Adaptive radiations arise through the rapid divergence of an ancestral species into a multitude of morphologically and ecologically differentiated taxa . This process is assumed to be driven by divergent natural selection and ecological speciation where the adaptation to different niches eventually results in the evolution of reproductive isolation . For example, specialization to certain food resources might lead to divergent habitat preferences, which in turn might isolate the populations reproductively [reviewed in ]. Specialization in diet is usually accompanied by morphological adaptations facilitating resource exploitation as has been shown in some textbook examples of adaptive radiation, e.g. the Darwin finches on the Galapagos Islands , the cichlid fishes in East African lakes [5–7], or the cosmopolitan tiger beetles .
Often, adaptive radiations are triggered by an altered adaptive landscape providing opportunity to invade previously not encountered ecological niches (e.g. after colonization of a new environment) or not accessible niches (e.g. after evolution of a 'key innovation') [9, 10]. Recent studies showed that these adaptive peak shifts might happen rapidly [reviewed in ], and raise the question of how the adaptive morphological change drives the shift from one peak to another on the adaptive surface [12, 13]. Mutation in coding and regulatory sequences and selection might not be sufficient to explain the rapidity of ecological adaptation seen in some instances . Adaptation from standing genetic variation is also not likely to apply to all cases of adaptive radiations, particularly those with only a small number of founders . Adaptive phenotypic plasticity might play a key role allowing populations to enter the 'realm of attraction' of a new adaptive peak, in which genetic assimilation occurs through directional selection favoring genotypes that produce even more extreme phenotypes than what would be possible by plastic response of the ancestral genotype alone [16, 17]. Baldwin discussed this topic already in 1896 and described it as 'a new factor in evolution' [18, 19]. Although its importance meanwhile became evident, phenotypic plasticity and genetic assimilation were dismissed as being unimportant during the modern evolutionary synthesis . There has been a recent resurgence of interest in these phenomena [21–25], but the link to diversification is still little explored and under debate [26–28]. Not many investigations of phenotypic plasticity in model systems for speciation research, such as cichlid fishes, have been attempted (but see [29–33]).
The Neotropical Midas Cichlid species complex (Amphilophus spp.), is recognized among evolutionary biologists for its rapid phenotypic diversification and speciation [6, 34]. This species complex has its center of its distribution in Nicaragua, and is comprised of an array of very young species that inhabit both the large Nicaraguan lakes, and several volcanic crater-lakes that contain small scale adaptive radiations [35, 36]. The large Nicaraguan lakes, characterized by relatively turbid and shallow waters, have repeatedly acted as source populations for the colonization of nearby crater-lakes newly formed in the calderas of extinguished volcanoes. In these lakes the Midas cichlids encountered novel environmental conditions - i.e. presence of deeper zones and clearer water - and speciated in situ [34, 35, 37–41]. Crater-lake species have separated along depth and benthic-limnetic axes [34, 35], with the open water column apparently being the first novel habitat invaded. Also, the Midas cichlid species have differentiated in their trophic adaptations. Usage of food sources like stonewort, Aufwuchs, evasive invertebrate prey, fish or snails differs species-specifically . The Midas cichlids species, as well as other Neotropical and Old World cichlids, often differ in the relative degree of hypertrophy of a second set of jaws in the throat - the pharyngeal jaw - derived from branchial arch components and important for food mastication [reviewed in ]. Specialization for feeding on hard-shelled prey like snails, mussels, or crustaceans (durophagy) through this hypertrophy of the pharyngeal jaw apparatus (PJA) has been found to be a common axis of differentiation in crater-lake Midas cichlids as well as in other cichlid groups [5, 31, 32, 34, 42–44]. Its frequency and independency of acquisition across the phylogenetic tree suggests an important role of this adaptation in cichlid speciation [, , reviewed in ]
The Midas cichlid species in the crater lakes are often well differentiated in the trophic apparatus and only a few thousand years old [34–37]. The trophic polymorphism in the Midas crater-lake species could be derived from standing genetic variation, since the polymorphism is present in the large lakes, too [31, 32, 38, 41]. However, the probably limited number of colonizing individuals would render a scenario of the evolution of trait divergence subsequent to colonization also plausible. This scenario is arguably more likely for remote crater-lakes with a monophyletic Midas cichlid assemblage, e.g. Lake Apoyo (see ). A plausible scenario could be that the divergence in the pharyngeal jaw apparatus in the crater lake Midas cichlid species might have been initiated by phenotypic plasticity in the ancestor. Reproductive isolation might then have occurred via habitat isolation through the heterogeneous distribution of snails in Nicaragua's volcanic crater-lakes, where densities appear to be dependent on depth and substrate type . During times of low food availability otherwise opportunistic individuals adapted for durophagy might confine to areas of high snail density and thereby encounter mates non-randomly in respect to their pharyngeal jaw type [31, 32, 46, 47]. If the ancestor of derived species was phenotypically plastic in ecologically relevant traits, this plasticity might have triggered the diversification. The "flexible stem" model, proposed by West-Eberhard , predicts that the directions in phenotypic space in which plasticity is expressed influence the trajectories of phenotypic evolution via genetic accommodation, similar to evolution along "genetic lines of least resistance" . Therefore, it also predicts the outcomes of adaptive radiations seeded by the same ancestor and evolving in similar environments to be similar in terms of their phenotype composition.
In several cichlid fish species (family Cichlidae), plasticity in different traits has been demonstrated: Meyer experimentally induced changes in the oral jaw morphology in the Neotropical cichlid Parachromis managuensis by feeding different diets , a similar procedure was followed by Bouton and coworkers using the African cichlid Neochromis greenwoodi . The Lake Victoria cichlid Haplochromis pyrrhocephalus was almost driven to extinction by the upsurge of the introduced, predatory Nile perch in the 1980s, but was able to adapt morphologically to the new environmental conditions of high predatory pressure and eutrophication in only two decades . It has been interpreted that the speed and complexity of these morphological changes relied on a joined action of phenotypic plasticity and genetic change. The molluscivorous Astatoreochromis alluaudi naturally exhibits molariform pharyngeal jaws (i.e. stout, broad and strong jaw-bones with wide and flat teeth) . However, when raised on soft artificial food under laboratory conditions , in natural conditions in lakes not inhabited by snails , or in lakes inhabited by snails but also with a molluscivorous competitor present , they develop less stout pharyngeal jaws with cuspid teeth (papilliform).
Specializations matter most during ecological "crunch times", when resource availability is low and opportunistic feeding is precluded [42, 46]. The ability to exploit resources then at all or more efficiently than other species can, matters for the individual's survival. But specializations come with a trade-off. The specialization of being able to feed on particular diets especially efficiently often comes at the cost of being much less efficient when dealing with alternative diets. Apparently, such a trade-off exists in the Neotropical Midas Cichlid (Amphilophus cf. citrinellus) between two different types of pharyngeal jaws, molariform and papilliform. Individuals with papilliform lower pharyngeal jaws are more effective when dealing with soft food items . Individuals with molariform jaws, on the other hand, can crack larger and harder snail shells and do this faster than papilliform individuals .
These cases of phenotypic plasticity, the basis of lacustrine cichlid radiations on trophic specialization [44, 55, 56] and the possible causal linkage of plasticity and diversification [23, 30, 31, 57] call for examination of adaptive phenotypic plasticity in trophic traits in an adaptive radiation of cichlids comprising species differentiated in these traits. The lower pharyngeal jaw (LPJ) might constitute 'an ideal component of cichlid trophic morphology' to be investigated in this respect . Preferably, the case in study should have a known and young history, involve colonization of new habitats and tests for plasticity in the ancestral or similar to the ancestral source population.
Here, we tested in a common garden experiment the developmental plasticity of the lower pharyngeal jaw of Amphilophus citrinellus (Günther, 1864) exposed to diets differing in hardness. Earlier work  had suggested that the species in this species complex are phenotypically plastic and that the abundance of molariform fish correlates with the abundance of their major prey item, hard-shelled snails.
The experiment was performed on a laboratory stock derived from the crater Lake Masaya, which was bred in captivity for several decades. Although Lake Masaya is a volcanic crater-lake, its A. citrinellus population is very close to the populations of the Lake Nicaragua - which is probably the ancestral source population of most crater-lake radiations - in terms of body shape  and phylogenetic relationships . Furthermore, it has been suggested that Lake Masaya might have been colonized as recently as 450 years ago .
We investigated whether the development of pharyngeal jaws differed between three types of diets: (1) intact snails with shell, (2) peeled snails without shell, and (3) finely ground up whole snails frozen in pellets, from which fish could nibble off the thawed, soft outer layer when those were given into the water. We aimed to verify whether a hard diet could induce changes in the pharyngeal jaw of the fish, and whether the generation of robust pharyngeal jaws with stout teeth (molariform jaws) was determined by higher calcium content in the diet, or by mechanical stimulation of the jaws when crushing hard food items.
Our study finds that diet can induce changes on the trophic apparatus of the Midas cichlids, and that this changes are related to the mechanical stimulation of the jaws.