Most multicellular animals exhibit one of two forms of symmetry: radial, in which multiple planes of symmetry can be drawn across an organism; or bilateral, where a single plane of symmetry, the sagittal plane, bisects an organism into mirrored halves . Bilateral symmetry is a synapomorphy of the bilateria, a taxonomic group that encompasses most animal phyla. Despite its utility as a diagnostic character, however, symmetry is not ubiquitous across all organ systems. For example, there is marked asymmetry in the patterning of the brain, heart and visceral organs in vertebrates and the genes that regulate the asymmetric morphogenesis of these structures (for example, nodal, lft1, pitx2) are well known . Likewise, a myriad of asymmetries have evolved in normally paired structures among various vertebrate lineages. Owls have evolved asymmetrical ears, which differ in size and placement on the skull, making them more effective auditory predators . The eyes of flatfish migrate over the midline of the body during development and the oral jaws develop asymmetrically such that adults can lie on the benthic substrate and ambush prey . Conspicuous craniofacial asymmetries are also evident in narwals , fruit bats  and a group of snail-eating snakes . The prevalence of laterality in nature suggests that bilateral symmetry may, in fact, be more superficial than originally thought and, while much is known about the developmental genetic basis for normal asymmetric development of the visceral organs and brain, comparatively little is known about the genetic basis of laterality in normally paired structures.
Asymmetries are typically differentiated according to their causal origin and tend to be grouped into three classes. The first is fluctuating asymmetry, where the breaking of symmetry is a consequence of developmental 'noise' and lacks a strict genetic basis. Asymmetries of this type are normally distributed around a mean symmetrical form . The second type of asymmetry is antisymmetry, where the nature of the asymmetry (that is, which traits are affected) is genetically determined, but the side in which the trait manifests itself is purported to be environmentally determined [9, 10]. The random environmental determination of handedness results in an equal bimodal distribution on either side of a symmetrical mean . The third category is directional asymmetry, in which both the trait of interest and handedness are genetically determined. Directional asymmetries are found in populations as skewed unimodal distributions, with populations biased towards a particular side. The evolution of directional asymmetries is thought to occur either directly from a symmetrical ancestor or as a progression from symmetry to antisymmetry to directional asymmetry [9, 11].
Scale-eating cichlids from Lake Tanganyika present a striking example of an evolved asymmetry in mouth direction. The Perissodini are a monophyletic cichlid tribe from Lake Tanganyika whose evolutionary history is marked by an ecological expansion from a deep-water generalized predator to shallow-water specialists that feed almost exclusively on scales (lepidophagy) . Within the Perissodini, Perissodus species exhibit jaw asymmetries that are dimorphic [13–18], with mixed populations of both 'lefty' individuals, that attack the left side of prey species with mouths angled off to the right, and 'righty' individuals, that correspondingly attack the right side with mouths bent to the left . The nature of this asymmetry has been attributed to sided differences in the length of the jaw joint (that is, the left-side is longer in lefty individuals)  but little more detail has been offered.
In P. microlepis lefty and righty morphs are maintained through frequency-dependent selection, where the minority morph experiences a higher fitness than the majority morph as a consequence of a preferential prey avoidance of the more abundant morph . The relative frequency of each morph fluctuates around a mean of 0.5 and the presence of both morphs appears to be an evolutionary stable state . This system is a commonly cited example of antisymmetry, as jaw asymmetry is bimodal and there appears to be no species level bias in handedness . Although the handedness of antisymmetric traits is generally assumed to be environmentally determined, a genetic basis has been suggested for jaw laterality in P. microlepis [14, 19], as well as for the freshwater goby Rhinogobius flumineus  and the herbivorous cichlid Neolamprologus moorii , suggesting widespread heritable laterality in mouth direction among Perciformes. Jaw laterality, therefore, does not appear to fit neatly into any one category of asymmetry . The bimodal distribution of mouth direction in P. microlepis populations is what would be expected for an antisymmetric trait . However, a genetic basis for jaw handedness is more consistent with a directional asymmetry . This apparent paradox has led to some debate in the literature concerning the causal origin of this trait .
Here we explore the evolution of craniofacial morphology and laterality among scale-eating cichlids. Using geometric and traditional morphometric techniques, we show that the evolution of the Perissodini involved discrete shifts in craniofacial shape that are correlated with foraging habitat and that sided differences in craniofacial anatomy are evident in certain species that feed exclusively on scales, consistent with the lateralization of feeding mechanics. In P. microlepis, we observe jaw laterality early in development and identify a conserved locus segregating with craniofacial handedness in East African cichlids. Our data are consistent with the hypothesis that jaw laterality evolved in Tanganyikan scale-eaters due to selection on a conserved locus for handedness and that, as species became increasingly specialized to feed on scales, selection favoured an elaboration of this asymmetry through the evolution of sided differences in jaw shape.