CORE Metadata, citation and similar papers at core.ac.uk Journal of Morphology Provided by Digital.CSIC

The Morphology of Pre-hatching Embryos of orientalis (Amphibia: : )

For Peer Review Journal: Journal of Morphology

Manuscript ID: draft

Wiley - Manuscript type: Research Article

Date Submitted by the n/a Author:

Complete List of Authors: Perez, Oscar; Pontificia Universidad Catolica del Ecuador, School of Biological Science Lai, Ngan; University of California, Berkeley, Integrative Biology Buckley, David; University of California, Berkeley, Integrative Biology del Pino, Eugenia; Pontificia Universidad Catolica del Ecuador, School of Biological Science Wake, Marvalee; Univ of California, Integrative Biology

Keywords: , direct development, bone mineralization, tooth crowns

John Wiley & Sons Page 1 of 36 Journal of Morphology

1 2 3 4 5 6 7 8 9 10 11 The Morphology of Pre-hatching Embryos of Caecilia orientalis (Amphibia: 12 13 Gymnophiona: Caeciliidae) 14 15 16 17 18 For Peer Review 19 20 21 22 23 Oscar D. Pérez 1, Ngan Betty Lai 2, David Buckley 2, Eugenia M. del Pino 1, and Marvalee 24 25 2 26 H. Wake * 27

28 29 30 1 School of Biological Sciences, Pontificia Universidad Católica del Ecuador, Quito, 31 32 33 Ecuador 34 35 2Department of Integrative Biology and Museum of Vertebrate Zoology, University of 36 37 California, Berkeley, CA 94720-3140 38 39 40 41 42 43 44 Running head: Pre-hatching caecilian embryo morphology 45 46 47 48 49 *Correspondence to: 50 Marvalee H. Wake 51 52 Department of Integrative Biology 53 3060 VLSB 54 University of California 55 Berkeley, CA 94720-3140 USA 56 e-mail: [email protected] 57 Phone: 1510 642-4743 58 59 60 1 John Wiley & Sons Journal of Morphology Page 2 of 36

1 2 3 ABSTRACT 4 5 A clutch of advanced embryos of the direct-developing caecilian Caecilia orientalis 6 7 (Caeciliidae: Gymnophiona: Amphibia) was collected in the field in Ecuador. Specimens 8 9 were cleared and stained in order to evaluate skeletal development (each element of the 10 11 12 chondrocranium, dermatocranium, jaws and teeth, hyobranchial apparatus, vertebrae) at 13 14 near-hatching. Because it is established that development is correlated with reproductive 15 16 modes in a number of features, we included comparison with taxa that represent the 17 18 For Peer Review 19 major reproductive modes and all of the modern normal tables and ossification 20 21 sequences. The embryos most closely resembled stage 47/48 ramaswamii, 22 23 24 an Indian caeciliid, and stage 45-48 Hypogeophis rostratus , a Seychellian caeciliid, both 25 26 direct developers, in details of bone mineralization, chondrocranial degeneration, and 27 28 vertebrogenesis, and stage 42 H. rostratus in external features (gills, pigmentation, etc.). 29 30 31 They were less like pre-hatchlings of Ichthyophis kohtaoensis , an ichthyophiid with free- 32 33 living larvae, and fetuses of the viviparous caeciliid Dermophis mexicanus and the 34 35 viviparous typhlonectid Typhlonectes compressicauda at comparable total lengths. A 36 37 38 correlation of developmental features with mode of life history is implied. A noteworthy 39 40 feature is that the direct-developing C. orientalis has an armature of multiple rows of 41 42 teeth on the lower jaw with tooth crowns that resemble the “fetal” teeth of viviparous 43 44 45 taxa and that are covered with a layer of oral mucosal epithelium until full development 46 47 and eruption, but the upper jaw bears a single row of widely spaced, elongate, slightly 48 49 50 recurved teeth that resemble those of the adult. 51 52 53 54 Key words: caecilian, direct development, bone mineralization, tooth crowns 55 56 57 58 59 60 2 John Wiley & Sons Page 3 of 36 Journal of Morphology

1 2 3 (Amphibia: Gymnophiona) are elongate, limbless, tailless or nearly so 4 5 6 that inhabit most of the tropical regions of the world. They are not often 7 8 observed because they are fossorial or secondarily aquatic to semi-aquatic. However, 9 10 11 information is accruing about their biology and the nature of their adaptive radiation. In 12 13 particular, their reproductive biology is of interest because it includes several different 14 15 modes. Apparently all caecilians have internal fertilization via the male intromittent 16 17 18 organ inserted intoFor the vent of Peerthe female to transportReview sperm. Eggs may be laid 19 20 terrestrially shortly after fertilization, with the mother guarding the clutch until hatching, 21 22 when larvae wriggle into streams for a relatively lengthy period up to one year before 23 24 25 metamorphosing and returning to land. Alternatively, development through 26 27 metamorphosis ensues before hatching and a juvenile emerges, obviating the aquatic, 28 29 free-living larval stage. In addition, a number of species are viviparous, retaining the 30 31 32 developing embryos in the maternal oviducts through metamorphosis, so that juveniles 33 34 are born. Yolk is resorbed early during the gestation period, which may be 7-11 months 35 36 37 depending on the species, and nutrient secretions from the mother’s oviductal mucosa 38 39 nourish the fetuses, which actively “forage” in the oviducts (see Wake, 1977ab; 40 41 Himstedt, 1996; Exbrayat, 2000, 2006 for summaries). Recently, it has been observed 42 43 44 that two distantly related direct-developing caeciliid species, the east African 45 46 Boulengerula taitanus and the South American Siphonops annulatus, have precocial 47 48 hatching and the young eat the skin and its secretions of the mother (Kupfer, et al. 2006; 49 50 51 Wilkinson et al., 2008). 52 53 Caecilia orientalis (Taylor, 1968) occurs in the eastern lowlands of 54 55 Ecuador, and its reproductive mode was unknown until the discovery of an egg clutch at 56 57 58 59 60 3 John Wiley & Sons Journal of Morphology Page 4 of 36

1 2 3 the Yanayacu Biological Station in January 2001 (Funk et al., 2004). The rare availability 4 5 6 of a clutch of living embryos of the direct-developing C. orientalis allowed examination 7 8 of several aspects of their biology. For example, an observation derived from the single 9 10 11 clutch of C. orientalis has yielded information about caecilian evolution. The lamina- 12 13 associated polypeptide 2 (LAP2) is an integral protein of the inner nuclear membrane in 14 15 mammals, fish and frogs (reviewed by Dechat et al., 2000; Prüfert et al., 2004). LAP2 16 17 18 expression in somaticFor cells of C.Peer orientalis and Review C. guntheri is strikingly different from 19 20 that of anuran and urodele (del Pino et al., 2002). Caecilian somatic cells 21 22 include three LAP2 isoforms with electrophoretic motilities comparable to those of the 23 24 25 LAP2 α, β, and γ of mammals (del Pino et al., 2002). Mammalian cells express mainly 26 27 LAP2 α, β, and γ whereas in somatic cells of the frog Xenopus laevis only LAP2 β 28 29 30 isoforms have been detected (Lang et al., 1999). Consequently, LAP2 α is considered a 31 32 mammalian character, and has not been detected in cells of frogs, fish and the chick (del 33 34 35 Pino et al., 2002; Lang et al., 1999; Prüfert et al., 2004). Further analysis of the LAP2 36 37 isoforms found in somatic tissues of caeciliids and other caecilians is required to 38 39 determine whether the caeciliid LAP2 pattern of somatic cells is molecularly similar to 40 41 42 that of mammals. We mention this example to support the necessity of fieldwork for 43 44 molecular studies and the multidisciplinary approach to research on rare taxa. 45 46 We present information on the developmental morphology of the embryos of C. 47 48 49 orientalis based on examination of living and preserved members of that one clutch 50 51 known to science. Caecilian development is known from only a few normal tables and 52 53 54 descriptions of various stages of embryos and larvae for very few species. Because so 55 56 57 58 59 60 4 John Wiley & Sons Page 5 of 36 Journal of Morphology

1 2 3 little is known, this report on a clutch at a particular stage of development presents new 4 5 6 information to contribute to the comparative developmental morphology of caecilians. 7 8 MATERIALS AND METHODS 9 10 11 Specimens 12 13 Three adults (two males and a female), an uncharacterized individual, and an egg 14 15 16 clutch were found under a large decomposing log at the Yanayacu Biological Station, 17 18 Napo Province, EcuadorFor at 00°35’S Peer 77°53’ W.Review The station is located in a region of cloud 19 20 forest at an altitude of approximately 2100 m on the east side of the Cordillera Oriental of 21 22 23 the Andes. The adults and two embryos were preserved and deposited in the Museum of 24 25 Zoology of the Pontificia Universidad Católica del Ecuador (QCAZ). The remaining five 26 27 embryos were maintained for further study. The simultaneous occurrence of adults and 28 29 30 the egg clutch under a log allowed identification of the specimens as Caecilia orientalis 31 32 (see Funk et al., 2004) . 33 34 35 Culture medium and fixation 36 37 The embryos of C. orientalis were cultured by two methods: (1) the egg clutch 38 39 was maintained in a humid chamber that consisted of a 10 cm Petri dish with a bottom 40 41 42 covered by wet filter paper. The egg clutch was placed over a small piece of plastic foil to 43 44 prevent the egg jelly from sticking to the filter paper. (2) An embryo, dissected from the 45 46 jelly capsule, was maintained in 0.15x Steinberg's solution (58 mM NaCl, 0.65 mM KCl, 47 48 49 0.85 mM MgSO 4, 5 mM Tris (pH 8.0), 0.34 mM Ca(NO 3)2 ,[Rugh, 1962]). 50 51 Two methods of fixation also were used: for bone and cartilage staining, the 52 53 embryos were fixed in 10% formalin in phosphate buffered saline (PBS: 137 mM NaCl, 3 54 55 56 mM KCl, 1.5 mM KH 2PO 4, 7 mM Na 2HPO 4, pH 7.4). One embryo was fixed for 57 58 59 60 5 John Wiley & Sons Journal of Morphology Page 6 of 36

1 2 3 scanning electron microscopy in MEMFA buffer at room temperature for 12 hours, and 4 5 6 stored in methanol at -20 °C (Harland, 1991). 7 8 Clearing and staining 9 10 11 Fixed embryos were eviscerated and stained with Alizarin Red and Alcian Blue. 12 13 The specimens were cleared in 0.5 % KOH and stored in glycerol at – 20 °C according to 14 15 (Jegalian and de Robertis, 1992). 16 17 18 MaintainingFor the clutch Peer Review 19 20 One embryo was released with forceps from the jelly capsule and was immersed 21 22 in 0.15x Steinberg’s solution, a dilute saline solution used for the culture of the aquatic 23 24 25 embryos of frogs. The Petri dish was tilted to allow the hatchling to remain in the aquatic 26 27 medium or to move away to the dry environment. That hatchling was processed for 28 29 molecular analysis of proteins (del Pino et al., 2002), and for the morphological study of 30 31 32 the skeletal system. 33 34 In the absence of information about the developmental biology of C. orientalis , 35 36 37 the remaining embryos were cultured in a humid chamber at room temperature 38 39 (approximately 20 °C) for four days. Fungal growth was discovered that covered the jelly 40 41 capsules, and the embryos died. Two dead embryos were released from the jelly capsules 42 43 44 and were fixed and stained for bone and cartilage. The remaining embryos were fixed in 45 46 10% formalin within the jelly capsules. 47 48 SEM 49 50 51 The left ramus of the lower jaw was excised. The fixed tissue was dehydrated 52 53 through a series of 80%, 95% and 100% ethanol concentrations followed by critical point 54 55 drying in an Autosamari-815 critical point dryer. The dried tissue was mounted onto 56 57 58 59 60 6 John Wiley & Sons Page 7 of 36 Journal of Morphology

1 2 3 aluminum stubs with conductive carbon tape then sputter coated to 8.4 nm with iridium 4 5 6 in a Med 020 sputter coater. Observations and photomicrographs were made with a 7 8 Philips XL-30 scanning electron microscope. Photomicrographs were labeled using 9 10 11 Adobe PhotoShop. 12 13 Staging the embryo 14 15 We followed Müller et al.’s (2005) assessment that the normal tables and 16 17 18 developmental sequencesFor reported Peer for species Review that are egg-laying with free-living larvae 19 20 (e. g., Dünker et al., 2000) and viviparous species (e. g., Wake and Hanken, 1982; 21 22 Sammouri et al., 1990) are not appropriate for comparison with direct-developing taxa. 23 24 25 Therefore, we refer to Brauer’s (1899) evaluation of development in the direct- 26 27 developing Hypogeophis rostratus and, like Müller et al. (2005), used Brauer’s ‘stages’ 28 29 for our assessment of the state of development of our embryo of C. orientalis. 30 31 32 RESULTS 33 34 Caecilia orientalis 35 egg clutch 36 37 The egg clutch consisted of nine egg capsules. Seven capsules contained embryos; 38 39 two capsules were empty and collapsed. The egg capsules were elongated with an 40 41 42 average length of 10.4 mm and a mean diameter of 8.4 mm (Funk et al., 2004), and were 43 44 connected to each other by thick cords of egg-jelly (Fig. 1A). The jelly cords formed a 45 46 central knot to which all of the capsules were attached, similar to the clutches of 47 48 49 Ichthyophis glutinosus (Sarasin and Sarasin, 1887-90; Breckenridge and de Silva, 1973; 50 51 Breckenridge and Jayasinghe, 1979) , I. kohtaoensis (Himstedt, 1996), G. carnosus 52 53 (Seshachar, 1942), and S. annulatus (Gans, 1961), for example. The mother moves the 54 55 56 eggs in the nest, leading to the formation of the central knot of egg cords, in I. 57 58 59 60 7 John Wiley & Sons Journal of Morphology Page 8 of 36

1 2 3 kohtaoensis (Himstedt, 1996). A similar feature of parental care may occur in C. 4 5 6 orientalis. 7 8 When an embryo was released from its capsule in the lab, it rapidly moved away 9 10 11 from the water in its tilted Petri dish. It reacted in a similar way whenever it was 12 13 immersed in the solution. The hatchling was vigorous and moved with an undulating 14 15 motion. The water avoidance behavior observed and the empty capsules suggest that the 16 17 18 embryos of this eggFor clutch were Peer near the time Review of hatching. In fact, the empty jelly 19 20 capsules suggest that two embryos already hatched. 21 22 The hatchling measured 40.0 mm total length (TL). The body was light brown 23 24 25 with obvious melanocytes, and the small eyes were darkly pigmented (Fig. 1A-D). It had 26 27 three pairs of gills with numerous branches and an obvious vascular network. Two gills 28 29 on each side were large (left: 3.4 mm long with 25 filaments and 2.6 mm with 19 30 31 32 filaments; right: 4.1 mm long with 28 filaments and 3.4 mm with 21 filaments) and the 33 34 third gills were very small (left: 0.7 mm long with 7 filaments, right: 0.3 mm with 4 35 36 37 filaments) (Fig. 1BCD). The hatchling had a small dorsal tail fin (Fig. 1B), comparable to 38 39 that of stage 36 embryos of Ichthyophis kohtaoensis (Dünker et al., 2000). 40 41 42 43 44 Developmental osteology 45 46 The skull 47 48 Chondrocranium: The chondrocranium is still well formed at stage 47 (Figs. 49 50 51 2AB, 3ABC). There is limited chondroclastic activity despite some ossification. The 52 53 ventral trabeculae and parachordals are well established; the orbital cartilages are 54 55 elongate and supported by stout pilae; posteriorly the orbital cartilages connect to the 56 57 58 59 60 8 John Wiley & Sons Page 9 of 36 Journal of Morphology

1 2 3 taenia marginales, which connect posteriorly to the large otic capsules (Figs. 2AB, 3AC). 4 5 6 Anteriorly, large nasal capsules attach to the trabeculae ventrally and the orbital 7 8 cartilages dorsally (Figs. 2AB, 3A). The capsules appear to lack copulae anteriorae, and 9 10 11 the prenasal process is short and blunt (Fig. 3A). The capsule cartilage is eroded 12 13 ventrally, but not yet becoming invested with bone (Fig. 3C). The otic capsules are 14 15 cartilaginous and complete, have started ossification around their peripheries, and are 16 17 18 attached to the taeniaeFor and the Peerparachordals (Figs.Review 2AB, 3A). The taenial extension that 19 20 curves ventrally and connects to the capsule is ossifying and the cartilage is resorbing, as 21 22 is that of the parachordals between the postoptic and preotic pilae (Figs. 2AB, 3A). The 23 24 25 quadrate is cartilaginous medially and dorsally, with some bone investing it, but its 26 27 articular component is well ossified and the cartilage is resorbed, except for the articular 28 29 cap Ffigs. 2AB, 4A). A pterygoid process is not apparent. The columella/stapes is 30 31 32 largely cartilaginous with some slight ossification beginning on the anterior head (Fig. 33 34 2AB). Meckel’s cartilages remain pronounced, with large medial bosses and a shaft that 35 36 37 extends to the level of the articulation, but its retroarticular process is nearly fully 38 39 resorbed (Figs. 2AB, 4A). Extensive dermal ossification around the Meckel’s cartilage 40 41 elements is established (see below). The occipital condyles are ossified (Figs. 2AB, 3A), 42 43 44 and the ossifications represent the exoccipital elements. We see no evidence of a chorda 45 46 dorsalis extending into the skull cavity, but, despite the limited cartilage resorption of the 47 48 posterior part of the skull, it may have been resorbed. 49 50 51 Dermatocranium: In dorsal view, lateral strips of bone representing the frontals (over the 52 53 eyes) and the parietals (much the longer) have formed (fig. 2B); posteriorly, the parietal 54 55 has some ossification that is spreading medially. There is an ossification site anterior to 56 57 58 59 60 9 John Wiley & Sons Journal of Morphology Page 10 of 36

1 2 3 the frontal slip that likely represents the nasal. The squamosal appears to be forming 4 5 6 from a medial low, lateral slip that overlies the quadrate and is barely mineralizing 7 8 anteriorly. The maxilla is well formed, with a strong dentigerous process bearing a 9 10 11 number of teeth in a single row and a flared flat plate anterior to the orbit (Figs. 2B, 3A). 12 13 Paired premaxillae are formed; the dentigerous processes are well ossified, each bearing 14 15 4 teeth, two attached to pedicels and two crowns forming between the well-developed 16 17 18 teeth (fig. 4B). OssificationFor ofPeer the dorsal processes Review has commenced. Ventrally, the 19 20 vomers are small mineralizing patches that lack dentigerous processes (Fig. 3A), and the 21 22 palatines are well ossified dentigerous arcs that bear a single row of teeth (Fig. 3AC). 23 24 25 The palatines have faintly staining ossification that extends from the dentigerous process 26 27 toward the maxilla (fig. 3AC), but the elements are not yet fully fused to the maxilla. The 28 29 os basale and the sphenoid process are beginning to mineralize as thin struts of 30 31 32 ossfication (Fig. 3A). 33 34 Lower jaw: The lower jaws are well developed. The pseudangular with the 35 36 37 articulation facet and the retroarticular process are well ossified, covering the short 38 39 Meckel’s cartilage retroarticular process which is nearly fully eroded by chondroclasts 40 41 (Figs. 3AB, 4A). A cartilage cap remains on the articular surface; it contacts the 42 43 44 cartilaginous end of the quadrate (Fig. 4A). The dentigerous plate is stout and extends 45 46 from the anterior end of the jaw nearly to the articular facet (Fig. 2AB). It bears several 47 48 rows of numerous teeth of several different shapes (see Dentition ). 49 50 51 Hyobranchial apparatus: The hyobranchium is nearly completely formed (fig. 52 53 3D); the cerathyals are fused to a small remnant of the basihyal by long descending 54 55 processes; the medial halves of the blades are flared, tapering laterally. The paired 56 57 58 59 60 10 John Wiley & Sons Page 11 of 36 Journal of Morphology

1 2 3 ceratobranchial (CB) I is fused to a lighter-staining basibranchial I remnant at its lateral 4 5 6 edges. The blades are moderately broad and taper slightly laterally. Ceratobranchials II 7 8 are free, fused medially to the lighter-staining remnant of basibranchial II, the blade 9 10 11 width similar to that of CB I. Ceratobranchials III and IV are fused together broadly 12 13 medially, obliterating any remnant of CB III-IV, and the blades of both elements are 14 15 fused continuously from medial to lateral except at their ends, where CB III still have 16 17 18 broad, free ends, andFor the recurved Peer part of IV isReview expanded and ventral, rather than lateral. 19 20 A small remnant of CB V appears to be fused medially to the margin of each CB IV. 21 22 These elements are filling in with cartilage to form the rounded plates typical of the adult 23 24 25 cartilaginous hyobranchium. Paired laryngeal cartilages are centered between CB III-IV. 26 27 See figure 3D. 28 29 Dentition: As observed in cleared and stained specimens, all of the teeth are 30 31 32 pedicellate. Those on the lower jaw are ‘fetal’, in the sense that none have the elongate 33 34 recurved unicuspid teeth of an adult. The crown shapes are all elongate but variously 35 36 37 have peg-like narrowing apices, to spatulate apices, to apices that broaden slightly and 38 39 have small spicule-like projections (Fig. 3B), reminiscent of early fetal teeth in the 40 41 viviparous Gymnopis multiplicata and Dermophis mexicanus (Wake, 1976, 1977ab, 42 43 44 1980). However, all shapes can occur in a given row of teeth, unlike the situation in the 45 46 latter taxa. Teeth posterior on the dentigerous process are as well developed and 47 48 mineralized as those medially (Figs. 2AB, 4AB). There appear to be at least four rows of 49 50 51 teeth on the jaws (Fig. 4B), similar to the aggregations seen in fetuses of viviparous 52 53 caecilians. However, there is a single row of teeth, both fixed to pedicels and 54 55 ‘replacement’ crowns, in the maxillary-premaxillary arcade (Fig. 3B) and on the palatines 56 57 58 59 60 11 John Wiley & Sons Journal of Morphology Page 12 of 36

1 2 3 (Fig. 3AC), similar to the adult condition and more well formed than in fetuses of 4 5 6 viviparous taxa observed. Premaxillary teeth are widely spaced, elongate but not 7 8 recurved structures (Fig. 3B). The apices of the crowns are rounded rather than pointed, 9 10 11 the latter the adult condition. Newly mineralizing crowns from alternate tooth loci occur 12 13 between pedicellate teeth (Fig. 3B), presumably the ‘replacement’ teeth that will become 14 15 associated with pedicels when the currently attached crowns are shed. The palatine teeth 16 17 18 have much shorter Forcrowns that Peer are flat with lateralReview points, some associated with pedicels, 19 20 some newly developing. 21 22 Scanning electron microscopy of the dentition of the lower jaw reveals that many 23 24 25 teeth that appear to be fully erupted are still covered with a thin layer of cellular 26 27 epithelium that extends over the elongate tooth crowns (Fig. 4C). Five rows of teeth are 28 29 present medially, three postero-laterally. Typical of caecilian ‘fetal’ dentitions, the teeth 30 31 32 on the labial margin are more fully developed and erupted than those closer to the lingual 33 34 margin of the jaw. Only the nearly fully developed crowns have penetrated their 35 36 37 epithelial covering (Fig. 4D). Those of the three labial-most tooth rows are bare of the 38 39 epithelium, which surrounds only the bases of those crowns (Fig. 4D). Newly erupted 40 41 crowns are covered with cellular debris (Fig. 4D), which is not present on ‘older’ crowns. 42 43 44 Crown shapes are nearly uniform, with a basal crown stalk that expands to a slightly 45 46 bulbous, expanded apex that bears four to six low spicules, more pronounced on some 47 48 crowns than on others (fig. 4D). The labial face of each crown is slightly cupped. 49 50 51 The vertebrae and ribs 52 53 The embryo has 125 vertebrae in various stages of development (Fig. 5AB), most 54 55 advanced in anteriormost vertebrae in the typical highly cephalized pattern of caecilians 56 57 58 59 60 12 John Wiley & Sons Page 13 of 36 Journal of Morphology

1 2 3 (Wake and Wake, 2000). Taylor (1968) in his description of the species reported that 4 5 6 120-128 vertebrae are present in adults (based on x-rays of two specimens). Our count of 7 8 125 vertebrae in the stage 45 embryos indicates that all vertebrae present in the adult are 9 10 11 formed by that stage of development. The first three vertebrae have extensive 12 13 ossification evident (Fig. 2A). The cartilaginous neural arch of the atlas is fully invested 14 15 with bone, the centrum and neural arch rudiments of the axis have a thin coat of bone, but 16 17 18 retain considerableFor cartilage, andPeer the next vertebra Review has a well ossified centrum with 19 20 mineralization of neural arch pedicels and rudiments. The atlas retains a notochordal 21 22 rudiment that projects anteriorly, but it does not reach the foramen magnum (Fig. 3A). 23 24 25 Behind those, centrum ossification is evident in at least 40 vertebrae (Fig. 5A), strongest 26 27 in the anterior half, progressively slighter and more restricted in the posterior half, then 28 29 more posteriorly ossification is present only investing the anterior half of the centrum 30 31 32 (Fig. 5A) (posterior sclerotomite-half of the more anterior somite: see Wake and Wake, 33 34 2000). Ventral keels are apparent, although not well developed. Only the anteriormost 35 36 37 neural arches have significant ossification. Clearly, centrum ossification precedes that of 38 39 the neural arches. More posteriorly, vertebrae are cartilaginous with well-formed centra, 40 41 neural arches, and rib-bearing processes (Fig. 5B). Cartilage density varies along the 42 43 44 column, being most extensive anteriorly, and a boss of dense cartilage typically forms at 45 46 the juncture of the apex of the neural arch. In the most posterior vertebrae, the centrum is 47 48 a slender cartilaginous structure that forms a continuous ring with the neural arches, and 49 50 51 the notochordal cartilage is prominent, both between vertebrae and through the centers of 52 53 the centra (Fig. 5B). Dorsal and ventral rib-bearers have developed on the anterior 54 55 vertebrae and are associated with bifid rib heads; the ribs of the anterior-most vertebrae 56 57 58 59 60 13 John Wiley & Sons Journal of Morphology Page 14 of 36

1 2 3 have grown longer and more curved latero-ventrally (Fig. 5A). On the vertebrae of the 4 5 6 posterior half of the column, neither rib heads nor rib-bearers have differentiated, and the 7 8 connection is continuous, with rib struts more poorly developed, especially in length (Fig. 9 10 11 5B), concomitant with the antero-posteriorly graded development of the entire column. 12 13 14 15 DISCUSSION 16 17 18 Comparison withFor development Peer in other species Review of caecilians 19 20 The embryos in the clutch of C. orientalis present a mosaic of developmental 21 22 23 features. Because we have only one stage of development, we cannot comment on the 24 25 ossification sequence relative to that of other caecilians, except to compare the presence, 26 27 developmental state, or absence of elements. Furthermore, because our embryos are in a 28 29 30 relatively late stage of osteogenesis, we cannot definitively assess some issues of 31 32 homology of certain relevant elements (e. g., some of those of the lower jaw and of the 33 34 35 anterior membrane bones of the skull). The difficulty is compounded because there 36 37 currently are only three relatively comprehensive accounts of skull development available 38 39 in the literature, those of Wake and Hanken (1982) for the viviparous, New World 40 41 42 caeciliid Dermophis mexicanus, Müller et al. (2005) for the direct-developing, Indian 43 44 caeciliid Gegeneophis ramaswamii , and Müller (2006) for the direct-developing, 45 46 Seychellean caeciliid Hypogeophis rostratus. Each of these presents summary 47 48 49 discussions, and those by Müller include excellent, current summaries of the history of 50 51 ideas about skull development and of differences in reported sequences. Wake (2003) 52 53 also reviewed the data and the ideas. The incomplete developmental sequence of the 54 55 56 direct-developing, Seychellian Hypogeophis rostratus (which included a stage of 57 58 59 60 14 John Wiley & Sons Page 15 of 36 Journal of Morphology

1 2 3 Grandisonia [then Hypogeophis ] alternans , a form with free-living larvae) that Marcus 4 5 6 (1933), Eifertinger (1933) and Marcus et al. (1935) described and interpreted was highly 7 8 influential in considerations of comparative skull development for some time, preceding 9 10 11 the work of de Beer (1937) and many others, until Wake and Hanken (1982) and then 12 13 especially Müller (2006) questioned both data and interpretations. Each species that is 14 15 reported presents new information about skull development, so we believe it appropriate 16 17 18 to compare the stageFor of development Peer of the skeletonReview of C. orientalis with that of the 19 20 paucity of other taxa for which there is information. 21 22 Because it is now well established that there is a correlation of some features of 23 24 25 development in amphibians with their general reproductive mode (Wake and Hanken, 26 27 1982; Hanken, 1992; Wake, 1982, 1989, 1993, 2006; Wake and Dickie, 1998; Müller, 28 29 2006), we compare our data for C. orientalis with the direct-developing H. rostratus 30 31 32 (Brauer, 1897, 1899; Marcus, 1933; Marcus et al., 1935; Müller, 2006) and G. 33 34 ramaswamii (Müller et al., 2005), and then with the viviparous D. mexicanus (Wake and 35 36 37 Hanken, 1982) (all terrestrial caeciliids, but from the Seychelles, India, and Central 38 39 America, respectively), and then in part with the aquatic, viviparous typhlonectid T. 40 41 compressicauda (Sammouri et al. [1990], external morphology only; Wake et al. [1985], 42 43 44 chondrocranium of a few stages only) at the comparable stage of general development 45 46 (given our caveats about staging) . Our hypothesis is that the state of skeletal 47 48 development of C. orientalis will most closely resemble that of the similarly direct- 49 50 51 developing caeciliids, then that of the viviparous caeciliid, and least like that of the 52 53 aquatic, viviparous typhlonectid, which is very different in several respects from the 54 55 56 57 58 59 60 15 John Wiley & Sons Journal of Morphology Page 16 of 36

1 2 3 viviparous caeciliid. We briefly consider the effects of phylogenetic relationship, 4 5 6 reproductive modes, and biogeographic sphere. 7 8 Predictably, not all external characters ‘fit’ a particular stage in diverse taxa. We 9 10 11 found that our specimens of C. orientalis generally agree with stage 42 of Hypogeophis 12 13 rostratus in terms of external morphological features (three gills, one reduced, on each 14 15 side; pigmentation; head morphology; etc.), although it resembles ‘stage’ 48 in yolk 16 17 18 resorbtion. However,For the embryo Peer most closely Review resembles the direct-developing 19 20 Gegeneophis ramaswamii of ‘stage’ 47-8 in many aspects of skeletal development, 21 22 although there are interesting variations in ossification states. 23 24 25 Because we are comparing only one example of development (two embryos at 26 27 nearly the same stage of development from a single clutch), we must consider those 28 29 embryos in the context of the several stages reported for the comparator taxa. The more 30 31 32 complete reports of skull development in comparator taxa allows cross-stage comparison 33 34 in order to “locate” the point in the developmental trajectory of C. orientalis. The state of 35 36 37 ossification of the skull of C. orientalis is much like that of a stage 47/8 G. ramaswamii , 38 39 with most of the same elements present (see Table 4, Müller et al., 2005). That stage in 40 41 turn resembles that of 45-48 of H. rostratus , so Müller’s (2006) comparison of G, 42 43 44 ramaswamii, H. rostratus, and D. mexicanus is apt. We see several similarities of C. 45 46 orientalis to G. ramaswamii and to H. rostratus in terms of state of choncrocranium 47 48 development and degeneration as endochondral bone development ensues, and dermal 49 50 51 elements invest the skull and lower jaw. However, we do not see separate prootics or 52 53 lacrimals present in both G. ramaswamii ( Müller et al., 2005) and H. rostratus (Müller, 54 55 2006), nor do we see indications that such elements are already fused to others; this might 56 57 58 59 60 16 John Wiley & Sons Page 17 of 36 Journal of Morphology

1 2 3 reflect either 1] their absence, 2] development later, which would be unusual, or 3] fading 4 5 6 staining. We find that in one embryo of C. orientalis the os basale is a scant but fully 7 8 distributed sheet of mineralization, unlike that in the comparator taxa. This suggests that 9 10 11 ossification of the entire os basale occurs virtually uniformly (probably except for the 12 13 elements surrounding the brainstem-spinal cord), in contrast to the pattern of the dorsal 14 15 membrane bone elements in which mineralization spreads medially from lateral, linear 16 17 18 sites of ossification.For Furthermore, Peer the posterior Review region of the skull of C. orientalis has 19 20 considerably less degeneration of the chondrocranium (e. g. of the otic capsules, etc.) 21 22 than does that of G. ramaswamii at the comparable stage . The prominent and elongate 23 24 25 chorda dorsalis of the stage 45 (and preceding) G. ramaswamii and H. rostratus is not 26 27 present in our embryos, either because it is not so extensively developed, or because it 28 29 has already resorbed. 30 31 32 However, more substantial differences exist in postcranial development, at least 33 34 between C. orientalis and G. ramaswamii (Müller [2006] restricted his analysis of 35 36 37 ossification in H. rostratus to that of the skull and hyobranchial apparatus). In C. 38 39 orientalis, the atlas, the axis, and the next two vertebrae are well ossified, the axis-atlas 40 41 complex being well ossified including the neural arches, the latter two vertebrae having 42 43 44 ossified centra and neural arch pedicels. The first 40 or so vertebrae have ossifying centra 45 46 with mineralization progressively slighter posteriorly, as noted above. The anterior 60 or 47 48 so vertebrae have rib-bearers and ribs, progressively less well developed further away 49 50 51 from the head. Cartilage of the neural arches and centra of the posteriormost 30 vertebrae 52 53 is weakly stained, and the elements are more fragile in appearance, retaining a prominent 54 55 notochord. In contrast, in G. ramaswamii (Müller et al., 2005) , at stage 38 the atlas and 56 57 58 59 60 17 John Wiley & Sons Journal of Morphology Page 18 of 36

1 2 3 the anterior vertebrae are cartilaginous; at stage 40, neural arches of all vertebrae are 4 5 6 almost fully developed, and the atlas and first 70 vertebrae have ossified centra; at stage 7 8 45 all neural arches are chondrified, and all centra except those of the last five vertebrae 9 10 11 and most neural arches are ossified; at stage 47/8 all neural arches are ossified and the 12 13 cartilaginous ribs are well developed, but not ossified; at stage 49, there is complete 14 15 ossification of all vertebrae and ribs with no vestiges of cartilage. These differences 16 17 18 suggest that developmentFor in C.Peer orientalis may Review be somewhat more cephalized than in G. 19 20 ramaswamii , with slower development of postcranial elements relative to that of the 21 22 skull. 23 24 25 The teeth of the lower jaw of the near-hatching embryos of C. orientalis strongly 26 27 resemble fetal teeth in viviparous species in having several rows with development of 28 29 those rows from the lingual aspect so that the first-developed rows are more labial, and 30 31 32 newer ones more lingual; addition is also antero-medial to postero-lateral, as is typical of 33 34 both fetal and adult dentitions. Crown shape in lower jaw teeth of C. orientalis also 35 36 37 resembles that of fetal teeth, particularly those of the aquatic, viviparous typhlonectids 38 39 Typhlonectes natans and T. compressicauda (Wake, 1976, 1977a; Hraoui-Bloquet and 40 41 Exbrayat, 1996), and to a limited extent, the paroral teeth of apparently newborn 42 43 44 Scolecomorphus vittatus (Loader et al., 2003) . The pattern of development with the 45 46 extensive epithelial covering of the elongated crown before eruption has not been noted 47 48 previously for caecilians (or amphibians in general, to our knowledge). This suggests that 49 50 51 the epithelium is stretched over the crown as it develops, and only is broken through at 52 53 the end of crown development, based on our SEMs of newly erupted teeth (Fig. 4CD). 54 55 Conversely, the single row of widely spaced teeth on the upper jaw arcade and the 56 57 58 59 60 18 John Wiley & Sons Page 19 of 36 Journal of Morphology

1 2 3 elongate peg-like shape of the crowns does not resemble fetal teeth so much, but is 4 5 6 perhaps more similar to the peg-like teeth of B. taitanus , whose hatchlings forage on the 7 8 skin of the mother (Kupfer et al., 2006 ). However, those of C. orientalis are not bicuspid 9 10 11 like those of B. taitanus. Teeth in newly hatched young of other direct-developers that 12 13 bear a resemblance to the fetal teeth of pre-birth viviparous species have been reported 14 15 for Caecilia (see Wake, 2003) . In any case, these teeth differ markedly from the adult 16 17 18 dentition in crown Forshape (elongate, Peer monocuspid, Review and recurved) and in numbers of rows (a 19 20 single row on the premaxillary-maxillary and vomeropalatine arcades, and on the 21 22 dentary, with a very few teeth in an “inner mandibular row”, called the splenial). Tooth 23 24 25 crown morphology is a poorly explored area of systematic and functional biology of 26 27 caecilians. 28 29 Comparison of the developmental stage of C. orientalis with development in the 30 31 32 viviparous Dermophis mexicanus is limited by 1) having only the one stage of the former, 33 34 and 2) the use of total length rather than character-defined states in Wake and Hanken’s 35 36 37 (1982) description of development. However, the comparison that Müller et al. (2005) 38 39 made of development in G. ramaswamii with that of the data for D. mexicanus gives a 40 41 significant point of departure, because of the above-noted similarities (and differences) of 42 43 44 C. orientalis to G. ramaswamii. We cannot ascertain the sequence of development in C. 45 46 orientalis, as we have noted, but the absence of prootics and lacrimals is similar to the 47 48 condition in D. mexicanus but not G. ramaswamii (given our statement above that our C. 49 50 51 orientalis resembles a stage 47/48 G. ramaswamii ). The lower jaw elements are well 52 53 ossified but not yet fully fused, similar to the other two direct-developing species, 54 55 56 57 58 59 60 19 John Wiley & Sons Journal of Morphology Page 20 of 36

1 2 3 suggesting that development may be faster in D. mexicanus (possibly associated with its 4 5 6 early intraoviductal feeding). 7 8 Comparison with the chondrocranium of T. compressicauda basically points up 9 10 11 those features that distinguish its chondrocranial structure from that of other caecilians 12 13 studied, including the enlarged and nearly enclosed nasal capsules, the flange of cartilage 14 15 that extends from the orbital cartilage and the nasal capsule to roof the orbit, and the 16 17 18 lateral walls of theFor braincase beingPeer extensively Review cartilaginous. T. compressicauda lacks a 19 20 prenasal process, a lamina perpendicularis of the mesethmoid, and a septum nasi. The 21 22 anterior part of the chondrocranium is extensively cartilaginous, more so than in other 23 24 25 caecilians reported (Wake et al., 1985), and this is especially apparent at Stage III-5 (42 26 27 mm TL), a stage that would seem to be comparable to that of our C. orientalis. The 28 29 degeneration of the palatoquadrate, Meckel’s cartilage, the parachordal plate and the 30 31 32 occipital arches is less extreme than in C. orientalis . We noted previously that crown 33 34 shape in lower jaw teeth of C. orientalis resembles that of fetal teeth of some viviparous 35 36 37 taxa, particularly those of the aquatic, viviparous typhlonectids Typhlonectes natans and 38 39 T. compressicauda (Wake, 1977a; Hraoui-Bloquet and Exbrayat, 1996), These 40 41 typhlonectids have several rows of fetal teeth on a tooth plate formed of the fused tooth 42 43 44 pedicels surmounting the large medial bosses of the cartilage. The plate of typhlonectids 45 46 is unlike that of other caecilians that have fetal or ‘larval’ dentitions, and concomitantly is 47 48 not present in C. orientalis . Aside from the tooth crown shapes and perhaps the relatively 49 50 51 early development of the palatoquadrate and the articular bones (M.H. Wake, pers. obs.), 52 53 there are few obvious correlates with reproductive mode or comparisons with other taxa 54 55 for either T. compressicauda or C. orientalis that would give any reliable phylogenetic 56 57 58 59 60 20 John Wiley & Sons Page 21 of 36 Journal of Morphology

1 2 3 signal regarding development, but that is largely a consequence of the paucity of material 4 5 6 for C. orientalis and the absence of published ossification sequences for T. 7 8 compressicauda and for most other caecilians at this time. 9 10 11 Consequently, we see more similarities of our specimens of C. orientalis to G. 12 13 ramaswamii and H. rostratus at comparable stages than to either D. mexicanus or T. 14 15 compressicauda. This greater resemblance among the direct-developing taxa may well 16 17 18 have more to do withFor the difficulties Peer of having Review only one stage of C. orientalis and limited 19 20 information about ossification in T. compressicauda than with any correlation with 21 22 reproductive mode. Müller et al. (2005) also commented on the insufficiency of data on 23 24 25 caecilian ossification sequences to develop hypotheses about the evolution of ossification 26 27 patterns and any correlates with life history in caecilians. We await more ontogenetic 28 29 material for caecilians, especially Latin American taxa, and look forward to the time that 30 31 32 the several laboratories working on caecilian biology will have sufficient material for 33 34 collaborative assessments of caecilian development, ecology, behavior, life history, and 35 36 37 relationships. 38 39 40 41 Acknowledgements 42 43 44 We thank W. Chris Funk, George Fletcher-Lazo, Fernando Nogales-Sornosa, 45 46 Diego Almeida-Reinoso, and Luis E. Coloma for the donation of the C. orientalis 47 48 material. Alejandro Sánchez provided information on the method for bone and cartilage 49 50 51 staining. Lai appreciates the training provided by the Electron Microscopy Laboratory, 52 53 University of California, Berkeley. del Pino thanks the Pontificia Universidad Católica 54 55 56 57 58 59 60 21 John Wiley & Sons Journal of Morphology Page 22 of 36

1 2 3 del Ecuador and Wake the National Science Foundation (awards IBN 02-12027 and EF 4 5 6 03-34939) for support. 7 8 9 10 11 12 13 14 15 16 17 18 For Peer Review 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 22 John Wiley & Sons Page 23 of 36 Journal of Morphology

1 2 3 Literature Cited 4 5 6 Brauer A. 1897. I. Beiträge zur Kenntniss der Entwicklungsgeschichte und Anatomie der 7 8 Gymnophionen. Zool Jahrb Anat 10:389-472. 9 10 11 Brauer A. 1899. Beiträge zur Kenntniss der Entwicklung und Anatomie der 12 13 Gymnophionen. Zool Jahrb Anat 12:477-508. 14 15 Breckenridge WR, de Silva GIS. 1973. The egg case of Ichthyophis glutinosus 16 17 18 (Amphibia,For Gymnophiona). Peer Proc Ceylon Review Assoc Adv Sci 29:71. 19 20 Breckenridge WR, Jayasinghe S. 1979. Observations on the eggs and larvae of 21 22 Ichthyophis glutinosus. Ceylon J Sci 13:187-202. 23 24 25 deBeer GR. 1937. The Development of the Vertebrate Skull. Oxford University Press, 26 27 Oxford, UK. 28 29 Dechat T, Vlcek S, Foisner R. 2000. Lamina-associated polypeptide 2 isoforms and 30 31 32 related proteins in cell cycle-dependent nuclear structure dynamics. J Struct Biol 33 34 129:335-345. 35 36 37 del Pino EM, Sáenz FE, Pérez OD, Brown FD, Ávila M-E, Barragán VA, Haddad N, 38 39 Paulin-Levasseur M, Krohne G. 2002. LAP2 (Lamina-associated polypeptide 2) 40 41 expression in fish and amphibians. Int J Dev Biol 46:327-334. 42 43 44 Dünker N, Wake MH, Olson WM. 2000. Embryonic and larval development in the 45 46 caecilian Ichthyophis kohtaoensis (Amphibia, Gymnophiona): A staging table. J 47 48 Morphol 243:3-34. 49 50 51 Eifertinger L. 1933. Die Entwicklung des knöchernen Unterkiefers von Hypogeophis . 52 53 Beitrag zur Kenntnis der Gymnophionen XX. Zeit Anat Entwicklungsgesch 54 55 101:534-552. 56 57 58 59 60 23 John Wiley & Sons Journal of Morphology Page 24 of 36

1 2 3 Exbrayat J-M. 2000. Les Gymnophiones Ces Curieux Amphibiens. Boubée, Paris. 4 5 6 Exbrayat J-M. 2006. In Reproductive Biology and Phylogeny of Gymnophionans. 7 8 Jamieson BMG series ed., Exbrayat J-M volume ed. Science Publishers, Inc., 9 10 11 Enfield, NH. 12 13 Funk WC, Fletcher-Lazo G, Nogales-Sornosa F, Almeida-Reinoso D. 2004. First 14 15 description of a clutch and nest site for the genus Caecilia (Gymnophiona: 16 17 18 Caeciliidae).For Herpetol ReviewPeer 35:128-130. Review 19 20 Gans C. 1961. The first record of egg laying in the caecilian Siphonops paulensis 21 22 Boettger. Copeia 1961:490-491. 23 24 25 Hanken J. 1992. Life history and morphological evolution. J Evol Biol. 5:549-557. 26 27 Harland RM. 1991. In situ hybridization: an improved whole-mount method for X. laevis 28 29 embryos. Meth Cell Biol. 36:685-695. 30 31 32 Hraoui-Bloquet S, Exbrayat J-M. 1996. Les dents de Typhlonectes compressicaudus 33 34 (Amphibia, Gymnophiona) au cours du développement. Ann Sci Nat Zool Paris 35 36 37 17:11-23. 38 39 Himstedt W. 1996. Die Gymnophionen. Westarp Wissenschaften, Marburg. 40 41 Jegalian BC, De Robertis EM. 1992. Homeotic transformations in the mouse induced by 42 43 44 overexpression of a human Hox 3.3 transgene. Cell 71:901-910. 45 46 Kupfer A, Müller H, Antoniazzi MM, Jared C, Greven H, Nussbaum RA, Wilkinson M. 47 48 2006. Parental investment by skin feeding in a caecilian . Nature 49 50 51 440:926-929. 52 53 54 55 56 57 58 59 60 24 John Wiley & Sons Page 25 of 36 Journal of Morphology

1 2 3 Lang C, Paulin-Levasseur M, Gajewski A, Alsheimer, M, Benavente R, Krohne G. 4 5 6 (1999). Molecular characterization and developmentally regulated expression of 7 8 Xenopus lamina-associated polypeptide 2 (XLAP2). J Cell Sci 112:749-759. 9 10 11 Loader SP, Wilkinson M, Gower DJ, Msuya CA. 2003. A remarkable young 12 13 Scolecomorphus vittatus (Amphibia: Gymnophiona: Scolecomorphidae) from the 14 15 North Pare Mountains, Tanzania. J Zool Lond 259:93-101. 16 17 18 Marcus H. 1933. ZurFor Entstehung Peer des Unterkiefers Review von Hypogeophis . Beitrag zur 19 20 21 Kenntnis der Gymnophionen XX. Anat Anz 77:178-184. 22 23 Marcus H, Stimmelmayr E, Porsch G. 1935. Die Ossifikation des Hypogeophisschädels. 24 25 Beitrag zur Kenntnis der Gymnophionen XXV. Gegenbaurs Morphol Jb 76:375- 26 27 28 420. 29 30 Müller H, Oommen OV, Bartsch P. 2005. Skeletal development of the direct-developing 31 32 caecilian Gegeneophis ramaswamii (Amphibia: Gymnophiona: Caeciliidae). 33 34 35 Zoomorphol 124:171-188. 36 37 Müller H. 2006. Ontogeny of the skull, lower jaw, and hyobranchial sleleton of 38 39 Hypogeophis rostratus (Amphibian: Gymnophiona: Caeciliidae) revisited. J 40 41 42 Morphol 267:968-986. 43 44 Prüfert K, Winkler C, Paulin-Levasseur M, Krohne G. 2004. The lamina-associated 45 46 α 47 polypeptide 2 (LAP2) genes of zebrafish and chicken: No LAP2 isoform is 48 49 synthesized by non-mammalian vertebrates. Eur J Cell Biol 83:403-411. 50 51 Rugh R. 1962. Experimental Embryology: Techniques and Procedures. 3 rd edition. 52 53 54 Burgess Publishing, Minneapolis, MN. 55 56 57 58 59 60 25 John Wiley & Sons Journal of Morphology Page 26 of 36

1 2 3 Sammouri R, Renous S, Exbrayat JM, Lescure J. 1990. Développement embryonnaire de 4 5 6 Typhlonectes compressicaudus (Amphibia, Gymnophiona). Ann Sci Nat Zool 7 8 Paris 11:135-163. 9 10 11 Sarasin P, Sarasin F. 1887-90. Ergebnisse Naturwissenschaftlicher Forschungen auf 12 13 Ceylon in den Jahren 1884-1886. Band 2: Zur Entwicklungsgeschichte und 14 15 Anatomie der ceylonesischen Blindwühle Ichthyophis glutinosus . C W Kreidel’s 16 17 18 Verlag, Wiesbaden.For Peer Review 19 20 Seshachar BR. 1942. The eggs and embryos of Gegenophis carnosus Bedd. Cur Sci 21 22 11:439-441. 23 24 25 Taylor EH. 1968. Caecilians of the World A Taxonomic Review. University of Kansas 26 27 Press, Lawrence, KS. 28 29 Wake DB, Wake MH. 1986. On the development of vertebrae in gymnophione 30 31 32 amphibians. Soc Zool France 43:67-70. 33 34 35 Wake MH. 1976. The development and replacement of teeth in viviparous caecilians. J 36 37 Morphol 148:33-64. 38 39 40 Wake MH. 1977a. The reproductive biology of caecilians: an evolutionary perspective. 41 42 Pp. 73-102 In Reproductive Biology of the Amphibia. S Guttman and D Taylor, 43 44 45 eds. Plenum Press Publ., New York. 46 47 48 Wake MH. 1977b. Fetal maintenance and its evolutionary significance in the Amphibia: 49 50 Gymnophiona. J Herpetol 11:379-386. 51 52 53 Wake MH. 1980. Fetal tooth development and adult replacement in Dermophis 54 55 mexicanus (Amphibia: Gymnophiona): fields versus clones. J Morphol 166:203- 56 57 58 216. 59 60 26 John Wiley & Sons Page 27 of 36 Journal of Morphology

1 2 3 Wake MH. 1982. Diversity within a framework of constraints. Amphibian reproductive 4 5 6 modes. Pp. 87-106 In Environmental Adaptation and Evolution. D Mossakowski, 7 8 G Roth, eds. Gustav Fischer, Stuttgart. 9 10 11 Wake MH. 1989. Phylogenesis of direct development and viviparity. Pp. 235-250 In 12 13 Complex Organismal Functions: Integration and Evolution in Vertebrates. DB 14 15 16 Wake, G Roth, eds. John Wiley and Sons, Chichester. 17 18 For Peer Review 19 Wake MH. 1993. Evolution of oviductal gestation in amphibians. J Exp Zool 266:394- 20 21 413. 22 23 Wake MH. 2003. The osteology of caecilians. Chapter 6, pp. 1811-1878 In 24 25 26 Amphibian Biology, vol. 5, Osteology. H Heatwole, M Davies, eds. Surrey 27 28 Beatty and Sons, Pty. Ltd., Chipping Norton, Australia. 29 30 Wake MH. 2006. A brief history of research on gymnophionan reproductive 31 32 33 biology and development. Pp. 1-37 In Reproductive Biology and Phylogeny of 34 35 Gymnophionans. Jamieson BMG series ed., Exbrayat J-M volume ed. Science 36 37 38 Publishers, Inc., Enfield, NH. 39 40 Wake MH, Dickie, R. 1998. Oviduct structure and function and reproductive modes in 41 42 amphibians. J Exp Zool 282:477-506. 43 44 45 Wake MH, Exbrayat J-M, Delsol M. 1985. The development of the chondrocranium of 46 47 Typhlonectes compressicaudus (Gymnophiona), with comparison to other species. 48 49 J Herpetol 19:68-77. 50 51 52 Wake MH, Hanken J. 1982. The development of the skull of Dermophis mexicanus 53 54 (Amphibia: Gymnophiona), with comments on skull kinesis and amphibian 55 56 57 relationships. J Morphol 171:203-223. 58 59 60 27 John Wiley & Sons Journal of Morphology Page 28 of 36

1 2 3 Wake MH, Wake DB. 2000. Early developmental morphology of vertebrae in caecilians 4 5 6 (Amphibia: Gymnophiona): resegmentation and phylogenesis. Zool Anal Compl 7 8 Syst 103:68-88. 9 10 11 Wilkinson M, Kupfer A, Marques-Porto R, Jeffkins H, Antoniazzi MM, Jared C. 2008. 12 13 One hundred million years of skin feeding? Extended parental care in a 14 15 16 Neotropical caecilian (Amphibia: Gymnophiona). Biol Let 4:38-361. 17 18 For Peer Review 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 28 John Wiley & Sons Page 29 of 36 Journal of Morphology

1 2 3 Figure Legends 4 5 6 7 8 Figure 1. Late embryos of Caecilia orientalis. A. Clutch with embryos in egg 9 10 11 membranes bound together by coiled strings of ‘egg cases’. B. Embryo extracted 12 13 from egg membrane. Note slight lateral compression of terminus of body. C. 14 15 Close-up of head of embryo. Note vascular supply of gills and extent of 16 17 18 pigmentationFor of skin, especiallyPeer concentrations Review of melanocytes in annular 19 20 margins. D. Dorsal view of anterior of embryo to illustrate head shape and 21 22 biramous filamented gills. Scale bars: A, B, D = 2.0mm; C = 1.25mm. 23 24 25 Abbreviations: ag = anterior gill ramus; agl = anterior gill ramus; ap = annular 26 27 groove pigmentation ; cd = coiled egg case strands; e = eye; em = embryo; m = 28 29 mouth; op = otic placode; pg = posterior gill ramus; pgl = left posterior gill; pgr = 30 31 32 right posterior gill. 33 34 Figure 2. A. Right lateral view of head of cleared and stained embryo. B. Left lateral 35 36 37 view. Mineralization is Alizarin Red stained; cartilage alcian blue. Scale bars = 38 39 0.6mm. Abbreviations: ar = articular; at = atlas; d = dentary; f = frontal; ma = 40 41 maxilla; moc = mineralized otic capsule; nc = nasal capsule; oc = otic capsule; or 42 43 44 = orbital cartilage; p = parietal; pan = pila antotica; ppor = pila postorbitalis; q = 45 46 quadrate; s = stapes; sn = solum nasi; tm = taenia marginalis. 47 48 Figure 3. Ventral elements of the skull. A. Ventral view of entire skull. B. Frontal view 49 50 51 of nasal region, premaxillary and pseudodentary teeth. C. Palatine shelf (see text). 52 53 D. Hyobranchial apparatus. Note that posteriormost ceratobranchials have not yet 54 55 completed fusion with the third ceratobranchial, and cartilage has not filled in to 56 57 58 59 60 29 John Wiley & Sons Journal of Morphology Page 30 of 36

1 2 3 smooth to the rounded shape of the compound element of the adult. Scale bars: A 4 5 6 = 1.0mm; B = 0.25mm; C = 0.65mm; D = 0.5mm. Abbreviations: ac = arytenoid 7 8 cartilage; bh = basihyal; cbI = ceratobranchial I; cbII = ceratobranchial II; cbIII = 9 10 11 ceratobranchial III; cbIV = ceratobranchial IV; cbV = ceratobranchial V; ch = 12 13 ceratohyal; dt = dentary tooth; ept = ectopterygoid; mps = maxillopalatine shelf; 14 15 nc = nasal capsule; pal = palatine; pm = premaxilla; pmt = premaxillary tooth; pp 16 17 18 = prenasal Forprocess; ob Peer= os basale; oc Review= otic capsule. 19 20 Figure 4: Articulation and dentition. A. Articular region; note extent of ossification of 21 22 elements. B. Tooth crowns and pedicels of lower jaw. C. Scanning electron 23 24 25 micrograph (SEM) of tooth of lower jaw. Note epithelial covering. D. SEM of 26 27 lower jaw tooth erupted from its epithelium. Scale bars: A = 0.25mm; B = 28 29 0.15mm; C, D = 50µm. Abbreviations: ac = articular cartilage; ar= articular; dr = 30 31 32 dentigerous ramus; h = hinge; ma = maxilla; mc = Meckel’s cartilage; ome = oral 33 34 mucosal epithelium; q = quadrate; rp = retroarticular process; tc = tooth crown; tp 35 36 37 = tooth pedicel. 38 39 Figure 5. Vertebrae. A. Mid-column vertebrae. The centra have begun to ossify; the 40 41 cartilaginous rib-bearers and ribs are well formed. B. Posteriormost vertebrae, 42 43 44 illustrating extreme cephalization of caecilian development such that 45 46 posteriormost elements are least developed. In more anterior vertebrae, the 47 48 centrum is well formed, neural arches are connected and have medial crests; 49 50 51 posteriorly, the centra and neural arch elements form circles, and are not 52 53 completely linked in terminal vertebrae. The notochord is still present, but 54 55 resorbing. Scale bars: A = 0.3mm; B = 0.7mm. Abbreviations: c = centrum; na = 56 57 58 59 60 30 John Wiley & Sons Page 31 of 36 Journal of Morphology

1 2 3 neural arch; ncr = notochordal rudiment; ns =neural spine; r = rib; rb = rib-bearer; 4 5 6 z = zygapophysis. 7 8 9 10 11 12 13 14 15 16 17 18 For Peer Review 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 31 John Wiley & Sons Journal of Morphology Page 32 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 For Peer Review 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 83x109mm (600 x 600 DPI) 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Page 33 of 36 Journal of Morphology

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 For Peer Review 19 20 21 22 23 24 25 172x80mm (600 x 600 DPI) 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Journal of Morphology Page 34 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 For Peer Review 19 20 170x53mm (600 x 600 DPI) 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Page 35 of 36 Journal of Morphology

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 For Peer Review 19 20 21 22 23 24 25 26 27 28 29 162x93mm (600 x 600 DPI) 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Journal of Morphology Page 36 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 For Peer Review 19 20 21 22 23 24 25 26 27 28 29 30 170x106mm (600 x 600 DPI) 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons