JOURNAL OF MORPHOLOGY 231:11–28 (1997)

Cellular Migration and Morphological Complexity in the Caecilian Brain

ANDREA SCHMIDT1 AND MARVALEE H. WAKE2* 1Brain Research Institute, University of Bremen, 28334 Bremen, Germany 2Department of Integrative Biology and Museum of Vertebrate Zoology, University of California, Berkeley, California 94720

ABSTRACT The morphology of the tectum mesencephali and the medial pallium is studied in species representing the six families of caecilians (Amphibia: Gymnophiona) in order to determine whether differences in brain morphology are related to function, phylogenetic history, or life history strate- gies. In general, the caecilian tectum is characterized by simplification in having little to no lamination and few migrated cells. The degree of morpho- logical complexity differs between species and between brain regions. Our data suggest that changes in brain morphology are due to a mosaic of different influences. We did not find a strict correlation between visual system reduc- tion and tectal morphologies. However, phylogenetic effects exist. The great- est degree of morphological complexity is found in members of the Rhinatrema- tidae, a family that is considered basal to the lineage. Thus, simplification of brain morphology in caecilians must be considered a secondary or derived rather than a primitive feature. Direct development and miniaturization are correlated with the greatest simplification in the tectum mesencephali and medial pallium. There is a relationship between differences in brain morphol- ogy and heterochrony in caecilians, as in other amphibians. J. Morphol. 231:11–27, 1997. r 1997 Wiley-Liss, Inc.

The morphology of the brain of members layer and a superficial fiber layer containing of the three modern orders of amphibians— very few migrated cells (Roth et al., ’90b; frogs (Anura), salamanders (Caudata) and Schmidt and Wake, ’91). In salamanders, caecilians (Gymnophiona)—differs not only not only the tectum mesencephali but also in the relative sizes of different brain re- other regions are affected by the reduction of gions but also in the degree of morphological cell migration (Roth et al., ’93; Wake et al., complexity of the parts. The caecilian brain ’88). Only the medial pallium always shows is characterized by a large telencephalon, cell migration (Roth et al., ’93). In the past, and a very small tectum mesencephali, an simple morphology was interpreted to be a important area for sensory integration that primitive (Herrick, ’48; Leghissa, ’62) and receives primary visual projections. Small unspecialized state. However, a recent phy- size is in concordance with the reduction of logenetic analysis suggests that simple mor- the visual system in these animals (Man- phology is an apomorphic trait and due to teuffel and Himstedt, ’85; Wake, ’85). The secondary, derived simplification (Roth et tectum mesencephali also differs between al., ’92, ’93). frogs and salamanders. The tectum of an- Major factors that trigger brain morpho- urans is large, expanded, and multiply lami- genesis are 1) stimulation by afferents and nated (Fig. 2a). It consists of nine alternat- 2) changes in the time course of development ing cellular and fiber layers (Potter, ’69). The (Finlay et al., ’87; Roth et al., ’93; Schmidt superficial fiber layers (layers 7–9) contain a and Roth, ’93). Caecilians are characterized substantial number of migrated cells (up to by a reduction of the entire visual system. 30%). These layers resemble the tectum of The degree of reduction differs interspecifi- most other groups of vertebrates (Vanegas et cally (Wake, ’85). It is currently not known al., ’84). In contrast, the tectum of sala- manders and caecilians (Fig. 2b,c) is very *Correspondence to: M.H. Wake, Department of Integrative simple. It has only a periventricular cellular Biology, University of California, Berkeley, CA 94720-3140. r 1997 WILEY-LISS, INC. 12 A. SCHMIDT AND M.H. WAKE

Fig. 1. Cladogram illustrating currently accepted hypothesis of relationships of families of caecilians (Order Gymnophiona). Species examined in this study are listed below the family name. whether or how the reduction of the visual ’78). The genome sizes of caecilians (Ses- periphery affects the morphological complex- sions and Wake, unpublished) lie between ity of the tectum and what secondary effects those of salamanders, which are character- on the development of other brain regions ized by large genomes and a very simple might arise. Is, for example, the expansion tectal morphology (Roth et al., ’90a, ’93), and of the telencephalon and the elaborated olfac- those of frogs, which have small genomes tory system (Schmidt and Wake, ’90) in these and a multiply laminated tectum. Changes animals indirectly related to the reduction in the time course of development also occur of the visual system? in the context of differences in life history In urodeles, simplified brain morphology strategies. For example, some frogs and sala- is considered to be due to a change in the manders that have direct development (fer- time course of development, in the sense tilized eggs laid on land, development that late ontogenetic stages are retarded or through metamorphosis before hatching so lost (Roth et al., ’93; Schmidt and Roth, ’93). that a free-living larval period does not oc- In consequence adult animals retain juve- cur) are characterized by an acceleration or nile characters in the adult stage and are condensation of early ontogenetic stages. The thus considered paedomorphic. The retarda- more common and presumably plesiomor- tion of development in these animals is phic life history mode is shown by species thought to be due to a decrease in rates of that lay eggs that develop into larvae, char- cell metabolism and cell proliferation that is acterized by a suite of characters involving a correlated with an increase in genome and free-living aquatic period before a relatively cell size (Sessions and Larson, ’87; Roth et profound metamorphosis into the juvenile/ al., ’90a; Szarski, ’76, ’83; Cavalier-Smith, adult form. Among salamanders the most CAECILIAN BRAIN 13 simplified brains in terms of degree of lami- (West Africa, both viviparous) (see Table 1 nation and cellular migration occur in minia- for reproductive modes). turized, direct developing, derived species (Roth et al., ’90a, ’94). Methods The gymnophione amphibians show a di- versity of structure-function modifications The morphology of the tectum was exam- and life history patterns that is not seen in ined in heads of adult specimens, obtained other amphibians, at least to the same ex- from numerous individuals, museums, and tent. These features include limblessness, collections (see Wake, ’86, ’92). Material was visual reduction, and viviparity (retention of formalin-fixed and preserved in 70% etha- embryos with maternal nutrition following nol. Heads were removed, decalcified, and resorption of yolk, and birth of fully meta- embedded in paraffin and serially sectioned morphosed young). In order to determine (transversely or horizontally) at 7–10 µm. whether differences in brain morphology are Every third slide was stained with picro- related to function and/or phylogenetic ponceau, hematoxylin-eosin, or Mallory’s history and/or life history strategies, we in- azan according to standard procedures (Hu- vestigated the morphology of the tectum mes- manson, ’79). Since an exact identification of encephali and the medial pallium of repre- tectal borders can be made only from trans- sentative species of the six families of caecilians, verse sections, only those species for which which differ in both the degree of reduction we had such sections were studied quantita- of the visual system and in life history strat- tively (E. bicolor (N 5 2), D. mexicanus egies. (N 5 3), T. natans (N 5 1), B. boulengeri (N 5 1), and S. uluguruensis [N 5 1]). Every MATERIALS AND METHODS third section was drawn with the aid of a Species examined camera lucida. From these drawings we de- Tectal morphology was studied in repre- termined total cell number and number of sentatives of all six families of the Order migrated cells in the superficial fiber layer Gymnophiona (Fig. 1) (see Duellman and and calculated the percentage of white mat- Trueb, ’86; Nussbaum and Wilkinson, ’89; ter. Areas (i.e., the periventricular cellular but also see Hedges et al., ’93). Families and layer and the white matter) were measured species studied include Rhinatrematidae using a graphic tablet (Summagraphics, Aus- (northern South America), Epicrionops pe- tin, TX) and its program. Each migrated cell tersi and Epicrionops bicolor (both with an was counted. In order to approximate the aquatic larval stage); Ichthyophiidae (south- genome size, we determined the nuclear size east Asia), Ichthyophis kohtaoensis (aquatic by calculating the mean diameter of nuclei larval stage); Uraeotyphlidae (India), Uraeo- of eight to twelve cells/section, ten sections typhlus narayani (aquatic larval stage in- per specimen (genome size is highly corre- ferred from condition of congener [Wilkin- lated with nuclear and cell size; see Discus- son, ’92]); Caeciliidae (Central and South sion and Table 1). Tables 2 and 3 present America, Africa, Seychelles, India), Caecilia analyses of comparative organization of the occidentalis (Peru, life history not known), tectum, telencephalon, and the visual sys- Dermophis mexicanus (Central America, vi- tem, emphasizing degrees of differentiation, viparous), Gymnopis multiplicata (Central complexity, and cell migration for the former America, viviparous), Hypogeophis rostra- two components and components present tus (Seychelles, direct development), Geotry- (complexity) in the latter (see Tables and petes seraphini (Ghana, viviparous), Sylva- Discussion). Because we found large differ- caecilia grandisonae (Ethiopia, semiaquatic ences in the numbers of migrated cells in larval stage [Largen et al., ’72]), Bouleng- different tectal regions (medial zone and lat- erula taitana (Kenya, direct development eral zones), we did calculations for the whole [Nussbaum and Hinkel, ’94]), B. boulengeri tectum as well as separately for the medial (Tanzania, life history unknown, perhaps and lateral tectum. Cell number within the direct development [Nussbaum and Hinkel, periventricular cellular layer was deter- ’94]), Idiocranium russelli (Cameroon, direct mined by counting cells in different tectal development); Typhlonectidae (SouthAmerica), regions of the same size (four counts for each Typhlonectes natans and Chthonerpeton indis- cross-section). Using the cell numbers of tinctum (both viviparous); Scolecomorphi- these volumes, we extrapolated the cell num- dae (East and West Africa), Scolecomorphus bers of the total volume of the periventricu- kirkii and Scolecomorphus uluguruensis lar cellular layer. Brains that were sectioned 14 A. SCHMIDT AND M.H. WAKE

Fig. 2. Transverse sections of tectal hemispheres in ences in the number of migrated cells (open arrows) the anuran Bufo bufo (a), the urodele Salamandra sala- within the superficial fiber layer (SF). The thick line mandra (b), and the caecilian Typhlonectes natans (c). transecting the tectal layers indicates the thickness of The frog tectum is multilayered; those of the sala- the periventricular gray. Bars 5 100 µm. mander and the caecilian are bilayered. Note the differ- horizontally were studied from photomicro- sections. The degree of morphological differ- graphs at specific levels of the tectum and by entiation was evaluated by counting the comparing the degree of cellular migration number of fiber layers within the periventric- and morphological differentiation in these ular gray. No measurements were made from CAECILIAN BRAIN 15 TABLE 1. Life history modes and tectal cell nuclear a periventricular cellular layer and a super- diameters of caecilians ficial fiber layer that contains retinal affer- Nuclear ents, dendrites of tectal neurons, tectal effer- Life diameter in Standard 1 ents, and migrated cells that are mainly Taxon history micrometer deviation concentrated in the lateral part of the tec- Rhinatrematidae tum. However, there are differences among Epicrionops bicolor o 6.8 1.27 Epicrionops petersi o 7.1 0.82 the caecilian species examined in the com- Ichthyophiidae plexity of the cellular layer as well as in the Ichthyophis number of migrated cells that occur in the kohtaoensis o 7.5 0.45 superficial fiber layer. In some species the Uraeotyphlidae Uraeotyphlus periventricular gray is homogeneous (Fig. narayani o(i) 6.8 0.54 3a,b,c), whereas in other species it is tra- Scolecomorphidae versed by fibers that, however, in most cases Scolecomorphus kirkii v 6.2 0.81 do not constitute continuous layers (Fig. Scolecomorphus 3d,f). Those species that have the greatest uluguruensis v 5.6 0.52 degree of morphological complexity show dif- Caeciliidae ferent layers (Fig. 3e,g,h). The caudal tec- Gymnopis multipli- cata v 8.1 0.53 tum has a greater degree of lamination than Dermophis mexi- the rostral tectum. We distinguish an epen- canus v 7.1 1.11 dymal layer, an adjacent fiber layer, an in- Geotrypetes sera- phini v 5.8 0.89 ner cellular layer, an adjacent fiber layer, an Sylvacaecilia gran- outer cellular layer, and a superficial fiber disonae o 7.0 0.53 layer containing migrated cells (Fig. 3e,g,h). Boulengerula bou- Since nothing is known about the cytoarchi- lengeri d(i) 5.9 0.69 Boulengerula tecture of the caecilian tectum and because taitana d 5.8 0.66 different layers in the caecilian tectum can- Hypogeophis ros- not easily be homologized with those in the tratus d 5.7 0.56 Idiocranium russelli d 5.6 0.54 tectum of frogs and salamanders, we do not Caecilia occiden- adopt the descriptive numbering used for talis ? 6.8 0.63 the tectal layers of frogs and salamanders Typhlonectidae (Potter, ’69; Roth, ’87). Typhlonectes natans v 7.7 0.92 Chthonerpeton indistinctum v 7.7 1.16 Species-specific variation 1d, direct development; i, inferred; o, oviparous, larvae; v, vivipar- Descriptions are organized phylogeneti- ity; ?, unknown. cally, from basal families to the more derived (see Fig. 1). Figures 3 and 4 compare morpho- brains that were sectioned horizontally (E. logical differentiation in the tectum, Figures petersi, U. narayani, C. occidentalis, G. multipli- 5–7 neuronal migration in tectum (Fig. 5) cata, H. rostratus, G. seraphini, S. grandiso- and telencephalon (Figs. 6, 7). Consequently, nae, B. taitana, and S. kirkii), because of non- the reader should note the figure order as it comparability with other data, so our results applies to these species-level descriptions are more qualitative than quantitative for these and then refer to the figures again relative species. Numbers of specimens examined per to the discussion of each region. species ranged from one to five, most often one because so few specimens of these rela- Family Rhinatrematidae tively rare taxa are available for sectioning. The small sample sizes and substantial Epicrionops bicolor (Fig. 3e,g,h). E. bi- variation of meristic measures precluded color possesses a morphologically well-orga- cross-species statistical analysis. nized tectum. The morphologically most com- plex region of the tectum mesencephali is RESULTS the lateral zone of the tectum; here, several Descriptive morphology of the tectum tectal layers are distinguishable. The peri- mesencephali ventricular gray is divided into three cellu- In general, the caecilian tectum appears lar layers and two fiber layers. We distin- rather simple in that it consists primarily of guish an ependymal layer, a subsequent 16 A. SCHMIDT AND M.H. WAKE

TABLE 2. Morphological organization of the tectum mesencephali and the visual system, and cellular migration in the tectum mesencephali and the telencephalon in caecilians1 Tectal Visual Tectal Telencephalic Taxon organization class organization class migration class migration class Rhinatrematidae Epicrionops bicolor 3331 Epicrionops petersi 3331 Ichthyophiidae Ichthyophis kohtaoensis 2322 Uraeotyphlidae Uraeotyphlus narayani 2 3 2–3 3 Scolecomorphidae Scolecomorphus kirkii 1222 Scolecomorphus uluguruensis 1123 Caeciliidae Gymnopis multiplicata 1133 Dermophis mexicanus 2–3 3 3 3 Geotrypetes seraphini 2333 Sylvacaecilia grandisonae 2–3 3 2 2 Boulengerula boulengeri 1122 Boulengerula taitana 1122 Hypogeophis rostratus 2322 Idiocranium russelli 1–2 3 1 1 Caecilia occidentalis 2–3 2 3 2 Typhlonectidae Typhlonectes natans 2 3 2–3 3 Chthonerpeton indistinctum 2 2 2–3 3

1Categories 1–3 indicate simple, moderate, and complex classes of organization, migration, and differentiation; simple is often secondary rather than primary. See text for discussion. inner fiber layer, a cellular layer (thickness better distinction of different cellular layers of about five cell diameters), an outer fiber in the caudal than in the rostral tectum. layer, an outer cellular layer (thickness of Epicrionops petersi (Fig. 4c). As in E. about seven cells), and the superficial fiber bicolor, the caudal part of the tectum mesen- layer that contains migrated cells, which are cephali in E. petersi is morphologically more far more numerous in the lateral than in the complex than the rostral part. In the caudal medial zone. In contrast to the lateral zone, part we distinguish an ependymal layer, an there is no clear distinction of different cellu- adjacent fiber layer, an inner cellular layer lar layers in the medial zone. There is a that is separated from an outer cellular layer by a fiber layer, and a superficial fiber layer containing migrated cells. TABLE 3. Tectal migration relative to the amount of 1 white matter in caecilian brains Family Ichthyophiidae % Migrated cells % White matter Ichthyophis kohtaoensis. The tectum Taxon TT MT LT TT MT LT2 mesencephali of I. kohtaoensis is rather un- Rhinatrematidae differentiated. It consists of an ependymal Epicrionops layer that is separated from the main cellu- bicolor 9.5 1.5 11.7 65.5 57.4 65.5 lar layer by a very thin fiber layer. The main Scolecomorphidae periventricular cellular layer is a more or Scolecomorphus uluguruensis 3.7 1.5 5.1 55.8 50.4 59.1 less homogeneous layer of loosely packed Caeciliidae cells. In the lateral part of this layer, there Dermophis mexi- are some islets of fibers that traverse the canus 9.6 3.3 15.5 67.3 50.2 76.1 periventricular gray but do not constitute a Boulengerula boulengeri 3.8 0.5 7.3 63.1 48.5 72.2 continuous fiber layer. The superficial fiber Typhlonectidae layer contains only a few migrated cells that Typhlonectes occur mainly in the lateral zone. natans 5.0 2.0 7.8 60.3 53.2 65.8

1Compare tectal cell migration percentages to nuclear/cell sizes Family Uraeotyphlidae and life history modes in Table 1. 2LT 5 lateral zone of the tectum; MT 5 medial zone of the Uraeotyphlus narayani. The morphologi- tectum; TT 5 total tectum. cally most complex region of the tectum in Fig. 3. Morphological differentiation in the tectum inner cellular layer, an outer fiber layer, an outer cellu- mesencephali. Transverse sections. Boulengerula bou- lar layer, and the superficial fiber layer containing mi- lengeri (a–c) has a homogeneous periventricular gray in grated cells. There is a greater degree of morphological the rostral (a) as well as in the caudal (b) tectum. In T. complexity in the caudal (h) than in the rostral tectum natans (d,f) there are subdivisions within the periven- (g). Photomicrographs for panels c, d, e show a higher tricular gray without definition of layers. Epicrionops magnification of the periventricular gray in B. bouleng- bicolor (e,g,h) has the greatest degree of morphological eri (c), T. natans (d), and E. bicolor (e). a, b, f, g, h: Bars 5 differentiation. Starting from the ventricle, we distin- 100 µm. c, d, e: Bars 5 50 µm. guish an ependymal layer, a subsequent fiber layer, an 18 A. SCHMIDT AND M.H. WAKE

Fig. 4. Morphological differentiation in the tectum petersi (c) has the greatest degree of morphological mesencephali. Horizontal sections. Gymnopis multipli- differentiation. The tectum and diencephalon of G. sera- cata (a) has a homogeneous periventricular gray. In G. phini are illustrated at low magnification in panel d. seraphini (b,d), some fibers occur within the periventric- Bars 5 100 µm. ular gray without clear definition of layers. Epicrionops

U. naryani is the lateral zone of the caudal layer, a thin outer cellular layer, and the super- tectum. Here we distinguish an ependymal ficial fiber layer containing migrated cells. In layer, an adjacent inner fiber layer, a thick the rostral tectum, we observe only an ependy- inner cellular layer, a subsequent outer fiber mal layer, a fiber layer, a thick homogeneous CAECILIAN BRAIN 19

Fig. 5. Degree of neuronal migration within the tec- tectum of S. grandisonae conjoined to show that the tum mesencephali. Horizontal sections. Idiocranium rus- greatest degree of neuronal migration is in the lateral selli (a) has limited neuronal migration; Scolecomor- part of the ventral tectum. An exception to that general phus kirkii (b) has moderate neuronal migration; T. condition for caecilians is C. occidentalis (f), in which natans (c) has extensive neuronal migration. Illustra- migration is greater in the dorsal tectum. Bars 5 100 µm. tions for panels d and e are the dorsal (d) and ventral (e) cellular layer, and the superficial fiber layer lar layer. Moderate numbers of migrated that contains a moderate number of migrated cells are more concentrated in the lateral cells. zones but are also present in the medial zones. Family Scolecomorphidae Scolecomorphus kirkii (Fig. 5b). The tec- Family Caeciliidae tum consists of one homogeneous periventric- Boulengerula boulengeri (Fig. 3a,b,c). B. ular cellular layer and a superficial fiber boulengeri has a very simple tectal morphol- layer that contains a few migrated cells. ogy in that we find only an ependymal layer, Scolecomorphus uluguruensis. As in S. a well-differentiated inner fiber layer, a thick kirkii, there are no subdivisions of the loosely homogeneous periventricular gray, and few packed homogeneous periventricular cellu- migrated cells in the lateral zone. 20 A. SCHMIDT AND M.H. WAKE Boulengerula taitana. The tectum mes- tectum. The superficial fiber layer contains encephali of B. taitana is rather simple. In only very few migrated cells. general, we distinguish an ependymal layer, a fiber layer that is better expressed in the Sylvacaecilia grandisonae (Fig. 5d,e). rostral than in the caudal tectum, a homoge- Morphological complexity is greatest in neous periventricular cellular layer, and a the caudal tectum. In this part, we distin- superficial fiber layer containing migrated guish an ependymal layer, a thin fiber layer, cells that are most numerous in the central two cellular layers that are separated by an tectum. outer fiber layer, and a superficial fiber layer. More rostrally, there is no distinction of dif- Gymnopis multiplicata (Fig. 4a). The ferent layers within the periventricular gray morphology of the tectum mesencephali in matter. S. grandisonae possesses few mi- G. muliplicata is rather simple in that we grated cells. Most migrated cells are found distinguish only an ependymal layer, a sub- in the lateral part of the central tectum. sequent fiber layer that is more elaborated Caecilia occidentalis (Fig. 5f). In this spe- in the rostral than in the caudal tectum, a cies, a great degree of morphological com- thick periventricular gray matter that does plexity is found in the dorsal tectum. We not have any further lamination, and a su- distinguish an ependymal layer, a few fibers perficial fiber layer containing many mi- between the ependymal layer and a subse- grated cells. quent thin cellular layer (which is not con- Dermophis mexicanus. The tectum mes- tinuous), an outer fiber layer, a thick cellular encephali in Dermophis has a moderate to layer, and the superficial fiber layer that great degree of morphological complexity. contains many migrated cells. We observe an ependymal layer, an adjacent Hypogeophis rostratus. The tectum mes- fiber layer, a periventricular gray matter encephali is morphologically more complex that is subdivided into two cellular layers in the caudal than in the rostral part and in with a fiber layer lying between them, and a the lateral than in the medial part. In the superficial fiber layer that contains many caudal tectum we distinguish an ependymal migrated cells in the lateral zone of all parts layer, a well-differentiated fiber layer, a thick of the tectum. There are few migrated cells inner cellular layer where cells are loosely in the medial zone. Although Dermophis pos- packed, an outer cellular layer where cells sesses a subdivision of the periventricular are more densely packed, and a superficial gray similar to that of Epicrionops, this sub- fiber layer containing migrated cells. There division is not as elaborated as in Epicrion- are very few fibers between the inner and ops. outer parts of the cellular layer. In the ros- Geotrypetes seraphini (Fig. 4b,d). Tectal tral tectum there is no distinction of differ- morphology in this species is similar to that ent parts of the thick periventricular cellu- in H. rostratus. The caudal tectum is morpho- lar layer, and there are fewer migrated cells logically more complex than the rostral tec- within the superficial fiber layer. tum. Within the caudal tectum, we distin- guish an ependymal layer, a very thin fiber Family Typhlonectidae layer, a thick inner cellular layer with a low Typhlonectes natans (Figs. 3d,f, 5c). T. cell density, a thin outer fiber layer and an natans has a moderate degree of morphologi- outer cellular layer where the cell density is cal complexity. We distinguish an ependy- higher than in the inner layer, and a superfi- mal layer, a well-differentiated fiber layer, a cial fiber layer containing migrated cells. cellular layer, and a superficial fiber layer The number of migrated cells is highest in that contains a moderate number of mi- the central tectum. In the rostral tectum, we grated cells. Migrated cells occur mainly in distinguish an ependymal layer, a well- the lateral zone of the tectum; only few are differentiated fiber layer, a thick homoge- found in the medial zone. There is a slight neous cellular layer, and a superficial fiber distinction of two different cellular layers layer with migrated cells. within the periventricular cellular layer, Idiocranium russelli (Fig. 5a). I. russelli separated by few fibers. However, there is no possesses the greatest reduction of the tec- clear distinction of an outer fiber layer within tum; it is very small and very poorly lami- the periventricular gray. nated. There is only a thin fiber layer that Chthonerpeton indistinctum. The mor- divides the periventricular gray into two phology of the tectum in C. indistinctum is thin layers in the ventralmost part of the similar to that of T. natans in that there is a CAECILIAN BRAIN 21

Fig. 6. Degree of neuronal migration in the telencephalon. Transverse sections; all photos are of comparable regions of the telencephalon. Epicrionops bicolor (a) has the least migration; B. boulengeri (b) has moderate migration; S. uluguruensis (c) has extensive neuronal migra- tion. Bars 5 100 µm. slight but not well-expressed distinction of geri, and S. grandisonae have more mi- subdivisions of the periventricular gray mat- grated cells than does I. russelli (Fig. 5a) but ter. Moderate numbers of migrated cells are still few migrated cells (Fig. 5). E. petersi, G. present in the superficial fiber layer. multiplicata, G. seraphini (Fig. 4b,d), and C. Neuronal migration in the tectum occidentalis (Fig. 5f) possess a substantial mesencephali and telencephalon number of migrated cells comparable to their Tectum mesencephali (Fig. 5; see also Figs. abundance in E. bicolor and D. mexicanus. 3, 4) The degree of cellular migration in H. rostra- tus and C. indistinctum is comparable to The degree of cellular migration differs in that of T. natans (Figs. 3d,f, 5c). In general, the caecilians examined (Figs. 3–5; Table 2). Our quantitative studies reveal that there migrated cells occur primarily in the lateral are species that have substantial cell migra- zone of the tectum. There are few migrated tion (i.e., E. bicolor and D. mexicanus)(.9%). cells in the medial zone of the tectum (Fig. 5). T. natans possesses a moderate number of Quantitative measurements indicate that migrated cells (5%), whereas limited cell mi- the degree of neuronal migration is corre- gration (,4%) exists in S. uluguruensis and lated with the relative size of the white B. boulengeri. The lowest number of mi- matter (Table 3). Species in which the per- grated cells occurs in I. russelli, which has centage of the white matter is relatively very few or even no migrated cells (Fig. 5a). high (67.3% in D. mexicanus and 65.5% in E. I. kohtaoensis, S. kirkii, B. taitana, B. boulen- bicolor) have the greatest degree of neuronal 22 A. SCHMIDT AND M.H. WAKE

Fig. 7. Degree of neuronal migration in the telen- have extensive neuronal migration. Photomicrographs cephalon. Horizontal sections; all photos are of compa- for panels d and e are higher magnifications of the rable regions of the telencephalon. Idiocranium russelli medial pallium of I. russelli (d) and of T. natans (e). (a,d) has the least migration; S. grandisonae (b) has Bars 5 100 µm. moderate migration; U. narayani (c) and T. natans (e)

migration (9.6% in D. mexicanus and 9.5% and limited neuronal migration (0.5–3.3%). in E. bicolor). T. natans has a moderate The lateral zone of the tectum is character- percentage of white matter (60.3%) and is ized by a higher percentage of white matter characterized by a moderate degree of cellu- (59.1–76.1%) and more migrated cells (5.1– lar migration (5%). Further, a lower percent- 15.5%). age of white matter is correlated with a lower degree of neuronal migration (i.e., S. Telencephalon (Figs. 6, 7) uluguruensis with 55.8% and 3.7%, respec- There are great differences in the degree tively). An exception is B. boulengeri, which of cellular migration in the telencephalon of possesses a relatively large amount of white the species examined (Tables 2, 3; Figs. 6, 7). matter (63.1%) but only little neuronal mi- The least cellular migration in the telen- gration (3.8%). The correlation between the cephalon was found in E. bicolor (Fig. 6a) degree of neuronal migration and the rela- and E. petersi. Boulengerula taitana, B. bou- tive size of the white matter not only fits lengeri (Fig. 6b), I. russelli (Fig. 7a,d), Sylva- interspecifically but also for different tectal caecilia grandisonae (Fig. 7b), C. occiden- regions. In all species examined quantita- talis, S. kirkii, H. rostratus, and I. tively, the medial zone of the tectum has a kohtaoensis have more, but still not many, low percentage of white matter (48.5–57.4%) migrated cells. T. natans (Fig. 7e), C. indis- CAECILIAN BRAIN 23 tinctum, S. uluguruensis (Fig. 6c), U. nara- tectum mesencephali of salamanders, even yani (Fig. 7c), G. seraphini, D. mexicanus, though the degree of lamination and the and G. multiplicata have a substantial num- number of migrated cells in the most devel- ber of migrated cells. oped caecilian tectum (E. bicolor) are greater than in most salamanders. In salamanders, Nuclear/cell size (Table 1) only ambystomatids (e.g., Ambystoma mexi- Nuclear sizes range from 5.6 µm (S. ulugu- canum [personal observation]) possess a de- ruensis) to 8.1 µm diameter (G. multipli- gree of lamination and cellular migration cata) (Table 1). The correlation of nuclear that is similar to that of E. bicolor. In the size with genome size and cell size has been past, the complexity of brain morphology discussed extensively (Cavalier-Smith, ’78, often was used as a parameter for evaluat- ’82; Horner and Macgregor, ’83; Sessions, ing brain evolution. A complex brain was ’84; Sessions and Larson, ’87; Szarski, ’76, regarded as representing a derived condi- ’83). Because of the very limited cytoplasm tion, while a simple morphology was thought surrounding the nuclei of the brain cells to be the plesiomorphic condition (Herrick, measured, we use nuclear diameters as a ’48; Bullock, ’84). Following these argu- referent to cell size in this discussion. In ments, the anuran tectum would be the most general, animals that have a free-living derived, while the salamander tectum would aquatic larval stage in their life histories (E. represent the plesiomorphic condition; the bicolor,E. petersi, I. kohtaoensis, S. grandiso- caecilian tectum would be morphologically nae, and (larval stage presumed) C. occiden- intermediate between that of those two or- talis and U. narayani) are characterized by ders. However, a phylogenetic analysis of large cells (6.8–7.9 µm). Viviparous species tectal morphology among vertebrates (Roth are divided into two groups, with the New et al., ’93) revealed that the simple morphol- World taxa G. multiplicata, D. mexicanus, T. natans, and C. indistinctum possessing large ogy is an apomorphic trait that must be cells (7.1–8.1 µm) and the African S. kirkii, considered a secondary simplification rather S. uluguruensis, and G. seraphini possess- than a primitive feature. Thus, in contradic- ing small cells (5.6–6.2 µm). Species that tion to the above-mentioned hypothesis, the have direct development (B. taitana, B. bou- laminated tectum represents the primitive lengeri (presumably), H. rostratus, and I. condition, whereas the simplified tectum is russelli) are characterized by small cells (5.6– derived. A comparison of tectal morphology 5.9 µm). in different caecilian species clearly sup- Cell sizes correlate with body sizes to some ports the latter hypothesis. E. bicolor and E. degree only for the small species examined. petersi, members of the Rhinatrematidae, However, the cells of I. russelli, a truly min- the basal family of the lineage (Duellman iaturized species (Wake, ’86), are no smaller and Trueb, ’86; Nussbaum and Wilkinson, than those of small but less dramatically ’89; Hedges et al., ’93) (Fig. 1), have the miniaturized species. Similarly, there is con- highest degree of morphological complexity siderable variation in body size and propor- and migration. In these species all tectal tions among the ‘‘large-celled’’ taxa with layers that have been described in the an- aquatic larvae and among the viviparous uran tectum (Potter, ’69; Sze´kely and La´zar, African ‘‘small-celled’’ forms. Given the small ’76) can be distinguished, even though they number but wide range of taxa examined, it are not as elaborated as in anurans. The appears that reproductive mode (see Table 1 tectal morphology of many derived species and Discussion) and biogeography are bet- (i.e., B. boulengeri, S. kirkii, and S. uluguru- ter correlated with cell size than is body size. ensis) is rather simple, in the sense that only one thick periventricular cellular layer and DISCUSSION a superficial fiber layer can be distinguished. Simplification of the tectum mesencephali There are only a few migrated cells. These in caecilians—a secondarily derived results are in concordance with comparative process? studies on the tectum mesencephali of sala- Our studies indicate that the tectum mes- manders. Roth et al. (’93) demonstrated that encephali in caecilians not only is small com- in salamanders the simplest tectal morphol- pared to other brain regions but is character- ogy is found in those species that are highly ized by a low degree of morphological derived (i.e., the bolitoglossines) (see also complexity. It more closely resembles the Roth and Schmidt, ’93). 24 A. SCHMIDT AND M.H. WAKE Simplification of the caecilian tectum lamination in the caudal than in the rostral mesencephali—a consequence of tectum (Fig. 3g,h). This cannot be explained heterochrony? by paedomorphosis, because the caudal tec- Comparative studies of tectal morphology tum is the last part to develop. There are two in vertebrates in general, as well as in sala- possible routes by which a greater degree of manders and frogs in particular (Roth et al., lamination in the caudal tectum could occur. ’92, ’93, ’94), suggest that the simplification First, a condensation of early ontogenetic of the nervous system is a consequence of processes and an extension of late processes paedomorphosis (late developing features of might have occurred in these animals. Sec- outgroups are retarded or even missing in ond, in caecilians most aspects of the reduc- members of the lineage in question). A wide- tion of the visual system appear to occur in spread influence of paedomorphosis on sala- later ontogenetic stages (Wake, ’85). In lar- mander evolution has been described for vae the visual system is better developed many morphological traits (Wake, ’66) and than in adults. Owing to the fact that the also exists for the brain (Roth et al., ’92, ’93; caudal tectum is that part of the tectum Schmidt and Roth, ’93); for example, al- which develops latest, it might have a greater though the general pattern of tectal develop- potential for plasticity at late developmental ment in the anuran Rana temporaria and stages than does the rostral tectum that the salamander Pleurodeles waltl is similar, develops earlier. Thus, as a consequence of certain developmental traits, especially those reduction of the visual system at late devel- that appear at later ontogenetic stages of opmental stages, tectal afferents represent- Rana (e.g., the migration of cells to the super- ing other than visual modalities might have ficial tectal layers), are reduced in Pleurode- been expanded in the caudal tectum in some les (Schmidt and Roth, ’93). The tectum of caecilian species. frogs and salamanders develops along a gra- In frogs and salamanders, a negative cor- dient from rostral to caudal as well as from relation exists between morphological com- lateral to medial; that is, the lateral part of plexity of the brain and genome and cell size the rostral tectum is the region where cellu- (Roth et al., ’92, ’94). This correlation may be lar proliferation, lamination, and cellular explained by the fact that increases in ge- migration occur earliest, and the medial part nome sizes generally lead to decreases in of the caudal tectum is the part where these rates of cell metabolism, cell proliferation, processes occur latest. In adult salamanders and differentiation (Sessions and Larson, in general, there are very few migrated cells. ’87; Szarski, ’76, ’83; Cavalier-Smith, ’78, There are no differences concerning the dis- ’82; Horner and Macgregor, ’83). Substantial tribution of migrated cells in the medial and differences in genome sizes exist among am- lateral tectum. This suggests that some early phibians. Frogs have small genomes (0.9–19 as well as later developmental processes are pg DNA/haploid nucleus [Mahony, ’86; Mo- suppressed in these animals. In caecilians, a rescalchi, ’90; Olmo, ’83]), while salamanders substantial number of migrated cells occurs are characterized by large genomes (13.7–83 in the lateral part of the tectum, which sug- pg [Hally et al., ’86; Olmo, ’83; Sessions and gests that only the phase of cell proliferation Kezer, ’91]); the genome sizes of caecilians within the medial proliferative zone that lie between those of frogs and salamanders corresponds to the last proliferative phase in (from 8–26 pg [Sessions and Wake, unpub- frogs is not expressed. In caecilians, the lat- lished data]). In general, the degree of lami- eral and the medial parts of the tectum not nation and cellular migration in caecilians is only differ in number of migrated cells, but in concordance with the concept that the also in the degree of lamination. For ex- degree of paedomorphosis is correlated with ample, in E. bicolor, the lateral part of the cell size. However, this correlation does not tectum has a clear distinction of different fit a comparison among caecilian species. In cellular and fiber layers, while the medial contradistinction to frogs and salamanders, part does not. Thus, retardation of late onto- which show a negative correlation between genetic processes in caecilians is not as ex- genome size and cell size and the degree of treme as it is in salamanders. In contrast to morphological complexity, there is a positive the situation in salamanders, in some caecil- correlation between cell size and morphologi- ian species (G. seraphini, H. rostratus, U. cal complexity in caecilians. Those species narayani, E. bicolor, E. petersi, S. grandiso- that have the smallest cell sizes have brains nae, B. taitana) there is a greater degree of that are relatively simple, whereas species CAECILIAN BRAIN 25 that have large cells are characterized by a repatterning such as that described for di- greater degree of morphological complexity. rect-developing salamanders (Wake and This suggests that there are additional fac- Roth, ’89), occurs. While changes in genome tors that influence brain morphology. Stud- size lead to a general slow-down of develop- ies in frogs and salamanders (Roth et al., ment that affects all structures in the same ’94) revealed that in salamanders one impor- way, patterns of development that affect dif- tant parameter related to the degree of lami- ferent structures in different ways require nation and cell migration is brain size. Brain additional factors. Studies on the salamander size is determined by the rate of cellular P. waltl (Schmidt and Bo¨ger, ’93) suggest proliferation, the rate of neurite outgrowth, that ontogenetic repatterning and miniatur- and the duration of development. Thus, the ization are under hormonal control. An in- degree of morphological complexity might crease in thyroxine during the aquatic lar- simply be a function of the time course of val stage in that species leads to these processes. In addition to changes in miniaturization and the development of a genome size, factors that are involved in mosaic of characters that resembles that of these processes are changes in the synthesis bolitoglossine salamanders. of growth factors and the stimulation by Even though there is no strict correlation afferents. between developmental strategy and brain Changes in rates of development of those morphology, our studies on cell migration processes necessarily lead to changes in mor- and lamination in the tectum mesencephali phological complexity. We suggest the follow- and telencephalon show that direct develop- ing evolutionary scenario for secondary sim- ment, together with minaturization, may plification of the caecilian brain. We start lead to the greatest simplification of the with the simplest form of developmental de- brain (compare together data in Tables 1 lay—all processes are affected equally. Ani- and 2). Idiocranium russelli possesses a tec- mals with large cells and a presumed low tum that is rather simple, and it has the rate of cellular proliferation would require a lowest degree of cellular migration in the longer time to reach a certain level of mor- tectum. This is in concordance with data phological complexity than animals with obtained in salamanders (Roth et al., ’90a, smaller cells and a higher rate of cellular ’94) showing that the simplest brain mor- proliferation. If the total time of develop- phology (in terms of lamination and degree ment is the same in both groups, those ani- of cellular migration) occurs in direct-devel- mals possessing small cells would be morpho- oping miniaturized species. Similarly, most logically more complex than would animals species of caecilians that are characterized with large cells. However, if development by a high degree of morphological complex- stops early in animals with small cells, they ity are species that have an aquatic phase of would have the same degree of morphologi- development with free-living larvae. cal complexity as an animal with large cells The most striking feature in viviparous and a longer developmental time. If develop- caecilians is the degree of cellular migration ment stops very early in species possessing within the telencephalon. All species that small cells, the degree of complexity would are characterized by extensive cellular mi- be even less than in animals with large cells. gration within the telencephalon (D. mexica- This may be the case in I. russelli, a minia- nus, G. multiplicata, G. seraphini, and T. turized progenetic (see Wake, ’86) caecilian natans) are viviparous (Tables 1, 2). These species that possesses very small cells. Since are species that have a poor or moderate cell size is correlated with the rate of cell degree of tectal morphological complexity. In proliferation (Sessions, ’84), the occurrence contrast, the least cellular migration within of small cells in miniaturized species may be the telencephalon occurs in E. bicolor and E. a compensatory effect for an early cessation petersi, those species that have the greatest of development. Developmental trajectories degree of lamination within the tectum and that affect different processes in different are oviparous with free-living larvae. It re- ways complicate this simple model. If some mains an open question whether intrinsic processes are retarded while, simultaneously, constraints in the development of the brain others are accelerated, a mosaic of simpli- exist, in the sense that the reduction of the fied (remaining from early ontogenetic morphological complexity in one brain re- stages) and advanced (representing later on- gion is correlated with an increase in com- togenetic stages) characters, an ontogenetic plexity in another brain region, or whether 26 A. SCHMIDT AND M.H. WAKE the simplification of the telencephalon in E. moderate to great morphological complexity, bicolor and E. petersi is due to a retardation except I. russelli, which has minimal lamina- of the development of the telencephalon par- tion. Most of these taxa also have rather ticular to these species, probably due to pae- extensive (E. bicolor, E. petersi, U. narayani, domorphosis, or whether the simplification G. seraphini, H. rostratus, D. mexicanus, of the telencephalon represents a plesiomor- T. natans) or moderate (I. kohtaoensis and phic condition in caecilians. S. grandisonae) tectal cell migration. The Visual system reduction and tectum specimens of S. grandisonae examined for morphology our neuroanatomical work were larvae, which appear to have better developed vi- Gymnophione amphibians are character- sual systems than do adults (Wake, ’85). ized by a reduction of the visual system, ranging from loss of the components of ac- Idiocranium russelli again may constitute a commodation and covering the eye with skin special case; it has a well-developed visual to the eye being represented only by an system but has minimal lamination and the amorphous retina in an orbit deeply embed- least number of migrated tectal cells. Ex- ded in musculature (Wake, ’85). Stimulation treme paedomorphosis (for caecilians) may by afferents has been reported to be a major be implicated in having a well-developed eye factor to trigger the morphogenesis of brain but a very simplified tectum. Taxa with well- regions (Finlay et al., ’87). An hypothesis developed visual systems, therefore, have that the reduction of the visual system is all degrees of morphological complexity and correlated with concomitant reduction or of cell migration in the tectum; species with simplification of tectal morphology is test- reduced visual systems all have limited to able with our data. We recognize three no lamination and moderate to extensive groups of species, based on components of tectal cell migration. Several aspects of the reduction of the visual system (Table 2). biology of the animals may influence these Group 1 has the most reduced eyes (extrin- generalizations. For example, the closely re- sic musculature absent, except for the de- lated S. uluguruensis and S. kirkii of the rived retractor tentaculi; lens absent, rudi- Family Scolecomorphidae have rather differ- mentary, or amorphous; optic nerve attenuate ent visual system morphologies, but their or highly attenuate; retina reduced, net-like tectal structure is remarkably similar. It is or amorphous). Taxa included in group 1 are possible that there is a developmental effect, S. uluguruensis, G. multiplicata, B. taitana, in that the brain may be more conservative and B. boulengeri. These taxa are also the than the visual system in its modification. A species that have the least tectal lamina- similar situation may obtain with the two tion. However, they have moderate to exten- members of the Family Typhlonectidae ex- sive cell migration in the tectum. Group 2 amined, T. natans and C. indistinctum, which includes forms with moderately reduced eyes have somewhat different visual systems but (two to four extrinsic muscles present in very similar tectal morphologies. It is not addition to the retractor tentaculi; lens dif- surprising that the two species examined of fuse to cellular; optic nerve intact; retinal Epicrionops, which have very similar eye layers well developed but with reduced num- bers of cells): C. occidentalis, S. kirkii, and development, also have virtually the same C. indistinctum. These taxa vary in morpho- tectal morphologies. Based on this small logical complexity in the tectum, S. kirkii taxonomic and morphological sample (but having very little, C. indistinctum having inclusive of representatives of three of the moderate, and C. occidentalis extensive mor- five families of caecilians), we see no strict phological complexity. The two latter taxa correlation of visual system reduction with have considerable numbers of migrated tec- tectal morphologies. This may be the case tal cells. S. kirkii possesses a moderate num- because visual afferents are not the only ber of migrated cells. Forms with well- afferents reaching the tectum. Afferents be- developed eyes (all extrinsic muscles present; longing to other systems also may regulate lens well developed; optic nerve intact; retina tectal development. In this context, the cor- well developed) constitute group 3 and in- relation between the relative amount of tec- clude E. bicolor, E. petersi, I. kohtaoensis, tal neuropil and the degree of cellular migra- U. narayani, D. mexicanus, H. rostratus, tion is important. In those species in which G. seraphini, S. grandisonae, I. russelli, and the percentage of white matter is relatively T. natans. The tecta of these taxa all have high, cell migration is more extensive than CAECILIAN BRAIN 27 in species that have a lower percentage of Humason, G.L. (1979) Animal Tissue Techniques, 4th white matter. An exception is B. boulengeri, ed. San Francisco: W.J. Freeman. Largen, M.J., P.A. Morris, and D.W.Yalden (1972) Obser- which has a moderate amount of white mat- vations on the caecilian Geotrypetes grandisonae Tay- ter but no more migrated cells than a species lor (Amphibia Gymnophiona) from Ethiopia. Monit. with relatively little white matter (i.e., S. Zool. Ital. 8:185–205. uluguruensis). This suggests that a certain Leghissa, S. (1962) L’evoluzione del tetto ottico nei bassi vertebrati. Arch. Anat. Embriol. 67:343–413. size of the tectal neuropil is a necessary but Mahony, M. (1986) Cytogenetics of Myobatrachid Frogs. not sufficient precondition for cellular migra- Ph.D. dissertation, Macquarie University, Ryde, New tion. South Wales, Australia. In summary, our data suggest that changes Manteuffel, G., and W. Himstedt (1985) Retinal projec- in brain morphology are consequences of a tions in the caecilian Ichthyophis kohtaoensis (Am- phibia, Gymnophiona). Cell Tissue Res. 239:689–692. mosaic of different influences. Whereas Morescalchi, A. (1990) Cytogenetics and the problem of changes in the peripheral sensory system do lissamphibian relationships. In E. Olmo (ed): Cytoge- not necessarily lead to different brain mor- netics of Amphibians and Reptiles. Basel: Birkha¨user, phologies, phylogenetic characteristics and pp. 1–19. life history strategies that are related to Nussbaum, R.A., and H. Hinkel (1994) Revision of East African caecilians of the genera Afrocaecilia Taylor changes in patterns of development may in- and Boulengerula Tornier (Amphibian: Gymnophiona: fluence the modification of brain morphol- Caeciliaidae). Copeia 1994:750–760. ogy. Nussbaum, R.A., and M. Wilkinson (1989) On the classi- fication and phylogeny of caecilians (Amphibia: Gym- ACKNOWLEDGMENTS nophiona), a critical review. Herpetol. Monogr. 3:1–42. Olmo, E. (1983) Nucleotype and cell size in vertebrates: We thank Gloria Wurst, Kath Thomas, A review. Bas. Appl. Histochem. 27:227–256. Willy Bemis, Kristen Nygren, and Enrique Potter, H.D. (1969) Structural characteristics of cell and Lessa for preparing specimens and Gisela fiber populations in the optic tectum of the frog (Rana Kru¨ ger for photographic assistance. We ap- catesbeiana). J. Comp. Neurol. 136:203–232. Roth, G. (1987) Visual Behavior in Salamanders. Berlin: preciate helpful discussions with Gerhard Springer. Roth and David Wake and their comments Roth, G., and A. Schmidt (1993) The nervous system of on early drafts of the manuscript. This work plethodontid salamanders, insight into the interplay was supported by the National Science Foun- between genome, organism, behavior, and ecology. Her- dation (M.H.W.: BSR92-07555) and by the petologica 49:185–194. Roth, G., B. Rottluff, W. Grunwald, J. Hanken, and R. German Science Foundation (A.S.: DFG, Linke (1990a) Miniaturization in plethodontid sala- Schm 833/3-2). manders (Caudata: Plethodontidae) and its conse- quences for the brain and the visual system. Biol. J. LITERATURE CITED Linn. Soc. 40:165–190. Bullock, T.H. (1984) The future of comparative neurol- Roth, G., C. Naujoks-Manteuffel, and W. Grunwald ogy. Am. Zool. 24:693–700. (1990b) Cytoarchitecture of the tectum mesencephali Cavalier-Smith, T. (1978) Nuclear volume control by in salamanders: A Golgi and HRP study. J. Comp. nucleoskeletal DNA, selection for cell volume and cell Neurol. 278:181–194. growth rate, and the solution of the C-value paradox. Roth, G., U. Dicke, and K. Nishikawa (1992) How do J. Cell Sci. 34:247–278. ontogeny, morphology, and physiology of sensory sys- Cavalier-Smith, T. (1982) Skeletal DNA and the evolu- tems constrain and direct the evolution in amphib- tion of genome size. Annu. Rev. Biophys. Bioeng. 11: ians? Am. Nat. 139:S105–S124. 273–278. Roth, G., C. Naujoks-Manteuffel, K. Nishikawa, A. Duellman, W.E., and L. Trueb (1986) Biology of Am- Schmidt, and D.B. Wake (1993) Paedomorphosis and phibia. New York: McGraw Hill, Inc. simplification in the nervous system of salamanders. Finlay, B.L., K.C. Wikler, and D.R. Sengelaub (1987) Brain Behav. Evol. 42:137–170. Regressive events in brain development and scenarios Roth, G., J. Blanke, and D.B. Wake (1994) Cell size for vertebrate brain evolution. Brain Behav. Evol. predicts morphological complexity in the brains of 30:102–117. frogs and salamanders. Proc. Natl. Acad. Sci. U. S. A. Hally, M.K., E.M. Rasch, H.R. Mainwaring, and R.C. 91:4796–4800. Bruce (1986) Cytophotometric evidence of variation in Schmidt, A., and T. Bo¨ger (1993) Thyroxine induces a genome size of desmognathine salamanders. Histo- decoupling of ontogenetic processes in urodeles: Conse- chemistry 85:185–192. quences for the visual system. In N. Elsner and Heisen- Hedges, S.B., R.A. Nussbaum, and L.R. Maxson (1993) berg (eds): Genes—Brain—Behaviour. Proceedings of Caecilian phylogeny and biogeography inferred from the 21th Go¨ttingen Neurobiology Conference. Stutt- mitochondrial DNA sequences of the 12S rRNA and gart: Thieme, p. 762. 16S rRNA genes (Amphibia: Gymnophiona). Herpet. Schmidt, A., and G. Roth (1993) Patterns of cellular Monogr. 7:64–76. proliferation and migration in the developing tectum Herrick, C.J. (1948) The Brain of the Tiger Salamander. mesencephali of the frog Rana temporaria and the Chicago: Chicago University Press. salamander Pleurodeles waltl. Cell Tissue Res. 272: Horner, H.A., and H.C. Macgregor (1983) C-value and 273–287. cell volume: Their significance in the evolution and Schmidt, A., and M.H. Wake (1990) Olfactory and vom- the development of amphibians. J. Cell Sci. 63:135– eronasal systems of caecilians (Amphibia: Gymnoph- 146. iona). J. Morphol. 205:255–268. 28 A. SCHMIDT AND M.H. WAKE

Schmidt, A., and M.H. Wake (1991) Phylogenetic changes In H. Vanegas (ed): Comparative Neurology of the in the morphological differentiation of the caecilian Optic Tectum. New York: Plenum, pp. 93–120. tectum mesencephali. In N. Elsner and H. Penzlin Wake, D.B. (1966) Comparative osteology and evolution (eds): Synapse—Transmission—Modulation. Stutt- of the lungless salamanders, family Plethodontidae. gart: Thieme, p. 335. Mem. So. Calif. Acad. Sci. 4:1–111. Sessions, S.K. (1984) Cytogenetics and Evolution of Sala- Wake, D.B., and G. Roth (1989) The linkeage between manders. Ph.D. dissertation, University of California, ontogeny and phylogeny in the evolution of complex Berkeley. Sessions, S.K., and J. Kezer (1991) Evolutionary cytoge- systems. In D.B. Wake and G. Roth (eds): Complex netics of bolitoglossine salamanders (Family Pleth- Organismal Functions: Integration and Evolution in odontidae). In D.M. Green and S.K. Sessions (eds): Vertebrates. New York: Wiley, pp. 361–377. Amphibian Cytogenetics and Evolution. San Diego: Wake, D.B., K. Nishikawa, U. Dicke, and G. Roth (1988) Academic Press, pp. 89–130. Organization of the motor nuclei in the cervical spinal Sessions, S.K., and A. Larson (1987) Developmental cord of salamanders. J. Comp. Neurol. 278:195–208. correlates of genome size in plethodontid salamanders Wake, M.H. (1985) The comparative morphology and and their implications for genome evolution. Evolu- evolution of the eyes of caecilians (Amphibia, Gymno- tion 41:1239–1251. phiona). Zoomorphology 105:277–295. Szarski, H. (1976) Cell size and nuclear content in Wake, M.H. (1986) The morphology of Idiocranium rus- vertebrates. Int. Rev. Cytol. 44:93–111. selli (Amphibia: Gymnophiona) with comments on min- Szarski, H. (1983) Cell size and the concept of wasteful iaturization through heterochrony. J. Morphol. 189:1– and frugal evolutionary strategies. J. Theor. Biol. 105: 16. 201–209. Sze´kely, G., and G. La´za´r (1976) Cellular and synaptic Wake, M.H. (1992) Patterns of peripheral innervation of architecture of the optic tectum. In R. Llina´s and W. the tongue and hyobranchial apparatus in caecilians Precht (eds): Frog Neurobiology. Berlin: Springer, pp. (Amphibia: Gymnophiona). J. Morphol. 212:37–53. 407–434. Wilkinson, M. (1992) On the life history of the caecilian Vanegas, H., S.O.E. Ebbesson, and M. Laufer (1984) genus Uraeotyphlus (Amphibia: Gymnophiona). Her- Morphological aspects of the teleostean optic tectum. petol. Jour. 2:121–124.