The innovations in the cichlid pharyngeal jaw may have facilitated the utilization of novel prey and ultimately led to the unparalleled trophic diversification of these fishes (Liem 1973). Although most bony fishes have pharyngeal gill arches modified to process prey (Liem 1986; Wainwright 1989), the cichlid pharyngeal jaw uniquely exhibits novel upper pharyngeal jaw joints, a ‘muscular sling’ and suturing between the two-fifth ceratobranchial elements (figure 1). This suturing results in a functionally fused lower pharyngeal jaw (LPJ) (Liem 1973), but the degree of fusion via suturing is highly variable (Kullander 1998). As the cichlid LPJ both resists and exerts forces during prey processing, this variability in LPJ fusion should relate to forces exerted on prey (Stiassny & Jensen 1987; Kullander 1998). To examine the trophic consequences of cichlid pharyngeal jaw fusion, I investigated whether greater LPJ suturing enhances the ability to exploit durable prey in the trophically polymorphic Herichthys minckleyi and tested the evolutionary association in Heroine cichlids between LPJ suturing and crushing hard-shelled molluscs.

Figure 1 The labroid pharyngeal anatomy, implicated as a key innovation (Liem 1973) has three basic components: (a) the muscular sling formed from the pharyngeal muscles, (b) two synovial joints in the upper pharyngeal jaw and (c) a single ‘fused’ (more ...)

Suturing integrates the two halves of the LPJ into a single unit. Because the pharyngeal jaw is the primary location of prey processing in most bony fishes (Liem 1973), pharyngeal jaw modifications frequently determine, what prey fishes can exploit (Wainwright 1989; Galis & Drucker 1996; Grubich 2003). The importance of pharyngeal jaw modifications to cichlid trophic divergence is clearly evident in species that are polymorphic in their pharyngeal morphology. For example, in H. minckleyi (Kornfield & Taylor 1983), one pharyngeal morphotype specializes in crushing molluscs (molariform), while the other pharyngeal jaw type (papilliform) utilizes prey that require less force to process (Hulsey et al. 2005). These alternative pharyngeal morphotypes (figure 2a,b) occur sympatrically (Kornfield & Koehn 1975), interbreed (Kornfield & Taylor 1983) and differ in little other than what prey they utilize (Sage & Selander 1975). Of all the fresh water prey, molluscs that molariform H. minckleyi and other cichlids crush should require the most forceful pharyngeal bite to process (Vermeij & Covich 1978; Vermeij 1987). Therefore, if greater pharyngeal jaw fusion is important in adapting the jaw to exploit more durable prey, molariform H. minckleyi should have more heavily sutured LPJs and better resist compressive forces encountered when crushing molluscs.

Figure 2 The dorsal view of LPJs of H. minckleyi. (a) The papilliforms are specialized to shred plants. (b) The molariforms are specialized to crush snails and are diagnosable by their large molariform teeth.

Adaptations for exploiting particular prey likely arise within populations (Meyer 1990; Schluter & Nagel 1995), but macroevolutionary divergence is often the best evidence available that a particular innovation was key to a clade's diversification (Simpson 1953). Linking modifications within a species like H. minckleyi to changes among species would increase the mechanistic understanding of how LPJ fusion may have generally influenced cichlid trophic divergence. As molluscivory has likely evolved multiple times in the Heroine cichlid clade that includes H. minckleyi (Winemiller et al. 1995; Hulsey et al. 2005), this cichlid lineage is ideal for examining the macroevolutionary association between greater cichlid LPJ fusion and crushing durable mollusc shells.

I examined cichlid LPJ fusion using morphometrics, empirical tests of the strength of the LPJ, diet analyses and phylogenetic comparative methods. I first quantified external suturing in H. minckleyi molariforms and papilliforms. Then I tested if, the forces that split the LPJ along the suture differed between molariforms and papilliforms and if suturing may reduce damage to the pharyngeal apparatus when this species crushes molluscs. I also quantified LPJ suturing and the extent of specialization on molluscs in 31 other Heroine cichlids and examined the phylogenetically corrected correlation between LPJ suturing and molluscivory.

Figure 3 There is substantial variation in suturing of the ventral LPJ in cichlids. In (a) A. robertsoni, the two halves of the LPJ meet but there is no external suture, while other species (b and c) exhibits extensive suturing. (b) To estimate the suture length, (more ...)

To estimate the percentage of the LPJ that was sutured, the total length of the jaw, L_T, from anterior tip of the keel to the posterior centre of the jaw was measured to the nearest 0.1mm using calipers (figure 3b). Then, the linear distance from the anterior-most initiation of suturing to the posterior-most end, L_S, along the centre of the jaw was measured (figure 3c). If, there was no suture, L_S was zero. Subsequently, the suturing of the dried LPJ was photographed using a digital microscope. The image was then imported into NIH Image (http://rsb.info.nih.gov/nih-image/about.html) and two measurements were taken. First, the winding length, L_W, of the suture was traced digitally (figure 3c) and its length calculated. Then, the linear length, L_S, between the anterior-most point, where suturing of the jaw began and the posterior-most end of jaw suturing was calculated using a straight line. The winding length was then divided by the linear length. This value was multiplied by the percentage of the length of the LPJ that was sutured in order to calculate the percentage the suture increased in the perimeter of contact, P_L (equation (2.1)), between the two halves of the pharyngeal jaw as compared to an unsutured jaw

2.1

I also determined if the LPJ in the H. minckleyi pharyngeal morphotypes resisted different amounts of compressive force. For 12 molariforms (Range: 80–147.2mm SL) and 12 papilliforms (Range: 77.1–145.4mm SL), wild-caught fishes were sacrificed with an overdose of MS222, their SL measured and their LPJ removed by dissection. Subsequently, the LPJ was laid on the lower force plate of an Accuforce Cadet force gauge (0–1000Newton, Ametek, Inc. Pennsylvania, USA) with the dorsal side of the jaw facing the upper force plate. I then placed a Mexithauma quadripaludium snail, approximately 5mm in shell length, aperture down, on the dorsal crushing surface of the LPJ to mimic the position of a snail during a pharyngeal crushing event. This snail is frequently crushed by molariform H. minckleyi (Hulsey et al. 2005). The snail and LPJ were then squeezed between the opposing force plates of the force gauge until, the LPJ split in half along the suture. Using SL as a covariate, an ANCOVA was used to test if the force recorded in Newtons (N) needed to split the LPJ of each morphotype along the suture was statistically distinguishable. Finally, the compressive resistance of each morphotype's LPJ was compared to the maximum force molariforms exert to crush snails in the wild (n=33, logN=1.34 (logSL)−0.75, R²=0.69) using an ANCOVA with SL as a covariate. This maximum molariform crushing force was estimated from snail opercula taken from gut contents of wild-caught fishes (Hulsey et al. 2005).

Figure 4 The phylogenetic relationships, evolution of molluscivory and photos of LPJ suturing in representative Heroine species. The percent contribution of molluscs to the diet of each species and the mean PL are depicted next to the species name. The subscript (more ...)

The phylogenetic independent contrast analyses were performed in several ways. First, the correlation of P_L and arcsin transformed values of molluscivory were examined using the ‘crunch’ algorithm that assumes all variables are continuous. Then, because many species had zero or low contribution of molluscs to the diet, I also examined the correlation when mollusc feeding was treated as a categorical variable. Species that included less then 2% molluscs in their diet were categorized as (0) non-molluscivores and species with greater then 2% molluscs in their diet as (1) molluscivores. Then, with the ‘brunch’ algorithm, which allows dichotomous variables to be tested in an independent contrast framework in CAIC, it was determined if P_L was consistently greater in those clades coded as molluscivores. Finally, because the P_L of the jaws in H. minckleyi appeared to be slightly correlated with SL, I reanalysed the above correlations using the residuals of a reduced major axis regression between average SL of specimens and P_L. Also, the presence of greater than 2% molluscs in the diet was mapped onto the phylogeny using parsimony (Maddison & Maddison 2000) to provide an initial estimate of the number of times molluscivory may have evolved in Heroine cichlids.

In the polymorphic H. minckleyi, the amount of suturing adds to the list of putatively adaptive characteristics that differ between molariform snail crushers and papilliform plant processors (Kornfield & Taylor 1983; Liem & Kaufman 1984; Hulsey et al. 2005). Molariform H. minckleyi possess the greatest average amount of external suturing of any Heroine cichlid examined (figures 3c and 4) and much greater suturing throughout ontogeny than papilliforms. However, it is unclear, if the amount of suturing is genetically controlled in this species or in any other Heroine. Although, the different pharyngeal morphologies in H. minckleyi appear to have some genetic basis (Trapani 2003), much of the pharyngeal jaw variation may be a phenotypically plastic response to crushing the unusually robust snail prey found in their native habitat (Vermeij & Covich 1978; Liem & Kaufman 1984; Hulsey et al. 2005).

In H. minckleyi, the LPJ consistently split along the suture prior to extensive damage occurring to the remainder of the jaw. It took substantially more force to split the molariform LPJ along the suture than the force used by molariforms to crush snails in the wild. Whereas a molariform that is 150mm in SL can produce 115N of crushing force (Hulsey et al. 2005), its LPJ can resist approximately 600N of crushing force before splitting along the suture. A similarly sized papilliform LPJ resists only about 60N before splitting at the suture. Because greater suturing should act to reduce damage to the jaw and damage to the feeding apparatus should be selected against (Vermeij 1987; Van Valkenburgh 1988), greater LPJ fusion through increased suturing may be a critical modification of the molariform jaw. The minimal fusion of the papilliform LPJ and its inability to resist substantial crushing forces may explain why papilliforms rarely utilize snails as prey (Hulsey et al. 2005). Strikingly, if papilliforms generated the forces that molariforms frequently use to crush molluscs, papilliform H. minckleyi would likely split their LPJ into separate halves at the suture.

Specialization in molluscs probably evolved several times within Heroine cichlids (Winemiller et al. 1995, figure 4). Twelve of the 32 species included herein had gut contents composed of more than 5% molluscs by volume and molluscivory may have evolved at least six times, independently. Notably, molluscivory has also evolved numerous times in other cichlid clades such as Aequidens (Winemiller et al. 1995), Astatoreochromis, Lamprologus (Liem 1973) and Serranochromis (Winemiller et al. 1995). Suturing of Heroine cichlid jaws is also exceptionally variable (figure 4) and external suturing has probably evolved and been lost multiple times in Heroines, as it has in other cichlids (Kullander 1998). Pharyngeal musculature and other skeletal elements are undoubtedly important in adapting the cichlid pharyngeal jaw to mollusc crushing (Liem 1973; Hulsey et al. 2005; Wainwright 2005). However, there is a clear relationship between greater suturing and molluscivory both among evolutionarily independent lineages of Heroine cichlids and within H. minckleyi. Micro- and macroevolutionary lability in LPJ fusion likely makes it easy for cichlids to evolve to exploit durable prey.

Cichlids share several morphological innovations in their pharyngeal jaw with wrasses (Labridae), damselfish (Pomacentridae), parrotfish (Scaridae) and the surfperches (Embiotocidae). These major groups of fishes were once included in a putatively closely related group named the Labroidei (Kaufman & Liem 1982) that is now recognized as being paraphyletic (Streelman & Karl 1997). Monophyly of the Labroidei was originally hypothesized because of their structurally similar pharyngeal jaws, although, the LPJ in non-cichlid labroids exists as a single fused bone that lacks a suture (Liem 1986; Stiassny & Jensen 1987). Like cichlids with extensive LPJ suturing, many wrasses and surfperch feed upon hard-shelled crustaceans or molluscs (Liem 1986; Randall et al. 1997) and parrotfish pharyngeal jaws are likely modified to break down extremely tough prey such as coral (Bellwood 1995). As in cichlids with extensive LPJ suturing, labroid LPJ fusion may have facilitated the exploitation of durable prey (Wainwright et al. 2004; Wainwright 2005). Most bony fishes and, therefore, the ancestors to all groups with a labroid pharyngeal jaw likely had a divided LPJ with no suturing (Liem 1986; Wainwright 1989). However, LPJ fusion through suturing is not limited to cichlids. Suturing of the two ceratobranchials composing the cichlid LPJ resemble the condition found in a several non-Labroidei fish clades such as Gerres, Lepomis and Pogonias that lack a muscular sling but frequently consume molluscs (Stiassny & Jensen 1987; Galis & Drucker 1996; Grubich 2003). The repeated coupling of greater LPJ fusion and durophagy during teleost evolution makes the apparent pharyngeal jaw convergence between cichlids and other labroids functionally less surprising.

Both the splitting of LPJs in H. minckleyi and the phylogenetic comparative analyses indicate a more extensively fused jaw is advantageous for the exploitation of durable prey. However, the variation in cichlid suturing suggests there are probably disadvantages to completely fused jaws. When manipulating prey, less fusion might facilitate independent movement or greater bending of the two halves of the LPJ (Wainwright 2005). Fused jaws may also decrease the pharyngeal gape and hinder swallowing larger prey like fishes (Wainwright 1987). Piscivory is rare in the other labroid groups (Randall et al. 1997; Wainwright et al. 2004) and cichlid groups that are primarily piscivorous like Cichla, Crenicichla and Petenia (Chavez-Lomeli et al. 1988; Winemiller et al. 1995; Hulsey & García de León in press) all have limited LPJ fusion (Kullander 1998). Having a flexibly sutured LPJ as opposed to a single bony element may have permitted cichlids to exploit both large prey types such as fishes as well as hard-shelled prey like no other group of aquatic vertebrates.

Evaluating if a complex key innovation (sensu Simpson 1953) like the cichlid pharyngeal jaw is responsible for trophic diversification is difficult. Evolutionary novelties in organismal design are not always replicated (Levinton 1988; Padian 2001) and novel structures cause changes that could themselves be interpreted as innovations (Cracraft 1990). However, when variability in transitional phenotypes exists, what favoured the evolution of a putative innovation can be tested (Greene 1983). Furthermore, verifying that a phenotype is favoured under realistic ecological circumstances can provide support for the hypothesis that particular character states, like greater LPJ fusion, are crucial to specialization on particular prey (Wainwright 1987). In order to evaluate the evolutionary and ecological consequences of their unique trophic apparatus, it may be key to more critically assess the structural and functional consequences of diversity in the cichlid pharyngeal jaw.

F. J. García de León and his students helped to collect fishes in Mexico. B. O'Meara, R. Carlson, P. Wainwright, T. Streelman and T. Near provided comments. An NSF Predoctoral Fellowship, the UC Mexus Doctoral Fellowship and the National Geographic Society provided support. The Field Museum of Natural History, Texas Memorial Museum and Florida Museum of Natural History provided specimens. Cichlids were collected under scientific permits issued by the Mexican Government (PPF 230401-613-03 and DAN 02939).