A Morphometric Comparison of Gray Treefrogs, Hyla chrysoscelis and H. versicolor, from Ohio

Thirteen morphometric characters were compared between two cryptic species of gray treefrogs from Ohio. Discriminant function analysis using snout-vent length, shank length, and thigh length correctly classified 76.7% of diploid Hyla chrysoscelis and tetraploid//. versicolor. Ohio diploids are more similar in size to those in Texas than to those of Michigan and Wisconsin or southeastern states. Tetraploid treefrogs in Ohio are intermediate in size to those of Texas, North Carolina, and Wisconsin. Latitudinal and longitudinal clinal variation in snout-vent length is apparent in both species. OHIO J. SCI. 90(4):98-101,1990 INTRODUCTION Several investigations have addressed morphometric differences between the two currently recognized species within the North American complex of gray treefrogs. Tetraploid Hyla versicolor averages slightly larger but is otherwise indistinguishable from its diploid progenitor H. chrysoscelis. The two species can be differentiated by measuring nuclear DNA content (Bachman and Bogart 1975), chromosome number (Wasserman 1970), by LDH electrophoresis (Ralin and Selander 1979), and by comparison of the temperature corrected pulse rate of the advertisement call (Zweifel 1970). Johnson (1961) measured three characters on sexually mature individuals and showed that morphometric differences between species exist. Tetraploid male H. versicolor were significantly larger on average than diploid male H. chrysoscelis based upon snout-vent length (SVL) and ratio of tibia length/ femur length. However, overlap in size range between species precluded their differentiation based upon these characters. Ralin (1968) also found overlap between species in SVL, and could not use SVL to separate species in Texas. Jaslow and Vogt (1977) noted a statistically significant difference in SVL between diploids and tetraploids in Wisconsin. However, Little (1983) did not detect differences using discriminant function comparisons of nine "size-free" morphological variables between West Virginia and south-central Ohio populations of H. chrysoscelis and H. versicolor. Ralin and Rogers (1979) used discriminant function analysis of 13 morphometric characters on individuals from 12 populations of H. chrysoscelis and H. versicolor to separate eastern diploid populations (eastern Texas, South Carolina, Georgia, Ohio, and Mississippi) from western diploid populations (central Texas). Southern tetraploid populations (Texas) were intermediate between eastern and western diploid populations but were more similar to eastern diploids. Ralin and Rogers (1979) suggested that populations of H. chrysoscelis in the Midwest may be intergrade populations between eastern and western diploids and that Appalachian and northern midwestern populations of H. versicolor may resemble tetraploid 'Manuscript received 20 November 1989 and in revised form 29 March 1990 (#89-29). populations from Texas (southern H. versicolor) more than those in the Northeast (New York, New Jersey). The purposes of the present research were to determine the degree of accuracy with which gray treefrogs in Ohio could be classified using morphometric characters, and to compare Ohio material with that available from other portions of the distributional range of the complex. MATERIALS AND METHODS Collections of 72 diploid and 100 tetraploid male treefrogs were made between May and July 1985 to 1987 from 13 localities in nine counties of Ohio. Collecting localities approximated a transect extending from Ashtabula County in the northeast to Clermont and Preble Counties in the southwest (Matson 1988). Calling males were identified by sound spectrographic analysis of the temperature corrected pulse rate of the advertisement call. Thirteen external morphological measurements were made on each preserved treefrog, and each measurement was made three times on different days in an attempt to minimize measuring errors. Characters selected are similar to those described by Ralin and Rogers (1979) and were measured on the right side. They are defined as follows: I) snout-vent length (SVL), distance from tip of snout to posteriodorsal edge of the cloaca; 2) thigh length, distance from lateral edge of vent to knee; 3) shank length, the distance from knee to articulation of tibio-fibula with fibulare; 4) foot length, distance from articulation of tibiofibula with fibulare to tip of the fourth (longest) toe; 5) upper arm length, distance from articulation of humerus with pectoral girdle to the elbow; 6) forearm length, distance from elbow to articulation of radio-ulna with the carpals; 7) hand length, distance from articulation of radio-ulna with carpals to tip of the third (longest) finger; 8) toepad width, width of toepad on the third finger; 9) head length, distance from tip of snout to posterior edge of rim surrounding the tympanum; 10) eye-to-nostril length, distance from naris to anterior corner of the eye; II) nostril-to-lip length, vertical distance from nostril to edge of the upper lip; 12) tympanum diameter, horizontal width of tympanum; and 13) head width, widest measurement of the head, usually distance between corners of the mouth. Measurements 8, 11 and 12 were measured to the nearest 0.01 mm with an ocular micrometer; all other OHIO JOURNAL OF SCIENCE T. O. MATSON 99 measurements were made to the nearest 0.1 mm with a dial caliper. Mean values of the three data sets were used in all analyses. The Student's /-statistic was used to compare means for each character between species. Direct and Mahalanobis methods of discriminant function analysis (DFA) were used to derive an unstandardized canonical discriminant classification function. Normality tests for skewness and kurtosis were significant for three characters in H. versicolor and for two different characters in H. chrysoscelis; however, skewness was detected only in shank length in H. chrysoscelis among characters used in the direct discriminant function presented here. Skewness of the shank measurement was considered to not adversely affect the homogeneity of variance assumption of DFA because the level of significance was only slightly greater than 0.05 (P > 0.05). DFA was run on an IBM 3081D mainframe computer using the program Discriminant by SPSS, Inc. (1983). The F to enter and F to remove equalled 1.7, the approximate F-value required for significance with P = 0.01. Specimens were deposited in the herpetology collection of The Cleveland Museum of Natural History. RESULTS Differences between species were small for the characters measured, and comparison of means between species using the Student's /-statistic was significant (P < 0.05) for hand length, head width, nostril-to-lip, and SVL only (Table 1). Few females of either species were obtained in the field (n = 2 H. chrysoscelis; n = 5 H. versicolor); they averaged larger than males in both species but were few in number and were excluded from DFA.