Ecology of Chytridiomycosis in Rainforest Stream Frog Assemblages of Tropical Queensland

Ecology of Chytridiomycosis in Rainforest StreamFrog Assemblages of Tropical QueenslandDOUGLAS C. WOODHAMS∗ AND ROSS A. ALFORDSchool of Tropical Biology, James Cook University, Townsville, QLD 4811, Australia

Abstract: In the wet tropics of Queensland, Australia, eight species of stream-dwelling frogs have experi-enced population declines. Some declines were associated with an emerging infectious disease of amphibians(chytridiomycosis) caused by the fungus Batrachochytrium dendrobatidis. We examined the spatial and tem-poral pattern of infection prevalence in a sample of frog populations. Infected adults and tadpoles of allspecies were found, and infections occurred at every site. Infection prevalence varied among species and wasalways < 10.0% in adults but ranged from 0.75 to 76.0% in tadpoles. In this system tadpoles and adults ofsome species may act as disease reservoirs, experiencing avirulent infections, whereas other hosts (decliningspecies) experience virulent infections. Infection prevalence was higher during the cool, dry winter season(May to September) and at high elevations (600–800 m), suggesting regulation by environmental conditions,including temperature and precipitation. We found no relationships between infection prevalence and meanbody condition, fluctuating asymmetry of hind limbs, population density, or the presence of metamorphosingtadpoles and juvenile frogs. Although it is not certain whether chytridiomycosis was responsible for past frogpopulation declines in the wet tropics of Queensland, the pathogen is now endemic. Our data indicate thatat the landscape level, environmental conditions have strong effects on host-pathogen dynamics. These effectsinteract with species-specific behavior or immune function and may be important underlying determinantsof chytridiomycosis epizootics and emergence.

Key Words: amphibian, Batrachochytrium dendrobatidis, chytrid, disease ecology, infection prevalence, reser-voir host, tadpole

Ecologıa de la Quitridiomicosis en Ensambles de Ranas de Selva Lluviosa en Queensland Tropical

Resumen: En el tropico humedo de Queensland, ocho especies de ranas que habitan en arroyos han experi-mentado la declinacion de sus poblaciones. Algunas declinaciones se asociaron con una enfermedad infecciosa(quitridiomicosis) emergente en anfibios causada por el hongo Batrachochytrium dendrobatidis. Examinamosel patron espacial y temporal de la prevalencia de la infeccion en una muestra de poblaciones de ranas.Encontramos adultos y renacuajos de todas las especies infectados, y las infecciones ocurrieron en todos lossitios. La prevalencia de la infeccion vario entre especies y siempre fue <10.0% en adultos pero vario entre0.75 y 76% en renacuajos. En este sistema, los renacuajos y adultos de algunas especies pueden fungir comohospederos reservorio, experimentando infecciones no virulentas, mientras que otros huespedes (especies endeclinacion) presentan infecciones virulentas. La prevalencia de la infeccion fue mayor durante el perıodoinvernal fresco, seco (mayo a septiembre) y en elevaciones altas (600-800 m), lo que sugiere la regulacionpor condiciones ambientales incluyendo temperatura o precipitacion. No encontramos relaciones entre laprevalencia de infeccion y la condicion corporal media, la asimetrıa fluctuante en extremidades inferiores, ladensidad poblacional o la presencia de renacuajos en metamorfosis y ranas juveniles. Aunque es incierto si laquitridiomicosis fue responsable de declinaciones pasadas en el tropico humedo de Queensland, el patogenoahora es endemico. Nuestros datos indican que, a nivel de paisaje, las condiciones ambientales tienen efectos

∗Current address: Department of Microbiology and Immunology, A-5301, Medical Center North, Vanderbilt University, Nashville, TN 37232-2363,U.S.A., email [email protected] submitted August 4, 2004; revised manuscript accepted December 28, 2004.

1449

Conservation Biology 1449–1459C©2005 Society for Conservation BiologyDOI: 10.1111/j.1523-1739.2005.00236.x

1450 Ecology of Chytridiomycosis Woodhams & Alford

fuertes sobre la dinamica hospedero-patogeno. Estos efectos interactuan con el comportamiento o la funcioninmune de cada especie y pueden ser factores determinantes subyacentes en la epizootia y emergencia de laquitridiomicosis.

Palabras Clave: anfibio, Batrachochytrium dendrobatidis, ecologıa de enfermedad, hospedero reservorio,prevalencia de infeccion, quitridio, renacuajo

Introduction

Wildlife disease ecology is currently receiving renewed at-tention because of the obstacles pathogens may pose forthe conservation of rare species (Cleaveland et al. 2001).Recently, emerging infectious diseases have been associ-ated with amphibian population declines (e.g., Berger etal. 1998; Daszak et al. 1999; Bosch et al. 2001; Green etal. 2002; Speare 2003). Sick and dying frogs have beenfound with chytridiomycosis during regular monitoringsurveys at times of mass mortality (Berger et al. 1998,1999, 2004; Lips et al. 2003a, 2004). As monitoring forBatrachochytrium dendrobatidis (Bd) infections has in-creased, it has been found over an increasing geographicrange, which now includes the United States (Bradley etal. 2002; Green et al. 2002), Spain (Bosch et al. 2001), Cen-tral America (Lips 1999; Lips et al. 2003a, 2004), Africa(Weldon & du Preez 2004), and Australia (Berger et al.1998). Chytridiomycosis has been classified as an emerg-ing infectious disease of amphibians (Daszak et al. 1999).Because chytridiomycosis is associated with amphibianpopulation declines, it is listed as a key threat in Australiaunder the Environment Protection and Biodiversity Con-servation Act 1999 (Speare et al. 2001).

Within the wet tropics region of northeastern Queens-land, at least eight species of stream-dwelling frogs haveexperienced population declines (McDonald & Alford1999; Table 1). One species, the green-eyed treefrog(Litoria genimaculata), has recovered, but seven oth-ers are still listed as endangered (Northern QueenslandThreatened Frogs Recovery Team 2001; Table 1). Theendangered frogs have greater fidelity toward streamhabitats than do common species (McDonald & Alford1999). They also have lower fecundity and greater habi-tat specialization than average for rainforest frogs withaquatic larvae (Williams & Hero 1998). At sites monitoredin the wet tropics following the population crashes ofadults, tadpoles continued to survive and metamorphose(Richards et al. 1993; McDonald & Alford 1999). Whereendangered frogs are extant in stable numbers, they arerestricted to low-elevation (< 400 m) areas of their his-torical ranges (McDonald & Alford 1999). Frogs infectedwith Bd have been found in rainforest streams of tropicalnorthern Queensland (Speare & Berger 2000; Table 1).

To determine the current extent and prevalence ofchytridiomycosis, we monitored frog assemblages asso-ciated with rainforest streams (Table 1). We developed

five hypotheses regarding the patterns we expected toobserve and evaluated their fit to the data: (1) body con-dition (proportion of weight to length) of adult frogs iscorrelated with infection prevalence; (2) level of fluctu-ating asymmetry (developmental instability, Alford et al.1999) is correlated with infection prevalence; (3) infec-tion prevalence is density dependent and therefore frogdensity is correlated with infection prevalence; (4) meta-morphosing tadpoles provide an influx of susceptibleor infected hosts, resulting in increased infection preva-lence; and (5) environmental conditions strongly influ-ence infection prevalence, and this increases with coolerconditions in the winter and at high elevations. By testingthese hypotheses we developed a better understanding ofthe underlying determinants of chytridiomycosis at bothlandscape and population levels.

Methods

Frog Surveys

We surveyed three high- and three low-elevation rainfor-est stream sites at three latitudes in northeastern Queens-land (Table 2) for frogs and tadpoles. Between August2000 and March 2003 we surveyed each site twice eachyear during the warm, wet season (November to March)and twice each year during the cool, dry season (Mayto September). With three exceptions, the sites werewithin the historical ranges of all the frog species westudied (Cogger 2000; Northern Queensland ThreatenedFrogs Recovery Team 2001). The Australian lace-lid (Nyc-timystes dayi) has not been recorded from Ethel Creek(lowlands), and the common mistfrog (L. rheocola) hasnot been recorded from Ethel and Birthday creeks. Ateach survey we sampled 400-m sites (sections of stream)for 2 consecutive nights for adult frogs and on 1 dayfor tadpoles. During the nightly surveys weather condi-tions were recorded, and for each individual observed,the species, sex, snout-vent length, weight, position alongtransect, microhabitat type, temperature, and relative hu-midity were recorded. Following the protocol suggestedby Alford et al. (1999) for analysis of fluctuating asym-metry we measured the tibia-fibula segment of each hindlimb of adult frogs three times, resetting the calipers be-tween each measurement. We gave each frog an iden-tification number, clipped one or two toes in a pattern

Conservation BiologyVolume 19, No. 5, October 2005

Woodhams & Alford Ecology of Chytridiomycosis 1451

Table 1. Status and historical declines of frogs of rainforest streams in the wet tropics bioregion of Queensland, Australia.a

Common name Species Current status Decline description Decline period

Green-eyed treefrog Litoria genimaculatab common declines at high elevations 1990–1994subsequent recovery

Stoney-creek frog Litoria lesueurib common no declinesc n.a.Armoured mistfrog Litoria lorica endangered disappeared throughout range 1991Waterfall frog Litoria nannotisb endangered disappeared above 400 m, 1989–1993

stable at low elevationsMountain mistfrog Litoria nyakalensis endangered disappeared throughout range 1990Common mistfrog Litoria rheocolab endangered disappeared above 400 m, 1989–1994

possibly recoveringGreat barred frog Mixophyes shevilli common no declines n.a.Australian lace-lid Nyctimystes dayib endangered disappeared above 400 m, 1989–1994

stable at low elevationsSharp-snouted dayfrog Taudactylus acutirostrisb endangered disappeared throughout range 1989–1994Northern tinker frog Taudactylus rheophilus endangered disappeared throughout range 1989–1991

aData from Ingram & McDonald 1993; McDonald & Alford 1999; Northern Queensland Threatened Frogs Recovery Team 2001; n.a. in declineperiod, not applicable.bSpecies previously reported susceptible to chytridiomycosis (Speare & Berger 2000).cDeclines only in Blackall and Conondale ranges south of wet tropics 1978–1984 (Ingram & McDonald 1993). Recent genetic research suggeststhat frogs in the Litoria lesueuri species complex within the range of this study are most likely L. jungguy or L. wilcoxii (Donnellan & Mahoney 2004).

unique to each survey site, and preserved the toe clips in70% ethanol for histology and skin infection diagnosis.

To avoid transmitting disease among frogs, we capturedindividuals in new plastic sandwich bags and weighed andmeasured them in the bags. For toe clipping, we used ei-ther a sterile scalpel blade or scissors sterilized by spray-ing with ethanol and flaming. We applied topical antisep-tic (Betadine) in rare instances when bleeding occurred.Following contact with frogs, we washed our hands thor-oughly with sand from the stream. Before moving to an-other site, we sterilized our rubber boots with bleach andthoroughly cleaned and dried nets and other equipment.These practices were adopted according to the HygieneProtocol for the Control of Diseases in Frogs (NSW Na-tional Parks and Wildlife Service 2000).

During daytime surveys for tadpoles we recorded thestream width every 50 m, weather conditions, and loca-tion along transect, water depth, temperature, and sub-strate for each group of tadpoles found. Some tadpoleswere collected in plastic bags and later euthanized in

Table 2. Descriptions of three upland and three lowland rainforest stream sites.a

Latitude Longitude Elevationb Areac

Site name (deg min sec) (deg min sec) (m) (m2)

Ethel Creek, Mt. Spec National Park NP S 18◦59′08′′ E146◦12′35′′ 160 7340Birthday Creek, Paluma State Forest S 18◦58′54′′ E146◦10′02′′ 800 3500Windin Creek North, Wooroonooran NP S 17◦21′56.7′′ E145◦42′53.7′′ 750 2502Frenchman Creek, Wooroonooran NP S 17◦18′32.3′′ E145◦55′15.8′′ 40 2984Kirrama Bridge 1, Kirrama State Forest S 18◦12′11′′ E145◦53′00′′ 100 2596Kirrama Bridge 11, Kirrama State Forest S 18◦12′55′′ E145◦47′48′′ 850 1878

aUpland and lowland sites are paired at each of three latitudes.bElevation measured at the beginning of the 400-m transect. Sites at 750–850 m were considered upland.cArea ( for frog density calculations) represents the 400-m transect length multiplied by the average width of the stream (measured every 50 m).

chlorbutanol (at a concentration of 7 g/L water) and pre-served in 70% ethanol for determination of developmentalstages and infection status.

Histology

Histology and diagnosis methods followed Berger et al.(2000). We made a mean of 37 (range 6 to 105) serial ver-tical cross sections (5 µm) from each toe, and a mean of 11(range 6 to 78) serial sections from each tadpole mountedwith the mouthparts downward. We diagnosed infectionstatus on sections only if keratin was visible around theteeth of tadpoles or if the keratinized layer of skin on thetoes was present. All sections were scanned at 150x mag-nification and suspicious areas, including rough areas orareas of hyperkeratosis, were examined at 600x magnifi-cation. In total, five trained examiners diagnosed 44,258sections from samples collected during this project from765 adult frogs and 780 tadpoles. For consistency, oneexaminer (D.C.W.) examined some sections from each

Conservation BiologyVolume 19, No. 5, October 2005

1452 Ecology of Chytridiomycosis Woodhams & Alford

individual frog and tadpole sampled. We classified eachsample as positive, negative, or suspicious for infectionwith Bd. At least two trained examiners confirmed eachpositive included in analyses of infection prevalence.

Statistical Analyses

We first examined patterns of infection prevalence andecological data by visual inspection of summary statisticsgenerated in Microsoft Excel 2000. We used SPSS (ver-sion 10, SPSS Inc., Chicago, Illinois) for most statisticaltests and StatXact (version 4, Cytel Sofware, Cambridge,Massachusets) for determining 95% binomial (Clopper-Pearson) confidence intervals of sample infection preva-lence. We compared infection prevalence among adults offive species and tadpoles of four species with chi-squaredcontingency tests in Microsoft Excel 2000.

We used binomial logistic regression in SPSS to exam-ine the effects of seven independent variables and pair-wise interactions on the probability of infection. We in-cluded five categorical variables in the analysis: species(green-eyed treefrog, stoney-creek frog, waterfall frog,common mistfrog, and Australian lace-lid; Table 1), year(2000, 2001, 2002), elevation (upland or lowland; Table2), season (wet or dry), and life-history stage (adult ortadpole). Latitude and longitude were included as con-tinuous variables. We did not include variables exclusiveto one life-history stage (i.e., sex, body condition, fluc-tuating asymmetry). Main-effect variables were excludedfrom the model if the Wald statistic indicated that theydid not significantly predict the probability of infectionstatus (p > 0.05). We then added each pair-wise inter-action to the model individually and tested each interac-tion for significance. Finally, we added all significant vari-ables into the model, tested for overall model significancewith a model chi-square test, and used the Hosmer andLameshow goodness-of-fit test to determine whether themodel’s predicted values fit the observed data. As a mea-sure of effect size, we used Nagelkerke’s R2 to estimatethe percentage of variation explained. This is not a defini-tive test of hypotheses; rather, it is an exercise in modelbuilding to see how our data could best be explained bythe variables we believed on theoretical grounds couldbe important.

Parametric statistical tests generally have more powerto detect departures from the null hypothesis (Sokal &Rohlf 1981) and were used wherever possible. However,we used a nonparametric analog if the distribution of dataerror was not sufficiently known, if variances in the er-ror were not homogeneous as indicated by Levene’s testin SPSS, or if categorical, count, or if derived (i.e., pro-portional) data were used that would cause a violationof the assumptions of parametric analysis. We used themost powerful test that met the assumptions in each case,even if this meant using a nonparametric test for overall

differences among categories and parametric tests of dif-ferences between pairs.

To determine whether body condition as a proxy forindividual health was related to infection prevalence,we compared weight-to-length ratios (McDonald 1990;Brown & Brown 1998) between infected and uninfectedindividuals with an independent samples t test. We alsotested for larger-scale effects with a correlation analysisbetween infection prevalence and mean body conditionacross samples with Kendall’s tau. We used Kendall’s taubecause it does not require a linear relationship betweenvariables or bivariate normal distributions and weightseach difference in ranks equally, and thus it may be betterthan Spearman’s correlation when there are many tieddata (Sokal & Rohlf 1981).

To determine whether infection prevalence was posi-tively associated with levels of fluctuating asymmetry, wemeasured levels of fluctuating asymmetry of hind limbsaccording to Palmer and Strobeck (1986, 2003) and Alfordet al. (1999) with SPSS. We tested for correlation betweenfluctuating asymmetry and infection prevalence of sam-ples.

If disease transmission was density dependent, or ifhost density affected some other aspect of the host–disease interaction, there should be a correlation betweeninfection prevalence and host density. We tested for thepresence of such correlations with Kendall’s tau betweeninfection prevalence and frog density in samples for allspecies combined and for each species individually.

With the reorganization of the immune system at meta-morphosis, frogs may be particularly vulnerable to infec-tion (Carey et al. 1999). To determine whether infectionprevalence increased in populations at times when meta-morphosis was occurring, we examined the correlationbetween the infection prevalence in each sample and thenumber of metamorphs in that sample with Kendall’s tau.We compared the infection prevalence in metamorphswith that in adults for each species with independent ttests.

Experiments on the biology of Bd (Woodhams et al.2003; Piotrowski et al. 2004) indicate that its growthand development and the progress of chytridiomycosisin frogs are affected by temperature and humidity. Thesefactors vary with site, elevation, and season and throughtime. We investigated how infection prevalence variedwith environmental conditions by examining relation-ships between infection prevalence and environmentaldata taken from National Centers for Environmental Predi-cation/National Center for Atmospheric Research Reanal-ysis Project data provided by the Climate Diagnostics Cen-ter (Boulder, Colorado) (http://www.cdc.noaa.gov/cdc/reanalysis/reanalysis.shtml). We downloaded monthlymean values of maximum and minimum air temperaturesand precipitation rate at ground level from a 210 × 210km grid area centered over Tully, Queensland (145.5◦E,17.5◦S). These data were converted from “netCDF” files to

Conservation BiologyVolume 19, No. 5, October 2005

Woodhams & Alford Ecology of Chytridiomycosis 1453

Figure 1. Prevalence ofBatrachochytrium dendrobatidisinfection of adult and tadpole stages offrogs of rainforest streams in theQueensland wet tropics. ∗Forty-eightpercent of Litoria rheocola tadpoleswere reported with infections from thesame site and time period as this study(Retallick 2002).

“csv” files with the program R (version 1.4.1, The R Devel-opment Core Team, http://www.r-project.org/), and thedata were then manipulated in Excel. We expected preva-lence to be higher at times and places with temperaturesclose to the optimal growth conditions for the amphibianchytrid in culture (17◦–25◦ C; Piotrowski et al. 2004).

Results

Patterns of Bd Infection Prevalence

At least some individuals of all species were infected withBd. Infected individuals of some species were also foundat every site. Infection prevalence varied substantiallyamong species and between adult and tadpole stages (Fig.1). The mean detection efficiency (the proportion of sec-tions in which Bd was detected if it was detected in atleast one section) for Bd in tadpole mouthparts was 0.77and the minimum was 0.11. For adults the mean detec-tion efficiency was 0.70 (minimum 0.06). Infection preva-lence in adult frogs and tadpoles differed significantlyamong species (adults: χ2

4 = 14.439, p < 0.05; tadpoles:χ2

3 = 312.88, p < 0.001). Combining individuals withinspecies across sites and times showed that mean adult in-fection prevalence ranged from 10.0% in Australian lace-lids to 0.75% in waterfall frogs. In tadpoles, mean preva-lence ranged from 76.0% in great barred frogs (Mixophyesshevilli) to 1.09% in stoney-creek frogs. Because of thisvariation it is not appropriate to compare the grand meanof adults with the grand mean of tadpoles of all speciescombined; larger sample sizes from species with highprevalences would increase the mean, and lower samplesizes would decrease the mean. The same is true withinspecies in which infection prevalence varies widely withseason or other factors, although in this case the numberswere not so disparate as to become meaningless whenaveraged and compared with other species. One species,green-eyed treefrog, occurred and was abundant across

all the sites and was used to examine patterns of infectionprevalence in space and time.

Binary Logistic Regression

Significant variables in the logistic regression model in-cluded season, life-history stage, species, elevation, andyear (Table 3). Latitude and longitude were excluded. Ofthe 10 pair-wise interactions tested, 3 were statisticallysignificant: species by season, life-history stage by season,and life-history stage by species (Table 3). Odds ratios in-dicated that the odds of observing an infected frog were4.3 times greater in the cool, dry season than in the warm,wet season, and season also interacted with species andlife-history stage. Tadpoles had a 2.8 times higher likeli-hood of infection than adults, although life-history stageinteracted with species. The odds of infection at low ele-vations were 0.39 times that at high elevations. Accordingto Nagelkerke’s R2, 24.6% of the variation was explainedby the whole model. This model was significant (modelchi-square test, χ2

16 = 102.529, p < 0.001) and showed noevidence of lack of fit (Hosmer and Lameshow goodness-of-fit test, χ2

8 = 2.168, p = 0.975).

Hypothesis Tests

The sizes of infected and uninfected adult, male green-eyed treefrogs were not significantly different (Mann-Whitney U tests, p > 0.05). Because body condition (theratio of weight to snout-vent length) varied among species(Kruskal Wallis test, χ2

4 = 293.61, p < 0.001), all testson the relationship between body condition and infec-tion status were done within species. Body conditiondid not differ significantly between infected and unin-fected green-eyed treefrogs, stoney-creek frogs, waterfallfrogs, common mistfrogs, or Australian lace-lids (Mann-Whitney U tests, p > 0.05). Green-eyed treefrog sampleinfection prevalence did not significantly correlate withsample mean body condition (Kendall’s tau, r22 = 0.179,p = 0.265). Body condition did vary seasonally. Among

Conservation BiologyVolume 19, No. 5, October 2005

1454 Ecology of Chytridiomycosis Woodhams & Alford

Table 3. Binary logistic regression model examining the effect of independent variables on the probability of infection with Bd (Batrachochytriumdendrobatidis).

Variable Coefficient SE Walda df p Odds ratio

Life-history stage 1.020 0.312 10.690 1 0.001 2.774Species 12.375 4 0.015

Litoria wilcoxii/jungguy −2.225 0.726 9.393 1 0.002 0.108Litoria nannotis −1.467 0.604 5.896 1 0.015 0.231Litoria genimaculata −1.477 0.581 6.469 1 0.011 0.228Nyctimystes dayi −2.291 0.858 7.130 1 0.008 0.101

Season 1.458 0.354 16.958 1 <0.001 4.299Elevation −0.948 0.339 7.793 1 0.005 0.388Year 6.557 2 0.038

2000 1.130 0.443 6.511 1 0.011 3.0952001 0.397 0.326 1.483 1 0.223 1.488

Intercept −2.144 0.602Univariate model 38.630 9 <0.001

Species × season 19.150 4 0.001Life-history stage × season 6.819 1 0.009Life-history stage × species 27.527 2 <0.001Whole modelb 102.529 16 <0.001

aModel chi-square test used for univariate and whole models.bBuilt by adding significant pair-wise interactions to the best univariate model.

five sample groups of male green-eyed treefrogs measuredfrom Birthday Creek between March 2001 and Febru-ary 2003, condition differed significantly and was betterduring the warm, wet season than the cool, dry season(ANOVA, F 4,62 = 3.889, p = 0.00698; Fig. 2). Althoughbody condition varied with season, fluctuating asymme-try did not. Fluctuating asymmetry was usually signifi-cantly > 0 and was always < 0.26 mm, or 0.62% of thelength of the hind limb for green-eyed treefrogs at Birth-day Creek. We did not find any statistically significantdifferences in the fluctuating asymmetry of hind limblengths among species, among samples (samples for fluc-tuating asymmetry analysis include a single species mea-

Figure 2. Environmentalconditions and adult Litoriagenimaculata infectionprevalence in the Queenslandwet tropics from January 2000to July 2003. Units for the y-axisare in the key. Optimal growthtemperature of Batrachochytriumdendrobatidis is 17–25◦C(Piotrowski et al. 2004).

sured on a single stream survey), between upland andlowland elevations, or between infected and uninfectedfrogs. Mean fluctuating asymmetry with measurement er-ror extracted for each sample was not significantly cor-related with mean infection prevalence of each sample(Kendall’s tau, r33 = 0.2196, p = 0.104). The mean fluc-tuating asymmetry for all individuals combined (percentlimb length excluding measurement error) ranged from0.76% in Australian lace-lid males to 0.19% in waterfallfrog females.

The density of adult frogs (abundance per area ofstream) of most species was higher during the cool, dryseason surveys with the exception of Australian lace-lid,

Conservation BiologyVolume 19, No. 5, October 2005

Woodhams & Alford Ecology of Chytridiomycosis 1455

Table 4. Number of newly emerged metamorphs (juvenile frogs) in34 cool, dry season and 34 warm, wet season surveys.

Cool, Warm,Species dry season wet season

Litoria nannotis 18 12Litoria genimaculata 16 5Litoria rheocola 8 4Litoria wilcoxii/jungguy 5 1

Total 47 22

which was more common on transects during the warm,wet season. The green-eyed treefrog was the only speciesfound at all sites. Its density varied among sites (KruskalWallis test, χ2

4 = 10.979, p = 0.0268) and was higherduring surveys at upland sites than at lowland sites (inde-pendent samples t test, t25 = −2.072, p = 0.0487). Thestoney-creek frog occurred at all sites except Kirrama up-lands, which was the only upland site where waterfall frogoccurred. The common mistfrog occurred at one uplandsite, Windin North. The Australian lace-lid did not occurat any upland site. The density of frogs in a sample was notcorrelated with the prevalence of infection in the samplefor any species or for all species combined (Kendall’s tau,r25 = 0.0392, p = 0.792). Infected animals were not con-sistently present at particular locations along transects ondifferent sampling trips.

Juvenile frogs, or metamorphs, were present through-out the year, and emergence was not exclusively sea-sonal (Table 4). More newly emerged metamorphs ofeach species were found, however, during the cool, dryseason than the warm, wet season (paired samples ttest, t3 = 3.783, p = 0.032). The infection prevalenceof metamorphs did not differ significantly from that ofadults, and the number of metamorphs present in a sam-ple was not correlated with the sample infection preva-lence (Kendall’s tau, r30 = 0.0222, p = 0.8784).

Environmental conditions were correlated with infec-tion prevalence in adult green-eyed treefrogs, such thatinfection prevalence was higher during the cool, dry sea-son and at higher elevations (Fig. 2). Although humiditywas high throughout the year at stream sites, the optimal-temperature conditions for growth of Bd occurred moreoften in the cool, dry season and at higher elevations.

The prevalence of infection at sites was associatedwith elevation (Fig. 2). For each species and life-historystage that occurred at both the upland (750–850 m)and lowland site (40–160 m) at any latitude, infectionprevalence was always higher at the upland site. Sea-sonal effects on infection prevalence were also apparent(Fig. 2). Adults showed higher infection prevalence dur-ing the cool, dry season than during the warm, wet sea-son. Tadpoles showed the same pattern. Infection preva-lence was higher in tadpoles than in adults during the dryseason but was higher in adults than in tadpoles during

the wet season. There was also a pattern of higher in-fection prevalence in green-eyed treefrog adults over twocool, dry seasons than in three warm, wet seasons, andthis occurred at both upland and lowland sites, with in-fection prevalence consistently higher at the upland sites(Fig. 2).

Discussion

Disease Reservoirs

The effects of disease reservoirs on communities and eco-systems are well documented (Richards et al. 1999; Clea-veland et al. 2001; Gog et al. 2002; Haydon et al. 2002). Forexample, canine distemper virus is carried in domesticdogs and causes disease outbreaks in a number of threat-ened hosts such as African wild dogs (Lycaon pictus),black-footed ferrets (Mustela nigripes), and seals (Cleave-land et al. 2001). McCallum and Dobson (1995) suggestthat if a disease is affecting an endangered population,infection may be present in other hosts, including com-mon species. Infection should occur at higher prevalencein (resistant) species with stable populations than in (sus-ceptible) species in which infection leads to mortality andrapid removal from the population. Batrachochytriumdendrobatidis in the Australian wet tropics may use tad-poles as reservoir hosts with high infection prevalence atevery site. These reservoirs could provide the source forinfectious zoospores, which then infect hosts that are sus-ceptible to chytridiomycosis. In this way, high virulencecould occur in adults of some frog species (endangeredspecies) without inhibiting the evolutionary fitness of thepathogen at the time of disease emergence.

Batrachochytrium dendrobatidis infection of tad-poles may have sublethal effects and reduce host fitness(M. J. Parris, unpublished data), but disease-related mor-tality is rarely noted in tadpoles. Tadpoles of waterfallfrogs and great barred frogs may be reservoir hosts inthis system because they have infection prevalences of55% and 76%, respectively (Fig. 1), and appear to be re-sistant to disease. Common mistfrog tadpoles are also inthis group, with 33 (48%) of 69 tadpoles infected whencollected from lowland stream sites, including French-man Creek (Retallick 2002), which we also surveyed. De-formed mouthparts and other signs of disease were notnoted in tadpoles of any of the species examined, againsuggesting low pathogenicity in tadpoles, although de-formities have been noted in other species (Lips 1999;Fellers et al. 2001), and Retallick (2002) noted signs ofstarvation in moribund tadpoles.

Currently, amphibian chytrids are known to occur onlyon amphibian hosts. No nonamphibian hosts or restingspores have been found (Longcore et al. 1999; Morehouseet al. 2003). Bd sporangia can, however, survive up to 7weeks in lake water ( Johnson & Speare 2003) and thus

Conservation BiologyVolume 19, No. 5, October 2005

1456 Ecology of Chytridiomycosis Woodhams & Alford

can potentially survive for some time as saprobes. If Bdcan survive and reproduce on other hosts or materialsin the environment, such as fish or reptile scales, birdfeathers, or aquatic arthropods, these may also functionas disease reservoirs. This would drastically change ourunderstanding of the host–pathogen dynamics of this dis-ease.

Bd Infection Prevalence

In our surveys of frogs and tadpoles at rainforest streamsall the species we encountered were susceptible to in-fection by Bd. Histological studies indicate that ventralsurfaces and feet are most consistently infected in frogswith chytridiomycosis (Berger et al. 1999). Given that anaverage of 37 sections were examined per individual adultfrog and mean detection efficiency was > 70% for adultsand tadpoles, it is unlikely that many light infections weremissed if present on the tissues examined. Because serialsections should be autocorrelated and not independent,more rigorous analyses were not possible.

The univariate factors we studied that most influencedthe prevalence of infection included (in decreasing or-der of importance) season, life-history stage, elevation,species, and year. Additionally, seasonal effects interactedwith species and life-history stage, and life-history stageinteracted with species (Table 3). Species and life-historystages differed in habitat use, behavior, and immune func-tion (D.C.W. et al., unpublished data). These factors maycontribute to environmental effects that appear to be in-volved in disease emergence because the pathogen ismore prevalent at times and places of optimal growthconditions of temperature and limited rainfall but highhumidity

These results correspond to infection prevalence pat-terns found among adults of the same species at 30 sitesin the Australian wet tropics (K. R. McDonald, R. Speare,D. Mendez, and A. B. Freeman, unpublished data). In atranslocation experiment, Retallick (2002) collected mistfrog adults and tadpoles from lowland rainforest streamsites in North Queensland and relocated them to largefield enclosures with running stream water at both low-and high-elevation sites. These frogs were susceptible tochytridiomycosis, particularly at high elevations and dur-ing the cool, dry season. Batrachochytrium dendroba-tidis infections of adult frogs kept in the enclosures weremore prevalent at high elevations and during the cool,dry season.

Several other investigators have commented on theseasonal nature of chytridiomycosis. Berger et al. (2004)found higher incidence of fatal chytridiomycosis in east-ern Australia during the cool, dry season than during thewarm, wet season. This trend was repeated each year ofthe study between 1996 and 2000, and 53% of cases inwild frogs occurred in July or August. A study of Eungellatorrent frogs (Taudactylus eungellensis) and sympatric

stoney-creek frogs at Eungella National Park, Queenslandfound higher infection prevalence in the cooler months( June through August) each year between 1994 and 1998(Retallick et al. 2004).

Environmental conditions might act directly to affectpathogen growth or virulence on frogs or might act indi-rectly by changing the nutritional state or immune com-petence of frogs or altering population densities or op-portunities for disease transmission.

Body Condition and Fluctuating Asymmetry

Environmental and genetic stress can be estimated usingmeasures of fluctuating asymmetry (Leary & Allendorf1989; Tsubaki 1998; Palmer & Strobeck 2003). Body con-dition measures such as the ratio of weight to SVL may pro-vide a more sensitive indicator of stress for some speciesand circumstances (McCoy & Harris 2003). Variation ininfection prevalence might be due to genetic stress. In ad-dition, body condition and infection may be more directlycorrelated with level of fluctuating asymmetry (Moller1996a). We did not, however, find any consistent differ-ences in the fluctuating asymmetry of hind limb lengthsbetween species, samples, seasons, or elevations or anycorrelation with condition or infection status.

Body condition differed among species, samples, sea-sons, and elevations but was not correlated with infectionprevalence. These results suggest that although the nutri-tional status of frogs depended on season and elevation,these changes did not alter levels of fluctuating asym-metry or susceptibility to Bd infection. Conversely, theyalso suggest that infection with Bd in adult frogs does notresult in changes in fluctuating asymmetry or body condi-tion in the wild. Exposure of Hyla chrysoscelis and Bufowoodhousei tadpoles to Bd, however, causes an increasein development period, a decrease in metamorph bodymass, altered community interactions, and increased fluc-tuating asymmetry in fore and hind limbs (M. J. Parris,unpublished data). We analyzed fluctuating asymmetrybased on only one trait, hind limb length. It is possiblethat measurements of one trait are not sufficient for stud-ies of fluctuating asymmetry in frogs. Studies of otherspecies suggest that although some traits may show dif-ferences in fluctuating asymmetry, others do not (Moller1996b; Gallant & Teather 2001).

Host Density

The prevalence of some infectious wildlife diseases varieswith host density (Lavigne & Schmitz 1990; Swinton etal. 2001; Green et al. 2002). Density-dependent transmis-sion and spatial or temporal host population structuremay be an important component in models of pathogenswith direct life cycles (de Jong et al. 1995; Lipsitch et al.1995; Hess et al. 2001), such as the amphibian chytrid.We found no evidence, however, that host density affectsthe prevalence of chytridiomycosis.

Conservation BiologyVolume 19, No. 5, October 2005

Woodhams & Alford Ecology of Chytridiomycosis 1457

Metamorphosis

Metamorphosis may be a crucial point in the ecologyof chytridiomycosis. Individuals transform from tadpolesthat may be resilient to disease (Berger et al. 1999) intoterrestrial juveniles that can be highly susceptible. In atleast one species, Dendrobates tinctorius, Bd infectionscan persist through metamorphosis and cause mortalitywithin weeks of metamorphosis (Lamirande & Nichols2002). In other species it is possible that infections areeliminated by the loss of the keratinized mouthparts be-fore infection can occur on the developing keratinizedskin. Whether infections persist through metamorphosisor are eliminated, frogs appear to be particularly suscep-tible during and immediately after metamorphosis to in-fection from water or cohabiting tadpoles. Therefore, thepresence of emerging tadpoles could increase the risk ofdisease transmission to adult frogs.

Although our data show that the number of meta-morphs in a sample was not correlated to sample in-fection prevalence and metamorphs appeared in streamsthroughout the year, there was a tendency for more meta-morphs to appear during the cool, dry season. Thesemetamorphs, however, were not infected more often thanadults. In 56 L. rheocola metamorphs that had metamor-phosed in stream enclosures throughout the year at anearby study area, only one was probably infected (Retal-lick 2002). Many of these, however, died of chytridiomy-cosis after a few months within the enclosures (Retallick2002).

Host Behavior

Host frogs also exhibit a variety of behaviors such as so-cial congregations, seasonal and diel activity patterns, ag-gressive interactions, and contact with stream water (Mc-Donald & Alford 1999; Hodgkison & Hero 2001, 2002)that may result in variation in disease transmission effi-ciency. This may help explain the differences in infec-tion prevalence or conservation status among species.Retallick (2002) showed that adult and metamorph L.rheocola in field enclosures varied their behavior withseason. Frogs chose elevated perches more often in thewarm, wet season but chose aquatic microhabitats dur-ing the cool, dry season. This behavior may increase therisk of infection during the cool, dry season and corre-sponds with higher infection prevalence. Although manyecological variables are related to environmental condi-tions, the hypotheses that poor frog condition, stress thatcauses increased fluctuating asymmetry, high density, andmetamorphosis increase infection prevalence and poten-tially disease emergence were not supported by our data.This suggests that infection prevalence is actually affectedby environmental conditions and not some related factor.Species-specific differences in behavior or immune func-tion, however, may also influence host infection status.

Management Implications

The management of endangered rainforest stream frogsrequires more than knowledge of potential underlyingmechanisms of disease or past population declines. Anapproach that may be useful in the adaptive managementof these species and also a test of the theory that chytrid-iomycosis is the proximal cause of disease would be toattempt to prevent a decline in an experimental popula-tion (or cause reestablishment of endangered species) bytreating or removing the most highly infected groups orreservoir hosts in a population. In some upland streamswhere declines have occurred in the wet tropics, captureand treatment or removal of the large and long-lived tad-poles of the common species M. shevilli may be a viableexperimental management project.

In addition to helping understand the underlying mech-anisms of disease emergence, information about diseaseecology and the prevalence of infection in stream frogassemblages suggests that chytridiomycosis may regu-late frog populations and assemblage structure by caus-ing population declines or preventing reestablishmentof certain species. Variation in environmental conditionsprovides the primary explanation for patterns of infec-tion prevalence at the landscape level. Still needed is abetter understanding at the population level of species-specific behaviors and immune functions that influencehost-pathogen dynamics.

Acknowledgments

We are very grateful for the support of S. Townsend andfield volunteers B. Brinkman, M. Bauman, J. Cauchi, B.Hegedus, M. Kusrini, K. Lewthewaite, K. Mitchell, F. Mar-shall, E. Mitchell, T. Sawin, F. Sinnegar, and K. Trudgen.We thank A. Backer, S. Jellinek, T. Langkilde, and B. Mottfor field and laboratory assistance and S. Reilly and D.Mendez for histology and Bd diagnosis training. We thankR. Rowe for assistance with R software and L. Berger andR. Retallick for discussion of data in preparation for publi-cation. This research was carried out under animal ethicspermits granted by James Cook University (permits A451and A699-01). Field work was conducted under ScientificPurposes Permits and Take, Use, Keep or Interfere withCultural or Natural Resources Permits WISP00443102,WITK00441702, F1/000375/01/SAA, F1/000386/00/SAA, F1/000375/00/TA, N0/001257/96/SAC, and N0/001257/96/SAA, granted by the Queensland Environmen-tal Protection Agency and Queensland Parks and WildlifeService. Permits to collect biological or geologicalmaterial from Queensland state forest, timber reserves,and other state lands (1573, 1615, 1733) were grantedby the Queensland Forestry Service and Department ofPrimary Industries. This research was funded by a sub-contract to R.A.A. from Integrated Research Challenges

Conservation BiologyVolume 19, No. 5, October 2005

1458 Ecology of Chytridiomycosis Woodhams & Alford

in Environmental Biology (IRCEB) grant IBN-9977063from the U.S. National Science Foundation to J. P. Collins.D.W. was supported by Doctoral Research Scholarship,Supplementary Internal Research Account Scholarship,Sigma-Xi, The Scientific Research Society Grant-in-Aid-of-Research, and an International Post-Graduate ResearchScholarship from James Cook University.

Literature Cited

Alford, R., K. Bradfield, and S. Richards. 1999. Measuring and analysingdevelopmental instability as a tool for monitoring frog populations.Pages 34–43 in A. Campbell, editor. Declines and disappearancesof Australian frogs. Environment Australia, Canberra.

Berger, L., et al. 1998. Chytridioycosis causes amphibian mortality asso-ciated with population declines in the rain forests of Australia andCentral America. Proceedings of the National Academy of ScienceUSA 95:9031–9036.

Berger, L., et al. 2004. Effect of season and temperature on mortality inamphibians due to chytridiomycosis. Australian Veterinary Journal82:434–439.

Berger, L., R. Speare, and A. Hyatt. 1999. Chytrid fungi and amphibiandeclines: overview, implications and future directions. Pages 23–33in A. Campbell, editor. Declines and disappearances of Australianfrogs. Environment Australia, Canberra.

Berger, L., R. Speare, and A. Kent. 2000. Diagnosis of chytridiomycosis inamphibians by histologic examination. Zoo’s Print Journal 15:184–190.

Bosch, J., I. Martinez-Solano, and M. Garcia-Paris. 2001. Evidence of achytrid fungus infection involved in the decline of the common mid-wife toad (Alytes obstetricans) in protected areas of central Spain.Biological Conservation 97:331–337.

Bradley, G. A., P. C. Rosen, M. J. Sredl, T. R. Jones, and J. E. Longcore.2002. Chytridiomycosis in native Arizona frogs. Journal of WildlifeDiseases 38:206–212.

Brown, C. R., and M. B. Brown. 1998. Intense natural selection on bodysize and wing and tail asymmetry in Cliff Swallows during severeweather. Evolution 52:1461–1475.

Carey, C., N. Cohen, and L. Rollins-Smith. 1999. Amphibian declines: animmunological perspective. Developmental and Comparative Im-munology 23:459–472.

Cleaveland, S., G. R. Hess, A. P. Dobson, M. K. Laurenson, H. I. McCallum,M. G. Roberts, and R. Woodroffe. 2001. The role of pathogens inbiological conservation. Pages 139–150 in P. J. Hudson, A. Rizzoli, B.T. Grenfell, H. Heesterbeek, and A. P. Dobson, editors. The ecology ofwildlife diseases. Oxford University Press, Oxford, United Kingdom.

Cogger, H. G. 2000. Reptiles and amphibians of Australia. Reed NewHolland, Sydney.

Daszak, P., L. Berger, A. A. Cunningham, A. D. Hyatt, D. E. Green, andR. Speare. 1999. Emerging infectious diseases and amphibian popu-lation declines. Emerging Infectious Diseases 5:1–23.

de Jong, M. C. M., O. Diekmann, and H. Heesterbeek. 1995. How doestransmission of infection depend on population size? Pages 84–89in D. Mollison, editor. Epidemic models: their structure and relationto data. Cambridge University Press, Cambridge, United Kingdom.

Donnellan, S. C., and M. J. Mahoney, 2004. Allozyme, chromosomaland morphological variability in the Litoria lesueuri species group(Anura: Hylidae), including a description of a new species. AustralianJournal of Zoology 52:1–28.

Fellers, G. M., D. E. Green, and J. E. Longcore. 2001. Oral chytridiomy-cosis in the mountain yellow-legged frog (Rana muscosa). Copeia101:945–953.

Gallant, N., and K. Teather. 2001. Differences in size, pigmentation,and fluctuating asymmetry in stressed and nonstressed northernleaopard frogs (Rana pipiens). Ecoscience 8:430–436.

Gog, J., R. Woodroffe, and J. Swinton. 2002. Disease in endangeredmetapopulations: the importance of alternative hosts. Proceedingsof the Royal Society of London Series B 269:671–676.

Green, D. E., K. A. Converse, and A. K. Schrader. 2002. Epizootiologyof sixty-four amphibian morbidity and mortality events in the USA,1996–2001. Annals of the New York Academy of Sciences 969:323–339.

Haydon, D. T., S. Cleaveland, L. H. Taylor, and M. K. Laurenson. 2002.Identifying reservoirs of infection: a conceptual and practical chal-lenge. Emerging Infectious Diseases 8:1468–1473.

Hess, G., S. Randolph, P. Arnberg, C. Chemini, C. Furlanello, J. Har-wood, M. Roberts, and J. Swinton. 2001. Spatial aspects of diseasedynamics. Pages 102–118 P. J. Hudson, A. Rizzoli, B. T. Grenfell, H.Heesterbeek, and A. P. Dobson, editors. The ecology of wildlife dis-eases. Oxford University Press, Oxford, United Kingdom.

Hodgkinson, S., and J.-M. Hero. 2001. Daily behavior and microhabitatuse of the waterfall frog, Litoria nannotis in Tully Gorge, easternAustralia. Journal of Herpetology 35:116–120.

Hodgkison, S., and J.-M. Hero. 2002. Seasonal behavior of Litoria nan-notis, Litoria rheocola and Nyctimystes dayi in Tully Gorge, NorthQueensland, Australia. Pages 29–39 in A. Nattrass, editor. Frogs inthe community: proceedings of the Brisbane Symposium. Queens-land Frog Society, Brisbane.

Ingram, G. J., and K. R. McDonald. 1993. An update on the declineof Queensland’s frogs. Pages 297–303 in D. Lunney and D. Ayers,editors. Herpetology in Australia. Surrey Beatty and Sons, Sidney.

Johnson, M. L., and R. Speare. 2003. Survival of Batrachochytrium den-drobatidis in water: quarantine and disease control implications.Emerging Infectious Diseases 19:922–925.

Lamirande, E. W., and D. K. Nichols. 2002. Effects of host age on suscep-tibility to cutaneous chytridiomycosis in blue-and-yellow poison dartfrogs (Dendrobates tinctorius). Pages 3–13 in R. G. McKinnell andD. L. Carlson, editors. Proceedings of the sixth international sympo-sium on the pathology of reptiles and amphibians 2002. Universityof St. Paul Printing Services, Saint Paul, Minnesota.

Lavigne, D. M., and O. J. Schmitz. 1990. Global warming and increasingpopulation densities: a perscription for seal plagues. Marine Pollu-tion Bulletin 21:280–284.

Leary, R. F., and F. W. Allendorf. 1989. Fluctuating asymmetry as anindicator of stress: implications for conservation biology. Trends inEcology & Evolution 4:214–217.

Lips, K. R. 1999. Mass mortality and population declines of anurans atan upland site in western Panama. Conservation Biology 13:117–125.

Lips, K. R. 2004. Amphibian population declines in montane southernMexico: resurveys of historical localities. Biological Conservation119:555–564.

Lips, K. R., D. E. Green, and R. Papendick. 2003a. Chytridiomycosis inwild frogs from southern Costa Rica. Journal of Herpetology 37:215–218.

Lips, K. R., J. D. Reeve, and L. R. Witters. 2003b. Ecological traits predict-ing amphibian population declines in Central America. ConservationBiology 17:1078–1088.

Lipsitch, M., E. A. Herre, and M. A. Nowack. 1995. Host population struc-ture and the evolution of virulence: a “law of diminishing returns.”Evolution 49:743–748.

Longcore, J. E., A. P. Pessier, and D. K. Nichols. 1999. Batrachochytriumdendrobatidis gen. et sp. nov., a chytrid pathogenic to amphibians.Mycologia 91:219–227.

McCallum, H., and A. Dobson. 1995. Detecting disease and parasitethreats to endangered species and ecosystems. Trends in Ecology &Evolution 10:190–194.

McCoy, K. A., and R. N. Harris. 2003. Integrating developmental stabilityanalysis and current amphibian monitoring techniques: an experi-mental evaluation with the salamander Ambystoma maculatum.Herpetologica 59:22–36.

McDonald, K. R. 1990. Rheobatrachus Liem and Taudactylus Straughan

Conservation BiologyVolume 19, No. 5, October 2005

Woodhams & Alford Ecology of Chytridiomycosis 1459

and Lee (Anura: Leptodactylidae) in Eungella National Park, Queens-land: distribution and decline. Transactions of the Royal Society ofSouth Australia. 114:187–194.

McDonald, K., and R. Alford. 1999. A review of declining frogs innorthern Queensland. Pages 14–22 in A. Campbell, editor. De-clines and disappearances of Australian frogs. Environment Australia,Canberra.

Moller, A. P. 1996a. Parasitism and developmental instability of hosts: areview. Oikos 77:189–196.

Moller, A. 1996b. Sexual selection, viability selection, and developmen-tal stability in the domestic fly Musca domestica. Evolution 50:746–752.

Morehouse, E. A., T. Y. James, A. R. D. Ganley, R. Vilgalys, L. Berger, P.J. Murphys, and J. E. Longcore. 2003. Multilocus sequence typingsuggests the chytrid pathogen of amphibians is a recently emergedclone. Molecular Ecology 12:395–403.

Northern Queensland Threatened Frogs Recovery Team 2001. Recov-ery plan for the stream-dwelling rainforest frogs of the wet tropicsbiogeographic region of north-east Queensland 2000–2004. Reportto Environment Australia, Canberra. Queensland Parks and WildlifeService, Brisbane, Queensland.

NSW (New South Wales) National Parks and Wildlife Service. 2000.Threatened species management information circular no. 6: hygieneprotocol for the control of disease in frogs. NSW National Parks andWildlife Service, Sydney.

Palmer, A. R., and C. Strobeck. 1986. Fluctuating asymmetry: measure-ment, analysis, patterns. Annual Review of Ecology and Systematics17:391–421.

Palmer, A., and C. Strobeck. 2003. Fluctuating asymmetry analysis revis-ited. Chapter 17 in M. Polak, editor. Developmental instability (DI):causes and consequences. Oxford University Press, Oxford, UnitedKingdom.

Piotrowski, J. S., S. L. Annis, and J. E. Longcore. 2004. Physiology of Ba-trachochytrium dendrobatidis, a chytrid pathogen of amphibians.Mycologia 96:9–15.

Retallick, R. 2002. Using experimental translocations to learn aboutdeclines in Queensland’s frog populations. Final report. QueenslandParks and Wildlife Service, Atherton.

Retallick, R. W. R., H. McCallum, and R. Speare. 2004. Endemic infectionof the amphibian chytrid fungus in a frog community post-decline.PLoS Biology 2:e351.

Richards, A., J. Cory, M. Speight, and T. Williams. 1999. Foraging in apathogen reservoir can lead to local host population extinction: acase study of a Lepidoptera-virus interaction. Oecologia 118:29–38.

Richards, S. J., K. R. McDonald, and R. A. Alford. 1993. Declines inpopulations of Australia’s endemic tropical rainforest frogs. PacificConservation Biology 1:66–77.

Sokol, R. R., and F. J. Rohl. 1981. Biometry. 2nd edition. W.H. Freemanand Company, San Francisco.

Speare, R. 2003. Summary of formidable infectious diseases of am-phibians. Amphibian Diseases Research Group, James Cook Uni-versity, Townsville, Australia. Available from http://www.jcu.edu.au/school/phtm/PHTM/frogs/formidable.htm (accessed March 2005).

Speare, R., et al. 2001. Nomination for listing of amphibian chytridiomy-cosis as a key threatening process under the Environment Protec-tion and Biodiversity Conservation Act 1999. Pages 163–187 in R.Speare and Steering Committee of Getting the Jump on Amphib-ian Disease, editors. Developing management strategies to controlamphibian diseases: decreasing the risks due to communicable dis-eases. School of Public Health and Tropical Medicine, James CookUniversity, Townsville, Australia.

Speare, R., and L. Berger. 2000. Chytridiomycosis in amphibians inAustralia. Amphibian Diseases Research Group, James Cook Uni-versity, Townsville, Australia. Available from http://www.jcu.edu.au/school/phtm/PHTM/frogs/chyspec.htm (accessed March 2005).

Swinton, J., et al. 2001. Microparasite transmission and persistence.Pages 83–101 in P. J. Hudson, A. Rizzoli, B.T. Grenfell, H. Heester-beek, and A. P. Dobson, editors. The ecology of wildlife diseases.Oxford University Press, Oxford, United Kingdom.

Tsubaki, Y. 1998. Fluctuating asymmetry of the oriental fruit fly (Dacusdorsalis) during the process of its extinction from the Okinawaislands. Conservation Biology 12:926–929.

Weldon, C., and L. H. du Preez. 2004. Decline of the Kihansi spraytoad, Nectophrynoides asperginis, from the Udzungwa mountains,Tanzania. FROGLOG 62:2–3.

Williams, S., and J. M. Hero. 1998. Rainforest frogs of the Australian wettropics: guild classification and the ecological similarity of declin-ing species. Proceedings of the Royal Society of London Series B265:597–602.

Woodhams, D. C., R. A. Alford, and G. Marantelli. 2003. Emerging dis-ease of amphibians cured by elevated body temperature. Diseasesof Aquatic Organisms 55:65–67.

Conservation BiologyVolume 19, No. 5, October 2005