INTRODUCTION

Behavioral and sensory lateralization is widely recognized in various vertebrates (Roger and Andrew, 2002). Studies of lateralization have focused primarily on humans and higher vertebrates, mainly examining cerebral lateralization and related behavioral or sensory differentiation between right and left sides. However, morphological asymmetry, such as differences between the right and left sides of kinetic or sensory systems, has not been explored in depth, except for differences in the brain and nervous system. Recent studies of laterality in fishes have revealed that behavioral laterality is always associated with morphological asymmetry, which is characteristically represented as dimorphism in populations, i.e., every population is composed of both lefty and righty individuals at different rates. The goals of this paper are (1) to demonstrate that all fishes have laterality and, therefore, that the origins of laterality are at least as old as those of fish, and (2) to discuss the significance of the maintenance of dimorphic laterality in most populations. Specifically, we direct our attention to predator-prey interactions in biotic communities as the primary factor responsible for the maintenance of dimorphism of laterality in fish.

The first and most notable example of laterally asymmetric bodies in fishes was found in scale-eating cichlids in Lake Tanganyika (Liem and Stewart, 1976; Hori, 1991, 1993; Hata et al., 2013), in which laterality was associated with hunting techniques. Each population consists of righty (the right side of the head faces slightly forward, and the mouth opens to the left) and lefty (vice versa) individuals. Righty (lefty) individuals attack the right (left) sides of their prey fish from behind to tear off scales (Hori, 1993; Takeuchi et al., 2012, 2016). The behavioral laterality associated with bilateral asymmetry has also been found in fishes from a wide range of trophic levels and taxa, including cichlids (Hori et al., 2007; Takeuchi and Hori, 2008), gobies (Seki et al., 2000; Yasugi and Hori, 2011), bettas (Takeuchi et al., 2010), largemouth bass (Nakajima et al., 2007; Yasugi and Hori, 2011), characins (Hata et al., 2011), poeciliids (Matsui et al., 2013), and anglerfish (Yasugi and Hori, 2016). Nearly all of the individuals in these studies could be classified as either righty or lefty individuals.

Laterality appears to have a genetic basis, as indicated by studies on parents and their offspring in wild species (a goby: Seki et al., 2000; two cichlids: Hori et al., 2007) and in breeding experiments (a cichlid: Hata et al., 2012; medaka: Hata et al., 2012; and zebrafish: Hata and Hori, 2012), in which righty and lefty offspring were born in a Mendelian ratio with left dominant over right. Additionally, genetic studies have identified genomic loci corresponding to mandibular asymmetry (Stewart and Albertson, 2010).

The degree of laterality in individuals has been quantified based primarily on two measures: (i) differences between the right and left mandibles (Hori, 1993; Seki et al., 2000; Takeuchi and Hori, 2008; Yasugi and Hori, 2011; Takeuchi et al., 2012; Hata et al., 2013), more specifically, the height of the posterior ends of the right and left mandibles (HMPE) (Hori et al., 2007; Hata et al., 2013) (Fig. 1A), and (ii) the angle θ formed by the head and vertebrae (Hata et al., Yasugi and Hori, 2011; Nakajima et al., 2007) (Fig. 1B), in which the head of each individual inclines laterally, such that one side of the head faces forward (Hori et al., 2007). Laterality identified in every individual using these two methods has been consistent (Yasugi and Hori, 2011; Hata et al., 2013).

In general, lateral asymmetry in species or populations can be divided into three categories based on the frequency distribution of measured laterality: fluctuating asymmetry (FA) with a unimodal and symmetrical distribution, directional asymmetry (DA) with a unimodal distribution shifted from symmetry, and anti-symmetry (AS) with a bimodal distribution (Palmer and Strobeck, 1992). Recent studies examining the asymmetry of fishes have shown bimodal distributions, suggesting AS (Hori et al., 2007; Nakajima et al., 2007; Takeuchi and Hori, 2008; Seki et al., 2000; Takeuchi et al., 2010, 2012, 2016; Hata et al., 2011, 2013; Yasugi and Hori, 2011, 2012; Matsui et al., 2013). Here, to ascertain the type of asymmetry, we examined frequency distributions of two measures in two species of bony fishes—amberjack and catfish—as arbitrary representatives of dominant taxa in marine and freshwater environments, respectively, as well as in one cartilaginous fish, a stingray.

Fig. 1.

Bone measurements used to identify fish laterality. (A) The height of the mandible at the posterior end (HMPE) is the distance from A (the deepest point on the suspensoriad articulation facet of the anguloarticular) to RP (the ventral corner of the retroarticular process). (B) Ventral view of a skeleton showing data points for head angle (θ), defined as the angle between the line extended from V_III–B (dashed red line) and P–B. B: the sagittal posterior end of the basioccipital; P: the center of the posterior portion of the parasphenoid; V_III: the sagittal posterior end of the third vertebra. Both panels show the Japanese amberjack (Seriola quinqueradiata).

Next, to assess the prevalence of lateral asymmetry among fishes and to identify trends in the degree of laterality during evolution, we measured the laterality of individuals from at least one arbitrary representative species from all 62 extant orders of fishes (Nelson, 2006). The species examined and the measurements of the 62 orders are listed in the electronic supplementary material ( Supplementary Table S1 (10.2108.zsj.34.267.s1.pdf)). We measured mandibular asymmetry and head angle (as described above) in 10–12 individuals from one to several populations of each species, except for five rare species in which only a few individuals (n < 6) were available.

MATERIALS AND METHODS

Degree of lateral asymmetry during growth

To compare the degree of lateral asymmetry among fishes of various ages, we collected 60 largemouth bass (Micropterus salmoides) individuals of various lengths from a single population in Lake Biwa, Japan, in 2004–2008. We measured the standard length, fish age by counting circuli on the opercular bone (Menon, 1950; Campana, 2001), IAS, and θ of every individual. Then we used analysis of variance (ANOVA) to assess the effects of age on IAS and θ.

Fig. 2.

Frequency distributions of (A) the index of asymmetry (IAS) of the mandibles and (B) head angle (θ ) of the Japanese amberjack (Seriola quinqueradiata, n = 100; upper) and Amur catfish (Silurus asotus, n = 110; lower). Lines show the probability densities derived from the three models (see Methods). The model selected by the lowest Akaike information criterion (AIC) value is indicated by the thick line.

Fig. 3.

Degrees of laterality among extant teleost fishes. (A) Laterality quantified by the absolute values of the index of asymmetry (IAS) of the mandibles and the head angle (θ) in each representative species from all extant orders of teleost fishes. Index numbers correspond to the extant orders of fishes in Nelson (2006). Species are grouped into six categories (deep-sea fishes and five phylogenetic groups) indicated by colors. Each axis was standardized by the mean and standard deviation (s.d.) of all original values. (B) Comparison of the distance of laterality intensity (DLI) between the seven species of deep-sea fishes and the other teleost fishes (range, outlier, lower and upper quartiles, and median). P-values are from the Wilcoxon rank sum test. **, P < 0.01. (C) Comparison of DLI among five phylogenetic groups (range, outlier, lower and upper quartiles, and median).

RESULTS

DISCUSSION

Our results indicate that AS laterality is universally shared among fishes, but that the degree of asymmetry has become increasingly weaker in more derived groups. However, functional (behavioral) laterality seems not to be reduced even in derived groups, since many fishes have prominent behavioral laterality, including cichlids (Hori, 1993; Takeuchi and Hori, 2008), gobies (Seki et al., 2000), largemouth bass (Nakajima et al., 2007; Yasugi and Hori, 2011), and bettas (Takeuchi et al., 2010), all of which belong to derived groups. This cryptic trend in morphological laterality should be considered in depth. Here, we discuss 1) why morphological and behavioral laterality has developed in fish and why fish have AS rather than DA laterality; 2) the relationship between the two measures we used to evaluate morphological laterality, its implications for body form, and how this relates to the progressive weakening of morphological laterality in fish evolution; 3) the origin and legacy of fish laterality; and 4) the relationship between laterality in fishes and that in aquatic invertebrates.

Behavioral advantage has been proposed as a driving force behind lateral asymmetry among fishes. The general fish body plan facilitates a bilateral motor pattern. The body is propelled by side-to-side undulations. When a fish makes a dashing movement, it first bends its body either to the right or left and subsequently straightens it (Wilga, 2005). The locomotive speed of this initial motion is crucial in predator- prey interactions (Cantalupo et al., 1995; Takeuchi et al., 2012; Yasugi and Hori, 2012). If one side of the body is mechanically dominant, i.e., has more muscle and effective support tissue than the opposite side, the initial use of this side can produce stronger propulsion and higher speed than that generated by non-specialists with completely symmetrical bodies (Takeuchi et al., 2012). Therefore, morphological asymmetry and laterally disproportionate usage (behavioral laterality) may be associated and mutually reinforcing.

Although asymmetrical bodies seem advantageous, why are fish dimorphic rather than monomorphic? Behavioral laterality inevitably leads to biased directions of attack (Nakajima et al., 2007; Yasugi and Hori, 2012) and evasion (Izvekov et al., 2009; Yasugi and Hori, 2012). These biases may provoke negative, frequency-dependent selection. For instance, of the two lateral morphs among scale-eaters, the numerically dominant morph was more likely to fail at feeding than the less common morph, probably because prey fishes are more vigilant on the side more frequently attacked by the major morph (Hori, 1993; Takahashi and Hori, 1994, 1998). Among pursuit-type piscivores, predation success was higher if the laterality of the predator and its prey were opposite (so-called ‘cross predation’) than if they were the same (‘parallel predation’) (Yasugi and Hori, 2011, 2012). Conversely, when luring-type piscivores face prey fish during a predation event, parallel predation was more successful (Yasugi and Hori, 2016). This difference in success also seems to cause an alternating advantage between righties and lefties in both predator and prey fishes and results in the dynamic maintenance of laterality (Nakajima et al., 2004). This advantage is shared by predators that attack the same prey fishes and by prey fishes that share a common predator (Takahashi and Hori, 2005). Natural selection through predator- prey interactions may play a major role in maintaining anti-symmetric dimorphism in fish communities (Stearns and Hoekstra, 2000; Hedrick, 2005; Sinervo and Calsbeek, 2006). We suggest that this is the very reason why morphological laterality has progressively weakened during fish evolution, which we discuss below.

The two measures of fish laterality used here represent the most distinctly asymmetric features. The morphological implications of head angle are intuitive. If the body is asymmetric due to differential development of the right and left sides, the more developed side may be structurally dominant over the opposite side. The more developed side should, therefore, be convex, as suggested in Fig. 1. Zebrafish were found to have a trunk muscle volume that was laterally biased in relation to the laterality of the fish (Heuts, 1999), although there was a significant bias towards escaping in one direction at the population level. The morphological implication of the IAS of the HMPE can be also understood by applying the principle of leverage; i.e., the HMPE acts as a line between the effort point (i.e., where the ligament is attached) and the fulcrum point of the lever (i.e., the articulation of the mandible) (Westneat, 1990). Thus, the difference between the right and left HMPE may cause the force and speed of opening to differ between the right and left mandibles, causing the mouth to twist as it opens. Because of this morphological asymmetry, the fish can open its mouth toward one side to a greater extent than through head inclination alone. In any lateralized behavior, these factors may work together. For example, in the shrimp-eating cichlid (Neolamprologus fasciatus; Takeuchi and Hori, 2008) and the scale-eating cichlid (Perissodus microlepis; Takeuchi et al., 2016), each morph exhibits an advantage (greater hunting success) when the fish uses the dominant side of its body during foraging. The direction to which the body was bent, the eye used to aim at the target, and the dominant side of the mandible were all consistent.

The negative relationship between the two measures found among extant bony fishes (Fig. 3A) suggests a tradeoff between asymmetry in the mandibles and head angle. Although this relationship may reflect the constraints imposed by morphogenesis, it may be ultimately affected by differences in lifestyles and modes of locomotion among fish groups. Fishes with a large head angle, such as silvereye, John Dory, and puffer fish (#42, 54, and 60, respectively, in Fig. 3A), are deep-bodied, whereas those with high IAS values, such as herring, salmon, and pickerel (#27, 35, and 36, respectively, in Fig. 3A), have slender bodies (defined as having a high body length to diameter ratio). Perciforms are highly variable in body shape. Among them, slender fishes, such as amberjack, are adapted to prolonged high-speed swimming and a pelagic lifestyle, whereas deep-bodied fishes, such as sea bream, are adapted to quick turns and a demersal lifestyle (Azuma, 2006). This relationship between body shape and swimming style may apply to all bony fishes except those with extreme body shapes, such as deep-sea fishes and eels. Thus, we propose that deep-bodied fishes are suited to tilting the head to either side during asymmetric behaviors, whereas slender fishes are suited to having asymmetric mandibles due to their streamlined bodies and smaller head angles adapted to rapid and prolonged cruising. A more symmetrical bodyline is adaptive for high-speed swimming (Gosline, 1971). One of the fastest-moving fishes, the flying fish Prognichthys brevipinnis, has the smallest head angle ( Supplementary Table S1 (10.2108.zsj.34.267.s1.pdf)).

The universality of morphological laterality in fishes suggests a monophyletic origin. We hypothesize that the evolution of fish laterality began with lancelets or agnathans, and that laterality has been maintained over the course of fish evolution. Ecological factors, particularly predator-prey interactions, are likely key drivers in the evolution of lateral dimorphism in every fish population, which should be maintained by frequency-dependent selection mediated through minority advantage (Hori, 1993; Stearns and Hoekstra, 2000; Hedrick, 2005; Sinervo and Calsbeek, 2006). Although laterality is inevitable for effective locomotion in every fish species, they must also conceal their own laterality from appearance, as such laterality would indicate a preferred attacking direction of predators or escape direction of prey. Laterality may therefore have become increasingly cryptic during fish evolution. In this context, the extreme asymmetry in deep-sea fishes (Fig. 3A) may be explained as follows; since they live in the dark and rely only slightly on vision, they have little need to conceal their morphological laterality.

In addition to the origin of laterality, the remarkable laterality of sarcopterygians (coelacanths and lungfish; #61 and 62, respectively, in Fig. 3A), which share a common ancestor with terrestrial tetrapods (amphibians, reptiles, birds, and mammals), provides another implication for the universality of laterality, particularly with respect to the traits inherited by tetrapods from fish. Lateralized behaviors have been evaluated in animals ranging from fish to mammals (Roger and Andrew, 2002) with emphasis on brain lateralization, which is thought to facilitate enhanced performance or faster responses (Vallortigara and Rogers, 2005; Dadda et al., 2010). By contrast, the laterality of behaviors in fishes, including hunting techniques (Hori, 1993; Nakajima et al., 2007; Takeuchi and Hori, 2008; Hata et al., 2011; Yasugi and Hori, 2012; Takeuchi et al., 2012, 2016), scouting behaviors (Matsui et al., 2013), and social behaviors (Takahashi and Hori, 2008; Takeuchi et al., 2010), are correlated with the morphological laterality that we have demonstrated here. This documentation of morphological laterality in fish may lead to a deeper understanding of laterality in other vertebrates.

Morphological AS has also been known in aquatic invertebrate groups, including Arthropoda and Mollusca (Palmer, 2005). Our recent studies on shrimp (Takeuchi et al., 2008), crayfish (Tobo et al., 2012), and cuttlefish (Lucky et al., 2012) demonstrated that their morphological AS is associated with behavioral laterality, which is involved with escaping from predators or attacking prey. Like fish, these animals move actively in aquatic environments. They suffer minimal effects from gravity due to buoyancy, which may allow them to develop bilateral body asymmetry. By contrast, in terrestrial environments, the effect of gravity may force animals to have a laterally symmetric body supported by limbs in a balanced arrangement. Morphological laterality in terrestrial animals may thus have been weakened, although functional (behavioral) laterality seems to remain, as many studies indicate (e.g., Rogers and Andrew, 2002). Although it is difficult to speculate on the phylogenetic relationship of laterality between aquatic invertebrates and fishes, it is possible that all members of Bilateria share laterality as a homology, given the commonness of morphological laterality among invertebrate phyla (Palmer, 2005). In addition, Arthropoda and Mollusca are phylogenetically older than fishes (Moody and Zhuravlev, 2001), and behavioral laterality seems to have already existed among Cambrian trilobites and/or their predators (Babcock, 1993); fish did not appear until the end of this period. Therefore, it is possible that laterality developed first in Arthropoda, and then probably in Cephalopoda (e.g., Nautiloidea in the Ordovician: Babcock, 1993). The Cambrian also saw the explosive development of predator- prey interactions, which were driven by newly acquired eyesight in most animal phyla (Gould, 1989). We speculate that the development of the anti-symmetric laterality in Bilateria may be closely linked to such interspecific interactions where appearance is crucial to detect prey and/or escape from predators. Although it is unclear whether the laterality of fish is homologous with that of aquatic invertebrates, the anti-symmetric laterality of fish may have been developed through predator-prey interactions in ancient aquatic communities. Understanding the laterality of fish may also shed new light on the community structures of aquatic ecosystems.