Cnemidophorus tesselatus (Neaves, 1971)

TAYLOR, HARRY L., COLE, CHARLES J., HARDY, LAURENCE M., DESSAUER, HERBERT C., TOWNSEND, CAROL R., WALKER, JAMES M. & CORDES, JAMES E., 2001, Natural Hybridization Between the Teiid Lizards Cnemidophorus tesselatus (Parthenogenetic) and C. tigris marmoratus (Bisexual): Assessment of Evolutionary Alternatives, American Museum Novitates 3345, pp. 1-1 : 1-

CNEMIDOPHORUS TESSELATUS

This is the largest of the four species at Arroyo del Macho; snout–vent length = 88.6 ± 1.66, 59–109 (40). Zweifel (1965) provid­ ed the original descriptions of color pattern variation among populations presently assigned to pattern class E. This variation involved the degree to which the lateral stripes (a pair of pale stripes extending from dorsal edge of the ear opening to anterodorsal aspect of the thigh) were disrupted by vertical black bars. The lateral stripes were absent in some populations while other populations had a predominance of individuals with relatively intact lateral stripes, resembling pat­ tern class C in this respect. This variation within pattern class E (as presently con­

ceived) has been dealt with, consistently and conveniently, by using Zweifel’s (1965) depiction of the geographic distributions of col­ or pattern classes, rather than actual color pattern characteristics, to assign certain individuals and populations to pattern class. As an example of the problem, refer to Dessauer and Cole (1989: fig. 1A, 1B ). Although they referred the individual in their figure 1A to pattern class C, this specimen would be identified as a representative of pattern class E (with relatively intact lateral stripes) based on its sampling locality in the northern Rio Grande drainage (Zweifel, 1965: fig. 1 ). This problem is currently under study.

The Arroyo del Macho population of C. tesselatus is distinguished from the other three sympatric species of Cnemidophorus by a sharply contrasting dorsal pattern of pale stripes and their transverse expansions on black dark fields ( figs. 2A , 4 ). One color pattern element, evidence of lateral stripes, can be used to distinguish members of C. tesselatus from hybrids at this locality. Ontogenetic development can reduce definition of the lateral stripes in two ways. Fundamentally, dark fields subtending the lateral stripes can become subdivided into linear series of dark­field remnants through localized loss of melanin; this produces a series of transverse pale bars of the same hue as the stripes. Because pale bars and dark­field remnants can be in register above and below the stripes, there is loss of stripe definition where pale bars pass through the stripes. Similarly, stripes are disrupted by black bars where localized production of melanin in the pale stripes causes fusion of dark­field remnants above and below the stripes. However, the disruption of lateral stripes is reduced in the Arroyo del Macho population, and evidence of lateral stripes persists in all individuals of this population. Individual variation in dorsal color pattern is particularly evident in the vertebral dark field ( fig. 4 ). A vertebral stripe is usually evident on the nape, extending posteriorly for variable distances before fragmenting into a series of pale elements. Lateral expansions of the vertebral stripe or its fragmented elements establish a sequence of contacts with the paravertebral stripes. These expansions can be highly exaggerated in the final color pattern as a longitudinal series of

transverse pale bars connecting the paravertebral stripes to form a ladderlike pattern. The ventral surfaces are immaculate or nearly so ( fig. 3A ); i.e., there are no black spots in the central gular region, although there may be one or two spots laterally. A few small black spots (often confined to single scales) may be scattered on the chest.

CNEMIDOPHORUS TIGRIS MARMORATUS There is sexual dimorphism in body size, with males being significantly larger than females in our sample (t 29 = 2.469, P = 0.02); snout–vent length of males = 80.7 ± 1.81, 63–92 (21); snout–vent length of females = 73.3 ± 2.24, 57–80 (10). Individual variation in this subspecies can be observed in the relative prominence of pale stripes or their remnants in the dorsal pattern, ranging from clearly evident stripes to a loss of stripe identity in an irregular reticulum of pale and dark sectors. All stripes may be fragmented to various degrees. The double nature of the pale vertebral stripe (or its remnants) is usually evident, thus resembling its hybrid derivatives, pattern classes Colorado D and New Mexico D of C. tesselatus , in this feature. Pale stripes, or significant segments thereof, are retained in the dorsal pattern of many adult individuals of C. tigris marmoratus in the Pecos River drainage, and these populations were described as C. marmoratus reticuloriens by Hendricks and Dixon (1986). All individuals in our samples lack lateral stripes ( fig. 2C ). Pastel hues embellish the ventral pigmentation, with pink shades added to the gray throat and chest, shifting to orange or tan shades on the ventrolateral areas of the abdomen. Individuals larger than 78 mm snout–vent length have black spots and blotches on the gular and chest regions ( fig. 3C ), with considerable variation in their size and number. Although C. tigris marmoratus was uncommon at the hybridization site, it was a strong candidate for the paternal parent of the hybrids because of two shared color pattern features (see below).

HYBRIDS

Males and females had similar snout–vent lengths (t 18 = 0.194, P = 0.85): males = 89.2

± 3.15, 56–97 (12) and females = 88.4 ± 2.85, 71–97 (8). The dorsal color pattern ( fig.

2B) is most similar to C. tesselatus ( fig. 2A ), but the hybrid pattern tends to be coarser, with larger dark and pale sectors (compare figs. 4–6 ). There is no vertebral stripe, and, from the shoulder region posteriorly, the vertebral dark field is either rather open or bridged by transverse pale elements. Ventrally, hybrids are intermediate to parental species in degree of black­spotting ( fig. 3B ). As currently understood (Lowe et al., 1970a; Cuellar and McKinney, 1976; Taylor et al., 1989; Walker et al., 1989a, 1989b), a lizard with male reproductive structures that has a color pattern resembling a normal parthenogenetic species is either a triploid or tetraploid hybrid—cytological states that are attained only through fertilization of unreduced eggs from diploid or triploid parthenogens by haploid, Y­bearing sperm from males of bisexual species.

COLOR PATTERN AND IDENTIFICATION OF HYBRIDS

Although there is color pattern variation among the hybrids ( figs. 5 , 6), all bear a strong resemblance to their maternal parent, C. tesselatus ( fig. 4 ). There are, however, several color pattern features that distinguish hybrids from C. tesselatus and implicate C. tigris marmoratus as their paternal parent. The most obvious difference is dichotomous—lateral stripes (or obvious remnants) are present in the Arroyo del Macho population of C. tesselatus and absent in hybrids and C. tigris marmoratus ( fig. 2 ). In addition, some hybrids (particularly males) have subtle suffusions of pink on the throat and chest and yellowish­tan on the lateral abdominal surfaces. These are pigmentation features of C. tigris marmoratus . Certain hybrids also resemble C. tigris marmoratus in having small black spots in the central gular region and on the chest. However, the intensity of pastel hues and the size and number of black ventral spots are reduced in such hybrids compared to C. tigris marmoratus ( figs. 2 , 3 , 5 , 6).

As a group, hybrids differ from C. tesselatus in having a greater number of interruptions of the dorsolateral stripes (IDLS; t 26.747 = 5.438, P <0.0005; table 1) and paravertebral stripes (IPVS; t 58 = 5.730, P <0.0005; table 1). The presence of a stripe or smaller pale segments in the vertebral dark field of C. tesselatus ( fig. 4 ) and a more open vertebral dark field in hybrids ( figs. 5 , 6) will distinguish many individual hybrids from C. tesselatus . However, a quantification of this feature (PSVF, number of midsagittal pale segments in the vertebral dark field) did not

distinguish the two groups (t 55 = 1.754, P = 0.08; table 1).

UNIVARIATE MORPHOLOGICAL COMPARISONS

Sexual dimorphism was lacking in meristic characters of the hybrids (all P s> 0.09) and representatives of C. tigris marmoratus (all P s> 0.14) from the vicinity of Arroyo del Macho, permitting us to pool the sexes into single samples for each group. For seven of the eight scalation characters, there was one significant difference among the pairwise comparisons of hybrids, C. tesselatus , and C. tigris marmoratus (table 1). Hybrids differed significantly from both parental species in number of granules around midbody and in subdigital finger lamellae. However, the pattern of nonsignificant differences (resemblance) between hybrids and parental species was also informative. All 20 hybrids from Arroyo del Macho had abruptly enlarged mesoptychial scales (those bordering the gular fold anteriorly), thereby resembling C. tesselatus rather than C. tigris marmoratus in this character. Hybrids also resembled C. tesselatus in number of subdigital toe lamellae and gular scales, but resembled C. tigris marmoratus in number of femoral pores, circumorbital scales, and lateral supraocular granules (table 1). Cnemidophorus tesselatus , itself a product of past hybridization, also exhibited a mosaic pattern, resembling its maternal progenitor ( C. tigris marmoratus ) in some characters and its paternal progenitor ( C. gularis septemvittatus ) in others (Parker, 1979b).

MULTIVARIATE MORPHOLOGICAL ANALYSES

Principal Components Analysis

The first three principal components summarized 64.1% of the variation found in eight meristic characters used in the PCA of C. tesselatus , C. tigris marmoratus , and their hybrids. Significant differences in principal

components scores (treating these variables as univariate characters) among the three groups are shown in table 1. A comparison of component loadings (table 2) to characters with nonsignificant differences between hybrids and one of the two parental groups (table 1) indicated that principal component 1 was expressing similarities of hybrids to C. tigris marmoratus in number of femoral pores, circumorbital scales, and lateral supraocular granules. Principal component 2 expressed similarities of hybrids to C. tesselatus in number of subdigital toe lamellae and gular scales.

The pattern of variation was depicted in an ordination of principal components scores on the first two axes ( fig. 7 ). There was considerable overlap in principal components scores of all three groups, presumably reflecting shared parentage; i.e., that C. tigris marmoratus was the maternal parental species of C. tesselatus and the paternal parental species of the hybrids from Arroyo del Macho. The third principal component reflected the distinctively low numbers of granules around midbody in hybrids compared to both parental species (table 1), and it effectively separated hybrids from parental species on this axis.

Phylogenetically (because C. tesselatus is also a hybrid derivative of C. tigris marmoratus ), the hybrids at Arroyo del Macho have approximately 67% of their genes from C. tigris marmoratus and approximately 33% of their genes from C. gularis septemvittatus . Proximately, and more important in terms of phenotypic expression, the triploid hybrids have 100% of the genes of C. tesselatus and 50% of the genes of C. tigris marmoratus ; the matriclinous nature of the multivariate expression of meristic characters is revealed by the canonical variate analysis that follows.

TABLE 1 Descriptive Statistics of Morphological Characters and Tests of Significancea Among the Parthenogenetic

Species Cnemidophorus tesselatus , the Bisexual Species C. tigris marmoratus , and Their Hybrids

CANONICAL VARIATE ANALYSES

Each specimen was assigned to its correct a priori group in a quadratic canonical variate analysis of eight meristic characters, and hybrids were significantly different from both parental species in both canonical variates (table 1). The separation of the three groups on discriminant axes is depicted in figure 8 . We ran a second canonical variate analysis (linear) with three characters excluded from the model (number of circumorbital scales, number of lateral supraocular granules, and number of gular scales) in order to achieve homogeneous covariance matrices (P = 0.06). This analysis provided squared Mahalanobis distances (D 2) among centroids of the three a priori groups for quantifying the degree of multivariate resemblance among

TABLE 2 Factor Loadings for Three Principal Components

Derived from Meristic Variation Among the Parthenogenetic Species Cnemidophorus tesselatus , the Bisexual Species C. tigris marmoratus , and Their Hybrids

hybrids and parental species. The hybrid group resembled maternal C. tesselatus more closely than paternal C. tigris marmoratus — D 2 values were 11.0 between hybrids and C. tesselatus and 19.9 between hybrids and C. tigris marmoratus . In comparison, the D 2 value between C. tesselatus and C. tigris marmoratus was 18.9. Because C. tesselatus has one set of chromosomes from C. tigris marmoratus and the hybrids have two sets of chromosomes from C. tigris marmoratus , one might assume that the resemblance to C. tigris marmoratus would increase (i.e., D 2 values would decrease) with each additional set of chromosomes inherited from this taxon. However, the distances of 18.9 (one set of C. tigris marmoratus chromosomes in C. tesselatus ) and 19.9 (two sets of C. tigris marmoratus chromosomes in the hybrids) are evidence of an absence of significant additive interactions among the tigris genomes and substantiate the matriclinous nature of phenotypic expression in the hybrids; i.e., there is a disproportionate resemblance of hybrids to the female parent (see also Parker, 1979b).

The overall dorsal color pattern of the Arroyo del Macho hybrids most closely resembles that of the maternal parent, C. tesselatus . As an example of the unpredictability that seems to be associated with the genus Cnemidophorus , there is an example where the opposite is true. Hybrids between the diploid

Figures

Fig. 1. A. A view to the east showing the principal gypsum outcrop and south­facing slope at the hybridization locality north of Roswell, Chaves County, New Mexico. Scattered dark shrubs in the center of the figure are mesquite (Prosopis juliflora) in a desert grassland; C. exsanguis and C. inornatus were

Fig. 2. Dorsolateral views of three representative whiptail lizards (Cnemidophorus) from Arroyo del Macho. A. Diploid unisexual C. tesselatus, AMNH R­146638, body length 97 mm. B. Triploid C. tesselatus X C. tigris marmoratus hybrid male, AMNH R­146693, body length 100 mm. C. Diploid bisexual C. tigris marmoratus male, AMNH R­146653, body length 94 mm. All photographed June 19, 1998.

Fig. 4. Representatives of Cnemidophorus tesselatus from the Arroyo del Macho hybridization site. A. AMNH R­146615 (91 mm SVL); B. AMNH R­146618 (83 mm SVL); C. AMNH R­146619 (85 mm SVL); D. AMNH R­146623 (86 mm SVL); E. AMNH R­146627 (89 mm SVL); F. AMNH R­ 146626 (90 mm SVL).

Fig. 3. Ventral view of the same lizards arranged in the same sequence as in figure 2.

Fig. 5. Representative Cnemidophorus tesselatus X C. tigris marmoratus hybrid females from the Arroyo del Macho hybridization site. A. AMNH R­146689 (89 mm SVL); B and E. AMNH R­146687 (93 mm SVL); C and F. AMNH R­146688 (91 mm SVL); D. AMNH R­146681 (84 mm SVL).

Fig. 7. Pattern of morphological variation expressed by the distribution of scores on the first two principal components extracted from a correlation matrix of eight meristic characters. Symbols depict scores of 84 individuals collected in the vicinity of Arroyo del Macho, Chaves Co., New Mexico: O, 38 C. tesselatus; Z, 26 C. tigris marmoratus; ·, 20 hybrids. Ellipses define the 95% confidence limits of each group.

Fig. 8. Pattern of morphological distinctiveness expressed by the distribution of canonical variate scores derived from a quadratic canonical variate analysis of eight meristic characters. Symbols depict 84 individuals collected in the vicinity of Arroyo del Macho, Chaves Co., New Mexico: O, 38 C. tesselatus; Z, 26 C. tigris marmoratus; ·, 20 hybrids. Ellipses define the 95% confidence limits of each group.