An Automated Low-Cost Swim Tunnel for Measuring Swimming Performance in Fish

In the last 60 years, swim tunnels have been extensively used to study metabolism, oxygen consumption, and swimming activity of fish. ¹ Swim tunnels mostly consist of a cylinder or rectangular chamber where the fish is obligated to swim against a water flow. ² In a typical swimming performance assay, water velocity is progressively increased up to the point in which the fish is unable to swim. The time swam by the fish before exhaustion is related to the velocity of the water flow to calculate the critical swimming speed (U_crit). ³

In this study, we describe a user-friendly, automated, modular, and low-cost swim tunnel that permits to study swimming performance of one or more fish separately, as well as in a small group of individuals. By means of an interdisciplinary and intersectoral collaboration between biologists, engineers, and computer scientists, we designed a swim tunnel that (1) minimizes rambling flow paths; (2) is automated and can be programmed to control water velocity with a predetermined time schedule; and (3) integrates the setup with a fish tracking software and an automated pump controller to stop the flow when the fish reaches maximum swimming speed (Supplementary Figs. S1–S4). Moreover, the chamber of our swim tunnel can be rapidly and easily modified in width and length to fit species with different body size or to create a wide arena to investigate shoaling behavior. The control architecture of the smart swim tunnel is based on a computer program that acts both as a human–machine interface and a controller. It collects data from a camera, a pH-meter and a water temperature sensor, and it additionally provides a control signal for the pump, by means of an Arduino board, which is able to generate a 0–10 V signal for the pump controller (Supplementary Fig. S4).

To validate our swim tunnel, we assessed swimming performance of four different species: two strains of zebrafish (AB and Ariosto), medaka, guppy, and Somalian cavefish (Fig. 1A, B and Supplementary Movie S1). We applied a protocol of incremental changes in water velocity, from 20 to 80 cm s⁻¹ in 10-min steps. Thanks to the automatic tracking software, we recorded the fatigue as the point at which a fish stopped swimming and contacted the rear section of the tunnel for >5 s (Fig. 1C). U_crit (body length per second, BL s⁻¹) of the four species (N = 10 per group) is shown in Figure 1D. Mean U_crit (±SEM) of the two zebrafish strains was comparable (AB: 44.85 ± 6.23 cm s⁻¹; Ariosto: 56.26 ± 8.92; Mann–Whitney U-test, p = 0.47). Interestingly, in the Ariosto strain two subgroups with different U_crit were discriminated (Mann–Whitney U-test, p = 0.01). This last result could be due to heightened behavioral individual differences in the wild strain, in line with data in other species. ⁴ The mean U_crit of medaka (18.61 ± 1.06 cm s⁻¹), guppy (9.25 ± 0.49 cm s⁻¹), and cavefish (9.28 ± 0.5 cm s⁻¹) was significantly lower in comparison to zebrafish (one-way ANOVA, F_{4, 49} = 22.69, p < 0.0001; Tukey's multiple comparison test, p < 0.01). Our estimation of swimming performance of guppies and zebrafish was broadly overlapped with that of prior studies with a similar setting,^5–7 suggesting that the setup described in the present article allows reliable measurements of swimming activity and performance of fish. With other type of swim tunnels, some differences were observed. However, it is unclear whether these variations were due to differences in the chamber^8,9 or in some characteristics of the fish that can affect swimming such as genetic background, maintenance conditions, morphology, and physiological traits.^6,7,10–12

OPEN IN VIEWER

FIG. 1.

(A) The fish species. (B) The swim tunnel. (C) The video tracking software. (D) The critical swimming speed (U_crit; BL · s⁻¹) of the selected species or strain, where the mean ± SD (N = 10 per group, black lines) is displayed in the center of the dot plot of U_crit. See also Supplementary Data.

The measurement of swimming behavior, energetics, and kinematics is an important part of fish biology research^13,14 and swimming tunnels are now consistently used also in applied studies on pharmacology, ¹⁵ ecology, ¹⁶ toxicology, ¹⁷ and biomedical research. ¹⁰ Our system could help researchers in these fields with a cost-efficient solution for characterization of zebrafish and medaka mutants and testing biologically active compounds. Furthermore, differently from other Brett-type swimming tunnels,^1,3,14,18 our modular system gives the opportunity to simultaneously test multiple specimens (i.e., mutants vs. wild type) and to investigate shoal's swimming behavior. With the possibility to enrich the system with additional features, such particle image velocimetry for visualizing the flow physics around swimming fish ¹⁴ or respirometer to measure oxygen consumption, ¹⁹ our study could also provide the base for more complex research on fish swimming metabolism and mechanic.