SMITHSONIAN MISCELLANEOUS COLLECTIONSVOLUME 99, NUMBER 20
FURTHER NOTES ON MEXICAN SNAKESOF THE GENUS SALVADORA
BYHOBART M. SMITH
(Publication 3630)
CITY OF WASHINGTONPUBLISHED BY THE SMITHSONIAN INSTITUTIONFEBRUARY 21, 1941

SMITHSONIAN MISCELLANEOUS COLLECTIONSVOLUME 99. NUMBER 20
FURTHER NOTES ON MEXICAN SNAKESOF THE GENUS SALVADOR
A
BYHOBART M. SMITH
'4WW(Publication 3630)
CITY OF WASHINGTONPUBLISHED BY THE SMITHSONIAN INSTITUTIONFEBRUARY 31, 1941
BALTIMORE, MD., U. 8. A,
FURTHER NOTES ON MEXICAN SNAKES OFTHE GENUS SALVADORABy HOBART M. SMITHA review of the Mexican Salvadnra in the National Museum,originally intended to fill out data on hairdil, has resulted in the dis-covery of two undescrihed forms, one of them particularly interest-ing as illustrating an entirely new direction of evolution in the(jrahauiiac group of the genus. As might be ex})ected, this novelspecies originates in an isolated area not previously known to harborany member of the group.In addition, a third undescrihed form, closely related to anotherisolated species recently described, was discovered by Dr. E. H.Taylor and Richard Taylor in an area so remote from that occupiedby its close relative that its existence there is most remarkable.Through the courtesy of Dr. Taylor this form also is described herein.The systematics of Salvadora is a highly interesting study, largelybecause of the unusual multiplicity of characters. This multiplicityat first led to confusion, as for example, in the era in the 1920's whenbut "Jiexalepis" and "grahaniiac" were distinguished in the check-list area. Careful study of much larger collections in recent years,however, has shown the existence of a surprising number of trulyrecognizable forms.The present review was undertaken and largely completed beforeit was known that the genus was in the process of review by theprincipal student of the group. Mr. Bogert most generously coop-erated with me during the completion of the study, however, furnish-ing data on a number of specimens, and moreover giving me thebenefit of certain conclusions reached by him with many more datathan I have accumulated. In all cases, however, I have adhered to myoriginal conclusions with regard to the status of the various Mexicanforms, although I am aware that Mr. Bogert will have corrections andrevisions to make. For this most generous aid and cooperation I ammost grateful.I am also indebted to Dr. E. H. Taylor for loan of specimens andhelp with data; to Dr. Norman E. Hartweg and K. P. Schmidt forhelp with data and in formulating ideas ; to Dr. Howard K. Gloyd forSmithsonian Miscellaneous Collections, Vol. 99, No. 20
2 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 99loan of specimens; and finally to William Stickel, who has criticizedthe work, made various helpful suggestions, and contributed to theconstruction of the key. The drawings have been prepared by mywife.The entire study has been completed during tenure of, and withgreatly appreciated assistance from, the Walter Rathbone BaconScholarship. SALVADORA BOGERTI, new speciesFigures i, 2Type.—U.S.N. TNI. No. 30296, female, from "Tehuantepec," col-lected by Francois Sumichrast.Diagnosis.—Rostral slightly enlarged, edges somewhat free ; dor-solateral dark stripes terminating on nape, not divergent, not cross-ing temporal region ; lateral dark stripe not fused anteriorly withdorsolateral stripe ; 9 supralabials ; 2 loreals ; anterior section of nasalseparated from second supralabial ; frontal in contact with, or nar-rowly separated from, preocular ; posterior chin shields rather widelyseparated ; ventral and caudal count low.Description of type.—Rostral somewhat enlarged, with slightlyfree edges ; seen from the front, the sides of the rostral diverge alittle ; anterior edge of rostral, seen from above, nearly straight ; maxi-mum width of rostral slightly greater than its distance from frontal
;
length of frontal subequal to length of parietals; distance of frontalfrom tip of snout three-fourths its length ; frontal narrowly in con-tact with preocular on one side, narrowly separated on other ; 9supralabials, the fifth and sixth entering orbit ; anterior section ofnasal separated from second supralabial ; 2 loreals, the lower muchthe smallest and wedged between preoculars, supralabials and upperloreal ; 2 preoculars ; 2 postoculars ; temporals 2 -1- 2 + 3 ; seventh supra-labial in contact with postoculars; lO-ii infralabials ; posterior chinshields much smaller than anterior, separated from each other by arelatively large, elongate scale.Ventrals 188; subcaudals 87, tail complete; scale rows 17-17-13;the third scale row dropped at the it 7th ventral on one side, at the1 1 6th on the other; the paravertebral rows are dropped at the 141stventral.Maxillary teeth 11 +3.Total length, 463 mm. ; tail length, 115 mm.Middorsal light stripe (cream in color) one and two half-scale rowswide at nape, one scale row wide at posterior end of body, its edgesrather sharply defined ; dorsolateral dark stripes slate gray, covering
NO. 20 MEXICAN SNAKES OF THE GENUS SALVADORA S^^IT1I 3two and two half -scale rows anteriorly, two and one-half rows pos-teriorly ; these stripes terminate anteriorly at the nape, and do notdiverge nor pass through the temporal region ; most of the scales inthe dorsolateral dark stripe have the anterolateral edge light colored(usually concealed), while the anteromedial and posterolateral edgesare black ; only the posteromedial edge is gray as the rest of the scale ;a narrow lateral dark stripe anteriorly occupies the third scale rowand the extreme outer edges of the fourth row ; posteriorly (poste-
ViG. I.—Cephalic scutellation ofSalvadora bogerti, from type. Fig. 2.—Pattern of head andneck of Salvadora bogerti, fromtype. Scale pattern hypcithetical,based on that of tincata.
rior to point at which the third scale row drops) the lateral stripe ison the second and third rows ; the lateral stripe disappears anteriorlyat the nape, and does not merge wnth nor approach the dorsolateraldark stripes ; posteriorly the lateral stripe disappears at the anus.Ventral surfaces immaculate.Comparisons.—The caudal count, number of maxillary teeth, andseveral characters of the cephalic scutellation link this species withwhat might be termed the grahamiae group, and exclude it from thegroup or groups formed by the species mexicana, Icmn'iscata, and
4 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 99piilchcrriuM. However, the peculiar combination of characters pos-sessed by it make dubious its closest relationship within the grahainiaegroup.The locality data borne by the specimen (Tehuantepec) are not pre-cise, but it may be assumed that the specimen was taken in somemountain range west of the Isthmus of Tehuantepec and perhaps inthe area northwest of the city of Tehuantepec. It can be stated withsome degree of assurance that the semiarid lowlands about the cityof Tehuantepec do not harbor this species, else it probably would haveappeared with the recent intensive collecting in that area.Assuming this, it is apparent that the specimen comes from anarea from which no others of the group are known. Furthermore,the mountain ranges of this area form a relatively compact groupisolated faunistically and physiographically from the nearest otherareas represented by specimens of the group.Sahadora bogerti is comparable to hairdii, known from centralMexico as far south as southern Puebla, in two important characters
:
color pattern and the separation of the anterior section of the nasalfrom the second supralabial. The two most noteworthy similar fea-tures of the color pattern are: first, that the median light stripenarrows posteriorly to a width of one scale row; and second, thatthe dorsolateral dark stripes do not diverge anteriorly nor passthrough the temporal region, but terminate on the nape. These twocharacters, held in common with hairdii, differentiate it from allother members of the group. The species is easily distinguishablefrom hairdii by the possession of 9 supralabials, higher number ofmaxillary teeth (9 + 3 normally in hairdii), enlarged rostral, 2 loreals,and probably by a lower average ventral and caudal count (the lowestcounts in hairdii females are exactly the counts of hogerti).The other Mexican species, intcruiedia, with its subspecies richardi,most closely situated geographically, differs widely from hogerti incolor pattern, having the lateral stripe fused with the dorsolateralon anterior third of body, dorsolateral stripes passing through eye, alower number of supralabials, single loreal, higher number of sub-caudals, and separation of the antepenultimate supralabial from thepostoculars.From the geographically distant liiicata, hcxalcpis, grahainiae, andtheir subspecies, hogerti dift'ers markedly. S. h. hcxalepis, h. ceteris,and h. virgidtea are the only forms of the genus normally with twoloreals, and they moreover have the posterior chin shields widely sep-arated as in hogerti, but from these hogerti may be distinguished incolor pattern, the nasal-second labial character, smaller rostral, much
NO. 20 MEXICAN SNAKES OF THE GENUS SALVAnORA—SMITH 5lower ventral and caudal count, etc. -b\ Ji. dcscrticola is perhaps moreeasily comparable to bogcrti than any other form, as the scutellationof one is within the range of variation of the other, except for thenasal-second supralabial character and form of rostral, but the color-ation is very different.
Fig. 3.—Pattern of head aiulneck of Salvadora bairdii, fioniU.S.N.M. No. 56576, J.ilisco.Scale pattern hypothetical, basedon that of lincata. Note greatsimilarity of pattern in bairdiiand bogerti, both of which haveirregular black areas in thedark stripes.
Fir>. 4.— Pattern of head andneck of Salvadora lineaia. fromU.S.N.M. No. 105304, 17 mileswest of Santa Caterina, NuevoLeon. Texas specimens showa greater divergence of thedorsolateral dark stripes; thecondition as shown is more orless characteristic of Mexicanspecimens.S. grahainiac and lincata are so completely different from bogertithat comparisons are unnecessary.Finally, the present specimen exhibits one character which, if foundconstant in the species, occurs in no other of the group. That is thecontact or near contact of the frontal with the preocular. So uni-versally are these scales separated in members of the grahainiac groupthat I would consider the condition in the type of bogerti anomalousif it were not for the fact that the same condition occurs frequently inIcmuiscata and nicxicana.
6 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 99Within the grahamiae group two smaller groups are apparent—onecontaining intermedia and hexalepis (with their subspecies), theOther containing bairdii, grahamiae, and lineata. Group I is charac-terized by normal possession by its members of 1 1 + 3 teeth and a pat-tern which involves fusion of the lateral stripe with the dorsolateralon the anterior part of the body. Some interesting gradations in ven-tral and caudal counts are apparent in this group. In group II theteeth are normally 9+ 3 or 10+ 3, and the lateral stripe, when pres-ent, remains distinct from the dorsolateral throughout its length,S. bairdii is by far the most distinct of group II and it is certainlythe most primitive of either group, but its relationship with this grouprather than with group I is apparent.In determining the relationship of bogcrti to these two groups,therefore, it is necessary to choose between emphasis on number ofteeth, or on pattern, for the one links it with group I, the other withgroup II. It has been demonstrated in many groups of animals thatcertain details of pattern frequently are more stable than anatomicalcharacters. I believe that this is another example of that phenomenon,since to me the pattern and geographical position of bogcrti seem moresignificant and to link that species more definitely with bairdii thanthe considera1)le difference in number of teeth would imply.SALVADORA INTERMEDIA RICHARDI, new subspeciesFigure 5Type.—E. H. Taylor-H. M. Smith collection No. 23470, male,collected i mile north of Tehuacan, Puebla, August 8, 1940, byRichard Clark Taylor.Diagnosis.—Rostral somewhat enlarged, with edges slightly free,its anterior border (viewed from above) nearly straight ; 8 suprala-bials ; antepenultimate supralabial separated from postoculars ; maxil-lary teeth 1 1 + 3 ; lateral stripe fused with dorsolateral stripe, onanterior third of body, poorly defined ; anterior section of nasal sepa-rated from second supralabial ; ventrals 189 in male type.Description.—Rostral somewhat enlarged, with slightly free edges
;
seen from the front, the sides of the rostral are nearly straight ; an-terior edge of rostral, seen from above, nearly straight ; maximumwidth of rostral slightly greater than its distance from frontal ; lengthof frontal subequal to length of parietals; distance of frontal fromtip of snout nearly equal to length of frontal ; 8 supralabials, fourthand fifth entering orbit ; anterior section of nasal separated fromsecond supralabial; 2 loreals on one side (the lower small and flat).
NO. 20 >rKXICAN SNAKES OF THE GENUS SALVADORA SMITH 7
1 on the other ; second supralabial broadly in contact with loreals
;
2 preoculars on one side, 3 on other (the extra scale is split fromthe lower part of the upper preocular) ; antepenultimate supralabialseparated from postoculars by a small scale ; 2 postoculars ; temporals2 + 3 (2+ 2); 10 infralabials ; posterior chin shields much smallerthan anterior, separated from each other anteriorly by a small scale
I'lG. 5.—Pattern of head anc'neck of Salvadora intermediarichardi, from tyije. This pat-tern is duplicated in i. inter-media. Scale pattern hypotheti-cal, based on that of lineata.Note similarity between thisand lineata in character of pig-mentation; both of these showthe peculiar individual scalepattern on posterior part ofbody consisting of a white an-terolateral and black antero-medial edge.
and a portion of end of an anterior chin shield, posteriorly by 2 smallscales.Ventrals 189; subcaudals indeterminate, 82 without tip; scale rows1 7-1 7-1 3 ; third row dropped at the 1 15th ventral ( 1 14th on one side) ;paravertebral rows dropped at 122nd ventral.Maxillary teeth 11 +3.
8 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 99
Total length, 86 1 mm. ; snout to vent, 657 mm.Middorsal light stripe three scales wide at neck, gradually becomingnarrower posteriorly, at posterior end of body one and two half-scale rows wide (the half-scale rows are dusky) ; dorsolateral darkstripes fused with lateral stripes anteriorly, the dark color extendingto middle of third scale row ; lateral stripe begins to show very dimlyat about the 40th ventral, but only at the point at which the third rowdrops does it become clearly defined ; posterior to this point thelateral row follows the third scale row to near the anus, where itdisappears ; the considerable lateral extension of the dorsolateralstripes contributes to the indistinctness of the lateral stripe ; darkstripes diverging anteriorly and passing over temporal region to eye
;
ground color (slate gray, with a bluish tinge) extending onto ends ofventral surfaces ; head somewhat brownish.Anterior portion of belly white ; edges of ventrals becoming pinkin middle portion of belly ; posterior third of belly light salmon pink ;tail pink, becoming white toward tip (in preserved specimen).Anterolateral edges of scales in dorsolateral stripes white, Ih.eiranteromedial edges black (concealed).Comparisons.—This form is very close to intermedia, agreeingwith that in all pertinent details of color and pattern (including thepink belly), in form of rostral, number of labials and number ofmaxillary teeth. It differs from interuicdia primarily in ventralcount, the known range of variation in male intermedia being 175to 181. It is impossible to state whether there are differences in caudalcoimts.In the cephalic scutellation, it is most notable that the anteriorsection of the nasal is separated from the second supralabial inrichardi, in contact in intermedia; secondarily it may be pointedout that the loreals are 1-2 and the preoculars 2-3.These characters are very meager, but in their totality imply asignificance; one or two such differences would not be notable; butthe union in this specimen of so many, though minor, differences,coupled with the fact that the specimen comes from an area remotefrom that inhabited by intermedia, should be significant. Only futurecollecting will show whether the characters are constant.The present specimen brings the range of intermedia as a wholevery close to the known range of bairdii, which has been taken just12 miles north of Tehuacan. The ranges of the two possibly overlapslightly in the area north of Tehuacan. However, it is very probablethat, in general, bairdii does not extend south of Tehuacan, norrichardi very far north of Tehuacan, because this city coincidentally
NO. 20 MEXICAN SNAKES OF THE GENUS SALVADORA SMITH 9
is on the edge of a low, arid region which extends southward intothe dry valleys of the Balsas Basin. The area is rugged and very dry,whereas the extensive plains north of Tehuacan are flat and becomeincreasingly moist toward the north as elevation of the plains in-creases toward Puebla. It is noteworthy that, so far as is known,hairdil does not occur in extremely arid regions in any other partof its range. Furthermore, /'. interuicdia is confined, as far as known,to a mountainous habitat, similar in rainfall to that of hairdii, butisolated on all sides by extremely arid or tropical zones.The subspecies richardi links hairdii and intermedia a little moreclosely, structurally as well as geographically. The presence in richardiof a higher number of ventrals, and the separation of the secondsupralabial and the anterior section of the nasal, are two steps towardthe conditions exhibited by hairdii, which I consider the most primi-tive of the entire grahainiae group. It is rather obviously specializa-tion in i. intermedia which produces low ventral count (lowest inthe entire grahamiae group) and at the same time the naso-labialcontact. The latter specialization is carried throughout all forms ofgroup I, with the exception of richardi.SALVADORA HEXALEPIS CELERIS, new subspeciesType.—U.S.N.M. No. 40043, female, from San Bias, Sinaloa.Collected by J. N. Rose, March 28, 1910.Diagnosis.—Rostral much enlarged ; lateral dark stripe fused withdorsolateral at middle of neck ; 9 supralabials ; normally i or 2 loreals
;
ventrals 200 and 205 in known specimens ; no trace of cross-barredefl^ect in pattern. Maxillary teeth 11+3.Description of type.—Rostral much enlarged, with very free edges,its anterior margin (seen from above) nearly straight; anterior sec-tion of nasal broadly in contact with second supralabial; latter sepa-rated from loreal ; 2 loreals on one side, i on other ; 2 preoculars2 postoculars; 9-10 supralabials, the antepenultimate separated frompostoculars ; temporals irregular, 2-I-2 + 3.Infralabials ii-ii; maximum length of anterior chin shields sub-equal to that of posterior chin shields ; latter scales broadly separatedon median line by 2 scales anteriorly, 3 posteriorly.Dorsals in 17- 17- 13 rows, the third row dropped at the 131stventral (129th on other side), the paravertebral dropped at the 158thventral (162nd on other side) ; ventrals 200; subcaudals yy.Maxillary teeth II -1-3.
lO SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 99Median light stripe three scale rows wide over most of bodv. oneand two half -scale rows wide posteriorly ; dorsolateral dark stripecovering two scale rows over most of body, only one posteriorly, itsedges serrate and corresponding rather exactly to the scales in therows it follows ; lateral dark stripe involving mainly the fourth row,
Fig. 6.—Pattern of head andneck of Sah'adora hexalepisdeserticola, from U.S.N.M. No.22201, Ft. Huachuca, Arizona.This pattern is duplicated inh. ccleris. Scale pattern hypo-thetical, based on that of lineata.Posteriorly the lateral line dropsto the third row.
Fig. 7.—Pattern of Salvadarah. hexalepis, from U.S.N.M.No. 17500, Tucson, Arizona.Scale pattern hypothetical,based on that of lineata. LTp-per, head and neck; lower,middle of body. Note simi-larity in individual scale patternbetween this and deserticola,both of which have the edgesof the scales white, (not soin (irahamiae, lineata, and inter-media); these two (all formsof hexalepis) lack the whiteanterolateral and black antero-niedial corners characteristic of(he other contrasted forms.but also a small part of the third, on anterior part of body, descendingposteriorly to the third row after loss of the third scale row anteriorly
;
dorsolateral stripe diverging anteriorly and passing through eyelateral stripe fusing with dorsolateral stripe at middle of neck.Ventral surfaces immaculate.
NO. 20 MEXICAN SNAKES OF THE CENUS SALVADORA SMITH IIComparisons.—^This subspecies ' is most closely related to descrfi-cola. In coloration the two cannot be distinguished. Their only dif-ference lies in ventral count, desert icola varying between 187 and 192ill 9 females, and from 184 to 195 in 11 males. Since these femalecounts average somewhat lower than males, the fact that the San Biasfemale has 200 is of special significance.The only difference from typical licxalepis which can be stated defi-nitely at present is in color pattern. In hcxalcpis the pattern is typi-cally barred, i.e., there are at least alternating darker and lighter areasin the dorsolateral and lateral dark stripes. Usually the coincidingdarker patches of the two stripes on a side are more or less fusedinto a larger, dim, dark spot or band. There is no indication of thesecross bars in either deserticola or celeris.The range of celeris presumably includes extreme southern Sonoraas well as northern Sinaloa. The area is closely approached bydeserticola, a specimen of which I have seen from Batopilas. Chi-huahua. Mr, Bogert has kindly informed me of another specimen,also in the R. T. Moore collection (No. 17449)' which approachesstill more closely the known range of celeris. This is from Yecorato.Sinaloa, which locality is probably in the same mountainous type ofcountry as Batopilas ; San Bias and Ahome are on a flat coastal plain.Intergradation of celeris and deserticola very likely occurs a short dis-tance west of Yecorato, and very possibly celeris and hexalepis ashort distance south of Guaymas, where numerous forms of mammalsand lizards are known to intergrade or terminate their ranges. Theline south of Guaymas is very well defined, and marks the boundaryof the Arizonian and Sinaioan biotic provinces ; the line to the westmarks the boundary of the Durangan province, through the northernpart of which the range of deserticola extends.KEY TO MAINLAND MEXICAN SALVADORA
1. Lateral dark stripe involving first row of dorsal scales 2Lateral dark stripe not involving first row of dorsal scales on any partof body 32. Stripes continued to occiput; ventrals immaculate, 196 to 208 innumber IcmniscataStripes interrupted anteriorly, cross bars on neck ; spots on outer mar-gins of anterior ventrals; ventrals 182 to 192 in number me.vicana
^ Probably identical with this is a specimen from Ahome, Sinaloa, in the'collection of R. T. Moore at the California Institute of Technology (No.17943)- C. M. Bogert has kindly given me data on this specimen, which has205 ventrals, 2 loreals, 9 supralabials (fifth and sixth entering eye).2 Bogert informs me that this specimen has 190 ventrals and i canthal.
12 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 99
3. Only one pair of dark stripes on bod}', these not involving third row ofscales on any part of body 4Stripes involving third row of scales (figs. 2, 3, 4, 5, 7) 54. Antepenultimate supralabial separated from postoculars ; maxillary teeth1 1+3; stripes involving third and fourth rows of scales, bifurcatedposteriorly (dorsolateral and lateral), but not well defined 5Antepenultimate supralabial usually in contact with postoculars ; maxil-lary teeth usually 10+3; only dorsolateral stripes present, well defined,not involving either the third or the fourth scale rows anteriorly...g. grahamiac5. Dorsolateral dark stripes terminating on nape and not passing throughtemporal region (figs. 2, 3) 6Dorsolateral dark stripes diverging on neck and passing through tem-poral region to eye (figs. 4 to 7) 76. Loreals 2; rostral with slightly free edges, anterior margin (seen fromabove) nearly straight; maxillary teeth 11+3; 9 supralabials (figs, i, 2).bogcrtiLoreal i ; edges of rostral not free, its anterior margin (seen from above)distinctly convex ; maxillary teeth usually 9-I-3 ; usually 8 supralabials(fig. 3) bairdii7. Lateral dark stripe fusing with dorsolateral dark stripes on neck(figs. 5, 6, 7) 8Lateral dark stripe distinct and separate from dorsolateral throughoutits length, generally disappearing free on neck, rarely discernible totemporal region (fig. 4) grahamiac liiicata8. Rostral strongly enlarged, with prominently free edges ; 9 supralabialsor more; antepenultimate labial generally in contact with postoculars. . 10Rostral moderately enlarged, with somewhat free edges; 8 supralabials;antepenultimate labial usually separated from postoculars 99. Anterior section of nasal in contact with second supralabial ; ventrals182 or less i. intermediaAnterior section of nasal separated from second supralabial ; ventrals190 in single specimen (fig. 5) intermedia richardi10. A distinct barred effect in pattern; lateral stripe fused with dorsolateralover all of neck (fig. 7) h. hexalepisPattern not barred, the stripes uninterrupted in intensity ; lateral stripesfused with dorsolateral over anterior half or third of neck (fig. 6) 1111. Ventrals 195 or less (fig. 6) h. deserticolaVentrals 200 to 205 (in two specimens) h. celcris