Work carried out since the late 1970s has provided key insights into the comparative biomechanics, kinematics, behaviour and neurobiology of fish escape responses. An escape response is an ecologically important behaviour used by fishes to evade predation and aggression via rapid swimming movements. With environmental change expected to affect the physiology and biomechanics of aquatic ectotherms, there is a growing interest in understanding how environmental stressors affect the swimming performance and behaviour of fishes during escape responses, particularly in the context of predator–prey interactions. As the study of fish swimming continues to expand, there have been repeated calls to standardise experiments and reporting practices to facilitate integrative and comparative studies. Here, we provide a set of practical guidelines for conducting, analysing and reporting experiments on escape responses in fish, including a reporting checklist to assist authors undertaking these experiments. These resources will facilitate executing and reporting escape response experiments in a rigorous and transparent fashion, helping to advance the study of fish swimming in an era of rapid environmental change.

Escape responses are high-energy swimming bursts used by fishes to escape predation and aggression (Box 1). Comparative physiologists and biomechanists interested in fish locomotion have a long history of studying escape responses, starting with pioneering work based on hand tracings of body movements (Weihs, 1973; Webb, 1975, 1976). Steady improvement in methodology brought about by digital video recording, affordable high-speed cameras and data extraction software has since increased general interest in the study of escape responses, allowing important insights in fields including biomechanics, functional morphology, predator–prey ecology, muscle physiology and neurophysiology (Eaton et al., 2001; Domenici et al., 2011). In parallel to these developments, there has been a growing interest in understanding how anthropogenic stressors affect whole-organism performance in fishes, particularly in the context of predator–prey interactions and climate change (Domenici et al., 2019).

Box 1. A brief overview of escape responses in fish

Escape responses are a type of startle response induced by a sudden, threatening stimulus. Fishes exhibit different types of startle responses (Domenici and Hale, 2019), including withdrawal, freezing and escaping. Escape responses are caused by the contraction of fast-glycolytic muscle fibers, typically resulting in a rapid bend of the body (often into a C-shape), followed by burst swimming. Because escape responses involve short but rapid bouts of acceleration, they are often referred to as ‘fast starts’; however, not all fast starts are escape responses (Domenici and Hale, 2019).

Escape responses by fish were originally considered as a highly stereotypic behaviour triggered by the activation of large reticulospinal cells – the Mauthner cells (or M-cells) – and other related neurons in the hindbrain (Korn and Faber, 2005). In fact, escape responses exhibit a wide range of kinematics (Domenici and Hale, 2019), and not all are controlled by M-cells (Domenici and Hale, 2019; Hecker et al., 2020). Escape responses generally comprise three stages (Weihs (1973): (1) the preparatory stroke; (2) the propulsive stroke; and (3) a variable stage involving continuous swimming, coasting and/or deceleration. Research has focused on the first two stages (Walker et al., 2005). However, stage 2 is not always present in an escape response (reviewed in Domenici and Hale, 2019). Being able to identify these stages is important for measuring both behavioural and kinematic components of escape responses (Table 1). Stage 1 begins at the onset of the escape response and ends when the rotation of the head changes the direction of the turn (double-bend responses) or the body stops bending (single-bend responses). Stage 2 begins at the end of stage 1 and ends when the rotation of the head stops or changes the direction of the turn.

The sequence of events in a typical Mauthner-mediated escape response is as follows. A fish perceives the threatening stimulus, and the sensory neurons excite the Mauthner cell ipsilateral to the stimulus; this cell inhibits the contralateral Mauthner cell while also exciting the contralateral axial musculature. The resulting muscle contraction causes the head and tail to bend in stage 1. In this stage, there is little motion of the fish's centre of mass. Stage 2 is characterized by a return flip of the tail, leading to forward acceleration of the body, although some thrust can be produced during stage 1 (Tytell and Lauder, 2008). The onset of stage 2 is not directly activated by the Mauthner cells (Domenici and Hale, 2019), and stage 2 is only present in double-bend responses (see Domenici and Hale, 2019). Given the variability of stage 3, it is generally not considered in measurements of escape performance (see ‘Data extraction’). Recent studies have shown substantial variability in the neural control, timing and kinematics of escape responses (Domenici, 2010b; Domenici and Hale, 2019). For a more detailed description of the diversity of fish escape responses, see Domenici (2010a,b, 2011) and Domenici and Hale (2019).

Fast starts (i.e. escape responses and predatory strikes) are often classified based on the shape of the body (C-shape or S-shape) at the end of stage 1. C-starts were initially considered synonymous with escape responses, and S-starts with predator strikes. However, fish sometimes escape using S-starts and attack prey using C-starts (for a review, see Domenici and Hale, 2019).

Table 1.

Performance variables measured during an escape response

Performance variables measured during an escape response
Performance variables measured during an escape response

Methods in experimental biology are increasingly employed to address interdisciplinary questions aimed at evaluating and predicting how organisms respond to human-induced environmental change (Stillman, 2019; Hof, 2021). However, there is also a pressing need to ensure that trainees and researchers venturing into new fields have access to resources allowing them to understand and rigorously apply these methodologies (Killen et al., 2021; Roche et al., 2022).

Here, we provide practical guidelines for carrying out escape response experiments in fishes across fields spanning behavioural ecology, ecophysiology, biomechanics and ecomorphology. We outline six important steps: (1) considering important species-specific characteristics, (2) designing an appropriate experimental setup and protocol, (3) recording escape responses, (4) extracting data from videos, (5) analysing data and (6) reporting methods and results (Fig. 1). We also provide a checklist to assist researchers in reporting their methods and results transparently and to facilitate study replication and evidence synthesis.

Fig. 1.

The steps involved in conducting, analysing and reporting escape response experiments in fish, with important considerations at each step. Considerations marked with an asterisk are detailed or expanded upon in the Supplementary Materials and Methods.

Fig. 1.

The steps involved in conducting, analysing and reporting escape response experiments in fish, with important considerations at each step. Considerations marked with an asterisk are detailed or expanded upon in the Supplementary Materials and Methods.

Swimming behaviour prior to an escape

Fish swimming behaviour has an important influence on methods of measuring escape performance and the ease with which experiments are carried out. Three broad categories of swimming behaviour can be considered for the purpose of escape response experiments: (1) continuous swimmers – fish that swim continually and often live in pelagic habitats; (2) intermittent swimmers – fish that swim in a stop-and-go fashion and often live near the substrate or other structures in demersal habitats; and (3) occasional swimmers – fish that spend most of their time immobile and often live in benthic habitats. The procedures to standardise a fish's position and motion prior to stimulation vary among these behaviour and habitat categories.

Continuous swimmers should be stimulated to elicit an escape response during constant, slow swimming, when fish are undisturbed. Some continuous swimming species do not behave naturally in aquaria, and natural swimming behaviour can be induced by providing a gentle current (<0.5 BL s−1) in a circular tank or a flow tunnel against which fish can swim at a constant, slow speed (Marras and Domenici, 2013). Many continuous swimmers are schooling species and testing them in a school rather than individually might be more ecologically relevant (Webb, 1980; Domenici and Batty, 1997; Short et al., 2020). Escape responses in intermittent swimmers should be triggered while they are still or swimming slowly. The experimenter should standardise the swimming behaviour of these species prior to stimulation or account for variation in swimming speed in the statistical analyses (see ‘Building statistical models’). Occasional swimmers tend to rest on the substrate and remain still for extended periods of time, often near or inside a shelter. Their position at the time of stimulation can be controlled by creating a shadow in a small area of the tank or by using a refuge that can be lifted prior to stimulation (see ‘Stimulus type and operation’).

Body size

Fish body size must be considered when deciding on the size of the experimental arena and the frame rate of the camera used to record escape responses. Body size also affects key escape variables (table 1 in Domenici and Hale, 2019). Unless size is a variable of interest in the study design (e.g. Domenici and Blake, 1993; Hale, 1996; Wakeling et al., 1999), researchers should aim to use fish of similar sizes across experimental treatments (e.g. within a body length range of 10%). A common practice to account for size differences among fish is to measure relative swimming speed in body lengths per second (BL s−1). Relative swimming speed can be useful for comparing results across studies; however, small fish are capable of much higher relative swimming speeds than large fish (Videler, 1993). Therefore, absolute values should always be presented alongside measures of relative swimming speed.

Arena size and shape

A key feature of any experimental setup to measure escape performance is the size of the experimental arena. The arena size must be sufficient for the test fish to engage in natural swimming behaviour and escape at maximum performance when startled. If the arena is too small, proximity to the walls can affect the fish's reaction and impede its movements. The minimum distance between the fish and the arena wall at the time of stimulation should be at least two body lengths to avoid wall effects on swimming performance and trajectory (see Eaton and Emberley, 1991; Mirjany et al., 2011). In contrast, if the arena is too large, the time spent by the fish in the camera's field of view will be limited (see ‘Video recording’), lengthening the time needed to conduct a trial. In general, the diameter of the arena should be 6–10 times the body length of the test fish, with continuous swimmers requiring larger arenas than intermittent or occasional swimmers.

Arena shape is another important consideration when designing an effective experimental setup. Holding tanks and aquaria are often rectangular, offering corners where shy species or individuals can hide and remain immobile. Circular tanks avoid this problem – although they do not eliminate problems associated with thigmotaxis (fish remaining close to or against the walls). If a circular tank is not available for species that tend to hide in corners, an acrylic sheet can be bent into a cylinder, fastened and placed inside a rectangular arena (e.g. Gingins et al., 2017).

Water depth, temperature and oxygen levels

Allowing a fish to display maximum escape performance requires that the experimental arena be free of physical and physiological constraints. Three key characteristics of the test water require consideration.

Water depth

The depth of the water in the arena should allow fish to swim without contacting the arena floor or the water surface. However, restricting the water depth in the arena facilitates kinematic measurements in two dimensions. A suggested rule of thumb is a water depth of 3–4 body depths for occasional swimmers and 4–5 body depths for intermittent or continuous swimmers. When the water depth exceeds this level, a mirror or an additional camera can be used to record vertical movements for kinematic analyses in three dimensions (see Supplementary Materials and Methods) or to exclude trials with vertical movements above a certain threshold (e.g. one body depth; Roche, 2021).

Water temperature

Water temperature affects escape performance (reviewed in Wakeling, 2006; Domenici, 2010a) and should therefore be constant throughout the experiments (within ±1°C of the set temperature). A stable temperature should be achieved without disturbing the fish, by either working in a temperature-controlled environment or using a continuous water exchange with an external, temperature-controlled water bath.

Dissolved oxygen

Hypoxia can lower escape performance (Domenici et al., 2007), and dissolved oxygen levels should be maintained above 90% by bubbling air into the arena or water bath using an air stone. The air pump should be turned off or the air stone removed from the arena prior to stimulation to avoid disturbance.

Lighting and contrast

Extracting high-quality kinematic data from videos of escape responses requires that the experimental arena be moderately and homogeneously lit, creating contrast between the animal and the background. Light reflection on the water surface should be avoided as it creates glare that can interfere with motion tracking (see ‘Data extraction’).

Multiple flood lights (typically two to four) can be used to uniformly illuminate the experimental arena. LED lights are preferable to halogen lights, which generate a lot of heat, and to neon lights, which flicker. Flood lights should be placed above the arena, outside the arena walls and facing down at an angle that avoids direct reflection into the camera lens. Uniform lighting can also be achieved by orienting the lights upward and illuminating a white panel above the arena. Alternatively, LED light strips can be placed around the top of the arena. Another option is to position light strips or floodlights below an experimental arena with a white bottom. This setup will illuminate the area around the test fish, creating a high contrast between the dark body of the fish and its white surroundings. The arena must be raised above the light source to achieve homogeneous illumination, and paper sheets can be positioned above LED (not halogen) lights to act as diffusers. The arena background should provide an acceptable level of contrast between the fish's body and its surroundings to facilitate tracking.

To facilitate data extraction from recorded videos, the experimenter should test the tracking software on multiple sample videos before commencing data collection (see ‘Data extraction’). This step is important so that the setup can be modified if tracking proves challenging (see Sridhar et al., 2019).

Acclimation to the experimental arena

Acclimation is needed when transferring a test fish from its holding tank to the experimental arena. The time required for proper acclimation can vary depending on the species and individual, and the handling procedure used for transferring fish. Work on cod (Artigas et al., 2005) and zebrafish (Ramsay et al., 2009) indicates that ventilation rates and cortisol levels return to control levels 60–90 min after fish are handled with nets. For some species, using a water-filled container rather than a dip net to avoid air exposure when transferring individuals might reduce stress levels and acclimation time (Brydges et al., 2009). Several escape response studies have used acclimation times of 30–60 min (e.g. Marras et al., 2011; Schakmann et al., 2021), but preliminary trials should ideally be carried out on a given test species to assess how different acclimation times affect individuals' responsiveness to stimulation. Similarly, the possibility of habituation or fatigue should be investigated when carrying out repeated stimulations to determine an appropriate rest period between trials (e.g. Jornod and Roche, 2015).

Stimulus type and operation

Various approaches can be used to elicit an escape response, including acoustic, mechanical, visual and tactile stimuli. Information on each stimulus type and their pros and cons is provided in the Supplementary Materials and Methods.

When operating a stimulus, regardless of its type, delimiting a restricted area of the experimental arena in which the test fish is stimulated facilitates video recording (see ‘Camera field of view’) and helps to standardise the test fish's distance to the stimulus (Domenici and Batty, 1997). For mechano-acoustic or visual stimuli, placing the stimulus close to one of the arena walls helps position the ‘stimulation area’ towards the centre of the arena. If preliminary trials indicate that fish do not approach the pre-defined stimulation area 1–2 h post release into the arena, slightly shading this area with the use of a mesh net can help induce proximity (Turesson et al., 2009). For occasional swimmers such as gobies, a shelter can also be provided in the stimulation area, which can be lifted prior to stimulation (Kimura et al., 2022).

Standardising the orientation of the test fish relative to the stimulus is important, because body orientation can affect the perceived strength of the stimulus, escape directionality and the fish's turn angle, which influences escape duration (Table 1; Domenici and Blake, 1993, 1997; Kimura and Kawabata, 2018). Orientation relative to the stimulus ranges between 0 and 180 deg; it is calculated immediately prior to the onset of stage 1 (Box 1) as the angle between the fish's body axis, i.e. the straight line joining the tip of the snout to the fish's centre of mass (the location at which forces are assumed to act), and the line joining the centre of the stimulus to the fish's centre of mass. Standardisation can be achieved by stimulating fish only within a certain range of orientations relative to the stimulus (e.g. 60–120 deg) and/or including orientation as a covariate in the statistical model(s) (see ‘Building statistical models’).

Avoiding disturbances

Undesirable visual or auditory stimuli in the environment can affect how a test fish perceives and reacts to the stimulus – for example, by altering its responsiveness or response latency (see Table 1 for definitions). To avoid disturbing the test fish, it is important to separate the arena and experimenter using an opaque partition or screen (e.g. Marras et al., 2011). If the camera displays a live video feed, the experimenter can observe the fish's movements on an external monitor and operate the stimulus from behind the screen. When this is not possible, a small opening in the screen can allow the experimenter to observe the arena without disturbing the fish. Unwanted disturbance prior to stimulation, such as from the physical activation of the stimulus, should be avoided as it can bias response latency (see Supplementary Materials and Methods).

Camera position

When recording escape responses in two dimensions, the camera should be placed above or below the location of the stimulus delivery in the experimental arena (Fig. S1). The camera should be far enough from the area to minimise image distortion around the edges of the frame, or images should be corrected for distortion (e.g. with MATLAB's undistortImage or fisheye correction available in most video processing software, such as Adobe Premiere). Ideally, the camera should be controlled remotely. If not, it should be activated by the experimenter without disturbing the test fish. If a high-speed camera has a limited amount of recording time (i.e. due to loop recording) and/or cannot be controlled remotely, it is advisable to position the camera in front of the setup through a hole in the screen and record escape responses using a mirror angled at 45 deg above or below the experimental arena (e.g. Gingins et al., 2017).

The camera, experimental arena and stimulus should not be physically connected to each other. This is to avoid the camera shaking when the stimulus is released, early stimulation of the test fish if the stimulus release mechanism is connected to the arena, or disturbing the test fish when replacing a battery or SD card in a camera connected to the arena.

Camera field of view

The precision of the digitization process for extracting data from escape response videos is related to the precision of the digital image recorded by the camera in pixels. Therefore, high-speed cameras with a high resolution are preferred, and the optical zoom should be used to record only a specific area of interest within the experimental arena. The camera's field of view should be restricted to the ‘stimulation area’ (see ‘Stimulus operation’) because filming the entire arena reduces the resolution of kinematic measurements. The behaviour of the test fish beyond the stimulation area can be monitored by an additional camera with a lower temporal resolution (e.g. 30 Hz). Ideally, the width and height of the camera's field of view should be approximately 4–5 times the body length of the test fish (Fig. S2). A scale (e.g. a ruler or grid) is needed in this area to allow kinematic measurements (see ‘Data extraction’).

Camera frame rate

Most studies of escape responses are based on high-speed video recording with frame rates of 240–1000 Hz (the frame rate of standard video is 24–30 Hz). High frame rates are needed to capture rapid body motions that last only a few milliseconds. Ideally, the camera frame rate should be set to capture a minimum of five frames during stage 1 of the escape response (i.e. <20 deg resolution for a 90 deg turn), allowing measurements of instantaneous rather than mean locomotor performance. Instantaneous performance is akin to a snapshot in time throughout the escape response, requiring multiple camera frames. In contrast, mean performance is an average value that can be based only on two frames, one at the start and one at the end of the event. The number of frames recorded during stage 1 of the escape response depends not only on the camera frame rate but also on the duration and total angle of the turn performed by the fish.

The duration of the turn during stage 1 depends primarily on the size of the fish and the temperature of the water, with smaller fish and higher temperatures leading to faster turns, which require higher frame rates (Wakeling, 2006). Although fish species may also be important, data reviewed by Domenici (2001) show that over 90% of the variation in turning rate observed among aquatic vertebrates (most of which were fish) is due to body size (R2=0.92–0.95). For example, the average turning rate of a 5, 10 and 25 cm fish is around 5500, 3000 and 1500 deg s−1, respectively (Domenici, 2001). Parson et al. (2011) found that turning rates decrease with size even when considering invertebrate species. Therefore, a typical 90 deg turn for fishes of these sizes will produce approximately 4, 7 and 14 frames if recorded at 240 Hz (Fig. S3). For this reason, frame rates of 500 Hz or higher are recommended for fish below 5 cm. Larval fishes typically require frame rates of 1000 Hz, whereas frame rates of 200–250 Hz are generally suitable for ecological studies of larger fish (>15 cm). For biomechanical studies that require detailed measurements of velocity or acceleration, frame rates of 500 or 1000 Hz are usually needed, regardless of fish size. Fig. S3 indicates the frame rate needed for a given angular resolution during stage 1 as a function of fish size.

The temporal resolution of the camera will also affect the precision with which response latency can be determined. Minimum response latencies are in the order of 5–20 ms (Eaton and Hackett, 1984; Domenici and Hale, 2019) and independent of fish size (Turesson and Domenici, 2007). Therefore, a minimum frame rate of 240 Hz is recommended for measurements of escape latency regardless of fish size.

Identifying a fish's centre of mass

The point typically used as a reference for the measurement of distance–time performance variables (Table 1) is the centre of mass of the fish. Several methods have been used to estimate the location of a fish's centre of mass (Fig. 2). One commonly used approach in biomechanical studies is to divide the fish's body into many small segments and calculate the centre of mass as the average location of the fish, weighted by the density of the different body segments (see Supplementary Materials and Methods). R and MATLAB code for three methods that follow this approach can be found in Tytell (2023), allowing users to calculate and track the true centre of mass, the volume centre of mass and the area centre of mass.

Fig. 2.

A comparison of a fish's swimming speed (velocity) during the first two stages of an escape response using four methods to determine the location of the fish's centre of mass (CoM): the true CoM, the volume CoM, the area CoM and the ‘stretched-straight’ CoM. Note that the true, area and volume CoM can be located outside the fish's body when it is curved. The stretched-straight method is the least accurate from a biomechanical perspective, but the simplest and most ecologically relevant, as predators tend to target the visual CoM on a prey's body.

Fig. 2.

A comparison of a fish's swimming speed (velocity) during the first two stages of an escape response using four methods to determine the location of the fish's centre of mass (CoM): the true CoM, the volume CoM, the area CoM and the ‘stretched-straight’ CoM. Note that the true, area and volume CoM can be located outside the fish's body when it is curved. The stretched-straight method is the least accurate from a biomechanical perspective, but the simplest and most ecologically relevant, as predators tend to target the visual CoM on a prey's body.

The ‘stretched-straight’ method is another approach for calculating a fish's centre of mass. This method is the least accurate from a biomechanical perspective (Fig. 2), but is the simplest and the most ecologically relevant, because predators tend to target the visual centre of mass on a prey's body (Webb and Skadsen, 1980; Walker et al., 2005). The experimenter identifies the location of the centre of mass on euthanised, rigid specimens when the body is stretched straight (see Supplementary Materials and Methods) and tracks that point during the escape response. The centre of mass can be physically marked on live fish prior to an experiment (e.g. with a small piece of reflective tape temporarily glued onto the body; Domenici et al., 2008) or digitally identified in each video frame using a cubic spline algorithm (Tytell, 2023), which has the advantage of being a non-invasive technique. The stretched-straight centre of mass is generally located approximately 0.35–0.40 BL from the tip of the fish's snout (Webb, 1978).

Escape performance variables

Early work on escape responses tended to focus on variables related to locomotor performance, such as swimming speed, acceleration and turning radius (Weihs, 1973; Webb, 1975, 1976). Subsequently, it became apparent that non-locomotor (i.e. behavioural) performance is also critically important in affecting escape success, including responsiveness to the threat, the timing of the response and the direction of the escape (Walker et al., 2005; Fuiman et al., 2006; Domenici, 2010a). Importantly, however, the influence of different escape performance variables on survival is likely to be context dependent and species specific (Domenici and Hale, 2019). Thus, the choice of variables to measure depends on the species and question(s) being investigated. Table 1 provides an overview of commonly measured escape performance variables.

Distance–time variables (Table 1) are often used to measure locomotor performance during an escape response, and include distance travelled as well as speed and acceleration (i.e. the first two derivatives of distance with respect to time). Owing to the process of taking numerical derivatives (see discussion in Van Breugel et al., 2020 and Walker, 1998), errors in distance measurements are compounded when calculating values of speed and acceleration: acceleration is the noisiest of the three variables, whereas distance travelled is the least noisy. For this reason, smoothing (the process of removing noise or jitter from a signal to reveal an underlying trend) is recommended when computing maximum values of speed and acceleration (see ‘Smoothing’). Importantly, measurement errors are more likely to occur when extracting distance measurements from low-resolution (i.e. pixelated) images, which makes the centre of mass difficult to accurately identify and track.

Time frame of the analysis

Another important consideration when assessing distance–time variables is the time interval over which to take measurements. It is generally recommended to record distance–time variables within a fixed time rather than within kinematic stages, because predator reactions are likely to be constrained by time rather than by the kinematic stages of the prey's escape response. For example, if one assesses maximum swimming speed based on performance values achieved by the end of stage 1, one would find that the greatest speed tends to occur during escapes with a longer stage 1, simply because of the longer time available to achieve high speeds in these responses. Similarly, large fish achieve higher speeds than small fish by the end of stage 1 because they take longer to complete this stage. Hence, as suggested by Webb (1976), a fixed time is typically used for measuring escape distance and maximum speed. This fixed time can be chosen as the average duration of stages 1 and 2 across all escape responses in a given group or study (Domenici et al., 2008). Maximum acceleration can be measured as the peak acceleration at any point in time during the escape response (Domenici et al., 2008).

Manual versus automated data extraction

Measuring the motion of a fish's centre of mass can be done using manual or automated tracking methods. Manual tracking requires physically marking the centre of mass on test fish before running experiments, e.g. using elastomer tags or reflective tape (Domenici et al., 2008), and is therefore restricted to tracking the ‘stretched-straight’ centre of mass. Manual tracking is more time consuming than automated tracking but has the benefit that it can sometimes accommodate low-quality video resulting from challenging field conditions (e.g. insufficient contrast, presence of glare disturbances), which complicate automated tracking solutions. Several options of free software are available to implement manual tracking, including the MTrackJ package in ImageJ (Meijering et al., 2012), Kinovea (www.kinovea.org), DLTdv (https://biomech.web.unc.edu/dltdv; available now without a MATLAB license; Hedrick, 2008) and Tracker (https://physlets.org/tracker). The user must mark the centre of mass in subsequent camera frames, for example by clicking on its location, and the software outputs Cartesian coordinates (X and Y), which can be converted into kinematic measurements using an appropriate scale.

Automated tracking solutions are provided by computer-vision techniques for object detection, which rely on algorithms to track objects through frame sequences (Dell et al., 2014). Numerous options exist in the computer-vision literature, but often require a programming background to implement (Panadeiro et al., 2021). The DeepLabCut machine-learning algorithm (Mathis and Mathis, 2020) and machine learning through DLTdv can automatically track many points on animal bodies without markers, and are becoming more user friendly. Other freeware such as Kinovea (www.kinovea.org) and Tracktor (Sridhar et al., 2019) can automatically track single fish in noisy environments with minimal coding skills.

Smoothing

Smoothing is essential to estimate velocity and acceleration during an escape response (Walker, 1998). Derivatives tend to amplify noise in measurements, and, without appropriate smoothing, can lead to incorrect results. Smoothing splines produce the smoothest possible curve through a set of points given a certain error and are recommended to estimate acceleration provided the approximate accuracy of a digitising technique is known (Walker, 1998). It is also possible to use smoothing regressions (Lanczos, 1956) and low-pass digital filters, such as Butterworth or Chebyshev filters, with cut-off frequencies corresponding to durations less than 10–20% of the stage 1 duration (e.g. if stage 1 is 50 ms, then cut-off frequencies greater than 100–200 Hz are recommended). Finally, one can use running mean or running median filters with durations less than 10–20% of the duration of stage 1 (e.g. with a frame rate of 1000 Hz and a stage 1 of 50 ms, the duration of the filter should be less than 5–10 frames long). Example code is provided in Tytell (2023).

Selecting response variables

Escape response experiments allow measuring numerous locomotor and non-locomotor performance variables (Table 1), and researchers are often interested in comparing these measures among species, populations or treatment groups. Given that speed and acceleration are inherently noisy, they are generally considered less-reliable measures of distance–time performance than escape distance (Domenici and Blake, 1997). However, both can have an important fitness value because they relate to energetics, and therefore should be considered or analysed if a researcher is interested in muscle power or energy consumption (e.g. Walker et al., 2005). Performance measures that are recorded but not included in statistical models can be presented in a table for descriptive purposes (e.g. Roche, 2021).

Building statistical models

The choice of statistical model(s) to examine relationships (e.g. how does size or body/fin shape affect escape performance) or compare groups of fishes depends largely on a study's experimental design and the nature of the response variables examined. For example, whether each test fish is stimulated once (Domenici et al., 2008) or multiple times (Gingins et al., 2017) in an experiment affects whether repeated measurements must be accounted for. Repeated stimulations are often used in studies of escape response because obtaining multiple measurements increases the likelihood of observing maximum performance. Linear mixed-effects models can be used to accommodate repeated measurements on individuals and allow examination of how multiple stimulus presentations affect performance, for example, through habituation (Marras et al., 2011; Roche et al., 2016; Roche, 2021). Most escape performance measures are expected to generate Gaussian (i.e. normal) error distributions. Exceptions are response latency, which can be bimodal (Domenici and Batty, 1997), and responsiveness and directionality, which follow a binomial distribution – these can be modelled using generalised linear models. Variation in fish body size, motion (swimming speed) and position (distance and orientation relative to the stimulus) prior to stimulation can be controlled for by including these variables as covariates in the statistical model(s) (e.g. Gingins et al., 2017; Roche, 2021).

Methods and results reporting

Evaluating whether an experiment was rigorously conducted is only possible if the methods and results are transparently and comprehensively reported. Unfortunately, studies in experimental biology often suffer from a lack of reporting consistency and underreporting of methods and results (Clark et al., 2013; Marqués et al., 2020; Killen et al., 2021). To address this issue, we provide a checklist of key information that should be presented when authors report escape response experiments (Table 2). Reporting checklists are valuable tools to assist authors in designing and reporting studies; to help editors and readers assess the reliability of a study's findings; and to facilitate study replication and evidence synthesis such as meta-analysis (Parker et al., 2018; Killen et al., 2021).

Table 2.

A checklist of criteria for reporting the methods and results of escape response experiments in fish

A checklist of criteria for reporting the methods and results of escape response experiments in fish
A checklist of criteria for reporting the methods and results of escape response experiments in fish

Published studies of escape response experiments in fish are steadily increasing, partly owing to the greater accessibility of high-speed cameras but also owing to a heightened interest in how environmental stressors affect whole-animal performance and predator–prey interactions (Domenici et al., 2019). As the study of fish escape responses continues to expand, standardising experiments and reporting practices is critical to facilitate integrative and comparative studies, as well as evidence synthesis. Rigorous methods and reporting practices strengthen not only the biological conclusions derived from a study's results, but also their transparency and reproducibility (Ihle et al., 2017; Aaron and Chew, 2021). The guidelines we have provided here are intended to help researchers design, execute and report escape response experiments, particularly students and primary investigators entering this rapidly evolving field of research. Importantly, we acknowledge that our own previous work on escape responses is imperfect and contains design and reporting deficiencies. This Commentary has been an opportunity to reflect on improvements going forward, and we hope that it will be as useful to others as it has been for us.

We thank Sandy Kawano and an anonymous reviewer for helpful comments on the paper.

Data availability

The data and code to associated with this paper are publicly available from Zenodo (Tytell, 2023): https://doi.org/10.5281/zenodo.7577129.

Aaron
,
J.
and
Chew
,
T.-L.
(
2021
).
A guide to accurate reporting in digital image processing – can anyone reproduce your quantitative analysis?
J. Cell Sci.
134
,
jcs254151
.
Artigas
,
M.
,
Skjæraasen
,
J.
,
Utne-Palm
,
A.
and
Nilsen
,
T.
(
2005
).
Recovery from handling stress in Gadus morhua
.
J. Fish Biol.
67
,
384
-
391
.
Batty
,
R.
(
1989
).
Escape responses of herring larvae to visual stimuli
.
J. Mar. Biol. Assoc. U.K
69
,
647
-
654
.
Brydges
,
N. M.
,
Boulcott
,
P.
,
Ellis
,
T.
and
Braithwaite
,
V. A.
(
2009
).
Quantifying stress responses induced by different handling methods in three species of fish
.
Appl. Anim. Behav. Sci.
116
,
295
-
301
.
Cade
,
D. E.
,
Carey
,
N.
,
Domenici
,
P.
,
Potvin
,
J.
and
Goldbogen
,
J. A.
(
2020
).
Predator-informed looming stimulus experiments reveal how large filter feeding whales capture highly maneuverable forage fish
.
Proc. Natl. Acad. Sci. USA
117
,
472
-
478
.
Clark
,
T. D.
,
Sandblom
,
E.
and
Jutfelt
,
F.
(
2013
).
Aerobic scope measurements of fishes in an era of climate change: respirometry, relevance and recommendations
.
J. Exp. Biol.
216
,
2771
-
2782
.
Dell
,
A. I.
,
Bender
,
J. A.
,
Branson
,
K.
,
Couzin
,
I. D.
,
de Polavieja
,
G. G.
,
Noldus
,
L. P. J. J.
,
Pérez-Escudero
,
A.
,
Perona
,
P.
,
Straw
,
A. D.
,
Wikelski
,
M.
et al. 
(
2014
).
Automated image-based tracking and its application in ecology
.
Trends Ecol. Evol.
29
,
417
-
428
.
Dill
,
L. M.
(
1974
).
The escape response of the zebra danio (Brachydanio rerio) I. The stimulus for escape
.
Anim. Behav.
22
,
711
-
722
.
Domenici
,
P.
(
2001
).
The scaling of locomotor performance in predator–prey encounters: from fish to killer whales
.
Comp. Biochem. Physiol. A
131
,
169
-
182
.
Domenici
,
P.
(
2002
).
The visually mediated escape response in fish: predicting prey responsiveness and the locomotor behaviour of predators and prey
.
Mar. Freshwat. Behav. Physiol.
35
,
87
-
110
.
Domenici
,
P.
(
2010a
).
Context-dependent variability in the components of fish escape response: Integrating locomotor performance and behavior
.
J. Exp. Zool. A
313A
,
59
-
79
.
Domenici
,
P
. (
2010b
).
Escape responses in fish: kinematics, performance, and behavior
. In
Fish Locomotion. An Eco-Ethological Prespective
(ed.
P.
Domenici
and
B. G.
Kapoor
), pp.
123
-
170
.
Enfield, NH
:
Science Publishers
.
Domenici
,
P
. (
2011
).
Fast start
. In
Encyclopedia of Fish Physiology: from Genome to Environment
, Vol.
1
(ed.
A. P.
Farrell
), pp.
587
-
596
.
San Diego
:
Academic Press
.
Domenici
,
P.
and
Batty
,
R. S.
(
1997
).
Escape behaviour of solitary herring (Clupea harengus) and comparisons with schooling individuals
.
Mar. Biol.
128
,
29
-
38
.
Domenici
,
P.
and
Blake
,
R. W.
(
1991
).
The kinematics and performance of the escape response in the angelfish (Pterophyllum eimekei)
.
J. Exp. Biol.
156
,
187
-
205
.
Domenici
,
P.
and
Blake
,
R. W.
(
1993
).
The effect of size on the kinematics and performance of angelfish (Pterophyllum eimekei) escape responses
.
Can. J. Zool.
71
,
2319
-
2326
.
Domenici
,
P.
and
Blake
,
R. W.
(
1997
).
The kinematics and performance of fish fast-start swimming
.
J. Exp. Biol.
200
,
1165
-
1178
.
Domenici
,
P.
and
Hale
,
M. E.
(
2019
).
Escape responses of fish: a review of the diversity in motor control, kinematics and behaviour
.
J. Exp. Biol.
222
,
jeb166009
.
Domenici
,
P.
,
Standen
,
E. M.
and
Levine
,
R. P.
(
2004
).
Escape manoeuvres in the spiny dogfish (Squalus acanthias)
.
J. Exp. Biol.
207
,
2339
-
2349
.
Domenici
,
P.
,
Lefrançois
,
C.
and
Shingles
,
A.
(
2007
).
Hypoxia and the antipredator behaviours of fishes
.
Philos. Trans. R. Soc. B Biol. Sci.
362
,
2105
-
2121
.
Domenici
,
P.
,
Turesson
,
H.
,
Brodersen
,
J.
and
Brönmark
,
C.
(
2008
).
Predator-induced morphology enhances escape locomotion in crucian carp
.
Proc. R. Soc. B Biol. Sci.
275
,
195
-
201
.
Domenici
,
P.
,
Blagburn
,
J. M.
and
Bacon
,
J. P.
(
2011
).
Animal escapology I: theoretical issues and emerging trends in escape trajectories
.
J. Exp. Biol.
214
,
2463
-
2473
.
Domenici
,
P.
,
Allan
,
B. J. M.
,
Lefrançois
,
C.
and
McCormick
,
M. I.
(
2019
).
The effect of climate change on the escape kinematics and performance of fishes: implications for future predator–prey interactions
.
Conserv. Physiol.
7
,
coz078
.
Eaton
,
R. C.
and
Emberley
,
D. S.
(
1991
).
How stimulus direction determines the trajectory of the Mauthner-initiated escape response in a teleost fish
.
J. Exp. Biol.
161
,
469
-
487
.
Eaton
,
R. C.
and
Hackett
,
J. T
. (
1984
).
The role of the Mauthner cell in fast-starts involving escape in teleost fishes
. In
Neural Mechanisms of Startle Behavior
(ed.
R. C.
Eaton
), pp.
213
-
266
.
New York
:
Plenum Press
.
Eaton
,
R. C.
,
Lee
,
R. K. K.
and
Foreman
,
M. B.
(
2001
).
The Mauthner cell and other identified neurons of the brainstem escape network of fish
.
Prog. Neurobiol.
63
,
467
-
485
.
Fuiman
,
L. A.
,
Rose
,
K. A.
,
Cowan
,
J. H.
and
Smith
,
E. P.
(
2006
).
Survival skills required for predator evasion by fish larvae and their relation to laboratory measures of performance
.
Anim. Behav.
71
,
1389
-
1399
.
Gingins
,
S.
,
Roche
,
D. G.
and
Bshary
,
R.
(
2017
).
Mutualistic cleaner fish maintains high escape performance despite privileged relationship with predators
.
Proc. R. Soc. B Biol. Sci.
284
,
20162469
.
Hale
,
M. E.
(
1996
).
The development of fast-start performance in fishes: escape kinematics of the chinook salmon (Oncorhynchus tshawytscha)
.
Am. Zool.
36
,
695
-
709
.
Hamming
,
R.
(
2012
).
Numerical Methods for Scientists and Engineers
.
New York
:
Dover Publications
.
Hecker
,
A.
,
Schulze
,
W.
,
Oster
,
J.
,
Richter
,
D. O.
and
Schuster
,
S.
(
2020
).
Removing a single neuron in a vertebrate brain forever abolishes an essential behavior
.
Proc. Natl. Acad. Sci. USA
117
,
3254
-
3260
.
Hedrick
,
T. L.
(
2008
).
Software techniques for two-and three-dimensional kinematic measurements of biological and biomimetic systems
.
Bioinspir. Biomim.
3
,
034001
.
Hof
,
C.
(
2021
).
Towards more integration of physiology, dispersal and land-use change to understand the responses of species to climate change
.
J. Exp. Biol.
224
,
jeb238352
.
Ihle
,
M.
,
Winney
,
I. S.
,
Krystalli
,
A.
and
Croucher
,
M.
(
2017
).
Striving for transparent and credible research: practical guidelines for behavioral ecologists
.
Behav. Ecol.
28
,
348
-
354
.
Jornod
,
M.
and
Roche
,
D. G.
(
2015
).
Inter- vs intra-individual variation and temporal repeatability of escape responses in the coral reef fish Amblyglyphidodon curacao
.
Biol. Open
4
,
1395
-
1399
.
Katzir
,
G.
and
Camhi
,
J. M.
(
1993
).
Escape response of black mollies (Poecilia sphenops) to predatory dives of a pied kingfisher (Ceryle rudis)
.
Copeia
1993
,
549
-
553
.
Kawabata
,
Y.
,
Akada
,
H.
,
Shimatani
,
K.-I.
,
Nishihara
,
G. N.
,
Kimura
,
H.
,
Nishiumi
,
N.
and
Domenici
,
P.
(
2023
).
Multiple preferred escape trajectories are explained by a geometric model incorporating prey's turn and predator attack endpoint
.
Elife
12
,
e77699
.
Killen
,
S. S.
,
Christensen
,
E. A. F.
,
Cortese
,
D.
,
Závorka
,
L.
,
Norin
,
T.
,
Cotgrove
,
L.
,
Crespel
,
A.
,
Munson
,
A.
,
Nati
,
J. J. H.
,
Papatheodoulou
,
M.
et al. 
(
2021
).
Guidelines for reporting methods to estimate metabolic rates by aquatic intermittent-flow respirometry
.
J. Exp. Biol.
224
,
jeb242522
.
Kimura
,
H.
and
Kawabata
,
Y.
(
2018
).
Effect of initial body orientation on escape probability of prey fish escaping from predators
.
Biol. Open
7
,
bio023812
.
Kimura
,
H.
,
Pfalzgraff
,
T.
,
Levet
,
M.
,
Kawabata
,
Y.
,
Steffensen
,
J. F.
,
Johansen
,
J. L.
and
Domenici
,
P.
(
2022
).
Escaping from multiple visual threats: modulation of escape responses in Pacific staghorn sculpin (Leptocottus armatus)
.
J. Exp. Biol.
225
,
jeb243328
.
Korn
,
H.
and
Faber
,
D. S.
(
2005
).
The Mauthner cell half a century later: a neurobiological model for decision-making?
Neuron
47
,
13
-
28
.
Lanczos
,
C.
(
1956
).
Applied Analysis
.
Englewood Cliffs, NJ
:
Prentice Hall
.
Lefrançois
,
C.
,
Shingles
,
A.
and
Domenici
,
P.
(
2005
).
The effect of hypoxia on locomotor performance and behaviour during escape in Liza aurata
.
J. Fish Biol.
67
,
1711
-
1729
.
Marqués
,
G.
,
Pengo
,
T.
and
Sanders
,
M. A.
(
2020
).
Imaging methods are vastly underreported in biomedical research
.
eLife
9
,
e55133
.
Marras
,
S.
and
Domenici
,
P.
(
2013
).
Schooling fish under attack are not all equal: some lead, others follow
.
PLoS One
8
,
e65784
.
Marras
,
S.
,
Killen
,
S. S.
,
Claireaux
,
G.
,
Domenici
,
P.
and
McKenzie
,
D. J.
(
2011
).
Behavioural and kinematic components of the fast-start escape response in fish: individual variation and temporal repeatability
.
J. Exp. Biol.
214
,
3102
-
3110
.
Mathis
,
M. W.
and
Mathis
,
A.
(
2020
).
Deep learning tools for the measurement of animal behavior in neuroscience
.
Curr. Opin. Neurobiol.
60
,
1
-
11
.
McCormick
,
M. I.
,
Fakan
,
E.
and
Allan
,
B. J.
(
2018
).
Behavioural measures determine survivorship within the hierarchy of whole–organism phenotypic traits
.
Funct. Ecol.
32
,
958
-
969
.
Meager
,
J. J.
,
Domenici
,
P.
,
Shingles
,
A.
and
Utne-Palm
,
A. C.
(
2006
).
Escape responses in juvenile Atlantic cod Gadus morhua L.: the effects of turbidity and predator speed
.
J. Exp. Biol.
209
,
4174
-
4184
.
Meijering
,
E.
,
Dzyubachyk
,
O.
and
Smal
,
I.
(
2012
).
Methods for cell and particle tracking
.
Methods Enzymol.
504
,
183
-
200
.
Mirjany
,
M.
,
Preuss
,
T.
and
Faber
,
D. S.
(
2011
).
Role of the lateral line mechanosensory system in directionality of goldfish auditory evoked escape response
.
J. Exp. Biol.
214
,
3358
-
3367
.
Paglianti
,
A.
and
Domenici
,
P.
(
2006
).
The effect of size on the timing of visually mediated escape behaviour in staghorn sculpin Leptocottus armatus
.
J. Fish Biol.
68
,
1177
-
1191
.
Panadeiro
,
V.
,
Rodriguez
,
A.
,
Henry
,
J.
,
Wlodkowic
,
D.
and
Andersson
,
M.
(
2021
).
A review of 28 free animal-tracking software applications: current features and limitations
.
Lab. Anim. (NY)
50
,
246
-
254
.
Parker
,
T. H.
,
Griffith
,
S. C.
,
Bronstein
,
J. L.
,
Fidler
,
F.
,
Foster
,
S.
,
Fraser
,
H.
,
Forstmeier
,
W.
,
Gurevitch
,
J.
,
Koricheva
,
J.
and
Seppelt
,
R.
(
2018
).
Empowering peer reviewers with a checklist to improve transparency
.
Nat. Ecol. Evol.
2
,
929
-
935
.
Parson
,
J. M.
,
Fish
,
F. E.
and
Nicastro
,
A. J.
(
2011
).
Turning performance of batoids: limitations of a rigid body
.
J. Exp. Mar. Biol. Ecol.
402
,
12
-
18
.
Ramsay
,
J. M.
,
Feist
,
G. W.
,
Varga
,
Z. M.
,
Westerfield
,
M.
,
Kent
,
M. L.
and
Schreck
,
C. B.
(
2009
).
Whole-body cortisol response of zebrafish to acute net handling stress
.
Aquaculture
297
,
157
-
162
.
Roche
,
D. G.
(
2021
).
Effects of wave-driven water flow on the fast-start escape response of juvenile coral reef damselfishes
.
J. Exp. Biol.
224
,
jeb.234351
.
Roche
,
D. G.
,
Careau
,
V.
and
Binning
,
S. A.
(
2016
).
Demystifying animal ‘personality’ (or not): why individual variation matters to experimental biologists
.
J. Exp. Biol.
219
,
3832
-
3843
.
Roche
,
D. G.
,
Raby
,
G. D.
,
Norin
,
T.
,
Ern
,
R.
,
Scheuffele
,
H.
,
Skeeles
,
M.
,
Morgan
,
R.
,
Andreassen
,
A. H.
,
Clements
,
J. C.
,
Louissaint
,
S.
et al. 
(
2022
).
Paths towards greater consensus building in experimental biology
.
J. Exp. Biol.
225
,
jeb243559
.
Schakmann
,
M.
,
Becker
,
V.
,
Søgaard
,
M.
,
Johansen
,
J. L.
,
Steffensen
,
J. F.
and
Domenici
,
P.
(
2021
).
Latency of mechanically stimulated escape responses in the Pacific spiny dogfish, Squalus suckleyi
.
J. Exp. Biol.
224
,
jeb230698
.
Short
,
M.
,
White
,
P. R.
,
Leighton
,
T. G.
and
Kemp
,
P. S.
(
2020
).
Influence of acoustics on the collective behaviour of a shoaling freshwater fish
.
Freshwat. Biol.
65
,
2186
-
2195
.
Soto
,
A.
,
Stewart
,
W. J.
and
McHenry
,
M. J.
(
2015
).
When optimal strategy matters to prey fish
.
Integr. Comp. Biol.
55
,
110
-
120
.
Sridhar
,
V. H.
,
Roche
,
D. G.
and
Gingins
,
S.
(
2019
).
Tracktor: image-based automated tracking of animal movement and behaviour
.
Methods Ecol. Evol.
10
,
815
-
820
.
Stillman
,
J. H.
(
2019
).
Heat waves, the new normal: summertime temperature extremes will impact animals, ecosystems, and human communities
.
Physiology
34
,
86
-
100
.
Turesson
,
H.
and
Domenici
,
P.
(
2007
).
Escape latency is size independent in grey mullet
.
J. Fish Biol.
71
,
253
-
259
.
Turesson
,
H.
,
Satta
,
A.
and
Domenici
,
P.
(
2009
).
Preparing for escape: anti-predator posture and fast-start performance in gobies
.
J. Exp. Biol.
212
,
2925
-
2933
.
Tytell
,
E. D
. (
2023
).
Code for computing and tracking a fish's centre of mass during an escape response using R or Matlab (v1.0.0). Zenodo. https://doi.org/10.5281/zenodo.7577129
.
Tytell
,
E. D.
and
Lauder
,
G. V.
(
2008
).
Hydrodynamics of the escape response in bluegill sunfish, Lepomis macrochirus
.
J. Exp. Biol.
211
,
3359
-
3369
.
Van Breugel
,
F.
,
Kutz
,
J. N.
and
Brunton
,
B. W.
(
2020
).
Numerical differentiation of noisy data: A unifying multi-objective optimization framework
.
IEEE Access
8
,
196865
-
196877
.
Videler
,
J. J.
(
1993
).
Fish Swimming
.
London
:
Chapman & Hall
.
Wakeling
,
J. A
. (
2006
).
Fast-start mechanics
. In
Fish Biomechanics
(ed.
R. E.
Shadwick
and
G. V.
Lauder
), pp.
333
-
368
.
San Diego
:
Academic Press
.
Wakeling
,
J. M.
,
Kemp
,
K. M.
and
Johnston
,
I. A.
(
1999
).
The biomechanics of fast-starts during ontogeny in the common carp Cyprinus carpio
.
J. Exp. Biol.
202
,
3057
-
3067
.
Walker
,
J. A.
(
1998
).
Estimating velocities and accelerations of animal locomotion: a simulation experiment comparing numerical differentiation algorithms
.
J. Exp. Biol.
201
,
981
-
995
.
Walker
,
J.
,
Ghalambor
,
C.
,
Griset
,
O.
,
McKenney
,
D.
and
Reznick
,
D.
(
2005
).
Do faster starts increase the probability of evading predators?
Funct. Ecol.
19
,
808
-
815
.
Webb
,
P. W.
(
1975
).
Acceleration performance of rainbow trout Salmo gairdneri and green sunfish Lepomis cyanellus
.
J. Exp. Biol.
63
,
451
-
465
.
Webb
,
P. W.
(
1976
).
Effect of size on fast-start performance of rainbow-trout Salmo gairdneri, and a consideration of piscivorous predator–prey interactions
.
J. Exp. Biol.
65
,
157
-
177
.
Webb
,
P. W.
(
1978
).
Fast-start performance and body form in seven species of teleost fish
.
J. Exp. Biol.
74
,
211
-
226
.
Webb
,
P. W.
(
1980
).
Does schooling reduce fast-start response latencies in teleosts?
Comp. Biochem. Physiol. A
65
,
231
-
234
.
Webb
,
P. W.
(
1986
).
Body and fin form and strike tactics of four teleost predators attacking fathead minnow (Pimephales promelas) prey
.
Can. J. Fish. Aquat. Sci.
41
,
157
-
165
.
Webb
,
P.
and
Skadsen
,
J.
(
1980
).
Strike tactics of Esox
.
Can. J. Zool.
58
,
1462
-
1469
.
Webb
,
P. W.
and
Zhang
,
H.
(
1994
).
The relationship between responsiveness and elusiveness of heat-shocked goldfish (Carassius auratus) to attacks by rainbow trout (Oncorhynchus mykiss)
.
Can. J. Zool.
72
,
423
-
426
.
Weihs
,
D.
(
1973
).
The mechanism of rapid starting of slender fish
.
Biorheology
10
,
343
-
350
.

Competing interests

The authors declare no competing or financial interests.

Supplementary information