SALAMANDRA 43 1 129-138 Rheinbach, 20 August 2007 ISSN 0036-3375

Comparison of anuran acoustic communities of two habitat types in the Danum Valley Conservation Area, ,

Abstract. We compared advertisement calls of assemblages in two different habitats, (i) an Open area along a dirt road with ponds and secondary vegetation; (ii) a fast flowing stream in primary forest. Eleven frog species were recorded and significant differences in the dominant call frequencies between the two observed frog communities were discovered. Stream-breeding species produce higher frequencies than species occurring in the roadside habitat. Noisy habitats have an influence on dominant frequency and demand acoustic adaptations to increase the signal-to-noise ratio. Selective logging represents a major threat to stream-breeding anurans in Sabah. Pollution of clear water threatens the stream-dependent herpetofauna. Keywords. Amphibia, Anura, conservation, vocalisation, Malaysia.

Introduction GER 1969, DUELLMAN1978, INGER& VORIS 1993, GILLE~PIEet al. 2004). Apart from be- Selective logging has been the main cause of havioural and morphological adaptations to disturbance to tropical forests in Southeast the environment, frog assemblages also show Asia (MARSH& GREER1992, WILLOTTet al. acoustic differences. A prime factor sepa- 2000). Concession-based timber extraction rating species within anuran assemblages and plantation establishment result in high- is vocaiization (DUELLMAN& TRUEB 1986). ly fragmented forests (MCMORROW& TAL- Acoustic partitioning through temporal and IP 2001, CURRANet al. 2004) and increase spectral features minimizes competition and river sediment yields (DOUGLASet al. 1993). allows conspecific females to recognize indi- The extent to which biodiversity is altered vidual signalling males (HÖDL1977, GARCIA- in selectively logged forest is an important RUTLEDGE& NARINS2001, GERHARDT8r conservation issue. Fragmentation, degrada- HUBER2002). The dominant frequency of a tion and conversion of rainforests affect the call is partially constrained by its body tropical herpetofauna (ALCALAet al. 2004). size (KIMEet al. 2000). Snout-vent length and About half the frog species in Southeast Asia dominant frequency are negatively correlated are restricted to riparian habitats and devel- (LITTLEJOHN1977, DUELLMAN& PYLES 1983, op in streams (INGER1969, ZIMMERMAN& RYAN& BRENOWITZ1985). Sounds produced SIMBERLOFF1996). Most anuran stream-side by waterfalls and fast-flowing streams con- communities in are known to breed tribute to the ambient noise level. Distinct, in clear, turbulent water and are absent in acoustic habitat properties ("melotops") de- streams with silt bottoms and lacking riffles mand different adaptations on vocal- and torrents (INGER& VORIS1993). Several izations. In this study, species composition anuran community studies have focused on was determined for two habitats within the environmental conditions such as vegeta- Danum Valley Conservation Area (DVCA), tion, elevation, rainfall and microhabitat (IN- a fast flowing stream in the primary forest

0 2007 Deutsche Gesellschaft für Herpetologie und Terrarienkunde e.V (DGHT) http:llwww.salamandra-journal.com INGER et al. and a roadside habitat with secondary veg- ondary vegetation, ephemeral roadside pools etation, and spectral and temporal features of filling seasonally with water, and varying de- the acoustic signals were compared. grees of human disturbance. In both habitats iooo m long and y-shaped transects were established and marked with coloured tape. Material and methods We performed visual and acoustic encounter Study area surveys. The search areas also included adja- Cent riverbanks or vegetation up to a distance From 10 January to 13 March 2006, we studied of 3 m to the water- or road-edge. Sampling frog communities in DVCA, Sabah, Malaysia. was performed independent of prevailing DVCA has an area of 43,800 ha of lowland weather conditions. Each transect was sam- tropical rainforest and is located in the upper pled at least 20 times by two observers during catchment area of the Segama River. The Da- the day and during nocturnal censuses. Re- num Valley Field Centre (DVFC) is located peated controls of the Same transect on con- on the Western edge of the DVCA (4'57'40'' secutive days were avoided to ensure inde- N, ii7"48'oon E; all but 9% of the area is below pendence of samples. To determine all call- 760 m above sea level) from where a 2.8 km ing species, a 24-hour time period was sam- long trail leads south to the Tembaling water- pled four times during one month along both fall. The rainy season in Sabah extends from transects. After locating a vocalizing male, December to March (Northeast Monsoon) call recordings were made from distances of with rainfall peak during February (monthly i to 5 m using directional and surround mi- total 788 mi). Annual precipitation is near crophones (Sennheiser Me 66, AKG D 190 E) 3,000 mm with year-to-year fluctuation of and DAT- recorders (Sony DAT-Rec. TCD- about ioo mm. Mean annual temperature D8). Microhabitat temperature and humid- is 26.7"C with mean maximum 30.9"C and ity were measured with a digital thermohy- mean minimum 22.5"C. Daily temperature grometer (Testo 610 GM) before each record- fluctuations are noticeably more pronounced ing. Frogs were located at their resting sites, than the year-to-year fluctuations. In the pe- and if possible, captured and placed in plastic riod from January to March the highest re- bags. Snout-vent length and snout-urostyle corded temperature was 32.5OC and the low- length were measured using a Wiha calliper est was 20.3"C; mean relative humidity at (I0.1 mm). To differentiate between individ- the DVFC was 97.0% at 8:ooh and 89.7% at uals we photographed and compared dorsal 14:ooh. Daily rainfall and temperature data colour Patterns. All individuals were released from January to March 2006 derived from after taking their measurements. the Danum Valley weather station, which is maintained by the staff of the Royal Society SE Asia Rain Forest Research Programme. Data analysis

We digitized and analysed the recorded calls Sampling methods with the sound-analysis Software Raven 1.2.1 at a sampling frequency of 44.1 Hz and with We focused on two macrohabitat types. One a mono 16 bit PCM Input and a 10 Hz update is pristine, the other disturbed. The Tembal- rate at normal speed. Power spectra, sono- ing River within the primary dipterocarp for- grams and oscillograms of one to five adver- est at about 900 m elevation is a fast flowing tisement calls were analyzed for each frog freshwater stream with rapids, waterfalls and and the average dominant frequency, mini- no siltation. The selected disturbed habitat mum and maximum frequency, call duration, was along a dirt road in the DVFC with sec- note duration, repetition rate and number of Anuran acoustic communities in Sabah, Malaysia notes, when applicable were calculated for (Fig. 5), otilophus (Fig. 6), and each individual. Spectrogram views were Bufo nsper (Fig. 7), Meristogenys orphnocne- produced with a Hann-filter with a sample mis (Fig. 8), Microhyla sp. (Fig. g), Staurois size of 512 and a 3-dB filter bandwidth of 124 natator (Fig. io), Staurois latopalmatus (Fig. Hz. The correlation between body size (SVL) ii), respectively. No calls could be recorded and dominant frequency was calculated with for species without frequency data (Tab. 1). A a Spearman non-parametric correlation (p 5 total of 213 calls were analyzed. The dominant 0.05). ANOVA (p 5 0.05) was used to test for call frequencies of the anuran species are not statistical differences, in SVL (mm) and dom- dependent on the SVL of the vocalizing in- inant frequency (Hz) of advertisement calls, dividual~(N = 11; C = -0.555; p < 0.077; r2= among frog assemblages in two habitats. The 0.30) (Fig. 13). Meristogenys orphnocnemis, a influence of SVL and habitat on the dominant species found in the stream habitat produced frequency of the frog species was calculated the highest calling frequency (7,205.0 Hz). by an ANCOVA (p 5 0.05). Since the sample Frog species from the torrent are above the size for some anuran species tested was N = 1, the influence of SVL on dominant frequency was calculated twice: once corrected accord- ing to number of individuals and once with- out correction. All statistical tests were pro- duced with SPSS for Mac 11.0.4. ' 01 ' 02 ' 03 ' 0:4 ' 0:s '

Results 15 10 5 During the sampling period 20 frog species were found. Eleven frog species were record- kHz 0,1 O,Z 0,3 0,4 0,5 ed, six along the disturbed area and five with- in the pristine habitat: Fejervarya nicobarien- 2. 'pectrogram (frequenc~LkHzI) arid wave- form (rel. amplitude [kUnit] = proportional to sis 11, Feje""ar~a limflocharis (F&. 2)> pressUre of the recording) of the ahertise. Rhaco~korus 31, Pol~~edates ment ca11 of Fejervarya limnocharis. Temperature leucomystax (Fig. 4), Polypedates macrotis during recording 24.6 "C,

Fig. 1. Spectrogram (frequency [kHz]) and wave- Fig. 3. Spectrogram (frequency [kHz]) and wave- form (rel. amplitude [kUnit] = proportional to form (rel. amplitude [kUnit] = proportional to sound pressure of the recording) of the advertise- sound pressure of the recording) of the advertise- ment call and non-collected specimen of Fejer- ment call and non-collected specimen of Rhacoph- varya nicobariensis. Temperature during record- orus dulitensis.Temperature during recording 26.3 ing 25.4 "C. "C. DORISPREININGER et al.

Fig. 4. Spectrogram (frequency [kHz]) and wave- Fig. 6. Spectrogram (frequency [kHz]) and wave- form (rel. amplitude [kUnit] = proportional to form (rel. amplitude [kUnit] = proportional to sound pressure of the recording) of the advertise- sound pressure of the recording) of the advertise- ment call and non-collected specimen of Polype- ment call and non-collected specimen of Poly- dates leucomystax. Temperature during recording pedates otilophus. Temperature during recording 25.0 "C. 25.4 'C.

Fig. 5. Spectrogram (frequency [kHz]) and wave- Fig. 7. Spectrogram (frequency [kHz]) and wave- form (rel. amplitude [kunit] = proportional to form (rel. amplitude [kUnit] = proportional to sound pressure of the recording) of the advertise- sound pressure of the recording) of the adver- ment call and non-collected specimen of Poly- tisement call and non-collected specimeil of Bufo pedates macrotis. Temperature during recording asper. Temperature during recording 24.6 "C. 24.9 "C.

calculated regression except Bufo asper and = 11; df = I; F = 7.315, P = 0.024). Dominant Microkyla sp. All frog species in the road- frequency is not dependent on the SVL (N = side habitat are below the calculated regres- 11; df = 1; F = 3.822, p = 0.082) but the influ- sion. The largest frog species was Polypedates ence of the habitat is significant (N = 11; df = otilopkus with a dominant call frequency of I; F = 7.031, P = 0.029). If the number of ana- 1,033.6 Hz. No frog species with a dominant lyzed individuals is included, the dependen- call frequency below 1,000 Hz was recorded. cy of the dominant frequency, the influence Frequencies of advertisement calls of anuran 0f SVL (N = 11; df = 1; F = 0.519, p = 0.049) species found in the stream transect are high- and the influence of the habitat (N = 11; df = er than call frequencies recorded of species 1; F = 11.435, p = o.010) become more signifi- occurring in the roadside habitat (Fig. 14) (N cant. Aiiuran acoustic communities in Sabah, Malaysia

Fig. 8. Spectrogram (frequency [~HZ])and wave- Fig. 10. Spectrogram (frequency [~HZ])and wave- form (rel. amplitude [kUnit] = proportional to form (rel. amplitude [kunit] = proportional to sound pressure of the recording) of the advertise- sound pressure of the recording) of the advertise- inent call aild non-collected speciinen of Meristog- ment call and non-collected specimen of Staurois enys orphnocnemis. Temperature during recording natator. Ternperature during recording 25.3 "C. 24.8 "C.

Fig. 9. Spectrogram (frequency [kHz]) and wave- Fig. 11. Spectrogram (frequency [kHz]) and wave- form (rel. amplitude [kUnit] = proportional to form (rel. amplitude [kUnit] = proportional to sound pressure of the recording) of the advertise- sound pressure of the recording) of the advertise- ment call and non-collected speciinen of Micro- ment call and non-collected specimen of Staurois hyla sp. Temperature during recording 25.3 "C. latopalmatus. Temperature during recording 25.2 0,-

Discussion grams of streams show high sound pressure levels in low frequencies and lower sound The comparison of two local frog assem- pressure levels in frequencies above 3,000 blages in different habitats shows that adver- Hz (HÖDL& AMEZQUITA2001, FENG et al. tisement calls of frog species living in noisy 2006, authors' unpublished data). Environ- environments constrain higher frequencies. mental constraints should have an effect on Fast flowing streams and waterfalls produce the evolution of signals (NARINS8r ZELICK low-frequency-dominated noise which may 1988). Selection favors signals that minimize present a selective force for stream-breeding the effects of background noise and inter- species (HÖDL8r AMEZQUITA2001, FENGet fering signals from other species (ENDLER al. 2006). The high-frequency calls exceed 1993). Riparian frog species are acoustically the constant ambient noise level. Spectro- well adapted to their environment and show DORISPREININGER et al.

Fig. 12. a) I„. varya nicobariensis, b)Fejervarya limnocharis, C) Rhacophorus dulitensis, d) Polypedates leucomystax, e) Polypedates macrotis, f) Polypedates otilophus, g) Bufo asper, h) Meristogenys orphnoc- nemis, i) Microhyla sp., j) Staurois natator, k) Staurois latopalmatus. Photos: M. BÖCKLE, W. HÖDL, T. REIS. Anuran acoustic communities in Sabah, Malaysia

cies found in the disturbed roadside habitat call with lower frequencies and less constant noise is emitted by the surrounding environ- ment. This gives rise to the question why a high proportion of species breeds close to fast flowing streams (INGER1969). ZIMMER- MAN & SIMBERLOFF(1996) showed that re- productive mode and developmental habitat are strongly associated with phylogeny. The Same limited Set of ecological and reproduc- tive traits is shared by ranids, pelobatids and rhacophorids. Environmental conditions and interspecific interactions may have led to ad- aptations, but local selection on species can not be counted independent when closely re- lated species occur in the Same community. Phylogeny was not included in our analysis and a certain bias may result from the fact that two species of the genus Staurois are present at the stream habitat. Advertisement calls are important determinants of repro- ductive success and could have evolved in a common ancestor. However, we were able to record five species of the farnily of ranids, three at the stream (Staurois latopalmatus, S. natator and Meristogenys orphnocnemis) and two along the roadside (Fejervarya nico- bariensis and E limnocharis). All three ranid species recorded along the stream produce advertisement calls in which the dominant frequencies are above the calculated linear regression line (Fig. 13). Although morpho- logical and phylogenetic limitations con- strain frog vocalizations, selective pressures strong habitat affiliations. With the exception + such as environmental factors have resulted of Bufo asper the mean dominant frequency in a diversity of advertisement calls (DUELL- was above 3.9 kHz for frogs calling in the un- MAN & TRUEB1986). Acoustic partitioning disturbed habitat. A possible explanation for and increasing the signal-to-noise ratio are the call frequency in B. asper could be that it prerequisites for frogs living along torrents prefers stream banks, with small riffles that and streams. The influence of the habitat on produced less background noise. Addition- advertisement calls should emphasize the ally, dominant frequency is determined by importance of microhabitat composition on SVL. Larger frogs call with lower frequencies frog assemblages. Most recorded species are than smaller frogs (LITTLEJOHN1977, KIME threatened by habitat loss (Tab. 1). Deforesta- et al. 2000). The sampling size was proba- tion and subsequent siltation of streams and bly too small to produce the dependency of rivers are the major threats to most stream- SVL on dominant frequency. Only weighted breeding species in Sabah. Out of twelve spe- data were able to show the influence. All spe- cies found along the stream, three are clas- DORISPREININGER et al.

~Meristogenysoiphnocnemis

@

0 Fejervatya limnocharis Bujo asper

SVL [mm]

Fig. 13. Influence of SVL on dominant frequency of the species advertisement call (y = -55.49~+ 5,472.71). Frog species of the stream transect are marked with a Square, those from the roadside habitat with a circle. completely lacked all the species known to breed along clear and fast flowing streams. Selective logging changes the water chemis- try considerably in nearby streams. Sediment yields of streams are 18 times higher within the next five months in selectively logged ar- eas (DOUGLASet al. 1992,1993). The increase in sedimentation levels has a clear effect on larval anurans and on reproduction of sever- al stream-breeding species. Frog assemblages as those found at the Tembaling River show the importance of conservation areas like Danum Valley. On the other hand more mi- crohabitats are created through reduced im-

Roadside Stream pact logging which attracts frog species that Fig. 14. Dominant call frequencies of the anuran are normally not found in primary forests species from two different habitat types. Species (e.g., Polypedates macrotis, Fejervarya nico- along the stream call with higher frequencies than bariensis and Rhacophorus pardalis) (WONG frog species in the roadside habitat. in press). We believe that disturbed areas such as those around the Danum Valley Field sified as "near threatened according to the Center and adjacent roads contain suitable IUCN Red List Status (Tab. I). INGER& VORIS ponds and breeding habitats for disturbance- (1993) found that a stream with a silt bottom tolerant breeders, but the large destruction Anuran acoustic communities in Sabah, Malaysia

Tab. 1. List of anuran species found including weight, snout-vent length (SVL), dominant call frequency (DF) and IUCN Red List status. Habitat: 1 = stream, 2 = road edge (numbers in parentheses are the ab- solute number of individuals measured); weight [g]; SVL [mm];dominant frequency, DF [Hz] (number of individuals/number of analyzed calls); major threat: HL = habitat loss, P = pollution of streams and rivers, ID = infrastructure development, N = none; Red List status according to IUCN et al. (2006). species habitat SVL DF major threat IUCN Red List status Bufo asper 1 76.2 (1) 1,031.2 (119) HLIP least concern Meristogenys orphnocnemis 1 30.2 (4) 7,205.0 (116) HLIP least concern Microhyla sp. 1 14.9 (5) 3,969.3 (4125) Staurois natator 1 33.9 (4) 4,746.0 (3119) HLIID least concern Staurois latopalmatus 1 47.7 (11) 5,149.4 (5120) HLIP least concern Ansonia longidigita 1 HLIP near threatened Ansonia spinulifer 1 3.2 (1) HLIP near threatened Leptobrachium abbotti 1 56.5 (5) HL least concern Limnonectes leporinus 1 106.8 (1) HL least concern Rana picturata 1 39.8 (2) HL least concern Pedostibes rugosus 1 HLIP near threatened Chaperina fusca 1 18.8 (1) HL least concern Chaperina fusca 2 HL least concern Polypedates leucomystax 2 30.0 (7) 2,056.7 (4135) N least concern Polypedates inacrotis 2 57.0 (5) 1,388.9 (113) N least concern Polypedates otilophus 2 81.1 (9) 1,033.6 (112) HL least concern Fejervarya limnocharis 2 33.4 (3) 1,274.0 (313) least concern Fejervarya nicobariensis 2 41.2 (6) 3,169.7 (5144) HLIN least concern Rhacophorus dulitensis 2 46.6 (4) 1,868.6 (5147) HL near threatened Rhacophorus pardalis 2 43.5 (6) HL least concern Occidozyga laevis 2 31.0 (10) N least concern and fragmentation of primary forest poses a References major threat to the characteristic herpetofau- na of Sabah. ALCALA,E.L., A.C. ALCALA& C.N. DOLINO (2004): and reptiles in tropical rainforest fragments on Negros Island, the Acknowledgements Philippines. - Environ. Conserv., 31: 254-261. CURRAN,L.M., S.N. TRIGG,A.K. MCDONALD,D. We thank the Danum Valley Management Com- ASTIANI,Y.M. HARDIONO,P. SIREGAR,I. CA- mittee, the Economic Planing Unit, the Sabah NIAGO & E. KASISCHKE(2004): Lowland forest Wildlife Department and the Universiti of Ma- loss in protected areas of Indonesian Borneo. laysia Sabah for giving us the opportunity to work - Science, 303: 1000-1003. in the Danum Valley Conservation Area. Special DOUGLAS,I., T. GREER,K. BIDIN8r M. SPILSBURY thanks go to the Royal Society (RS) and GLEN (1993): Impacts of rainforest logging on riv- REYNOLDSfor their support and commitment to er Systems and communities in Malaysia and make this project possible. JULIA FELLINGand Kalimantan. - Global Ecol. Biogeogr. Lett., 3: KAROLINESCHMIDT assisted during our fieldwork and MARKUSGMEINER did the photo editing. We 245-252. are very thankful for the statistical help of ADOLFO DOUGLAS,I., T. SPENCER,T. GREER,K. BIDIN8r AMEZQUITA.ANNA WONG acted as our local col- W. SINUN(1992): The impact of selective com- laborator. Financial support was given by the Uni- mercial logging on stream hydrology, chemis- versity of Vienna (Brief Scientific Stays Abroad) try and Sediment loads in the Ulu Segama Rain and the DGHT (WILHELM-PETERS-Fonds).The Forest, Sabah, Malaysia. - Phil. Trans. R. Soc. Department of Evolutionary Biology (University Lond. B, 235: 397-406. of Vienna) provided the necessary field equip- DUELLMAN,W.E. (1978): The biology of an equa- ment. Last but not least we thank the DVCA and torial herpetofauna in Amazonian Ecuador. - RS staff for keeping our Spirits up. Misc. Publ. Mus. Nat. Hist. Univ. Kansas, 65: 1-352. - Washington. [www.globalamphibians.org, latest accessed 15 December zoo6] DUELLMAN,W.E. & A.R. PYLES(1983): Acoustic resource partitioning in anuran communities. KIME,N.M., W.R. TURNER& M.J. RYAN(2000): The transmission of advertisement calls in Central American frogs. - Behav. Ecol., 11: 71- DUELLMAN,W.E. & L. TRUEB(1986): Biology of amphibians. - McGraw Hill, New York. 83. ENDLER,J.A. (1993): Some general comments on LITTLEJOHN,M.J. (1977): Long range acoustic the evolution and design of auimal communi- communication in anurans. - pp. 263-294 in: TAYLOR,D. H & S.I. GUTTMAN(eds.): The re- cation Systems. - Phil. Trans. R. Soc. Lond. B, 340: 215-225. productive biology of amphibians. - Plenum Press, New York. FENG,A.S., P.M. NARINS,C.-H. Xu, W.-Y. LIN,Z.- MARSH,C.W. & A.G. GREER(1992): Forest land- L. Yu, Q. QIU,Z.-M. Xu & J.-X. SHEN(2006): use in Sabah, Malaysia: An introduction to Da- Ultrasonic communication in frogs. - Nature, num Valley. - Phil. Trans. R. Soc. Lond. B, 335: 331.339. GARCIA-RUTLEDGE,E.J. & P.M. NARINS(2001): Shared acoustic resources in an Old World frog McMo~~ow,J. & M.A. TALIP(2001): Decline of forest area in Sabah, Malaysia: Relationship to community. - Herpetologica, 57: 104-116. state policies, land Code and land capability. - GERHARDT,H.C. & F. HUBER(2002): Acoustic Global Environ. Change, ii: 217-230. communication in insects and anurans. - Univ. Chicago Press, Chicago and London. NARINS,P.M. & R. ZELICK(1988): The effects of noise on auditory processing and behavior in & GILLESPIE,G.R., D. LOCKIE,M.P. SCROGGIE amphibians. - S. 511-536 in: FRITZSCH,B., M. D.T. ISKANDAR(2004): Habitat use by stream- RYAN,W. WILCZYNSKI,T. HETHERINGTON& breeding frogs in south-east Sulawesi, with W. WALKOWIAK(Hrsg.): The evolution of the some preliminary observations on community auditory System. - John Wiley & organization. - J. Trop. Ecol., 20: 439-448. Sons, New York. HÖDL,W. (1977): Ca11 differences and calling site RYAN,M.J. a E.A. BRENOWITZ(1985): 7he role of Segregation in anuran species from central body size, phylogeny, and ambient noise in the - Amazonian floating meadows. Oecologia, evolution of bird Song. - Amer. Nat., 126: 87- 28: 351-363. 100. & HÖDL,W. A. AMEZQUITA(2001): Visual sign- WILLOTT,S.J., D.C. LIM, S.G. COMPTON& S.L. aling in anuran amphibians. - pp. 121-141 in: SUTTON(2000): Effects of selective logging on RYANM.J. (ed.): Anuran communication. - the butterflies of a Bornean rainforest. - Con- Smithsonian Inst. Press, Washington. serv. Biol., 14: 1055-1065. INGER,R.F. (1969): Organization of communi- WONG,A. (im Druck): The impact of forestry prac- ties of frogs along small rain forest streams in tices on frog communities in Sabah, Malaysia. Sarawak. - J. Anim. Ecol., 38: 123-148. - PhD Dissertation. Universiti Putra Malaysia, INGER,R.F. & H.K. VORIS(1993): A comparison Kota Kinabalu, Malaysia. of amphibian communities through time and ZIMMERMAN,B. & D. SIMBERLOFF(1996): An his- from place to place in Bornean forests. - J. torical interpretation of habitat use by frogs in Tropical Ecology, g: 409-433. a Central Amazonian forest. - J. Biogeogr., 23: IUCN, Conservation International, and Nature- 27-46. Serve. (2006): Global Amphibian Assessment.

Manuscript received: 24 August 2006 Authors' addresses: DORISPREININGER, MARKUS BÖCKLE, WALTER HÖDL, Department for Evolu- tionary Biology, University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria, E-Mails: doris- [email protected];[email protected]; [email protected].