An Annular Chamber for Aquatic Animal Preference Studies

Introduction

Aquatic animals tend to select or prefer temperatures that provide optimal conditions for growth (Jobling 1981), aerobic metabolism (Bryan et al. 1984), or both. The temperatures they select can be influenced by several factors, including oxygen concentration (Bryan et al. 1984), acclimation temperature (Garside and Tait 1958; Stauffer 1981), nutritional state (Javaid and Anderson 1967a; Magee et al. 1999), environmental light intensity (Kwain and McCauley 1978), age (Kwain and McCauley 1978), season (Roed 1979; Clark and Green 1991), time of day (Reynolds and Casterlin 1979), salinity (Garside et al. 1977), and even the presence of cover (Bevelhimer 1996). The numerous interacting variables and the importance of understanding aquatic animal temperature preferences have catalyzed many thermal preference studies.

Thermal preference experiments are normally conducted with thermal preference chambers in a laboratory, although a few field studies have been attempted (Shrode et al. 1982). Three classes of preference chamber have been used: (1) chambers (usually rectangular) that produce a horizontal gradient (Javaid and Anderson 1967b; Hesthagen 1979); (2) chambers (often cylindrical) that produce a vertical gradient (Garside et al. 1977; Kwain and McCauley 1978; Kellogg and Gift 1983), and (3) electronic shuttleboxes that produce a horizontal gradient controlled by the movement of the organisms (McCauley et al. 1977; Reynolds and Casterlin 1979). An additional class of preference chamber, the Y-maze, has been used to study the responses of aquatic organisms to other types of gradients, such as those created by pH and waterborne pollutants (Laughlin et al. 1978). Chamber-related differences may affect the experimental results (McCauley et al. 1977; Kwain and McCauley 1978), although McCauley and Pond (1971) reported no significant difference in the temperatures selected by rainbow trout Oncorhynchus mykiss in preference chambers with horizontal and vertical temperature gradients.

The presence of confounding variables such as light intensity, available or perceived cover, and water depth can influence the results of preference experiments. All three classes of chambers mentioned above are affected by one or more of these variables. Vertical chambers take advantage of thermal stratification and offer the advantages of relatively deep water, low amounts of available cover, and negligible currents but have the disadvantages of differential pressures and light intensities. Chambers using horizontal gradients offer the advantages of uniform light intensity and no water pressure differential but may be compromised by the presence of uneven cover (corners), currents and the shallow water (<3 cm deep) required to prevent stratification. Shuttleboxes allow the organism to regulate the temperature in a horizontal gradient but may have other problems because they are relatively shallow and have features that provide different amounts of perceived cover.

In addition to the variables mentioned above, the behavior of the study organism should be considered when selecting a preference chamber. Mobile, free-swimming organisms perform well in any of the three classes of chambers, though pelagic organisms may be affected by relatively shallow horizontal or shuttlebox-type chambers. Free-swimming organisms that spend most of their time on or just above the substrate, such as the sand goby Pomatoschistus minutus (Hesthagen 1979) and the white catfish Ameiurus catus (Kellogg and Gift 1983), are more likely to behave in a natural manner in a horizontal gradient that extends along the substrate. Tsukuda and Ogoshi (1985) and Kivivuori (1994) noted that slow-moving organisms like the crayfish Astacus astacus and the planarian Dugesia japonica may not be able to move quickly enough to avoid unfavorable temperatures and hence may become immobilized in temperatures outside of their preferred range. This makes the use of shuttlebox-type chambers problematic.

When we decided to measure the temperature preferences of fishes found in the Sacramento–San Joaquin estuary in northern California, including delta smelt Hypomesus transpacificus and juvenile steelhead, we realized that none of the common preference chamber types would be suitable. Delta smelt are delicate, pelagic species that often injure themselves when placed in tanks with corners (Swanson et al. 1996). Therefore we needed a chamber that would be suitable for such organisms.

Our objectives were to develop and test a preference chamber that would be suitable for use with small, delicate fish. We had the following design objectives: (1) There would be no corners or other form of cover that could injure the fish or affect their temperature selections; (2) the chamber would be uniformly illuminated to prevent fish congregating in areas with lower light intensities, and (3) the water in the chamber would be at least 5 cm deep.

Methods

Annular chamber materials and design

The annular preference chamber consisted of four concentric walls forming four low cylinders centered on a square baseplate (Figure 1). The outer wall was solid, whereas the other walls allowed water to pass through holes located below the water surface. The space between the two outermost walls formed the mixing channel and was divided into eight sections (Figure 2). The space between the second and third outermost walls formed the swimming channel for the organisms (Figure 2). The space between the third and fourth outermost walls formed the effluent channel, which was also subdivided into eight separate compartments (Figure 2). A 5-cm-diameter center drain was located in the geometric center of the acrylic base (Figures 1 and 2).

We constructed the chamber of acrylic sheets, although other materials, such as PVC, could be substituted. The base was an acrylic square 122 cm on a side and 1.9 cm thick. Four 0.32 cm wide × 0.32 cm deep grooves cut into the base held the four concentric walls, which were made of 0.32-cm-thick acrylic. The outer wall was 100 cm in diameter and had a circumference of 314.5 cm. The dividers in the mixing channel were 0.64 cm thick × 11.9 cm high × 9.9 cm wide; those in the effluent channel were 0.64 cm thick × 10.9 cm tall × 9.9 cm wide. Dividers in both mixing and effluent channels were set on 45° radials from the center of the chamber from each other. Water overflowed through V-notch weirs along the inner walls of the swimming and mixing channels if flows exceeded 1.5 L/min per compartment.

Water distribution system materials and design

The chamber was complemented by a four-reservoir water distribution system (Figure 2), in which 20-L HDPE buckets were used as constant-head distribution reservoirs. Two reservoirs provided water at an intermediate (ambient) temperature, one chamber delivered heated water, and one reservoir delivered chilled water. Water left the reservoirs through eight 0.3175-cm-diameter hose-barb outlets located 3.175 cm above the base and through one 3.175-cm-diameter overflow outlet located 17.78 cm above the base of each reservoir. Hose-barb outlets were connected to 0.3175-cm-diameter irrigation hoses with Bowsmith model 70 nozzles at the downstream ends. As shown in Figure 2, four nozzles from each reservoir delivered water to a single central mixing compartment, two nozzles delivered water to the compartment to the left of the central one, and the remaining two nozzles delivered water to the compartment to the right of the central one. Nozzle flows were adjusted to 0.4 L/min by suspending the distribution reservoirs 106 cm above the annular chamber floor. Nozzle delivery rates were kept constant by making sure that water was always overflowing from reservoirs, thus ensuring a constant amount of head pressure for the smaller hose-barb outlets.

Testing procedures

We conducted two tests of the annular preference chamber. The first test was conducted without organisms to determine whether the apparatus created a repeatable temperature gradient. We then conducted a second test using hatchery-reared Feather River strain juvenile steelhead (anadromous rainbow trout) to observe the behavior of fish in the chamber.

In preparation for the first tests, the swimming channel was divided into 48 virtual segments (no physical divisions were made) of equal size. Cold (10.5°C), ambient (19.8°C), and hot (28.6°C) water was then pumped into the distribution reservoirs and the temperature gradient was allowed to form. A mercury thermometer was used to measure surface water temperatures against the inner wall, the center, and the outer wall of the swimming channel in each segment. A similar set of measurements were made at the bottom of each segment. This gave us 6 measurements per segment and a total of 288 measurements per test. We then replicated this process six times to check the repeatability of the gradient.

Measurement of steelhead thermal preferences

Ten juvenile steelhead (mean weight ± SE: 32.7 ± 6.2 g; standard length: 116.9 ± 7.9 mm) were used in the live organism test. Steelhead were supplied by the Feather River State Fish Hatchery in Oroville, California, and had been reared in 16 ± 0.1°C well water on a satiation ration of Silvercup floating steelhead pellets during the summer of 1999. Fish were reared indoors on a natural photoperiod (latitude 38.55°N, June–August) with natural and artificial lighting.

Thermal preferences of individual steelhead were determined by using the annular preference chamber in which a 20°C thermal gradient (10–30°C) was established. The cold-water reservoir received 10°C water, the intermediate temperature reservoirs received 20°C water, and the hot water reservoir received 30°C water. Steelhead were transferred to the annular preference chamber at their rearing (i.e., acclimation) temperature and allowed 1 h to recover from handling before the gradient was established. Fish locations were monitored remotely via a charge-coupled devide (CCD) camera connected to a video monitor. Fish locations were noted every 10 min for 1 h. Water temperatures at those locations were measured with YSI model 401 thermistors connected to a YSI 44TD telethermometer. If fish positioned themselves between two thermistors, then the temperature at that location was measured with a mercury thermometer once the fish had been removed from the chamber at the end of the experiment.

Statistical analyses

Data from both trials were summarized by using Microsoft Excel and analyzed with JMP 5.0 for Macintosh (SAS 2002). Two-way analyses of variance (ANOVAs) were used to determine whether horizontal or vertical position within each segment significantly affected water temperature (i.e., was there vertical or horizontal stratification within segments). The alpha level was set at 0.05. No formal statistical tests were conducted on the juvenile steelhead data.

Results

The annular preference chamber consistently produced an 18°C gradient (Figure 3). Despite some slight variation in temperature among horizontal and vertical locations within each segment, neither horizontal or vertical locations significantly affected the water temperature, indicating that stratification did not develop (ANOVA, P > 0.05). Water currents in the annular chamber flowed radially inwards from the mixing channel to the center drain; the velocity of water flowing around the circumference of the swimming channel was not detectable with a Marsh-McBirney model 201D current meter (0.02 m/s detection limit). Steelhead tested in the chamber selected temperatures between 13.3°C and 26.0°C, with a mean (±SE) selected temperature of 18.1 ± 0.3°C and a modal selected temperature of 18.2°C (Figure 4).

Discussion

Our annular preference chamber met or exceeded our design objectives. Our initial tests successfully demonstrated that we could establish a thermal gradient in the chamber and that the chamber could be used to collect temperature preference data on fish. The annular design allowed us to present experimental organisms with a uniform environment with no gradients of lighting or pressure and no cover that could influence the temperatures they selected. Because the radial flow of water from the exterior towards the interior of the chamber did not provide a single focus point for rheotaxis, we were able to control for another possible confounding variable.

Water temperatures recorded during the initial tests of the annular chamber fluctuated slightly more than we would have liked. These fluctuations can be attributed to temperature fluctuations in the water supplies, which were regulated within 1°C of the target temperature. We recommend that future water supplies be thermostatically regulated within 0.1°C of the desired temperature. We also did not establish as distinct a gradient as we would have liked, probably because we used water at three different temperatures. The almost 20°C difference between our heated and chilled water sources resulted in a sharp temperature gradient, such that an organism could potentially find itself in a mix of temperatures. Future studies can avoid this by using water supplies that are closer to each other in temperature or by increasing the diameter of the chamber to increase the size of the area occupied by each temperature.

An additional drawback of the design as used in the pilot studies is that the amount of chamber space at the middle temperatures is roughly double that available at the extremes. This raises the possibility that a fish randomly selecting positions throughout the chamber (ignoring temperature) would “show” a preference for a midrange temperature, purely because of the greater percentage of those temperatures available. One possible solution to this would be to compare the amount of time fish spent at a particular temperature to the amount of chamber available at that temperature to show that the distribution is not random, perhaps by using a chi-square test. Alternatively, by using only two source temperatures (two distribution reservoirs per temperature), one should be able to create equal amounts of space at each temperature.

Temperatures selected by juvenile anadromous rainbow trout in the annular preference chamber matched most values reported in the literature in studies that used vertical, horizontal, and shuttlebox-type chambers. Kwain and McCauley (1978) reported that 1-, 2-, and 6-month-old rainbow trout acclimated to 15°C subsequently selected mean temperatures of 16.9, 17.8, and 16.9°C, respectively, in a vertical gradient apparatus. Juvenile rainbow trout acclimated to 15°C by Javaid and Anderson (1967b) had mean (±SD) preferred temperatures of 17.5 ± 1.4°C when placed in a horizontal preference chamber. McCauley and Pond (1971) measured temperatures selected by juvenile rainbow trout acclimated to 15–20°C in both horizontal and vertical preference chambers and reported mean preferred temperatures of 18.4 ± 1.2°C and 18.4 ± 2.7°C, respectively. Because of the preliminary nature of our results, we recommend that further testing be conducted to establish a dependable temperature preference for this rainbow trout stock. Additionally, because of the sharp gradients present in the chamber, the fish may have encountered regions where the ΔT was 2–3°C over a distance of 18 cm. Because the size of the fish relative to this potential ΔT is quite similar, the fish could have experienced a gradient along their bodies. To avoid this possibility in future studies, we recommend that the dimensions of the chamber be adjusted so that each distinct temperature zone is at least 1.5–2 times as long as the study organisms. This problem can also be reduced if water sources of only two temperatures are used.

In summary, the annular preference chamber proved to be a useful alternative to the more traditional rectangular chamber with a horizontal temperature gradient. Because the annular chamber should present test organisms with a more uniform environment than a rectangular chamber, we recommend that chambers based on this design be used to test the temperature selection of aquatic organisms that do not require a deep-water column. Our prototype annular chamber had a relatively narrow swimming channel (10 cm) and we elected to set the maximum water depth at 5 cm for the initial tests. One could easily increase the width and depth of the swimming channel to accommodate larger organisms. The annular chamber can also be used for preference experiments using other environmental gradients (e.g., salinity, dissolved oxygen, planktonic [prey] organisms, dissolved or suspended contaminants). Since the completion of the initial tests, our prototype chamber has been used to measure the temperature preferences of more steelhead, delta smelt, and green sturgeon (Acipenser medirostris), as well as the salinity preferences of green sturgeon.

Acknowledgments

The design of this annular preference chamber stems from a senior project by D.K.F. submitted to the University of California, Davis, Department of Biological and Agricultural Engineering in partial fulfillment of the requirements for the Bachelor of Science in Engineering degree. We thank T. Reid for initial inspiration; T. Mussen, S. Chun, D. Cocherell, L. Kanemoto, A. Kawabata, T. MacColl, and C. Marks for chamber testing and animal care assistance; J. Miles and R. Piedrahita for engineering guidance; J. Mehlschau and the staff of the Biological and Agricultural Engineering Shop for chamber construction; P. Lutes and the Center for Aquatic Biology and Aquaculture staff for facility support; G. Dethloff, M Bevelhimer, M. Brandt, J. Holloway, M. Kondratieff, J. Watters, and an anonymous reviewer for manuscript comments; and the California Department of Water Resources and the University of California Agricultural Experiment Station (Grant no. 3455-H to J.J.C.) for research support. All research was conducted in accordance with the guidelines of the University of California, Davis, Animal Care and Use Committee (Protocol no. 7605).