ANMERICAN MUSEUM NOVITATES Published by NUmber 1285 THE AMERICAN MUSEum OF NATURAL HIsTORY May 14, 1945 New York City

TWO ADDITIONAL RACES OF THE PATCH-NOSED SNAKE, SALVADORA HEXALEPIS BY CHARLES M. BOGERT INTRODUCTION

Some years ago an attempt was made to emphasis will be placed on taxonomy per determine the affinities and distributions of se, and the chief purpose of the present the various popuilations composing the paper is to dispose of descriptions of new genus Salvadora. The results of this survey subspecies. (Bogert, 1939a) proved to be highly inade- It must be added, pending detailed dis- quate. Subsequent emendations have ap- cussion of the clines in ventral and caudal peared in the papers by Smith (1941), by counts, that the nature of the variation in Bogert (1939b), by Schmidt (1940), and by these characters suggests that the number Hartweg (1940). As far as the species of ventral and subcaudal scutes, and hence hexalepis is concerned, the most important the number of vertebrae in hexalepis, is not paper is that of Schmidt who recognized the closely correlated with other characters. In specific distinctness of Salvadora hexalepis view of these facts, the race celeris, described and pointed out my error (Bogert 1939a) by Smith in 1941 and diagnosed as having in considering the populations of hexalepis more ventrals than deserticola, has been surrounding isolated populations of S. referred to the synonymy of the latter by grahamiae to be intergrades between the Bogert and Oliver (1945). The ventral two species. Schmidt recognized these counts in S. h. hexalepis are highest in low- "intergrades" as a subspecies of hexalepis land populations and lowest in specimens which he described as Salvadora hexalepis from the foothills. A similar situation deserticola. This interpretation is undoubt- seems to exist in deserticola, a snake inhab- edly correct, although Schmidt's concept of iting foothills and higher valleys throughout the range of deserticola was inadequate. most of its range, with the mean for ventral Stimulated by these corrections, I have counts being lower than that of other sub- undertaken a second revision of the genus. species. However, where its range extends Much additional material has become avail- onto the coastal plain in northern Sinaloa able, some of it from critical regions. In and southern Sonora, the ventral counts fact, over twice as many specimens have rise just as they do in hexalepis in the north- been tabulated, with welcome additions western corner of Sonora. from nearly all parts of the range of the On the basis of a sample composed of genus. This new material has thrown over 375 specimens assigned to the species doubt upon the locality data of a few older hexalepis, the ranges of the subspecies are specimens that led to misguided notions interpreted as mapped in figure 10, based published in 1939. It has also made it pos- upon "spot" maps to be published later. sible to re-evaluate the significance of char- In connection with the map it should be acters either neglected or erroneously as- noted that throughout the southern por- sessed in previous papers, and to outline tion of the range, snakes of this species do the ranges of species as well as subspecies not occur in mountain or plateau regions with greater completeness. above 4000 feet in elevation, whereas in In the revision now in preparation con- the Mojave Desert a few specimens have siderable attention will be devoted to been taken at elevations slightly exceeding clines or geographical gradients. Little 6000 feet. Mountain "islands" inhabited 2 AMERICAN MUSEUM NOVITATES [No. 1285 by Salvadora g. grahamiae occur within the H.T.W. The late Lieut. Harold T. Woodall ranges of both hexalepis and deserticola, and of Long Beach, California L.M.K. Dr. Laurence M. Klauber of San the range of the latter also overlaps that of Diego, California Salvadora bairdi in Chihuahua. M.V.Z. Museum of Vertebrate Zoology, Statistical analyses other than simple University of California means are not used in these preliminary de- R.H. Dr. Ross Hardy of Dixie Junior College, Utah scriptions, but such methods will be used to S.U.M. Stanford University Natural His- interpret data when all representatives of tory Museum the genus are being considered in the survey S.D.S.N.H. San Diego Society of Natural His- now under way. tory U.C.L.A. University of California, Los An- geles ABBREVIATIONS USED U.S.N.M. United States National Museum Data presented herein have been drawn U.M.M.Z. University of Michigan Museum of from material in the collection of the Ameri- Zoology can Museum as well as from specimens Z.N.P. Zion National Park generously lent by the institutions and listed below. Abbreviations To the individuals in charge of these individuals collections and to numerous others who used precede the names. have lent material not used directly in the A.N.S.P. Academy of Natural Sciences of preparation of this report sincere thanks Philadelphia Dr. M. A.M.N.H. American Museum of Natural His- are extended. L. Klauber and Dr. tory Hobart M. Smith courteously supplied B.Y.U. Brigham Young University scale counts and other data used herein C.A.S. California Academy of Sciences and to them I am especially indebted. C.U. Cornell University The drawings were prepared by Mr. Dim- G.W.H. Mr. George W. Hanley of Bakers- field, California itry Alexandroff whose careful attention to G.C.N.P. Grand Canyon National Park detail is gratefully acknowledged.

DIFFERENTIATION IN SALVADORA HEXALEPIS As now understood S. hexalepis con- Salvadora hexalepis klauberi,' new sub- tains but three races: (1) hexalepis, oc- species cupying the central portion of the range of the species, from Nevada and Utah BAJA CALIFORNIAN PATCH-NOSED SNAKE southward to Sonora and Baja California; Figures 1-4 (2) virgultea, with a limited range in TYPE: No. 20912 in the collection of L. southwestern California and northwestern M. Klauber, from Cape San Lucas, Baja Baja California; and (3) deserticola, inhab- California, Mexico. Collected by Fred iting an ill-defined region at the eastern ex- Lewis and preserved on August 6, 1933. tremity of the territory occupied by the DIAGNOSIS: A subspecies of hexalepis species. Mor-e intensive study indicates which differs from both S. h. hexalepis and that two additional populations formerly S. h. virgultea in more commonly having included with S. h. hexalepis, (1) that in two supralabials instead of one reaching the peninsular Baja California south of the eye. Furthermore it differs from virgultea San Pedro de Martir Mountains, and (2) in having the light vertebral stripe three that inhabiting the arid region loosely de- scales in width, and continued onto the top fined as the Mojave Desert, are sufficiently of the head. From S. h. deserticola it differs well differentiated from populations in in more commonly having the loreal di- contiguous areas to warrant their nomen- vided, and in having the lower lateral stripe clatorial recognition. Accordingly they are ' Named for Dr. Laurence M. Klauber whose ex- described with tensive collections and writings have contributed diagnosed and below, so much to our knowledge of patch-nosed snakes in ranges as indicated on the map (fig. 10). th. 3 )u; 1 Ar es t. 1945] RACES OF SALVADORA HEXALEPIS 3

FIG. 5

FIG.2

FIG.3 F I G. 7

FIG.4 FIG. 8 Figs. 1-3. Dorsal, lateral, and ventral views of the head (X 1) of the type specimen of Salvadora hexalepis klauberi, new subspecies. Fig. 4. Pattern at midbody of the type specimen (not drawn to scale) of S. h. klauberi, characteris- tic of the race. Figs. 5-7. Dorsal, lateral, and ventral views of the head (X 1) of the type specimen of Salva- dora hexalepis mojavens8s, new subspecies. Fig. 8. Midbody pattern of the type of S. h. mojavensis (not drawn to scale), characteristic of the majority of specimens in the subspecies. well defined on the third and fourth scale tral on the right resulting in the final re- rows. duction to 13 rows. The scales of the sec- DESCRIPTION OF THE TYPE: An adult ond to fifth scale rows over the anal region male with an over-all length of 1125 mm., and on the base of the tail are strongly with the tail (255 mm. long and not quite keeled; otherwise all dorsal scales are complete) comprising 23 per cent of the smooth, with vestiges of double apical pits total length. The dorsal scale rows are visible on some of them. There are 201 19-17-15-13, the first reduction resulting ventrals, not including a small triangular from the loss of the fourth row on the neck scale at the junction with the gulars. The at the level of the eighth ventral. The row anal plate is divided, and the caudals num- which then becomes the fourth row at mid- ber 91 with perhaps as many as 10 missing body is lost at the 128th ventral (right) with the tip of the tail. and at the 125th ventral (left). The sev- The scales on the top of the head include enth row is suppressed above the 132d ven- the large, recurved rostral with the lateral tral on the left and above the 128th ven- edges grooved and projecting, paired inter- 4 AMERICAN MUSEUM NOVITATES [No. 1285 nasals, prefrontals, a frontal and a pair of row is immaculate on either side, but faint parietals, as depicted in figure 1. traces of brown are present on the second The nasals are divided, with the ante- row. The ventral surface, from the mental rior section in contact with the second su- plate to the end of the tail, is uniformly pralabial. There are two loreals on each white with a faint tinge of yellow. side, the upper larger than the lower. Preoculars are 2-2, postoculars are 2-2, and SUMMARY OF THE PARATYPES there are no suboculars. The temporals are SCUTELLATION: Of the specimens sum- irregular but essentially 2 + 3 on the right marized here and listed individually be- and 2 + 2 on the left. There are nine su- low, 17 are males and 20 are females. Data pralabials on each side, of which the third drawn from the type are included because and fourth reach the loreals, and the fifth it represents part of the hypodigm. and sixth reach the ocular. The infrala- The dorsal scale row formula is normally bials are 11-11, the anterior three of which 19-17-15-13. However, on one female are in contact with the first pair of chin from Miraflores, the fourth row normally shields. The posterior pair of chin shields dropped on the neck is retained posteriorly are larger than the anterior pair and sepa- to the ninety-first ventral, giving a mid- rated by three scales, with an extra scale body count of 19. On another female on the left as shown in figure 3. from Cape San Lucas there are extra rows There are 11 teeth in the right maxilla, on the neck which are carried throughout followed after an interspace by three en- the length of the body, resulting in the larged teeth. On the left maxilla there are formula 21-19-17-15. Two additional 12 teeth anterior to the three enlarged females retain the seventh row as far as the teeth. The teeth in other series have not anus, and the resulting formula, 19-17-15, been examined on the type. One hemipe- is also found on a single male, the only one nis has been removed, and on the right the of the sex abnormal in scale row reductions. organ is partly everted so that no satisfac- The first reduction on the neck takes place tory examination could be made. between levels of the eighth and twenty- The pattern consists of a light clay-col- fourth ventrals on males (mean for 11 ored vertebral stripe three scale rows in sides 12.5), and, excluding abnormal indi- width flanked on either side by a brown viduals described above, between the sev- stripe embracing five scale rows. The up- enth and nineteenth ventrals on females per margins of the scales in the fourth row (mean for 9 sides 11.8). The second reduc- and the lower margins of those in the fifth tion, that involving the suppression of the are lighter so that the brown stripe is os- fourth row above the ventrals posterior to tensibly split at midbody (fig. 4). Anteri- midbody, occurs between the 118th and orly, however, the light margins disappear 137th ventrals on males (mean 128.2 for and the two portions of the stripe are con- 29 sides), and between the 121st and 138th fluent at the forty-fourth ventral. The on females (mean 131.2 for 36 sides). The brown stripe becomes narrower on the ventral scales vary from 191 to 203 in males neck and extends onto the temporal region, (mean 198.3 for 17 specimens), and from whereas the light vertebral band broadens 192 to 204 on females (mean 197.1 for 20 to about eight scales in width on the occiput specimens). The tip or more of the tail is where virtually the same color extends onto missing on so many specimens that sub- the dorsal cephalic plates. Posterior to the caudal counts for only 12 males and 13 fe- lateral scale reduction the band is clearly males can be summarized. Males vary divided, with the lower portion confined to from 91 to 103, mean 95.7; females from 86 the third scale row. Only the upper mar- to 100, mean 91.6. gin of the fourth row is brown, and the up- Supralabials are normally 9 (89.5 per per stripe falls principally on the fifth row, cent) but 8 occur on 3.9 per cent and 10 are diminishing in width as it extends onto the found on 6.6 per cent of 76 sides tabulated. tail. The lower portion of the stripe fades On 81.7 per cent of the sides, two suprala- out anterior to the anus. The first scale bials, usually the fifth and sixth and rarely 1945] RACES OF SALVADORA HEXALEPIS 5 the sixth and seventh, reach the eye. On ever, the pattern closely corresponds to the remaining sides only the sixth labial that described for the type. The light ver- reaches the eye. On 92.9 per cent of the tebral stripe is uniformly three scales wide, specimens the third and fourth labials reach and the brown lateral stripe is split anter- the loreal, the third, fourth, and fifth on iorly almost to the neck. The only excep- 5.7 per cent, and on only 1.5 per cent the tion is a specimen (L.M.K. No. 1068) from second, third, and fourth are in contact El Mlrmol (or Onyx Mine) just south of with the loreal. The infralabials are more latitude 300 N. This specimen is slightly commonly 10, but 11 are present in about darker, with the lower scale rows darker 40 per cent of the specimens. The loreal is than the ventral surface. The sixth supra- divided in 88.1 per cent of the sides, usually labial reaches the orbit on the right side, into two scales, although three are present and two suboculars present on the left on 9.2 per cent of the total. Single loreal side occlude all supralabials from the plates occur on 12.9 per cent of the sample. orbit. This is one of two specimens from The preoculars and postoculars are uni- Baja California in which the latter char- formly 2-2, and one subocular is present on acter occurs; it is rare in virgultea but pres- one or both sides of only six specimens out ent on one side of a Baja Californian speci- of 38. The total number of preoculars, men of the race. On the whole, therefore, postoculars, and suboculars varies from the El Mirmol specimen seems to be an only four to five, therefore, with 89.5 per intergrade. It is virtually identical with cent of 76 sides tabulated having four. The intergrades between virgultea and hexalepis, temporals are more commonly 2 + 3, but particularly those taken in the eastern foot- 2 + 2 are present on about 20 per cent of hills of San Diego County. the specimens, and 3 + 3 sometimes occurs. SEXUAL DIMORPHISM: As in all races of DENTITION: On 54 maxillae the formula S. hexalepis, the strongly keeled scales in 11 + 3 is represented on 70.4 per cent of the lateral rows above the anal region are the sample; the formula for the remainder present only on mature males. Despite is 12 + 3. Complete counts of all teeth the fact that the means for relative tail have been made on four specimens with the lengths, as well as for ventral counts, are following results: palatine teeth vary from substantially the same for both sexes, the 7 to 10 (mean 8.6), pterygoid 14 to 18 small samples indicate a difference in the (mean 14.4), and teeth on the dentary from mean number of subcaudals that may be 15 to 18 (mean 16.1). statistically if not biologically significant. DIMENSIONS: The largest specimen ex- Also it may be noted in the summary above amined is a male, from San Pedro at the that the fourth scale row is retained for a south end of the peninsula, which has a to- mean distance of three ventrals more on fe- tal length of 1147 mm., and perhaps 10 mm. males than it is on males. The largest are missing from the tip of the tail. The snakes of the species occur within the range largest female, with over-all dimensions of of S. h. klauberi, and males appear to attain 1135 mm., is from La Paz, a few miles greater lengths than those of the other sex. north of San Pedro. The ratio of tail to Of 18 females and 16 males that could be total length averages 0.23 in males with ex- measured with sufficient accuracy, only one tremes of 0.22 to 0.27. For females the female exceeds 1100 mm. in over-all di- mean is the same, but extremes are from mensions in contrast to four males that 0.22 to 0.24. exceed this figure. PATTERN: Some of the older specimens GEOGRAPHICAL VARIATION: In the ma- are too bleached, and a few modern speci- terial available, most of it from the cape mens permitted to remain too long in region, there is no indication of any gra- formalin are so badly discolored and dark- dient or trend that could be detected. ened, that they are virtually worthless However, more material from the peninsula for pattern characters. On most freshly north of La Paz would be required before preserved patch-nosed snakes from Baja mean differences would become apparent. California south of latitude 300 N., how- INTERGRADATION: The transition from 6 AMERICAN MUSEUM NOVITATES [No. 1285 virgultea to klauberi appears to be rather Salvadora hexalepis mojavensis, new abrupt, both in pattern as well as in scale subspecies characters. The diagnostic character, con- sisting of two labials reaching the eye, is MOJAVE DESERT PATCH-NOSED SNAKE found on 89.5 per cent of the klauberi sam- Figures 5-9 ple, whereas it occurs on but one of 11 in- TYPE: No. 63000 in the collection of the dividuals belonging to the subspecies vir- American Museum of Natural History, gultea from Baja California north of lati- taken at Deadman Point, 11.5 miles south- tude 30° N. Intergradation with hexalepis east of Victorville, at the south end of the presumably occurs at the northeastern Granite Mountains in San Bernardino corner of the range of klauberi. No speci- County, California, by C. M. Bogert, June mens of hexalepis are known from Baja 19, 1935. California, although there is every reason DIAGNOSIS: A subspecies of hexalepis to believe that hexalepis occurs in the distinguished from other racial populations northeastern corner of the peninsula. A of the species in more commonly having specimen from San Matias represents the suboculars which prevent any supralabials easternmost record for the species in this from reaching the eye. Also it differs from region, but the individual is referred to other subspecies in having an obsolescent virgultea on the basis of its low ventral pattern, with less contrast in shade between count (189), the pattern being bleached be- the vertebral and lateral stripes, and the yond recognition. The El MArmol speci- stripe on the third and fourth rows paler men mentioned above more closely re- than that on the sixth and seventh. sembles klauberi in pattern characters than DESCRIPTION OF THE TYPE: Adult male. it does virgultea. Yet the nature of the Length over all 800 mm., with the tail (175 difference is such that it would be nearly mm.) comprising 22 per cent of the total impossible to distinguish klauberi-virgultea length. The dorsal scale rows are 19-17- intergrades from hexalepis-virqultea inter- 15-13, with the fourth row dropping out grades, until several individuals from the on the neck at the point above the ninth critical area become available. (right) and tenth (left) ventrals, and the RANGE: The peninsula of Baja Califor- reduction to 15 resulting from the loss of nia, Mexico, from approximately latitude the fourth row at the 125th ventral (right) 300 N. southeastward to the Cape (Cabo and the 128th ventral (left); further re- San Lucas). duction results from suppression of the The subspecies is based upon 38 speci- seventh rows near the 150th ventral. mens including the type and paratypes Scales in the third to fifth rows above the with locality data as follows: anal region are keeled; otherwise all are Five miles north of San Xavier Mission smooth with vestiges of double apical pits (L.M.K. No. 30386); San Ignacio (M.V.Z. not easily seen. The ventrals number 203, Nos. 13565, 10670, 10671, L.M.K. No. 3827, U.M.M.Z. No. 76463); 20 miles west of Santa the subcaudals 94, not including four par- Rosalia (U.S.N.M. No. 37517); 3 miles south tial subcaudals on either side of the vent. of Canipole, northwest of Loreto (L.M.K. No. The anal plate is divided. The scales on 30385); Loreto (L.M.K. No. 30387); La Paz the head include the recurved rostral with (C.A.S. No. 45958, U.S.N.M. Nos. 12637, 12637a, 12637b, 12638, M.C.Z. Nos. 37228, lateral edges grooved and projecting (figs. 37229); San Pedro (C.A.S. Nos. 45951, 45952); 5-6), a pair of triangular internasals, a pair Buena Vista, Eureka Rancho, Palmas Bay, of larger prefrontals, a large frontal, large (C.A.S. No. 45953); Todos Santos (U.S.N.M. supraoculars, and a pair of parietals. No. 37518, C.A.S. No. 45957); Eureka (M.V.Z. The nasals are with the anterior No. 11838); vicinity east of Santiago (C.A.S. divided, No. 45954); Miraflores (U.S.N.M. No. 64581, section in contact with the second suprala- C.A.S. Nos. 45955, 45965); Santa Anita (S.U.M. bial. The loreal is divided on both sides, Nos. 4297, 4298); San Jos6 del Cabo (C.A.S. with the upper portion larger than the Nos. 45949, 45950); Cape San Lucas (C.A.S. No. lower. Preoculars are suboculars 45959, U.S.N.M. No. 5347, L.M.K. Nos. 20466, 2-2, 2-2, 20858, 20511, 20912 type); "Baja California" postoculars 2-2. The temporals are 2 + 3, (A.N.S.P. Nos. 5487, 5488). 2 + 3. The supralabials are 9-9, with the RACES OF SALVADORA HEXALEPIS 7 third and fourth reaching the loreals; none brae, mean 8.3 on males and 10.5 on fe- of them is in contact with the eye. The males (six of each sex tabulated). The infralabials are 11-11, the first four touch- second reduction, involving the suppres- ing the anterior chin shields. The posterior sion of the fourth row at midbody, occurs pair of chin shields are separated by two between the 105th and 136th ventrals on scales followed by three scales. males (mean 125.6 for 62 sides) and On the right maxilla there are 11 teeth between the 114th and 141st ventrals on (12 on the left) slightly increasing in size females (mean 127.9 for 38 sides). The posteriorly and followed after an interval loss of the sixth or seventh rows is more er- by three enlarged teeth. Other teeth on ratic, and not satisfactorily tabulated ow- the type have not been examined. ing to the fact that the row may be inter- The hemipenis, examined as dissected in rupted before being permanently sup- situ, extends to the twelfth subcaudal. The pressed. The ventrals are as follows: for sulcus is single with one large and three males the extreme range is 194 to 208, mean smaller spines inserted at the level of the 200.1; for females 191 to 208, mean 200.5. fourth subcaudal on one side only. Be- Caudals on males (tails complete on only 27 tween the sixth and seventh caudals there specimens) range from 82 to 98, with a are six spines in diagonal rows, three on mean of 92.4; on 16 females with complete either side of the sulcus when the organ is tails from 80 to 95, mean 85.1. Suprala- split opposite the sulcus. These are fol- bials are distributed as follows: 8 (5.7 per lowed by spinules which merge distally cent), 9 (80.5 per cent), 10 (12.2 per cent), with reticulated calyces having crenulated 11 (1.6 per cent) for 123 sides recorded. edges. On 71.0 per cent of the sides tabulated the The ill-defined pattern consists of a gray supralabials are excluded from the ocular vertebral stripe, three scales in width at by suboculars; a single supralabial, usually midbody, flanked by a light brown stripe the sixth, reaches the eye on 26.6 per cent represented by scales with darker centers of the sample, and on 2.4 per cent two la- on the sixth and seventh rows (see fig. 8). bials reach the eye. (See table 1 for com- A secondary lateral stripe, somewhat parisons of samples representing the five lighter in shade than the upper one, falls on subspecies of hexalepis.) Normally the the third and fourth scale rows. Anteriorly third and fourth supralabials reach the the lateral stripes merge and become faint loreals (84.4 per cent), but on 4.2 per cent on the neck, while the vertebral stripe, inso- of the sides tabulated the second and third far as its boundaries can be determined, are in contact with the loreals, the third widens to seven scale rows in width on the only in 2.1 per cent, and the third, fourth, occiput. Posteriorly, as a result of scale and fifth in 9.3 per cent. The infralabials reduction, vestiges of the lateral stripes fall are normally 10 or 11, usually with the on portions of the fifth and sixth, and on the first four touching the anterior chin shields. third scale rows, respectively. The under- In this subspecies the loreal is very rarely side is immaculate pale yellow, with the divided (0.8 per cent); more often there same color extending onto the lower scale are two loreals (91.1 per cent), sometimes row on each side. three (6.5 per cent), or uncommonly four (1.6 per cent). On 92.8 per cent of the SUMMARY OF THE PARATYPES sides there are two preoculars, sometimes SCUTELLATION: Of the specimens for three (6.3 per cent), and a single preocular which the data below are summarized, 40 is present in one side of one specimen. The are males and 22 are females, while two are number of suboculars varies from zero to heads only, of which the sex cannot be as- three, but usually there are two. The certained. Data drawn from the type are postoculars vary from two to four, with also included. The dorsal scale formula is the vast majority (92.8 per cent) having uniformly 19-17-15-13, with the first re- two., Because the suboculars occlude the duction occurring on the neck in the region labials from the eye, and because it is some- between the fifth and nineteenth verte- times an arbitrary decision whether indi- 8 S AMERICAN MUSEUM NOVITATES [No. 1285 vidual scales are called suboculars, it is San Bernardino County, California, where convenient to tabulate the total number of intergradation with hexalepis occurs, some oculars in comparisons of samples. The specimens are marked with darker cross temporals are scale-like and on 73.3 per bars. In the southwestern portion of the cent of the sides consist of two scales fol- range (see fig. 10) these dark cross bars are lowed by three (2 + 3), but 2-2 are present not clearly defined and they do not obscure on 17.9 per cent of the sides, and 3 + 3, the vertebral stripe. A more strongly 4 + 3, 3 + 4, and 3 + 2 sometimes occur. barred pattern is characteristic of speci- DENTITION: Counts for 78 maxillae have mens from Utah, and from the Grand Can- been tabulated, but the number of teeth yon and the Painted Desert in Coconino in other bones has been determined on County, Arizona. On these the vertebral both sides of only three specimens. On all stripe is irregular and broken by the dark specimens examined the anterior maxillary bars which may also extend onto the lower teeth are subequal and vary in number scale rows and even onto the ventrals (fig. from 10 to 12, invariably followed by a dia- 9). In Nevada a similar pattern is found stema and three enlarged teeth. The on one specimen from Mount Grant, Min- eral County, and traces of cross bars are present on specimens from Clark County. On a very few individuals the secondary stripe on the third and fourth rows may be scarcely evident on the third row so that the pattern may resemble that found on S. h. deserticola. SEXUAL DIMORPHISM: Whereas the means for ventral counts and tail length ra- tios are virtually identical, there are evi- dently differences between the sexes in sub- IG. 9 Fig. 9. Midbody pattern of S. h. mojavensi8 caudal counts and in the position of the (B.Y.U. No. 1499, from St. George, Washington lateral scale row drop at midbody, as County, Utah). This aberrant pattern is char- races acteristic of specimens from Utah and Coconino noted above. As in other of hexalepis, County, Arizona; it also occurs on one specimen strongly keeled scales are present above from Mount Grant in Mineral County, Nevada. the anal region on sexually mature males, whereas the scales in the same region formula is 11 + 3 on 78.5 per cent of the are smooth in females or very faintly keeled sample, 12 + 3 on 15.4 per cent, and 10 + on larger specimens. Males probably ex- 3 on 5.1 per cent. Means and extremes ceed the females in size, since nearly one- (in parentheses) for the teeth in other third of 31 males measured exceed in total bones are as follows: palatine 6.6 (6-8), length the largest female in the sample. pterygoid 11.8 (9-13), dentary 16.1 (15-18). GEOGRAPHICAL VALIATION: As noted DIMENSIONS: The largest specimen ex- above, the specimens with barred rather amined is a male taken "20 mniles from Mo- than simple lineate patterns occur around jave," Kern County, California, measuring the periphery of the range, except on the 941 mm. over all. The largest female, from southeast. Likewise specimens with the the same county, 20 miles south of Inyo- sixth, or with the fifth and sixth, suprala- kern, is 802 mm. in total length, with a bials reaching the eye tend to occur most portion of the tail missing. Ratios of tail frequently in peripheral local populations. to total length vary from 0.21 to 0.24 in Utah specimens represent an aberrant pop- males and females, with a mean of 0.23 for ulation, with five out of seven specimens both sexes. being asymmetrical in the character. On PATTERN: On the majority of specimens seven specimens from Washington County, the pattern approximates that described Utah, the sixth labial reaches the eye on for the type. However, around the periph- nine sides, and the fifth and sixth reach the ery of the range, except in southeastern eye on two sides. Only three sides con- 1945]^ RACES OF SALVADORA HEXALEPIS form to the condition characteristic of the in the region centering around Antelope majority of the specimens referred to mo- Valley tend to be darker than mojavensis javensis. A single labial reaches the eye from the center of the range. It may be also in the majority of specimens taken noted that while these specimens are suffi- along the southern edge of the range. But ciently light in coloration to be referred to all specimens from the Grand Canyon and mojavensis, most of them lack the complete the Painted Desert in northern Arizona row of suboculars that occlude the sixth conform in having the labials separated supralabial from the orbit. from the eye. Intergradation with S. h. hexalepis takes In but one other character is there much place over a wider area, and the pattern evidence of geographical variation; the changes involved in the transition of one mean ventral counts tend to be high'est in race with the other are so subtle that it is San Bernardino and Inyo counties, with impossible to select any individual speci- the lowest individual counts occurring at mens which might be designated as inter- the extreme north end of the range, in grades. A specimen from the Chemehuevis Nevada and in Utah. Two females each Mountains in Arizona has a pattern more with but 191 ventrals, the minimum count or less intermediate between that of hexa- for the subspecies, are from Sutcliffe, Ne- lepis and that of mojavensis, and the sixth vada, and from Zion National Park in labial reaches the eye on one side but a Utah. Insufficient material is available, complete row of suboculars is present on however, to work out gradients with satis- the other. This specimen has somewhat factory results, there being too few speci- arbitrarily been referred to mojavensis. On mens from any single county. a similar basis three specimens from the INTERGIRADATION: Where the range of Providence Mountains, which lie somewhat mojavensis abuts that of virgultea in north- farther north on the California side of the ern Los Angeles County, California, rela- Colorado River, are referred to hexalepis; tively few specimens have been taken. these specimens uniformly have the sixth However, one specimen, U.M.M.Z. No. supralabial reaching the eye, but their pat. 84910 from Lake Hughes (a small sag pond terns are essentially those of mojavensis. on the San Andreas Fault and separated Thus, southeastern San Bernardino from the Mojave Desert by a low ridge), County, California, represents a region is intermediate in pattern characters, al- where the diagnostic characters are though in scutellation it more closely ap- shuffled, so to speak, and to some extent proximates virgultea. The light vertebral the mixture of scale characters evident in stripe is three scales wide at midbody and this zone extends along the Colorado River broadens to seven scales in width on the to its junction with the Virgin River in occiput where it merges with the slightly Nevada and continues into southwestern darker coloration of the cephalic plates. Utah. In contrast the Coconino County A well-defined, dark, lateral stripe extends population in Grand Canyon and the onto the temporal region, and at midbody Painted Desert is uniform in lacking the it includes scale rows three to seven, with suboculars, but is atypical in having the the split in the stripe becoming apparent as barred pattern. a result of light margins on the upper por- In the southwestern corner of the range, tion of scales in the fourth row and the mojavensis ranges as far south as Twenty- lower margins of those in the fifth row. nine Palms, and a specimen (no longer ex- The snake is 1075 mm. over all (tip of tail tant) from Quail Springs near the summit missing) and thus exceeds in length any of the Little San Bernardino Mountains specimen of mojavensis. In pattern char- was quite typical in pattern and scalation. acters this Lake Hughes individual re- The escarpment on the south is steep, how- sembles snakes of the species' from southern ever, and all patch-nosed snakes taken in Baja California. Coachella Valley in Riverside County are The few other snakes taken along the referable to S. h. hexalepis. southwestern border of the Mojave Desert RANGE: Principally the Mojave Desert, 10 AMERICAN MUSEUM NOVITA TES [No. 128S from Washoe County, Nevada, on the No. 8177). MINERAL COUNTY: 10 miles east of north, southeastward across southern Ne- Basalt (U.M.M.Z. No. 84909); 7.5 miles north- west of Hawthorne (M.V.Z. No. 20396); Cot- vada to Washington County, Utah, and tonwood Creek, Mount Grant at an elevation through the Grand Canyon to portions of "below 7400 feet" (M.V.Z. No. 17130). NYE. the Painted Desert in Coconino County, COUNTY: Near Skeleton Hills (S.U.M. No. Arizona. On the south mojavensis merges 10324); near Bullfrog, Armagosa Desert (S.U. M. No. 10325); Tonopah (M.V.Z. No. 17598). with S. h. hexalepis in Mojave County, CLARK COUNTY: 20 miles north of Boulder Dam Arizona, and in southeastern San Bernar- (M.V.Z. No. 19263); Virgin River (U.S.N.M. dino County, California. In the south- No. 18062); Las Vegas (A.M.N.H. No. 9681), western portion of the range in California 19 miles southeast of Indian Springs near Corn Creek (L.M.K. No. 25386); 2 miles east of intergradation with S. h. virgultea occurs Indian Springs (M.V.Z. No. 33722); 13 miles in northern Los Angeles County, and pos- west of Indian Springs (L.M.K. No. 25357). sibly in central Kern County where the range extends through Tejon Pass into CALIFORNIA INYO COUNTY: Daylight Pass, Funeral Moun- southern San Joaquin Valley. tains (L.M.K. No. 32827); Johnson Canyon The following specimens comprise the (M.V.Z. No. 6689); 3 miles southwest of Wild- paratypic series: rose Station (L.M.K. No. 34100), and Towne's Pass (L.M.K. No. 25384), all in the Panamint UTAH Mountains; Maturango Spring (U.S.N.M. No. WASHINGTON COUNTY: Zion National Park 18060), Mountain Spring (C.A.S. No. 65499, (Z.N.P. no number, B.Y.U. No. 2880); St. M.V.Z. Nos. 33720, 33721), Shepherd Canyon George (B.Y.U. Nos. 1499, 1097); Santa Clara (U.S.N.M. No. 18059), and Homeward Canyon (R.H. 2 specimens, no number); and "Cotton- (C.A.S. No. 65498) in the Argus Mountains; wood Canyon" presumptively in the county Linnie (L.M.K. No. 28578); Amargosa Borax (U.S.N.M. No. 9101). Works (U.S.N.M. No. 18061). SAN BERNAR- DINO COUNTY: Near Cave Spring, Avawatz ARIZONA Mountains (S.U.M. No. 10354); Mountain COCONINO COUNTY: Tanner's Spring near Pass (L.M.K. Nos. 33446, 4400) and Wheaton Tuba City (M.V.Z. No. 2681); 9 miles west of Springs (L.M.K. No. 4401) in the Ivanpah Cameron (L.M.K. No. 34439); 15 miles north Mountains; Klinefelter (L.M.K. No. 10692); 7 of Cameron (M.V.Z. No. 19206); Tonto Plat- miles west of Red Pass, northeast of Barstow form, Bright Angel Trail, Grand Canyon (L.M.K. No. 35896); Hawes (L.M.K. No. (G.C.N.P. Nos. R-29, 76). MOJAVE COUNTY: 33951); Phelan (U.C.L.A. No. 17); Twenty- Mojave Valley (U.S.N.M. No. 4770); Fort nine Palms (L.M.K. No. 8503). KERN COUNTY: Mojave (M.V.Z. No. 8286); White Hills, 28 Dove Springs, Red Rock Canyon (U.C.L.A. miles north of Chloride (L.M.K. No. 25385); No. 87); 20 miles south of Inyokern (L.M.K. Lucky Star Mine, Chemehuevis Mountains No. 26130); 3 miles west of Rosamond (U.C.- (S.D.S.N.H. No. 17315). L.A. No. 461); "20 miles from Mojave" (C.U. No. 3625); mouth of Kern Canyon (H.T.W. no NEVADA number); 2 miles north of Grapevine Station WASHOE COUNTY: Sutcliffe near Pyramid (L.M.K. No. 25864), Kern River, near Bakers- Lake (C.A.S. No. 40506). CHURCXHILL COUNTY: field (G.W.H. no number). Los ANGELES 7.5 miles east of Frenchman's Flat (M.V.Z. No. COUNTY: Wilsona (U.C.L.A. No. 57); Harold 16179). LYON COUNTY: Near Nordyke (S.U.M. (U.C.L.A. No. 58).

DISCUSSION On the basis of pattern and scale char- in the samples was made upon the basis of acters it is possible to segregate five races pattern characters, whereas differentiation of Salvadora hexalepis. Pattern characters in scale characters rarely occurs in pre- are most variable in hexalepis and mojaven- cisely the same zone as intergradation in sis, and least variable in deserticola. pattern characters. In southeastern Ari- Similarly the diagnostic scale characters zona the difference in pattern between hexa- are least variable in deserticola, and most lepis and deserticola is virtually a dichoto- variable in hexalepis and mojavensis. To mous one, and it is closely correlated with some extent the greater variation in scale loreal characters. However, the number characters may be attributed to the fact of labials reaching the eye, usually two in that the primary segregation of specimens deserticola and one in hexalepis, is also of RACES OF SALVADORA HEXALEPIS 111 diagnostic value. But along the eastern fore, to the extent that one portion of the periphery of the range of hexalepis, particu- range is better represented than are other larly in Pima and Pinal counties, two portions. labials reaching the eye is not a rare char- It follows that a simple tabulation of acter in specimens having the pattern of characters for the whole sample does not hexalepis. In fact it is found on 50 per cent provide an adequate summary; the true of the specimens from the two counties. picture is not represented. On the other Similarly a high percentage of the speci- hand the small number of specimens avail- mens in the mojavensis sample having one able from most counties within the range of labial instead of no labials reaching the eye the species hexalepis precludes adequate are from the southern portion of the range. treatment of the data if specimens from re- In addition, the two local populations, in- stricted areas are segregated for considera- habiting (1) the valleys of the Virgin River tion. Hence there is no completely satis- and its tributaries and (2) the Grand Can- factory alternative at present, but the yon appear to represent aberrant offshoots table below must be interpreted in the in pattern characters, the more northern light of the previous discussion. one retaining scale characters of hexalepis The number of supralabials in contact whereas the strain in Grand Canyon has with the eye, or, more precisely, in contact the scale characters of mojavensis. with the ocular scale covering this organ, In view of these facts it is instructive to varies more or less independently of most note that the statistical validity of diag- other scale characters. In pait the nature nostic characters is dependent upon the of the contact is correlated with the number sources of material composing the sample. of suboculars, but the correlation is not so Had the bulk of the specimens comprising perfect as one might expect. The number the hexalepis sample been taken in Pima or of supralabials anterior to the eye, of Pinal counties the supralabial-eye char- course, determines which of these reach the acter would have appeared to be of little eye, but not how many. Two supralabials value. Or had most of the specimens as- beneath the eye may be coalesced into one, signed to mojavensis been taken in north- but this abnormality is so rare as to be of ern Los Angeles County, only pattern little importance. As may be observed in characters would have served to differenti- table 1, the number of supralabials reach- ate the race. ing the ocular is of diagnostic importance, Valid use of statistical methods presup- particularly when it is borne in mind that poses random sampling, and certainly the the primary segregation of specimens com- material available to taxonomic workers is prising the sample was based upon pattern ordinarily not consciously "selected." On characters. Despite the fact that changes the other hand larger series of specimens in supralabial characters lag behind the pat- tend to be assembled in more densely popu- tern changes in the zones of intergradation lated areas, and to this extent our samples between some races, the majority of speci- are often biased. For examples, the ma- mens from each of three regions conform jority of the specimens of klauberi were se- to one of three conditions encountered in cured in the south end of the peninsula, the supralabial-ocular character. the bulk of the available virgultea are from By means of this character it is possible San Diego County, California, and more to segregate two pairs of populations, and specimens of hexalepis are from Pima a third single population. A directional County, Arizona, particularly from the trend from south to north is apparent, with vicinity of Tucson, than from any other a single northern population most poorly county within the range of the species. differentiated. In essence the variation is More than one-third of the specimens as- clinal, but the change occurs in steps rather signed to deserticola come from southeast- than being continuous. On a statistical ern Arizona. All samples available in this basis a "step cline" is indicated, with lat- study, with the possible exception of that eral populations differentiated on pattern from the Mojave Desert, are biased, there- characters at the south end of the cline. 12 AMERICAN MUSEUM NOVITATES [NO. 1285

In the tabulation which follows the sub- in the vertical column, with the forms on species are arranged in the order of their the western edge of the range preceding geographic occurrence from north to south those on the east.

TABLE 1 SUPRALABIALS IN CONTACT WITH THE OCULAR (Percentage of the sample in each class) Number of Sides Subspecies Tabulated 0 1 2 3 S. h. mojavensis 124 71.0 26.6 2.4 0 S. h. vltrultea 236 5.1 84.3 10.6 0 S. h. hexalepis 192 4.2 72.4 23.4 0 S. h. klauberi 76 0 18.4 81.6 0 S. h. deserticola 148 1.4 2.7 94.5 1.4 On the basis of these data, as well as S. h. mojavensis at the north end of the other evidence that need not be discussed cline is the most poorly differentiated, and here, it may be hypothesized that deserticola probably the most recently evolved. represents the most primitive population.

SUMMARY Five subspecies of Salvadora hexalepis, from south to north as a "step cline" when two of them described herein, are recog- populations are segregated primarily on nized. The known ranges of the races are pattern characters. A key to the races is mapped, and it is shown that one scale supplied. character of diagnostic importance varies

LITERATURE CITED BOGERT, CHARLES M. HARTWEG, NORMAN 1939a. A study of the genus Salvadora the 1940. Description of Salvadora intermedia, patch-nosed snakes. Publ. Univ. new species, with remarks on the California Los Angeles Biol. Sci., vol. grahamiae group. Copeia, pp. 256- 1, pp. 177-236, 2 figs. in text, pls. 3-7, 259. tables 1-3, 2 maps. 1939b. Notes on snakes of the genus Salva- SCHMIDT, KARL P. dora with a redescription of a neg- 1940. Notes on Texan snakes of the genus lected Mexican species. Copeia, pp. Salvadora. Zool. Ser. Field Mus. 140-147, fig. 1. Nat. Hist., vol. 24, pp. 143-150, figs, BOGERT, CHARLES M., AND JAMES A. OLIVER 13-15. 1945. A preliminary analysis of the herpeto- fauna of Sonora. Bull. Amer. Mus. SMITH, HOBART M. Nat. Hist., vol. 83, pp. 297-426, text 1941. Further notes on Mexican snakes of figs. 1-13, pls. 30-37, table 1, maps the genus Salvadora. Smithsonian 1, 2. Misc. Coll., vol. 99, pp. 1-12, figs. 1-7. 1945] RACES OF SALVADORA HEXALEPIS 13

IFIG. 10 "7'f 1I Fig. 10. Map of southwestern United States and northwestern Mexico, showing distributions of the subspecies of Salvadora hexalepis. The presence of S. h. hexalepis in northeastern Baja Cali- fornia, indicated on the map, is hypothetical; no specimens are yet available from the area, although it is relatively certain that S. h. hexalepis occurs there. Parallel lines through portions of adjoining: ranges indicate known regions of intergradation. 14 AMERICAN MUSEUM NOVITATES [No. 1285

KEY TO THE SUBSPECIES OF Salvadora hexalepis A. Two supralabials (usually the fifth and sixth) reaching the eye. B. Loreal single; lower lateral stripe on fourth scale row; subcaudals on males 76 to 87, on females 66 to 82; ventral scales usually less than 192 (82 per cent).deserticola Schmidt BB. Loreal divided; lower lateral stripe on third and fourth scale rows; sub- caudals on males 91 to 103, on females 86 to 100, ventral scales usu- ally more than 192 (94 per cent).klauberi Bogert AA. One supralabial (usually the sixth) reaching the eye. B. Light vertebral stripe three scales in width; top of head gray; lat- eral stripe distinctly separated at midbody ...... hexalepis (Cope) BB. Light vertebral stripe one and two half scales in width, top of head brown, lateral stripe not or feebly separated at midbody. virgultea Bogert AAA. No supralabials reaching the eye (owing to the presence of suboculars). B. Lower lateral stripe on third and fourth scale rows fainter than the upper stripe on the sixth and seventh rows; the entire pattern is sometimes obsolescent or, in the eastern portion of the range, cross bars obscure the vertebral stripe.mojavensis Bogert