The dental system of †Kazanichthys viatkensis (Actinopterygii, Acrolepididae) from the middle Permian of European Russia: palaeobiological and palaeoecological inferences

Among ray‐finned fishes (Actinopterygii), the crushing, durophagous feeding strategy first evolved in the early Carboniferous period, with the †Eurynotiformes possessing dentitions with single layers of partially to fully fused blunt teeth. In the †‘Platysomidae’ (Permian), a new form of crushing dentition evolved (phyllodonty), in which multiple layers of superimposed crushing teeth developed intraosseously, within the jaw. The phyllodont durophagous dentition is also recovered from later‐occurring taxa originating mainly in the Mesozoic, such as the †Bobastraniiformes, the neopterygians †Pycnodontiformes and Ginglymodi, and in the teleost group †Phyllodonta. By comparison, †Kazanichthys viatkensis, an actinopterygian from the middle Permian of European Russia, is characterized by a third, putatively durophagous dentition, with anterior conical teeth and closely packed molariform teeth on the buccal dental plates (a potential similarity with eurynotiforms). Whereas the conical teeth are similar to those of basal actinopterygians, the molariform teeth superficially resemble teeth of some teleosts (Characiformes, Tetraodontiformes), but are unique among known fossil and living Actinopterygii in being crowned by anastomosing, sharp apical ridges. Teeth are ankylosed to the jaw and acrodont in implantation. There is neither evidence of plicidentine, nor cavities corresponding to intraosseous crypts. Most replacement teeth formed extraosseously, differing from the phyllodont dentition, but similar to several more phylogenetically basal actinopterygians. The dental system morphologically resembles recent Sparidae (Teleostei; Perciformes), possibly indicating a similar trophic adaptation. Based on these comparisons and patterns of wear, we propose that †K. viatkensis was a generalist durophagous feeder, with the ability to switch prey types with its unique and complex dentition.

T H E earliest examples of durophagy (crushing prey items) among actinopterygian fishes occurred in the early Carboniferous, after the Hangenberg extinction event (Sallan & Coates 2013;Salamon et al. 2014;Friedman et al. 2018), with a later rapid increase in durophagous actinopterygian diversity occurring at the beginning of the Mesozoic (B€ urgin 1999;Bellwood 2003). Among Permian fish, the †'Platysomidae' is traditionally referred to as durophagous. Some taxa, such as †Platysomus schultzei, †Schaefferichthys leuderensis, †Bobastrania ladina, †B. antiqua and †B. morderi have phyllodont tooth plates (Zidek 1992;B€ ottcher 2014), and this type of tooth plate is also seen in numerous unassigned tooth plates (Johnson & Zidek 1981;Schultze 1985;Fracasso & Hovorka 1987) or in those provisionally referred to ' †Platysomus' (see Zidek (1992) for a review), consisting of several layers of superimposed, rounded teeth. Others such as †Platysomus deposits (Minikh & Minikh 2009;Wordian, lower Guadalupian). Most species are known only from fragmentary remains, whereas relatively complete (but mostly damaged) skeletons of †K. viatkensis are known. †Kazanichthys viatkensis differs from †K. golyushermensis in the less pronounced sculpturing of the scales, a weakly dissected free field, and very short interridge grooves. It differs from †K. uralensis in the smaller size of the bone protruding below the ganoine layer along the lower and posterior margins of the scale and in a smaller number of pores on the free field (Bakaev 2022). Unfortunately, all known skeletons were seriously damaged during excavation, therefore there is no qualitative description of the skull. Kazanichthys was interpreted as a Permian actinopterygian fish from the East European Platform specialized in durophagy (Esin 1995a(Esin , 1995b(Esin , 1997. However, this conclusion was made on the basis of only the 'molariform' tooth shape, without consideration of the dental system as a whole. It has been shown that the shape of fish teeth alone is a poor predictor of feeding mode, given that fishes with similar dentitions may have different feeding modes, or those with disparate teeth may feed on the same food types (Harder 1975;Machado-Allison & Garcia 1986). There are several ways to reconstruct the diet of extinct vertebrates: analysis of the fossilized remains preserved in the gastrointestinal tract (e.g. Kogan & Licht 2013;Bajdek et al. 2016;Nied zwiedzki et al. 2016); functional morphology of the feeding apparatus (e.g. Lauder 1980Lauder , 1982Taylor 1992;Vullo et al. 2017;Ziermann et al. 2019); stable isotope analyses (Roche et al. 2013); and inferences from the body shape (Kogan et al. 2015) and from extant relatives (e.g. Grande & Bemis 1998). However, detailed morphological analysis of the jaws and teeth, as well as dental wear, still provide important clues as to the range of possible food sources for a given fossil taxon, and they can also inform on dietary changes associated with an animal's life cycle. For this reason, we present the first indepth investigation of the dental system, based on better preserved skeletons along with isolated remains, with the aim of determining the trophic adaptations of †K. viatkensis. Notable differences with respect to other durophagous fish dentitions, for example, the dominance of extraosseous tooth development, signal a new style of durophagy in this Permian taxon.

RESULTS
A completely preserved skull of †K. viatkensis is unknown. Only the anterior part of the upper jaw is preserved in the holotype (PIN 5802/15; Fig. 2). More complete skulls are known but are damaged, making a complete reconstruction of the jaws and interpretation of the outlines of individual bones difficult. Despite this, the available material shows the main features of the dental system. The teeth can be classified according to shape and size into two main types: conical and molariform (e.g. Burress 2016, fig. 1). The conical teeth are unicuspid (Figs 3-6); a smooth acrodin cap and collar ganoine bearing longitudinally elongated microtubercles are visible (Fig. 3D, F, G).
There are differences in the morphology of the conical teeth, which can be divided into caniniform and small conical teeth. The caniniform teeth are massive, broadbased, with a finely pointed acrodin cap that occupies approximately 25% of the tooth length (Fig. 3G). Their shape ranges from flattened conical and straight (Fig. 3F) to acute conical, and lingually curved (Figs 3C, 5E). These are present on the premaxilla, maxilla (Fig. 3A) and dentalosplenial (anteriorly; Fig. 5E). The outer dental arcade bears only one caniniform tooth row, as can be seen in the maxilla of the †K. viatkensis holotype (Fig. 3A, C, E). The small conical teeth are straight, short, with a distinct acrodin cap that occupies approximately 25% of the tooth length. Below the cap, the shaft is distinctly expanded. They are not arranged in distinct rows (Fig. 3) and they are attached to the premaxilla as well as to the dentary and the coronoids of the lower jaw (Fig. 5E), and also the anterior dermal bones (presumably vomer, dermopalatine and ectopterygoid) of the palate of the upper jaw. The small conical teeth are not distinguished by a clear boundary from pointed molariform teeth (see below), and the teeth gradually change from one type to another moving rostrocaudally.
The molariform teeth are divided into pointed and blunt shapes. The pointed molariform teeth are superficially similar to the small conical teeth, but much larger, massive and broader. A distinct acrodin cap occupies approximately 10% of the tooth length (Figs 6-8). Below the cap, the shaft expands sharply. The blunt molariform teeth are massive and broad, crowned by raised, anastomosing sharp ridges . The contact between the acrodin cap and collar ganoine is indistinct and difficult to trace, suggesting that these teeth lack the acrodin layer. The molariform teeth of both subtypes are randomly arranged and so closely packed that some teeth in the middle of the plates have irregular or polygonal shapes in occlusal view, presumably due to crowding during tooth replacement events (Figs 6B, E, H, 7D, J, 8D, E, G). The molariform teeth are attached to the buccal dental plates , which presumably represent the prearticular of the lower jaw and dermometapterygoid and entopterygoid of the upper jaw. Molariform teeth are also attached to the posterior part of the maxilla and dentalosplenial (Fig. 6E, F).
In addition to teeth, there are small, conical, lingually oriented odontodes with acrodin caps. These structures are distinguished from the true teeth by the lack of any signs of replacement, and by their position relative to the rest of the dentition: they are restricted to the dermal bones of the jaws, in the vicinity of, but external to, the dentition (maxilla, Fig. 3A, B). Odontodes from the back of the jaws do not have an acrodin cap, but are crowned with short, anastomosing ridges, similar to the molariform teeth closest to them.
In most specimens we observed the functional tooth replacement pits of shed teeth (Figs 3E, 8E, G, I), distinguishable from postmortem breakage by the absence of tooth walls with rectangularly broken edges and the presence of numerous pits with scalloped borders (i.e. traces of resorption) (Fig. 3E). In some specimens the isolated acrodin caps (indicated by a smooth surface), with small parts of the tooth shafts, but without tooth bases, are visible in the empty replacement pits (Figs 3A, C, E, 8G, I) or lingual to the outer marginal row (Fig. 3A, C, E). We interpret these as replacement teeth, which were set to replace functional teeth before the animals died (see below).
The conical teeth have only minor traces of in vivo wear. The most robust molariform teeth show slightly more significant in vivo wear, but some dental plates have much stronger wear: acrodin caps are completely abraded, which results in the formation of a continuous flat surface ( Fig. 8B, E, F), oriented nearly perpendicular to the long axis of the tooth and parallel to the basal bone. In some cases, the wear reaches the dentinal tubules. We think that this wear was not postmortem because only the hardest parts of the dental plate (the apexes of the teeth) are heavily worn, while the softer bone base does not show signs of wear.
There is a clear external boundary between what we interpret as the collar ganoine and the acrodin cap. However, the acrodin cap is indistinct in blunt molariform teeth. We propose that the acrodin cap is either not formed or is completely covered with collar ganoine in these teeth (Shellis & Miles 1976;Smith 1992;Richter & Smith 1995;Sasagawa et al. 2012). Development of these has been investigated in extant actinopterygians, which can be compared to †K. viatkensis. In these extant taxa, the collagenous matrix of the acrodin cap is formed by the odontoblasts (ectomesenchymal cells that also form the dentine) with some contribution from the ameloblasts (epithelial cells that typically form enamel in other vertebrates). Matrix vesicles associated with the odontoblasts, and the crystals within the vesicles, mineralize the acrodin region, contributing to the formation of enameloid. The cells of the dental epithelium remove the collagen at a later stage and stimulate further crystal growth. Dentine is deposited during late enameloid formation (Sasagawa et al. 2009), while the matrix of collar ganoine is deposited by dental epithelium cells upon the surface of acrodin (usually covering only the base of the acrodin cap) and at the boundary between the epithelial cells and the dentine-depositing odontoblast layer (Sasagawa & Ishiyama 2005;Sasagawa et al. 2009Sasagawa et al. , 2012. Therefore, the boundary between the acrodin and dentine is usually not clear, the transition is smooth, but the boundary (both external and internal) between the collar ganoine and the acrodin/dentine is clear (Figs 3G, 6C, 7H). Thus, the collar ganoine covers the lower part of the formed acrodin and dentine, which leads to a usually obvious external boundary between collar ganoine and acrodin (Sasagawa et al. 2012).
In the extant Polypterus, initially (during the formation of the acrodin cap and dentin shaft), the external surface of the tooth germ is covered with a layer of differentiated ameloblasts, which, as described above, are involved in the maturation and mineralization of the acrodin cap. The ameloblasts then produce the collar ganoine matrix, after which most of the ameloblasts are involved in ganoine maturation. However, some of the ameloblasts (which cover the acrodin cap) begin to degrade (Sasagawa et al. 2012). For this reason the matrix does not mineralize and the acrodin cap is not covered with collar ganoine. However, the blunt molariform teeth of †K. viatkensis are completely covered with ganoine ( Fig. 7H), which indicates a deviation from the standard basal ray-finned fish process. Unfortunately, with our current material we cannot tell whether acrodin is present or has been lost. In the basal actinopterygians, both acrodin cap and collar ganoine are present (e.g. Kawasaki et al. 2021, fig. 1) in Polypteridae (e.g. Polypterus) and in Holostei (e.g. Lepisosteus (Kawasaki et al. 2022) and Amia (bowfins)), but are lost in Acipenseriformes (Acipenser, Polyodon).

Tooth attachment and implantation mode
Actinopterygians have four classical modes of tooth implantation: protoacrodont, acrodont, acro-protothecodont and pleurodont (Gaengler 2000). However, as noted by Bertin et al. (2018), tooth implantation should not be conflated with tooth attachment; the former categorizes the geometry of tooth attachment (depth and symmetry), whereas the latter characterizes the nature of that attachment (fused or fibrous). Like most early actinopterygians, the acrodont implantation mode is seen in †K. viatkensis, in which the base of the tooth does not extend into the underlying jawbone and is instead attached at the crest of each tooth-bearing element. The attachment mode of †K. viatkensis teeth is by ankylosis (fusion of the tooth to the bone by mineralized attachment tissue), which is the plesiomorphic condition for actinopterygians (Fink 1981;Bemis et al. 2019) and is characteristic of most fossil osteichthyans as well as recent polypterids and lepisosteids (e.g. Germain & Meunier 2017; Fig. 9A-B, D-F), and Amia. In ankylosis the dentine tooth base co-ossifies with the jaw bone via an intermediary attachment tissue, classically referred to as 'bone of attachment' (Fink 1981;Gaengler 2000). Ankylosis is often associated with the presence of plicidentine, which may increase the surface area for tooth attachment and better resist stresses associated with feeding in ankylosed teeth (Preuschoft Germain & Meunier 2020). Previously, it was suggested that the simplexodont plicidentine (a subtype of dentine folding pattern) is linked to the durophagous diet. But we were unable to find plicidentine of any kind in teeth of †K. viatkensis, which should be observable in jaws with broken teeth (Figs 3E, 6I, J, 7H). Nevertheless, the shagreen described above may represent the location of a comparable intermediary attachment tissue.

Tooth replacement mode
In the majority of both modern (  (Figs 3, 8G, I); this appears to be the dominant form of tooth replacement in †K. viatkensis, and involves the conical teeth (Fig. 3) as well as the more molariform teeth (Fig. 8). Apparently, a similar tooth replacement process is seen, for example, in Erpetoichthys and Lepisosteus; CT scans of the skulls and jaws of these taxa show very clearly the functional teeth in the jaws (Fig. 9A, D; Johanson 2023), and associated with these, mineralized tooth tips. In Erpetoichthys these are consistently positioned at an angle with respect to the functional teeth (e.g. Fig. 9C, small red arrows). One tooth captured in section, appears to show development of the replacement tooth towards the base of the former functional tooth (Fig. 9C, white arrow). Along with these mineralized tips, several of the functional teeth show external resorption, which eventually breaches the pulp, forming a resorption pit (Fig. 9C, red arrowheads), occurring consistently at the base but sometimes occurring further up the tooth. As described in the closely related Polypterus, marginal tooth germs of replacement teeth form posterolingually to the functional teeth and when the teeth are resorbed and shed, replacement teeth migrate in their position and become ankylosed to the bone; tooth germs develop only after resorption of the previous tooth, on top of existing pulpal openings (Clemen et al. 1998;Wacker et al. 2001;Vandenplas et al. 2014). In Lepisosteus (Fig. 9D-F), comparable extraosseous tooth replacement is occurring, including one large functional tooth that is being resorbed at its base, with plicidentine still remaining on the jaws. A very small mineralized tooth tip is developing in association with this resorbing tooth (Fig. 9F). More posteriorly, there are two positions where it appears that the functional tooth has been completely lost, and the replacement teeth are more developed and being added to a round shallow concavity, representing the previous point of attachment of the functional tooth. Given these comparisons, it seems that tooth replacement in †K. viatkensis is extraosseous, as in primitive F I G . 9 . Extraosseous tooth replacement in non-teleost actinopterygian fishes, CT scans. A-C, Erpetoichthys calabaricus Smith, 1865: A, skull, right lateral view; B, close-up of upper and lower jaws, small red arrows indicate mineralized tips of newly developing teeth; C, virtual section through the skull to expose the internal jaw surfaces; red arrows indicate new teeth; red arrowheads indicate functional teeth being resorbed, beginning near their bases; white arrow indicates a tooth captured in section that appears to show development of the replacement tooth towards the base of the former functional tooth. D-F, Lepisosteus sp.: D, skull, right lateral view; E, close-up of jaws; small red arrows indicate mineralized tips of newly developing teeth; F, close-up of jaws, slightly rotated to show dorsal lower jaw; from left to right, the first two small red arrows indicate mineralizing teeth developing towards the concavity from which the functional tooth has been lost; when the tooth base forms, acrodont attachment to the jaw will occur; the third red arrow shows the functional tooth being resorbed at its base (with plicidentine remaining at this stage); there is a small mineralized tooth tip nearby, representing the replacement tooth. White arrows indicate cranial direction. Scale bars represent: 2 mm (A-C); 5 mm (D-F).
actinopterygians. Lateral surfaces of broken bones do not show any traces of bony crypts below the functional teeth, or of buried tooth-like odontodes. Moreover, the shallow dental plates do not provide enough space for intraosseous tooth development, further supporting our interpretation of extraosseous tooth formation and complete basal resorption of the teeth in †K. viatkensis. Also, there is no evidence of gap-filling tooth replacement in †K. viatkensis (large, damaged teeth being replaced by multiple small teeth), as was recently described for the pycnodonts, unique for this particular group of fossil neopterygians (Collins & Underwood 2021). Instead, †K. viatkensis had one-for-one tooth replacement. We suggest that the extraosseous manner of tooth replacement in †K. viatkensis indicates that the replacement tooth germ either developed without the contribution of a successional or permanent dental lamina, or else with a very short one, compared with intraosseous development in which the dental lamina extends into the bony jaw (Bemis & Bemis 2015). Although mineralized tooth crowns along the jaw were suggested to indicate extraosseous replacement (Trapani 2001), the position of a replacement tooth crown close to the base of a functional tooth also supports this hypothesis (Fig. 3A, E, F). When the dental lamina is absent or short, the germ of the new tooth forms at the base of the functional tooth and then resorbs the functional tooth before moving into position (Vandenplas et al. 2014; Bemis & Bemis 2015). Resorption pits can best be seen as characteristic scalloped margins at the base of the tooth in Figure 3E, or as aggregates of large Howship's lacunae, which formed polygonal depressions on the external tooth surface in Figure 6J. These two patterns of resorption may indicate osteoclast activity both inside the tooth and externally; the latter can occur as a result of resorption in one tooth also affecting those on either side, as demonstrated recently for reptiles (LeBlanc et al. 2021 fig. 2A). We suggest that the tooth replacement mode in †K. viatkensis was probably close to the pattern of Polypterus: the tooth row was functioning simultaneously, usually without free replacement pits (Fig. 3B-E). The exchange of teeth occurred in every second position in a relatively short period.

Tooth-like odontodes
Tooth-like odontodes are present near the main dentition in †K. viatkensis (Fig. 3), but it is not clear whether they are growing over the dentition, which is characteristic of early bony fishes (e.g. Chen et al. 2016, 2020) аnd some early tetrapods (Haridy et al. 2019). In †K. viatkensis there is a clear morphological distinction between these odontodes closer to the dentition, which are tooth-like in morphology, and those further from the dentition. These have a very different morphology, being larger and less tooth-like (Fig. 3A, B).

Functional interpretation
Dietary reconstructions of fossil individuals based on stomach contents are relatively common and can provide information on feeding (see Klug et al. 2021 for a recent review), nevertheless this information is incomplete, which also is the case even in extant fish research (Bellwood et al. 2019). Even when complete information is available for the teeth, the actual behaviour can be quite different from the interpretation (Mihalitsis & Bellwood 2021 for an example in carnivores). Investigators assume that sharp conical teeth are for puncturing, but in practice they may also be used for lateral cutting. And, even when material is found in the gut, it provides evidence only of ingestion, not that it was actively consumed as food. Many extant fish have sediment in the gut, but this was probably ingested incidentally along with other, targeted, material. For example, 'herbivorous' parrotfishes target particulates but also ingest algae, large numbers of copepods and vast quantities of sediment at the same time (Kramer et al. 2013). Furthermore, large and harder fragments persist longer than smaller ones, potentially giving the false impression of a more durophagous condition. Thus, even with living forms, it is difficult to identify the main dietary components.
The trophic reconstruction of Kazanichthys is complicated by the fact that there are no durophagous examples among modern 'lower' actinopterygians (e.g. Polypteryformes, Lepisosteiformes (Fig. 9), Amiiformes, Acipenseriformes). Recent teleosts (including durophagous taxa) have a more mobile jaw apparatus than lower actinopterygians (Schaeffer & Rosen 1961;Lauder 1980) and thus functional analogues for Kazanichthys do not exist. This makes functional interpretations for the diet and function of the dentition in Kazanichthys difficult. However, some speculative reconstruction is still possible.
Today there are two main indicators for durophagy: jaw and tooth morphology. Jaw lever ratios (and relative jaw size) provide a crude correlate with durophagy. Unfortunately, lower jaws of Kazanichthys cannot be compared directly with the recent material (Bellwood et al. 2003;Bellwood & Hoey 2004) because of the poor preservation of the available material. A weak jaw would be inconsistent with Bellwood's indicator of herbivory and durophagy; a short jaw would support it. Molariform teeth are closely correlated with herbivory in extant forms. There are multiple independent examples of a correlation between hard prey harvesting and molariform teeth. The problem is that tooth form alone must be considered with care (e.g. Purnell & Darras 2015), with the most complete evidence provided by a full dentition. This distinction between the function of an isolated tooth and a dentition is summarized in Mihalitsis & Bellwood (2019). The critical point is that it is the position in the jaw, not only the shape of a tooth, that matters when inferring function: bite force varies along the jaw related to a variety of factors including where muscles are positioned along the jaw; this will affect tooth morphology.
†Kazanichthys viatkensis was previously classified as a specialized durophage (Esin 1997) but it is difficult to prove strict durophagy because this morphology does not preclude other types of food, including soft-bodied prey (Purnell & Darras 2015), and it is much more correct to speak of the ability to feed in a durophagous manner. Moreover, fish with a heterodont dental system can eat various types of food (Vandewalle et al. 1995;Smithwick 2015;Potter et al. 2022).
Monocuspid teeth with non-flattened cusps are bestsuited for capture and/or to hold prey (Mihalitsis & Bellwood 2019. The conical teeth of †K. viatkensis resemble those of most basal actinopterygians, including Polypterus, Erpetoichthys and Lepisosteus (Clemen et al. 1998;Wacker et al. 2001;Stamberg 2016Stamberg , 2020 (Fig. 9). Basal actinopterygians with conical teeth, including Polypterus (Bartsch 1997), are predators (Schaeffer & Rosen 1961;Lauder 1980Lauder , 1982. In some cases, the food object is known: the Devonian fossil actinopterygians Pointed molariform teeth are good for crushing hard prey (Crofts & Summers 2014), although the acrodin cap is quite thin and could break when crushing very hard food. The molariform tooth with an apical cusp is known in some platysomids, but in this case the cusp is much blunter. The blunt molariform teeth of †K. viatkensis with an anastomosing pattern of sharp ridges are extremely peculiar and differ from the teeth of all other fishes. However, there are some functional analogues. The most similar molariform teeth are those of the gymnodont fish Avitoplectus from the Eocene of India, with raised spokes radiating inward from the emarginated peripheral edge of the crown (Bemis et al. 2017). The spokes are connected to each other, forming a circle of sharp ridges. The authors suggest that Avitoplectus fed on hard-shelled prey and possibly plant material. Modern sparids, which have molariform teeth and feed on plant material, grind plants with flat, tightly occluded teeth (Purnell & Darras 2015). At first glance, raised spokes would prevent tight occlusion and grinding of plant material. However, the early moradisaurine Captorhinikos valensis had morphologically similar buccal teeth: each tooth had a pointed apex, ringed by a low enamel ridge (LeBlanc et al. 2015). Moradisaurines were adapted to grinding high-fibre plant material and we believe that †Kazanichthys could also process plant material, although much less efficiently, given that it was not capable of propalineal jaw movements.
In addition, raised sharp ridges, unlike a single cusp, would not allow the individual to concentrate force and increase stress in a restricted area, and would give no advantage in the crushing of hard prey over flat molariform teeth. In contrast, such ridges would be too fragile and would often break when eating prey with a very hard covering.
As mentioned above, the preservation of †K. viatkensis skeletons does not allow an accurate calculation of the lower jaw mechanical advantage. Mechanical advantage is the ratio of the length of the in-lever (jaw joint to insertion of force-producing muscles such as the adductor) to the out-lever (distance from jaw joint to a biting point along the jaw, often the most anterior tooth); high mechanical advantage is related to a jaw optimized for biting (effectively a shorter, deeper jaw), while low mechanical advantage indicates less optimization for biting and more for speed of jaw closing: a more elongate jaw ( ) measured the mechanical advantage with the out-lever positioned at the most anterior and posterior parts of the dentition of Eurynotis, finding that posteriorly, the mechanical advantage was much higher than anteriorly, suggesting greater force transmission and bite posteriorly.
The overall proportions of the skull of †K. viatkensis indicate that the lower jaw was not deep. Initially, this supports the suggestion that the jaws of †K. viatkensis had a lower mechanical advantage and the species was not a durophage, but instead fed on soft-bodied prey. However, its molariform teeth were at the back of the jaw, allowing us to predict that the jaw-closing mechanical advantage would be significantly different posteriorly compared with anteriorly, given that the anterior part of the jaws bears conical teeth. A similar relationship was found in Eurynotus, previously recognized as the oldest durophagous ray-finned fish: the anterior jaw-closing mechanical advantage was more similar to carnivorous basal ray-finned fish than to durophagous lungfishes, with the reverse being the case posteriorly (Friedman et al. 2018). This distribution of mechanical advantage matches the observed morphology of the †K. viatkensis lower jaw dentition, suggesting a potential ability to process relatively hard food.
In our opinion, the dental system of †K. viatkensis is suited for feeding on a range of prey types, including soft-bodied invertebrates, small fishes, amphibian larvae, as well as shelled prey such as crustaceans and thinshelled molluscs. Soft-bodied invertebrates are not found in Shikhovo-Chirki. But soft-bodied organisms are extremely rare in the geological record, and the Shikhovo-Chirki sites are not Konservat-Lagerst€ atten, where such organisms can be preserved. The conical teeth from the anterior part of the jaws were seemingly adapted for capture and/or to hold soft prey or perform some manipulation of hard prey. The molariform teeth were suited for crushing hard prey. However, the relative fragility of cusps and crack-patterned ridges, as well as the small or rare facets of wear (which only occasionally appear to be well-developed on some dental plates), indicate that the fish fed on relatively thin-shelled prey. Cusps and crack-patterned ridges are capable of piercing (cusps) or rupturing (ridges) not very hard, or well mineralized, but relatively flexible and thin mineralized coverings. This could include relatively thin, but composite shells of arthropods (chitin, proteins and calcium carbonates) (Fabritius et al. 2009) or, less likely, molluscs (aragonite, proteins) (Marin et al. 2018). This consideration is indirectly confirmed by the fact that Palaeozoic faunas were heavily dominated by archaeogastropods and by bilaterally symmetrical bellerophontaceans, which were relatively vulnerable to predation by crushing or other shelldestructive modes of attack (Vermeij 1977). With uncertainty over feeding in the fossil record in general, explanations for the very different degrees of tooth wear of different dental plates may be speculative. However, we suggest that wear differences may be due to †K. viatkensis switching its feeding mode depending on whether softbodied or solid prey was available during different seasons of the year.
Overall, the dentition of †K. viatkensis is unique among durophages. It possesses tooth replacement (and so differs from the eurynotiforms), but this replacement is extraosseous, which differs from the intraosseous replacement of the phyllodont dentition. Given the extraosseous replacement, †K. viatkensis is more similar to extant basal actinopterygian taxa, although the lack of durophagous feeding in these taxa makes comparisons difficult. The feeding type described for †K. viatkensis is most similar to that of more derived teleosts, including some sparids (Perciformes), such as Sparus aurata (Hadj-Taieb et al. 2013), Acanthopagrus butcheri (Potter et al. 2022) and Acanthopagrus berda (Shilta et al. 2018). These fishes have anterior caniniform teeth and posterior molariform teeth, and feed on molluscs, worms, crustaceans, small fish and aquatic plants. Sparids use anterior caniniform teeth for gripping and manipulating prey, and posterior molariform teeth for crushing hard prey. This enables them to feed on a wide range of benthic prey and to switch prey types. However, compared with the described sparids, Kazanichthys was able to eat less solid prey.

The middle Permian actinopterygian †Kazanichthys
viatkensis has a heterodont dental system: anterior conical teeth (similar to the teeth of basal actinopterygians, e.g. extant Polypterus) and closely packed molariform teeth on the buccal dental plates (superficially resembling teeth of some teleosts, e.g. the Eocene gymnodont †Avitoplectus). The molariform teeth are unique among actinopterygians in being crowned by anastomosing apical ridges.
2. †Kazanichthys viatkensis demonstrates the acrodont implantation mode and ankylosis attachment mode (plesiomorphic condition for actinopterygians), without plicidentine organization. 3. Most replacement teeth formed extraosseously, similar to basal actinopterygians, but differing from the modern and Mesozoic durophagous actinopterygians (in which teeth formed intraosseously). 4. The anterior conical teeth of †K. viatkensis are bestsuited for capture and/or to hold prey. Posterior molariform teeth were capable of piercing (cusps) or rupturing (crack-patterned ridges ridges) semi-hard, relatively flexible and thin mineralized coverings of arthropods or molluscs, and also process plant material. 5. The dental system of †K. viatkensis morphologically resembles that of recent Sparidae (Teleostei; Perciformes), which have anterior caniniform teeth and posterior molariform teeth, and which feed on both soft-bodied and solid prey, and aquatic plants. We propose that †K. viatkensis had a similar trophic adaptation and was a generalist durophagous feeder, with the ability to switch prey types. A. G. Menshikov for collecting and preparing material. We are indebted to K. S. Morshnev for providing comparative specimens of recent herbivorous fishes. We thank the Natural History Museum Core Research Labs for CT scanning and access to workstations and software, and Brett Clark and Vincent Fernandez for assistance. This study was funded by grants from the Russian Foundation for Basic Research No. 17-04-01937, 19-34-90040 and 21-54-10003, and has been supported by the Kazan Federal University Strategic Academic Leadership Program. Funding is also provided by the Royal Society, International Exchanges Fellowship IEC/R2/202001. This work was also funded by a subsidy allocated to Kazan Federal University for the state assignment 671-2020-0049 in the sphere of scientific activities. Finally, we would like to thank two anonymous referees who commented on an earlier draft of this manuscript.
Author contributions. ASB originally conceived this project, collected and curated specimens, photographed specimens and prepared figures, and prepared the first draft of the paper; ZJ CT scanned the Erpetoichthys and Lepisosteus specimens and prepared figures; ASB, ZJ and ALB discussed the interpretation of the results, and all authors contributed to the final draft of the manuscript.

DATA ARCHIVING STATEMENT
CT scan data for the specimens Erpetoichthys calabaricus