An interspecific comparison between morphology and swimming performance in cyprinids

Authors

  • G.-J. Yan,

    1. Laboratory of Evolutionary Physiology and Behavior, Chongqing Key Laboratory of Animal Biology, Chongqing Normal University, Chongqing, China
    Search for more papers by this author
  • X.-K. He,

    1. Laboratory of Evolutionary Physiology and Behavior, Chongqing Key Laboratory of Animal Biology, Chongqing Normal University, Chongqing, China
    Search for more papers by this author
  • Z.-D. Cao,

    1. Laboratory of Evolutionary Physiology and Behavior, Chongqing Key Laboratory of Animal Biology, Chongqing Normal University, Chongqing, China
    Search for more papers by this author
  • S.-J. Fu

    Corresponding author
    • Laboratory of Evolutionary Physiology and Behavior, Chongqing Key Laboratory of Animal Biology, Chongqing Normal University, Chongqing, China
    Search for more papers by this author

Correspondence: Shi-Jian Fu, Laboratory of Evolutionary Physiology and Behavior, Chongqing Key Laboratory of Animal Biology, Chongqing Normal University, Chongqing 400047, China. Tel.: +862365910701; fax: +862365910701;

e-mail: shijianfu9@hotmail.com

Abstract

Flow regimes are believed to be of major evolutionary significance in fish. The flow regimes inhabited by cyprinids vary extensively from still flow regimes to riptide flow regimes. To test (i) whether flow-driven swimming performance and relevant morphological differentiation are present among fish species and (ii) whether evolutionary shifts between high-flow and low-flow habitats in cyprinids are associated with evolutionary trade-offs in locomotor performance, we obtained data on both steady and unsteady swimming performance and external body shape for 19 species of cyprinids that typically occur in different flow regimes (still, intermediate and riptide). We also measured the routine energy expenditure (RMR) and maximum metabolic rate (MMR) and calculated the optimal swimming speed. Our results showed that fish species from riptide groups tend to have a higher critical swimming speed (Ucrit), maximum linear velocity (Vmax) and fineness ratio (FR) than fish from the other two groups. However, there was no correlation between the reconstructed changes in the steady and unsteady swimming performance of the 19 species. According to the phylogenetically independent contrast (PIC) method, the Ucrit was actively correlated with the MMR. These results indicated that selection will favour both higher steady and unsteady swimming performance and a more streamlined body shape in environments with high water velocities. The results suggested that steady swimming performance was more sensitive to the flow regime and that for this reason, changes in body shape resulted more from selective pressure on steady swimming performance than on unsteady swimming performance. No evolutionary trade-off was observed between steady and unsteady swimming performance, although Ucrit and MMR were found to have coevolved. However, a further analysis within each typically occurring habitat group suggested that the trade-off that may exist between steady and unsteady swimming performance may be concealed by the effect of habitat.

Introduction

The term ‘trade-off’ is usually understood to mean a necessary negative correlation between two aspects of performance or fitness under the condition that they cannot be simultaneously maximized due to competing demands on the organism (Futuyma & Moreno, 1988; Vanhooydonck et al., 2001; Roff & Fairbairn, 2007). Swimming performance in fish species can be classified as steady (commonly employed while holding station in a water current to seek favourable food items) or unsteady (commonly employed during predator avoidance and navigation in structurally complex environments) (Webb, 1984). Thus, the trade-off between the two primary swimming modes, steady and unsteady swimming, is presumed to exist in fish species employing relatively coupled locomotor systems (i.e. the same morphological structures are used for propulsion during both steady and unsteady swimming), as is the case in almost all of the cyprinids (Webb, 1984; Domenici, 2003; Langerhans, 2006). This general trade-off has long been hypothesized to play an important role in the ecology and evolution of fish (Webb, 1984; Domenici, 2003; Langerhans, 2006). Many ecological and evolutionary studies have sought evidence of evolutionary trade-offs between pairs of traits among birds and amphibians (Badyaev et al., 2002; Arendt, 2003; Poulin & Mouillot, 2004). Stephens & Wiens (2008) has defined ‘evolutionary trade-off’ as an evolutionary increase in one aspect of performance or fitness (relative to the ancestral state) associated with an evolutionary decrease in a related aspect of performance or fitness on the same branch of a phylogeny. Most research on the ecology and evolution of fish has centred on features of fish design rather than aspects of locomotor performance (Langerhans & Reznick, 2010). To our knowledge, no previous study has investigated the evolutionary trade-off between steady and unsteady swimming performance in fish species.

Flow regimes are generally believed to be of major evolutionary significance in fish. The flow regimes that fish inhabit vary extensively, from the low-flow regimes of ponds, lakes, backwaters and calm tide pools to the high-flow regimes of swift streams, rapid rivers and wave-swept near-surface oceanic waters (Langerhans, 2009). In high-flow environments, fish need a high steady swimming performance to maintain their position and perform routine tasks. It could be achieved by either high respiratory capacity as usually suggested by maximum metabolic rate or streamlined body and hence low swimming cost during steady locomotion (see the next paragraph). In low-velocity environments, depending on complex features of the habitat (e.g. predator evasion, prey capture and structurally complex environments), fish require an enhanced ability to perform unsteady locomotion. The structure-driven divergence of locomotor performance has been implicated in numerous cases of intraspecific differentiation (Ehlinger & Wilson, 1988; Robinson & Wilson, 1994; Smith & Skúlason, 1996; Hendry et al., 2002; McKinnon & Rundle, 2002), but only a few studies of interspecific differentiation have been performed on the basis of a compilation of available data sets (Domenici, 2003). Thus, the aim of this study was to investigate the variation in swimming performance among fish species that prefer habitats with different flow regimes and moreover, if there exists variation in swimming performance, to determine whether evolutionary shifts between high-flow and low-flow habitats in cyprinids are associated with evolutionary trade-offs in locomotor performance.

Because both steady swimming performance and unsteady swimming performance in fish species are important for survival in the face of natural selection, it appears that fish could simply optimize both steady and unsteady capacities to solve many of their problems. However, constraints are inevitable, and no form exists that can simultaneously optimize both swimming modes. From a hydromechanical perspective, fish face a dilemma in the form of a functional trade-off, that is, the optimal functional design for steady swimming performance reflects those features that maximize thrust and minimize drag, whereas the optimal functional design for unsteady swimming performance reflects those features that maximize thrust and stability. The features that maximize thrust and minimize drag include a high aspect-ratio lunate caudal fin, a narrow caudal peduncle and a large anterior body depth, whereas the maximum thrust and stability are typically produced by a deep body and a large low aspect-ratio caudal fin (Webb, 1984; Domenici et al., 2008). In this context, it appears that natural selection may favour different locomotor modes in different flow regimes and, hence, demand appropriately matched morphological characteristics (Peres-Neto & Magnan, 2004). Thus, another aim of this study was to investigate the variation in morphology among fish species that prefer different habitat flow regimes and its relationship between swimming performance.

The cyprinids are one of the largest families of vertebrates in the world. The family has a wide geographical distribution, including mainland Eurasia, Japan, the East Indies, Africa and North America. There are approximately 532 species of cyprinids within approximately 132 genera in China, and their phylogenetic relationships are well documented (Wang, 2005). Due to the benefits resulting from considerable morphological and physiological diversity, cyprinids exploit a wide variety of habitats (Howes, 1991). The empirical support for the role of water flow in fish evolution is nontrivial (Langerhans et al., 2003; McGuigan et al., 2003; Hendry et al., 2006). For this reason, we will focus on water flow as the common selective agent for fish species in this study, and our predictions will focus on the divergent evolution of morphology and locomotion based on hypotheses regarding the form of natural selection on swimming performance in different water flow regimes. For this study, we obtained data on the locomotor performance, maximum metabolic rate, metabolic cost during swimming and morphology of 19 species of cyprinids with different typically occurring habitats (see Table 1 for details) to test whether flow-driven swimming performance and relevant morphological differentiation are present among fish species and whether evolutionary shifts between high-flow and low-flow habitats in cyprinids are associated with evolutionary trade-offs in locomotor performance. This analysis will be based on both conventional and a phylogenetic approach.

Table 1. The classificatory information of the fish species studied
SpeciesGroupDistributionCollected cites
DataaInvestigationbHabitatVelocity
  1. a

    The data were The Fishes of Sichuan, China (Ding, 1994). L: lake, P: pond, R: River, B: backwater or slow flow water area of river, M: mountain steam.

  2. b

    Our investigation on local fish catchment and information from local fishermen. The number in brackets is the water velocity (m s−1).

  3. c

    The fish species were bought from local fishermen (other fish species were collected by hook-and-line angling).

Chinese false gudgeonStillL, BL (0)L0
Small fat minnowStillL, BB (< 0.2)B< 0.2
Topmouth gudgeonStillP, LP, L (0)L0
Chinese bitterlingStillP, L, BL (0)L0
Rose bitterlingStillL, BB (< 0.2)B<0.2
Bighead carpcStillL, RL (0)L0
Silver carpcIntermediateL, RL, R (0–2)L0
Grass carpcIntermediateL, RL, R (0–3)L0
Black carpcIntermediateL, RL, R (0–2)L0
Chinese breamcIntermediateL, RL, R (0–2)L0
Wild carpIntermediateL, RL, B (0–2)L0
Chinese hook snout carpRiptideR, MR, M (1.5–4)R1.5–4
Pale chubRiptideR, MR, M (1.5–4)M1.5–4
QingbocRiptideRR (1.5–4)R1.5–4
Sharp-jaw barbelcRiptideRR (1.5–4)R1.5–4
Mountain carpsRiptideM, RM (1.5–4)M2–4
Rock carpcIntermediateBR (0–2)B0–1
Crucian carpcStillP, L, BP, L, R (0–1)L0
Common carpcIntermediateP, L, RP, L, R (0–3)L0

Materials and methods

Experimental animals and holding conditions

All fish species were collected from Wu River water system. The bighead carp Aristichthys nobilis, crucian carp Carassius auratus, black carp Mylopharyngodon piceus, rock carp Procypris rabaudi, common carp Cyprinus carpio, qingbo Spinibarbus sinensis, sharp-jaw barbel Onychostoma sima, Chinese bream Parabramis pekinensis, grass carp Ctenopharyngodon idellus and silver carp Hypophthalmichthys molitrix were collected from local fishermen. The Chinese false gudgeon Abbottina rivularis, small fat minnow Sarcocheilichys parvus, topmouth gudgeon Pseudorasbora parva, rose bitterling Rhodeus ocellatus, Chinese bitterling Rhodeus sinensis, mountain carp Schizothorax wangchiachii, Chinese hook snout carp Opsariichthys bidens, pale chub Zacco platypus and wild carp Hemiculter leucisculus (Fig. 1) were collected by hook-and-line angling (Table 2). Based on our investigation on local fishery catchment, water velocity where we (or local fishermen) collected the fish and published reference, we classified the 19 species into three groups, that is, still, intermediate and riptide (Table 1). All experiments were conducted according to the Guidelines on the Humane Treatment of Laboratory Animals established by the Ministry of Science and Technology of the People's Republic of China. During the season when the measurements were taken, the water temperature in the field was approximately 25 °C (23–27 °C). Individuals of each fish species obtained from the local fisherman were selected and introduced into an indoor 250-L recirculating rearing system filled with well-aerated tap water at a temperature of 25 °C for at least 1 week. During the holding period, the fish were fed daily to satiation with a commercial diet. For the fish collected directly from the field, individuals were maintained in a 40-L portable holding tank for 48 h at 25 °C after capture before any measurements were taken.

Table 2. The body length and body mass of the fish species studied
TaxonNHabitat typeBody length (cm)Body mass (g)
Steady swimmingUnsteady swimmingSteady swimmingUnsteady swimmingSteady swimmingUnsteady swimming
Chinese false gudgeon66Still10.3 ± 0.310.3 ± 0.315.9 ± 1.715.9 ± 1.7
Small fat minnow87Still5.86 ± 0.175.88 ± 0.413.05 ± 0.263.29 ± 0.77
Topmouth gudgeon79Still5.88 ± 0.415.86 ± 0.163.29 ± 0.773.05 ± 0.25
Chinese bitterling78Still7.61 ± 0.237.58 ± 0.208.94 ± 0.808.87 ± 0.69
Rose bitterling66Still5.28 ± 0.175.28 ± 0.173.93 ± 0.343.93 ± 0.34
Bighead carp88Still6.93 ± 0.116.80 ± 0.306.19 ± 0.276.20 ± 0.41
Silver carp88Intermediate7.45 ± 0.117.44 ± 0.106.09 ± 0.246.29 ± 0.35
Grass carp88Intermediate5.51 ± 0.144.83 ± 0.162.66 ± 0.192.13 ± 0.16
Black carp78Intermediate7.89 ± 0.167.14 ± 0.267.05 ± 0.476.41 ± 0.21
Chinese bream84Intermediate6.41 ± 0.135.25 ± 0.104.49 ± 0.232.50 ± 0.17
Wild carp88Intermediate11.6 ± 0.411.6 ± 0.417.4 ± 1.717.4 ± 1.7
Chinese hook snout carp77Riptide9.22 ± 0.399.04 ± 0.3911.7 ± 1.611.0 ± 1.6
Pale chub66Riptide8.85 ± 0.328.85 ± 0.3210.75 ± 1.0810.75 ± 1.08
Qingbo88Riptide6.30 ± 0.105.24 ± 0.084.80 ± 0.312.85 ± 0.14
Sharp-jaw barbel78Riptide5.49 ± 0.115.41 ± 0.162.74 ± 0.182.65 ± 0.26
Mountain carps79Riptide8.77 ± 0.249.13 ± 0.268.59 ± 0.709.81 ± 0.89
Rock carp88Intermediate6.28 ± 0.185.93 ± 0.143.64 ± 0.333.29 ± 0.29
Crucian carp68Still5.95 ± 0.255.41 ± 0.246.22 ± 0.565.35 ± 0.44
Common carp88Intermediate6.10 ± 0.635.61 ± 0.075.15 ± 0.054.83 ± 0.08
Figure 1.

Chart of the phylogenetic relationships extracted and summarized from published studies (Wang, 2005), with morphological information for the fish species investigated in this study.

Measurement of unsteady locomotor performance (fast-start)

After at least 48 h of acclimation to the experimental conditions, the fish were slightly anaesthetized with neutralized tricaine methane sulphonate (MS – 222, 50 mg L−1), and a small white plastic ball (diameter: 1 mm) was attached to the dorsal side of each fish by a suture at the centre of mass (CM) position, as determined by a pilot experiment (He et al., 2011). The fish were individually transferred to an experimental glass tank (40 × 40 × 15 cm) 1 h prior to being startled. Reference grids, 1 cm square, were attached to the floor of the tank. The sides of the experimental tank were covered with black paper to prevent the fish from seeing the approaching stimulus. The water in the tank was 10 cm deep. After the fish had been allowed to acclimate for 1 h, an escape response was elicited by an electrical impulse (0.5 V cm−1; 30 ms, as determined by a previous study by He et al., 2011). The impulse was manually triggered when the fish was in the holding position at the centre of the filming zone. A high-speed camera (Basler A504K; 500 frame s−1) was used to record the entire time course of the procedure (time span: 3 s). The recording was initiated as soon as the LED (synchronized with the electrical stimulus) was illuminated (0 ms).

After the screen was finished, the images were digitized with TpsUnil (Rohlf, 2004) and TpsDig (Rohlf, 2005) software (http://life.bio.sunysb.edu/morph). The CM locomotion track of the fish fast-start movement was then obtained. The following parameters were calculated based on the CM locomotion track: the maximum linear velocity (Vmax), the escape distance during the initial 120 ms (S120 ms) and the maximum linear acceleration (Amax). The Vmax was calculated from the maximum distance moved by the CM in 2 ms (two consecutive video frames), the Amax was calculated from the derivative of Vmax, and the S120 ms was calculated from the distance the fish moved during the first 120 ms.

Measurement of steady locomotor performance

Critical swimming speed (Ucrit) and swimming oxygen consumption rate (math formulaO2)

After at least 48 h of holding under the experimental conditions, the fish were transferred individually into a Brett-type swimming tunnel respirometer to measure the Ucrit and the math formulaO2 (Brett, 1964). The specifications of the respirometer are described in detail in previous papers (Fu et al., 2009, 2011; Pang et al., 2011a,b). To eliminate the effect of handling stress, the fish were introduced into the water tunnel individually and left in the water at a low water velocity (6 cm s−1) for 8 h. The water velocity was then increased in 6 cm s−1 increments from the initial value of 6 cm s−1 every 20 min until the fish became fatigued. Fatigue was defined by the time at which the fish failed to move from the rear honeycomb screen of the swimming chamber for 20 s (Lee et al., 2003). The Ucrit was calculated for individual fish using Brett's equation (Brett, 1964):

display math(1)

where V is the highest speed at which the fish swam for the full time period (cm s−1), ΔV is the velocity increment, T is the prescribed period of swimming per speed, and t is the time that the fish swam at the final speed (min). The swim tunnel was designed to switch between a closed mode and an open mode, the former for respirometry and the latter to replenish oxygen levels. In the closed mode, a small volume of water was drawn from the sealed respirometer by a peristaltic pump, forced past a dissolved oxygen probe housed in a sealed temperature-controlled chamber, and then returned to the respirometer. The oxygen concentration (mg L−1) was recorded once every 2 min. The oxygen consumption rate (math formulaO2, mg kg−1 h−1) of individual fish while swimming was calculated from the depletion of oxygen according to the equation:

display math(2)

where St and S0 (mg L−1 min−1) represent the decrease in the water's dissolved oxygen content in the pipe per minute with and without fish, respectively, obtained from the linear regressions between time (min) and dissolved oxygen content (mg L−1); 3.45 is the volume (L) of the respirometer; and m is the body mass (kg) of the fish. The math formulaO2 was adjusted to a standard body mass of 1 kg by a body mass coefficient of 0.75 (Reidy et al., 2000; Fu et al., 2005). The maximal math formulaO2 during the Ucrit test was defined as the maximum metabolic rate and used as an indicator of respiratory capacity (math formulaO2 max).

Routine energy expenditure (RMR), optimal swimming speed (Uopt) and swimming efficiency coefficient (E)

For each fish species, the relationship between the math formulaO2 (y) and swimming speed (x) can generally be described by the equation

display math(3)

where a is the corresponding math formulaO2 at zero swimming speed calculated from equation 2, that is, the RMR; b is a constant, which can be used as an index of swimming efficiency, that is, the higher the b value, the more marked the increase in the swimming math formulaO2 with increased swimming speed. The optimal swimming speed (Uopt) was calculated according to the following equation (Weihs, 1973):

display math(4)

Morphometrics

The fish were euthanized with an overdose of MS-222 and pinned next to a ruler on a white polystyrene board. A digital photograph of the left side of each specimen next to the ruler was taken with a digital camera and then analysed with the thin plate spine (tps) software package (http://life.bio.sunysb.edu/morph). We used tpsDig and tpsUtil for data acquisition and editing. After first concatenating all photographs into a single file, we placed landmarks on 13 morphological features in each image (Fig. 2). We then used tpsReg (Rohlf, 2003) to perform a regression analysis and correlate the morphological differences with the independent variables, that is, the typically occurring habitats, the Ucrit value and the Vmax value. Using the image, we also measured the following morphological parameters: the body length (L), body depth (H), caudal fin depth (L1), the area of the caudal fin (S1), caudal peduncle (S2), anal fin (S3), dorsal fin (S4) and lateral body (S). We then calculated the fineness ratio (L/H), the ratios of these fin and caudal peduncle areas to the lateral body (Sx/S) and the aspect ratios of the caudal fin (L12/S1) (Fig. 2).

Figure 2.

Positions of the thirteen landmarks used for the examination of lateral body shape. (1) the tip of the premaxilla, (2) the point where the operculum intersects the dorsal body outline, (3–4) the points where the anterior- and posteriormost edges of the dorsal fin contact the outline of the body, (5) the beginning of the caudal fin at the dorsal outline, (6) the posteriormost edge of the upper caudal fin, (7) the fork of the caudal fin, (8) the posteriormost edges of the lower caudal fin, (9) the beginning of the caudal fin at the ventral outline, (10–11) the points where the posterior- and anteriormost edges of the anal fin contact the outline of the body, (12) the anteriormost edge of the right pelvic fin, 13) the point where the operculum intersects the ventral body outline. Using the landmarks in the image, we measured the body length (L), body depth (H, 3–12), caudal fin depth (L1, 6–8), the area of the caudal fin (S1), caudal peduncle (S2), anal fin (S3) and dorsal fin (S4). We then calculated the fineness ratio (L/H), the ratio of each area to the lateral body area (Sx/S) and the aspect ratio of the caudal fin (L12/S1).

Data analysis and statistics

We conducted both standard and phylogenetic anovas to test for differences in Ucrit and fast-start swimming performance, the fineness ratio and MMR variables between species inhabiting environments of varying water velocity. When performing standard analysis, the body length was used as the covariate for swimming performance parameters and body mass was used as the covariate for metabolic parameters. We used a two-way ancova to test whether variation in the math formulaO2 was related to swimming speed and habitat using body mass as the covariate. We used the pdsimul and pdanova programs (Garland et al., 1993) to perform phylogenetic anovas. Using these programs, we simulated trait evolution as Brownian motion with the means and variances of the simulations set to the means and variances of the original data. We performed 1000 simulations, producing a null distribution of F statistics against which the F value from the actual data could be compared to assess statistical significance (i.e. to determine how different the observed patterns were from that expected via genetic drift alone). We constructed a best-estimate phylogenetic hypothesis for this group of species based on previous morphological and molecular studies (Fig. 1). All branch lengths were set equal to one.

We also used both Pearson's correlation and a phylogenetically independent contrast (PIC) (PDTREE) to test the correlation of Ucrit with MMR and Vmax. The PIC for each variable was calculated with all branch lengths set to a value of one. The regression between morphological and other parameters was performed with tpsRegr. All values are presented as the means ± SE, and  0.05 is used as the level of statistical significance. We also performed power test (GPower 3.1) for statistical analysis of correlation between Vmax and Ucrit.

Results

Unsteady locomotor performance

Both a conventional ancova (F2,16 = 11.2, = 0.001; F2,16 = 12.0, = 0.001) and a pdanova (= 0.020, = 0.019) suggested that there were significant differences in Vmax and S120 ms among fish species preferring different habitats (Table 3). Both the Vmax and S120 ms of the fish species that typically occur in the riptide habitat were significantly higher than those of the fish species that typically occur in the intermediate and still habitats (Table 3). However, both a conventional ancova (F2,16 = 0.192, = 0.827) and a pdanova (= 0.948) suggested that there were no significant differences in Amax among fish species preferring different habitats (Table 3).

Table 3. The unsteady and steady swimming performance and metabolic parameters of fish species typically occur in different habitats (mean ±S.E.)
HabitatUnsteady swimming performanceSteady swimming performanceMetabolic parameters
Vmax (m s−1)Amax (cm s−2)S120 ms (cm)Ucrit (cm s−1)Uopt (cm s−1)E (10−3)RMR (mgOh−1)MMR (mgO2 h−1)
  1. Values in each column without a common superscript differ significantly (< 0.05).

  2. The Uopt, E and RMR were calculated from the equation listed in Table 4.

  3. The CV is the critical value of the pdanova.

  4. Vmax, maximum linear velocity; Amax, maximum linear acceleration; S120 ms, escape distance during the initial 120 ms; Ucrit, critical swimming speed; Uopt, optimal swimming speed; E, swimming efficiency coefficient; RMR, routine energy expenditure; MMR, maximum metabolic rate.

Riptide1.52 ± 0.09a96.9 ± 5.2102 ± 7a59.9 ± 0.4a35.6 ± 2.3a28.6 ± 1.9b57.0 ± 11.2375 ± 13a
Intermediate1.22 ± 0.06b103 ± 782.1 ± 5.7b49.4 ± 3.2b32.2 ± 3.2ab33.2 ± 3.6ab51.9 ± 7.2296 ± 22b
Still1.09 ± 0.07b98.3 ± 9.871.5 ± 3.5b41.3 ± 2.7c25.3 ± 1.5b40.4 ± 2. 4a44.8 ± 5.9248 ± 13b
ancova
Covariate effectF2,16 = 3.85F2,16 = 0.715F2,16 = 6.07F2,16 = 0.238F2,16 = 0.063F2,16 = 0.022F2,16 = 0.072F2,16 = 0.150
= 0.069= 0.411= 0.026= 0.633= 0.805= 0.885= 0.792= 0.704
Main effectF2,16 = 11.2F2,16 = 0.192F2,16 = 12.0F2,16 = 9.42F2,16 = 3.94F2,16 = 3.70F2,16 = 1.87F2,16 = 5.67
= 0.001= 0.827= 0.001= 0.002= 0.042= 0.050= 0.189= 0.015
Error MS0.02741913349.242.75.80 × 10−53102.96 × 103
Error d.f.1515151515151515
pdanova CV = 8.43CV = 8.91CV = 9.16CV = 9.26CV = 9.24CV = 9.42CV = 12.5CV = 8.73
= 0.020= 0.948= 0.019= 0.045= 0.296= 0.309= 0.419= 0.142

Steady locomotor performance

U crit

Both a conventional ancova (F2,16 = 9.42, = 0.002) and a pdanova (= 0.045) suggested that there were significant differences in Ucrit among fish species preferring different habitats (Table 3). The Ucrit of the riptide group was significantly higher than that of the intermediate group, and the Ucrit of the intermediate group was significantly higher than that of the still group.

math formulaO2

The math formulaO2 increased significantly with increased swimming speed in all fish species (Fig. 3) (< 0.001). In each species, the relationship between oxygen consumption and swimming speed was fitted by least squares using equation 4 (Fig. 3). The values of the corresponding constants and r2 are shown in Table 4. A conventional ancova suggested that there were significant differences in the MMR of fish species preferring different habitats (Table 3) (F2,16 = 5.67, = 0.015). The MMR of the fish species preferring riptide habitat was significantly higher than those of the fish species preferring intermediate and still habitats (Table 4). However, a pdanova suggested that there were no significant differences in MMR among fish species preferring different habitats (= 0.142). Both a conventional anova (F2,16 = 1.87, = 0.189) and a pdanova (= 0.419) suggested that there were no significant differences in RMR among the different groups (Table 3).

Table 4. Regression equations and r2 values for the plots of oxygen consumption vs. swimming speed using equation 3
 Ucrit (cm s−1)a (RMR)b (E) R 2 Uopt (cm s−1)Error d.f.Error MS P
Chinese false gudgeon34.2 ± 0.544.3 ± 4.40.040 ± 0.0040.72524.8330.067<0.001
Small fat minnow28.3±1.171.7±11.40.034±0.0090.30929.6330.159<0.001
Topmouth gudgeon45.5±1.334.0±4.00.046±0.0040.72621.9510.146<0.001
Chinese bitterling42.8±1.249.9±6.40.035±0.0050.53828.6470.170<0.001
Rose bitterling47.6±1.755.1±8.60.034±0.0050.47829.9450.235<0.001
Bighead carp45.1±3.326.8±2.30.048±0.0030.81121.1580.093<0.001
Silver carp53.4±1.447.6±4.00.026±0.0020.65237.9700.108<0.001
Grass carp43.2±1.989.9±5.70.023±0.0020.66544.4470.043<0.001
Black carp43.6±1.335.7±3.00.047±0.0030.80921.3550.084<0.001
Chinese bream54.2±1.364.0±2.70.029±0.0010.86634.2810.031<0.001
Wild carp59.3±1.644.0±4.90.035±0.0030.63629.0770.211<0.001
Chinese hook snout carp60.3±1.476.6±5.00.024±0.0020.73841.5670.063<0.001
Pale chub58.8±2.083.9±4.90.025±0.0020.81040.3570.045<0.001
Qingbo59.7±0.819.7±1.30.030±0.0030.89233.3780.078<0.001
Sharp-jaw barbel59.7±1.150.8±3.50.034±0.0030.83829.2680.070<0.001
Mountain carps60.9±1.253.7±5.30.030±0.0040.65833.6690.148<0.001
Rock carp36.0±2.032.8±2.70.046±0.0030.81421.9410.061<0.001
Crucian carp45.4±1.430.7±3.50.047±0.030.77121.2440.123<0.001
Common carp56.3±1.750.4±4.70.027±0.010.59436.6730.141<0.001
Figure 3.

The oxygen consumption curve of the studied fish at different swimming speeds.

Uopt and E

A conventional anova suggested that there were significant differences in both the Uopt and E values of the different groups (Table 3) (< 0.05). The Uopt of the riptide group was significantly higher than that of the still group (ancova, F2,16 = 3.94, = 0.042), and the E of the riptide group was significantly lower than that of the still group (ancova, F2,16 = 3.70, = 0.05). Neither variable differed significantly between the intermediate group and the other groups (Table 3). However, a pdanova suggested that there were no significant differences in both Uopt (= 0.296) and E (P = 0.309) among fish species preferring different habitats.

The relationship between Ucrit and Vmax and energetic parameters

The raw values of Ucrit and Vmax were positively correlated (R2 = 0.250, = 0.029) (Fig. 4). However, based on a PIC, there was no significant relationship between Ucrit and Vmax given equal branch lengths (R2 = 0.186, = 0.459) (Fig. 4). However, the power of Pearson's correlation was rather low because it was only 0.11 at α error probability < 0.05 level and only 0.58 even at α error probability < 0.459 level. Both a conventional Pearson's correlation (R2 = 0.783, < 0.001) and a PIC (R2 = 0.516, = 0.001) suggested that Ucrit was significantly positively correlated with the MMR (Fig. 4).

Figure 4.

Regression analyses of (a): raw values of Ucrit and Vmax; (b): raw values of Ucrit and MMR; (c): reconstructed changes in Ucrit on each branch vs. reconstructed changes in Vmax; (d): reconstructed changes in Ucrit on each branch vs. reconstructed changes in MMR. Square: Still; Triangle: Intermediate; Circle: Riptide.

Morphology

The result of tpsRegr suggested that there were significant differences in morphology among the different habitats (F22,3388 = 17.49, < 0.001) (Table 5). The fish species that typically occur in the riptide habitat showed a relatively streamlined body shape (Fig. 5). Furthermore, the caudal fin AR did not differ significantly among fish species preferring different habitats (F2,16 = 0.049, = 0.953) (Table 6). Both an ancova (F2,16 = 0.772, = 0.480; F2,16 = 1.475, = 0.260; F2,16 = 3.334, = 0.053; F2,16 = 0.291, = 0.751; F2,16 = 2.001, = 0.170) and a pdanova (= 0.803; = 0.938; = 0.324; = 0.534; = 0.564) suggested that there were no significant differences in FR or Sx/S among different groups(Table 6).

Table 5. Relationships between morphology and the relevant parameters, significance test results from tpsRegr
 Habitat U crit V max A max S 120 ms U opt b
F17.494.635.306.015.987.546.83
d.f.22,338822,338822,338822,338822,338822,338822,3388
P <0.001<0.001<0.001<0.001<0.001<0.001<0.001
Table 6. The effect of habitat flow regime on morphology
HabitatStill N = 7Intermediate N = 7Riptide N = 5 ancova pdanova
Covariate effectMain effectError d.f.Error MSCV P
FR3.53 ± 0.363.89 ± 0.254.19 ± 0.16

F = 2.076

= 0.772

F = 0.772

= 0.480

150.3919.580.803
S1/S (%)15.7 ± 0.917.0 ± 0.814.9 ± 0.7

F = 0.890

= 0.360

F = 1.475

= 0.260

15< 0.0018.790.938
Caudal fin AR4.08 ± 0.234.20 ± 0.094.15 ± 0.12

F = 0.063

= 0.319

F = 0.049

= 0.953

150.1908.670.912
S2/S (%)9.23 ± 0. 708.27 ± 0.799.79 ± 0. 17

F = 0.167

= 0.688

F = 3.334

= 0.053

15< 0.0019.170.324
S3/S (%)6.00 ± 0.805.83 ± 0.456.03 ± 1.54

F = 0.062

= 0.807

F = 0.291

= 0.751

150.00110.10.534
S4/S (%)11.23 ± 1.259.35 ± 2.008.48 ± 1.10

F = 4.311

= 0.055

F = 2.001

= 0.170

150.0018.790.564
Figure 5.

Morphological changes in fish typically occur in different habitats. (a) still; (b) intermediate; (c) riptide.

Discussion

In this study, we examined 19 species of cyprinids and used phylogenetic methods to test interspecific variation in locomotor performance and morphology and to investigate the relationship between these variations and the habitat flow regime. Our study demonstrated a marked divergence in steady swimming performance and unsteady swimming performance among different groups. The fish species from riptide groups tend to have higher Ucrit and Vmax values rather than the higher Ucrit and lower Vmax that we expected. The improved steady swimming performance in the riptide groups was, in part, associated with a more streamlined body shape and hence improved swimming efficiency and higher aerobic metabolic capacity, as suggested by the MMR results. Depending on the PIC, there was no evolutionary trade-off between the steady and unsteady swimming performance of the 19 species, but the Ucrit and MMR showed a pattern of coevolution. However, a further analysis within each typically occurring habitat group showed that the relationship between Ucrit and Vmax is close to the critical value of negative correlation (trade-off), suggesting that the trade-off between steady and unsteady swimming performance may be concealed by the effect of habitat.

The habitat-specific interspecific variation in steady and unsteady swimming performance and the possible underlying morphological and physiological mechanisms

Steady swimming performance

The Ucrit changed significantly among groups in our study. As we expected, the fish species from the riptide groups exhibited higher Ucrit values than the fish species from the intermediate and still habitat groups. The average Ucrit of the fish species that typically occur in riptide habitat was 21.3% higher than that of the fish species that typically occur in intermediate habitat and 45.0% higher than that of the fish species that typically occur in still habitat. Habitat-specific Ucrit variation at the intraspecific level has been documented in several fish species (Kolok & Farrell, 1994; Fu et al., 2012). For example, fish from tributary streams exhibited a 24% higher Ucrit than fish from the main stream (Fu et al., 2012). A previous study of interspecific locomotor variation in turtles indicated that both speed and endurance were habitat specific (Stephens & Wiens, 2008). However, to our knowledge, the current study is the first comprehensive investigation of habitat-specific swimming performance at the interspecific level in fish. The habitat-specific swimming performance may be partly due to plasticity (i.e. training effect) of different velocities in the different habitats. However, a previous study found that flow water training showed no effect on swimming performance at 25°C in qingbo (Pang et al., 2013). Furthermore, the MMR also showed a significant variation among different groups, a result identical to that observed for Ucrit. In addition, fish species also showed marked morphological changes concurrent with those in Ucrit, that is, the fish species from the riptide habitat groups exhibited more streamlining body shape and hence higher swimming efficiency than the fish species from the intermediate and still habitat groups (Fig. 5). These findings suggest that the difference in Ucrit was due, at least in part, to the improvement in aerobic metabolic capacity and higher swimming efficiency as a result of more streamlining of body shape in the riptide groups compared with the other two groups. The intraspecific habitat-specific Ucrit and relevant variation in morphology have also been documented in certain fish species (Fu et al., 2012). These results suggested that steady swimming performance and hence body shape were significantly affected by the selective pressure of the flow regime at both the inter- and intraspecific levels. These results agreed, in part, with our prediction that selection will favour steady swimming and hence the corresponding morphological characteristics in environments with high water velocities.

Unsteady swimming performance

Most previous intraspecific studies have found a significant difference in unsteady swimming performance between fish species living in habitats with different predator densities, with marked concurrent morphological change (Domenici et al., 2008). Interspecific study of the relationship between locomotor and feeding morphology in fishes has also been found (Higham, 2007). However, no previous interspecific studies of flow-driven unsteady swimming performance and concurrent morphological change were found in the literature. In this study, contrary to our expectations, the fish species from the riptide groups exhibited a higher Vmax than the fish species from the intermediate and still habitat groups. The reason for this result may be that the vortex dynamics and high water velocities select for higher unsteady swimming performance to facilitate similar physiological activities, such as the avoidance of predators and obstacles. However, the fish species from the riptide groups did show a more streamlined body shape, which is suitable for steady swimming performance rather than for unsteady swimming performance. Furthermore, there was no significant variation in the caudal region. This finding suggested that morphological change reflects compromises between unsteady and steady swimming performance. Thus, a compromise in morphology endows the riptide groups with both high steady and high unsteady swimming performance. However, the cost of this compromise may be a high expenditure of maintenance energy, as the RMR in the riptide groups was 27% higher than that of the still groups (although it was not significantly different) (Table 3).

Selection type and performance responses

In addition to the significant habitat-specific difference in the steady swimming performance of cyprinids, another interesting phenomenon is that the steady swimming performance (Ucrit) varied 68.0% and 64.7% in the still and intermediate groups, whereas it showed little variation in the riptide group (3.57%). This result may suggest that selective pressure leaves few opportunities for steady swimming performance in a high-velocity environment. However, unsteady swimming performance (Vmax) showed similar 56.2%, 59.4% and 39.1% variation in the still, intermediate and riptide groups, respectively. Furthermore, the Vmax of fish species in the riptide habitat was 39% higher than that in the still habitat, whereas the Ucrit of fish species in the riptide habitat was 45% higher than that in the still habitat. It appears that steady swimming performance was more sensitive to the selective pressure of the water flow conditions in the habitat than unsteady swimming performance. Previous studies at the intraspecific level have also supported this hypothesis. For example, in blacknose dace (Rhinichthys atraturus), a decrease in the water velocity from 22.6 to 2.1 cm s−1 induced a 33% decrease in Ucrit (Nelson et al., 2003). The Ucrit of rainbow fishes (Melanotaenia eachamensis and Melanotaenia duboulayi) from a lake habit was approximately 23% less than that of these same species in a stream habitat (McGuigan et al., 2003). The study about sticklebacks found that compared to anadromous sticklebacks, the freshwater sticklebacks have weaker prolonged swimming performance and better bust swimming performance (Taylor & McPhail, 1986). However, most previous studies have found significant variation in unsteady swimming performance in fish living in habitats with different predator densities rather than in habitats with different flow regimes (Domenici et al., 2008). Intraspecific studies in Chinese hook snout carp (Opsariichthys bidens) and pale chub (Zacco platypus) in our laboratory have suggested that the flow regime, rather than the stress of predation, drives variation in steady swimming performance and the corresponding morphological changes, whereas predation stress, rather than the flow regime, drives variation in unsteady swimming performance and the corresponding morphological changes among fish populations from different habitats (Fu et al., 2012; Fu et al., in press). These results may indicate that different selective pressures in nature may produce different patterns of divergence in locomotor performance and that steady swimming performance is more dependent on the flow regime, whereas unsteady swimming performance is more dependent on the predator density. Further investigations of this hypothesis might be of interest.

The evolutionary trade-off between steady and unsteady swimming in cyprinids

The trade-off between Ucrit and fast-start capability has long been hypothesized to be widespread and to play an important role in the ecology and evolution of fish (Webb, 1984; Langerhans et al., 2007). The predicted trade-off in swimming performance has been investigated at the intraspecific level and has been confirmed by several studies (Schliewen et al., 2001; Barluenga et al., 2006). However, only a few studies have been performed at the interspecific level, and these previous interspecific studies of the trade-off between steady and unsteady swimming performance have all ignored the effects of phylogeny. To our knowledge, this study is the first to measure the evolutionary trade-off between steady and unsteady swimming performance among fish species. Contrary to our expectations, we found no significant relationship between the relative change in Ucrit and Vmax in this study (Fig. 4d), although the raw values of Ucrit and Vmax were positively correlated (Fig. 4a). This finding may suggest that no evolutionary trade-off between steady and unsteady swimming performance occurred in the 19 fish species selected for study. However, when we tested the evolutionary trade-off between Ucrit and Vmax within each group, we found a significant correlation between Ucrit and Vmax in the riptide (R2 = −0.971, = 0.029) and a nonsignificant but high negative correlation (trade-off) in intermediate group (R2 = −0.770, = 0.074) (Fig. 6). Furthermore, because of the low sample size within each group, the power of the correlation in intermediate group was rather low (0.61). Thus, these findings may indicate that the evolutionary trade-off between steady and unsteady swimming performance was environment dependent and that the trade-off may occur among fish species inhabiting similar environments.

Figure 6.

Regression analyses of reconstructed changes in Ucrit on each branch vs. reconstructed changes in Vmax within each group. (a) still; (b) intermediate; (c) riptide.

In conclusion, this study has found significant changes in steady and unsteady swimming performance changes among fish preferring different flow regimes. Significant changes in morphology were more consistent with changes in steady swimming performance than with changes in unsteady swimming performance. Furthermore, although no evolutionary trade-off between steady and unsteady swimming performance was found in the 19 species of cyprinids selected for study, it indicates that the evolutionary trade-off between steady and unsteady swimming performance may occur among fish species inhabiting similar environments. These results suggest that selective pressure on swimming performance yields fish with different swimming strategies and body shapes in habitats with different water velocities. They also suggest that the flow regime plays an important role in the evolutionary divergence of locomotor performance in fish. Therefore, further studies of locomotor trade-offs should recognize the importance of the flow regime and select experimental subjects that inhabit similar environments.

Acknowledgments

This study was funded by grants from the National Science Foundation of China (NSFC 31172096), the Key Project of Natural Science Foundation of CQ (cstc2013jjB20003) and the Project of Chongqing Science & Technology Commission (No. CSTC, 2010CA1010) granted to S.J.F.

Ancillary