The present study provides the first description of the development of the naso-palatal complex in a member of the Dactyloidae. At the same time it is one of the most comprehensive anatomical descriptions of the squamate naso-palatal complex, taking into account morphogenesis of both the superficial palate and internal structures of the snout, such as the nasal cavity and the VNO. For a better description of the analyzed developmental events we refer to some structures which have previously been overlooked, namely the subconchal fold and the anterior segment of the palate. They are involved in the formation of the choanal groove and the duct of the VNO respectively. Moreover, we propose new terminology for the maxillary fold and different parts of the choanal tube. The former seems to be “forgotten” in contemporary literature.
Early developmental phase
The paired nasal pits in tetrapods, originating from the nasal placodes, are located on the lateral sides of the developing head. Each of them is bounded by the lateral and medial nasal prominences [6,17,20]. The results of our study showed that the nasal pit in the brown anole embryos is present very early, at stage 2. The vomeronasal pit appears at stages 3–5 (Figs. 2F, H and H’), emerging from the medial wall of the nasal pit. The primordial Stammteil, extraconchal space, and choanal tube were distinguishable in the nasal pit at the end of the early developmental phase of the naso-palatal complex. Moreover, the maxillary prominence terminates just posterior to the nasal pit at this time (Fig. 2C).
The beginning of the developmental sequence of analyzed structures in the brown anole (the origin of the vomeronasal pit from preexisting nasal pit, and the later anterior extension of the maxillary prominence) seems to be similar to that described in other Tetrapoda [6,17,20].
Initial fusion and formation of the vestibulum
The medial nasal prominences fuse and form the single frontonasal mass at stage 7 of the brown anole. A similar frontonasal mass is present also in embryos of other squamates, turtles and birds [17,66–68].
It is generally believed that the fusion of the nasal prominences result in the formation of the vestibulum of the nasal cavity and separation of the external naris from the primitive outer choana [16,17,69]. However, the importance of the maxillary prominence in this process may be underestimated since the studies utilizing optical projection tomography (OPT) showed that the fusion between facial prominences, which initiates the formation of the primary palate, varies across amniotes. In turtle (Emydura subglobosa) the medial nasal prominence fuses with the lateral nasal prominence, but in chick the initial fusion with the medial nasal prominence involves the maxillary prominence only . In crocodilians and mammals the initial fusion involves all facial prominences . A similar situation was found in our study in the brown anole, despite the significant differences in shape of the facial prominences. Abramyan et al.  reported that in lizards (bearded dragon, veiled chameleon, and whiptail lizard), the initiation of the primary palate formation does not engage the maxillary prominence as in turtle, but the external naris becomes closed by the fusion of the nasal prominences and reopens in later development. However, it seems that the complete fusion described by Abramyan et al.  in lizards is a combination of two events: the formation of the nasal plug (anteriorly) and the “classical” fusion (posteriorly; if present), where the epithelial seam is replaced by the mesenchymal bridge (see [68,70]). Thus we are unable to say whether the condition observed by Abramyan et al.  in lizards is truly different from that observed by us in the brown anole, in which all facial prominences initiate the fusion of the primary palate. The description provided by Parsons  indicates that obliteration of the external naris, and thus formation of the nasal plug, may occur relatively later in development of turtle (Chrysemys) than in snakes and other squamates. This is consistent with our observations on the brown anole, since the formation of the nasal plug starts before the initial fusion of the facial prominences and formation of the vestibulum. In fact, at the time of separation of the external naris and the primitive outer choana, which occurs at stage 8 of the brown anole (Movie S1), the external naris and the vestibulum are sealed by the well developed nasal plug. Similar “delay” in formation of the nasal plug, which has been noted in turtle, may occur in birds (see ). These facts may explain the description of Abramyan et al. , who did not find the closed external naris in turtle and chick in stages at the time of initial fusion of facial prominences.
The presence of the embryonic nasal plug is a common transient embryonic feature of amniotes. It has been found in snakes [66,69,72], lizards [18,73], turtle , Sphenodon , birds [71,72] and mammals [72,75] including human [76,77]. The nature of the nasal plug is poorly understood. It is described as mass of epithelial cells , mass of isodiametric cells , peridermal plug [72,78], or structure formed by keratinocytes . It has been proposed that it forms between the nasal prominences from the epithelial seam, which does not undergo disintegration [71,72], or as a proliferation of the “invaginated” oral epithelium , which in fact may be considered as walls of the presumptive vestibulum. It seems to us that these explanations are not mutually exclusive. In fact, the presumptive vestibulum containing the nasal plug at stages 6 and 7 of the brown anole (Figs. 3B and S2A) resembles a multilayered medial edge epithelial seam in the human embryonic soft palate (see ), although in the anole the basal layer, which forms the walls of the future vestibulum, can easily be distinguished. Distribution of the nasal plug is restricted to the vestibulum in later embryonic life [21,71,72].
The functional significance of the nasal plug formation is unknown. Weber  concluded that the nasal plug is functionless and, in contrast to the auditory meatal plug, does not protect undifferentiated sensory epithelium. As he stated, the amniotic fluid may still reaches the lumen of the main nasal cavity through the open choana (see also ). Moreover, despite the fact that recanalization of the vestibulum takes place shortly before hatching in the brown anole (this study) and in other extant sauropsids , in some mammals including human, the reduction of the nasal plug occurs relatively early [71,72,76]. Nevertheless, we suspect that the formation of the nasal plug may be crucial for vestibulum formation.
Recent studies on chick have suggested that the canalization of the vestibulum and the external naris is caused by “combination of cellular remodeling, apoptosis, as well as non-apoptotic necrosis” . In the brown anole the vestibulum takes the relatively simple form of a long tube, oriented almost horizontally at late stages (17–19; Figs. 10A, 11A). Elongated vestibulum has been found in the other studies on the brown anole , in Anolis carolinensis , A. cristatellus, and possibly in A. pulchellus, A. evermanni, and A. stratulus . In contrast, the vestibulum of A. roquet (described there as A. alligator) has been found to be short . The shortened vestibulum seems to be present also in A. equestris (see Fig. 12B in ). As was shown, the morphology of this structure may exhibit some variations in these lizards. The great variation of this structure was also found in other squamates. For instance, a shortened vestibulum occurs in snakes and Gerrhonotus (Anguidae), but it may be well developed and take an almost vertical orientation as in Phrynosoma  (Iguania). It has been suggested that the elongation of the vestibulum in squamates is associated with deserticolous habitats  and it has been suspected that such feature as well as “less direct communication with the exterior” may decrease the evaporation rate . It is worth mentioning that the vestibulum of adult squamates excludes inspired foreign bodies, such as soil particles, from the respiratory tract. This is possible due to hyperplasia of its epithelium (which occurs in some forms) and mucoid secretion of the lateral nasal gland, which discharges between the vestibulum and anterior part of the main nasal cavity [5,17]. The function of cleansing the nose may be improved by the length and shape of the vestibulum, and by the presence of a network of sinusoids associated with the reticulum of smooth muscle fibers (spongy sinusoidal tissue, cavernous tissue) surrounding this part of the nasal cavity, which probably close the vestibulum during intumescence [21,22,33].
Formation of the choanal groove and homology of the choanal folds
The term choanal groove is often used to describe the groove on the palatine bone in the region of the choana [38,80]. Here we use this term exclusively for a soft tissue structure [5,16,21] as defined above. The formation of the choanal groove in squamates is associated with the separation of the VNO from the nasal cavity and separation of the main nasal cavity from the oral cavity . The results of our studies showed that in the brown anole embryos all three events occur due to the closure of the primitive inner choana by the fusion of the subconchal fold with the vomerine cushion. Just prior to formation of the choanal groove, the duct of the VNO can be distinguished. The choanal groove, as well the VNO duct, constitute the persistent portion of the primitive outer choanal tube. Our interpretation differs from the previous descriptions of choanal groove formation which indicated that this structure develops by the fusion of the maxillary prominences (or choanal fold) with the vomerine cushion [6,21]. It was also suggest that the same fusion event is responsible for the obliteration of the primitive outer choanal tube just posterior to the VNO duct, resulting in the shortening of the adult choanal groove. In fact, the choanal groove is shortened anteriorly in many adult squamates and does not reach the VNO duct [5,6].
It seems that the formation and the obliteration of the choanal groove constitutes two different processes in squamates. As we showed, in the brown anole the former involves the fusion of the vomerine cushion with the subconchal fold, which originates from the lateral nasal concha. The lateral nasal concha originates in turn from the lateral nasal prominence (see also ). With the formation of the upper lip, the lateral nasal prominence and the nasolacrimal groove become hardly distinguishable from the rest of the snout (Fig. 6B). This condition may explain previous misinterpretation about the participation of the maxillary prominence, rather than lateral nasal prominence, in formation of the choanal groove.
Despite the absence of the choanal groove – in most snakes [6,36,40] it has been shown to occur in the grass snake embryo (and probably in embryos of other snakes) and extends from the VNO duct to the outer choana . Thus it is reasonable to assume that also in non-ophidian autarchoglossans, in which the choanal groove does not reach the VNO duct , the anterior part of the choanal groove becomes obliterated during a separate process. It seems that in contrast to the formation of the choanal groove its obliteration may indeed engage the maxillary prominence (choanal fold), as has been suggested [6,21].
In Gekkota and Iguania, the choanal groove is confluent with the duct of the VNO. We found that in the late stages of the brown anole the choanal groove extends along most of the length of the main nasal cavity. The outer choana and the outer choanal tube are located well posteriorly (Fig. 11A and Figs. 12A, A’). Interestingly, at stage 19 most of the choanal groove was found to be separated from the oral cavity, by the fusion of the medial edge of the choanal fold with the vomerine cushion (Figs. 12A’, B). Our findings agree with the previous description of the brown anole . The closed choanal groove has been noted also in A. cristatellus and possibly in A. pulchellus, A. evermanni, and A. stratulus . Bellairs & Boyd , who based their description on A. equestris, A. lineatopus and A. valencienni, stated that closure of the choanal groove is restricted to the anterior part of the choanal groove, “a little behind the level of the organ of Jacobson.” As far as we know the formation of the closed channel in a place of choanal groove has never been noted in other squamates and such condition is likely a synapomorphy of Dactyloidae.
The presence of the choanal groove was considered to be a synapomorphy of squamates . In fact, in Sphenodon, the only extant member of Rhynchocephalia, commonly considered to be a sister group to Squamata [28,82–84], the nasal cavity widely communicates with the oral cavity through the long outer choana [5,16,17]. Bellairs & Boyd  suggested that only a short choanal groove is present anteriorly. However, a long outer choana is also present in many representatives of Iguania [21,33].
The floor of the choanal groove and the lateral margin of the outer choana is formed by the choanal fold, which is supported by the ectochoanal cartilage [6,21]. The squamate choanal folds (sometimes called the palatine processes) seem to be considered as a structures homologous to the crocodilian and mammalian palatal shelves [see 3–5], which in these taxa form “true” secondary palate and well developed nasopharyngeal duct [69,86,87]. In mammals the palatal shelves arise as vertically-oriented structures, then become reoriented horizontally and fuse with each other over the tongue to form the secondary palate, which anteriorly fuses to the primary palate and the ventral part of the nasal septum [87,88]. Crocodilian palatal shelves grow horizontally from the beginning (except their posteriormost parts) and they merge to the nasal septum (anteriorly) and to each other (posteriorly) [86,89].
A review of the German literature published around 1900 indicates that at least two kinds of medial soft-tissue outgrowths of reptilian maxillary prominence can be distinguished [5,81,90]. This is consistent with our results, since we were able to find the maxillary fold and the choanal fold originating from it. Based on comparative embryological data, Fuchs  postulated that the squamate choanal folds are not homologous to the mammalian palatal shelves. On the contrary, he suggested that mammalian palatal shelves are homologous to the mediale Seitenfalten (here: the maxillary fold). Moreover, it was suggested that the homolog of the choanal fold is present in mammalian embryos. It is located dorsal to the (“secondary”) palatal shelve [5,81] and may be called as “primitive palatal shelve” (primitive Gaumenfalte) . Moreover, Fuchs  suggested that crocodilian secondary palate is formed by the homologs of choanal folds not the maxillary folds.
The presence of two rather than one medial soft-tissue outgrowth of the maxillary prominence in reptiles seems to be overlooked in current literature. Alternatively, the choanal fold could be considered as a part of the maxillary fold, since the former seems to emerge from the latter in the brown anole (compere Figs. 5B and 6A). However, both structures are relatively distinct from each other from the beginning of the late developmental phase of the naso-palatal complex (see Fig. 9H).
The initial vertical orientation of the mammalian palatal shelves has been discussed by Ferguson [86,89,92]. In fact so-called “palatal shelves” of sauropsids grow horizontally from the beginning. He stated that such condition is associated with the large muscular tongue. The alternative explanation may in fact support the view of Fuchs  and the elevation of maxillary folds (mediale Seitenfalten) to the horizontal orientation may constitute the mammalian synapomorphy.
In birds the palatal shelves approaches closely to each other, but do not fuse as in crocodilians . Interestingly in turtles the similar medial outgrowths of the maxillary prominence (palatal shelves or alternatively the choanal folds) are not formed in some taxa , but the main nasal cavity is well separated from the oral one, probably due to posterior extension of the primary palate . The homology between the amniote medial soft-tissue outgrowths of the maxillary prominence requires confirmation in future comparative studies.
Outer choanal tube and nasopharyngeal duct
Fuchs  recognized two parts (limbs) of the choanal tube: horizontal (“horizontaler Schenkel des Choanenganges”) and descending (“absteigender Schenkel des Choanenganges”). A similar division was applied by Parsons , who distinguished a “more or less horizontal dorsal portion lying ventral to the concha” and a ventral portion which “from near the lateral end of this dorsal portion” “extends ventromedially to the choana”. At the same time, he did not consider the choanal tube as a part of the conchal zone. Bellairs  and Bellairs & Boyd  distinguished the ventral conchal zone and restricted the term “choanal tube” to the descending limb of the choanal tube sensu Fuchs . In most non-ophidian squamates the horizontal limb of the choanal tube (or ventral conchal zone) leads to the descending limb of the choanal tube or choanal tube sensu Bellairs & Boyd  through the inner choana. The descending limb of the choanal tube runs ventromedially and opens to the oral cavity by the outer choana [6,22,64]. Here we employed division of the choanal tube into the inner and outer choanal tube (Figs. 1C, D), which correspondence to the two parts of this structure distinguished by Fuchs  and Parsons .
For embryonic structures we also introduced the term “primitive choanal tube” to define the structure connecting the primordial Stammteil to the oral cavity from about stage 5 of the brown anole (Figs. 2E–H). In later development of this species this structure gives rise to the vestibulum, the primitive outer choanal tube, and the primitive inner choanal tube. The primitive outer choanal tube gives rise to the primordial VNO duct, choanal groove, and the outer choanal tube. With the closure of the part of the primitive inner choana (stage 9/10) and the formation of the choanal groove, the primitive inner choanal tube may be called the inner choanal tube. Because the structure defined here as the outer choanal tube is usually recognized to be homologous to the nasopharyngeal duct of snakes and some lizards [6,21], we consider that only the inner choanal tube belongs the main nasal cavity. In snakes the outer choanal tube forms the nasopharyngeal duct separated ventrally from the oral cavity by the fusion of the vomerine cushion and choanal fold .
It is worth noting that the nasopharyngeal duct of snakes and some lizards is not homologous with the nasopharyngeal duct of mammals and crocodiles [21,94]. The ophidian nasopharyngeal ducts of both sides join together in their posterior course and enter the oral cavity into the median palatal trough (orbitonasal trough) [6,17,19,21]. In this case the inner and outer choana become widely separated . Some forms of the nasopharyngeal ducts (paired or single median) were also found in non-ophidian squamates (Xantusia, Dibamus chameleons, some skinks). This structure is formed by the medial extension of choanal folds or elongation of the vomerine cushion [21,33,39,40,67]. However, in most cases the nasopharyngeal ducts of lizards is incomplete ventrally and some authors did not distinguish it as a “snake-like” nasopharyngeal duct [6,21,33,39,40,68]. Stebbins  stated that in Anolis carolinensis “the nasopharyngeal duct situated well posteriorly” is present. It seems that Parsons  did not consider the condition of Anolis as the presence of the nasopharyngeal duct. Instead he noted that: “the choana is short and well posterior”. A similar interpretation was adopted in this study; thus the outer choanal tube was not considered to form the nasopharyngeal duct in the brown anole.
Lateral nasal concha and main nasal cavity
The nasal concha constitutes any projection into the nasal cavity of amniotes, but definition of this structure varies among different authors . In most squamates only a single concha is present, called the lateral nasal concha. It forms a space for the lateral nasal gland [17,21]. The lateral nasal concha of squamates is homologous to the posterior concha of Sphenodon . Due to the presence of the concha, 3 parts of the conchal zone of the nasal cavity in squamates can be distinguished: the extraconchal space, the Stammteil, and the choanal tube (the inner choanal tube). The primordium of the lateral nasal concha in the brown anole appears at the end of the early developmental phase (stage 5/6), when it emerges from the lateral nasal prominence (Fig. 2E) and thus before the separation of the external naris and the primitive outer choana at stage 8. The same developmental sequence was described in Thamnophis .
Previous studies indicated that the lateral nasal concha is absent in anoles [2,16,21,33]. We found that reduction of the concha is associated with the shift of the extraconchal space to the posteriormost part of the main nasal cavity and change of its orientation to vertical position. In later development, the extraconchal space becomes dilated and incorporated into the Stammteil (Figs. 10A, B and Figs. 11A, C). However, we found that even in the latest developmental stages (18–19) of the brown anole the remnant of the concha, called here the anterior extension of the lateral nasal concha, is still noticeable in the region of the lateral nasal gland, between the Stammteil and the inner choanal tube (Figs. 11A, D). It is likely that it undergoes further reduction in later ontogenesis. The reduction or lack of the concha was also found in other members of the Iguania including Acrodonta and Pleurodonta. Within the latter, the lateral nasal concha is absent not only in anoles, but also in the taxa classified as “Uma-morphotype” containing: Uma, Callisaurus, Holbrookia, Sceloporus, Uta, and Phrynosoma , which all belong to the Phrynosomatidae . Depending on a phylogenetic analysis it could be inferred that the lack of the lateral nasal concha occurred independently in the Phrynosomatidae and the Dactyloidae (see ) or the loss of this structure might have taken place in the common ancestor of the Phrynosomatidae and a clade containing Dactyloidae and Polychrotidae (see ). Still, little is known about the state of this character in other groups of Pleurodonta. In “Dipsosaurus-morphotype” lizards (Dipsosaurus, Iguana, Crotaphytus, Sauromalus, Ctenosaura) , some anguids and in monitor lizards the concha is attached dorsally or dorsolaterally. In such cases the extraconchal space was suggested to be absent, and instead the subconchal recess is distinguishable . However, in another study these two structures were considered to be homologous .
The primordium of the VNO appears as a vomeronasal pit emerging on the medial wall of the nasal pit. It appears before the fusion of the nasal prominences [17,69]. The vomeronasal pit is also described in crocodilian and bird embryos, but eventually disappears early in development and the VNO is not present in adult specimens . In crocodilians the vomeronasal pit is no longer noticeable after the fusion of the lateral and medial nasal prominences . In the brown anole, the vomeronasal pit was first found at developmental stages 3–5. Considering the level of association of the VNO with the nasal cavity and degree of its development in adult specimens a few types of the VNO may be distinguished. The VNO of amphibians and turtles is a part of the nasal cavity and the non-sensory part cannot be distinguished [97–99]. The mammalian VNO is connected to the nasal cavity and in some forms also to the oral cavity, and it consists of more or less developed sensory epithelium, which is thicker than the non-sensory one [4,97]. A non-sensory vestige of the VNO is present in humans and chimpanzees , but in some bats the VNO may be rudimentary or even absent . In Sphenodon the VNO takes a lens-shape form and there is no mushroom body. The VNO lumen is connected to the anterior part of the nasal cavity near the long outer choana [69,102]. However, some descriptions suggest that the VNO of this taxon communicates with both the oral cavity and the short choanal groove .
In adult squamates the VNO has no direct connection with the nasal cavity and its duct enters the oral cavity exclusively as in most squamate groups or the oral cavity and the choanal groove as in iguanians and gekkotans. In skinks and lacertids the choanal gutter is formed by the close association of the choanal groove with the lacrimal duct and is connected to the VNO duct by the lacrimal component . In the brown anole separation of the VNO from the nasal cavity occurs at developmental stage 9/10. At this stage the mushroom body is well distinguishable, but relatively small (Fig. 7B).
In adult squamates the vomeronasal sensory epithelium is generally strongly developed and the non-sensory epithelium covers the mushroom body [2,14,103]. The VNO can be reduced in some arboreal Iguania, especially in chameleons in which the mushroom body or even the entire organ is absent [16,39,69]. It was suggested that the anole VNO is: “present but less well-differentiated than in most lizards” . In fact we found that the VNO of the brown anole at stages 18 and 19 possesses a poorly developed mushroom body and weakly developed ventral channel (see Fig. 14). Histological study revealed medial localization of the dorsal dome in relation to VNO duct. Interestingly, the adult vomeronasal sensory epithelium has well-visible pigment aggregated within its central layer (Fig. 14C). Similarly as in brown anole embryos three layers of developing vomeronasal sensory epithelium were also found in other light microscopic studies [2,17,19] (e.g. Fig. 14A). The ultrastructural studies indicated that this epithelium contains different populations of cells [15,103,104]. Based on previous data and our studies we can suppose that the basal layer (L1) is a population of developing bipolar neurons and undifferentiated cells, the central layer (L2) is composed of supporting cells while the apical one (L3) includes the dendrites of bipolar neurons and protrusions of the supporting cells.
The sensory epithelium of the ophidian VNO is characterized by a peculiar columnar organization [15,17,19,103,104]. The columns are located basal to the thin layer of the supporting cells, and contain bipolar neurons and undifferentiated cells. Columns extend through most of the height of the entire vomeronasal sensory epithelium. Such columnar compartments are separated by the invaginated basal lamina and connective tissue containing the capillary network. Similar columnar organization may be present in non-ophidian squamates, but they are not such well developed as in snakes . For instance, Kratzing  stated that columns of connective tissue with a capillary network intrude the sensory epithelium of the VNO in Tiliqua, while in Takydromus only small, sparsely distributed blood vessels were found in vomeronasal sensory epithelium . The penetration of vomeronasal sensory epithelium by connective tissue, without formation of columns, was found in some mammals: rodents and rabbit . The columns of neurons have never been described for adult anoles. Neither columnar organization of VNO sensory epithelium nor blood vessels within bipolar and undifferentiated cells were detected in the present study.
The only indication of intrusion of connective tissue into the vomeronasal sensory epithelium might be the occurrence of the pigment granules within the layer of supporting cells, especially frequent in late stages of the brown anole (17–19), but such explanation seems to be unlikely. At stages 18 and 19 of the brown anole, fewer pigment granules occurred also within the basal layer containing bipolar neurons. The presence of connective tissue with associated pigment cells at the base of the sensory epithelium as well as between epithelial columns of bipolar neurons was found in newly hatched and adult Thamnophis . The presence of pigment cells in the connective tissue around the base of the VNO was noted previously in anoles, as in other squamates [2,21,97]. Nevertheless, the lack of even sparsely distributed blood vessels within the bipolar cell layer in the VNO of the brown anoles suggests that the intrusion of connective tissue into the vomeronasal sensory epithelium does not occur. Moreover, the association of such pigmentation with the supporting cell layer suggest that pigment granules observed here are homologous to the olfactory pigment of supporting cells, which has been found in some vertebrates, but usually reported for mammals and birds [106,107]. It has been suggested that the carotenoid component of such pigment may be involved in olfactory function in terrestrial vertebrates, in which olfactory pigment is present also in cells of Bowman’s glands . Interestingly, we did not observe similar pigmentation in the sensory olfactory epithelium of the brown anole and the pigment granules have not been noted in vomeronasal sensory epithelium in other vertebrates .
The mushroom body in squamates [14,15,103,104], the floor of the VNO in Sphenodon , and the lateral or dorsolateral wall of the organ in mammals [100,108,109] are covered with ciliated non-sensory epithelium. In general the luminal surface of mature vomeronasal sensory epithelium in tetrapods is covered with microvilli of dendrites and protrusions of supporting cells, and there are no cilia characteristic for the dendritic surface of olfactory sensory epithelium. We found that in the brown anole at late developmental stages (18 and 19) the apical surface of the vomeronasal sensory epithelium is covered with microvilli resembling the cilia (Fig. 14C’’). Similar images from light microscopy have been obtained for Takydromus (see Fig. 2 in Saito et al. ). Our preliminary studies based on transmission electron microscopy revealed that the apical surface of the vomeronasal sensory epithelium in late stages of the brown anole is covered by long and thin microvilli and only sparsely distributed cilia can be observed (unpublished observations).
Most tetrapods, except some plethodontid salamanders  and turtles , possess a lacrimal duct. In mammals, birds, crocodilians, Sphenodon and most amphibians, it extends from the lower eyelid and discharges into the nasal cavity near the external naris or more caudally, closer to the entrance to the VNO [20,111–115]. In squamates the lacrimal duct is usually directly associated with the VNO . Moreover, in most non-ophidian squamates it is connected with the choanal groove and, in some cases, also with the nasal cavity at the level of the outer choana [6,21,22]. Close association of the VNO and the lacrimal duct was also found in the Gymnophiona [20,116]. It is widely accepted that the lacrimal duct is involved in direct or indirect delivery of secretions of the orbital gland, including the Harderian gland secretion, into the VNO lumen [e.g. 24,42,114], except for crocodilians and birds, in which the VNO is transitory embryonic feature. It means that the secretion of the Harderian gland is involved in vomeronasal chemoreception, rather than eye lubrication [23,24,42,117]. In fact the flow of the Harderian secretion gland or orbital fluid to the VNO was experimentally tested in snakes  and frogs [111,115]. The other studies on snakes found that female pheromones in Thamnophis are soluble in the Harderian gland homogenate .
It seems that the origin of the lacrimal duct from the nasolacrimal groove ( = naso-optic furrow), located between the maxillary prominence and the lateral nasal prominence, is evolutionarily conserved in tetrapods [6,89,94,119]. In the brown anole, the primordium of the lacrimal duct was first present at stage 8. Interestingly, we found that only a small anteriormost part of the duct seems to originate from the nasolacrimal groove. The major part of the duct in the brown anole develops from the groove located between the primordial lower eyelid and the maxillary prominence (Figs. 8A, A’ and Movie S1). Such a peculiar condition could be explained by the fact that the eyes in this species seem to be very large in relation to the snout. In consequence, the developing eye may cover the posterior extent of the lateral nasal prominence and reduces the length of the nasolacrimal groove. This hypothesis needs to be confirmed by comparative morphometric studies, but the occurrence of the huge eyes of the brown anole in relation to the other squamates seems to be very clear (compare Fig. 1, stage 8 in , Fig. 4, stage 32 in , Fig. 2h in , and Fig. 2, stage 4 in .
The orbital end of the lacrimal duct of tetrapods usually forms two canaliculi or sometimes one [23,113,123]. Unfortunately, many studies concerning development of the nasolacrimal apparatus are based on relatively late developmental stages, when the lacrimal duct is well developed [e.g. 112–114]. Thus, still little is known about development of the lacrimal canaliculi. Here we show that the lacrimal canaliculi of the brown anole become distinguishable just after the formation of single lacrimal duct primordium.
The lacrimal duct of Sphenodon discharges exclusively into the nasal cavity [20,124]. In contrast to that condition, the lacrimal duct of most adult non-ophidian squamates establishes a relatively extensive connection with the choanal groove [6,34,125]. Here we show the steps of formation of such extensive connection in the brown anole, which is established by the lacrimal duct through its rostral plate (see Figs. 13A–D). In gekkotans, except pygopodids, the connection of the lacrimal duct and the choanal groove seems to be restricted to the anterior part of the latter [21,34]. The rostral end of the squamate lacrimal duct is usually connected to the VNO duct or enters the VNO lumen directly in some forms . The lacrimal duct of monitor lizards, pygopodids, some amphisbaenids, and at least one dibamid, reaches the VNO duct directly, and there is no communication with the choanal groove . The same condition occurs in adult snakes in which the choanal groove is absent [6,19,118]. Moreover, in pygopodids, amphisbaenids, and snakes, the lacrimal duct establishes a more intimate connection with the duct of the Harderian gland via lacrimal canaliculi or the Harderian gland discharges directly into the lacrimal duct [23,42].
In the latest developmental stages of the brown anole (18 and 19), the lacrimal duct is not connected to the outer choanal tube, but the posterior expansion of the rostral plate of the lacrimal duct runs backward toward the outer choana (Fig. 13D). This expansion is lined with the epithelium characteristic for the choanal groove (with the presence of cilia and goblet cells) (Fig. 12B); thus it could be considered as a part of the choanal groove rather than the lacrimal duct. In fact Bellairs & Boyd  suggested that “transition from stratified to ciliated epithelium” may approximate the border between the fused lacrimal duct and the choanal groove. However “ciliated respiratory type” epithelium was found in the anterior part of the shorter lacrimal duct in monitor lizards , goblet cells in the epithelial lining of the saccus lacrymalis in the lacrimal duct are present in crocodilians , and the goblet cells in addition to cilia have been found in the human lacrimal duct . Moreover, the changes in anatomy of the lacrimal duct observed in successive stages at late developmental phase suggest that the posterior part of the rostral plate of the lacrimal duct gradually grows posteriorly (see Figs. 13A–D) and thus the region under consideration does not represent the part of the choanal groove. If our interpretation is correct, then it could be assumed that the posterior expansion of the rostral plate of the lacrimal duct (white arrows in Fig. 13D) of the brown anole is involved in a different function than the rest of the lacrimal duct, and constitutes a functional extension of the choanal groove.
Importance of the olfactory system and functional remarks
Pratt  suggested that reduction or loss of functionality of olfactory systems in such forms as Anolis, arboreal agamids and chameleons corresponds to the “massive orbital development”. The sensory part of the nasal cavity was described as “almost non-sensory”, while the VNO was described as “reduced and completely non-sensory”. The Dactyloidae originates from an arboreal ancestor  and most extant species contain toe pads  which serve an adhesive function . However, many extant forms are not strictly associated with a fully arboreal lifestyle [29–32]. The brown anole occupies a “trunk-ground” niche [29,32]. Pratt’s  notion about the non-sensory character of the anole VNO was based on the anole described there as Anolis alligator (which probably is now A. roquet  and represents a trunk-crown ecomorph [128,129]). In fact, behavioral evidences, including that involving tongue extrusion, indicates that the accessory olfactory system may still be important in males of the brown anole for detection of female pheromones . Increases in the tongue flick rates were found in A. carolinensis during male-male encounters and in individuals which were transferred to novel habitats and exposed to foliage or air movements . The tongue extrusions were also observed in the field for A. trinitatis . The interpretation of such behavior as evidence for the importance of the accessory olfactory system may be problematic. The role of tongue extrusion in anoles, especially substrate licking, may mediate gustation rather than vomeronasal function, since taste buds of the tongue tip are found to be generally abundant in Iguanidae (“present” in A. carolinensis and “abundant” in A. bonairensis) . However, some findings suggest that lingual gustation in squamates cannot replace vomerolfaction and thus be responsible for chemical discrimination of prey or conspecifics .
Morphological studies also showed that the VNO is functional at least in A. garmani, A. grahami and A. lineatopus , which are classified as crown-giant, trunk-crown and trunk-ground ecomorphs respectively [32,128,129]. Some features of anole adult morphology may constitute adaptations to vomeronasal chemoreception. For instance, the closure of the choanal groove may efficiently deliver the secretion of the Harderian gland to the VNO duct and its vicinity. A similar condition is present in snakes in which the choanal groove is absent, but the anterior end of the lacrimal duct is exclusively connected with the VNO duct . The presence of a functional VNO in anoles representing different ecomorphs (including crown-giants and trunk-crowns) may suggest that Pratt’s  interpretation for A. roquet was not correct and the arboreal lifestyle in anoles may be associated with the reduction in size of the sensory epithelium of the VNO (and accessory olfactory bulb) rather than with complete loss of its chemosensory function. Moreover, the generalization of Pratt  is falsified by the fact that well-developed olfactory systems in arboreal geckoes are present [36,134].
It was suggested that reduction of olfactory organs in such representatives of Iguania as chameleons and agamas (and probably Anolis) is secondary, and the condition of Iguanidae, which are characterized by relatively well‑developed chemosensory abilities, is ancestral . In fact, our studies may indirectly confirm this notion at least for Pleurodonta, since the morphology of the adult brown anole probably does not represent the ancestral condition. The loss/significant reduction of the concha, characteristic for Anolis and the “Uma-morphotype”, but not for “Dipsosaurus-morphotype” including Iguanidae, seems to be secondary, since at the beginning of the middle developmental phase the concha and extraconchal space were easily distinguishable in the brown anole. Nevertheless, members of the Iguania in contrast to the rest of the squamates (Scleroglossa) are characterized by a less developed VNO , lower tongue-flick rate  and lower abundance of vomeronasal and olfactory receptor cells . The weakly or moderately developed chemosensory abilities in Iguania are difficult to explain, since the phylogenetic position of this group is still a matter of debate . Such a Sphenodon-like condition may be considered either as plesiomorphy according to morphological studies placing the Iguania as sister to the other squamates [e.g. 83], or reversal according to molecular phylogeny, which suggests the nested position of Iguania [e.g. 28,82].