Speciation Within the Chorus Frog Pseudacris Triseriata: Morphometric and Mating Call Analyses of the Boreal and Western Subspecies

Home , Boreal chorus frog, Chorus frog, Western chorus frog

Copeia, 1989(3), pp. 704-712

Speciation within the Chorus Frog Pseudacris triseriata: Morphometric and Mating Call Analyses of the Boreal and Western Subspecies

JAMES E. PLATZ

Data from mating call and morphological analyses of the boreal and western subspecies of the chorus frog Pseudacris triseriata were obtained along a transect from South Dakota to Oklahoma. Two call types are evident. At the northern end of the transect calls are long and low in pulse rate. Both traits had low variability relative to populations further south. The southern end of the transect contained populations with a much shorter call and pulse rates approximately twice those in the north. Identical long and short call types are evident in Col- orado as well. Limited data from Oklahoma indicates that the call of P. triseriata feriarum is more like that of P. t. maculata. Multivariate stepwise discriminant analyses performed on data sets containing only morphological measurements produced two clusters of populations which correspond on geographic grounds to two currently recognized subspecies, P. t. maculata and P. t. triseriata. When call data are included, the same two clusters are identified suggesting concor- dance between call type and morphology. The data indicate that the two entities represent separate species and that the zone of overlap is much more extensive than previously depicted.

FSEUDACRIS triseriata (Wied) is one of the extending north along the Atlantic as far as 1 smallest North American hylids. Accord- New Jersey. Pseudacris t. kalmi, the New Jersey ing to Tordoff and Pettus (1977), some live Chorusin Frog, is restricted in range, occurring excess of 6 yr. In eastern Colorado, Spencer in parts of Delaware, Maryland, southern New (1964) found it to be an anuran of limited va-Jersey, New York and eastern Pennsylvania. Ac- gility with few individuals moving more than cording to Conant (1975) it intergrades with P. 750 m from the hatch site. It is also the most t.feriarum in eastern Pennsylvania and at Staten widespread species of chorus frog in North Island, New York. America, occurring from the Gulf of Mexico to The criteria used to distinguish these four New York and southern Ontario in the east, subspecies hinge, to a large degree, on variable and from parts of Texas and Arizona in the dorsal skin patterns. The boreal form has short- southwest, north to Alberta and the Northwest er legs and often a dorsal pattern of spots in Territory in Canada. three longitudinal rows distinguishing it from Conant (1975) recognized four subspecies of the Western form which has longer legs and P. triseriata. Three of these have extensive three well defined dorsal stripes. The Upland ranges. The Boreal Chorus Frog, P. triseriata Chorus Frog has thin broken stripes differen- maculata, occupies the northwestern limits tiating of it from the New Jersey Chorus Frog, the range: from northwestern Ontario, east which to has broad well defined stripes. Winnipeg in Manitoba, and south to New Mex- Although the present nomenclature has re- ico, and east including parts of Nebraska, South mained stable for the past 25 yr, the literature Dakota, Minnesota and Wisconsin. Pseudacris reveals t. an earlier history of repeated revision. triseriata, the Western Chorus Frog, replaces Since the original description of Hyla triseriata maculata in the south resulting in extensive sym-(now P. t. triseriata) by Wied (1839), four dif- patry in a broad arc extending through most ferent of generic names have been applied, as have Nebraska, eastern South Dakota, central Min- numerous specific epithets for this and other nesota and northern Wisconsin and Michigan. subspecies now recognized as P. triseriata. For Along its southern limit, the western formdetailed reviews see Smith (1956), Harper overlaps with and is then replaced by the (1955),Up- Schwartz (1957) and Cook (1964). land Chorus Frog, Pseudacris triseriataferiarum: The first serious efforts to understand geo- common over most of the Gulf Coast states, graphic variation within P. triseriata were those

? 1989 by the American Society of Ichthyologists and Herpetologists PLATZ-CHORUS FROG SPECIATION 705 of Smith and Smith (1952) and Smith lation variation. (1956). To provide such demic infor- At that time boreal, western and upland mation, formsI have conducted preliminary studies were considered subspecies of P. nigrita. over three Smith breeding seasons. The results indi- (1956) rediagnosed the Boreal Chorus cate that Frog the taxonomic (P. status of P. triseriata is nigrita septentrionalis) as P. n. maculata. not fully Based resolved. on differences in: 1) breeding habitat; 2) the distinct nature of mating calls; and 3) the lack MATERIALS of AND METHODS intergradation among sympatric populations of P. n. nigrita and P. n. feriarum, Schwartz Call analysis.-In (1957) order to assess geographic recommended specific status for P. patterns n. feriarum for the boreal and western subspecies, as P. triseriata feriarum. He further 221 suggestedadult males representing 20 localities were that P. n. triseriata and P. n. maculata should be collected. Specific localities are given in Figure treated as subspecies of P. triseriata (i.e., P. t. 1. At the time of collecting, mating calls were triseriata and P. t. maculata, respectively). This obtained for 259 adults using Sony model TC- recommendation is based on evidence inter- 150 cassette recorders and microphones. Water preted as intergradation between P. t. feriarum temperature was recorded 2 cm posterior to an and P. t. triseriata reported by Smith (1956).individual as it called. Temperatures were de- Conant (1975) followed Schwartz (1957). termined For to the nearest 0.1 C using a Bailey convenience each of the four currently recog-BAT-12 fast reading telethermometer. All re- nized subspecies of P. triseriata will be referred cordings were made in night choruses during to in this paper as maculata, triseriata, feriarum, the breeding season (March-June). All calling or kalmi. individuals at a given locality were within my Smith's (1956) comparison of maculata hearing and distance. Each recording generally in- triseriata is critical to our current knowledge cluded of a minimum of one additional calling in- Pseudacris. It represents the only published dividual,study usually calling in a tandem mode. Af- of morphological variation, based on terlarge calls were recorded, each individual was cap- numbers of specimens collected over atured wide and placed in a labeled plastic bag coded geographic area. His conclusions rely onagainst ex- the taped calls and frozen in liquid ni- amination of over 800 individuals, including trogen 403 or placed on dry ice for later removal maculata from 112 localities in Canada and the of tissue samples. Specimens were subsequently United States, and 343 triseriata from 62 local- preserved in formalin and then transferred to ities (Smith and Smith, 1952) as well as 60%125 alcohol solution. presumed intergrades from 42 locations. Based A Kay Electrometrics Model 6061-B Sono- on his examination and measurements of these graph was used to generate sonograms. Three specimens, he differentiated present day macu- calls for each male were analyzed for dominant lata from triseriata in more specific terms frequencythan (DF), pulse number (PN), call dura- Conant (1975): maculata "differs by the tionpro- (CD), and pulse rate (PR). Because CD and portionately shorter tibia, femur, and foot; PR rel- are temperature dependent, recordings from atively shorter and narrower head; proportion- a test population composed of five adult male ately heavier body; and by the higher frequency maculata from an isolated pond (1 km south of of specimens with a spotted or mottled Glenpat- Cunningham Lake, Douglas Co., Nebras- tern." Examination of his table I indicates 343 ka) were made over a several week period in maculata and 430 triseriata were used to deter- order to obtain calls from a known group for a mine tibia-body length ratios. The resulting range of temperature values (10.5-17.7 C). mean values, given as percent of body length, These data (n = 39) were then used to deter- were 39.3 and 42.6, respectively. Computation mine the relationship between temperature and of confidence limits for the means (using his CD. The least squares regression was employed published SE of the mean times the appropriate and the resulting regression equation (Y = t value) indicates that the means are statistically -0.0974X + 2.277) was used to generate a different (P < 0.05). However, in this particular function, CD,, = CDamb. - (Tmb. - 14) comparison involving over 170 localities (sam- (-0.0974), to permit conversion of duration of ple sizes were not given), mean values and all SE subsequent calls to 14 C. Here "CD" refers were obtained by pooling all maculata vs all trise- to call duration in seconds and the subscript riata. This method resulted in very small "amb." SE refers to the temperature (T) of a re- values but precludes assessment of inter-popu- corded call or the duration (CD) from the son- 706 COPEIA, 1989, NO. 3

Fig. 1. Transect populations of Pseudacris triseriata maculata and P. t. triseriata. Populations connected by dashed lines represent P. t. triseriata. All others, except population 17, represent P. t. maculata. Sites collected are as follows: South Dakota: (1) Lake Oahe, Hwy. 1804, Potter Co., (2) Marsh Lake, Hamlin Co., (4) Springfield, Bon Homme Co. Iowa: (3) Lake Okoboji, Dickinson Co., (6) Wilson Island State Park, Potta- wattamie Co. Nebraska: (5) Beemer, Cuming Co., (7) Dodge Park, Omaha, Douglas Co., (8) 120th and Fort St., Omaha, Douglas Co., (9) Waterloo, Douglas Co., (10) Ogallala, Keith Co., (11) Cozad, Dawson Co. Kansas: (12) Concordia, Cloud Co., (13) Lawrence north, Douglas Co., (14) Lawrence south, Douglas Co., (15) Arkansas City, Cowley Co., (16) Hwy. 57, 15 km E St. Paul, Cherokee Co. Colorado: (18) Ft. Collins, Larimer Co., (19) 11 km W Fairplay, Park Co., (20) 3 km W Aroya, Lincoln Co. Population (17) represents P. t. feriarum from 2.5 km E Wagoner, Wagoner Co., Oklahoma.

ogram at the recorded temperature. Total PN was recorded to the nearest 0.1 mm. Measure- was counted from each sonogram. DF was mea- ments were read with the aid of a 3 x illumi- sured from the bottom of the base line to the nated magnifying lens. These included: body mid-point of the fundamental frequency band length (from the tip of the snout to the base of in kHz. CD also was measured directly from the the urostyle), head width (to the posterior mar- sonogram in mm and converted to seconds. PR gin of the tympanum), lower leg (tibia-fibula) was obtained by dividing the total number of length, head length (from tip of snout to the pulses per call by the projected duration of the rear angle of the jaw), snout depth (from ex- call (corrected to 14 C) in seconds. ternal naris to margin of upper lip), tympanum diameter, length of foot, and upper leg (femur) Morphometrics.-Morphometric measurements length. were made using needle point dial calipers. Each Multivariate stepwise discriminant analyses body measurement was made to the nearest 0.05 (Dixon, 1981) were performed on two data sets. mm with the exception of body length which The first contained morphometric measure- PLATZ-CHORUS FROG SPECIATION 707

kHz @ -i 19

32 ) - 1 , 9ia ii-,111001 ) . . i,tIttl))))l W i 2 , I ,I ( * 5 (feriarum) U X.. r . r ... 0 0,2 0.4 0.6 0.8 1,C Sec. ) ' 1e 14 triseriata 6 Fig. 2. Sonogram of male Pseudacris triseriata mac- macu 8t ulata mating calls. Two trimales seriata macu(animals lata 1 and 2) are calling in tandem. Recorded at Lake Oahe, Hwy. 1804,

Potter Co., South Dakota I wX on 29 May 1981. Water temperature was 23.3 0.2 0.4C. 0.6 0.8 1.0 1.2 1.4

Fig. 3. Call duration (corrected to 14 C) expressed ments for 221 adult in seconds. male specimens.Each horizontal The line sec-represents the range of values. Vertical lines represent means. Open boxes ond set contained morphometric as well as tem- represent ? 1 SD and a solid bar the 95% confidence perature corrected limits call for data each for mean. 158 The malescircled numbersfor to the left which data of both identify types populations were available. in accord with the locations given in Figure 1. Population sample sizes are given to the RESULTS right.

Call analysis.-Sonograms of mating calls (Fig. 2) representing 20 populations were measured elevation populations (18 and 19) from Colo- to determine the value of each of the parame- rado. Population 20 provides striking confir- ters (PN, DF, CD, PR). The mean, SD and 95% mation in Colorado of the same long and short confidence limits were obtained and plotted forcall types described among populations 1-16. each population in a north-south series. Mean PR were obtained by dividing the number of PN varies from 17.4 to 14.3 with no apparent pulses per call by the duration in seconds and clinal trend (r = 0.06, F .18 = 0.25, 0.9 > P are > therefore temperature dependent and in- 0.5), nor are there any apparent geographic pat-versely related to duration. In the four most terns correlated with subspecies designation. northernDF populations (1-4) the mean PR range varies clinally (r = 0.54, F,,18 = 7.13; P < 0.05) from 16.4 pulses/sec downward to 12.7 pulses/ with higher mean frequencies in the northern sec at Springfield, South Dakota. The mean of populations. Means range from 3.58 kHz to 3.28 the next locale (pop. 5), Beemer, Nebraska ex- kHz in South Dakota (pops. 1-2, and 4) to 2.91- ceeds twice that of population 4. Populations 3.11 kHz in Kansas at the southern end of the 6-9 have mean values similar to those of the transect (pops. 15-16). first four. Mean PR in population 10 (23.7 Both the mean CD (Fig. 3) and mean PR pulses/sec)(Fig. is again much higher. Populations 4) yielded patterns suggesting two call types. 11-16 tend to repeat this "low-high" pattern The four most northern sites (pops. 1-4) formwith the exception of population 11 which is a series possessing long calls ranging from intermediate0.91- between the two extremes. The 1.16 sec in length. One hundred km south, three at Colorado populations (18-20) again con- Beemer, Nebraska (pop. 5), the mean call length firm the existence of two call types; a low PR is 0.59 sec. Populations 6-9 have long calls, (pop. pop- 20) at Aroya and a much faster call at Ft. ulation 10 a short call. Population 11 is inter- Collins (pop. 18) and Fairplay (pop. 19). mediate between 10 and 12 which again isCall a data obtained for population 17 at Wag- long call. The mean values for duration in oner, pop- Oklahoma were not typical of the others. ulations 13-16 are short as are the two lower Although mean CD is within the range of tris- 708 COPEIA, 1989, NO. 3

K2 (N= 221) 2 2

11 , \, 1 10 ( '11 16 .. O' \ 1 ( ,I, 9 19 13 ''? 0*

? -' () e-- 1 2 o ---o ^ 12 -1 15 5 19 () I@ 10 -2 9 'i P. triseriata 14

? ) I- 0 8 -3 {-r~ 27 8 P. maculata

? I 13 I I . . I I . , K1 - - ' 11 -4 -3 -2 -1 0 1 2 3 4 5 ) @) I= -2 22 ( ' 5 (feriarum) Fig. 5. Canonical plot from multivariate stepwise discriminant analysis (BMDP7M, Dixon, 1981) based .-) ? I - @ - t 14 I ~~~( I 6, on the eight body measurements for 221 adult male Pseudacris triseriata. Both the K, and K2 axes are marked in units of 1 SD. The small circles represent popu- maculata triseriata lation means. The numbers refer to localities identified in the legend for Figure 1. Dotted and dashed lines connect the means of two clusters, each one

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 representing populations of a given species.

Fig. 4. PR (corrected to 14 C) in pulses/sec. Rep- resentation of the mean, range, standard deviation, seriata specimens have wider heads, longer legs, 95% confidence limits, localities and sample sizes are the same as Figure 3. and larger tympani than do maculata individuals. triseriata populations, PR is very low, as is char- Combined morphometric and acoustic analysis.- acteristic of maculata. Multivariate discriminant analysis was also performed on populations for which both mor- Morphometrics. -Stepwise discriminant analysis phological and acoustic data were available and (BMDP7M; Dixon, 1981) indicated that two axes included CD, PN and DF. PR was not included (K1 and K2; Fig. 5) explained 92% of the total in the analysis reported in Table 1 because it is dispersion (72% and 20%, respectively). Sepa- derived from PN divided by CD and therefore ration along the horizontal axis (Table 1) notis an independent parameter. When PR was heavily dependent on differences between the included the canonical plot is much the same as two groups in tympanum diameter, head width, the one in Figure 6, but with somewhat tighter and tibia-fibula length. Dispersion along the K2 within group clustering. axis is largely influenced by head width, tibia- The canonical plot (Fig. 6) shows tighter fibula length, and to a lesser extent by tympa- grouping within and greater separation be- num diameter and body length. In the canonical tween the two clusters than the morphological plot of the results (Fig. 5) two clusters of pop- data treated alone (Fig. 5). The K1 and K2 axes ulations within the transect are apparent. The (Table 1) account for 82% (K1 = 68%; K2 = group on the left (pop. 5, 10, 13, 15, and 16)14%) of the dispersion. Snout-urostyle distance correspond geographically to localities where has only minor importance and tibia length is one would expect to find triseriata (Conant, 1975) absent. Although head width and tympanum and the remaining cluster on the right (pops. diameter continue to contribute substantially in 1-4, 7-9, 11-12, and 14) corresponds to mac- separation along the K1 axis, CD becomes the ulata. To further clarify this relationship dashed major influence. Separation along the K2 axis lines on the map in Figure 1 connect popula- is still dependent upon head width and tympa- tions belonging to the triseriata group and in- num diameter and to a lesser extent snout-uro- dicate the northern limits (pops. 5 and 10). All style distance. However, CD continues to be the remaining populations in Figure 1 (except 17)dominant influence along the K2 axis. Mem- represent maculata populations. Pseudacris t. tri- bership within the two clusters is the same as PLATZ-CHORUS FROG SPECIATION 709

K2 TABLE 1. BMDP7M-COEFFICIENTS FOR CANONI- (N=158) CAL VARIABLES. The upper set of coefficients contain 4 values for the morphometric data set (n = 221) in- 14 3 dividuals alone. The lower set (n = 158) are for those individuals for which call data and morphometric data 5 . 3 were both available. The abbreviations for each vari-

1 - able are as follows: SVL = body length; TFL = tibia- 8d \ .12 fibula length; HW = head width; TYP = tympanum O- \ o10 diameter; CD = duration of call; PN = number of 13 21 /7/ -1 - 16 --o 11o. / pulses in a call; DF = dominant frequency of a call. 15 op 9 Each of the morphometric features is further defined -2- P triseriata 4 under Methods and Materials. K1 and K2 represent P. maculata -3 the first and second canonical axes respectively.

-4 Variable K, K, -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 (n = 221) Fig. 6. Canonical plot SVLfrom 0.59823 multivariate -0.24789 stepwise discriminant analysis TFL(BMDP7M, -0.67504 -1.44458 Dixon, 1981) based on the eight body measurements HW -2.05923 1.70151 and three call parameters (PN, CD and TYP DF) -2.28210 for -0.37919 each of 158 adult males. Populations are identified by number and correspond to localities (nlisted = 158) in Figure 1. The K, and K2 axes are the same asSVL Figure 0.27833 0.587165. Note that the two clusters of population HW means -1.60868 depicted -1.99833 here remain the same as Figure 5 inTYP terms -1.65344 of-0.86927 membership with the exception of the separation CD 8.20002 of-7.15832 population 14 which is intermediate between the two clusters. PN -0.39605 0.44765 DF 0.00028 -0.00028 that of Figure 5 with the exception of population 14 which shows equally low affinity to either of the two groups. umenting numerous sympatric populations of chorus frogs possessing the characteristics of DISCUSSION one or the other of two forms is consistent with the conclusion that each represents a separate Call analyses (CD and PR) alone demonstrate species. The data thus far support recognizing the existence of two call types, one replacing the boreal subspecies, P. t. maculata, as P. mac- the other across an extensive zone of north to ulata, and the western subspecies, P. t. triseriata south overlap. The differences are sometimes as P. triseriata. Preliminary data of Platz and abrupt between populations (e.g., populations Forester (1988), based on mating CD and PR 4-5 and 5-6). This pattern of marked differ- offeriarum (see this paper's discussion of pop- ences (pops. 18-19 vs 20) is seen in Colorado ulation 17) and kalmi suggest that the same may as well. be true for each of these two subspecies as well. Multivariate morphometric analysis of men- This conclusion is further supported by recent sural characters produced two clusters of pop- electrophoretic findings involving three of the ulations, independent of the call data indicating four subspecies (excluding maculata). Hedges the existence of two morphotypes. When mat- (1986) employed 33 loci to assess the genetic ing call data are incorporated in the analysis, affinities of North American hylids. A modifi- the same two clusters occur hence call type and cation of the Cavalli-Sforza distance method re- morphotype are concordant. sulted in distance values between triseriata and In several recent studies involving similar feriarum of 0.33 and between triseriata and kalmi zones of overlap among North American leop- of 0.32. These are twice the magnitude of the ard frogs, marked changes in morphology over difference betweenferiarum and P. brachyphona short geographic distances have proved to rep- (0.14) and of the same order as that between P. resent sympatric species pairs (Littlejohn and brachyphona and kalmi (0.14). Specific status for Oldham, 1968; Brown and Brown, 1972; Platz feriarum was suggested much earlier by Ralin and Platz, 1973; Platz, 1976; Frost, 1977). Doc- (1970) on the basis of laboratory crosses. 710 COPEIA, 1989, NO. 3

In light of my findings and those of Hedges females to choose between call types of conspe- (1986), the call data presented in Figures 3 andcifics and non-conspecifics. According to Little- 4 representing the South Dakota-Oklahoma john (1960, 1965), and Loftus-Hills and Little- transect warrant further discussion and inter- john (1971), to permit effective discrimination, pretation. If a conservative approach to current at least one parameter must differ by a factor range maps is adopted which presumes popu- approaching a two-fold difference, but in some lation 1 to represent allopatric maculata andcases discrimination may still be efficient with population 19 to be allopatric triseriata, and fur- smaller differences (Gerhardt, 1978). Assuming ther, that the two are behaving as species, then this is true for maculata and triseriata, then the the pattern of call variation, both in terms displacementof of mean values for maculata PR means and population variability, is of special downward in sympatry and the presence of trise- interest. riata individuals in the zone of sympatry with The four northern most populations of mac- extremely high PR (above 30 pulses/sec in pops. ulata each have higher mean values for CD (Fig. 5, 10 and 16) suggests character displacement. 3) which might be indicative of clinal variation, The concept of character displacement in the sampling error, or character displacement. Iforiginal (ecological) sense of Brown and Wilson clinal variation were operating one might then (1956) has also been applied to those repro- expect to see even higher means further south, ductive qualities which would retard gene ex- but this is not the case. Alternatively, it is pos- change between diverging sympatric popula- sible by chance to obtain four samples whose tions or what Blair (1955) referred to as means are each larger than the one to the north "reinforcement." Evidence for character dis- (P = 1/24 = 0.0625). However, this explanation placement in this second sense has been sought seems even less likely when the rest of the tran- among anurans for 30 yr during which a limited sect is considered, including the fact that vari- number of cases have been recorded. ability (as measured by the larger SD) is much Fouquette (1975) provided an additional re- higher in sympatry. Increased variability itself port involvingferiarum and nigrita in Alabama may be the result of hybridization or mixed and Georgia. His results differ from those of species sampling at single locales because thethis study in that differences between the two nature of the zone of overlap (highly variable taxa in mean CD, PN, and PR are more marked. dorsal pigment patterns) precluded assigning In sympatry, PR show no overlap in range val- individual specimens to subspecies. It is also pos- ues. PR means meet the minimum two-fold dif- sible that this pattern (mean values) among ference in value in all cases and some exceed a northern populations is the result of natural three-fold difference. Each of the two species selection within maculata favoring longer calls, can be readily identified visually in the field and suggesting that character displacement has re-no evidence of hybridization was found. Fou- duced the likelihood of mating with triseriata. quette's (1975) study involved two taxonomi- This last possibility merits consideration since cally well known congeners and a substantial it is within the zone of sympatry that the ex-number of sympatric populations thus elimi- tremes in high (maculata) and low (triseriata) natingin- two criticisms leveled at several earlier dividual values are encountered. These obser- reports of presumed cases of reproductive char- vations are more easily explained in this manner acter displacement. Clinal variation is an alter- than by assuming that matings between two native explanation for the marked upward shift species would produce hybrids with higher orin PR for the more southern populations of lower values than either parent species, rather feriarum along the Alabama-Georgia border. than the more likely case of producing inter- However, this seems less parsimonious on two mediate ones. grounds: 1) there is no evidence of clinal vari- PR (Fig. 4) reveals a similar overall pattern ation among allopatric populations ranging from with the lowest means and individual limits for central Kentucky south as far as central Ala- maculata occurring in sympatry (pops. 4, 9, bama and or approximately two-thirds of the lati- 12) and in general even greater variability thantude encompassed in his study; and 2) the shift the data for CD. toward higher PR is abrupt and coincident with The arguments advanced to explain the CDsympatry between nigrita andferiarum. There- patterns seen in populations 1-16 also apply fore,to his results currently represent the strong- PR. PR is an important parameter permitting est evidence for reproductive character dis- PLATZ-CHORUS FROG SPECIATION 711 placement in anurans. Despite this fact, LITERATURE CITED reproductive character displacement in anurans BLAIR, W. F. 1955. Mating call and stage of specia- seems rare and character displacement as a con- tion in the Microhyla olivacea-M. carolinensis com- cept remains controversial and incompletely plex. Evolution 9:469-480. understood (Paterson, 1978, 1981; Littlejohn, BROWN, L. E., ANDJ. R. BROWN. 1972. Call types of 1981). the Rana pipiens complex in Illinois. Science 176: The evidence presented here clearly suggests 928-929. secondary contact between maculata and trise- BROWN, W. L., AND E. O. WILSON. 1956. Character riata, each having diverged to a measurable de- displacement. Syst. Zool. 5:49-64. gree and therefore best explained by postulat- CONANT, R. 1975. A field guide to reptiles and am- phibians of eastern and central North America. ing reduced gene exchange between the two Houghton Mifflin Co., Boston, Massachusetts. entities. Limited data involving mating calls (see COOK, F. R. 1964. Additional records and a correc- pop. 17, Figs. 3-4) is consistent with Fou- tion of the type locality for the Boreal Chorus Frog quette's (1975) characterization of allopatric in northwestern Ontario. Canadian Field Naturalist feriarum. Both findings are an indication that 78:186-192. the call offeriarum is also distinct from that of DIXON, W. J. 1981. BMDP-77 Biomedical computer triseriata. programs. University of California Press, Berkeley, California. Although it is likely that some hybridization is occurring in the contact zone between mac- FOUQUETTE, M.J. 1975. Speciation in Chorus Frogs. I. Reproductive character displacement in the ulata and triseriata, my findings indicate that two Pseudacris nigrita complex. Syst. Zool. 24:16-23. call types are present and that the highest and FROST, J. S. 1977. Sympatry between Rana blairi and lowest values for PR and CD occur within what the southern form of leopard frog in southeastern I interpret to be a large area of overlap between Arizona (Anura:Ranidae). Southwest. Natur. 22: the two forms. When hybrid and parental types 443-453. can be individually identified from these pop- GERHARDT, H. C. 1978. Temperature coupling in ulations, differences in the means and range the vocal communication system of the Gray Tree limits for call parameters are likely to be more Frog, Hyla versicolor. Science 199:992-994. marked than they appear at present. The results HARPER, F. 1955. The type locality of Hyla triseriata. Proc. Biol. Soc. Wash. 68:155-156. from my call data are consistent with the infer- HEDGES, B. S. 1986. An electrophoretic analysis of ences drawn from the morphometric analysis in holarctic hylid frog evolution. Syst. Zool. 35(1):1- that each approach suggests the same two as- 21. semblies of populations. Taken together they LITTLEJOHN, M.J. 1960. Call discrimination and po- identify a broader area of overlap than cur- tential reproductive isolation in Pseudacris triseriata rently recognized in the literature, extending females from Oklahoma. Copeia 1960:370-371. as far west as Fort Collins, Colorado, and as far 1965. Premating isolation in the Hyla ewingi south as Lawrence, Kansas. complex. Evolution 19:234-243. Known zones of overlap involving large num- . 1981. Reproductive isolation: a critical re- bers of populations of each of the four subspe- view, p. 298-334. In: Evolution and speciation. W. R. Atchley and D. S. Woodruff (eds.). Cambridge cies pairs invite comparative studies involving University Press, Cambridge, England. biochemical, behavioral, morphological and , AND R. S. OLDHAM. 1968. The Rana pipiens acoustical approaches. Studies now underway complex: mating call, structure, and taxonomy. Sci- are expected to yield results pertinent to the ence 162:1003-1005. issues discussed here. LOFTUS-HILLS, J. J., AND M. J. LITTLEJOHN. 1971. Pulse repetition rate as the basis for mating call discrimination by two sympatric species of Hyla. ACKNOWLEDGMENTS Copeia 1971:154-156. PATERSON, H. E. H. 1978. More evidence against speciation by reinforcement. S. Afr. J. Sci. 74:369- This work was supported in part by National 371. Science Foundation grants DEB77-03254 and . 1981. The continuing search for the un- NSF-URP SPI-7826824 as well as Faculty De- known and the unknowable: a critique of contem- velopment Funds from the Creighton Univer- porary ideas on speciation. Ibid. 77:113-119. sity Graduate School. I thank B. Sullivan for PLATZ, J. E. 1976. Biochemical and morphological helpful comments and suggestions during the variation of leopard frogs in Arizona. Ibid. 1976: preparation of this manuscript. 660-672. 712 COPEIA, 1989, NO. 3

, ANDAND D.D. C. C. FORESTER. FORESTER. 1988. 1988. Geographic Geographic vari- vari- the Chorus Frogs,Frogs, PseudacrisPseudacris nigrita nigrita feriarum feriarum and and ation inin mating mating call call among among the the four four subspecies subspecies of Pseudacris of n.n. triseriata.triseriata. Amer. Amer. Midl. Midl. Nat. Nat. 48:165- 48:165- the choruschorus frog: frog: Pseudacris Pseudacris triseriata triseriata (Wied). (Wied). Ibid. Ibid. 180. 1988:1061-1065. SPENCER, A.A. W.W. 1964.1964. TheThe relationship relationship of of dispersal dispersal , AND A. L. PLATZ. 1973. Rana pipiens com-and migrationmigration toto genegene flowflow in in the the boreal boreal chorus chorus plex: hemoglobin phenotypes of sympatric and frog. al- Ph.D.Ph.D. dissert.,dissert., ColoradoColorado State State University, University, Fort Fort lopatric populations in Arizona. Science 179:1334- Collins, Colorado.Colorado. 1336. TORDOFF, W.,W., ANDAND D.D. PETTUS.PETTUS. 1977. 1977. Temporal Temporal sta- sta- RALIN, D. B. 1970. Genetic compatibility and a phy- bility ofof phenotypicphenotypic frequenciesfrequencies in in Pseudacris Pseudacris tri- tri- logeny of the temperate North American hylid fau- seriata. J.J. Herp.Herp. 11:161-168.11:161-168. na. Ph.D. dissert., University of Texas, Austin, WIED, M.M. P.P. 1839.1839. ReiseReise in in das das innere innere nord-amerika nord-amerika Texas. in den jahrenjahren 18321832 bisbis 1834.1834. 1st 1st ed. ed. J. J. Hoelscher, Hoelscher, SCHWARTZ, A. 1957. Chorus Frogs (Pseudacris nigrita Coblenz, Germany.Germany. Le Conte) in South Carolina. Amer. Mus. Novit. 1838:1-12. DEPARTMENT OFOF BIOLOGY,BIOLOGY, CREIGHTON CREIGHTON UNI- UNI- SMITH, P. W. 1956. The status, correct name, and VERSITY, OMAHA,OMAHA, NEBRASKANEBRASKA 67178. 67178. Accept- Accept- geographic range of the Boreal Chorus Frog. Proc. ed 9 Nov. 1988. Biol. Soc. Wash. 69:163-176. , AND D. M. SMITH. 1952. The relationship of

Copeia, 1989(3), pp. 712-718

Salamander Aggressiveness Increases with Length of Territorial Ownership

VANIA DA SILVA NUNES AND ROBERT G. JAEGER

The red-backed salamander, Plethodon cinereus, inhabits forest floors of eastern North America where adults establish territories. We conducted a controlled laboratory experiment to test the hypothesis that long-term territorial residents are more tenacious in territorial defense than are short-term residents; this hypothesis is commonly believed to apply to most types of territorial animals but has seldom been tested experimentally. In laboratory chambers, males of P. cinereus were allowed to establish territories for either four or 12 days before an intruder was introduced. The 12 d residents exhibited significantly more aggressive behavior and significantly less submissive behavior toward intruders than did 4 d residents. Both types of residents differed significantly from their respective controls (no salamander intruder) in certain aspects of agonistic behavior, but the two control conditions did not differ significantly for any behavior measured. We infer that increasing knowledge of a territory leads to more tenacious defense and, thus, that a long-term resident is unlikely to be ousted from a territory by an intruder.

EMPIRICAL studies often indicate that ter-this resolution of conflict often are not clear. ritorial residents generally win in conflicts Game theory has been used to investigate the with intruders (Davies, 1978; Figler and asymmetriesEin- that may occur between territorial horn, 1983) and thus retain their territories. owners and intruders (Parker, 1974; Maynard However, the behavioral interactions that lead Smith and Parker, 1976; Grafen, 1987), and

? 1989 by the American Society of Ichthyologists and Herpetologists