Hybridization Between the Endangered Unisexual Gray-Checkered Whiptail Lizard (Aspidoscelis dixoni) and the Bisexual Western Whiptail Lizard (Aspidoscelis tigris) in Southwestern New Mexico

Abstract Hybridization between the unisexual Aspidoscelis dixoni and the bisexual Aspidoscelis tigris punctilinealis in southwestern New Mexico is documented by observations and analyses of external morphology (coloration, size, scalation), chromosomes (karyotypes), nuclear gene products (allozymes), and mitochondrial DNA. The locality (Hidalgo County, Antelope Pass of the Peloncillo Mountains, centered at 10.5 km west of Animas), consisting of only a few square kilometers, is the only place where this particular unisexual clone of A. dixoni exists. Because of its extreme rarity in recent years, A. dixoni has been listed as an Endangered Species in New Mexico, and the status of its populations has received intense study. Today, the cause(s) of endangerment remains unknown, although we hypothesize that interspecific competition may be the problem. Aspidoscelis dixoni is a diploid unisexual species that normally reproduces by parthenogenetic cloning, as demonstrated here with genetic data from laboratory-reared lizards. However, fertilization of its eggs in Antelope Pass is possible if mating occurs with a male of the syntopic bisexual species A. tigris punctilinealis. The resulting hybrids closely resemble their maternal parent morphologically, but they are triploid and the females observed to date have been sterile. Aspidoscelis t. punctilinealis is a recent invader of southwestern New Mexico. It is the dominant species of whiptail lizard today in the low-elevation, semiarid habitat of creosote desertscrub in Antelope Pass. The present rarity of A. dixoni in Antelope Pass, in contrast to its abundance a few decades ago, may result from negative interactions with this dominant species, including asymmetrical destabilizing hybridization. Only a few other populations of A. dixoni are known to exist, each in a limited area in southwestern Texas, so there is a hiatus of nearly 500 km between the small and restricted populations in New Mexico and Texas. Comparative genetic data presented here indicate that although these populations are similar, the population in New Mexico represents a unique clone. It has three alleles at 3 nuclear gene loci (among 31 examined) that distinguish it from the Texan populations, and it lacks a microchromosome that occurs in Texan populations. In addition, in this paper we present new comparative genetic data confirming that the origin of A. dixoni itself was from a hybrid between an A. tigris marmorata ♀ × A. gularis septemvittata ♂, consistent with earlier studies.


INTRODUCTION
The gray-checkered whiptail lizard, Aspidoscelis dixoni (Scudday, 1973), occurs in extremely limited areas in New Mexico and Texas, with population centers widely separated from each other ( fig. 1). As its distribution in New Mexico is confined to a few square kilometers in Antelope Pass in the Peloncillo Mountains, centered about 10.5 km west of Animas, Hidalgo County (Degenhardt et al., 1996), and individuals were rarely seen in recent years, the New Mexico Department of Game and Fish (NMDGF) has listed it as an Endangered Species.
When first recognized as a distinctive entity or clone, A. dixoni was allocated to pattern class F of Cnemidophorus tesselatus (see Zweifel, 1965), known only from New Mexico. Later, Scudday (1973) named it Cnemidophorus dixoni (5 Aspidoscelis dixoni following the generic allocation of Reeder et al., 2002), and he recognized two pattern classes, A and B, both occurring mostly in Texas. Most recently,  and Cordes and Walker (2006) recognized the New Mexican population as a distinctive form, which they referred to as A. dixoni pattern class C, or A. dixoni C, which, with its extremely limited global distribution consisting of only a few square kilometers in southwestern New Mexico, is one of the most endangered lizards in the world.
Owing to the extreme rarity of A. dixoni C in New Mexico recently, one of us (C.W.P.) and the NMDGF conducted extensive ecological and capture-mark-release-recapture studies largely to determine the status of this lizard, which had been thought by some herpetologists in the 1980s to be extinct (James R. Dixon and colleagues were unable to find any in several attempts, personal commun.). This intense field program (Sias and Painter, 2002) continued for seven years (1987)(1988)(1989)(1990)(1991)(1992)(1993) and included a series of pit-fall traps in grids (without drift fences) and arrays (with drift fences) that extended across approximately 9 km of Antelope Pass. A total of 552 pit-fall traps in 31 grids and arrays were inspected every other day throughout the lizard activity season. During these 447,650 trap days, a total of 1,281 different individuals of the conspicuously abundant bisexual Arizona desert whiptail, Aspidoscelis tigris punctilinealis (see Taylor and Walker, 1996, for the use of this subspecific name) were captured in the study area. Although A. dixoni C previously was known to be reasonably abundant in the area (they were readily observed on any warm sunny day during lizard activity season in the sandy lower elevations of Antelope Pass in the 1960s; Zweifel, personal commun. and personal observations, C.J.C.), only 220 individuals of A. dixoni were found in the seven recent years of trapping (Sias and Painter, 2002). In addition, three of the lizards captured appeared morphologically similar to A. dixoni but with subtle aspects of color and pattern that suggested to the keen-eyed field assistants (Barney Tomberlin and Anthony Snell) that they might be hybrids between A. dixoni and A. t. punctilinealis, the only bisexual species of whiptail lizard found in Antelope Pass (addi- tional unisexual species found there were A. exsanguis [extremely rare in Antelope Pass but abundant elsewhere], A. sonorae [extremely rare in Antelope Pass but abundant elsewhere], and A. uniparens [abundant in Antelope Pass and elsewhere]). While we were studying these possible hybrids, several projects were underway that were relevant to this research: (1) C.J.C. was keeping two living female presumptive hybrids in the Aspidoscelis colony at the AMNH to determine whether they would lay eggs, and also some captive A. dixoni were reproducing there; (2) Dessauer et al. (2000) were studying extensive hybridization and gene flow in A. t. punctilinealis 3 A. tigris marmorata in passes immediately to the north of Antelope Pass, Peloncillo Mountains, Hidalgo County, New Mexico; (3) Taylor et al. (2001) were analyzing frequent production of sterile female hybrids between A. tesselata 3 A. tigris marmorata near Roswell, Chaves County, New Mexico; (4) Reeder et al. (2002) were determining that North American species of Aspidoscelis are not congeneric with South American species of Cnemidophorus; and (5) James E. Cordes (Louisiana State University at Eunice) kindly sent us some living individuals of Aspidoscelis dixoni of pattern classes A and B from Texas.
The specimens from James E. Cordes allowed us to conduct comparative genetic analyses (karyotypes, allozymes, mtDNA) of A. dixoni from throughout its known range. These data can now be understood in the context of the important comparative histocompatibility data recently published by Cordes and Walker (2006).
With all of the relevant studies listed above completed, we can now address the following questions in this report: (1) Were the presumptive hybrids of A. dixoni C 3 A. t. punctilinealis from New Mexico actually such hybrids? (2) If so, were the female hybrids sterile or could they clone themselves? (3) Do nonhybrid individuals of A. dixoni clone themselves in captivity? (4) Do the new data confirm that A. dixoni itself is of hybrid origin, and if so, are its ancestral species probably the same taxa that were proposed in earlier reports? (5) What is the genetic relationship between the New Mexican and Texan populations of A. dixoni and between both of these and A. tesselata (see Walker et al., 1997, for use of this specific name)? (6) Is it possible that recent competition and destabilizing hybridization with A. t. punctilinealis are the causes of endangerment of A. dixoni C in New Mexico, the only place where this clone exists?

FIELDWORK
Pit-fall traps, some with and some without drift fences to guide lizards into the traps, were used to capture animals alive (Sias and Painter, 2002;fig. 2 here; traps were modified from Campbell andChristman, 1982, andJones, 1986). Trapping stations in Antelope Pass, Peloncillo Mountains, centered approximately 10.5 km west of Animas, Hidalgo County, New Mexico, were established for varying periods of time at 31 different local sites within habitat thought to have been available to A. dixoni. Traps were open continuously between April 1 and October 15 each year from 1987 to 1993, and each trap was checked every other day during this period, for a grand total of 447,650 trap days. Lizards that were captured were taken to the laboratory for recording the following information: (1) weight (to the nearest 0.5 g with spring scales in 1987-1988, then to the nearest 0.05 g with an electronic balance in 1989-1993); (2) snout-vent length (to the nearest millimeter with a rule); and (3) sex and reproductive condition, when possible. Most individuals were uniquely toe-clipped and released at the capture site within 48 hr, although some were preserved for later study; presumptive hybrids and a few nonhybrids of A. dixoni C were maintained in captivity to determine whether they would reproduce and for examination of coloration, karyotypes, allozymes, mitochondrial DNA, and scalation (see below). See appendix 1 for a list of the specimens examined.

CAPTIVE MAINTENANCE
Lizards were maintained in captivity by methods that have been described elsewhere (Cole and Townsend, 1977;Townsend, 1979;Townsend and Cole, 1985). Nevertheless, some problems remain to be solved to maximize success in captive maintenance (Porter et al., 1994).

KARYOTYPE ANALYSIS
We used previously published methods (sodium citrate cell suspension, methanol and glacial acetic acid fixation, flame drying) for preparing and studying standard, giemsastained chromosomes (Cole, 1979). For A. dixoni C from New Mexico, we examined 21 cells at mitotic metaphase from bone marrow of three females. For A. dixoni A from Texas, we examined 26 cells at mitotic metaphase from bone marrow of three females captured in the field by James E. Cordes and two laboratory-reared offspring from two additional adults. For A. tigris, we have examined dozens of specimens from widespread localities representing many subspecies, including an individual from Antelope Pass, all of which are karyotypically identical to each other (e.g., Lowe et al., 1970b;Dessauer et al., 2000;Taylor et al., 2001). We also examined 12 cells at mitotic metaphase from bone marrow of one female hybrid of A. dixoni C 3 A. t. punctilinealis from New Mexico. Individuals karyotyped are specified in Specimens Examined (see appendix 1).

ALLOZYMES AND DNA
Genotypes at 31 nuclear gene loci, based on phenotypes of tissue protein activities on starch gels following electrophoresis, were determined for five wild-caught and three laboratory-reared specimens (all females) of A. dixoni C and two specimens of presumptive hybrids of A. dixoni C 3 A. t. punctilinealis. Five nonhybrid A. t. punctilinealis from Antelope Pass were examined also, and their genetic data were consistent with data from those punctilinealis that were among the 607 specimens of A. tigris analyzed from additional sites nearby (Dessauer et al., 2000). Three specimens of A. dixoni pattern class A and two of B from Texas were also compared with these specimens, and all lizards analyzed are listed in the Specimens Examined (see appendix 1).
Methods of collecting and storing tissues followed Dessauer et al. (1996a). Methods of preparing homogenates, conducting electrophoresis, localizing specific proteins, and scoring gel phenotypes, as well as the abbreviations for specific gene loci, followed Harris and Hopkinson (1976), Murphy et al. (1996), and, particularly for lizards of the genus Aspidoscelis, Dessauer et al. (2000). Dessauer et al. (1996b) used allele-specific oligonucleotide probes to determine haplotypes of mitochondrial DNA (mtDNA) present in certain individuals in order to confirm maternity of hybrids and of clonal species of hybrid origin.

MORPHOLOGY
Photographs and notes on colors and pattern were recorded prior to preservation of the lizards. Characters of size and scalation were noted as discussed by Cole et al. (1988) and Dessauer et al. (2000). Descriptive univariate statistics of the eight useful characters are provided later, in table 3.
Seven meristic characters for each individual were used in a principal components analysis (PCA) to generate scores extracted from the correlation matrix. This depicted the pattern of morphological variation without a priori identification of specimens to group. We then used our samples of A. dixoni C and A. t. punctilinealis as a priori groups and a step-wise selection of meristic characters (F probabilities of 0.15 for character entry and 0.20 for character removal; Costanza and Afifi, 1979) in a discriminant analysis (DA) to identify which parental taxon the hybrids (entered in the model as unclassified) most closely resembled. Finally, a canonical variate analysis (CVA) was run, using the same stepwise, character selection procedure and sam-ples of A. dixoni C, A. t. punctilinealis, and the hybrids as three a priori groups for maximally discriminating among the groups. We used SPSS 14.0 (from SPSS, Inc., Chicago, Illinois) and SYSTAT 11.0 (from SYSTAT Software, Inc., Richmond, California) for statistical analyses.

COLORATION AND IDENTIFICATION OF THE HYBRIDS
Dorsally and laterally, A. dixoni C has a ground color of small dark brown or black squares and rectangles separated by light stripes and cross bars in a checkered pattern (Degenhardt et al., 1996). Anteriorly, the light stripes and bars are grayish tan, but gradually they become a brighter tan or cream to yellow posteriorly. An orange-brown wash may be present on the posterior body ( fig. 3). Ventrally, the throat is light tan with a slight touch of orange and without dark spots. The chest is paler than the throat, with a few small black dots. There is a distinct tan color posterolaterally on the abdomen, but the belly is cream or white without black. The color beneath the arms, legs, and tail is pale tan.
Dorsally and laterally, A. t. punctilinealis has a brown to black ground color with grayish tan light spots and usually four similarly colored stripes. The dorsum is not a checkered pattern ( fig. 3). Laterally, individuals of this taxon are black with cream spots (light gray anteriorly), which may tend to coalesce into vertical rows, but the sides are not barred as typically they are in the subspecies A. t. marmorata. Ventrally, the throat and chest are black, which extends well onto the anterior abdomen ( fig. 4). The light parts of the abdomen, especially posterolaterally, are white, not tan. The color beneath the arms is black, beneath the legs, light gray with darker gray smudges, and beneath the tail, gray. Color notes were taken from one female presumed (and later verified) hybrid of A. dixoni C 3 A. t. punctilinealis (figs. 3, 4). Dorsally she had six light stripes and spots (grayish tan anteriorly, deeper tan posteriorly) on the black ground color. Laterally, some of the light bars were broken into spots. Ventrally, the throat was pale orange with a few round black dots. The chest was light orange with a few black spots. The abdomen was cream with a trace of tan posterolaterally and a few black spots laterally. The color beneath the arms was light orangish tan with light gray mottling; beneath the legs and tail was light tan.
Clearly, the hybrids are most similar to A. dixoni C, but they have some coloration characters that are intermediate between the two parental types (figs. 3, 4; table 1). In particular, their sides are less regularly checkered than in A. dixoni, their pale orange throat has round black dots, their chest has larger black spots, and their abdomen has only a trace of tan posterolaterally plus a few black spots.

Individuals of Aspidoscelis dixoni A from
Texas have a karyotype with a diploid number of 47 chromosomes consisting of two rather different haploid genomes that reflect its own origin by hybridization. As with other clones in the A. tesselata complex, the hybrid origin of the unisexual A. dixoni was apparently of A. t. marmorata U 3 A. gularis septemvittata - (Neaves, 1969;Parker and Selander, 1976;Densmore et al., 1989;Dessauer and Cole, 1989;Dessauer et al., 1996b). The karyotype of A. dixoni A is modified from the ancestral F 1 hybrid condition in one respect (see below), although, as seen in most clones of A. tesselata (see Dessauer and Cole, 1989: 57;Taylor et al., 2001: 24), it is close to the original hybrid state.
The haploid set of chromosomes originally inherited from the maternal A. t. marmorata (n 5 23) includes three large set I biarmed macrochromosomes + eight smaller set II biarmed intermediate-sized macrochromosomes + 12 set III microchromosomes (Lowe et al., 1970b;Taylor et al., 2001). The second largest chromosome in set I of the ancestral A. tigris has a dotlike satellite on the end of one arm, set off by the nucleolar organizer region (Ward and Cole, 1986), which is often difficult to see, and the third largest chromosome is the sex chromosome (Cole et al., 1969;Bull, 1978). In both A. dixoni A and C this haploid complement has been modified in that the X chromosome has apparently undergone centric fission and is now represented by two smaller set II telocentric macrochromosomes, as found in some clones of A. tesselata (see Taylor et al., 2001). Consequently, the pattern of macro-and microchromosomes in the modified ancestral tigris genome is 2 + 10 + 12 instead of the expected 3 + 8 + 12.
In A. dixoni also, the haploid set of chromosomes originally inherited from the paternal A. g. septemvittata (n 5 23) includes one large set I metacentric macrochromosome (which has a subterminal secondary constriction [nucleolar organizer; Ward and Cole, 1986] on one arm followed by an elongate satellite that is at least twice the size of the satellite in A. t. marmorata) +12 smaller set II intermediate-sized telocentric or subtelocentric macrochromosomes + 10 set III microchromosomes ( fig. 5). However, in A. dixoni C from New Mexico, one of the microchromosomes (possibly one from either parent) is apparently missing (for illustrative purposes, shown here as missing from the  fig. 3), with 2n 5 46 chromosomes. This taxon is a clone with its ultimate ancestor having been an F 1 hybrid between A. tigris marmorata (haploid complement of chromosomes in the upper two rows) 3 A. gularis septemvittata (haploid complement in the lower two rows of A). Note that the X chromosome that was ultimately inherited from marmorata (third largest, upper row) was apparently fissioned at the centromere (arrow) forming two additional set II-sized macrochromosomes (here arranged to suggest the ancestral unfissioned state in marmorata). Note also that the missing microchromosome of A. dixoni C is illustrated as missing from the marmorata genome (o at the right end of the second row), although we do not know from which genome it was actually lost. B, Triploid hybrid female of A. dixoni C 3 A. t. punctilinealis (AMNH R-148141, fig. 3), with 3n 5 69 chromosomes arranged in six rows. Rows 2 and 4 represent the haploid complement most recently inherited from A. t. punctilinealis, including its intact X chromosome (third largest). Rows 1, 3, 5, and 6 represent the two haploid complements inherited from A. dixoni C, including the fissioned X (arrow) chromosome and missing microchromosome (o), as in A, above. Bar represents 10 mm. marmorata complement; fig. 5). In contrast, the full ancestral complement of 22 microchromosomes is present in A. dixoni A from Texas. In addition, the two laboratoryhatched offspring of A. dixoni A had the same karyotype, comprised of unpaired sets of chromosomes, consistent with clonal inheritance in the laboratory.
In summary, the modified diploid karyotype of A. dixoni C consists of 46 chromosomes, with 3 large biarmed macrochromosomes + 22 smaller macrochromosomes (2 being the fissioned X chromosome) + 21 microchromosomes that were originally inherited as two morphologically different haploid genomes ( fig. 5). Considering the fissioned ancestral X chromosome from A. t. marmorata and the absence of one of the ancestral microchromosomes, the three specimens of A. dixoni C examined had a unique karyotype, and this remained constant in samples collected over the years spanning 1966-1987.
The karyotype of A. tigris punctilinealis is identical to that of A. t. marmorata, with n 5 23 chromosomes (Cole et al., 1969;Lowe et al., 1970b;Dessauer et al., 2000). Consequently, a hybrid between A. dixoni C and A. t. punctilinealis would be predicted to have 3n 5 69 chromosomes, including the cloned diploid set from the maternal A. dixoni C and the additional haploid set from the paternal A. t. punctilinealis. The suspected hybrid female karyotyped was, in fact, a triploid with precisely the combination of chromosomes expected from an offspring of A. dixoni C 3 A. t. punctilinealis ( fig. 5), having 6 large set I biarmed macrochromosomes + 30 smaller set II macrochromosomes (2 being the fissioned X chromosome from dixoni) + 33 set III microchromosomes.

REPRODUCTION IN THE LABORATORY
The two hybrid individuals of A. dixoni C 3 A. t. punctilinealis maintained alive in captivity were of adult size when preserved (93 and 104 mm in snout-vent length). One of the hybrids lived in good health in captivity for nearly 23 months and the other for more than 29 months, but neither ever appeared gravid nor did they lay any eggs during this period. In contrast, one A. dixoni C from New Mexico that was maintained alive in captivity laid a clutch of six eggs over a captive period of 8.5 months, although none hatched. Another individual of A. dixoni C from New Mexico laid six clutches of eggs (N 5 5-6 per clutch for a total of 33 eggs) over a period of 26 months, of which three eggs hatched. In addition, individuals of A. dixoni from Texas also reproduced in captivity.
Four P 1 generation (field-captured) lizards of A. dixoni A from Texas produced 22 clutches (N 5 2-6 per clutch for a total of 69 eggs, of which 23 hatched) over a period of up to 46 months per lizard; six F 1 lizards produced 28 clutches (N 5 1-5 per clutch for a total of 96 eggs, of which 28 hatched) over a period of up to 24 months per lizard; and five F 2 lizards produced 7 clutches (N 5 1-4 per clutch for a total of 18 eggs, of which none hatched) over a period of up to 10 months per lizard.
We conclude that the triploid hybrid females were not capable of parthenogenetic cloning, although their mothers were. This is consistent with the report of Taylor et al. (2001) concerning hybridization between A. tesselata U 3 A. t. marmoratafrom the vicinity of Roswell, Chaves County, New Mexico, for which a larger number of sterile triploid female hybrids was analyzed. As in that study, triploid hybrids between A. dixoni C and A. t. punctilinealis theoretically could be either male or female, depending on whether fertilization was by a spermatozoan bearing the X chromosome or the Y chromosome. The hybrid sex ratio should be 50:50, and it was (2 females:1 male).
Finally, although the Texan individuals of A. dixoni produced more offspring than the New Mexican lizards in captivity, one must not read too much into this comparison. Reproduction in captive Aspidoscelis is negatively affected by aspects of the captive environment in complex ways that are not fully understood (Townsend and Cole, 1985;Porter et al., 1994).

BIOCHEMICAL GENETICS
Protein electrophoresis of allele products representing 31 nuclear gene loci plus ASO dot-blot tests identifying the haplotype of mtDNA (table 2) provide extensive genetic a Genotypes are presented for the hybrids of A. dixoni 3 A. tigris punctilinealis (HYB) and for unisexual taxa (DIX C, TES, and DIX A and B), as these did not vary from individual to individual. For bisexual taxa (MAR, SEP, PUN, and BUR) only the most common alleles are listed, in order from the highest frequency to the lowest. If no a-allele is shown (e.g., EST2) we have found it in species of Aspidoscelis that are not included here. b Alleles are designated in alphabetical order according to decreasing anodal migration of their allozymes. For multilocus enzymes, loci are numbered in order of decreasing anodal migration of their isozymes. Methods follow Harris and Hopkinson (1976), Murphy et al. (1996), and Dessauer et al. (2000;EST2 in particular); s, cytosolic enzyme; m, mitochondrial enzyme. Abbreviations for loci are as follows: ADH, alcohol dehydrogenase; IDDH, L-iditol dehydrogenase; LDH, L-lactate dehydrogenase; MDH, malate dehydrogenase; MDHP, malate enzyme; SOD, superoxide dismutase; AAT, aspartate aminotransferase; EST, esterase; PEP, peptidase; ADA, adenosine deaminase; ACOH, aconitase hydratase; MPI, mannose-6-phosphate isomerase; GPI, glucose-6-phosphate isomerase; PGM, phosphoglucomutase; TF, transferrin; ALB, albumin; and ASO, allele-specific oligonucleotide of mitochondrial DNA (mtDNA; M, A. tigris marmorata; P, A. tigris punctilinealis). Boldface alleles in A. dixoni at sMDH, PEPA, and MPI are distinctive in one way or another (see text). The following 11 loci were invariant (i.e., all individuals possessed one and the same allele in the homozygous state): G3PDH (glycerol-3-phosphate dehydrogenase, except for some variation found within A. burti); LDH2; mMDH; s and m IDH (isocitrate dehydrogenase); mSOD; CK1 (creatine kinase); AK (adenylate kinase); ESTD (except for some variation found within A. burti); PEPE (except for some variation found within A. burti); and HB (hemoglobin, except that this was not tested for A. dixoni class B from Texas). c Abbreviations for taxa and individuals are as follows: MAR, A. tigris marmorata; SEP, A. gularis septemvittata; DIX C, A. dixoni from New Mexico; HYB, hybrids of A. dixoni from New Mexico 3 A. tigris punctilinealis (PUN); TES, A. data for comparing specimens of A. dixoni of all three pattern classes, its close relatives, offspring, and hybrids involving A. dixoni. Alleles representing heterozygous phenotypes usually are expressed equally, with banding patterns clearly indicating ploidy and the dosage of each allele (Neaves, 1969;Neaves and Gerald, 1969;Dessauer and Cole, 1984). Occasionally, however, phenotypes for parthenogens of hybrid origin suggest that one allele at a polymorphic locus is more active than another at the same locus, as discussed for PEPA by Taylor et al. (2001). Scoring such loci can be clarified by examining control gels using homogenates of different tissues on which samples from both the parthenogen and hybrids are compared side by side.
The biochemical data (table 2) are based on six specimens of A. dixoni C from Antelope Pass, New Mexico, and three offspring produced by one of them in the laboratory; five A. dixoni (classes A and B) from Texas; six A. t. punctilinealis from Antelope Pass; two hybrids of A. dixoni C 3 A. t. punctilinealis; 11 A. tesselata E-C from Macho Draw, New Mexico (Taylor et al., 2001); 15 A. g. septemvittata; 4 A. b. burti; 20 A. burti stictogramma; and more than 600 additional A. tigris from the Animas Valley and vicinity of the Peloncillo Mountains, New Mexico, where A. t. punctilinealis and A. t. marmorata interbreed (Dessauer et al., 2000). These data provide evidence bearing on the following questions: (1) Is A. dixoni C of hybrid origin, consistent with earlier studies, and if so, what are its ancestral species? (2) Does A. dixoni reproduce by parthenogenetic cloning? (3) Are the presumptive hybrids between A. dixoni C 3 A. tigris punctilinealis from New Mexico actually such hybrids? (4) What is the genetic relationship between the New Mexican and Texan populations of A. dixoni and between these and A. tesselata?
IS A. DIXONI C OF HYBRID ORIGIN, AND, IF SO, WHAT ARE ITS ANCESTRAL SPECIES?: For the 31 protein loci analyzed (table 2), all of the specimens of A. dixoni C from New Mexico had alleles in the heterozygous state at 15 loci, or 48%. This very high level of heterozygosity, compared with about 5% in nonhybrids of bisexual taxa of Aspidoscelis (Parker and Selander, 1976;Dessauer and Cole, 1989), clearly indicates that A. dixoni had a hybrid origin. In addition, the data (table 2) are consistent with the ancestry of A. dixoni C being the result of hybridization between an A. t. marmorata U 3 A. g. septemvittata -, similar to that of the diploid parthenogen, Aspidoscelis tesselata (Neaves, 1969;Parker and Selander, 1976;Dessauer and Cole, 1989;Densmore et al., 1989; see below).
There have been few genetic data directly comparing A. tesselata and A. dixoni C published to date, so conclusions concerning their close relationship have been based mostly on morphology (Zweifel, 1965;Scudday, 1973), until very recently. The few data published on allozymes, for about five loci, are consistent with concluding that there is a close relationship between these taxa (Neaves, 1969). In addition, mtDNA of A. dixoni (Densmore et al., 1989;Dessauer et al., 1996b) clearly implicates A. t. marmorata as its maternal parent. Most important of all, a recent report on histocompatibility among all color pattern classes of A. dixoni and between tesselata pattern class E-C from Macho Draw, NM (see Taylor et al., 2001Taylor et al., , 2003; DIX AB, A. dixoni from Texas (class A and class B individuals were analyzed and they were identical at all loci, but individuals of class B could not be tested for TF, ALB, and ASO); BUR, A. burti burti and A. b. stictogramma. d The data for DIX C include several field-captured individuals plus three offspring produced by one of them in the laboratory (see Specimens Examined, appendix 1). As all individuals were identical, with the heterozygosities at all loci maintained from mother to offspring, this constitutes evidence of clonal inheritance in A. dixoni. e At sMDH, we have found the c-allele in some individuals of A. gularis gularis also, so it would not be surprising to discover that it also occurs in A. g. septemvittata.
f We did not have the opportunity to record data for four loci in individuals of A. dixoni class B from Texas: TF, ALB, ASO, and HB, the last of which was invariant in all of the other lizards.
g We did not test ASO for individuals of TES E-C from Macho Draw, but all other specimens of TES (all pattern classes) tested have had M (Dessauer et al., 1996b), so this is a reasonable assumption for these also. these and A. tesselata most strongly suggests that dixoni and tesselata share as a common ancestor one F 1 hybrid female that cloned herself (Cordes and Walker, 2006) and from which the pattern classes and other genetically variant clones differentiated owing to postformational divergence. The allozyme data presented here (table 2) are consistent with these data, indicating that the origin of A. dixoni was from a hybrid between A. t. marmorata U 3 A. g. septemvittata -.
We compared proteins representing 31 nuclear gene loci for A. dixoni C from New Mexico, A. t. marmorata, and A. g. septemvittata (table 2). For 29 of these loci, A. dixoni C is consistent with having had a hybrid origin with one gamete from A. t. marmorata and one from A. g. septemvittata, which is consistent also with its high heterozygosity. Of the 35 individual alleles identified in New Mexican dixoni, 33 (94%) were found in marmorata and septemvittata. The two exceptional alleles were the c-allele at sMDH and the second a-allele at MPI ( fig. 6C,D). Interestingly, at both the sMDH and MPI loci, the Texan dixoni (pattern classes A and B) differed from the New Mexican ones and the Texan samples had the expected hybrid genotype (A. t. marmorata 3 A. g. septemvittata) at both of these loci (table 2). Consequently, the allozymes of the Texan dixoni A and B more closely represent a perfect F 1 hybrid, with but one exceptional allele. That is the second dallele at PEPA in the Texan dixoni ( fig. 6B). Considering that the histocompatibility data strongly indicate that the New Mexican and Texan dixoni share one individual ancestral hybrid, and they share this with A. tesselata as well (Cordes and Walker, 2006), we conclude that these allelic differences reflected in allozymes are a result of postformational events, similar to those discussed by Taylor et al. (2003) for various clones of A. tesselata. For New Mexican dixoni C, the second a-allele at MPI could have stemmed from biased gene conversion (Hillis et al., 1991). The unexpected c-allele at sMDH in New Mexican dixoni C could have resulted from a postformational point mutation, and it is interesting that we have seen evidence for this same allele in some individuals of A. g. gularis, a close relative of A. g. septemvittata.
Early in this investigation, H.C.D. discovered the unexpected c-allele at sMDH in New Mexican A. dixoni and realized that the New Mexican and Texan specimens differed at this locus. Initially, we wondered whether these populations could have had separate hybrid origins, and if so, which species might be a candidate for the paternal ancestor for New Mexican A. dixoni C. Consequently, we added Aspidoscelis burti stictogramma to the comparisons because it is another large, spotted species of the sexlineata-group (Lowe et al., 1970b;Reeder et al., 2002) (table 2; ADH, sMDHP, sAAT, PEPA, ADA, MPI, PGM3, and ALB), although the c-allele occurs at sMDH in burti. Therefore, consistent with the histocompatibility data (Cordes and Walker, 2006), the allozyme data indicate that the ancestor of A. dixoni (all three pattern classes) ultimately was a hybrid between A. tigris marmorata U 3 A. gularis septemvittata -.
The haplotype of mitochondrial DNA in two of the specimens of A. dixoni C from New Mexico, the triploid hybrid of A. dixoni C 3 A. t. punctilinealis and two of the A. dixoni A from Texas was determined by Dessauer et al. (1996b). All of these specimens had the mtDNA of A. t. marmorata (table 2), as expected, owing to the maternal inheritance of mtDNA, because ultimately A. t. marmorata was the maternal ancestor of the clonal A. dixoni in its hybrid origin (A. t. marmorata U3 A. g. septemvittata -), and the mtDNA of A. dixoni was inherited by the recent hybrids with punctilinealis.
DOES A. DIXONI REPRODUCE BY PARTHENO-GENETIC CLONING?: Table 2 includes protein data from a field-captured adult A. dixoni C from New Mexico (AMNH R-148360, see Specimens Examined, appendix 1) and three offspring that she produced in the Aspidoscelis colony at the American Museum of Natural History (AMNH R-148364-66). The mother and all of her daughters (all from the same Fig. 6. Electrophoretic phenotypes representing products of four gene loci of Aspidoscelis. A, IDDH, a tetrameric enzyme, from liver homogenates of seven lizards. Note that the b-allele from PUN is very faint but present in the hybrid. B, PEPA, a dimeric enzyme, from muscle homogenates of seven lizards. Note that New Mexican dixoni differ from Texan dixoni and that the hybrid is triallelic. C, sMDH, a dimeric enzyme, from muscle homogenates of seven lizards. Note that New Mexican dixoni differ from Texan dixoni and the hybrid shows two doses of the a-allele. D, MPI, a monomeric enzyme, from muscle homogenates of eight lizards. Note that New Mexican dixoni differ from Texan dixoni. Letters below gels identify allozymes based on alleles present (table 2). Lanes for individual lizards are labeled beside the gel (with genotype) as follows: DIXA, A. dixoni A; DIXB, A. dixoni B; DIXC, A. dixoni C; DIXC 3 PUN, hybrid of A. dixoni C 3 A. t. punctilinealis; MAR, A. t. marmorata; SCA, A. g. scalaris, which at these loci are often the same as A. g. septemvittata; TESC, A. tesselata C-E; TESD, A. tesselata D; TESE-C, A. tesselata E-C; PUN, A. t. punctilinealis. Anode is to the right, and arrow in A indicates position of sample application. clutch of eggs) were identical in their genotypes at all loci, with the heterozygosity at the 15 heterozygous loci being preserved in all of the offspring. In addition, these individuals had the same genotype at all loci as the other five field-collected specimens of A. dixoni C from New Mexico. Clearly, this demonstrates that A. dixoni C normally reproduces by parthenogenetic cloning. ARE THE PRESUMPTIVE HYBRIDS BETWEEN A. DIXONI C 3 A. TIGRIS PUNCTILINEALIS ACTUALLY SUCH HYBRIDS?: Two of the lizards identified as hybrids in the field were tested biochemically (AMNH R-148141 and 148142). Both specimens were triploids and their genotypes were identical to each other (table 2). At each locus, these triploids had a combination of alleles that matched perfectly predictions expected for lizards that inherited the cloned, diploid genome from the maternal A. dixoni C (including her mtDNA) plus a haploid genome consisting of the alleles found commonly in the local A. t. punctilinealis males. For most loci, this conclusion was based on recording the relative band densities (reflecting gene dosage) on gels representing heterozygous loci for the diploids (both parental forms) and triploids, and the pattern was clear. At PEPA, the hybrids were most clearly triallelic ( fig. 6B), bearing the c-and d-alleles from dixoni C and the b-allele from punctilinealis. The triploid hybrids also had the unusual c-allele at sMDH ( fig. 6C), which occurred in all dixoni C from New Mexico, as opposed to those (pattern classes A and B) from Texas. In addition, for the following four loci, dixoni C is a homozygote, but punctilinealis commonly bears a different allele, and at each of these loci the hybrids were heterozygotes with the predicted genomes: IDDH ( fig. 6A), PEPB, PEPD, and TF (table 2). Consequently, the level of heterozygosity in the triploid hybrids (61%) was significantly higher than in their maternal parent (48%).
The specimens under discussion were collected in Antelope Pass of the Peloncillo Mountains, Hidalgo County, New Mexico. A few kilometers to the north there are additional passes (Granite Gap and Steins Pass) where there is extensive interbreeding between two bisexual taxa, A. t. punctilinealis and A. t. marmorata, apparently as a result of recent secondary contact (Dessauer et al., 2000), and a new hybrid zone involving these taxa could develop one day in Antelope Pass as well, as a specimen of A. t. marmorata (AMNH R-148140) was recently found west of Animas. One might question, therefore, whether there is evidence of recently contributed marmorata alleles in the hybrids. The four most informative loci by which the local A. t. marmorata and A. t. punctilinealis differ are sMDHP, EST2, PEPB, and TF; in addition, PEPD shows significant allelic frequency differences. At each of these loci, the older, original ancestral marmorata allele was found in a single dose as expected in the local A. dixoni C, and the punctilinealis allele was found in the recently formed triploid hybrids. All of the genetic data were consistent with concluding that the triploids were a consequence of a cloned, diploid ovum from an A. dixoni C having been fertilized by a haploid spermatozoan from an A. t. punctilinealis.
WHAT IS THE GENETIC RELATIONSHIP BETWEEN THE NEW MEXICAN AND TEXAN POPULATIONS OF A. DIXONI AND BETWEEN BOTH OF THESE AND A. TESSELATA?: Data in table 2 provide the first comparison of allele products from nuclear genes for samples of A. dixoni C from New Mexico and pattern classes A and B from Texas. In addition, these are compared with individuals of A. tesselata pattern class E-C from Macho Draw, New Mexico (see Taylor et al., 2003, for the E-C pattern class designation).
There were no allozyme differences found between individuals of A. dixoni of classes A and B from Texas. For the genes tested, they were identical (table 2). In addition, the Texan A. dixoni were nearly identical to A. tesselata class E-C, differing by only one allele, the second d-allele at PEPA (fig. 6B). For PEPA, tesselata has a cd genotype, which is consistent with inheritance of the c-allele from A. t. marmorata and the d-allele from A. g. septemvittata. As the cd combination is also found in A. dixoni C from New Mexico ( fig. 6B), we hypothesize that it occurred in the parthenogenetic ancestor of Texan dixoni and that their c-allele experienced biased gene conversion to d (Hillis et al., 1991), particularly as all of these classes are histocompatible (Cordes and Walker, 2006). A single allele each at sMDH and MPI also differ between the New Mexican and Texan samples of A. dixoni ( fig. 6C,D), with Texan dixoni A and B having the original ancestral hybrid condition. Histocompatibility and the data in table 2 are consistent with hypothesizing that the c-allele at sMDH in New Mexican specimens resulted from point mutation(s) and the second a-allele at MPI in New Mexican specimens resulted from biased gene conversion. As mentioned above, all available data considered together indicate that all of the population samples of A. dixoni and A. tesselata share a common ancestor that was a parthenogen of hybrid origin from a single hybrid zygote (Cordes and Walker, 2006). This is consistent also with the clonal diversity found within A. tesselata. For example, Taylor et al. (2003) discussed evidence of a common ancestor for three allozyme clones involving sACOH, MPI, and GPI of A. tesselata of pattern class C-E from Conchas Lake State Park, New Mexico. As with some of the clones of A. tesselata also, the clones of A. dixoni have extremely limited geographic distributions. For example, the clone of A. dixoni C in New Mexico is unique (allozymes and karyotype) and today is known to occur only in an area of a few square kilometers in Antelope Pass. Whether the color pattern class of the original F 1 hybrid of A. t. marmorata 3 A. g. septemvittata mostly resembled A. dixoni or A. tesselata E or neither remains unknown.

SIZE, SCALATION, AND MULTIVARIATE ANALYSES
Morphological data were recorded for 16 Aspidoscelis dixoni C, 15 A. t. punctilinealis, and 3 of their hybrids, all from Antelope Pass, Hidalgo County, New Mexico. As the first 2 hybrids found were females, only females of the bisexual A. t. punctilinealis were examined, to avoid possible influence of a Y chromosome. The third hybrid was a male, but none of this should matter, as Dessauer et al. (2000) showed that none of these characters is sexually dimorphic in the local A. tigris.
Univariate statistics for each character and results of statistical tests, including those on principal component scores and canonical variates, are presented in table 3. Variances were homogeneous across samples for all characters except GS (number of gular scales), so we used analysis of variance (ANOVA) and Tukey HSD post hoc tests to determine whether significant differences in these characters occurred among the three groups. We used a Kruskal-Wallis test (chi-square 5 2.354, 2 df; P 5 0.31) to verify the nonsignificant differences in GS indicated by the ANOVA (F 2, 27 5 1.361; P 5 0.27). These tests indicated that the hybrids differed significantly (P , 0.05) from their maternal parent (A. dixoni C) in the GAB, IFS, and PC2 characters, but they were similar to their maternal parent in the ILS, FP, SDL-T4, and PC1 characters. Each group (A. dixoni C, A. t. punctilinealis, and hybrids) was distinctive in both the CV1 and CV2 characters (table 3).
Three multivariate analyses (PCA, DA, and CVA) were used to determine the patterns of morphological variation between and among individuals and groups. Snout-vent length was excluded from these analyses because our samples did not include the same proportions of various age or size classes.
The first multivariate procedure, a PCA of meristic characters (factoring a correlation matrix), generated scores summarizing the pattern of variation among individuals independent of group membership. An ordination of principal component scores ( fig. 7) depicted the two parental taxa as being separated from each other on PC1, with hybrids resembling their maternal parent, A dixoni, on this axis. Factor loadings indicated that SDL-T4, ILS, FP, and COS were more important in summarizing variation on PC1 than PC 2, whereas GAB, IFS, and GS contributed principally to PC2 (table 4).
For the remaining multivariate procedures, we used step-wise selection of meristic characters (F probabilities of 0.15 for character entry and 0.20 for character removal) to isolate those meristic characters that contributed significantly to the discrimination between and among groups. The model for the DA used functions determined from meristic characters of the parental taxa, with hybrids included as unclassified in the analysis. Three of seven characters were selected as having value in discriminating the two a priori groups (ILS, FP, and SDL-T4, with function coefficients of 0.227, 0.162, and 0.785, respectively). All individuals of A. dixoni C and A. t. punctilinealis were classified to the correct a priori group, tigris having a mean CV1 of 24.394 (range, 26.366 to 22.447) and dixoni having a mean CV1 of 4.732 (range, 2.576 to 7.443). Hybrids resembled their mother, with each of the three hybrids being classified with a probability of 1.0 to the maternal A. dixoni C.
Our CVA incorporated the three hybrids as a third a priori group and a CVA model using five of seven characters (SDL-T4, FP, ILS, GAB, and GS) for distinguishing among the three groups (table 5). Eigenvalues of 27.119 and 1.672 summarized 94.2% and 5.8% of the univariate variation on axes CV1 and CV2, respectively. Each group was meristically distinctive ( fig. 8), and each individual was classified correctly to its a priori group. As also shown by the PCA, the first axis (CV1) separated A. dixoni C and the hybrids from A. t. punctilinealis, but, unlike the PCA, the CVA separated the hybrids from both parents on CV2. This CVA reiterated the meristic resemblance of the hybrids to A. dixoni C, their maternal progenitor tigris, and Their Hybrids Mean 6 SE are subtended by sample size and range limits. * 5 A. tigris is significantly different from the other two groups for this character; ** 5 A. dixoni and hybrids are significantly different from one another for this character; *** 5 hybrids are significantly different from the other groups for this character; **** 5 each group is significantly different from the other groups.
Character a Group A. dixoni Hybrids A. tigris Fig. 7. Pattern of morphological variation expressed by the distribution of scores on the first two principal components extracted from a correlation matrix of seven meristic characters of 30 specimens of Aspidoscelis from Antelope Pass, Hidalgo County, New Mexico. The specimens include representatives of two species (A. dixoni C and A. t. punctilinealis) and hybrids between them. Note that the three hybrids cluster most closely to their maternal parent.  (Mahalanobis D 2 between centroids 5 21.6) rather than to A. t. punctilinealis, their paternal progenitor (D 2 between centroids 5 138.4). The lizard photographs ( fig. 3) and statistical analyses (figs. 7, 8) show that the hybrids of A. dixoni C U 3 A. t. punctilinealis most closely resemble their clonal maternal parent, while differing significantly from their paternal parent. This pattern of hybrids morphologically resembling their clonal parent has been found in nearly all of the studies that have included karyotypic and/or biochemical genetic evidence of hybridization between clonal all-female species of Aspidoscelis and males of bisexual congeners, as follows: A. sonorae U3 A. t. punctilinealis - (Lowe et al., 1970a); A. sonorae U3 A. t. marmorata - (Cole, 1979;Hardy and Cole, 1998); A. laredoensis U 3 A. g. gularis - (Walker et al., 1989(Walker et al., , 1991; A. neotesselata U 3 A. sexlineata viridis - (Walker et al., 1990); A. tesselata U 3 A. s. viridis - ; A. neomexicana U 3 A. tigris - (Dessauer et al., 2000); A. tesselata U 3 A. t. marmorata - (Taylor et al., 2001); and A. neomexicana U 3 A. s. viridis - (Manning et al., 2005). In each of these instances of hybridization, the maternal parent's genomes were cloned, whether consisting of two or three haploid sets of chromosomes, whereas the paternal parent contributed one haploid set of chromosomes in the spermatozoan. Fig. 8. Pattern of morphological distinctiveness expressed by the distribution of canonical variate scores derived from a canonical variate analysis of five meristic characters of 30 specimens of Aspidoscelis from Antelope Pass, Hidalgo County, New Mexico. The specimens include representatives of two species (A. dixoni C and A. t. punctilinealis) and hybrids between them. Note that the three hybrids cluster most closely to their maternal parent.

DISTRIBUTION OF THE PARENTAL SPECIES AND HYBRIDS IN ANTELOPE PASS AND SUGGESTIONS FOR FUTURE MONITORING
For a period of seven years (1987-1993) C.W.P. maintained a survey of the lizards of Antelope Pass, Peloncillo Mountains, Hidalgo County, New Mexico (centered at approximately 10.5 km west of Animas), for the New Mexico Department of Game and Fish (Sias and Painter, 2002). The study focused on A. dixoni C, which is listed as an Endangered Species by the NMDGF. For this survey, grids and arrays of pit-fall traps were established at 31 sites in low-elevation habitats where A. dixoni C was most likely to be found, based on an earlier study by Zweifel (1965). The stimulus for this study was that the population of A. dixoni C was thought by some herpetologists to be extinct in the late 1980s or early 1990s (J. R. Dixon, personal commun.), although during the 1960s they were so abundant that one could go there in the appropriate season and see several individuals on any day with good weather (Zweifel, personal commun., and personal observations, C.J.C.).
Antelope Pass (figs. 2,9), at 1300-1500 m elevation, consists of rocky hillsides with slopes and bajadas covered with gravel. Bottomlands have sandy soil with gravel, and the arroyo bottoms are quite sandy. Overall, the habitat is Chihuahuan Desert desertscrub. Literally in and beside the main arroyo in the bottomland there are a few trees (desert hackberry, Celtis pallida, desert willow, Chilopsis linearis, juniper, Juniperus) and conspicuous shrubs, listed from most to least abundant (estimated), being mesquite (Prosopis glandulosa), Apache plume (Fallugia paradoxa), fourwing saltbush (Atriplex canescens), littleleaf sumac (Rhus microphylla), and some allthorn (Koeberlinea spinosa). The broad expanse of lowland up and away from the arroyo includes, listed from most to least abundant, creosotebush (Larrea tridentata), Prosopis glandulosa, cholla, yucca (Yucca elata), prickly pear (Opuntia), Koeberlinea spinosa, and a few barrel cacti (Ferocactus). ''Dwarf desert-holly (Perezia nana), snakeweed (Gutierrezia sarothrae), and zinnia (Zinnia acerosa) are common perennial herbs. Grass cover is sparse with muhly grass (Muhlenbergia sp.) and tobosa grass (Hilaria mutica) most abundant'' (Sias and Painter, 2002: 1). Plants on the hillsides are more widely scattered than in the lowlands, consisting mostly of Larrea, and here juniper trees are more frequent. Also present, mostly higher up on the rockier soils, are agaves (Agave), small cacti (including Mammillaria), and scattered ocotillos (Fouquieria splendens). Running from south to north are a few narrow, short, rocky-sloped, tree-supporting side canyons that do not have sandy lowlands with dense stands of Larrea and Prosopis. These side canyons, with rocky soils, appear to be unfavorable microhabitat for A. t. puncti-linealis, and one wonders whether they could constitute small refuges for A. dixoni.
During the survey period 1,281 different individuals of A. t. punctilinealis were found, not counting the many recaptures. Only 220 individuals of A. dixoni C were found during this period, as were 3 hybrids of A. dixoni 3 A. t. punctilinealis. Clearly, the two species use the same habitat frequently, as 1,104 of the A. t. punctilinealis were found at only two of the sites (referred to as East and West), and 47 of the A. dixoni were found at the same two sites. However, two other sites (Debbie and Heather, the westernmost sites) produced either 50 or 51 A. t. punctilinealis, respectively, but no A. dixoni (table 6).
There were nine sites that produced from one to nine individuals of A. t. punctilinealis each. Six of these also produced fewer than 10 A. dixoni C each (two produced 0), indicating that marginal habitats that were being avoided were similar for both species. However, the two exceptional sites (Cherry and Hill) produced 55 A. dixoni each. Thus, 110 (50%) of , and yet the occurrence of dixoni at many sites little used by punctilinealis, raises the question of whether dixoni is being displaced and replaced at optimal sites by punctilinealis. Consequently we listed (table 6) and plotted (figs. 10 and 11) the number of individuals found at different trap sites where A. dixoni C and/or A. t. punctilinealis were found. Note that zero to nine individuals of A. t. punctilinealis were found at 15 of the 18 sites (83%) where A. dixoni C occurred. Also, zero to eight individuals of A. dixoni C were found at 10 of the 15 sites (67%) where A. t. punctilinealis occurred.
The three hybrids were found at the following sites: (1) Cherry, on July 22, 1989 (a site where a total of 55 A. dixoni and only 1 A. t. punctilinealis were found); (2) East, on May 7, 1991 (where a total of 29 dixoni and 765 punctilinealis were found); and (3) Baja, on July 5, 1993 (where a total of 11 dixoni and only 1 punctilinealis were found). Interestingly,   (table 6). Triangles indicate sites where A. dixoni C was found, but not A. t. punctilinealis; squares indicate sites where A. t. punctilinealis was found, but not A. dixoni C; and dots indicate sites where both species were found. This was mapped with TOPO! (2003, National Geographic;www.nationalgeographic.com/topo). at two of these sites (Cherry, Baja) A. t. punctilinealis was far outnumbered by the allfemale A. dixoni. Consequently, a male punctilinealis would far more frequently encounter a female dixoni than a conspecific female, which would be conducive to hybridization. The third site (East) had relatively high numbers of both species, but A. t. punctilinealis far outnumbered A. dixoni, so the chances for hybridization (and significant gamete wastage for A. dixoni) would be high, owing to the abundance of males of punctilinealis. We hypothesize that the most secure sites for the long-term survival of A. dixoni without threat of destabilizing hybridization are those sites lacking A. t. punctilinealis. However, if those sites are ecologically marginal for dixoni and they have been displaced to those sites by competitive interactions, this would have a negative impact on the long-term survival of dixoni.
We do not know when A. dixoni arrived in Antelope Pass, assuming it immigrated from elsewhere. The nearest localities today where its ancestral bisexual species occur in sympatry are hundreds of kilometers to the east and southeast in the Chihuahuan Desert. Aspidoscelis t. punctilinealis, a Sonoran Desert taxon, recently arrived from the west (via the San Simon Valley, Cochise County, Arizona), probably during the last few hundreds or thousands of years (Dessauer et al., 2000), and the population in Antelope Pass represents nearly the easternmost geographic limit for this taxon. As discussed elsewhere (Dessauer et al., 2000;Taylor et al., 2001), males of A. tigris are aggressive breeders during the mating season, and they are less discriminating in mate selection than are most other lizards. Consequently, they occasionally mate with females of other species, including parthenogens. Because females of A. dixoni carry an ancestral genome from A. t. marmorata, with which A. t. punctilinealis freely interbreeds in the area of the Peloncillo Mountains (Dessauer et al., 2000), it is not surprising that punctilinealis males mate with dixoni also. The known female hybrids were apparently sterile, however, so such matings constitute gamete wastage.
Today, in the Roswell area of New Mexico in Macho Draw, A. t. marmorata hybridizes frequently with A. tesselata E-C (Taylor et al., 2001), and 50% of the hybrids are sterile females, 50% males. The unisexual tesselata remain abundant, however, possibly because they produce twice as many eggs per clutch than does the bisexual marmorata, and the marmorata are extremely few in number, confined to very limited microhabitat in the area.
In contrast, in Antelope Pass, the bisexual species far outnumbers the parthenogen. We hypothesize that A. dixoni C has been experiencing difficulty maintaining itself in Antelope Pass owing to interspecific competitive interactions and asymmetrical destabilizing hybridization with A. t. punctilinealis, with which it shares the same preferred microhabitat. This hypothesis could be tested by conducting additional trapping in Antelope Pass. Future trapping could be conducted with immediate on-site release, minimizing handling of the lizards (i.e., without transporting, weighing, measuring, or marking in a laboratory). This would little disturb the lizards and minimize the possibility of introducing disease into the population.
Two of the five important sites to monitor for continuing hybridization would be East and West, where 29 dixoni (with 765 punctilinealis) and 18 dixoni (with 339 punctilinealis), respectively, were found (table 6). The other three sites would be Baja, Cherry, and Hill, where only 1 punctilinealis was found along with much larger numbers of dixoni. At the last three sites, males of punctilinealis would have few conspecific females with which to mate. People involved with future monitoring projects must be alert and aware of the similarities of A. dixoni C and the hybrids, lest hybrids be overlooked owing to being misidentified as A. dixoni C.
Considering that the survey found 220 individuals of A. dixoni C and only three hybrids (Sias and Painter, 2002), the impact of destabilizing hybridization to dixoni may not appear to be high. Nevertheless, 1,281 individuals of A. t. punctilinealis were found in broadly overlapping habitats. The low number of hybrids plus the lack of small ones indicate that they are produced neither consistently nor annually, assuming that no hybrids evaded recognition as such. However, the following points may be relevant: (1) We do not know the extent of hybrid mortality in eggs and hatchlings that are not observed; (2) dixoni is declining in abundance; (3) some dixoni occur locally where no tigris or very few tigris occur; (4) dates of collection and sizes of the three hybrids when found indicate that three different egg clutches were involved; and (5) with the capability of sperm storage in lizards (Hardy and Cole, 1981, reported sperm receptacles in the oviducts of Aspidoscelis females, including those of parthenogens), one hybrid mating could have a negative impact on a parthenogenetic female's productivity for two or more seasons. In addition, as male tigris mate with multiple females in a breeding season, the gamete wastage is probably insignificant to A. t. punctilinealis, so this is asymmetrical destabilizing hybridization, having significant negative impact only on A. dixoni C.
There were eight sites in Antelope Pass where A. dixoni C were found and no A. t. punctilinealis were found (table 6). One might infer that these sites hold the greatest promise for the future survival of dixoni, but only 13% of their total number were found at these sites. This could indicate that these sites are of marginal habitat for both species, that dixoni was displaced to these suboptimal sites by the dominant punctilinealis, and therefore these sites are not very promising for the long-term survival of dixoni C. Alternatively, one might infer that those sites where A. dixoni outnumbered A. t. punctilinealis hold great promise for the future of dixoni C. For example, one of the hybrids was found at the Cherry site, where 55 A. dixoni and only 1 punctilinealis were found. However, at such sites, where males of punctilinealis have few conspecific females with which to mate, the frequency of hybridization, and its negative impact on A. dixoni, may be greater than at other sites.
Whatever the future fate may be for the unique clone of A. dixoni C in Antelope Pass, New Mexico, its long-term survival does not appear to be enhanced by the presence of A. t. punctilinealis. Also, considering the small numbers of A. dixoni C in scientific collections that were assembled over a period of decades, its absence from the pet trade, and its nonuse as a game species, the decline in this popula-tion of A. dixoni appears to be a result of natural causes, including desertification, perhaps influenced by effects of human activities for more than a century.

THE POSSIBLE ORIGIN OF A NEW SPECIES
The possibility exists that with continued hybridization before (and whether or not) A. dixoni C becomes extinct, an unusual triploid female hybrid of A. dixoni C 3 A. t. punctilinealis might someday be capable of parthenogenetic cloning, even if the majority of F 1 female hybrids are sterile. Should such a clone appear in the future, it would be a unique self-perpetuating entity in ploidy and combination of genomes from ancestral taxa, with a unique history, and this would constitute a new species. Nevertheless, we wonder whether the long-term survival of such a new species, should it arise, would also be seriously threatened by asymmetrical destabilizing hybridization with A. t. punctilinealis in Antelope Pass. Ironically, the demise of such a species could be brought on by its progenitor. In any event, personnel involved in future monitoring of sites in Antelope Pass should be aware of this possibility as they discover what may appear to be triploid hybrids.
SUMMARY AND CONCLUSIONS 1. Aspidoscelis dixoni pattern class C is highly endangered, known to exist only in a few square kilometers of southwestern New Mexico, approximately 10.5 km west of Animas, Hidalgo County.
2. This is a diploid unisexual species that normally reproduces by parthenogenetic cloning, and the cause(s) of its recent rarity in New Mexico has been unknown. In this report we present the first genetic comparisons of an adult female A. dixoni C and her offspring, which demonstrate clonal inheritance.
3. Additional clones of A. dixoni (pattern classes A and B) occur in the Chinati Mountains, Presidio County, Texas. These are disjunct by nearly 500 km from the New Mexican population, and they also occur in very limited areas.
4. In New Mexico, females of A. dixoni C occasionally interbreed with males of A. tigris punctilinealis, producing triploid hybrids, none of which is known to have reproduced. Considering the initial hybrid origin of A. dixoni itself, these hybrids have genomes from the following three taxa: A. tigris marmorata U; A. gularis septemvittata -; and A. tigris punctilinealis -.
5. Comparative genetics (karyotypes and allozymes) and morphological analyses (univariate and multivariate statistics) clearly demonstrate the ancestry of the hybrids between A. dixoni C U 3 A. t. punctilinealis from New Mexico.
6. In addition, we used the same analyses of allozymes to compare individuals representing the three pattern classes of A. dixoni from New Mexico and Texas.
7. Individuals of Texan A. dixoni A and B were identical to each other in allozymes.
8. However, A. dixoni C from New Mexico is unique in lacking one of the microchromosomes found in Texan populations of A. dixoni A and in having allelic differences at three allozyme loci (sMDH, PEPA, and MPI). Nevertheless, the New Mexican and Texan populations share possession of a fissioned X chromosome that was originally inherited from A. t. marmorata. These populations also share the same individual ancestral F 1 female hybrid (A. t. marmorata U 3 A. g. septemvittata -) with each other and with various pattern classes of A. tesselata-groups considered by some taxonomists to be diverse clones of one species (see discussion in Cordes and Walker, 2006). Whether the color pattern class of the original F 1 hybrid mostly resembled A. dixoni or A. tesselata E or neither remains unknown. 9. Aspidoscelis t. punctilinealis is a recent invader of Antelope Pass, Hidalgo County, New Mexico, yet it is by far the most dominant species of whiptail lizard there today. Its female hybrids with A. dixoni C are apparently sterile and may represent significant gamete wastage for A. dixoni, particularly if females produce multiple clutches of inviable or infertile offspring after mating.
10. The present rarity of A. dixoni C in Antelope Pass may result from negative interactions with A. t. punctilinealis, including competition and asymmetrical destabilizing hybridization (with significant negative impact on dixoni, but not on punctilinealis). 11. We suggest testing this hypothesis by continuing to monitor for A. dixoni and hybrids at any of the following five sites: Baja, Cherry, East, Hill, and West. At these sites, A. t. punctilinealis either far outnumbered A. dixoni C during the last trapping period or was far outnumbered by it, both conditions which may be conducive to hybridization.
12. Personnel in monitoring projects must be trained to recognize hybrid lizards of both sexes, as they can be very difficult to distinguish from parental taxa.

ACKNOWLEDGMENTS
This project could not have been completed without the dedicated and professional efforts of Barney Tomberlin and Tony Snell, who worked feverishly on the A. dixoni project in Antelope Pass, Hidalgo County, New Mexico. Not only did they first propose the correct identification of the very cryptic hybrid lizards of A. dixoni C 3 A. t. punctilinealis, but they untiringly visited the study area no fewer than 1,500 times over the course of seven years, and they individually recorded data on almost 9,000 amphibians and reptiles captured at or near the site. These two people deserve an award! During the A. dixoni project, C.W.P. also received extensive assistance from many colleagues and volunteers who helped maintain and check the pit-fall traps and record original observations. Individuals deserving special mention because of their extraordinary efforts include Tom Moisi, Gerold Merker, and volunteers from the Southwestern Research Station near Portal, Arizona, thanks to its Director at the time, Wade C. Sherbrooke (recently retired; current Director, Dawn S. Wilson). In addition, Dr. Sherbrooke assisted C.J.C. in his field and laboratory work in many ways while he conducted part of this work at the Southwestern Research Station. The late Vincent D. Roth and his gracious wife Barbara provided C.W.P. with laboratory facilities in their home during the early stages of this project.
Mrs. Adeline Hill kindly allowed us to establish trap sites on her ranch and grazing leases for the duration of the A. dixoni project, and Lance Tyson (NMDGF) provided coordinates for plotting sites on figure 11. The U.S. Fish and Wildlife Service, Albuquerque Ecological Services Office, provided a significant portion of the funding for this project through various federal aid grants to the New Mexico Department of Game and Fish Endangered Species Program. Bill Merhege and Mark Hakkila (Mimbres District, Bureau of Land Management, Las Cruces, New Mexico) provided considerable funding and logistical support also. In addition, the New Mexico Department of Game and Fish, Share with Wildlife Program, provided funds and other field support, and the Department granted the scientific collecting permits necessary for this work.
Carol R. Townsend and Margaret G. Arnold provided important care for the hybrid lizards while they lived at the Aspidoscelis Ranch at the American Museum of Natural History. In addition, Carol R. Townsend assisted in all other aspects of this work and Bartek Jablonski helped to analyze the karyotypes.
We also owe special thanks to James E. Cordes, Louisiana State University at Eunice, and his former major professor, James M. Walker, University of Arkansas, Fayetteville, who sent us living representatives of the two pattern classes of A. dixoni from Texas, as well as some tissue samples. Without these specimens, we would not have been able to compare the genetic characters of the disjunct New Mexican and Texan populations. Andrew H. Price (Texas Parks and Wildlife Department) arranged and assisted with collecting important comparative material in Texas, especially of A. g. septemvittata. In addition, George L. Bradley and Philip C. Rosen provided C.J.C. access to chromosome data from A. dixoni at the University of Arizona, which he had prepared and analyzed in 1966 as a graduate assistant to C. H. Lowe in unpublished collaborative research with Lowe, J. W. Wright, and R. L. Bezy. We also thank Richard G. Zweifel and two anonymous reviewers for helpful comments on the manuscript.