Lateralized behavior in the attacks of largemouth bass on Rhinogobius gobies corresponding to their morphological antisymmetry

Animals do not necessarily behave left–right symmetrically. Some vertebrates are known to show biases in their turning direction, limb usage, predator-escape response and use of sensory organs (Rogers and Andrew, 2002). In these animals, each individual favors either the left or the right side when it performs a behavior. Viewed across a population, such individual behavioral biases exhibit a unimodal distribution biased towards the left or right side, or a bimodal distribution with two peaks corresponding to the left and right sides (Vallortigara and Rogers, 2005). Such a left–right difference in behaviors, especially at the population level, has been thought to result from brain lateralization, and many studies have examined these lateral biases in various animals to explore the origins of brain lateralization.

There have been a few studies that have discussed the correlation between morphology and these behavioral biases, especially in lower vertebrates (Dill, 1977; Bisazza et al., 1998). However, recent studies in some fishes have demonstrated that the leftward or rightward bias in the behavioral direction of each individual corresponds to morphological asymmetry, specifically antisymmetry (Hori, 1993; Nakajima et al., 2007; Takeuchi and Hori, 2008; Takeuchi et al., 2010; Hata et al., 2011). This antisymmetry is defined as a dimorphism in which one side of the body is structurally and/or functionally more developed than the other side, and is distinguished from fluctuating or directional asymmetry (Palmer and Strobeck, 1986; Palmer, 1996). When a population shows this dimorphism, it consists of left-biased individuals with well-developed left sides (so their bodies are convex to the left) and mouths opening rightward (hereafter, ‘lefties’) and right-biased individuals with the opposite traits (hereafter, ‘righties’). Lateral biases in behaviors are not necessarily proximately related to the left–right differences of these traits. However, the biases tightly correspond to the morphological type identified by these traits. It is predicted that this correspondence between behavior and morphology may result from the favored use of one eye (Takeuchi et al., 2010) or lateralized neuronal circuits for controlling such behavior (Takeuchi et al., 2012) according to the morphological difference. Antisymmetry has been found in many fishes belonging to seven families in four orders (Mboko et al., 1998; Seki et al., 2000; Hori et al., 2007; Nakajima et al., 2007; Takeuchi and Hori, 2008; Takeuchi et al., 2010; Hata et al., 2011; Hata and Hori, 2012; Hata et al., 2012) since its first documented occurrence in the scale-eating cichlids (Perissodus spp.) of Lake Tanganyika (Liem and Stewart, 1976; Hori, 1991; Hori, 1993). These studies suggest that this dimorphism is not rare in fishes.

What effect can these lateral biases in behaviors have on the predator–prey interactions in fish? Some predators have a common bias in approach or attack directions according to their morphological type (lefty or righty), with lefties showing more clockwise or rightward behaviors and righties showing more counterclockwise or leftward behaviors (Hori, 1993; Nakajima et al., 2007; Takeuchi and Hori, 2008; Takeuchi et al., 2012). Similarly, prey may have lateral biases in the direction of their evasive response corresponding to their morphological asymmetry. For example, lefties may be able to detect or escape from predators more easily when they are approached from their left side than when they are approached from their right side. If so, predation success will differ according to the combinations of morphological types of predators and prey.

As support for this hypothesis, field studies of predator–prey interactions have demonstrated that some piscivorous fishes exploit more prey of the opposite morphological type (cross predation) than of the same morphological type (parallel predation) (Hori, 2000; Hori, 2007; Yasugi and Hori, 2011). Moreover, mathematical models demonstrate that this predominance of cross predation can maintain the dimorphism without fixing a population to either only lefties or only righties (Nakajima et al., 2004). However, the evasive responses of prey have not previously been studied with regard to morphological antisymmetry. Recognized lateral biases in fish behaviors have been mostly investigated from the viewpoint of brain lateralization. Additionally, many of these behavioral biases were observed in detour tests (Bisazza et al., 1998) or elicited by artificial stimuli using either individual predators or individual prey. To understand the mechanism of cross predation predominating over parallel predation, one must investigate the relationship of morphological antisymmetry with the predatory and evasive behaviors that play important roles in predation success during actual encounters.

The purpose of the present study was to test the hypothesis that lateral biases in predatory behavior and evasive response corresponding to morphological antisymmetry will generate the predominance of cross predation in predator–prey interactions in fish. The largemouth bass, Micropterus salmoides (Lacepède 1802), is a major predator in Lake Biwa, Japan, and mainly feeds on the freshwater goby, Rhinogobius sp. (Ministry of the Environment, Government of Japan, 2004). Morphological antisymmetry occurs in both species (Seki et al., 2000; Nakajima et al., 2007; Yasugi and Hori, 2011), and the predominance of cross-predation has been demonstrated in their predator–prey interaction (Yasugi and Hori, 2011). Accordingly, we carried out a behavioral test of predation events using these two species, and investigated the interactions between the predatory behavior of bass and the evasive response of goby with respect to their dimorphism.

Thirty-four largemouth bass [mean ± s.d.=11.2±1.4 cm standard length (SL)] and 91 Rhinogobius sp. (Orange type mean ± s.d.=3.4±0.4 cm SL) were collected around the west coast of Lake Biwa, Japan. These fishes were transported to our laboratory, and kept in a glass tank (60×30×45 cm) with an undergravel filter (10 h:14 h dark:light photoperiod). The water temperature of the tank was maintained at 22±1°C. Bass were kept individually in each tank and fed once every 4 days with a live goldfish, minnow or goby. The fish were acclimated to feed under the rearing conditions for at least 1 month; subsequently, to acclimate them to hunting in the observational arena, they were repeatedly fed in the observational arena for at least 1 month before the behavioral test as in the procedure for trials described below. They were treated with the permission of the Ministry of the Environment, Government of Japan (registration number 06000338). Gobies were kept together in one tank and fed daily with pellet food. The fish were acclimated to feed in the rearing environment for at least 1 week before the behavioral test. The treatment of these fishes during rearing and behavioral testing was in accordance with the Regulations on Animal Experimentation at Kyoto University.

The behavioral test was conducted in a polyethylene tank (78×170×38 cm; Fig. 1) from 2004 to 2006 and from 2010 to 2011. The tank was filled with aerated water to a depth of 19 cm (250 l, 22±1°C) and lit with four halogen lamps covered with tracing paper. The light intensity at the surface of the water in the tank was 2.2 klx, which was sufficient for largemouth bass to locate and attack prey (Howick and O'Brien, 1983; McMahon and Holanov, 1995). The tank consisted of two areas: preparatory areas and the observational arena. Preparatory areas were composed of two left and two right compartments (19.5×15 cm). Sliding doors (10×2.5 cm) connected adjoining preparatory areas and the preparatory areas with the observational arena. All predation events occurred in the observational arena.

To maintain a consistent hunger level in the largemouth bass, we always used subject fish that had been starved for 3 days. The day before the trial, each bass was introduced into the left or right compartment of the preparatory area. The sliding doors were kept closed, and the side of the compartment into which each bass was introduced was randomly selected. On the trial day, we anesthetized gobies with 1% phenoxyethanol solution. Then we took a photograph of the lower jaw of each goby from the ventral side under a digital microscope (VHX-100; Keyence, Osaka, Japan) to identify its laterality phenotype (the method of identification is described later). After taking a photograph, each goby was kept in a tank for at least 4 h until it recovered from the anesthetic.

For the trial, each goby was introduced into the center of the observational arena and held under a transparent plastic cover (7×7×1.5 cm) with a long stick (20 cm) attached. We waited 15 min for the goby to become still on the bottom of the tank and accustomed to the observational arena. In preliminary field observations in Lake Biwa, we observed largemouth bass approaching gobies only from behind (N=7). Thus, each goby was positioned with its tail to the sliding door of the compartment of the preparatory area in which a bass was contained. After 15 min, we used the attached stick to gently lift the plastic cover and then opened the sliding door, and recorded the sequence of predation events from above with a digital video camera at a 60 Hz field rate (GR-DV3000; Victor Company of Japan, Yokohama, Japan). When a goby was not stationary or attempted evasion before the beginning of the approach of the bass, the trial was interrupted and restarted with the introduction of the goby. One predation event was defined as the sequence of behaviors from the approach of a bass to its first strike at a goby, or until a change in the direction of the escape trajectory of a goby, regardless of its success. We repeated trials no more than five times for each largemouth bass, depending on its state, and gobies were used only once.

Generally, each sequence of predation behavior proceeded as follows. The largemouth bass swam into the observational arena through the sliding door, detected the goby, which stayed at the center of the arena at the bottom, and then repeated stop and go with orientation and positioning behavior, as if the fish were judging the accurate location of the goby (see Nyberg, 1971). After some reduction in the distance between predator and prey, the bass made a slow and circular approach to the goby (Fig. 2A). This approach was characterized by cruising with little undulation of the body. Bass approached by turning clockwise or counterclockwise from behind towards the goby's head. When the bass was sufficiently close to the goby, it reduced its cruising speed (and sometimes stopped). Then, the bass quickly bent its body (a ‘strike’; Fig. 2B), protruded its mouth and seized the goby. Three patterns of timing were observed when the bass started to strike and the goby attempted escape: strike preceding, simultaneous occurrence and escape preceding. In escape preceding, the bass often dashed without striking and chased after the goby (see Dill, 1973). When bass approached, gobies showed C-start evasive behavior. Initially, the goby quickly bent its body, which formed a C-like shape (stage 1; Fig. 2C). This body bend was followed by a counterturn and propulsion forward with shifting of the center of mass of the goby (stage 2), leading to subsequent escape (Fig. 2D) (Domenici and Blake, 1997).

We measured the distance, direction and speed in the strike behavior of bass and the evasive response of goby as described below (Fig. 2). The success of predation in each trial was also noted. Moreover, we measured the time until the bass started a strike or a dash behavior after gobies attempted to escape, quantified as the number of video frames. Such differences of the timing of behaviors are expected to relate to the success of predation by altering the ease of strike or escape. These data were obtained by replaying video records on Image Tracker PTV (Digimo Co., Osaka, Japan), using the tip of the snout as a position reference for both bass and gobies (Fuiman, 1993). Parameters taken from video records were defined as follows, grouped into three categories: approach of the bass (A), strike behavior of the bass (S) and evasive response of the goby (E).

To investigate the lateral biases in approach direction corresponding to morphological antisymmetry, we noted approach direction (A-1), clockwise or counterclockwise (Fig. 2A).

To compare approach locomotion among combinations of morphological type and approach direction, we calculated approach speed (A-2), the trajectory length from the starting location of the circular approach to the starting location of the strike or dash divided by the number of video frames during the approach.

Strike behaviors were measured when bass exhibited strikes to stationary gobies; namely, in strike preceding and simultaneous occurrence cases. To investigate the lateral difference in strike locomotion according to morphological antisymmetry and strike direction (noted as either leftward or rightward), we calculated body bending speed (S-1), the maximum velocity of the tip of the snout during the body bend (Fig. 2B).

To compare the distance at which bass started their strike, we calculated strike distance (S-2), the length between the tips of snouts of the bass and the goby at the start of a strike on the goby (Fig. 2B).

Evasive responses were measured when gobies reacted to the approach behavior of largemouth bass, namely, in escape preceding and simultaneous occurrence cases. We categorized the evasive behaviors of the goby into reactive muscle contraction (stage 1) and subsequent propulsion toward the eventual direction. The phase of propulsion was defined as the escape behavior from stage 2 to the sixth frame or the frame in which the direction of trajectory changed. This route was designated as the ‘escape trajectory’ (Domenici, 2010).

To compare the distance at which gobies reacted to the bass and started their escape between morphological types and approach directions, we calculated evasive distance (E-1), the distance between the tips of the snouts of the goby and the bass at the start of the evasive response (Fig. 2C).

To investigate lateral difference in the escape locomotion, we calculated stage 1 escape speed (E-2), the maximum velocity of the tip of the snout during the body bend in stage 1 (Fig. 2C), and escape trajectory speed (E-3), the mean swimming velocity along the escape trajectory (Fig. 2D).

And finally, to examine the relationship between morphological antisymmetry and the direction in which gobies escaped, we noted evasive direction (E-4), the left–right direction of the escape trajectory to the body axis of the goby (Fig. 2D).

The trait asymmetry of largemouth bass and gobies was measured to identify their morphological type. In bass, we adopted the left–right difference in the height of the mandible posterior ends (HMPEs) (Hori et al., 2007) of the lower jaw as an indicator of morphological antisymmetry (Yasugi and Hori, 2011). After all trials were complete, each bass was anesthetized with 2% phenoxyethanol solution to stop respiration. Subsequently, the entire fish was preserved in 10% formalin for fixation. We removed the lower jaws from fixed fish and soaked them in sodium hypochlorite (8.5–13.5% Cl) diluted twice with water to dissolve skin and muscle. The left and right HMPEs of every specimen were measured three separate times under a digital microscope (VHX-100; Keyence) to the nearest 0.01 mm. We used the medians of the three repeated measurements to reduce measurement error, and calculated an index of asymmetry (IAS): [2×(M_R–M_L)/(M_R+M_L)]×100, where M_R and M_L are the medians of the right and left HMPE, respectively. Individuals with positive IAS values (i.e. with the right HMPE larger than the left) were defined as righties, and those with negative IAS values were defined as lefties (Hori et al., 2007; Yasugi and Hori, 2011).

In gobies, the ventral side of the lower jaw was photographed under a digital microscope (VHX-100; Keyence) with strong illumination to make the posterior end of the dentary bone visible through thin skin and muscle. We measured dentary length, i.e. the distance between the tip and the posterior end of the dentary, using ImageJ (National Institutes of Health, Bethesda, MD, USA). This measurement point was slightly different from that adopted in our previous study (Yasugi and Hori, 2011), but a preliminary examination showed that these two traits had the same tendency in the difference between left and right sides; righties, defined by the difference in HMPE, had longer left than right dentaries, whereas this was reversed for lefties. Thus we calculated the IAS of the dentary as [2×(D_R–D_L)/(D_R+D_L)]×100, where D_R and D_L are the lengths of the right and left dentary, respectively. Individuals with positive IAS values were defined as lefties, and those with negative IAS values were defined as righties (Yasugi and Hori, 2011).

During the trial repetitions, largemouth bass mainly approached gobies from behind, but sometimes made frontal approaches, which was attributable to one of two causes. One was that the goby turned after the cover was lifted. The other was that the bass did not detect the goby immediately and moved to a location leading to a face-to-face encounter. To evaluate the orientation of a largemouth bass approach towards a goby, we measured the angle between the body axes of the bass and the goby at the start of the approach (orientation angle; Fig. 2A). Body axis was defined as the line connecting the tip of the snout and the midpoint between the eyes. The orientation angle was 0 deg when a bass was directly behind a goby, and 180 deg when a bass was directly in front of a goby. Subsequently, we defined trials with an orientation angle <90 deg as approaches from behind, and trials with an orientation angle >90 deg as approaches from the front. Data from the first two trials involving approach from behind in each bass were grouped into a ‘predation from behind’ data set and, to investigate the difference in predation events between encounter styles, we also grouped data from the first one or two trials involving the approaches from the front in each bass into a data set of ‘predation from the front’ (two individuals showed only one approach from behind trial, and one individual showed only an approach from the front). For details, see supplementary material Table S1.

We investigated the effects of combinations of bass and goby morphological types (their types were the same or different) on the success of predation by fitting a generalized linear mixed model (GLMM) (Schall, 1991), assuming the binomial error distribution (link=‘logit’). In this model, the orientation angle was also contained as a fixed factor. Predation success when the bass and goby types are the same indicates parallel predation, and predation success when their types are different indicates cross predation. Moreover, we constructed GLMMs to examine the effects of morphological antisymmetry and the interaction between behavioral measures. In the models for approach direction (counterclockwise/clockwise, A-1) and evasive direction (rightward/leftward, E-4), fish morphological type (righty/lefty) was a fixed factor, and the binomial error distribution (link=‘logit’) was assumed. The model for approach direction additionally contained the difference in preparatory area compartment into which each bass was introduced (left/right) as a random factor. In the models for approach speed (A-2), body bending speed (S-1), strike distance (S-2), evasive distance (E-1), stage 1 escape speed (E-2) and escape trajectory speed (E-3), fish morphological type (righty/lefty), behavioral direction (counterclockwise/clockwise or rightward/leftward), their interaction and body length were fixed factors, and a Gaussian error distribution (link=‘identity’) was assumed. Multiple comparison among the four categories of combinations of morphological types and behavioral directions was also performed. All of these models included the individual identity of bass as a random factor to avoid pseudoreplication because each bass repeatedly participated in trials (e.g. Hurlbert, 1984; Russell et al., 2002).

Statistical analyses were performed using R version 2.13.0 (R Development Core Team, 2011). Additionally, we used the packages lme4 version 0.999375-41 and nlme version 3.1-102 for GLMMs and multcomp version 1.2-7 for multiple comparison.

Among the 34 largemouth bass subjects, 15 were lefties and 19 were righties. The 91 goby subjects were composed of 57 lefties and 34 righties. The distribution of orientation angle is shown in Fig. 3A. In the 91 trials that we observed, predation success was different between trials in which bass and gobies were same morphological type and trials in which they were different types, and this tendency changed depending on the orientation angle [GLMM for success/failure, orientation angle, β=–0.025±0.012, z=–1.954, P=0.050; combination of morphological type (same), β=–3.078±1.131, z=–2.722, P=0.006; orientation angle × combination of morphological type, β=0.037±0.015, z=2.458, P=0.014; Fig. 3B,C]. Accordingly, we investigated the effect of combination of morphological types on predation success in each approach style. As a result, in predation from behind, trials in which bass and gobies were different morphological types showed higher predation success than trials in which they were the same type [N=64, GLMM for success/failure, combination of morphological type (same), β=–1.788±0.845, z=–2.116, P=0.034; Table 1]. Conversely, in predation from the front, trials in which their morphological types were the same seemed to show higher predation success than when their types were different (Table 1), but this was not supported statistically [N=27, GLMM for success/failure, combination of morphological type (same), β=13.55±6992.28, z=0.002, P=0.998].

Of the 64 trials of predation from behind, 10 were strike preceding, 13 were simultaneous and 41 were escape preceding. Of the additional 27 trials of predation from the front, four were strike preceding, seven were simultaneous and 16 were escape preceding (Table 2). These distributions of behavioral timing were not different between approach styles (Fisher's exact test, P=0.895). Successful predation most frequently occurred in strike preceding, was less frequent in simultaneous occurrences and was rare in escape preceding, regardless of approach style (Table 2). This suggests that predation tended to succeed when bass exhibited strike behavior in response to stationary gobies (strike preceding and simultaneous occurrence), and tended to fail when goby escape preceded bass strike (in this case, bass chased a goby swimming away and struck it). This tendency was statistically supported [N=91, GLMM for success/failure, bass strike timing (escape preceded strike), β=–13.365±5.362, z=–2.492, P=0.012; approach style (front), β=–5.681±3.451, z=–1.646, P=0.099].

Largemouth bass did not start their strikes until they were sufficiently close to gobies. Most bass strikes started when bass approached gobies at a distance of less than 6 cm (mean ± s.d.=4.1±1.6 cm, N=34; Fig. 4A). Gobies could start their evasive response at a greater distance (mean ± s.d.=5.0±2.2 cm, N=77; Fig. 4B). Furthermore, in escape preceding trials, the longer the distance between bass and gobies at the start of goby escape, the more the subsequent strike or dash of bass was delayed, but this did not differ between approach styles [N=77, GLMM, evasive distance, β=0.321±0.083, t=3.844, P<0.001; approach style (front), β=–0.336±0.390, t=–0.862, P=0.393; Fig. 4B].

Largemouth bass showed significant bias in approach direction (A-1), with lefties tending to approach the goby clockwise and righties tending to approach counterclockwise (N=64, GLMM; Table 3, Fig. 5). This tendency was not different between approach styles [N=91, GLMM, bass morphological type (righty), β=1.881±0.626, z=3.004, P=0.002; approach style (front), β=1.159±0.756, z=1.534, P=0.125; morphological type × approach style, β=–0.281±1.087, z=–0.259, P=0.795].

Approach speed (A-2) was not explained by bass morphological type, approach direction, their interaction, or the body length of bass (N=64, GLMM; Table 4). Also, no difference in approach speed was observed among the combinations of morphological type and approach direction.

Neither body bending speed (S-1) nor strike distance (S-2) were explained by bass morphological type, strike direction, their interaction, or the body length of bass (N=23, GLMM; Table 4). These did not differ among combinations of the morphological type and strike direction.

Evasive distance (E-1) when gobies were approached from behind was affected by their morphological type, approach direction and the interaction between these factors (N=54, GLMM; Table 4), although it was not explained by the body length of either the bass or the goby. Multiple comparison showed that the evasive distance of lefty gobies was longer when being approached clockwise than counterclockwise (β=–2.155±0.797, z=–2.702, P=0.033), whereas the evasive distance of righties was longer when being approached counterclockwise than clockwise (β=2.505±0.933, z=2.683, P=0.035; Fig. 6A). Additionally, when gobies were approached clockwise, lefties showed a longer evasive distance compared with righties (β=–2.844±0.837, z=–3.398, P=0.003), but no difference was observed between morphological types of gobies when they were approached counterclockwise.

Neither stage 1 escape speed (E-2) nor escape trajectory speed (E-3) were affected by goby body length, morphological type, evasive direction (E-4) or the interaction of the latter two factors (N=54 and 52, respectively, GLMM; Table 4). Evasive direction (E-4) was also not explained by goby morphological type (N=52, GLMM; Table 3).

Largemouth bass showed a lateral bias in approach direction: individuals approached gobies either clockwise or counterclockwise corresponding to their morphological antisymmetry (Fig. 5). This behavioral bias can explain the lateralized hook position demonstrated in Nakajima et al. (Nakajima et al., 2007). In their angling experiment, lefty bass were hooked on the left side of their mouths and righty individuals were hooked on the right side. Lefty bass will approach clockwise and attack as the left side of their mouth grazes a lure moving away, causing lefty bass to be hooked on the left side. Moreover, this behavioral bias seem not to result from a lateral difference in the physical mechanisms necessary for approach, because no difference in approach speed was observed between the frequent approach direction and the occasional approach direction. Similarly, this also does not appear to be related to the physical ease of the subsequent strike, as the body bending speed of bass showed no difference between strike directions for either lefty or righty bass. As for the proximate cause of this lateral bias in approach direction, we hypothesize that bass have laterally biased perception, i.e. they favor the use of one eye over the other for targeting a prey item. Such lateralization in eye use of predators has been found previously in fish (Rogers and Andrew, 2002). The preference of bass in using either the left or the right eye may be related to their morphological asymmetry.

For the behavioral traits of Rhinogobius gobies, when they were approached from behind, evasive distance differed between the left and right side corresponding to their morphological antisymmetry (Fig. 6A). Lefty gobies reacted to clockwise approaches of largemouth bass at a further distance compared with counterclockwise approaches, whereas righty gobies reacted to counterclockwise approaches at further distance compared with clockwise approaches. This left–right difference in evasive distance seems to result from lateralization in perception corresponding to morphological antisymmetry, e.g. functional difference between the left and right eyes or lateral lines for detecting predator approach from the left and right sides.

Moreover, evasive distance was closely related to whether gobies were able to avoid predation. Predation success became remarkably higher when largemouth bass started their strike against stationary gobies, compared with the cases in which gobies started evasive responses before bass strike (Table 2). However, bass did not start their strikes until they were sufficiently close to gobies (Fig. 4A), which means that there is an optimum range of distance for bass to start striking prey. Therefore, to avoid predation, gobies must start their evasive response by the time bass approach to this adequate distance for their strike. Additionally, when the evasion attempt preceded the start of bass strike, the evasive response at a further distance delayed the subsequent bass strike or dash (Fig. 4B). These results suggest that gobies reacting from a further distance will more easily escape from bass.

The lateral bias in approach direction, left–right difference in the distance of the escape response and distance-dependent predation success all lead to the expectation that predation will be more successful when a lefty bass encounters a righty goby or a righty bass encounters a lefty goby, but will be less successful when a lefty bass encounters a lefty goby or a righty bass encounters a righty goby. This expectation is consistent not only with the difference in predation success in our tests between combinations of morphological types of bass and gobies in predation from behind (Table 1), but also with field data on the predominance of cross predation between these two fishes (Yasugi and Hori, 2011). Therefore, we believe that the lateral biases in approach direction and in evasive response corresponding to morphological antisymmetry are the principal mechanisms causing the predominance of cross predation. These behavioral biases of each morphological type generate the predominance of cross predation, which leads to the maintenance of antisymmetric dimorphism through negative frequency-dependent natural selection (Nakajima et al., 2004; Yasugi and Hori, 2011).

This mechanism of predominance of cross predation will not be specific to this predator–prey relationship. Other predators, including scale-eating cichlids (Hori, 1993) and the shrimp-eating cichlid Neolamprologus fasciatus (Takeuchi and Hori, 2008), have lateral biases in the direction of their hunting behavior similar to that of the bass approach. Thus, this lateral bias in approach direction may be common in the hunting behavior of predators when stalking prey. Moreover, the largemouth bass is an exotic predator that was introduced into Lake Biwa in the 1970s (Ministry of the Environment, Government of Japan, 2004), and only a short time (<40 years) has passed since it first encountered the indigenous goby. Thus, the mechanism revealed in this study will be seen as long as predator and prey exhibit lateral biases in their behaviors, rather than have characteristically developed in the relationship between the bass and the goby.

Furthermore, we hypothesize that the bias in predation success is reversed in approaches from behind and approaches from the front. Clockwise approaches by the largemouth bass head for the left side of the goby in predation from behind and for the right side in predation from the front, and vice versa for the counterclockwise approach. Thus, in predation from the front, lefty gobies will react to a counterclockwise approach of the bass at a further distance than to a clockwise approach, whereas righty gobies will react inversely, which may lead to the predominance of parallel predation. In Fig. 6B, we show a comparison of evasive distance in predation from the front between morphological types and approach directions. Unfortunately, our hypothesis about evasive distance was not statistically supported in the present study, probably because of small sample size. However, predation from the front showed a tendency different to that of predation from behind (Table 1, Fig. 3). If lateral biases in predatory approach and evasive responses are common among other fish predators and prey, cross predation may dominate in predator–prey interactions where predators approach prey from behind, and parallel predation may dominate in interactions where predators approach prey from the front. Ambush-hunting and luring predators (Keenleyside, 1979) are expected to exhibit frontal encounters with prey. The examination of the relationship between morphological types of such predators and prey fishes exploited by them may verify our hypothesis presented here.

In the present study, the escape direction of gobies did not exhibit leftward or rightward bias corresponding to their morphological antisymmetry. Previous studies have shown that atyid shrimps (Takeuchi et al., 2008) and crayfish (Tobo et al., 2012) have lateral biases in escape direction from vibration or contact stimuli corresponding to their morphological asymmetry. In our test, when a largemouth bass approached a goby, the bass traced an arc towards the head of the goby, rather than adopting a linear track, which would be the shortest distance to the goby. Such bilateral location on either the left or right side of the goby's head will likely be advantageous to the predation success of bass by preventing the goby from escaping in any direction at will [see the approach of bass to crayfish (Nyberg, 1971) and the attacks of tentacled snakes on prey fish (Catania, 2009)]. Actually, in the 64 trials involving an approach from behind, 14 gobies (22%) escaped towards the direction of the approaching bass, whereas 50 gobies (78%) escaped towards the direction in which the bass were absent. Therefore, it is natural that escape direction was strongly affected by bass approach direction, unlike in tests using only prey or a nondirectional stimulus (see Domenici et al., 2011). When prey are exposed to stimuli without left–right directionality, the lateral bias in escape direction may appear as an important factor in the predominance of cross or parallel predation. Such a stimulus is expected in predator–prey relationships when the predator attacks the prey from directly behind, e.g. with a chasing predator and cruising prey. Predators and prey meet in various scenarios, dependent on their ecology. Moreover, the factors playing important roles in predation success will differ with each encounter scenario. To reveal the mechanisms causing the predominance of cross and parallel predation, one must consider how predators and prey meet, and identify the lateral biases in the direction of behavioral traits that concern their encounter scenario.

We thank T. Komiya, M. Sasabe and H. Nishi for their assistance in sampling largemouth bass, and S. Takahashi for help with statistical analysis. We also thank Y. Takeuchi, D. Takahashi, K. Watanabe, T. Sota and other members of the Laboratory of Animal Ecology, Graduate School of Science, Kyoto University, for their support in this research. Finally, we deeply appreciate the valuable comments from the anonymous reviewers.