STATUS OF BULLFROGS AND NORTHERN LEOPARD FROGS AT FISU SPRINGS NATLONAL WILDLU'E REFUGE, JUAB COUNTY, UTAR 1 2 2 2 Matthew n. McKell ) , Scott Peterson , Keiko Kobayashi , Ryoko Miyazato , Donovan 2 2 3 Sherratt , Michael P. Oonov-dil\and Ten:y D. Schwaner • Running head: 'BULLFROGS AND LEOPARD FROGS AT FISH SPRlNGS 1Prcsent nddrc:;:;: Department ofZoo logy, Brigham Young University, Provo, Uf 84602. 2Dcpartment of Biology, Southern Utah University, Cedar City, ln' 84720 USA; telephone 435-586-7929, fax 435-865-8065. 3 Author to whom requests lor reprints should be sent. AnSTRACT.- Rutlfrogs (Rana catesbeiana) introduced into the western United States prey on other frogs, apparently, sometimes leading to the extinction of native amphibians. This hypothesis is confounded by alteroativcs, including differences in habitat disturbance, other potential predators, and competitors, as well as the dl~cts of ultraviolet radiation, parasites and pollutants. In most cases, a test requires situations where variables can be controlled. ln tllis paper we describe aspects of the ecology and natural history of bullfrogs introduced to Fisll Springs National Wildlite Refuge, a IIU!jor habitat for migratory waterfowl, and their apparent efl:ccts on putative native populations of leopard frogs also found in the area. Generally, we Jow1d size-age structure, growth, heba>~.or and reproduction of bullfrogs and leopard frogs to be similar to other populations of these species. However, both species have restricted patterns of distribution, abundance and diet that could be rdated to habitat differences and/or to predation by bullfrogs on leopard frogs. Because buUfrogs are non-native and non­ essential to the ecology ofthe refuge, their systematic removal with continued monitoring ofleopard frog distributions and ahundance, could provide a controlied test oftbe predation hypothesis. Key words: Rona catesheiana, R. pipiens, distribution, abundance, size, growth, age, ske/ctochronology, diet j The bullfrog, Rana C/1tesbeiana, is not native west of the Rocky Mountains, but has been successlully introduced into every state in that region (Stebbins 1985). nullfrogs were introduced to Fish Springs National Wildliti.~ Refuge (FSNWR), Juab County, Utah, in 1954, prior to the area being designated a federal wildlife refuge, to supply frog legs to local restaurants in nearby Salt Lake City (J. Banta pers. comrn.). Fish Springs is located within the natur<d r.mge oft.he northern leopard frog, Rana pipiens pipiens (Stebbins 1985). Whether this species v.>as al~o introducw or is native to the area is unclear; a natural population ofleopa rd frogs exists in Snake Valley just southwest of !'SNWR (Tiovingh 1997). General ecological, life history and behavioral studies (summarized in Bury and Whelan 1984, and Clarkson and DeVos 1986) indicated negative interspecific effects or bullfrogs on other ran.id frogs, including tadpole crowding that affected growth ofboth bullfrogs and other anurans (Licht 1967) and adult territoriality, possibly responsible for declines ofnativ e ranid populations in California (Moyle 1973). Hrurummon (1982a) found lower abundances of leopard frogs in at<:ll~ of Colorado inhabitw by buUfrogs. However, JcrUJings and Hays (1984) and Hayes and Jennings ( 1986) cited the lack of experimcntaJ evidence and suggested causal factors other than bullfrog introductions for these d<:elines. Su~quently, Hecnar and M'Closkcy (1996) documented increased abundances and distrib.utions ofnative ranids following the apparent extinction of bullfrogs from a national park in Canada. Similar studies 1bund reduced populations of red-legged frogs (Rona aurora) and ye11ow-legged frogs (R. boylii) in areas where bullirogs had been introduced (Kiesecker and Rlaustcilt 1997, 1998; Kupferberg 1997). Our study adds to tltis controversy by describing the distnbutions, abundances, sW:-age structure, gTO\vth, diet, and hahil~ of bullfrogs and Jeopard frogs at FSNWR. The descriptive results suggest a test ofth e null hypothesis of 5 Each captured frog was sexed, mta:;ured for snout to vent length (SVL in mm), weighed (WT in g), and toe-clipped with a unique mark (after Ferner 1979). We used the Peterson index (Canghley 1977) to ~tiroate population si7.es of bullfrogs in three transect~. Toe bones were prt:lit!rved in I 0 percent formalin cleaned of ~kin, muscle and tendons, de-minernlizcd overnight in 3 percent nitric acid, sectioned with a freeze­ microtome, stained with Ehrlich' ~ hematoxylin, and viewed a.nd photographed with a light microscope. We asswned lines of arrested growth (LAGs) formed during ~easonal inactivity and separating ~~a:;onal growth areas represented years since metamorphosis (F.stcoo.u ct al. 1996). We computed growth rote~ as SVLt2-SVLtdt2-ts, where t, = time of initial capture and t2 =time in days for frogs recaptured 20 or more da~ since initial capture (Andrews 1982). We flushed stomachs with water from a 50cc syringe fitted with a plastic tube. Stomach contents were preserved in 10 percent formalin and later tran~l~m:d to alcohol for analysis. Stomneh items were sort~d into groups, identified, :md measured for volume by water displacement in a graduated cylinder. Occurrence of a prey category was the number ofstomach:~ in which the item occurred divided by the total number of stomachs examined. Volume percentages were the total volume of a specific item divided by the total voiUlll\: for all samples. Statistical comparisons followed Sokal and Rohlf(\981). We uscu tltt: Kruskal­ Wallace test and Fcidman's method ofrandomized blocks to examine the abundances of fTogs in five transects over lour :sainpling periods. Spennnnn rank correlation tested the n.ssocwtio11 ofre lative abundances of bullfrogs and leopnrd frogs in the five transects. Chi-square tested the assumption of even distributions of frogs in Transects 4 and 5, and the null model!: I sex ratio for adult~. Analysis of cov<.tri;mc.;e (ANCOV/1.) compared the 6 functional relationship of length and weight in adult males and ti:ma!t:s and linear regression determined the amount of variation in growth rate explained by body length. RESULTS Distnbutions In February and March, we observed bullfrogs in springs directly connected to the main channel by permanent water flows (i.e., those m~ar transects 1 and 2, including House.Spring), and in Waller Spring connected to the main channel by a long canal system with reduced or intcrmillcnt water during late summer. We saw larval (>80 mm total length), juvenile <md adult bullirog$ in $prings adjacent to Transects I and 2, and the main charUlel, where water temperatures were > l9°C; bullfrogs were not seen in Avocet, Curle:;w, Shoveler, Egret, Ibis, Gadwall, Pintail and Harrison pools, or in roadside ponds with water temperatures below I o•c. From June to August, male bullfrogs called from springs, canals coru1ecting springs to the main channel, pools and roadside ponds adjacent to the main chrumel. We obseJ:Ved clasping pairs and egg masses only itl the main channel and canals connecting springs and adjacent pooL~ along Transects 1-4. Bullfrogs were absent from North Spd.ng, Deadmatl Spring, and the northernmost pools (Harrison, Pintail, Gadwall <md Ibis), at all times during the study. Leopard frogs became active in March-April. We observed them along the maio ch.am1el and in dense vegetation at the edge ofc<mals <md adjacent pools in Transects 1-5, when water temperatures were >l5°C. Leopard frog~ inhabited the easternmost margins of Egret Pool and Curlew Pool and adjacent roadside ponds (Fig. I). However, we rarely ob~erved leopard frogs in the springs or in canals connecting the springs with the main channel in Transects 1 and 2. 'I Abundances We counted a total of 77 1 bull !Togs and 165 leopard frogs in systematic surveys along tr.msects, monthly from April to July. For both species, the relative numbers of frogs (adjusted to observations per one kilometer t.ransect, Tabk I) did not differ among sampling dates (Kruskal-Wallis tests: bullfi:ogs, H - 1.58, df- 3, P = 0.66; leopard frogs, II = 4. 73, df .. 3, P- 0.19); among transects, however, we found highly significant differences in relnt.ive nbundauces of both species (Friedman's method for randomized blocks: bullfrogs, x2 = 11.8, df= 4, p < 0.025; leopard 6·ogs, x_2 = 23.5, df- 4, I' < 0.001). Total counts of bullfrogs and kopard !Togs per transect (Fig. 2), adjusted to observations pc::r km, show a significant negative relationship (Spearman rank correlation = -1.00, p < 0.001). In Transects I and 2, we marked 82 bull !Togs between April and July, and r~:eaptured only two frogs; however, eighteen of82 adult buUfrogs (SVL > 110nun) captured between August and October were recaptures. In Transect 5, we marked 15 bullfrogs and recaptured none between April and July; however, seven of 11 adults captured between August and October were recaprurcs. Applying a simple Petersen estimate (Caughlcy 1977) to these data gave population estimates and 95 percent confidence limits of358 ± 138 bullfTog.~ fur Transects 1 and 2, and 23 ± 81in Tr.msect5 (a difference in average densitic~ of 179 vs.l2 frogslkm, respectively). Although we marked 68 leopard frogs between April and July (51 !Tom Transect 5), and captured 52 (SVL > 60 mm SVL) between August and October (33 in Transect 5), only 3 were n:o.:aptured, all in Transect 5. These data do not give accurate density results (95% wnfidence limits exceed the population estimate by a fuctor of2, for Transect 5) due to the small number.; ofrecaptures. We arbitrarily designated bullfrogs as smaU, mcditun or large, in our April 1999, direct count, survey of the five transects. Later, when bullfrogs were caught and measured, small frogs were fuund to be juveniles in their second growing sea~on, post­ metamorphosis; medium and large bullfrogs represented adult~ ofvarious ages. Small bullfrogs C{)mprised 18 percent of totals in TrrulSccts 1-4, but were abt;t:nt in Transect 5.