For fishes, swimming performance is an important predictor of habitat use and a critical measure for the design of effective fish passage systems. Few studies have examined burst and prolonged types of swimming performance among several co-occurring species, and swimming performance in many fish communities is undocumented. In this study, we characterize both burst (c-start velocity) and prolonged speed (critical swim speed) across a poorly documented, co-occurring group of stream fishes within the Great Basin of the western USA. We documented the variation in swim speed associated with species, habitat, and body size. Body size had an overwhelming effect on both burst speed and prolonged speed, whereas habitat use, and species identity were not significant predictors. Among species, there is no evidence of a trade-off between burst swim speed and prolonged swim speed. Lack of a trade-off in performance between burst swim speed and prolonged swim speed among species may be due to unexpectedly high prolonged swim speeds exhibited by species that used substrate bracing behaviors. Incorporating body size and variation in behavior, such as substrate bracing behaviors, into fish passage models will likely be sufficient to ensure passage of all species without the need to account for species-specific swimming abilities. However, these results characterize the swimming performance for threatened and common fish species such that other comparisons can be made and species-specific studies can access accurate data.

2. Materials and Methods

We measured swimming performance in seven co-occurring stream fish species in the Great Basin of the western USA, representing four families and six genera. Species tested included: Cottus bairdii Girard (mottled sculpin), Catostomus platyrhynchus Cope (mountain sucker), Rhinichthys cataractae Valenciennes (longnose dace), Rhinichthys osculus Girard (speckled dace), Lepidomeda aliciae Jouy (southern leatherside chub), Richardsonius balteatus Richardson (redside shiner), Oncorhynchus clarkii Richardson (Bonneville cutthroat trout). These species were selected because they represent most of the native co-occurring species of stream fishes in this area. Additionally, L. aliciae and O. clarkii are endemic species that are of conservation concern in the state of Utah.  The benthic stream species group consisted of C. platyrhynchus, R. osculus, R. cataractae, and C. bairdii, the mid-water stream group consisted of O. clarkii, L. aliciae, and R. balteatus. All species were measured across the entire juvenile to adult size range except O. clarkii, which were only available in the juvenile size range.

2.1. Collection & Maintenance

We collected individuals for six of the seven species from wild populations in central Utah via elecrofishing, and we obtained individuals of O. clarkii from captive-reared populations (Table 1). Immediately after capture, we transported all individuals in aerated coolers containing water from the location of origin to laboratory facilities at Brigham Young University. We collected and tested all fish between 31 July 2007 and 23 October 2007 during low flow periods. Because of similar environmental conditions associated with the collection time (i.e. low water velocities, no extremes in temperature), all individuals are assumed to be similarly physically conditioned.

Table 1. Summary of sample sizes and collection locations for species used in the study. N represents number of fish. “Range” and “SE” are the range and standard error of the mean standard length of each fish species.

We completed all tests within one week after capture to ensure that performance of collected individuals would not reflect long-term acclimation to lab conditions. We tested only one species in a given week. During this week of testing fish, we housed them in large round tanks (1100 liter volume) in the Evolutionary Ecology Laboratories facility at Brigham Young University, Utah. We allowed fish to rest in laboratory conditions for at least 24 hours prior to the commencement of swimming trials. We fed fish small pellets of Silver Cup fish feed daily as needed, and we maintained photoperiod at 12:12 ld. We changed water in both the holding and testing facilities each week between swimming trials for each species.

Water temperatures in the holding aquaria were maintained at 17·0°C ± 0·5°C and near saturation with oxygen. This represents the mean water temperature of all sample sites during the collection period (range=14-20° C) and falls within the range of preferred temperatures for all species tested [41].

2.2. Burst Speed

We used a simulated predator attack in a laboratory observation arena to measure burst speed (Figure 1). The arena consisted of a 100 cm x 100 cm octagonal arena with 15 cm high walls. The center of the arena contained a 20 cm diameter clear-plexiglass cylinder that receded into the bottom of the arena, constraining individuals to the center of the observation arena while acclimating previous to the simulated attack. The attack consisted of the rapid projection of a model predator (adult brown trout, Salmo trutta) into the arena toward the test individual. A white cloth covering the observation arena eliminated outside disturbances and premature startling of acclimating fish. We maintained the water at 17·0°C ± 0·5°C, 15 cm depth, and near saturation with oxygen. 

Figure 1. Three-dimensional representation of observational tank used to measure burst speed.  VC, video camera; OA, observation arena; CC, confinement cylinder; MP, mock predator; HD, hinged doors; AR, aluminum runner.

For each burst swimming trial, we introduced a single individual into the clear confinement cylinder in the center of the tank, and allowed it to acclimate for 15 minutes.   After acclimation, the cylinder was lowered to the bottom of the arena and the mock-predator was rapidly propelled into the arena toward the test subject. Test fish were always facing the model before the mock attack was initiated. We recorded burst speed response from directly above the tank at 200 frames·sec^-1 using a high speed digital video camera (Phantom v4.2, Vision Research, Wayne, NJ USA). We measured burst speed with the aid of the Phantom Camera Control software v 8.4 (Vision Research Inc. 1992-2005).  This software electronically calculates the velocity of a moving object using the distance divided by time equation. Time is measured by multiplying the inverse of the framing rate by the number of frames recorded from start to finish of a user-defined video recorded event.  Distance is calculated by indicating a two-point distance from the starting and ending position of the measured object set to a user-defined distance scale. We used a 1 cm square grid on the bottom of the arena as a length reference to create the distance scale.

We measured burst speed using the insertion of the dorsal fin as a reference point.  The insertion of the dorsal fin is near the center of mass for an individual fish, which reduces variation in swimming performance due to undulations of the tail and head.  Burst speed occurs in three distinct stages [24]. Stage one consists of a unilateral contraction of muscles, bending the fish into a C-shape. Stage two consists of a strong propulsive stroke of the tail, projecting the fish forward and ends when the tail stroke reaches maximum exertion on the opposite side of the body. Stage three consists of a gliding or continuous swimming behavior. We measured burst speed (m·s^-1) from the end of stage 1 to the end of stage 2. Burst speed trials were always performed previous to prolonged speed trials.

2.3. Prolonged Speed

We quantified prolonged speed as the critical swim velocity at which a fish can no longer maintain position and becomes impinged on the downstream barrier of a Blazka-type swimming chamber [37]. The Blazka-type chamber consisted of a clear acrylic rectangular observation area (20 cm tall x 20 cm wide x 80 cm long) connected to a downstream reservoir and an upstream section designed to reduce turbulence (Figure 2).  An impeller-powered 5.6 kW motor situated between the reservoir and upstream section cycled water through the observation area. To reduce turbulence, all water passing through the pump was directed through the upstream section, which consisted of a plastic honeycomb with 7 mm wide openings held in place by a wire mesh with 1 mm wide openings. Following the upstream section, water passed through a contraction section which reduced the cross-sectional area and accelerated the flow into the observation section. Fish were restricted to the observation area by a plasic grid with 7 mm diameter round openings on the upstream end, and a metal screen with 7 mm square openings on the downstream end. During all trials, water was maintained at 17.0°C ± 0.5°C and near saturation with oxygen. We measured average water velocity in the swim chamber by averaging the velocity measurements of nine equally spaced quadrants across a cross section of the observation area with a Swoffer model 3000 flow meter.

Figure 2. Representation of the swimming chamber used for prolonged swimming speed tests.  FC, flow conditioner; CS, confinement section; US, upstream screen; OS, observation section; DS, downstream screen; R, Reservoir; P, pump. Arrows indicate current directions.

We calculated critical swimming velocity, or the velocity at which fish become fatigued, (prolonged speed, measured in m·s^-1) using the following formula [37]:

where, Ti was the time a fish was held at a specific current velocity (5 min.), Vp was the highest current velocity maintained for a full 5-min period (m·s^-1), Vi was the current velocity increment (0.1 m·s^-1), and Tf was the elapsed time at the fatigue velocity. We initiated trials by placing a single individual in the observation section for 15 min without flow.  After this acclimation period, water velocities were increased by 0.1 m·s^-1 every 5 min until the fish was impinged on the downstream barrier. Upon impingement, we gave several successive taps on the fish’s caudal peduncle to stimulate continued swimming. When an individual would no longer respond to stimulation following impingement, the swimming trial was terminated, and the time to fatigue and the velocity at fatigue were recorded [3,5]. After all trials were completed, we euthanized fish via overdose of MS-222 in accordance with IACUC protocols and direction from the state of Utah specified in the collecting permit, then we preserved all specimens in 70% ethanol. 

2.4. Statistical Analysis

We used simple linear regression of swim speed on total length (m) to characterize the change in swim speed with increasing body size for each species (Hypothesis 1). We used analysis of covariance (ANCOVA; Proc GLM, SAS Enterprise Guide version 7.15 HF8, SAS Institute Inc. Cary, NC, USA) to test for differences in burst speed and prolonged speed between benthic and mid-water species (Hypothesis 2). Prolonged speed (m·s^-1) and burst speed (m·s^-1) were response variables. We used species as a categorical predictor variable and total length (in m) of the fish as a covariate. We also included the interaction between species and total length to test for heterogeneous slopes among species (Proc GLM, SAS Enterprise Guide version 7.15 HF8, SAS Institute Inc. Cary, NC, USA). By including total length as a covariate, the model tested for differences in mean swim speed among species at the overall mean total length among all species (mean total length = 0.071 m). The total length range of specimens tested overlapped this mean value for all species. Sample sizes varied among species and are recorded in Table 1.

Missing values are indicated by "."

Key to species