The role of substrate holding in achieving critical swimming speeds: a case study using the invasive round goby (Neogobius melanostomus)


The swimming performance of fishes has generally been assessed using a stepped velocity test where the speed at fatigue is considered the critical swimming performance (U crit). Although this test was designed for fishes that swim in the water column, it has been applied to fishes that adhere to the substrate. Here we examined the extent to which substrate holding, slipping and swimming contributed to reaching U crit in an example substrate holding fish, the invasive round goby. A linear model indicated that each behavior contributed significantly to U crit, but that substrate holding was by far the biggest contributor (65.8 ± 3.9 % vs. 5.8 ± 0.9 and 28.4 ± 3.4 % slipping and swimming). We also used our behavioural analysis to determine the critical substrate holding speed (U hold: 28.6 ± 1.1 cm s−1). We conclude that the U crit test can be applied to substrate holding fish but that it is not just an indication of critical swimming speed as is often considered and must be interpreted with caution.


For half a century, the prevailing method for assessing the swimming performance of fishes has been the critical swimming performance (U crit) test (Brett 1964; Hammer 1995; Plaut 2001; Nelson et al. 2002; Kieffer 2010). This is a stepped velocity test, where fishes are brought to fatigue through incremental increases in water flow speed. For fishes that are active swimmers, such as the salmonids for which the test was pioneered (Brett 1964), the U crit test can reliably assess prolonged swimming performance (Beamish 1978), providing certain conditions are met (Nelson et al. 2002). For fishes that employ alternate strategies to advance or maintain their position in flowing water, this test may be of uncertain value, especially if interpreted in a metabolic context. Some fishes have adaptations that allow them to cling to the substrate (bottom) and swim intermittently to change position (Schoenfuss and Blob 2003; Blob et al. 2007; Deslauriers and Kieffer 2011; Deslauriers and Kieffer 2012). In these fishes, their point of fatigue will relate to flow speed, but it will result from a combination of energy exerted in substrate holding and swimming.

Despite being designed for active swimming species, U crit tests are also often carried out on species that are not obligate swimmers, such as flatfishes [e.g., a plaice (Priede and Holliday 1980), European flounder (Platichthys flesus), common dab (Limanda limanda), and lemon sole (Microstomus kitt) (Duthie 1982)), as well as shortnose sturgeon (Acipenser brevirostrum) (Deslauriers and Kieffer 2011; Deslauriers and Kieffer 2012), and the round goby (Neogobius melanostomus) (Hoover et al. 2003; Tierney et al. 2011). Additionally, of those fish that do routinely swim to fatigue, some may brake against substrate, such as Atlantic salmon (Salmo salar) (Keenleyside and Yamamoto 1962; Arnold et al. 1991). While U crit test conditions may be adapted to restrict substrate holding — e.g. Priede and Holliday (1980) inclined the swim tunnel forward to limit substrate holding — a better solution may be in parsing out the contributions that substrate holding and swimming make to U crit. Conceivably U crit could be described as the result of various behaviors that enable the fish to remain within water flow, and thus give estimates of energetic and evolutionary constraints on station or substrate holding.

In the current study, we used the round goby as an example species to help clarify the role of substrate holding and swimming in reaching U crit. Gobies are small fishes that count in their family (Gobiidae) some of the most remarkable and uniquely adapted athletes. For example, three species of Hawaiian gobies (Lentipes concolor, Awaous guamensis and Sicyopterus stimpsoni) may climb waterfalls up to 350 m high to reach their mating grounds, in part by using fused pelvic fins that function as a sucker (Schoenfuss and Blob 2003; Blob et al. 2007). This fin adaptation is conserved across Gobiidae and may play an important role in the dispersal of round goby in parts of Europe and North America where it is an invasive species (Jude et al. 1992; Charlebois et al. 1997). While they are adept at holding against the substrate, they do not appear to be outstanding sustained swimmers, as they reach U crit speeds far lower than other species at similar size [round goby U crit values range from 1.9–3.2 body lengths per second (BL/s) (Hoover et al. 2003; Tierney et al. 2011), vs. salmonids, which can exceed six BL/s (Ralph et al. 2012)]. An understanding of gobies ability to advance or maintain their position through challenging water flows is necessary for the design of hydrologic barriers to limit their spread (Hoover et al. 2003; Tierney et al. 2011).

Materials and methods


Gobies were caught by angling using a barbless hook in the Detroit River and were 16.9 ± 1.9 g in mass and 11.2 ± 0.4 cm total length at the time of testing (n = 23). Gobies were held for no less than 6 months in 20–22 °C filtered, dechlorinated municipal water aquaria under an 8:00:20:00 light dark cycle and were fed fish flakes and pellets (Wardley Essentials; Hartz, Secaucus, NJ). Any hook injuries would have been healed by the time of testing. Adult gobies were selected for testing based on similarity of size (juveniles were not tested), and were fasted for at least 24 h prior to swim tests. Experiments were approved by the University of Windsor’s animal care committee (AUPP # 07–01).

Swimming performance assessment

This is a follow-up study to Tierney et al. (2011). In the current study, we analyzed video of the round gobies taken during the U crit tests of the previous study and related their U crit to their substrate holding behaviors. In brief, U crit was measured using a modified 140 L Brett-type respirometer that had a 1 m long square tube with internal dimensions 10 × 10 cm as the test section. The rear gate of this section was constructed of stainless steel wire and a small voltage (≤ 5 V) was applied throughout the test to prevent resting against the rear of the flume. Water flow was driven by a 14 cm diameter ducted propeller (Bombardier, Montréal, QE) coupled to a 3 hp. digitally controlled variable speed electric motor (Marathon motor with a Lenze SMVector controller; Moncur Electric, Windsor, ON). Water speed was determined at the rear of the center tube (i.e. ahead of the downstream electrified gait), using a current meter (OTT, Kempten, Germany). Surveillance cameras (high resolution SX-920C-HR, 480 × 640 pixel; Matco; QE) were mounted underneath and to the side of the swimming chamber and video was recorded. Gobies were acclimated to the minimum flow (17.9 cm/s) and then brought through stepped increases in flow until they were unable to remain off the rear electrified grid for 5 s. The step height (7.54 cm/s) was chosen such that ten steps from the minimum flow would take the gobies to a speed approximately equal to one-half of the maximum burst speed observed earlier in static trials on separate gobies. Step height was held constant for all gobies, since swimming speed appeared independent of body size in the static trials (Tierney et al. 2011), and since gobies were of reasonably similar size. Step length (10 min) was chosen to capture the ability to maintain position in flow for a prolonged period (Farrell 2008). U crit was calculated as in Brett (1964) as U crit = Vf + Vi × (tfi / ti), where Vf is the speed of the last fully completed step, Vi is the speed increase of each step, tfi is the amount of time completed on the last step, and ti is the duration of each step.

Swimming behaviors were visually scored over the course of the middle (5–6 min) and end (8–9 min) minutes of each step. No observations were taken earlier in the steps to allow the gobies time to adjust their behaviour to the new water speed. During their swims, gobies exhibited one of three actions: ‘holding’, ‘sliding’ or ‘swimming’. Holding was apparent in an absence of ground movement; sliding was apparent in a substrate-associated rearward slide not involving caudal fin use; swimming was independent of the substrate and involved caudal fin use. These behaviours were scored continuously during the observation periods. For a subset of gobies (N = 11), the frequency of bursts was counted over 1 min at the 5th min of each step. A burst was considered a period of forward progress at a ground speed greater than 5 cm s−1 at the step velocity. Burst swimming (as opposed to steady swimming) was identified as a period of swimming in which there were more than three bursts per min, as in previous studies (MacNutt et al. 2006). The critical substrate holding velocity (U hold) was considered as the last flow speed at which substrate holding accounted for >50 % of the total activity. As substrate holding was the primary behaviour carried out by gobies at low water velocities, the term U hold is used to represent the speed at which gobies transition from primarily (>50 % of the time) substrate holding to primarily using other behaviours to maintain their position.

As the proportions of the behaviors did not differ between the sampling times (5th and 8th minutes; holding: F1,21 = 0.737, P = 0.411; slipping: F1,21 = 1.493, P = 0.250; swimming: F1,21 = 0.439, P = 0.522; two-way repeated measures analysis of variance using sampling time × step number), they were averaged. The duration of time spent carrying out each behavior during the U crit test was estimated by multiplying the proportion of time spent on each behavior at each velocity step by 10 min (step duration) and summing the values.

Data analysis

When investigating the proportion of time spent holding, slipping and swimming at each water velocity, velocity was considered as percent U crit rather than an absolute measure in order to better depict the contribution each behaviour made to an individual’s U crit value. To describe these relationships, sigmoidal, peak (Gaussian), and polynomial models were fitted to data of each behaviour over the course of the U crit test (Fig. 1) and a model was selected based on the lowest residual sums of squares; however, a model curve was only considered if all coefficients were statistically significant. To model the contribution of these behaviours to reaching U crit, a linear regression was performed between U crit and the estimated total duration spent conducting each behaviour during the entire test. The relative importance of each term to the overall model was calculated as the decomposed R2 using the Lindeman, Merenda and Gold method in the “relaimpo” package (Grömping 2006) in R (Core Team 2013).

Fig. 1

The utilization of substrate holding, slipping and swimming behavior during a critical swimming performance (U crit) test by individual round gobies (grey lines). The black curve indicates the summary sigmoidal (substrate holding and swimming) or Gaussian (slipping) curve and the blue lines show the 95 % confidence intervals. Water velocity was considered as %U crit as we were interested in the behaviors that occurred up to fatigue, and not the absolute U crit values, which were reported in Tierney et al. (2011). Curve equations: holding =0.997 / (1 + e(−(x - 86.0)/−8.91)) (R2 = 0.5036, F2,173 = 86.7320, P < 0.0001); slipping =0.121 × e (−.5 × ((x - 96.4)/15.0)2) (F2,173 = 25.8032, P < 0.0001); swimming =1.06 / (1 + e(−(x-91.6)/10.7)) (F2,173 = 87.5878, P < 0.0001); x = water velocity in %U crit; N = 23

A paired t-test was used to assess the difference between U crit and U hold. U hold was expressed as an absolute measure of velocity (cm s−1) because it is intended to be an absolute measure of substrate holding ability.

Linear regressions were performed between U crit and fish length to determine if length was a significant covariate of either term. Because the regression was not significant (R2 < 0.01, F1,21 = 0.07 , p = 0.79), length was not included as a covariate in any other analyses. All data analyses and presentation were completed using SigmaPlot 11 (Systat, San Jose, CA) except for the relative importance analysis in the swimming behavior-U crit model. All summary statistics are presented as mean ± standard error of the mean (SEM), and significance was accepted at α = 0.05.

Results and discussion

An example substrate holding fish, the round goby, demonstrated that U crit can be measured (34.8 ± 1.0 cm s−1), but that it was not only a measure of swimming ability as it is commonly considered. During U crit trials, the proportion of time gobies spent holding at low flow speeds was ~1, but this decreased in a threshold manner as flows increased (Fig. 1A). On average, the last flow speed at which gobies were able to hold the substrate for a majority of the time (U hold) was 28.6 ± 1.1 cm s−1. The U hold speed was lower than U crit (−18 %; t22 = 7.0, p < 0.001), indicating that other behaviors contributed to reaching U crit. Once gobies were no longer able to hold the substrate, swimming became their primary behavior (Fig. 1C). During the transition from primarily holding to primarily swimming, gobies would attempt substrate holding, but would slip backward and would burst forward and attempt to bottom hold again. This resulted in a peak in slipping behavior that corresponded with reduced time substrate holding and increased time swimming (Fig. 1B). When swimming became the primary behavior, gobies utilized a burst-and-coast gait, with greater than three bursts per min (Fig. 2). With respect to the total duration of the U crit test, substrate holding, slipping, and swimming on average accounted for 65.8 ± 3.9, 5.8 ± 0.9, and 28.4 ± 3.4 %, respectively. The duration of time spent utilizing each of these activities explained significant portions of U crit variance in a linear regression model (Table 1).

Fig. 2

The frequency of bursting events in relation to the proportion of time spent swimming during each velocity step in the U crit test. The utilization of a burst-and-glide mode of swimming is illustrated by points above the horizontal grey line at a burst frequency of three. N = 11 gobies; total observations =99

Table 1 A linear regression model of round goby critical swimming performance (U crit) as a function of the duration (min) of time spent substrate holding, slipping, and swimming during a U crit test

In two previous studies using U crit as an endpoint in this example substrate holding fish, it was suggested that U crit could be considered a measure of ‘critical station holding ability’ (Hoover et al. 2003; Tierney et al. 2011), based on the observation that is was the primary behavior utilized during the test. Here, however, we used a more detailed characterization of behavior during U crit tests to show that that while substrate holding ability contributes greatly to reaching U crit, partial substrate holding (slipping) and swimming also play important roles.

For fishes, a U crit value will reflect the water speed that a position can be maintained for a prolonged period, generally accepted to range between 20 s and 200 min (Beamish 1978), but it may not necessarily represent the critical ‘swimming’ velocity. Different fish species achieve U crit using different strategies, and as such U crit may be an indicator of the use of different swimming and holding abilities. Furthermore, the nature of the testing apparatus may influence the ability of a fish to carry out these strategies. For example, in salmonids tested in a swim tunnel that is of insufficient length to allow for unsteady, burst-and-coast swimming, U crit may be an indicator of gait transition speed (from steady to unsteady swimming) (Peake and Farrell 2006; Peake 2008). However, with a sufficiently long swim tunnel, fish may use burst-and-coast swimming, and so U crit, may be an indicator of prolonged aerobic and anaerobic swimming ability (Tudorache et al. 2007). The present study supports that U crit may be used in at least some substrate holding fishes to capture the proportions that substrate holding and swimming contribute to reaching critical water speeds.

Individual variation

Our results suggest that substrate holding ability was largely responsible for the round goby reaching U crit; however, we also found that there was substantial individual variation in the strategies used to achieve U crit. Some individuals rarely utilized swimming to reach U crit and relied almost exclusively on substrate holding (e.g. U hold = 98.3 % U crit), whereas others relied much more on burst-and-coast swimming to reach U crit (e.g. U hold = 60.7 % U crit). Whether this is a physiological, morphological or behavioural phenomenon remains to be investigated. The relevant message is that swimming ability can vary considerably across fish and cannot be ignored in setting flow barriers, or interpreting the physiological meaning of U crit in substrate holding fishes.


Using the invasive round goby as an example substrate holding fish, U crit was a combined measure of substrate holding and burst swimming abilities. If critical water velocity data were to be used to construct flow barriers, then further research is also needed on the effects of substrate on holding ability. In the present study, the substrate was arguably the most attachable substrate (Plexiglas). The expectation is that as substrate becomes coarser, holding ability decreases. However, a previous study noted that two species of gobies that burst swam up a waterfall vs. another that inched up, actually swam faster on a courser surface (Blob et al. 2006). In very course substrate, objects will provide refugia, where energy reserves may be replenished (gobies do prefer courser substratum; Young et al. 2010), which could enable longer migrations. Future research would benefit from the use of automated behaviour tacking software such as EthoVision (Noldus, Leesburg, VA), to more rapidly and precisely characterize substrate holding, slipping, sustained swimming, and bursting events.


  1. Arnold GP, Webb PW, Holford BH (1991) Short communication: the role of the pectoral fins in station-holding of Atlantic salmon parr (Salmo salar L.). J Exp Biol 156:625–629

    Google Scholar 

  2. Beamish FWH (1978) Swimming capacity. In: Hoar WS, Randall DJ (eds) Fish physiology, vol 7. Academic Press Inc., New York, pp. 101–187

    Google Scholar 

  3. Blob RW, Rai R, Julius ML, Schoenfuss HL (2006) Functional diversity in extreme environments: effects of locomotor style and substrate texture on the waterfall-climbing performance of Hawaiian gobiid fishes. J Zool 268:315–324

    Article  Google Scholar 

  4. Blob RW et al. (2007) Ontogenetic change in novel functions: waterfall climbing in adult Hawaiian gobiid fishes. J Zool 273:200–209

    Article  Google Scholar 

  5. Brett JR (1964) The respiratory metabolism and swimming performance of young sockeye salmon. J Fish Res Board Can 21:1183–1226

    Article  Google Scholar 

  6. Charlebois PM, Marsden JE, Goettel RG, Wolfe RK, Jude DJ, Rudnika S (1997) The round goby, Neogobius melanostomus (Pallas), a review of European and North American literature. Illinois-Indiana Sea Grant Program and Illinois Natural History Survey. INHS Special Publication No. 20, Illinois, p 76

  7. Deslauriers D, Kieffer JD (2011) The influence of flume length and group size on swimming performance in shortnose sturgeon Acipenser brevirostrum. J Fish Biol 79:1146–1155

    CAS  Article  PubMed  Google Scholar 

  8. Deslauriers D, Kieffer JD (2012) Swimming performance and behaviour of young-of-the-year shortnose sturgeon (Acipenser brevirostrum) under fixed and increased velocity swimming tests. Can J Zool 90:345–351

    Article  Google Scholar 

  9. Duthie GG (1982) The respiratory metabolism of temperature-adapted flatfish at rest and during swimming activity and the use of anaerobic metabolism at moderate swimming speeds. J Exp Biol 97:359–373

    CAS  PubMed  Google Scholar 

  10. Farrell AP (2008) Comparisons of swimming performance in rainbow trout using constant acceleration and critical swimming speed tests. J Fish Biol 72:693–710

    Article  Google Scholar 

  11. Grömping U (2006) Relative importance for linear regression in R: the package relaimpo. J Stat Soft 17:1–27

    Article  Google Scholar 

  12. Hammer CH (1995) Fatigue and exercise tests with fish. Comp Bioch Physiol 112A:1–20

    CAS  Article  Google Scholar 

  13. Hoover JJ, Adams SR, Killgore KJ (2003) Can hydraulic barriers stop the spread of the round goby? U.S. Army Corps of Engineers, U.S. Army Engineer Research and Development Center (ERDC), Vicksburg, MS., p 1–8

  14. Jude DJ, Reider RH, Smith GR (1992) Establishment of Gobiidae in the Great-Lakes basin. Can J Fish Aquat Sci 49:416–421

    Article  Google Scholar 

  15. Keenleyside MHA, Yamamoto FT (1962) Territorial behaviour of juvenile Atlantic salmon (Salmo salar L.). Behaviour 19:139–169

    Article  Google Scholar 

  16. Kieffer JD (2010) Perspective - exercise in fish: 50+ years and going strong. Comp Biochem Physiol 156A:163–168

    CAS  Article  Google Scholar 

  17. MacNutt MJ et al. (2006) Temperature effects on swimming performance, energetics, and aerobic capacities of mature adult pink salmon (Oncorhynchus gorbuscha) compared with those of sockeye salmon (Oncorhynchus nerka). Can J Zool 84:88–97

    Article  Google Scholar 

  18. Nelson JA, Gotwalt PS, Reidy SP, Webber DM (2002) Beyond Ucrit: matching swimming performance tests to the physiological ecology of the animal, including a new fish 'drag strip'. Comp Biochem Physiol 133A:289–302

    CAS  Article  Google Scholar 

  19. Peake SJ (2008) Gait transition speed as an alternate measure of maximum aerobic capacity in fishes. J Fish Biol 72:645–655

    Article  Google Scholar 

  20. Peake SJ, Farrell AP (2006) Fatigue is a behavioural response in respirometer confined smallmouth bass. J Fish Biol 68:1742–1755

    Article  Google Scholar 

  21. Plaut I (2001) Critical swimming speed: its ecological relevance. Comp Biochem Physiol 131A:41–50

    CAS  Article  Google Scholar 

  22. Priede IG, Holliday FGT (1980) The use of a new tilting tunnel respirometer to investigate some aspects of metabolism and swimming activity of the plaice (Pleuronectes platessa L.). J Exp Biol 85:295–309

    Google Scholar 

  23. R Core Team (2013) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna

  24. Ralph AL, Berli BI, Burkhardt-Holm P, Tierney KB (2012) Variability in swimming performance and underlying physiology in rainbow trout (Oncorhynchus mykiss) and brown trout (Salmo trutta). Comp Biochem Physiol 163A:350–356

    Article  Google Scholar 

  25. Schoenfuss HL, Blob RW (2003) Kinematics of waterfall climbing in Hawaiian freshwater fishes (Gobiidae): vertical propulsion at the aquatic–terrestrial interface. J Zool 261:191–205

    Article  Google Scholar 

  26. Tierney K, Kasurak A, Zielinski B, Higgs D (2011) Swimming performance and invasion potential of the round goby. Environ Biol Fish 92:491–502

    Article  Google Scholar 

  27. Tudorache C, Viaenen P, Blust R, de Boeck G (2007) Longer flumes increase critical swimming speeds by increasing burst-glide swimming duration in carp Cyprinus carpio, L. J Fish Biol 71:1630–1638

    Article  Google Scholar 

  28. Young JAM, Marentette JR, Gross C, McDonald JI, Verma A, Marsh-Rollo SE, Macdonald PDM, Earn DJD, Balshine S (2010) Demography and substrate affinity of the round goby (Neogobius melanostomus) in Hamilton harbour. J Great Lakes Res 36:115–122

    Article  Google Scholar 

Download references


We thank Marc St. Pierre of the University of Windsor Science Technical shop for his outstanding craftsmanship of the swim tunnel respirometer. This study was funded by NSERC grants to KBT and DMH.

Author information



Corresponding author

Correspondence to Keith B. Tierney.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Gilbert, M.J.H., Barbarich, J.M., Casselman, M. et al. The role of substrate holding in achieving critical swimming speeds: a case study using the invasive round goby (Neogobius melanostomus). Environ Biol Fish 99, 793–799 (2016).

Download citation


  • Swimming performance
  • U crit
  • Substrate holding
  • Gobiidae