Modification of Schooling Behavior in Larval Atherinid Fish Atherina mochon by Heat Exposure of Eggs and Larvae

Introduction

Cooling water for electrical power stations is often obtained from and discharged into the coastal marine environment (Langford 1983). During its transit through the cooling circuit, the water is heated about 10°C depending on the station. From the point of discharge, the thermal effluent mixes with surrounding water and progressively cools by dilution and heat loss to the atmosphere. The rate of cooling and the volume of coastal water affected by the thermal plume depend on the specific design and location of the discharge mechanism and the hydraulic characteristics of the receiving water.

When the discharge mixes with ambient water, it entrains planktonic organisms and exposes them to transient increases in temperature (Coutant 1972). As the plume contacts the bottom or shoreline during the dynamic mixing process, it similarly exposes sessile organisms to transient temperature increases. Such exposure may be important to fish populations when the thermal plume occurs in or near spawning areas. It is usually not directly lethal, except perhaps very close to the discharge point.

Biological criteria for regulating thermal discharges to the environment have been derived primarily from physiological research. Whereas research data and assessment techniques are well developed for evaluating the potentially lethal stresses of transient high-temperature exposures (Coutant 1970, 1972; NAS-NAE 1973; Brungs and Jones 1977), understanding of the potential long-term effects of sublethal transient temperature elevations on the early life stages of fish remains rudimentary. Such effects are generally not included in regulatory biocriteria. This is particularly true for the potentially subtle effects on behavioral and social functions such as schooling.

In the present study, we investigated the effects of heated effluents on the development of fish schooling behavior. Schooling is a vital adaptive behavior, ensuring species survival by facilitating life functions such as obtaining protection from predators, reproduction, sociability, finding food, and energy conservation (Pitcher and Parrish 1993). Indeed, the disruption of schooling may jeopardize the animal's success in its environment. Schooling develops gradually, with larval fish initially demonstrating patterns of fish-to-fish approach and orientation (Shaw 1960, 1961). Newly hatched fish approach each other head-on, with an almost immediate veering away or withdrawal from each other. As the larval fish grow older and increase in size, they exhibit head-to-tail approaches that persist in duration and develop into a sustained parallel orientation with the subsequent formation of schools (Williams and Shaw 1971; Williams 1976).

The experiments were conducted at the Department of Ecology, Stazione Zoologica Anton Dohrn, on the island of Ischia in the Bay of Naples, Italy. The Mergellina area of the bay receives heated discharges (among other pollutants) from two sources on the Naples coast, an electricity generating station and a steel plant. The study was designed to observe the effects of simulated heated effluents on the development of schooling behavior in local fish. The subjects were the atherinid little sand smelt Atherina mochon. We focused on the egg and larval stages because the thermal requirements of developing fish are often more exacting than those of adults (Brett 1956). The experimental design tested the hypothesis that a specified heat treatment of eggs, larvae, or both simulating exposure to a thermal discharge plume would cause measurable and statistically significant differences in the development of schooling behavior by early larvae.

Methods

Fish

Spawning adults of A. mochon were collected at Mergellina Harbor in the Bay of Naples. The males and females are externally indistinguishable except that ovulating females have greatly distended abdomens and many are larger in size than the males. Fertilization was accomplished in the field by “stripping,” that is, gently squeezing the abdominal walls of the adults to eject their gametes into vacuum flasks containing thoroughly aerated and filtered seawater. The sperm of one male was used to fertilize batches of eggs from seven females. Successful fertilization was indicated by the subsequent clumping of the eggs, which results from the extrusion of adhesive threads from the chorion, and a change in color from pale yellow to amber. We obtained 488 experimental larval fish for use in the experiment by this method.

At approximately the blastula stage of development, the embryos were individually separated by cutting the adhesive threads to ensure adequate aeration and the reduction of bacterial and mold growth. Batches of 15–20 separated embryos were placed in elongated plankton nets measuring 23 cm × 10 cm that were suspended in 40-L tanks. There were eight such tanks, each holding five to six nets. About 15 air stones in each tank provided continuous aeration and circulation around the nets. Two tanks were used for each of the four treatments, which comprised no thermal exposure (controls) and the thermal exposure of eggs, larvae, and both eggs and larvae.

A. mochon eggs normally hatch within 13–14 d; the eggs that received the heat treatments hatched in 11–12 d, however. The eggs were incubated in the plankton nets; after hatching, the larvae were maintained in free-swimming groups in twelve 40-L tanks for 1–35 d. The eggs and larvae were removed from these tanks for the heat treatments and returned as a group to their respective holding containers afterward.

The larval fish were fed to excess on newly hatched nauplii of brine shrimp Artemia spp. The seawater in the tanks was changed three times a week; it was always aerated and filtered through an Eaton water filter. The water in both the holding and experimental tanks was stabilized at a temperature of 20 ± 1°C (mean ± SD) and a mean specific gravity of 1.0278. The fish were reared in a laboratory that had no natural light; rather, low-intensity fluorescent room lighting was provided constantly.

Experimental design

The experiment consisted of comparing four groups (Table 1): (1) Group I (controls), in which eggs and larvae were reared at 20°C with no heat exposure; (2) Group II, in which eggs were exposed daily to a transient heat treatment beginning on the first day postfertilization and continuing until hatching (about 10 d); (3) Group III, in which larvae were exposed daily to transient heat from the first day posthatch to day 10; and (4) Group IV, in which eggs were exposed daily to transient heat from day 1 postfertilization and subsequently the larval fish were treated from day 1 posthatch until day 10. There were enough fish in all groups that all eggs and larvae could be used only once in behavioral testing, thus maintaining their naivete and reducing the variability in their behavior.

Table 1. The numbers of behavioral tests (four fish per test) and fish used to determine the effects of sublethal exposures of eggs, larvae, or both to transient daily increases in water temperature on the development of schooling behavior

Heat treatment

Eggs and larvae were given transient heat exposures below their lethal limit. The lethal limit of heat tolerance was determined by conducting several 96-h heating tests in which larvae aged 10–35 d were initially acclimated to 20°C, their normal ambient temperature during spawning and development. The temperature was then increased 1°C every 3 min until the fish became incapacitated. The limit of tolerance (i.e., the highest temperature at which no mortalities occurred) was found to be 28.5°C. This temperature was selected as the maximum for the transient heat treatments in the experiment.

The test groups were given experimental heat treatments at the same time of day for 10 consecutive days (20 d for the treatment of both eggs and larvae) in order to simulate exposure to heated discharges on a daily basis. For the heat treatment, the eggs and larvae were gently removed from their holding tanks by scooping them into a small cup and placing them in a 20-L water bath that was thermostatically controlled and programmed to provide repeatable temperature changes. The gradual and uniform increases in temperature permitted the subjects to adjust to the temperature changes. The total duration of a heat treatment was 1 h and 20 min. The water was warmed 8.5°C at a predetermined rate of increase over a period of 30 min; that is, starting at 20°C, the temperature was increased to 24.2°C after 5 min, to 27°C after 15 min, to 28°C after 22 min, to 28.2°C after 26 min, and to 28.5°C after 30 min. The fish remained in the tank at 28.5°C for 15 min, after which the temperature was gradually reduced to 20°C in a cooling bath over a period of 35 min. At that time, the fish were returned to their respective home tanks. Controls were also handled, but their water temperature was not changed.

Behavioral observations

The methods of testing and recording behavior were basically similar to those used in previous studies (Williams 1989, 1997). Four fish larvae that had not previously been observed for behavioral responses constituted a trial. Fish were observed in a circular tank that was 48 cm in diameter with a water depth of 6 cm. The tank was illuminated directly by a cold, circular, 10-W fluorescent light 30 cm in diameter that was suspended 38 cm above the tank. Prior to each test, the filtered and aerated seawater was changed.

Behavior was observed every 5 d beginning on day 10 posthatch (after which all heat treatments ceased) until day 35. Four test fish were individually and gently scooped with a small cup from their home tank and poured into the test tank at specific times. The first fish remained alone for 3 min; the second was introduced at the end of that period (the beginning of minute 4), the third 3 min later (the beginning of minute 7), and the fourth 3 min after that (the beginning of minute 10). The fish were introduced at the same location on the periphery of the tank. Observations were made of the behavior of the newly introduced fish during each of the three 3-min periods after the second, third, and fourth fish were placed in the tank. The behavior of all four fish was recorded for 12 min (minutes 13–24) and for 1 min each at the 30th and 36th minutes. A total of 23 min of recorded data were collected for each of the trials.

The Rustrak Event Recorder, a miniature recorder and keyboard designed on the pattern of the Esterline Angus multievent recorder (Tobach et al. 1962), was operated to simultaneously record the sequence, frequency, and duration of specific behaviors of the fish. Quantitative recordings were made of (1) the frequency of approach–withdrawal, in which one fry approaches to within 3 cm of another and one or both immediately veer away from each other; (2) the duration of parallel orientation, in which two or more fish mutually approach, orient into position parallel to each other, and school, the fish swimming in the same direction approximately 1–3 cm apart and synchronizing their swimming movements and speed; (3) the latency of first schooling, that is, the period of time that elapses before the formation of the first school (from the initiation of the test); and (4) the duration of the longest-persisting school (Williams and Shaw 1971; Williams 1976, 1989, 1997).

Detailed descriptions of various behaviors, such as the organization of schools and locomotor activity, were recorded in notes made during the testing. Observations of a less structured nature were also made prior to and after testing, when the fish were in their rearing tanks. Statistically significant differences among quantitative data from treatment groups and controls were determined by one-way analysis of variance (ANOVA) with Jandel SigmaStat statistical software (Jandel Scientific 1995). All comparisons of means were on a pairwise basis.

Results

Behavioral testing of larval fish 10, 15, 20, 25, 30, and 35 d posthatch demonstrated significant (P < 0.05) modifications in the development of schooling behavior in the three heat-treated groups compared with the controls. The schooling repertoire typical of A. mochon was displayed by the control groups and consisted of approach–withdrawals that diminished in number at later ages, followed by parallel orientation. During schooling, the fish swam in the same direction, with close interfish spacing of approximately 1 cm accompanied by synchronized swimming movements and speed. Intrinsic to these patterns of behavior were the reciprocal and positive biosocial attraction and interaction of the fish. In contrast to the controls, the heat-exposed fish exhibited significantly higher frequencies of approach–withdrawals that did not lead to schooling, decreases in the duration of parallel orientation, higher latencies for the formation of the first schools, and a shorter duration for the longest-persisting schools (Figure 1). Locomotor ability was diminished in the heat-treated fish and schooling was incohesive, with wide interfish spacing and the absence of synchrony in position, swimming movements, and speed. The fish sometimes appeared disoriented, with curved tail regions, body convolutions, and a tendency to rise in the water. Social interaction was dramatically reduced. In contrast to the clear differences between the controls and the heat-treated groups, there were few statistically significant differences among the fish in the three heat treatments.

The following narrative expands upon the data in the Figure 1. The behavioral observations are ordered by the age of the larvae when they were made.

Discussion

In laboratory studies, we showed that transient daily exposures of the eggs and/or larvae of A. mochon to above-normal but sublethal temperatures can retard the development of normal schooling behavior. The thermal exposures were designed to roughly simulate a slowly mixing plume of warmed water from a thermal discharge that cyclically wafts over a spawning and nursery area as tides change. The transient temperature changes represent what might occur as (1) eggs attached to a substrate are progressively enveloped by the thermal plume and then cooled again as currents shift and (2) larvae in the water column mix repeatedly with the warmed and progressively cooled plume of water.

To our knowledge, this is the first time such transient thermal exposures have been tested for their effects on the development of schooling behavior, although rearing environment (Williams 1976) and pollutants such as the insecticide Sevin (Weis and Weis 1974), methyl parathion (Williams 1989), and methyl mercury (Weis and Weis 1989, 1995; Ososkov and Weis 1996) have been shown to affect the development of schooling. The observed thermal sensitivity is somewhat surprising, however. Pelagic eggs and larvae often enter intertidal estuarine regions where the shallow waters are exposed to extensive diel temperature fluctuations due mostly to solar warming (although an 8°C shift in a short time period may be extreme). One might expect these young fish stages to be more tolerant of such temperature changes.

The results of heat treatment on the development of schooling behavior raise two sorts of biological questions: (1) their importance for the recruitment success of the exposed fish and the continued success of local populations if recruitment is decreased and (2) the cause(s) of the retardation in development and the observed abnormal behaviors. They also raise resource management questions about the general wisdom of using coastal waters near spawning areas for cooling and about ways of designing and locating thermal discharges to minimize their sublethal biological effects.

Alteration of the development of schooling may be significant for recruitment success and ultimately for local populations of the affected species, although this study does not address the longer-term significance. Schooling is a common behavior in fish. The advantages of aggregating in schools have been studied extensively (Pitcher and Parrish 1993). Schooling is a main defense against predators of pelagic fish, and it depends on well-coordinated swimming behavior by the members of the school to be successful. Interactions among fish develop gradually throughout larval and juvenile development (Shaw 1960, 1961; van Olst and Hunter 1970; Gallego and Heath 1994a). A critical period of vulnerability for late larval and early juvenile Atlantic herring Clupea harengus to visual predators (e.g., European whiting Merlangius merlangus) was shown to occur before and during the initial stages of the development of schooling behavior (Gallego and Heath 1994b), as the authors had predicted (Gallego and Heath 1994a). Thus, there may be a general survival advantage to developing schools quickly. When schools are not formed quickly, as with the heat-treated fish in this study, predation pressure at this stage may have important consequences for the recruitment to the adult population. Field verification of our results and their potential population consequences remain to be addressed.

The biological causes of the observed behavioral changes remain worthy of further study. However, our observations suggest that vision development is an important factor. The acuity of visual systems improves during ontogeny (see review by Fernald 1991). Visual acuity has been linked to the ability to detect predators and locate prey (Miller et al. 1993). Vision is unquestionably involved in schooling (Shaw 1960, 1961; Williams 1976). In our experiments, fish clearly exhibited behavior patterns indicating the use of vision to recognize potential schooling partners. Recognition of other fish (through initial head-to-head approach behavior followed by parallel pairing) seemed delayed by heat treatment, as did the formation of schools of more than two fish. But other sensory information, such as that derived from the lateral line, may be involved (Pitcher 1979). The interplay of the sensory mechanisms involved in forming and maintaining a schooling pattern is not well understood despite the results being observable and quantifiable, as in this study.

It is unlikely that electricity generating facilities will cease using coastal waters for cooling based on these results. However, the work does have practical application to the location and design of new or modified cooling-water discharges. A primary tenet of thermal discharge regulation, that heat should not be discharged into particularly important spawning and nursery areas, is substantiated (USEPA 1977). The work further supports the view that biological effects are minimized when discharges are introduced where dilution water is abundant, hydraulic mixing is rapid, and the temperature declines quickly to near the ambient level (e.g., when 90% of the temperature differential dissipates within a few seconds; Coutant 1972). Slowly mixing thermal plumes in which temperatures up to 8.5°C above the ambient level persist for fairly long periods can and should be avoided.

Acknowledgments

This publication is dedicated to the late Lucia Mazzella and the late Romilly Williams, both of whom were instrumental in the achievement of this research. We thank Professors Giorgio Bernardi, President, and Lucio Cariello, Director General, of the Stazione Zoologica Anton Dohrn and Maria Cristina Buia, Director of the Benthic Laboratory in Ischia, for providing Williams the opportunity to conduct the experiments and draft the manuscript. We are especially grateful for the assistance given by Valerio Zupo on statistical analyses. This paper was completed while Williams was on guest assignment at Oak Ridge National Laboratory courtesy of James Loar.