, it is possible that an individual fish's growth responds to environmental changes on a scale not measured by us. Perhaps this is indicated on a course scale by the negative relationship of growth with temperature range. In an ideal situation we would be able to monitor individual fish, their movements, determine their growth rates, and measure their response to environmental cues. On the other hand, some of the variability in cohort growth rates may have a genetic origin. Environmental conditions change daily, seasonally, and interannually in most estuaries, and species that produce offspring with variable growth rates may spread the risk of mortality due to any one environmental condition or suite of size-dependent predators. Fast growth in juvenile fishes is hypothesized to enhance survival through improved foraging capabilities and reduced vulnerability to size-selective predators (Miller et al. 1988; Sogard 1997). However, Pepin (1991) suggested that higher growth caused increased activity levels in pursuit of food, which could lead to more encounters with predators and, thus, higher mortality per unit growth for fast-growing individuals. In contrast, Rice et al. (1993) examined survival of larval fish as a function of the mean and variance of individual growth rates. An individual-based predation model suggested that increased growth variability among individuals resulted in higher cohort survival rates. In the Navesink River–Sandy Hook Bay estuarine system winter flounder are initially exposed to predators like sevenspine bay shrimp Crangon septemspinosa and blue crab Callinectes sapidus that prey on newly metamorphosed individuals under 20 mm in length (Witting and Able 1995). In addition, early spring in a northeastern estuary is a highly variable time of year, when temperature and food conditions may not always be conducive to growth. As the season progresses, other predators such as searobins (Triglidae), summer flounder P. dentatus, and bluefish Pomatomus saltatrix follow temperature fronts, arrive in the Navesink–Sandy Hook Bay estuarine system in May and generally prey on larger juvenile winter flounder exceeding 45–50 mm; (Manderson et al. 1999; Manderson et al. 2000). The presence of these larger predators is more variable in time and space in the estuary. By spawning over several months and having many offspring with variable growth rates, winter flounder may increase the probability that some individuals will survive the seasonal predation pressures and the innately variable environment of a temperate estuary.

Acknowledgments

The authors would like to thank Jeff Pessutti for long, cold hours of sampling for winter flounder; Susan Sogard, Kathy Lang, and Bruce Burns for teaching otolith preparation and reading techniques; and Mike Fogarty, Jeff Buckle, Fred Scharf, Mary Fabrizio, and Chris Chambers for editorial and analytical help. Tom Miller, Dawn Davis, and two anonymous reviewers provided useful comments to our manuscript. Finally we thank Michael Sherrill, Timmy Shaheen, Beth Phelan, John Rosendale, Al Bejda, and Frank Morello.