Header background


resources, reviews and more

Procedures for Lab Experiments in Tanks

Alan Temple wrote a blog, Setting Doses for Lab Experiments, which I followed with Setup for Lab Experiments in Tanks. That was followed by a short one, Size Matters, on the effect of fish size on the threshold voltage gradient and power density for immobilization or other responses. This blog discusses some aspects of how a tank study is conducted. Specifically mentioned are the fish themselves, the desired response to be assessed, how that response is to be evaluated, and two primary approaches for quantifying the results.

Selection of the fish species to be studied may be simple if research is focused on one or two species, or if a common species can be used as a surrogate for some valuable species. Perhaps a couple of target species, such as largemouth bass and bluegill sunfish, can be used to represent an assemblage of species to be surveyed in the Southeast US, for example. Once the species is/are selected, the size of fish for testing should be carefully chosen based on the overall study objectives. Important to consider is the availability of several fish of the same species and size for each trial. Depending on the method chosen, a treatment may require as few as 3-5 fish or up to about 20 fish or more. If you desire to compare about five or six treatments or more in a given trial, then one such study may require over 100 fish of the same species and of similar size. Such a sample size of healthy fish may require extensive planning for collecting, holding and caring for them until the study is to be conducted. An alternative may be to use hatchery fish; they are numerous, of uniform size, in good health and not stressed by capture from the wild and being held in captivity. Mesa and Schreck (1989) found that hatchery cutthroat trout were less stressed by electroshock than were wild cutthroats. This may mean that hatchery fish have higher electrical thresholds compared to wild fish, but this has not been tested. Just be aware that using hatchery fish may provide somewhat different results than using wild fish.

The fish response to be evaluated should be decided before the test. This may be immobilization by narcosis (slack muscles), loss of equilibrium (the fish rolls over, often on the bottom of the tank), or even just a twitch. Tetany (rigid muscles) is to be avoided as that may lead to fish injury or harm. Loss of equilibrium is easy to assess, if it occurs. Immobilization may be more difficult to determine. I have witnessed transient immobilization with a particular waveform; the fish seemed momentarily stunned then began swimming again, all within about four seconds. Another response to electrical stimulation is forced swimming (this may be with the fish upright or even a pseudo-swimming while upside down). The latter often requires a higher voltage setting than the upright forced swimming. Forced swimming may not even occur with the waveform being used. An important capture-prone response which may be evaluated is taxis, or attraction primarily to the anode.

The number of fish to be placed in the test tank at once must be decided. Most studies to determine threshold for immobilization are done with one fish in the test tank at a time. One fish is placed in the tank and allowed some time (a few seconds to minutes) to adjust to the tank and become calm prior to the electrodes being energized.  Orientation of the fish in the test tank is an important consideration. Generally, we let the fish face one electrode or the other before turning on the electricity. However, if one wishes to observe taxis to the anode, fish should face the cathode so their turn can be observed as a confirmation of taxis. A researcher may require that the fish all face the anode, for example, but generally this is not required. However, it may be wise to document to which electrode the fish was facing before the current is turned on. In some cases, fish are restrained in nets or baskets and in a particular orientation during the test. In one clever study (Bearlin et al. 2008), a light was placed over the tank and a structure was placed directly under the light such that it provided shade in the center of the test tank. The fish naturally sought that shaded area before the electrodes were energized. A uniform electrical field was mentioned in Setup for Lab Experiments in Tanks. Even though the voltage gradient should be the same anywhere inside the tank, the fish should be kept from touching the plate electrodes during the electrical exposure. Let me say one more thing about the number of fish in the test tank at once. Sometimes, it may be advantageous to place multiple fish in the test tank at one time. This is especially true if the desired response is attraction to the anode. With only one fish in the tank at a time, that fish may have been forced to the anode, or it may have just happened to swim that direction. If four fish were placed in the tank, and three or four swam directly to the anode, it is a clearer response than if just a lone fish swam to the anode.

How long should the current be left on while observing the fish response? This is a judgment call and may depend upon the question being asked or the field sampling being simulated in the tank. Studies have used current on times of three seconds, four seconds or more. The current on time should be established before the study, and it may depend upon the method being used. Some preliminary testing is important to learn what behaviors can be easily and consistently observed and at what intensities and durations they occur. Studies can be conducted with only one person, but it is preferable to have two or three fish observers during each treatment. The observers should decide immediately whether the desired fish response was achieved within the allotted treatment time. For example, a fish was subjected to a given voltage gradient for four seconds. Was it immobilized or not? The observers should be able to agree, and that consensus should be recorded. If possible, it is advisable to video the fish response in each treatment so that the original assessment can be confirmed or changed, as needed.

There are two main approaches to such studies. Kolz and Reynolds (1989) placed a fish in a tank and began with low voltage which was increased until the response was obtained. This was repeated for a few fish, and the average voltage gradient and power density were calculated for that treatment. Each fish was an independent assessment of threshold voltage gradient and power density. This approach may be termed the average or ramp-up method. If possible, the average of five fish works well with this method, but fewer fish may be used. Though it is best to use the average of at least three fish per treatment, I have used the average of two fish per treatment with good results, and there may be only one fish of a given size for a particular treatment. In contrast, Miranda and Dolan (2003) placed a fish in the test tank and administered a given voltage for three seconds. The fish either achieved the desired response or not. If the fish did so respond, the result was recorded as a 1. If the response was not achieved, the result was recorded as a 0. Those data were analyzed using logistic regression (called the logistic or binary method). From the regression equation, one may calculate the 50th percentile probability of response, or any other percentile. Thus, dose-response curves may be developed using the logistic regression method, and that is a big advantage. A disadvantage is that voltage gradients must be carefully selected and applied below and above the response threshold so that a suitable regression equation can be calculated. This often requires at least 15 fish per treatment, and perhaps 18-20 or more fish per treatment is preferred. None of the responses may be an estimate of the actual response threshold. If many fish of a given species and size are available, the logistic regression method may be the method of choice, especially if dose-response curves are desired. Where limited numbers of fish of the same size are available, the logistic regression method may be impossible, so the ramp-up or average method must be used. Actually, I have used both approaches with alligator gar, and the results appeared comparable for each method. There is a graph below of those data. It may be that the overlap in voltage gradient values which produce 0’s and 1’s is such that the software cannot produce a satisfactory solution; thus, no regression equation is generated.

Among four recent blogs, we have discussed the setup, doses, effect of fish size, and procedures for lab experiments in tanks. Few people may have the equipment and interest in conducting such studies. However, it is our hope that these have been useful in describing lab experiments so that you can appreciate how they are conducted, and it may help you in interpreting such studies described in reports and in journal articles. Lab experiments are not expected to translate perfectly to field sampling, but lab studies under controlled conditions in tanks may help you quickly find electrical waveforms and power levels that appear promising for field sampling and to eliminate those which are not. The most effective and efficient studies may use a combination of lab experiments and field sampling. We have used such approaches with Asian carp, northern snakeheads and other species.


Bearlin, A.R., S.J. Nicol and T. Glenane. 2008. Behavioral responses of Murray cod Maccullochella peelii peelii to pulse frequency and pulse width from electric fishing machines. Transactions of the American Fisheries Society 137:107-113.

Kolz, A.L., and J. B. Reynolds. 1989. Determination of power threshold response curves. U.S. Fish and Wildlife Service Technical Report 22:15-23.

Mesa, M.G., and C.B. Schreck. 1989. Electrofishing mark-recapture and depletion methodologies evoke behavioral and physiological changes in cutthroat trout. Transactions of the American Fisheries Society 118:644-658.

Miranda, L.E., and C.R. Dolan. 2003. Test of a power transfer model for standardizing electrofishing. Transactions of the American Fisheries Society 132:1179-1185.

Logistic regression method with a shallow response slope due to extensive overlap in the response to the voltage gradients.

Logistic regression method with a shallow response slope due to extensive overlap in response to the voltage gradients.


Logistic regression method with a steep response slope

Logistic regression method with a steep (or knife edge) response slope due to limited overlap in response to the voltage gradients.


Open circles (individual fish) are for the ramp-up method. Red circles are for the logistic or binary method (50th percentile of 15 fish).

Open circles (individual fish) are for the ramp-up method. Red circles are for the logistic or binary method (50th percentile of 15 fish). The results for the two methods appeared comparable. The logistic method required more fish and time to produce results.



Footer background

Drop us a line

Yay! Message sent. Error! Please validate your fields.
© 2015 Thread One Page. Imagery: Tom Rayner, Alan temple, Richard Pearson, Paul Godfrey, Roger Scott, John Rayner.