JOURNAL OF MORPHOLOGY 165:117-130 (1980)

Morphometrics of the Skeleton of mexicanus (Amphibia: ). Part 1. The Vertebrae, With Comparisons to Other MARVUE H. WAKE Department of Zoology, and Museum of Vertebrate Zoology, Uniuersity of California, Beikeley, California 37.20

ABSTRACT Morphometric analysis of vertebral structure in (Am- phibia: Gymnophiona) is presented. Ontogenetic variation in Dermophis mexi- canus is analyzed through the 100+ vertebrae composing the column. Vertebral structure in adult D. mexicanus is compared with that in Zchthyophis glutinosus and Typhlonectes compressicauda. Centra of the atlas, second, tenth, 20th, and 50th vertebrae grow at allometrically different rates in D. mexicanus, though the 20th and 50th are not significantly different. Growth appears significantly slower in several dimensions of anterior and posterior vertebrae relative to mid- trunk vertebrae in all three species. Mensural patterns throughout the entire column are similar in the terrestrial burrowers D. mexicanus and Z. glutinosus; patterns in the aquatic T. compressicauda differ substantially from those of the burrowing species and are strongly influenced by allometry. Of the 112 D. mexicanus examined. 13.4% had vertebral anomalies, usually fusions.

Some attention has been paid to vertebral atlas and 95-285 trunk vertebrae, lack a sac- structure in caecilians and to the development rum, and usually lack a tail (Taylor, ’68). Two of vertebrae, particularly in attempts to un- size samples of D. mexicanus (family Caecili- derstand variation among the three orders of idae) plus an ontogenetic series of juveniles Recent . A number of descriptions and adults were examined and compared to or figures of vertebrae occur in the literature, samples of I. glutinosus () and usually of only one or a few vertebrae from a T. compressicauda (Typhlonectidae). Thus, single specimen to represent a species; for three of the five families of caecilians were example, Chthonerpeton indistinctum (Wied- sampled; the species represent both terrestrial ersheim, 18791, Siphonops annulatus, Schis- and aquatic ways of life. tometopum thomensis, oxyurus, This paper, and another in preparation on Hypogeophis rostratus, and Z. glutinosus (Pe- the morphometrics of the skull of D. mexican- ter, 1894), Hypogeophis rostratus (Lawson, us, provide information on osteological varia- ’63), anterior four or five vertebrae of 24 spe- tion. It has been stated (see Lawson, ’63) that cies (Taylor, ’771, and the fossil Apodops pricei there is no significant variation in structure and six recent species (Estes and Wake, ’72). and proportions of vertebrae following the More attention has been paid to development atlas. I therefore wished to test the alternative of vertebrae than to their adult structure by hypothesis that there is significant variation several authors: Marcus and Blume (’26), Mar- among vertebrae in the column. I suspected cus (’34, ’371, and Lawson (’63, ’66) on H. that the structure of vertebrae in “neck,” body, rostratus; Ramaswami (’58) on I. ‘klutinosus”; and posterior regions of the column might be Mookerjee (’42) on an unidentified form, prob- significantly different, that growth pattern ably an Zchthyophis; Wake (’70) on Gymnopis might also vary, and any differences might be multiplicata and T. compressicauda, largely in correlated with locomotor (and, to some de- analysis of ordinal relationships. gree, feeding) functions. Further, since most Regional variation throughout the vertebral work on vertebrae has been typol- column of caecilians has not yet been exam- ogical, it seemed useful to provide data on ined, nor has the comparative ontogeny of linear, ontogenetic, and population variation vertebrae. In contrast to most other elongate in vertebral structure. An understanding of amphibians and reptiles, caecilians have an regional, age-related, and populational varia-

0362-252518011652-0117$02.60@> 1980 ALAN R. LISS. INC. 118 MARVALEE H. WAKE! tion allows more effective assessment of tax- that the position of the vertebrae and the onomic comparisons in the literature, such as quality of the x-ray allowed careful measure- those of Peter (1894) and Estes and Wake ment. Descriptive statistics were calculated (‘72). It also provides a frame of reference for using SPSS computer programs. allocation of isolated vertebrae, such as fossils, to body regions and possibly to species. These DESCRIPTION OF VERTEBRAE data also provide a baseline for further work Vertebral column on the functional morphology of the vertebral column and assessment of regional, ontoge- Vertebral counts in the popul@ion of D. netic, and interspecific variation in function. mexicanus range from 100 to 112 (x = 106.19, s = 2.03, mode = 105; Fig. 1). Contrary to Lawson’s (’63) impression for H. rostratus, MATERIALSANDMETHODS there is considerable regional variation in the For the study of ontogenetic and adult var- features and proportions of vertebrae, al- iation, material from a single population of D. though regions grade into one another, in all mexicanus collected near San Rafael Pie de la three species examined. The atlas is easily Cuesta, San Marcos, , was pre- distinguished from other vertebrae. In addi- pared. Five specimens from each of two size tion, anterior vertebrae are characterized by classes (total lengths 190k 5 mm and 370t 4 the presence of a foramen (or pair of foramina) mm) were prepared by inserting nichrome for the dorsal and ventral roots of the spinal wire through the neural arches of all verte- nerves (these pass between vertebrae in the brae of freshly killed , then macerat- rest of the column beginning at the 15th to ing and drying the bones. Vertebrae were thus klst vertebra in D. mexicanus, the fifth to maintained in order but freely movable. Meas- 11th in T. compressicauda, and the fourth in urements of neural arch length (NAL) and I. glutinosus) and a nuchal keel. Posterior anterior (AW) and posterior widths (PW), vertebrae have progressively reduced verte- width across transverse processes (TVPW), bral extensions such as the transverse proc- left parapophysis length (PPL), centrum esses, parapophyses, and the ventral keel so length (CL), posterior centrum depth (CHK), that posteriormost vertebrae are irregularly depth of central keel at mid-vertebra (K), shaped rings of bone around the end of the posterior vertebral height from neural arch spinal cord, composed only of neural arch and rim to bottom of centrum (PVH), and height centrum. Descriptions of vertebrae are based of neural (nuchal) keel (NK) were measured primarily on the D. mexicanus sample, but on an EPOl microscope mounted with a digi- apply to the three species examined, with tal micrometer which read to 0.001 mm. exceptions as noted. Means and 95% confidence limits were calcu- lated for both samples. All measurements were taken on all 106 vertebrae of one speci- 1 N=102 men and plotted against vertebral number in I x=106.1863 sequence. It was thereby determined that var- iation could be adequately evaluated from measurements of each of the first 20 vertebrae, then every fifth vertebra to the 80th, and finally alternate vertebrae to the end of the column. Spinal nerve foramina were also ex- amined. For comparison dried and strung ver- tebrae of an I. glutinosus (350 mm total length) of the primitive family Ichthyophiidae and two T. compressicauda (350 and 362 mm total length) of the aquatic family Typhlonec- tidae were also counted and measured. In addition, vertebrae were counted and length of the centrum of the atlas and second, tenth, 20th, and 50th vertebrae of each specimen was measured with a dial calipers on x-rays Number of vertebrae of 102 preserved specimens from 108 mm to 448 mm total length. These 102 specimens Fig. 1. Number of vertebrae vs. number of individuals were selected from-more than 300 x-rayed so in a sample of D.merzcanus from one population. MORPHOMETRICS OF CAECILIAN VERTEBRAE 119

Atlas. The atlas is modified to provide which reaches its maximum at vertebra 13 to large atlantal cotyles for articulation with the 16. The nuchal keel is absent, obliterated occipital condyles. The centrum and neural variously from vertebra 10 to 20, and having arch are significantly shorter (p < .05) than its greatest height in vertebra 4 through 9. those of succeeding vertebrae (Figs. 2 and 4). livo of the large D. mexicanus retain an an- The nuchal keel and ventral keel (except in terior notch through which the spinal nerve Ichthyophis, which has a slight keel) are ab- passes. sent, as are transverse processes and elongate parapophyses. A small ventral spinal root for- 50th Vertebra. The centrum of the 50th amen is present behind the rim of each cotyle. vertebra is not significantly longer than that There is no indication of a tuberculum interr of the 20th (p > 0.9, student’s T-test, Fig. 4). glenoideum. Anterior width of the neural arch Most dimensions are diminishing slightly or is significantly less than that of succeeding have plateaued (Figs. 5-9). vertebrae, but posterior dimensions are not Posterior Vertebrae. Most of the vertebral significantly different. The atlas and the ter- dimensions begin to decrease at about the minal three to six vertebrae are the only ones 80th vertebra. By the 100th vertebra, trans- in the column that do not bear ribs. verse processes (and ribs), parapophyses, and the ventral keel are no longer present. The Second Vertebra. The second vertebra has ventral keel is lost at about the 80th vertebra a significantly longer centrum (p < .001, stu- in juvenile D.mexicanus (Fig. 7)and I. gluti- dent’s T test, Fig. 4) and neural arch than the nosus, the 90th in T. compressicauda. Poste- atlas. The centrum is hourglass-shaped. The riormost vertebrae are neural arch-centrum vertebra has well developed transverse proc- rings around the terminal fibers of the spinal esses, elongate, anteriorly curving parapo- cord. They are irregularly shaped and usually physes (the ventral transverse processes of fused in sets of two or three vertebrae. Only Lawson, ’63) with dorsal diapophyses, short the last two to five vertebrae extend posterior postzygapophyses, a nuchal keel, a pro- to the end of the vent (up to ten in I. glutino- nounced ventral keel that expands and bifur- sus). There is no indication of a sacrum. cates anteriorly to join the parapophyses, and paired spinal nerve foramina anterior to the Ontogenetic variation in D. mexicanus transverse processes. In adults, centrum length increases among the first three vertebrae, plateaus to the tenth Tenth Vertebra. The tenth vertebra has a then increases steeply to the 20th. Length centrum longer than that of the second (p < stays about the same until the 60th, then .l, student’s T-test, Fig. 4). Other processes, begins to decline slowly, and precipitously as described for the second vertebra, are pres- after the 90th. Juvenile centrum length pla- ent and larger than those of the second ver- teaus from the third to the 40th vertebra, then tebra, except for the nuchal keel, which is declines gradually to the 90th, then declines smaller, having reached its greatest height in further to the terminal vertebra. vertebrae four to six. Diapophyses of anterior The neural arch length is proportionally vertebrae ofI. ghtinosus are more pronounced similar in the second through tenth vertebrae than those of the other species (Fig. 3). The of both juveniles and adults. There is a slight spinal nerve exit shows substantial variation, increase in length, in juveniles, of the tenth being paired in three adults of D. mexicanus, through 20th vertebra, a more precipitous single with a notch for the other root in one, increase in adults. Length then gradually de- and single (fused) in another; it is single in clines in both samples (Fig. 5). the juveniles of D. mexicanus examined. It is Widths of the neural arch show ontogenetic single in T. compressicauda, and in I. gluti- variation (Fig. 6). Posterior width increases nosus the nerve passes between vertebrae af- slightly through the first 20 vertebrae, then ter the fourth vertebra, which has paired an- declines in adults. Juveniles show a consistent terior foramina (Fig. 3). decrease from atlas to terminal vertebra. An- 20th Vertebra. The centrum of the 20th terior width increases slightly in adults vertebra is significantly longer than that of though the first 20 vertebrae, then declines, the tenth (p < .01) and of the second (p < more precipitously in the final 20. Juveniles .001), student’s T-test, Fig. 4). Measurements show nearly equal anterior widths among the of most comDonents are still increasing (Figs. first 20 vertebrae, then a gradual decrease

5-9), except for posterior vertebral height, through0 the entire column. 120 MARVALEE H. WAKE I 2 3 4 5 a n b n

d

e

f

Fig. 2. Vertebral structure in juveniles and adults of D. mezicunus. Column 1, atlases; column 2, second vertebrae; column 3, tenth vertebrae, column 4, 50th vertebrae; column 5, near-terminal vertebrae. Row a, dorsal view, juveniles; row b, dorsal view, adults; row c, lateral view, juveniles; row d, lateral view, adults, row e, ventral view, juveniles; row f, ventral view, adults. Bar = 2.0 mm. MORPHOMETRICS OF CAECILIAN VERTEBRAE 121

I 2 3 4 a

b

C

d

e

f

Fig. 3. Vertebral structure in I. glutinaws and T. compressicauda. Column 1, atlases; column 2, second vertebrae; column 3, tenth vertebrae; column 4,5Oth vertebrae; column 5, near-terminal vertebrae. Rows a-c, dorsal, lateral, ventral views ofl. glutinosus; rows d-f, dorsal, lateral, ventral views of T.compressicauda. Bar = 2.0 mm. 122 MARVALEE H. WAKE

Total length (mm)

Fig. 4. Correlation slopes (principal axis) for centrum length of atlas (--I, second vertebra ( .-.- I, tenth vertebra (- -I, 20th vertebra (-1, and 50th vertebra (-----I vs. total length of in an ontogenetic series of D. mexicanus. Differences in slopes indicate differences in growth rates, atlas slowest growing. ltuentieth and 50th vertebrae have similar growth rates. This is predicted by their nearly equal lengths, as shown in Figure 5A.

Transverse process width in adults is ap- Other measurements show a characteristic proximately the same for the second through increase among the first 20 vertebrae, a pla- 20th vertebrae of adults, then diminishes, teau or slight decrease from the 21st through more precipitously after the 80th. Juveniles the Both, then a marked decrease. In general, show a slight decrease in the anterior part of the decrease in size in posterior vertebrae is the column in that dimension, then a slightly much more precipitous in adults than in ju- increased diminution terminally, but not as veniles. Also, variance is greater in adults pronounced as in adults (Fig. 6). than in juveniles, as indicated by the greater Posterior vertebral height increases from 95% confidence intervals for most measure- the first through the 17th vertebrae in both ments on each vertebra in adults. This in- samples, much more in adults (increase in crease in variance with increase in size is mean, 3.45 mm to 4.39 mm) than in juveniles similar to that reported by Worthington ('71) (increase in mean, 2.10 mm to 2.41 mm) (Fig. in salamanders. 7),then gradually decreases in both. The height of the nuchal keel in adults Interspecific variation decreases from the seventh through the 20th The adult specimens of I. glutinosus, T. vertebrae, with large 9Wo confidence intervals compressicauda, and D. mexicanus selected for from the ninth on (since some adults lack the study are of similar total length to make keel from that vertebra on). The height of the comparisons more direct, but the ranges in keel increases in juveniles in the first through total lengths attained (I.glutinosus, 236-410 fifth vertebrae, then decreases, but with much mm [Taylor, '60, '681; 2'. compressicauda, less variation. The ventral keel increases in 262-800 mm [Moodie, '781; D. mexicanus, depth slightly in adults among the first 18 108-487 mm in the population sampled, to vertebrae, then it diminishes, but more grad- 600 mm for the species [Taylor, '681) suggest ually posteriorly. The keel depth in the ante- that though adult, individuals may not have rior column of juveniles is approximately the been of equivalent ages. I. glutinosus (Ichth- same, diminishing gradually through the 85th yophiidae) is a primitive egg-laying terrestrial vertebra, then increases slightly before being burrower that lives near streams in Sri Lanka lost (Fig. 7). (Ceylon). T.compressicauda (Typhlonectidae) MORPHOMETRICS OF CAECILIAN VERTEBRAE 123

0 0 0 - 4 - -0- 0 0 -0- -0- 4 0 4 -0- 0 *

t- 0

-0- 4- c

0

t t 0

t *

4 *

0

t 4

0 0

t t 0

-+ 4 I I 1 I I I I I I I I m cu - ww CIN A. bb

El 6. m

I 1 1 I 1 1 I 1 1 1 I 1 I I I I 1 I I I I 80 90 I00 MORPHOMETRICS OF CAECILIAN VERTEBRAE 125

a m

0 0 0 0 0 0

0

0

-a- *

0

Q

-0- I -I I I I I 1 I I I I I

UJUJ 126 MAFtVALEE H. WAKE is a viviparous, aquatic species of the Amazon species, with the slope of anterior increase in basin, while D. mexicanus (Caeciliidae) is a I. glutinosus slightly less steep than in D. viviparous terrestrial burrower inhabiting mexicanus. Both species decrease centrum lowland northern Central America. length posteriorly in comparable patterns. It Four sets of measurements are compared, appears that T.compressicauda increases cen- for they are representative of the patterns for trum length, though in progressively slighter all measurements in each species. Centrum increments, to a vertebra at about mid-body lengths show considerable variation among or just posterior, then decreases centrum adults of nearly the same size in the three lengths consecutively. In contrast, both I. glu- species. The curves that this measurement tinosus and D. mexicanus increase centrum describes for each vertebral column have length in anterior vertebrae, but have centra markedly different shapes (Figs. 5, 8, 9). The of very similar lengths through the 60 mid- atlas of T. compressicauda is shortest of the body vertebrae, then centrum lengths de- three, yet mid-body centra reach substantially crease posteriorly. “here is a step in the in- greater lengths (4.0-4.5 mm, 25th to 60th crease in length of anterior vertebrae at the vertebrae) than those of the other two species seventh to 10th vertebrae in all three species, (I glutinosus, 2.5-2.85 mm, 15th to 90th ver- then the slope of the curve rises. Thus, the tebrae; D. mexicanus, 3.03-3.25 mm, 18th to curve described by centrum length throughout 75th vertebrae). The range in size in this the vertebral column in T. compressicauda is series is 0.5 mm in T. compressicauda, 0.35 rounded, whereas those of the other two spe- mm in I. glutinosus, 0.25 mm in D. mexicanus, cies are flattened except at the extremes. over 35 of 98 mid-body vertebrae in T. com- The length of the neural arch through the pressicauda, 75 of 111 in I. glutinosus, and 57 column follows a pattern very similar to that of 110 in D. mexicanus. The slopes of the of the centrum for all three species (Figs. 5,8, increase and decrease in centrum lengths of 9). Neural arches are approximately 0.5 mm the first and last 20 vertebrae are much great- longer than centra for all vertebrae in I. glu- er in T. compressicauda than the other two tinosus and D. mexicanus, except for the atlas,

A.

1 I I 0 0 1 I I I 1 I I I 1 1 I 1

B.

0. 0 8

o! 1 1 I I I I I I I 1 I 0 20 40 60 80 100 Sequence of vertebrae

Fig. 8. A) Posterior width of neural arch (0)and width of vertebra across the transverse processes (i I) through the column of a specimen of I. glutinosus. B) Neural arch length (0) and centrum length (c ,) in I. glutinosus. MORPHOMETRICS OF CAECILIAN VERTEBRAE 127 in which the neural arch is 0.75 mm longer, The posterior width of the neural arch in D. and for the posterior 20 vertebrae, for which mexicanus increases through the tenth verte- the difference diminishes so that neural arch- bra, plateaus through the 20th, then dimin- es are nearly the same length as centra. In T. ishes (with some variation) to the 92nd ver- compressicauda neural arches through mid- tebra. The 93rd to 95th vertebra increase in body are 0.5-0.9 mm longer than centra. The width 0.25 mm, then width precipitously de- difference diminishes posteriorly. creases posteriorly. Posterior width is consist- Width across transverse processes (Figs. 5, ently less than that at the transverse process- 8, 9) reaches its greatest extent at the 20th es, usually 0.25 mm, except anterior to the vertebra in T. compressicauda and D. mexi- 19th vertebra, where the difference is less. canus, then gradually declines. In posterior- The pattern in T. compressicauda is very sim- most vertebrae the processes are barely dis- ilar to that of D. mexicanus (Figs. 6, 91, but cernable, so the measurement (for all verte- with a more gradual decline in width in pos- brae) is also one of vertebral width just ante- terior vertebrae. The pattern found in I. glu- rior to mid-vertebra. In I. glutinosus, however, tinosus differs substantially from that of the the greatest width is that of the third and other two species. Posterior width of the neur- fourth vertebrae; length of anterior vertebrae al arch is greater than that at the transverse diminishes to the tenth vertebra, remains processes in the first 20 vertebrae, and about about the same through the 20th, then grad- the same width through the rest of the column. ually declines through the posterior vertebrae. Greatest width is attained by the third and The processes in posterior vertebrae, as in fourth vertebrae, then it progressively dimin- other species, are barely discernable. The pos- ishes with a plateau from the ninth through terior 3-7 vertebrae in each species do not the 20th vertebra (Fig. 8). There is no indi- bear ribs; ribs are rudimentary in the three to cation that the "tail" vertebrae of Zchthyophis five vertebrae immediately anterior to these. (a few post-vent vertebrae are seen in x-rays,

. 0' A 0 0 .0 0 . . 0 . 0 0 . 0 . 0 . 0

I I I I I 1 I I I I 1

5 4 . ..0.. B 0 0% 0 . .Z 0 0 0 0 0 . .. 0 .....* 00 .. O 0 *. 000 00 :oooJ .. E3.*0 0 0. E 411 00" 0.

o! I 1 I I 1 1 I I I 1 0 20 40 60 80 I00 Sequence of vertebrae

Fig. 9. A) Posterior width of neural arch ( ') and width of vertebra across the transverse processes (0)through the column of a specimen of T. compressicuuda. B) Neural arch length (0)and centrum length (' ' ) in T. compressicauda. Compare withD. mexicanus (Figs 5-7) and I. glutinosus (Fig. 8). 128 hfARVALEE H. WAKE whereas these are not characteristic of non- Anomalies are summarized in Table 1. Of the ichthyophiids) are significantly different from 15 specimens from the population with an- posteriormost vertebrae of the other two spe- omalies, ten were specimens with total lengths cies, except perhaps the absence of any indi- greater than 320 mm; five were juveniles 201 cation of transverse processes. Functional im- mm or smaller. plications of these inter-specific differences are discussed below. Vertebral growth In D. mexicanus there is a highly significant Anomalies correlation between increase in centrum The most characteristic structural anomaly length and increase in total length (principal seen in the caecilian vertebral column in- axis correlation: atlas, r = .95; second verte- volves fusion of vertebrae (see Table 1).While bra, r = .97; tenth, 20th) and 50th vertebrae, this appears a normal state for posteriormost r = .98 for each; N = 102). The correlation vertebrae (of 112 D. mexicanus, 106 had two slope for the atlas is low relative to other or more fused vertebrae among the last six), vertebrae, and the slopes for the second, tenth, fusions more anteriorly are rare. Of the 112 and 20th increase proportionately to the atlas D. mexicanus, 13 had more anterior fusions. and to each in succession, suggesting increas- These were primarily among vertebrae in the ing growth rates for these vertebrae. The slope posterior half of the body, though two had for the 50th vertebra is similar to that of the fusions among the 20th to 24th vertebrae. Wo 20th, suggesting little difference in growth to four vertebrae were usually fused; only one rate (Fig. 4). This agrees with the pattern specimen had fusions in two body regions. The shown in Figures 5, 8, and 9, which show most marked situation was a small specimen differences in centrum length among anterior (111 mm total length) that had its posterior vertebrae, but little difference between the 40 vertebrae truncated and fused (possibly 20th and 50th vertebrae, indicating size dif- contributing to its small size). In one instance, ferences at one point in time that reflect the the right half of a vertebra was fused to the differences in growth rates suggested in Fig- preceding vertebra. ure 4. This also indicates allometric increase ?tyo other anomalies were observed. In one in size due to greater differential growth of specimen, an anterior vertebra was signifi- mid-body vertebrae relative to anterior (and cantly shorter than those surrounding it. In doubtless posterior) vertebrae. Functional im- three specimens, instances of two pairs of ribs plications of these differences are discussed per vertebra occurred. The transverse process- below. es of these vertebrae were broadened and There have been suggestions that vertebrae diapophyses either expanded or duplicated. number might increase with size (and age)

TABLE I. Vertebral Anomalies* Total length of Vertebral specimen number Anomaly 367 14 Truncate 366 20-21 Fused (2) 195 23-24 ” (2) 163 52-53 ’* (21 327 54-55 ” (21 201 68-72.74-75 ” (2 sets, 1 of 4, 1 of 2 vertebrae) 387 77-78, 88-90 ” (2 sets, 1 of 2, 1 of 3 vertebrae) 373 81-82 ” (21 329 88-89 ” (2) 190 90-91 ” (right 1/21 380 94-95 ” (21 384 98-99 ” (2) 111 60-100 ” (40);multiple ribs 327 54 387 62 * Size (total length) of specimens having vertebral anomalies, location of anomaly in the sequence of vertebrae in the column, and nature of the anomaly are presented. N = 112; 13.4% have anomalies. MORPHOMETRICS OF CAECILIAN VERTEBRAE 129

(Wiedersheim, 1879) or with the influence of (‘77) was correct in saying that caecilians have the “crawling” locomotion (Marcus “341, in a a “cervical” region and that the region is curiously Lamarckian explanation of in- modified to accommodate the stresses of bur- creased vertebral number and internal organ- rowing. However, the cervical region includes ization in limbless amphibians and reptiles). many more vertebrae than the four or five I suspect that the former question may have Taylor considered, and the region is not mod- arisen as morphologists observed the marked ified for pivotal movements of the cranium in cephalization of development in caecilian em- a general sense. If Taylor meant that “cervi- bryos, for posterior somites differentiate much cal” vertebrae represent an adaptation for later than more anterior ones. In the sample movement of the cranium in a single plane of 102 D. mexicanus, there is no correlation of and positioning the anterior end of the body, vertebral number with size (r = 0.1274). I concur. Differences in slopes of the anterior and DISCUSSION posterior vertebral dimensions in juveniles There are functional implications to the and adults of D. mexicanus appear to be the regional differences in vertebral structure and result of differences in growth. Anterior and to the interspecific differences in pattern. The posterior vertebrae appear to grow more slow- anterior 20 vertebrae in all three species and ly (Fig. 4) than mid-body vertebrae. In adults, in all sizes of D. mexicanus differ in several mid-body vertebrae have grown more relative respects from those of the rest of the column. to anterior and posterior vertebrae by exploit- The atlas is modified for articulation with the ing these differential growth rates over a skull and lacks ribs. Most of the anterior longer time than juveniles have had. The vertebrae have a longitudinal nuchal keel, adult curves are thus more pronounced for apparently produced according to the stress of several meristic features. the dorsal head musculature that attaches in Mid-body vertebrae, their ribs, and associ- the midline of these vertebrae. Parapophyses ated musculature are involved primarily in are short and widespread on these vertebrae, locomotion, as discussed by Marcus (’34), von and pre- and postzygapophyses are broader Schnurbein (’351, Gaymer (‘711, and Gans (’731, and flatter than those of more posterior ver- and are modified for flexibility in controlling tebrae. Observations of caecilians beginning movement on (or in) the substrate by having to burrow suggest that they stiffen the ante- well extended rib-bearers (less so in I. gluti- rior part of the column and position the head nosus than others), confined zygapophyses, for use as a trowel. The head is moved in an and greater length, making for longer posi- up-and-down burrowing motion, with the an- tional segments. Posterior vertebrae appear to terior vertebrae transmitting forces from re- have more limited locomotor function, but sistance of a more posterior body coil placed provide a positioning effect, especially in the on the surface of the substrate, and the resist- aquatic forms such as Typhlonectes, in which ance of the substrate to the head. There is the posterior part of the body has a dorsal little side-to-side motion of the head, and it is “fin” which may act as a rudder. Positioning usually due to movement of the entire anterior of the posterior part of the body is also impor- part of the body. Hence, the structure of the tant in copulation. anterior vertebrae permits little rotation of Major differences in pattern among these atlas on second vertebra, provides for “fixing” three species appear to be related to differ- vertebrae as a rod because of the association ences in growth rate of mid-body vertebrae of zygapophyses and parapophyses of adjacent and degree of extension of rib-bearing proc- vertebrae during burrowing, and the dorsal esses. The primitive, burrowing Zchthyophis longitudinal head musculature attaching to has the least accelerated midbody growth rate; the nuchal keel provides for forceful elevation the highly derived, aquatic Typhlonectes, the of the head during burrowing. Positioning of greatest rate (Figs. 8-9). In fact, Typhlonectes the head is also important within the burrow. appears to increase the growth rate to actual Observations of three species at rest in bur- mid-body, then decrease it slowly, rather than rows (under red light in glass containers) having an extended mid-body region composed indicate that much of the “resting time” is of vertebrae of equivalent size. This may be spent with the head in a slight enlargement an adaptation for aquatic locomotion and sine- of the burrow. The head is held elevated from wave generation in a surrounding medium. the floor of the burrow by contraction of the The rib-bearers of Zchthyophis are not as dorsal longitudinal head musculature and the extended as those of the other two species fixed position of the anterior vertebrae. This (Figs. 2, 3, 6, 8, 9). X-rays show that the ribs facilitates buccal respiration. Thus, Taylor of I. glutinosus are held more perpendicular 130 MARVALEE H. WAKE! to the vertebral axis than in the other species LITERATURE CITED and are shorter, broader structures. The men- Estes, R., and M.H. Wake (1972) The first fossil record sural data on transverse processes of Typhlo- of caecilian amphibians. Nature, 239:228-231. nectes relative to Dermophis are somewhat Gans, C. (1973) Locomotion and burrowing in limbless misleading, since the dimension is smaller in vertebrates. Nature, 2423414-415. Gaymer, R. (1971) New method of locomotion in limbless Typhlonectes. Though the measurement from terrestrial vertebrates. Nature, 234: 15&151. tip to tip of the processes is less in Typhlo- Lawson, R. (1963) The anatomy of Hypogeophis rostru- nectes than in Dermophis, the processes are at tus Cuvier (Amphibia, Apoda) Part I. The skin and least equivalently extended. The centra and skeleton. Proc. Univ. Durham Phil. Soc., 13:254-273. Lawson, R. (1966) The development of the centrum of neural arches of Typhlonectes are much more Hypogeophis rostrutlls (Amphibia, Apoda) with specid slender than those of Dermophis, as can be reference to the notochordal (intravertebral) cartilage. J. evaluated by inspecting the measurements of Morph., 118:137-148. the posterior width of the neural arch (Figs. Marcus, H. (1934) Beitrage zur Kenntnis der Gymno- phionen XXVII. iiber den Einfluss des Kriechens ad 6, 9). Wirbelzahl und Organgestaltung bei Apoden. Biol. Zent., It is probable that these several differences 54:5ia523. have to do with differences in the fine-tuning Marcus, H. (1937) Beitrag zur Kenntnis der Gymnophi- of locomotion. These conjectures await the test onen XXX. herMyotome Horizontal-septum und Rip- pen bei Hypogeophis und Urodelen. Zeit. Anat. En- of rigorous analysis using living animals rep- twgesch., 107:531-552. resenting several species and of various sizes Marcus, H., and W. Blume (1926) Beitrag zur Kenntnis within species. Yet, the meristic data for en- der Gymnophionen VII. iiber Wirbel und Rippen bei tire vertebral columns provide insight into the Hypogeophis nebst Bemerkungen uber Torpedo. Zeit. anat. Entwogesch., 80: 1-78. nature of ontogenetic and phylogenetic struc- Moodie, G.E.E. (1978) Observations on the life history tural variation. of the caecilian Typhlonectes compressicuudus (Dummeril and Bibron) in the Amazon basin. Can. J. Zool., ACKNOWLEDGMENTS 56: 10 05- 10 08. Mookerjee, H.K. (1942) On the development of the ver- The efforts of a number of people and insti- tebral column in Gymnophiona. Proc. 29th Ind. Sci. tutions made this work possible. W. C. Breck- Congr., 1942, pp. 15S160. enridge, John Cadle, Ted Papenfuss, Robert Peter, K. (1894) Die Wirbelsaule der Gymnophionen. Seib, David Wake, Thomas Wake, and the Ber. Naturf. Ges. Freiburg, 9t3S38. Ramaswami, L.S. (1958) The development of the apodan author collected specimens; Lorrie Kloster- vertebral column. Zool. Am.,161:271-280. man and Brian Sullivan x-rayed specimens; Taylor, E.H. (1960) On the caecilian specieskhthyophis Willy Bemis spent hours at the computer; glutimsus and Ichthyophis momhrous, with description Robert Colwell and Donald Straney provided of related species. Univ. Kansas Sci. Bull., 40:37-120. Taylor, E.H. (1968) The Caecilians of the World: A Tax- statistical advice; Phyllis Thompson drew the onomic Review. Kansas Univ. Press, Lawrence, 848 pp. vertebrae; Gene Christman executed the Taylor, E.H. (1977) Comparative anatomy of caecilian graphs; David Wake criticized the manuscript, anterior vertebrae. Univ. Kansas Sci. Bull., 51:21%231. and Teriann Asami-Oki typed it. The staff and von Schnurbein, A.F. (1935) Beitrag zur Kenntnis der Gymnophionen XXIII. Der Bewegungsapparat von Hy- students of the Museum of Vertebrate Zoology, pogeophis. Morph. Jahrb., 75:315330. University of California, Berkeley, were pa- Wake, D.B. (1970) Aspects of vertebral evolution in the tient while I used their measuring scope to modern Amphibia. Forma et Functio, 3:33-60. take the data. The Department of Zoology Wiedersheim, R. (1879) Die Anatomie der Gymnophi- onen. Verlag von Gustav Fischer, Jena. provided crucial funds for data analysis, and Worthington, R.D. (1971) Postmetamorphic changes in the National Science Foundation, through the vertebrae of Ambystom opacum Gravenhorst (Am- grant DEB 77-22642, supported these studies. phibia, Caudata). Sci:Ser. U&. Texas El Paso, 4:l-74.