"PROTOVIPERS" AND THE EVOLUTION OF SNAKE FANGS

KENNETHV. KARDONG Department of Zoology, Washington State University, Pullman, Washington 99164

Received September 9, 19i7. Revised August 2, 19i8

Anthony (1955) argued in his summary om toxicity. Certainly, long teeth (fangs) report on the venom apparatus of snakes are effective in delivering venom into a that the fang of viperid snakes evolved prey. It is possible however that injection from a maxillary tooth, specifically a tooth may not have been the initial selective at the posterior end of the maxilla. He factor favoring fang development. In- identified specific colubrid snake genera stead, enlarged teeth may have evolved which, when aligned into a morphocline initially to play a role in prey swallowing (sensu Maslin, 1952), showed an increas- and were secondarily modified into instru- ing similarity in maxillary tooth form to ments for venom injection (Kroll, 1976). the viperid solenoglyph fang. Since then, It is my purpose here 1) to identify possible a search for plausible nacent "protovipers" initial selective advantages of enlarged among the colubrids has continued teeth and, 2) to follow the implications of (Weaver, 1965; Kroll, 1976). enlargement in "protoviper" evolution. Certainly, colubrids are a promising group in which to search for clues to the evolution of the viperid fang and associ- ated apparatus. Over half (51.5%) of the Swallowinrr events were observed di- 309 species of colubrids examined by rectly and by cinematography followed by Marx and Rabb (1972) had enlarged teeth frame-by-frame analysis. Filming was at the posterior end of the maxilla and of done at speeds of 25, 50, or 200 frames these over one-quarter (26.8%) even had per second with a Hycam 16 mm cine- grooves in these enlarged teeth. The fam- matic camera before two 1000 watt tung- ily Colubridae includes many forms harm- sten photoflood lamps. Over two hundred less to humans, as well as some of the most meters of usable swallowing sequences toxic (e.g., Dispholidus, boomslang; Thel- were obtained of adult snakes of Pituophis otornis, bird snake). Toxicity, or at least melanoleucus, Crotaphopeltis hotamboeia, mild-toxicity, may be more widespread in and Xenodon merremii. Pituophis were colubrids than is appreciated. For in- fed mice and the two other species frogs. stance, some colubrids customarily con- .?\dditionally, single radiographs were tak- sidered non-venomous to humans occa- en of Pituophis and Crotaphopeltis at sionally deliver a painful bite (Heatwole various points in the swallowing cycle and and Banuchi, 1966; Willard, 1967; Johan- assembled into swallowing sequences bocke, 1974; Goellner, 1975). Since many much as described elsewhere (Kardong, colubrids have teeth similar in general ap- 1977). pearance to viperid fangs, and because Dissections were performed under a dis- some of these are mildly to highly ven- section microscope and skulls drawn with omous, many investigators have looked to the aid of a camera lucida attachment colubrids for viperid ancestors. (Wild). Teeth (shown in Fig. 4) were re- The changes leading to the evolution of moved from the posterior end of the max- the viperid fang have usually been attrib- illa from dried snake skulls, placed on uted to the selective advantages accruing mounting stubs, sputter coated with ap- from a progressively more efficient injec- proximately 30 nm of gold, and examined tion system together with increasing ven- with a scanning electron microscope FIG. 1. Three species of colubrids showing three degrees of relative length of posterior maxillaw teeth compared to other teeth borne by the maxilla (mx). .I)Pituophis melanolencus, B) Crotaphopeltis hotam- boeia, C) Dispholidus typus. THE EVOLUTIOS OF SSXKE FANGS 435

FIG. 2. Lateral view of Crotaphopeltis hotamboeia swalloaine the severed hind leg of a frog. Even& unfold left to right as the mavilla is carried forward and tilts In C, the mouth closes bringing the large posterior masillap teeth into the prey; in D the maxilla is retracted, pulling the pre>- into the mouth and, as it does, causing wringling (arrow) of the skin.

(ETEC .iutoscan UI). .inatomical termi- enlarged posterior maxillary teeth have nology follows Haas (1973). these secretion grooves or functional equivalents. For instance, in Thamnophis elegans vagrans (western terrestrial garter snake) the posterior maxillan teeth are Two aspects of the maxilla in colubrids enlarged as in Crotaphopeltis but lack a deserve mention because of the special sig- secretion groove or its equal. nificance they have for interpretation of Second, the maxilla of colubrids is "ro- tooth evolution. First, varying degrees of tatable" though much less so than in vi- posterior tooth enlargement can be found. perid snakes. In Pituophis the maxilla of For example, in Figure 1 all three are col- one side advances horizontally over the ubrid species yet among them there is con- prey during swallou7ing and is then siderable difference in relative size of teeth brought downward, its posterior end along the maxilla. In Pitz~ophis,the pos- swinging ventrally first so that rear teeth terior maxillary teeth are smaller than the re-engage the prey slightly before more anterior teeth. However, in Crotaphopel- anterior teeth. This is similar to the swal- tis, and especially in Dispholidus, the lowing motions of Elaphe described in posterior maxillary teeth are enlarged and greater detail b>- Xlbright and Selson the long posterior teeth bear a groove (1959a, b). down which secretions from the oral The maxilla of Crotaphopeltis describes glands can pass. Occasionally, the func- a similar displacement pattern (Fig. 2) ex- tional equivalent to such a secretion cept that during jaw advance over the groove is provided by the channel formed prey it experiences a greater degree of ro- between a pair of enlarged teeth (Taub, tation about its articulation with the pre- 1967). However, not all colubrids with frontal. This causes the maxilla to tip 436 KESSETH 1.. ELiRDOSG

FIG.3. Lateral photographs (top) of Crotaphopeltis hotamboeia showing the maxilla advancing forward (left)and during contact with prey (right) when mouth closes. Below each photograph is sketched the position of the maxilla (mx)relative to the prefrontal (pO about which it rotates.

sharply out of a horizontal plane (Fig. 3), vantage of the entire sweep of the retrac- resulting in upward swing of its anterior tion motion. end and downward swing of its posterior Long arc.-The posterior teeth are geo- end. As the posterior end of the maxilla metrically farther from the point of rota- is rotated ventrally during closure of the tion about the prefrontal than are the mid- jaws, the rear teeth are the first to pene- dle maxillary teeth. This means that trate into the prey and provide a firm pur- posterior teeth swing through a longer arc chase. Thus, in addition to whatever role than teeth closer to the point of rotation. they may play in prey capture, the en- Any object engaged by the posterior teeth larged posterior maxillary teeth of Crota- will likewise be advanced farther (longer phopeltis also function during swallowing arc) than if carried by more anterior max- to engage the prey firmly and aid in mov- illary teeth. The posterior teeth thus enjoy ing it past the angle of the jaws during a geometric advantage over other teeth of maxillary retraction. the maxilla in that they ride through a lon- ger arc and can thus advance the prey far- ther during each swing of the maxilla. Spikes.-If the prey presents special Engage first. -As described above, the mechanical problems in swallowing, then maxilla, during swallowing, experiences one design adaptation favored may be the displacement such that teeth borne on its elongation of the maxillary teeth. Certain- posterior end are first to engage and hence ly, other adaptive responses are conceiv- the first teeth to start moving the prey able (e.g., increase in tooth number, and1 along. If the prey was not advanced until or change in tooth orientation). However, mid or anterior teeth engaged it, then the lengthening of the teeth allows for much of the retraction motion of the max- deeper penetration. Further, with teeth of illa would have passed before the prey even length, the force of jaw closure is began to move. Since the posterior teeth experienced approximately equally by are the first teeth of the maxilla to engage each individual tooth. But, when a few the prey, the snake is able to take full ad- teeth (e.g., posterior maxillary teeth) proj- THE Ei.OLCTIOS OF SSAKE F.4SGS 43 7 ect beyond the even tooth row, this per- nodon) and viperid snakes. Ancestral to mits the force of jaw closure to be locally both these colubrid genera and to viperids concentrated upon these long teeth. Thus, is an hypothesized "protoviper" presumed a few long, projection teeth not only sink very heterodon- or senodon- like in jaw deeper but also concentrate forces to aid anatomy (Weaver, 1965). Some evidence their penetration. As a consequence, the in support of this phylogenetic relation- elongated projecting posterior maxillan ship comes from serological work (Minton teeth gain secure purchase on the prey. and Salanitro, 1972), but most of this pro- The specific kinds of special mechanical posed Heterodon-ayenodon to viperid af- problems favoring tooth enlargement are finity is based on the viper-like construc- not known. Perhaps the prey may be tion of the maxilla and associated features moist (e.g., amphibians), smooth skinned (Anthony, 1955; Mosauer, 1935; Weaver, (e.g., skinks), or inflated (e.g., toads) and 1965). However, many structures of Xe- thus offer only a slippery or uncertain sur- nodon and Heterodon tire !ikely special face to an even row of teeth. In non-con- adaptations for swallowing, not venom in- stricting snakes, the prey may continue to jection, and upon closer look do not re- struggle even as swallowing proceeds. semble viperid characters. This would also place a premium upon First, the posterior maxillary teeth are dental adaptations effective in retaining a dissimilar. The posterior maxillary teeth firm purchase on a thrashing victim. Fur- of Xenodon-Heterodon and fangs of vi- ther analyses of feeding mechanics and perids though lengthened are structurally prey type may in time eventually help to different and serve quite different func- identify the specific mechanical problems. tions. In viperids, the fangs function pri- Whatever these may prove to be, it is ap- marily during the strike to deliver venom parent, at least in Crotaphopeltis, that the and play only a very modest (Kardong, large posterior teeth are swung into a po- 1977) or no (Dullemeijer and Povel, 1972) sition so that they gain a firm purchase role during swallowing. On the other upon the prey and do transmit a force to hand, the posterior teeth are used not to the tissues during retraction. The long, deflate puffed-up toads as was once projecting posterior teeth function like thought (Pope, 1947; Morris, 1944; Min- spikes to give the snake a better grasp ton, 1944; Smith and White, 1955; Mc- upon the prey during swallowing. Alister, 1963), but to manipulate prey during swallowing in Heterodon (Kroll, 1976) and in Xenodon (personal obser- General.-The rotation of the maxilla vation). and geometric position of its rear teeth Second, Xenodon and Heterodon have would emphasize the role of posterior anatomical features of the skull that are teeth in swallowing; as I have argued quite different from those of viperids and above, tooth lengthening would be fa- even different from many other colubrids vored if the prey presents special mechan- (Figs. 4, 5). The curving forward of the ical problems during swallowing. If this ventral end of the M. adductor mandib- line of reasoning is sound, then the initial ulae externus supe&ialis jaw muscle near enlargement of posterior maxillary teeth its insertion has been claimed to be a represents an adaptation not to venom in- primitive condition in vipers (Weaver, jection, but to the requirements of swal- 19561. However. this character is not as lowing. This hypothesis has special impli- useful for assess'ment of phylogenetic af- cations for the interpretation of viperid finities as was earlier thought, because snake evolution. some very nonviper-like colubirds have Protovipers.--A close phylogenetic re- been found to have a similar direction for lationship has been claimed between cer- this muscle (Varkey, 1973; Cundall, 1974). tain colubrid genera (Heterodon and Xe- Though some variations do exist among 438 KENNETH V. KARDONG THE EVOLCTIOS OF SSXKE FASGS 439

colubrids and viperids (Varkey, 1973; However, the presence of toxicity and Kardong, 19731, the M. cervico-mandib- even a separate gland (Duvernoy's) pre- ularis is generally a single muscle that in- sumably for its secretion have been re- serts on the retroarticular process of the ported in a variety of colubrid species other mandible andlor joint capsule formed be- than Heterodon and Xenodon (Taub, tween quadrate and mandible (Haas, 1967). 1973). In .Yenodon, however, the M. cer- The central role of enlarged maxillary vico-mandibularis is divided, the posterior teeth and maxillary rotation in swallowing part inserts on the retroarticular process and the differences in tooth structure all and nearby joint capsule, but the anterior argue that Heterodon-Xenodon and vi- part inserts extensively on the quadrato- perids are not closely related. The dissim- maxillary ligament. In addition, dissimi- ilarities of muscle anatomy not only dis- larities occur in the quadrato-maxillary tinguish Heterodon-Xenodon from ligament itself. In viperids and most col- viperids but also from other colubrids. ubrids, this ligament attaches posteriorly These genera, both specialists on toads, on or near the retroarticular process of the then represent an evolutionary pathway mandible. However, in X. mewemii it at- not toward vipers, but towards a separate taches not to the mandible but to M. cer- line of colubrids adapted for grasping and vico-mandibularis and in Heterodon, the manipulating a unique prey item. Stated ligament is absent (Fig. 5). Further, in another wav.. , Xenodon and Heterodon both Xenodon and Heterodon the M. are unlikely "protovipers" in the sense of pterygoideus inserts over a much more being evolutionar?; intermediates between extensive area of the posterior mandible colubrids and vi~ers. than is common in other colubrids or Parallel evolution.-Selection pressure viperid snakes. relating to their role in swallowing would The distinctiveness of the liaamentum- favor an initial enlargement of posterior quadrato-mandibularis, .M. cervico-man- maxillary teeth. Such a state may then be dibularis, and M. pterygoideus may be the condition ancestral to at least two sub- especially noteworthy. It is these struc- sequent independent lines of evolution tures that attach in such a way as to in- that both involve further enlargement of fluence directly the mobility, rotation, and rear maxillary teeth. In one line, contin- deployment of the maxilla and its teeth. ued but increased involvement in the The distinctiveness of these structures swallowing mechanics of prey manipula- may reflect the specialized role they play tion would favor further enlargement per- in controlling the maxilla. Whether spe- haps with accompanying modification of cializations or not, however, they do rep- the jaw musculature. The result would be resent unique anatomical features of Xe- a snake such as Heterodon or Xenodon nodon and Heterodon compared to other and account for the presence of long max- colubrids and compared to viperids. illary teeth with specialized features of Finally, the presence of toxicity in Het- jaw musculature. In the second line, in- evodon (Bragg, 1960) has been presented volvement in the introduction of oral as evidence of anticipation of the viperid gland secretion could favor further tooth venom system in Heterodon (Kroll, 1976). enlargement and modification. Introduc-

FIG.1. Fang of a rattlesnake, Crotalus zliridis (.I)compared to the enlarged posterior maxillay tooth of Heterodon simus (B). Sote that this tooth in Heterodon is blade-like, more-or-less straight, and has a long posterior knife-like edge seen to better advantage in the stereopair tC) from a posterior aspect looking directly at the pointed tip. The fang, by contrast, is round, curved, and does not possess a long edge down one side. 440 KESNETH V. KARDONG THE EVOLUTIOS OF SNAKE FASCS 44 1 tion of any oral gland secretion that tran- this argument, the oral gland secretion quilized the prey would facilitate prey may also initially function to promote capture and thus be of selective advan- digestion as seems to be true in rattle- tage. .?\s a snake worked its jaws over a snakes (Thomas and Pough, 1977). Like struggling prey, elongated teeth would the initial selective forces favoring tooth be plunged and re-plunged into the tis- enlargement, the selective forces acting sues. This would open channels for en- upon the oral secretions may not have trance of oral gland secretion and even in- been related to their toxic properties but troduce any secretion flowing over or to their more immediate role in tooth adhering to the dental surfaces. An en- health, swallowing, or digestion. Xny as- zyme or chemical of the oral gland secre- sessment of oral gland function or evolu- tion that tranquilized or hastened prey tion should look not only to its toxicity but death would be of selective value by re- also to possible other rcles and properties. ducing the snake's efforts and risks in sub- duing the victim. In fact, any component of the oral gland secretion that enhanced Changes leading to the evolution of the envenomation would be favored once it viperid fang involved lengthening of rear could be introduced deeply enough to maxillay teeth and have usually been at- reach and be circulated by the vascular tributed to the adaptive value derived by system of the prey (Gans and Elliott, providing for more effective venom injec- 1968). Though any tooth in the mouth that tion. .As an alternative interpretation, it is breaks the prey's skin could conceivably argued here that an initial factor favoring carry oral gland secretions inward, the elongation was not venom injection but posterior maxillary teeth are the best suit- instead increased swallowing effective- ed for this because they are enlarged and ness. During swallowing, rear maxillary thus would penetrate deeper than the teeth are the first teeth of the maxilla to shorter anterior teeth. engage the prey upon each swallowing In both lines of evolution, enlargement cycle. Further, they are geometrically po- of the posterior maxillary teeth would re- sitioned more favorably than other max- sult and superficially appear to be an ex- illary teeth so as to swing through a longer ample of parallel evolution, but the func- retraction arc and hence move the prey tions (prey manipulation vs. injection) and farther. Consequently, year maxillary hence selective pressures underlying these teeth might be expected to be the ones trends would be quite different. most likely modified in any dental adap- Venom evolution.-If, as argued above, tations that served swallowing. Elonga- the selective pressure favoring initial tooth tion of a few teeth permits deeper pene- enlargement centered upon the use in tration into the tissues and hence increases swallowing and not injection, then the purchase of the jaws on the prey. These likely initial function of oral secretions factors, it is argued, all favored the initial was also different from that of killing elongation of posterior maxillary teeth. prey. Gans (1978) has suggested that one However, once elongated to enhance initial function for oral gland secretions swallowing effectiveness, they became a was that of conditioning teeth and pro- key feature preadapted for subsequent moting oral hygiene. To speculate with modification along several evolutionary

FIG.5. Lateral jaw muscles as seen in three colubrid snakes (first three) and viperid snake (bottom). .I) Heterodon simus B) .Yenodon rabdocephalus C) Xenodon merremii D) .-lgkistrodon halys. .\bbreviations as. .M. adductm mandibulae externus supelficialis: ap. .M. adductor mandibulae externus profundus; cg., .M. com- pressor glandulae: cm. .M. cen~icomandibularis;lqm, Ligamentum quadrato-mandibularis; vg, venom gland. 442 KENSETH V. KARDOYG pathways. One direction led toward the lier versions of this manuscript as well as Xenodon and Heterodon-like snakes in comments by C. Gans, H. W. Greene, which the manipulation function of the M. K. Hecht, J. C. Kroll, and R. C. elongated teeth during swallowing was Snyder. further enhanced. These snakes thus rep- resent not "protovipers" but simply a spe- cialized form of snake in which the teeth .\LBRIGHT. R. G.. AND E. M. NELSON. 1959a. aid prey manipulation during swallowing. Cranial kinetics of the generalized colubrid snake Elaphe obsoleta quadrivittata. I. Descriptive A second line of evolution led to viperid morphology. J. Morph. 105:393-239. snakes in which the emphasis shifts so that .~LBRIGHT,R. G.. ASD E. M. ELS SON. 19596. the elongated teeth now play a greater role Cranial kinetics of the generalized colubrid snake in venom injection. Although the super- Elaphe obsoleta quadrioittata. 11. Functional morpholom. J. Morph. 105:241-291. ficial impression would be that of two par- .\NTHOKY,J. 1955. Essai sur I'Evolution anato- allel lines of evolution involving rear max- mique de I'appareil venimeus des Ophidiens. illary tooth elongation, the selective forces .innales des Sciences Saturelles. Zoolopie acting in each line would be quite differ- 17:;-53. ent. In one, the adaptations would serve BRAGG,.4. ?;. 1960. Is Heterodon venomous? Her- petologica 16(2):12 1-123. to improve prey manipulation during CVNDALL,D. L. 1974. The cranial osteolom and swallowing while in the other to improve myology of the green snakes, genus Opheodrys. envenomation. Ph.D. dissertation, Uni\.ersity of Arkansas. Cni- Just as the mechanics of swallowing versity microfilm i4-28,078. DCLLEMEIJER,P., AND G. D. E. POVEL.1972. The could have been the initial factor favoring construction for feeding in rattlesnakes. Zoolo- tooth elongation, one might expect that gische mededelingen 47 :561-578. oral gland secretions also served initially GASS, C. 1978. Reptilian venoms: Some evolution- not a toxic function but as an aid in prey anconsiderations. In C. Gans and K. .;\. Gans handling or swallowing. ieds.), Biology of the Reptilia. Academic Press, London. In press. GASS. C.. AND LV. B. ELLIOTT. 1968. Snake ven- oms: production, injection, action. -4dvances in For assistance with radiology, my Oral Biol. 3:45-81. thanks are extended to R. D. Sande of the GOELLNER,R. 19;s. Garter snakes as toxic as rat- tler. St. Louis Herpetol. Soc. Newsl. 2(12):8-9. College of Veterinary Medicine at Wash- HA~s,G. 19i3. Muscles of the jaws and associated ington State University and to Howard structures in the rhynchocephalia and squamata. Ricketts with Radiology Research at the In C. Gans and T. S. Parsons (eds.), Biology of University of Washington. The scanning the Reptilia 4:285-490. electron microscopy was performed with HEATWOLE,H., AND I. B. BANUCHI.1966. En- venomation by the colubrid snake .Ilsophis par- the kind permission and assistance of Ar- toricensis. Herpetologica 22:132-134. thur L. Cohen, Director, Electron Micro- JOHASBOCKE,M. .M. 19i4. Effects of a bite from scope Center, Washington State Univer- Conophis lineatus (Squamata: Colubridae). Bull. sity. For the assistance in obtaining live Philadelphia Herp. Soc. 22:39. KARDONG,K. V. 1973. Lateral jaw and throat mus- specimens, I am happy to acknowledge culature of the cottonmouth snake Agkistrodon the help of Professor Paulo de Toledo Ar- piscivonts. Gegenbaurs Morph. Jahrb. 119:316- tigas, Universidade de S5o Paulo and 335. John Visser, Wildlife Documentaries, . 19i7. Kinesis of the jaw apparatus during South Africa. For permission to inspect swallowing in the cottonmouth snake, .4gkistro- don piscivonts. Copeia 19i'i:338-348. and/or dissect preserved specimens, thanks KROLL,J. C. 1976. Feeding adaptations of hognose are due to Walter Auffenberg of the Flor- snakes. Southwestern Natur. 20:537-557. ida State Museum, D. G. Broadley of the -mRX, H., AND G. B. ~BB.1972. Phyletic anal- Umtali Museum, and Hymen Marx of the ysis of fifty characters of advanced snakes. Field- iana, Zoology 63:1-32 1. Field Museum of Natural History. MASLIN,T. P. 1952. Morphological criteria of phy- I especially appreciate the thoughtful letic relationships. Syst. Zool. 1:49-70. comments of John H. Larsen, Jr. on ear- MCALISTER,\V. H. 1963. Evidence of mild toxicity THE EI-OLUTIOS OF SSAAE F.iSGS 443

in the saliva of the hognose snake (Heterodon). T.AL-B..A. 11. 196;. Comparative histological stud- Herpetologica 19: 132-137. ies on Duvernoy'.; gland of colubrid snakes. Bull. ~IISTOS.S. .A. 1944. Introduction to the stud>-of .Amer. Mus. Sat. Hist. 138:l-50. reptiles of Indiana. .Amer. Jlidl. Satur. 32:138- T~o>r;\s.R. G.. ASD F. H. POLGH. 1977. The role 477. of rattlesnake \-enom in prey dizestion. .Amer. >frsTos. S. A,. ASD J. A. SAL.~SITRO.1972. Sero- Zool. 17:947 [abstract). logical relationships among some colubrid snakes. I..ARKEY..A. 1973. Comparative cranial myology of Copeia 19;2:246-252. Sorth American natricine snakes. Ph.D. disser- MORRIS.P. .A. 1944. The>-hop and crawl. Jacques tation, Louisiana State University and .Agricul- Cattell Press. Lancater. 253 p. tural and Mechanical College. University micro- MOSACER.K. 1935. The myology of the trunk re- films. 73-27, 877. gion of snakes and its significance for ophidian \YE.AVER.I!-. G.. JR. 1965. The cranial anatomy of taxonomy and phylogeny. Publ. Univ. Calif. Los the hognosed snakes (Heterodonl. Bull. Florida .Angeles Biol. Sci. 1:81-120. State Mus. 9:275-304. POPE, C. H. 194;. Amphibians and reptiles of the it-ILL.ARD,D. E. 1967. Evidence for toxic saliva in Chicago area. Chicago Mus. Sat. Hist. 275 p. Rhadinea j'?ai~ilata (the yellow lipped snake). SMITH.H. 121.. .ASD F. S. I\-HITE. 1955. .idrenal Herpetologica 23:238. enlargement and ib significance in the hognose snakes (Heterodonl. Herpetologica 11:137-144.