The Case of the Golden Toad and the Harlequin Frog Author(S): J

Home , Glass frog, Golden toad, Myobatrachidae

Society for Conservation Biology

Amphibian Declines and Climate Disturbance: The Case of the Golden Toad and the Harlequin Frog Author(s): J. Alan Pounds and Martha L. Crump Source: Conservation Biology, Vol. 8, No. 1 (Mar., 1994), pp. 72-85 Published by: Wiley for Society for Conservation Biology Stable URL: http://www.jstor.org/stable/2386722 . Accessed: 20/03/2014 13:14

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Wiley and Society for Conservation Biology are collaborating with JSTOR to digitize, preserve and extend access to Conservation Biology.

http://www.jstor.org

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions ContributedPapers

AmphibianDeclines and Climate Disturbance:The Case ofthe Golden Toad andthe HarlequinFrog

J.ALAN POUNDS MonteverdeCloud ForestPreserve Tropical Science Center Apartado10165-1000 San Jose,Costa Rica MARTHA L. CRUMP* Departmentof Zoology Universityof Florida Gainesville,FL 32611, U.S.A.

Abstract: The endemic golden toad (Bufo periglenes) was Declinaci6n de anfibiosy perturbacionesclimaticas: El caso abundant in Costa Rica's MonteverdeCloud ForestPreserve del sapo dorado y la ranaharlequin in April-May 1987 but afterwardsdisappeared, along with localpopulations of theharlequin frog (Atelopus varius). We Resumen: El endemico sapo dorado (Bufo periglenes) era examine thepossible relationshipbetween these sudden de- abundante en la Reserva del bosque nuboso de Monteverde clines and unusually warm,dry conditions in 1987. For our en Costa Rica durante abril-mayo 1987 pero despues desa- analyses of local weatherpatterns, we define a 12-month parecio al mismo tiempoquepoblaciones locales de la rana (July-June)amphibian moisture-temperaturecycle consist- harlequin (Atelopus varius).Se examin6 la posible relacion ing offour periods: (1) late wet season; (2) transitioninto entre esta subita disminucion y las condiciones cclidas y dryseason; (3) dryseason; and (4) post-dry-season(early- secas de 1987. Para analizar los patrones locales de tiempo, wet-season)recovery. The 1986-1987 cyclewas theonly one se definio un ciclo de temperatura-humedadde 12 meses on record(of 20 analyzed) withabnormally low rainfall in (julio-junio) que consiste de cuatro periodos: (1) ultimos allfourperiods,and temperatureanomalies in 1987 reached meses de la estaci6n lluviosa; (2) transicion a la estacion recordhighs. Flow in local aquifer-fedstreams during the seca; (3) estacion seca; y (4) recuperacion despues de la dry season and post-dry-seasonrecovery period reached a estaci6n seca de la estacion lluviosa). El ciclo de record low. This climate disturbance,associated with the (principio el unico (de 20 analizados) que 1986-1987 El Nifio/SouthernOscillation, was more severe 1986-1987fue registrado recibio anormalmente en todos los cuatro than a similar eventassociated with the 1982-1983 El Nifo, precipitacion baja periodos,y las anomalias de temperaturasen 1987 fueron thoughthis earlier oscillation was the strongestof thepast las macsaltas registradas.Los caudales en las quebradas lo- century.Demographic data for one harlequin frogpopula- cales que se alimentan de manantialesfueron los mds bajos tion,gathered during thesetwo climatic events,support the registradosdurante la estaci6n seca y el periodo de recupe- hypothesisthat in 1987, shortlybefore the population col- racion. Este trastorno del clima, asociado con El Nifo/ lapsed, thefrogs underwent an unprecedentedshift in distri- Oscilaci6n del sur de 1986-1987, estuvo mcassevero que el eventoparecido asociado con El Nifo de 1982-1983, aun- * Ari- Currentaddress: Department of Biological Sciences, Northern que esta oscilacion anteriorfue la mds fuerte del ultimo AZ All zona University,Flagstaff 86011, US.A reprintrequests Datos una de ranas rec- shouldbe sentto thisaddress. siglo. para poblaci6n harlequines, olectados durantesestos dos trastornosdel la Papersubmitted November 20, 1992; revisedmanuscript accepted clima, apoyan May3, 1993. hipotesisque en 1987, poco antes de que la poblaci6n co-

ConservationBiology, Pages 72-85 Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions Pounds& Crump AmphibianDeclines and Climate Disturbance 73 bution within the habitat in response to desiccatingcondi- lapsarac las ranas cambiaron su distribuci6nen el habitat tions. The juxtaposition of these rare demographic events en respuestaa condiciones desecantes.La yuxtaposicion de suggeststhey were causally linked yet sheds little light on estos raros eventos demogrcficossugieren que estuvieron mechanisms underlyingthe sudden decline. While desicca- ligados causalmente,pero dicepoco de los mecanismosde la tion or directtemperature effects may have beenfactors lead- subita disminucion.Mientras que la desecacin y los efectos ing to high adult mortality,moisture-temperature condi- directosde la temperaturapudieron haber causado mortal- tions may have interactedwith some other,unidentified idad de adultos, tambien esposible que hubiera interaccion agent. We discuss two hypothesesconcerning possible syner- entrelas condicionesclimdticasy otro factor. Se discutendos gistic effects:In the climate-linkedepidemic hypothesis,mi- hipotesissobre una posible interaccion:En la hip6tesis de croparasitesare the additional agent. In the climate-linked epidemialigada al clima,los micropardsitosson elfactor adi- contaminantpulse hypothesis,atmospheric contaminants cional. En la hip6tesisde contaminaci6nligada al clima,con- scavengedby mist and cloud water in montane areas reach taminantesatmosfericos removidos por la llovizna y niebla critical concentrations when conditions are abnormally alcanzan concentracionescriticas cuando las condiciones warm and dry. son anormalmenteccilidas y secas.

Introduction seen outside theirbrief annual matingbouts (one to several 5-10-day periods fromMarch throughJune; Ja- In 1987, the golden toad (Bufo periglenes;Bufonidae), cobson & Vandenberg 1991; W. Guindon, personal endemic to elfincloud forestin Costa Rica's Cordillera communication)harlequin frogs had been abundantand de Tilaran,appeared safein the MonteverdeCloud For- conspicuous year-roundsince 1980, when we began est Preserve.During April-May, more than 1500 toads our observations(Crump 1986; Pounds & Crump 1987; gatheredto mate in temporarypools at Brillante,the Crump& Pounds 1989). Because annualadult survivor- principal known breeding site (Fig. 1; Crump et al. ship in bufonidsis generallymuch greater than 1% 1992). This spectacle had occurred annuallysince at (Kelleher & Tester 1969; Clarke 1977), the abruptna- least 1972, when the preservewas founded(W. Guin- tureof the declines suggestshigh adult mortality rather don,personal communication). But in 1988 and againin thanjust a lack of successfulbreeding and recruitment. 1989, onlya singletoad appeared at Brillante,and a few Environmentalfactors that could be implicated in othersgathered 4-5 km SE. A varietyof otheramphib- thesedeclines includeacid depositionor otherforms of ians in the area, includingthe harlequinfrog (Atelopus atmosphericcontamination, ultraviolet radiation, micro- varius; Bufonidae) and membersof six other families, parasites, and extreme weather conditions (Pounds became scarce at the same time. During 1990-1992, 1990, 1991). Althoughthe possibilityof an extraordi- despiteour intensivesurveys, no golden toads or harle- naryevent of acid precipitationin 1987 cannotbe ruled quin frogswere found. out, measurementsof pH in subsequentyears have re- Interestin thiscase has intensifiedwith the recogni- vealed no abnormalacidity (Crump et al. 1992; Clark, tion thatit is part of a global pattern(Barinaga 1990; personal communication).Moreover, while acidity is Blaustein& Wake 1990; Phillips1990; Vittet al. 1990; knownto killaquatic embryos(Pierce 1985) and affect Wyman1990; Wake 1991). Nevertheless,the manyre- the distributionof terrestrialadults (Wyman 1988; Wy- ports of amphibiandeclines have met with some skep- man & Jancola 1992), it has not been linked to high ticism. Because populations fluctuatenaturally (see adult mortality.The possibility,however, that other Bragg1960), an apparentdecline could be an artifactof kindsof atmospheric contamination were a factorin the short-termobservations. Pechmann et al. (1991) declines remains unexplored. Ultraviolet radiation pointedto largefluctuations in the abundanceof several could conceivablykill adult harlequin frogs, but golden species at a breedingpond in lowland South Carolina toads normallyhide in retreatsabout 95% of the time, over 12 years.A species was sometimesrare or absent emergingto breed beneaththe forestcanopy, typically one year and abundantthe next.Many populations that under heavy cloud cover. A varietyof microparasites have disappeared, however, including those at Mon- attackamphibians (such as viruses,bacteria, and proto- teverde,show no sign of recoveryafter several years. zoans; Hoffet al. 1984), yet it seems unlikelythat they Furthermore,studies of coastal plain populationsof the would suddenlybecome highlylethal unless theywere easternU.S. may have limitedrelevance to our under- interactingwith some otherfactor. standingof events in montaneand high-latituderegions, Crumpet al. (1992) focused on the possible role of where most amphibiandeclines have been observed unusual weather in the disappearance of the golden (Wake 1991). toad. Examiningprecipitation patterns, breeding pool Sightingsof both golden toads and harlequinfrogs at temperatures,and the timingof pool formation,they Monteverdedecreased by about 99% in the same year. hypothesizedthat warm, dry conditions since 1987 Whereas adult golden toads had not commonlybeen have discouragedbreeding attempts and thatthe toads

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions 74 AmphibianDeclines and Climate Disturbance Pounds& Crump

lar n 1_I A 600

S63 )r r 4Calr"S Uib

mar CCafibe

oS an os6

COST ICA Oc6-Io z\ Pacmico o S. .Oo mappea imeypp r fox r 0om 5 KmI Fo Claping

Figure 1. The Monteverderegion of Costa Rica's Cordillerade Tilardn Golden toads are known onlyfrom elfin forestalong the continentaldivide in and near theMonteverde Cloud ForestPreserve (Savage 1966). This habitat is characterizedby dwarfed,wind-sculpted trees heavily laden with epiphytes(Lawton & Dryer 1980) and belongs to the lower-montanerain-forest life zone of Holdridge (1967). Forestedand deforestedareas were mapped approximatelyfrom aerial photographs mightsimply be waiting in retreatsfor conditionsto lationsand the annual weather cycle, we ask if condi- improve.In view of more recent fieldwork, however, tionswere more extremein 1987 thanin otheryears on it seems unlikely that golden toads still survive in record. Our analyses emphasize the potential impor- large numbers.In thispaper, we focus on the possibil- tance of subterraneanwater, the source of springsand itythat warm, dry conditions in 1987 were a factorin seepages thatoften play a role in amphibianwater econ- high adult mortality,leading to the collapse of popula- omy.While the data span only 24 years,they cover the tions. 1982-1983 El Nifio,the strongestof the past century Whengolden toads and harlequinfrogs were lastseen (Cane 1983), and thus include some long-termex- in large numbersin 1987, weather at Monteverdewas tremes that are a good basis for comparisons.After- underthe influenceof the 1986-1987 El Niiio/Southem wards,we analyze harlequinfrog demographic data in Oscillation,which produced warm,dry conditions over lightof climaticpatterns and discuss possible mecha- muchof Costa Rica (Manso & Ramirez1987). Afterpro- nismswhereby climate disturbances could lead to am- vidingbackground on anuranmoisture-temperature re- phibiandeclines.

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions Pounds& Crump AmphibianDeclines and Climate Disturbance 75

AnuranMoisture-Temperature Relations near a smallaquifer-fed stream. The galleryforest along thisstream became extremelydry, and the frogsseldom Conditionsof the physical environment interact in ways venturedmore than 3 m fromthe water (Pounds & that make it difficultto isolate the effectof a single Crump1987; Crump& Pounds 1989). These terrestrial factoron anyspecies ofplant or animal.The interplayof frogs were poor swimmers and rarely entered the moistureavailability and ambienttemperature may be stream;they thus relied on moistureabsorbed from wet especiallycritical for amphibians.Because virtuallyall surfacesin the stream'ssplash zone. Because therewas physiologicalprocesses are temperature-sensitive,am- littlesurface runoff at thetime, stream flow reflected the bienttemperature is undoubtedlyan importantlimiting quantityof groundwater. As flowdeclined and moisture factorin itsown right,but itsimpact on watereconomy availabilitybecame patchy,the dispersionpattern of the mayalso be crucial. frogsbecame increasinglyclumped as theygathered in Mostamphibians readily lose wateracross theirmoist, the remainingwet areas, especially near waterfalls permeable skins,and the rate of loss is temperature- (Pounds & Crump 1987; Crump& Pounds 1989). dependent(Spotila 1972). A dehydratinganuran at high Althoughgolden toads do not gathernear aquifer-fed temperaturesmay be facedwith conflictingphysiologi- streamsin this manner,the water table in their elfin- cal demands.By increasingdermal mucous secretion,a foresthabitat often lies within 30-50 cm of the soil NorthAmerican bullfrog (Rana catesbeiana) keeps its surface(R. Lawton,personal communication). Seepages body temperaturebelow dangerous levels through are common.We have observed the toads coming and evaporativecooling, but at the same time accelerates going fromcrevices thatform near the roots of elfin- waterloss (Lillywhite1971). In a similarmanner, a giant foresttrees as theyare rocked by the tradewinds. The toad (Bufo marinus) in Panamamaintains its body tem- toads have also been excavated fromsmall tunnelsap- peraturebelow the daytimeambient temperature, yet parentlyleft by the decay oftree roots (R. Law,personal may die fromdehydration in 24-72 hours (depending communication).For toads in these subterraneanre- on body size) unless it has access to water (Zug & Zug treats,groundwater might ordinarily be an important 1979). source of moisture. To offsetevaporative losses, terrestrial frogs and toads absorbwater, mainly through hypervascularized skin in theventral pelvic region(Roth 1973). In the absence of TheAnnual Moisture-Temperature Cycle freewater, they may absorb moisturebound to the sub- strate,but the abilityto do so variesamong species and In Costa Rica,where geographiclocation relative to the depends on substratecharacteristics and moisturecon- centralaxis of mountainranges largely determines extent.In NorthAmerica, xeric-adapted spadefoot toads posure to seasonal air-flowpatterns, the climateregime (Scaphiopus), which regulateosmotic pressure of their on the Caribbeanslope differsfrom that on the Pacific body fluidsby storingurea, can absorb water at soil slope (Coen 1983). Because the Monteverde region moisturetensions up to. 15 atm (Ruibal et al. 1969). straddlesthe continentaldivide (Fig. 1), its climateis a Mesic-adaptedAmerican toads (Bufo americanus), on mix of these regimes. the otherhand, lose water to the soil at tensionsabove To quantifyseasonal patterns and compare them 1.5 atm (Walker & Whitford1970). among years,we analyzed rain-gaugeand stream-flow For terrestrialanurans in the seasonal tropics,prob- data fromboth the Caribbean and the Pacific slopes. lemswith water balance are mostlikely to occur during Rain-gaugedata were fromMonteverde (20 years) and the dryseason, when soils may develop moisturedefi- San Gerardo(14 years); stream-flowdata were fromLi- cits (Herrera 1985). High insolationand air tempera- bano (24 years) and Cairo (15 years;Fig. 1). Recording tures,associated with reduced cloud cover and low hu- stationson each slope were those nearest the golden midity,raise evapotranspirationrates and acceleratethe toad's habitat providing continuous, long-termdata. dryingof soils. In the cloud forestat Monteverde,where Standardrain gauges underestimatewindblown precip- the dryingeffect of frequent winds mayalso be a factor, itationyet provide data thatare suitablefor year-to-year soils can develop moisturedeficits in a matterof days comparisons.Stream-flow data give a more complete,if (N. Nadkarni,personal communication).An amphibian less localized,picture of hydrologicalflux. We also an- thatvacates a dryingretreat may increase its exposure alyzed air-temperaturedata (15 years) fromthe Mon- to high ambient temperaturesand desiccatingcondi- teverderecording station. tions. During November-April,when northeasterlytrade Wheresoil moisturedeficits prevail, subterranean wa- winds dominatethe weather, the Caribbeanside is wet- termay be an importantsource of moisture.During the ter than the Pacificside (Fig. 2a). As these moisture- 1983 dryseason, harlequin frogs at our studysite in the laden winds meet the Cordillerade Tilarianand flow headwatersof the Rio Lagarto(3 km west of the Mon- upward,they cool adiabatically,producing clouds, mist, teverde Preserve; Fig. 1) sat on wet rocks and logs and rain.Most of this advective precipitation falls on the

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions 76 AmphibianDeclines and Climate Disturbance Pounds& Crump

500A (Coen 1983). At many sites on the Caribbean slope, the A annual peak in rainfalloccurs at this time. That the peak occurs in October at San Gerardo reflects a 400 (Fig. 2a) Pacific influence. Likewise, the high November- December precipitation at Monteverde, compared to ' 300 sites furtherdown the Pacific slope, reflects a Caribbean influence. 200 During January-April,as cold frontsbecome less fre- I quent and the trade winds less intense, bouts of Carib- bean precipitation become increasingly uncommon, re- 100 sulting in relatively dry weather on both the Caribbean and the Pacific slope (Fig. 2a). Typically in April, local streams reach basal flow (Fig. 2b) and daytime temperatures reach their annual peak (Fig. 2c). 12 B Differences in rainfall between the two slopes are least pronounced during May-October (Fig. 2a), when passage of the intertropicalconvergence zone generates heavy convective rains throughout the region. Also, 0E 10 . i n o o "easterly waves" and tropical depressions produce advective precipitation sporadically during this period 4: (Herrera 1985). Heavy advective rains in September- October ("Pacific temporals"), which occur when cy- clonic winds associated with a Caribbean storm rotating near the Central American isthmus draw moisture off the Pacific, largely account for the annual rainfallpeak i 0 ( I and I V I (Coen 1983). a)Ca 2 During the dry season, an anuran's chances of avoid- ing dehydration may depend on more than just that 24 season's temperature and precipitation patterns. How 0O well the preceding wet season recharged groundwater, how quickly the dry season began, and how promptly early-wet-season rains bring recovery could all be critical. To take these factors into account, we define a 12-month,July-June amphibian moisture-temperature 0, 20 cycle consisting of four periods: (1) late wet season (July-October); (2) transitioninto the dry season (No- vember-December); (3) dry season (January-April); and (4) post-dry-season (early-wet-season) recovery Jan Mar May Jul Sep Nov (May-June). Figure2. The annual moisture-temperaturecycle in theMonteverde area Shaded are Ca- symbols for El Niiioand the 1986-1987Cycle ribbean-sloperecording stations, open ones for Pa- stations (see I and Values are cific-slope Fig. text). In the wet season during a typical El Niiio year, as at- means totals or + across-year (of monthly averages) mospheric pressures decrease over the Pacific and in- thestandard error variation. for between-year (A) crease over the Atlantic, the northeasterly trade winds Total (B) stream monthlyprecipitation. Average become unusually intense over Costa Rica, mimicking a maximum flow (discharge rate). (C) Averagedaily dry-season wind field pattern (Fernandez & Ramirez See text sizes. temperature. for sample 1991). These winds interferewith convective rains, especially on the Pacific slope, yet sometimes increase Caribbean(windward) slope and the continentaldivide. advective precipitation on the Caribbean slope. In the Some spills over to the upper Pacific(leeward) slope, dry season during El Niiio, temperatures are often un- but little reaches rain-shadowedareas furtherdown- usually high, reducing condensation rates and advective slope. precipitation while increasing rates of evapotranspira- This advective Caribbeanprecipitation peaks in No- tion. vember-Decemberwith the influxof Arcticcold fronts Duringthe 1986-1987 El Nifno,total precipitation for

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions Pounds& Crump AmphibianDeclines and Climate Disturbance 77

the July-June cycle in the Monteverde region was a riod ofhighest daytime temperatures, was similarto that record low on both the Caribbean and the Pacific slope duringthe 1982-1983 El Nifio(Fig. 3c). Yet excluding (Fig. 3a). Average stream flow during the dry season and April,which in 1987 was affectedby a cold front,this post-dry-season recovery period was likewise a record period was significantlywarmer than in 1983 (Mann- low on both slopes (Fig. 3b). Mean daily high tempera- Whitney U = 5396.5;p < 0.001). ture at Monteverde during March-June, the annual pe- For a closer look at dailyhigh temperatures, we plot- ted a timeseries of monthlymean anomaliesfor recent years (Fig. 4). This analysisrevealed two periods of 5000 strongpositive anomalies, signatures of the 1982-1983 A and 1986-1987 El Nifio events. The greatestpositive anomalies accompanied the 1986-1987 event. They - 4000 A were highestin March 1987, near zero in Aprilbecause of the cold frontat thattime, and thenstrongly positive 1988. 3000 0 throughFebruary A more detailed analysisof precipitationpatterns re- vealed thatthe 1986-1987 cycle was the only one on recordwith abnormally low precipitationin all fourpe- riods.This was true on both the Caribbeanand the Pa- ~ 31~~~~~0~~0 cificslope. On each slope, all othercycles analyzedhad I 20001000 I 0 I,, at least one period in which precipitationwas at or 0' B above the mean (one-samplet-tests; cx = 0.05). Only in 1986-1987 did the Monteverdearea experiencea weak followed an transitionto a i 80 late-wet-season, by abrupt harshdry season, followed by a delayedpost-dry-season recovery(Fig. 5). Monthlyrainfall and stream-flowdata illustratethe developmentof thisunusual cycle (Fig. 6). Convective U, 0 rains duringJuly-October were weak on both slopes, but more consistentlyso on the Pacificside. Advective Caribbean precipitation(November-April) was like- 24 , l . .I . I , . , . . . I . . , I . . wise low, except duringcold frontsin January and April. Because the wet season arrivedlate, rainfall was low in both and Basal streamflow occurred in 24 May June. May, ratherthan April as is usuallythe case, indicatingthat 0 24 water tables had continued to fall at a time when groundwateris normallybeing replenished. Stream flow o was a recordlow forMay at bothrecording stations and remainedsignificantly below averageuntil August 1987 (one-samplet-tests; p < 0.04). These analyseslead us to reject the null hypothesis thatthe 1987 dryseason was no more severe thanoth- , *,~ I , I I a a a ers on record. The of El Niiio have been 1974-'75 '78 -'79 '82-'83 '86 -'87 timing may important:the 1986-1987 eventaffected one complete July- JuneCycle July-Junecycle, whereas otherson record affectedpor- Figurelowest 3. values Variation on record. in moisture- (C) Average temperature daily maxi-conditionsof one or more cycles.Nevertheless, it seems that Tionsamong successive amphibianmoisture-tempera- timingalone cannot explain the severe impact of this ture cycles. Each cycle is 12 months (July-June); oscillation.Given the enormous thermalsignal of the symbols as in Figure 2. (A) Total precipitation per 1982-1983 El Niiio (Cane 1983), it is unclearwhy tem- cycle. (B) Average stream flow (discharge rate) peratureanomalies at Monteverdewere greaterduring for the second half of the cycle (January-June), the the 1986-1987 event. combined dry season, and post-dry-season recovery Examining temperaturesof golden toad breeding period Only the last 19 of the 24 years available pools, Crump et al. (1992) concluded that conditions for the Rio Canas are shown, but these include the after1987 were warmerthan in 1987. Our more de- lowest values on record. (C) Average daily maxi- tailed analyses,however, reveal that conditionswere mumnmum temperaturetemperaturefor for March-June, the period actuallywarmer in 1987 thanin subsequentyears (Figs. in each cycle with the highest daytime temperatures. 3c and 4). Crumpet al. (1992) based theirconclusion The average for 1991, not graphed, was 22.5?C. on low pool temperaturesrecorded duringthe early-

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions 78 AmphibianDeclines and Climate Disturbance Pounds& Crump

0 _>. I -

0 0

-2

Jul Jan Jul Jan Jul Jan Jul Jan Jul Jan Jul Jan Jul Jan Jul Jan 1983 1984 1985 1986 1987 1988 1989 1990 Figure 4. Anomalies of daytimeair temperaturesat Monteverdein recentyears. Each value is the deviation of the monthlymean of daily maximum temperaturesfrom the 15-yearaverage of thesemonthly means.

Aprilcold frontin 1987, when air temperaturesbriefly corded after 1987. This pattern accords with the results departedfrom the generalpattern of positive anomalies presented here. (Fig. 4). AlthoughCrump et al. (1992) did not empha- Because pools filled earlier in 1987 than in the sub- size the point,the pool temperaturesthey recorded in sequent three years, Crump et al. (1992) concluded that late Apriland May 1987 were higherthan any theyre- conditions in 1987 were wetter than years after 1987.

1982-83 1983-84 1984-85 1985-86 1986-87 1987-88 1988-89 500-

t, -

I00-500 D-

WT DR WT DR WT D R WT D R WT D R WT D R WT D R

Figure 5. Precipitationby period in recentJuly-Juneamphibian moisture-temperature cycles. The upper row is for the Caribbean slope, the lower row for thePacific slope. Bars give the average monthlyrainfall for each period of each cycle.Lines give the mean (+ standard error)across years,repeated for each cycle.W = late wet season; T = transitioninto the dryseason; D = dryseason; and R = post-dry-seasonrecovery.

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions Pounds& Crump AmphibianDeclines and Climate Disturbance 79

500/ .

400 - I I

300 I / * /

0 Q00 t2 -

B2Y B

0 q)

62 2O0--o

Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May

Figure 6 Month-to-month variation in precipitation and stream flow during the 1986-1987 cycle. Shaded symbols (left) are for the Caribbean slope, open ones (right)for thePacific slope. Squares are means (+ standard error)across years; circlesare for 1986-1987. (A) Total monthlyprecipitation. (B) Averagestream flow (dis- chargerate).

Our more detailed analyses,however, reveal thatcon- theless, it shows that the elfin cloud forestat Mon- ditionswere actuallydrier in 1987 thanin subsequent teverde,among the wettesthabitats in the area, did not and The in as years (Figs. 3a, 3b, 5). pools filled 1987 escape the disruptionof hydrologyassociated with the a resultof the April cold front.This kindof extratropical 1986-1987 El Niflo. disturbance,which increasescondensation rates and ad- The effectsof this climatic event on adult golden vective precipitation,is common duringthe northern toads are unknown,yet demographicdata forharlequin winter and spring but is otherwise unpredictablein frogssuggest important effects on adults.Before harle- is time.It not unusualfor golden toad breedingpools to quin frogsdisappeared from the Monteverdearea, they fill than did in and earlieror later they 1987, theyoften had been observed in the drainagesof the Rios Penias fillduring convective rather than advective rains (Jacob- Blancas, Guacimal,and Lagartos,from 700 m on both son & Vandenberg1991; W. Guindon,personal commu- slopes to 1700 m near the continentaldivide (Hayes et nication). al. 1989). The species's geographicrange is Costa Rica and westernPanama (Savage 1972). At our studysite in the headwatersof the Rio Lagarto,these diurnalfrogs BiologicalConsequences were active year-roundon rocks and logs near the stream (Pounds & Crump 1987; Crump & Pounds In 1987, golden toads emergedto breed duringthe tem- 1989). They ranged fromcommon to abundant(one porarydrop in temperaturesand bout of Caribbeanpre- frogper 12.5 m of streamto one frogper 0.496 m of cipitationassociated with the Aprilcold front(Figs. 4 stream)every year from1981, when we began our ob- and 6). Because of the warm, dry conditionsthat fol- servations,through 1987. lowed, however,the breedingpools dried,and virtually During the 1982-1983 El Nifio,as the stream ap- all eggs and tadpoles died (Crump et al. 1992). As dis- proached basal flow and temperaturesrose above nor- cussed earlier,this impact on earlylife stages probably mal,the observed density of frogs declined as manytook does not explainwhy the populationcollapsed. Never- refugein damp crevices (Crump & Pounds 1989). Be-

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions 80 AmphibianDeclines and Climate Disturbance Pounds& Crump cause the frogshidden in crevicesincluded a dispropor- 200 tionatenumber of females,the operationalsex ratio(in A exposed areas) was highlymale-biased. In 1987, we the March, sampled population during 150 when the warm, dry conditions associated with the O 0 1986-1987 El Nifiowere near theirpeak. The observed i ?@*? 0 *00? density(one frogper 0.496 m of stream)was a record high,4.4 timesgreater than predicted by a polynomial regressionmodel of seasonal variationin 1982-1983 (Fig. 7a). This was the last census before the popu- lationcrashed: in May 1988, densitywas at a recordlow z / ** ^?* 0 (one 40 m of and it had fallen ** frogper stream) byJune E to zero. (Harlequin frogsdisappeared at the same time fromthe drainagesof the Rios PefiasBlancas and Gua- 1.0 cimal; G. Bello, E. Cruz, and M. Fogden,personal communication.) Suspectingthat the highdensity in March 1987 might be a clue as to why the Rio Lagartopopulation had . * crashed,we consideredtwo hypotheses.First, if recruit- 0. 0 @0 ment were high in 1986, the population mighthave been in if con- unusuallylarge 1987. Second, warm,dry - | 0.8 ditionsin 1987 caused individualsto leave dryingcrev- * 0 ? 0' ices and gather in remainingwet areas closer to the stream, have been observabilitymight unusuallyhigh. 0 '. . .0. .. 0 Althoughwe have no data on 1986 recruitment,these hypothesesdiffer in theirpredictions concerning sex ratio. The first a male-biased predicts normal, opera- 0.69 tional sex ratio,while the second predictsone thatis less male-biased.To arriveat quantitativenull predictions,we again used a polynomialregression model of 0 10 20 30 40 seasonal variationin 1982-1983 (Fig. 7b). In March Oct'82 Dec'82 Mar'83 1987, the proportionof male frogs(0.293) was 60% Aug'82 May'83 lower than thatpredicted by this model. For.the first Census Number timeon record,females outnumbered males. igure 7. Seasonal variation in observed and These analysessupport the hypothesisthat harlequin density operational sex ratio of harlequin near the Rio were respondingin an un- frogpopulation frogs Lagarto theEl Nino/SouthernOscillation. Data are precedented fashion to the extreme moisture-tem- during conducted in 1982-1983 at the perature conditions shortly before the population from weeklysurveys Rio Lagarto site. (A) Number of per 200 m crashedand disappeared.The juxtapositionof these rare frogs of stream. Te curve is a eventssuggest they were causallylinked yet sheds little computer-generated for quadratic regression.The numberoffrogs observed lighton possible underlyingmechanisms. Likewise, the in late March 1987 (403) was too high to fit on this simultaneouscrash of golden toads and the local popu- graph. (B) Proportionthat were males. The com- lationsof other amphibianspecies suggeststhat all the puter-generatedcurve is for a 4? polynomial regres- declines were part of a singlephenomenon. sion. Theproportion of males in late March 1987 (0.293) was too low to fit on thisgraph. Discussion thatseveral years of dry weather culminating in a 1982- The El Nifio/SouthernOscillation affects weather on a 1983 drought(attributed to El Ninlo;Rasmusson & Wal- planetaryscale, sometimeswith devastatingbiological lace 1983) were responsiblefor various declines in Aus- consequences (Barber & Chavez 1983), yet its impact tralia.More case studies are needed to evaluate the on terrestrialand freshwaterecosystems are onlybegin- impactof these climateoscillations on amphibiancom- ning to be understood (Foster 1982; Grant & Grant munities. 1989). DifferentEl Niiio events could, depending on Many reported declines of amphibian populations theirtiming and magnitude,affect amphibians in differ- have takenplace on mountains(Corn & Fogleman1984; ent parts of the world. Osborne (1989) hypothesized Heyer et al. 1988; Czechura & Ingram 1990; Bradford

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions Pounds& Crump AmphibianDeclines and Climate Disturbance 81

1991), where climateis especiallyvulnerable to atmo- ment. Hypotheses concerning the role of moisture- sphericdisturbances. At middle latitudes,precipitation temperature conditions in this mortalitymay invoke di- increaseswith elevation;in the tropics,it peaks at mid- rect causation such as death due to desiccation or dle elevationson windward slopes (Coen 1983). Dis- temperature stress, or indirect causation involving an turbancesthat reduce rainfallor increase temperatures interaction with an additional factor. in montaneareas where amphibiansare adaptedto cool, wet conditionscould have severe consequences. MOISTURESTRESS HYPOTHESIS Many but not all such areas have been affected.On It is that rates of anuran water loss asso- Costa Rica's Cerro Chompipe (2200 m elevation),Ate- possible high ciated with conditions in 1987 lopus senex and Bufo holdridgeidisappeared at about warm, dry directly caused adult Several how- the same time as A varius and B. periglenes at Mon- high mortality. observations, cast doubt on this the Caribbean teverde,yet at Las Tablas (1800 m elevation) A chir- ever, hypothesis. First, affected El remained iquiensis and B. fastidiosus appear to be thriving(Bo- slope, although by Nifio, signifi- wetter than the Pacific 3 and lafios & Barahona, personal communication). One cantly slope (Figs. 6), yet from there as well possible explanationis thatsmall differencesin topog- harlequin frogs disappeared (M. Fog- with their raphycan translateinto big differencesin local climate den, personal communication). Compared the Rio and itsvulnerability to disturbance.Turbulence in mon- Pacific-slope counterparts,harlequin frogs along Peiias Blancas should have had more tane areas,for example, can produce local reversalsin (Fig. 1) opportu- nities to avoid did these wind direction(rotor effects);hence, prevailingwind dehydration. Why populations not show a with the on- directioncan differbetween sites separatedby only a graded response? Furthermore, set of wet-season rains in Pefias Blancas few kilometers(Coen 1983). Valley in leaf The decline of amphibiansat Monteverdecould be a 1987, red-eyed frogs (Agalychnis callidryas; at traditional sites but dis- repeatof history. In the case ofthe golden toad,perhaps Hylidae) appeared breeding thereafter com- a seriesof climate-induced extinctions, combined with a appeared shortly (M. Fogden, personal If these had survived the worst des- low potentialfor recolonization, had leftthe singlerelic munication). frogs did afterwards? population.On the otherhand, deforestation (Fig. 1) or iccating conditions, why they disappear otherhuman activities could be increasingthe impact of these disturbances.Although there are no detectable TEMPERATURESTRESS HYPOTHESIS trends of annual or mean decreasing precipitation While desiccating conditions were presumably uncom- streamflow rank > (Spearman correlations;p 0.05), mon after the onset of rains in 1987, strong positive minimumstream flow in has decreased May significantly temperature anomalies continued through February since 1966 in the Rio Canias rank (Spearman correlation, 1988 (Fig. 4). Unfortunately,little is known of the ther- = n = < There is a similarbut rs -0.73; 24;p 0.0001). mal tolerances of golden toads and harlequin frogs. It is trendfor the Rio Canlo statisticallynonsignificant Negro known, however, that the glass frog Centrolenella fleis- = n = 15; = A deteriorationof (rs -0.42; p 0.07). chmanni, which also declined sharply at Monteverde, local could be the of eco- hydrology reducing capacity has little ability to acclimate to temperature changes to buffernatural climate disturbances. systems (Brattstrom 1968). Narrowly adapted montane popula- Global could also be the warming increasing impact tions-especially small, genetically homogeneous ones of these disturbances Models of atmo- (Wyman 1991). restricted in altitudinal range-could be particularly circulation that spheric predict regional precipitation lacking in this kind of physiological plasticity will with the patterns change warming,though spatial (Brattstrom 1970). Such populations might be relatively resolutionof these is predictions poor (Schneider 1987; common on low tropical mountains such as the Cordil- & Woodwell El Nifio Houghton 1989). superimposed lera de Tilaran, where life zones are highly compressed on or altered higherglobal temperatures hydrological (the massenerhebung effect;Flenley 1974). cycles could result in moisture-temperatureextremes thatare unprecedentedin recenttimes. Also, by chang- CLIMATE-LINKEDEPIDEMIC HYPOTHESIS ing sea-surfacetemperatures and atmosphericcircula- tion,global warmingcould change the frequency,tim- In constructing hypotheses concerning adult mortality, ing,or intensityof El Nifioevents. a natural tendency is to focus on physiologically lethal conditions. Nevertheless, nonlethal but suboptimal conditions, which may affectbehavior, energy budgets, and HypothesesConcerning Mechanisms general vitality (including immune system response), could also lead to high adult mortality by altering the As discussed earlier,the abruptnature of the amphibian outcome of critical biological interactions. Canning et declines at Monteverdesuggests high adult mortality al. (1964) observed this sort of interplay in captive Eu- ratherthan just a lack ofsuccessful breeding and recruit- ropean toads (Bufo bufo) maintainedunder crowded

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions 82 AmphibianDeclines and Climate Disturbance Pounds& Crump conditions in England. Parasites, including the mi- at which agrochemicalsvolatilize from exposed sur- crosporidean protozoan Plistophora myotrophica, faces, increasingrates of input into the atmosphere caused high adult mortality, especially during warm (Hoff et al. 1992). Dryingof soils may also create a summers. Likewise, Brodkin et al. (1992) found that the suctiongradient that draws soil contaminantsto the sur- bacterium Pseudomonas aeruginosa caused high mor- face (the wick effect),where theycan then make their talityin leopard frogs(Rana pipiens) maintained under way into the atmosphere(Spencer & Cliath 1973). Sec- suboptimal conditions, including crowding and high ond, long periods with littleor no precipitationto re- temperatures. move toxic compoundsfrom the air would allow more Harlequin frogs illustrate how habitat patchiness re- timefor atmospheric concentrations to build (Glotfelty sulting from warm, dry conditions can cause a popula- 1978). Atmosphericresidence times of organiccontam- tion to shiftto a highly clumped dispersion pattern. To inants(assuming only chemical removal) mightrange a microparasite, such a shiftcould represent a decrease up to 70 days (Cupitt 1980). Third,bouts of finemist in interhost distance, facilitatingan epidemic (Dobson and cloud water deposition punctuatinga dry period & May 1986). Crump et al. (1992) speculated that a could effectivelyscavenge these contaminantsfrom the "non-species-specific pathogen" might have caused the atmosphereand deposit themwith minimaldilution in decline of amphibians at Monteverde. Alternatively,if montaneareas. Fog is much more efficientthan rainwa- the warm, dry conditions were a coupling factor,differ- ter at concentratingpesticides; for reasons stillpoorly ent microparasite species could have affectedthe differ- understood,droplets may contain residue levels up to ent amphibian populations. severalthousand times greater than predicted by Hen- ry's law, which describes the dissolutionof airborne in an ideal et al. CLIMATE-LINKEDCONTAMINANT PULSE HYPOTHESIS compounds liquid (Glotfelty 1987). Fourth,moisture deficits in cloud-forestsoils and epi- The possibility of a transient increase in the concentra- phytemats could diminishthe extent to which these tion of toxic environmental contaminants has not been toxic compoundsare dilutedonce theyarrive. Further- ruled out. It is clear that human activityin the lowlands more,between sporadic events of cloud water deposi- of Costa Rica affectsatmospheric chemistry in the high- tion, organic residues might become highlyconcen- lands. Each year in March-April, turbidityof the atmo- tratedon soil and plantsurfaces as moistureevaporates sphere increases, resulting in a visible haze at Mon- duringcloud dissipation(Glotfelty et al. 1987). Fifth, teverde. This condition has been attributed to partiallydehydrated amphibians that rehydratewhen temperature inversions at 2000-3000 m, combined toxic compounds are prevalentmight have relatively with wind erosion of soils and agricultural burning in low volumesof body fluids to dilutethem. A rehydrating the lowlands (Herrera 1985). Recent data from Mon- anuranmay absorb a volume ofwater equal to 60-70% teverde show that cloud water collected during these of its normalbody water content (Putnam & Hillman months contains abnormally high concentrations of ni- 1977). trates and phosphates (Clark & Nadkarni, personal com- Because of the prevalence of mist and cloud water munication). While this input of inorganic compounds deposition at high altitudes,this climate-linkedcon- probably does not explain Monteverde's amphibian de- taminantpulse hypothesisis consistentwith the mon- clines, it illustrates that long-range contamination does tanepattern of amphibian declines. The patchynature of occur. Precipitation at Monteverde might contain other, amphibiandeclines in montaneareas could relateto the more toxic, compounds, detectable only through spe- location of contaminantsources and the effectsof to- cific assays. pographyon wind currentsand cloud-flowpatterns. A potential source of long-range contamination in During a normal dry season, airborne contaminants Costa Rica is the massive and indiscriminate use of pes- mightarrive at sublethalbackground levels. During an ticides (insecticides, herbicides, fungicides), which are unusuallysevere dry season, a deadlypulse mightoccur. pervasive contaminants of air and water (Hartshorn et Iftarget compounds can be identified,their behavior in al. 1982; Hilje et al. 1987). Pesticides enter the atmo- the atmospherecan be modeled and their effectson sphere by driftduring application, wind erosion of de- amphibianscan be analyzed.Studies of thiskind of pos- posited residues, and volatilization (Seiber et al. 1989). sible synergismbetween climatedisturbance and envi- Once in the atmosphere, they are removed chemically, ronmentalcontamination might be an importantkey to by photolysis or reaction with atmospheric oxidants, or understandingthe global amphibiancrisis. physically, by dry deposition, adsorption onto aerosols, or rainfall(Glotfelty 1978; Cupitt 1980). A climate disturbance like that of 1986-1987 could Coda interact with atmospheric contamination in several ways to produce transiently lethal conditions for am- Since the completionof analysesfor this paper, condi- phibians.First, warm, dry conditions increase the rates tions at Monteverde have continued to be cooler and

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions Pounds& Crump AmphibianDeclines and Climate Disturbance 83 wetter than in 1987. El Nifio revisitedin 1991-1992 Brattstrom,B.H. 1968. Thermal acclimationin anuran am- (and is stillwith us in April1993), butwith significantly phibiansas a functionof latitudeand altitude.Comparative and 24:93-111. mildereffects than in 1986-1987. There is stillno sign, Biochemistry Physiology however,of eithergolden toads or harlequinfrogs. Brattstrom,B. H. 1970. Thermalacclimation in Australianam- phibians.Comparative Biochemistry and Physiology35:69- 103. Acknowledgments Brodkin,M.A., M.P. Simon,A.M. DeSantis, and K.J. Boyer. 1992. ofRana pipiens to doses of the bac- The MacArthurFoundation, Stanford University's Center Response graded teriumPseudomonas aeruginosa. Journal of Herpetology forConservation Biology, and the BrookfieldZoo pro- 26:490-495. vided financialsupport for J. A. Pounds; the National GeographicSociety supported the work of M. L. Crump. Cane, M. A. 1983. Oceanographicevents during El Nino. Sci- Personnelof the Tropical Science Centerand the Mon- ence 222:1189. teverdeCloud ForestPreserve with G. helped logistics. Canning,E. U., E. Elkan,and P. I. Trigg.1964. G. M. W. C. A. Plistophora my- Barboza, Bello, Fogden, Guindon, Lopez, otrophica spec. nov., causing high mortalityin the common Pereira,N. Obando, and M. Wainwrighthelped with toadBufo bufo L.,with notes on the maintenanceof Bufo and fieldwork. S. Barahona,E. Bello, G. Bello, F. Bolafios,K Xenopus in the laboratory.Journal of Protozoology11:157- 166. Clark,E. Cruz,M. Fogden,W. Guindon,R. Law, R. Law- Longino,and N. Nadkarnishared ton, J. unpublished Clarke,R. D. 1977. Postmetamorphicsurvivorship of Fowler's data and fieldnotes. J. Campbell provided weather data toad,Bufo woodhouseifowleri. Copeia 1977:594-597. fromthe Monteverde recording station. The Costa Rican electrical institute(I.C.E.) provided weather data for Coen, E. 1983. Climate.Pages 35-46 in D. H. Janzen,editor. Costa Rican natural of Chi- San Gerardo and hydrologicaldata fromCairo and Li- history.University Chicago Press, cago, Illinois. bano. S. Barahona,F. Bolafios,K Bolda, K Clark,G. Gor- man,W. Guindon,J. Harte,R. Ibafiez,D. and M. Lieber- Corn, P. S., and J.C. Fogleman. 1984. Extinctionof montane man, B. Meyers,R. Mummie,K Phillips,S. Rand, D. populationsof the northernleopard frog(Rana pipiens) in Robinson,J. Savage, S. Sargent,J. Vial, K Vliet,D. Wake, Colorado.Journal of Herpetology18:147-152. and R. valuable discussion.L. Mather Wymanprovided Crump,M. L. 1986. and site in a in D. Harrison Homing fidelity neotropical drew the map Figure 1; produced the frog,Atelopus varius (Bufonidae). Copeia 1986:438-444. remainingfigures. P. Feinsinger,M. Fogden,G. Gorman, F. Hensley,K Rehm-Switky,E. Vohman,and D. Wake Crump,M. L.,andJ. A. Pounds. 1989. Temporalvariation in the of a anuran. commentedon early drafts.A. Blaustein,M. Hayes,J. dispersion tropical Copeia 1989:209-211. S. and R. reviewedthe final Karr, Rand, Wymancritically Crump,M.L., F.R. Hensley,and K L. Clark. 1992. Apparent version.C. Rojas gave moral and logisticalsupport; she decline of the golden toad: Undergroundor extinct?Copeia and A. Pereiraedited the Spanishsummary. We dedicate 1992:413-420. thispaper to the memoryof Douglas C. Robinson.We that otherswill have the chance to know Costa Cupitt,L. T. 1980. Fate of toxic and hazardousmaterials in the hope air environment. EPA-600/S3-80-084.U.S. Rica's as well as did. Project Summary amphibians Douglas EnvironmentalProtection Agency, Environmental Sciences ResearchLaboratory, Research Triangle Park, North Carolina. LiteratureCited Czechura,G. V., and G.J.Ingram. 1990. Taudactylus diurnus and thecase ofthe Memoirsof the Barber,R. T., and F. P. Chavez. 1983. Biological consequences disappearingfrogs. Queens- ofEl Nifo. Science 222:1203-1210. land Museum 29:361-365. Dobson, A. P., and R. M. May. 1986. Disease and conservation. M. 1990. Where have all the Science Barinaga, froggiesgone? Pages 345-365 in M. E. Soule, editor. Conservationbiology: 247:1033-1034. The science of scarcityand diversity.Sinauer Associates, Sun- derland,Massachusetts. Blaustein,A. R, and D. B. Wake. 1990. Declining amphibian populations:A global phenomenon?Trends in Ecology and Fernindez,W., and P. Ramirez.1991. El Ninio,la oscilacion del Evolution5:203-204. sur y sus efectosen Costa Rica. Tecnologia en Marcha(Costa Rica) 11. D. F. 1991. Mass and extinctionin a Bradford, mortality high- editor.1974. Altitudinalzonation in Malesia.Mis- elevation of Rana muscosa of Flenley,J.R., population Journal Herpetol- cellaneous Series No. 16. of Hull of 25:174-177. University Department ogy Geography,Hull, Aberdeen, Scotland.

Bragg,A. N. 1960. Population fluctuationsin the amphibian Foster,R.B. 1982. Famine on Barro Colorado Island. Pages faunaof Cleveland county,Oklahoma during the past twenty- 151-172 in E. G. Leigh,Jr., A. S. Rand,and D. M. Windsor,ed- fiveyears. Southwestern Naturalist 5:165-169. itors.The ecology of a tropicalforest: Seasonal rhythmsand

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions 84 AmphibianDeclines and Climate Disturbance Pounds& Crump long-termchanges. Smithsonian Institution Press, Washington, Osborne, W. S. 1989. Distribution,relative abundance and D.C. conservationstatus of Corroboreefrogs, Pseudophryne cor- roboreeMore (Anura:Myobatrachidae).Australian Wildlife Re- Glotfelty,D. E. 1978. The atmosphereas a sink for applied search 16:537-547. pesticides.Journal of the Air Pollution Control Association 28:917-921. Pechmann,J. H. K, D. E. Scott,R. D. Semlitsch,J. P. Caldwell, L.J.Vitt, and J.W. Gibbons. 1991. Decliningamphibian popu- Glotfelty,D. E.,J. N. Seiber,and L. A. Liljedahl.1987. Pesticides lations:The problemof separatinghuman impacts from natu- in fog.Nature 325:602-605. ral fluctuations.Science 253:892-895.

Grant,B. R.,and P. R. Grant.1989. Evolutionarydynamics of a Phillips,K 1990. Where have all the frogsand toads gone? naturalpopulation: The large cactus finchof the Galapagos. Bioscience 40:422-424. Universityof Chicago Press,Chicago, Illinois. Pierce, B.A. 1985. Acid tolerancein amphibians.Bioscience Hartshorn,G., L. Hartshorn,A. Atmella,L. D. G6mez,A. Mata,L. 35:239-243. Mata,R. Morales,R. Ocampo, D. Pool, C. Quesada, C. Solera,R. G. A. Umafia,C. Villalobos,and R. Sol6rzano, Stiles,J. Tosi, Jr., Pounds,J.A. 1990. Disappearinggold. BBC Wildlife8:812- Wells. 1982. Costa Rica countryenvironmental profile: A field 817. study.Tropical Science Center,San Jose,Costa Rica. Pounds,J. A. 1991. Amphibianwatch: New clues in the case of Hayes, M. P., J.A. Pounds, and W.W. Timmerman.1989. An thedisappearing amphibians. Wildlife Conservation 94:16-18. annotatedlist and guide to the amphibiansand reptiles of Monteverde,Costa Rica. HerpetologicalCirculars 17:1-67. Pounds,J. A., and M. L. Crump. 1987. Harlequinfrogs along a montanestream: and therisk of Herrera,W. 1985. Clima de Costa Rica. EditorialUniversidad tropical Aggregation predation flies. 19:306-309. Estatala Distancia,San Jose,Costa Rica. by frog-eating Biotropica and S. S. Hillman.1977. of Heyer,W. R., A. S. Rand, C. A. Goncalvez da Cruz, and 0. L. Putnam,R.W., Activityresponses anuransto Peixoto. 1988. Decimations,extinctions, and colonizationsof dehydration.Copeia 1977:746-749. frogpopulations in southeastBrazil and theirevolutionary im- plications.Biotropica 20:230-235. Rasmusson,E. M., and J.M. Wallace. 1983. Meteorologicalas- pects of theEl Nifo/SouthernOscillation. Science 222:1195- Hilje, L., L. E. Castillo,L.A. Thrupp,and I. Wesseling.1987. El 1202. uso de los plaguicidas en Costa Rica. EditorialUniversidad Estatala Distancia,San Jose,Costa Rica. Roth,J.J. 1973. Vascularsupply to the ventralpelvic regionof anuransas related to water balance. Journalof Morphology Hoff,G. L., F. L. Frye,and E. R. Jacobson,editors. 1984. Dis- 140:443-460. eases of amphibiansand reptiles.Plenum Press, New York. Ruibal,R., L. Tevis,and V. Roig. 1969. The terrestrialecology Hoff,R. M.,D. C. G. Muir,and N. P. Grift.1992. Annualcycle of of the spadefoot toad Scaphiopus hammondii. Copeia polychlorinatedbiphenyls and organohalogenpesticides in air 1969:571-584. in southernOntario. 2. Atmospherictransport and sources. EnvironmentalScience and Technology26:276-283. Savage,J. M. 1966. An extraordinarynew toad (Bufo) from Costa Rica. Revistade Biologia Tropical 14:153-167. Holdridge,L. R. 1967. Lifezone ecology.Tropical Science Cen- San Costa Rica. ter, Jose, Savage,J.M. 1972. The harlequin frogs,genus Atelopus of Costa Rica and westernPanama. Herpetologica 28:77-94. Houghton,R.A., and G.M. Woodwell. 1989. Global climate ScientificAmerican 260:36-44. change. Schneider,S. H. 1987. Climatemodeling. Scientific American 256:72-79. Jacobson,S. K, and J.J.Vandenberg. 1991. Reproductiveecol- ogy of the endangeredgolden toad (Bufo periglenes).Journal Seiber,J. N., M. M. and J.E. Woodrow. 1989. Air- of Herpetology25:321-327. McChesney, borne residuesresulting from use of methylparathion, moli- and thiobencarbon rice in the Sacramento Cali- Kelleher,K E., and J.R. Tester. 1969. Homingand survivalin nate, Valley, fornia.Environmental and 8:577-588. the Manitobatoad, Bufo hemiophrys,in Minnesota.Ecology Toxicology Chemistry 50:1040-1048. Spencer,W. F., and M. M. Cliath.1973. Pesticidevolatilization soil. of Lawton,R., and V. Dryer. 1980. The vegetationof the Mon- as related to water loss from Journal Environmental teverdeCloud ForestPreserve. Brenesia 18:101-116. Quality2:284-289.

Lillywhite,H. B. 1971. Thermalmodulation of cutaneousmu- Spotila,J. R. 1972. Role of temperatureand water in the ecol- cus dischargeas a determinantof evaporativewater loss in the ogy of lunglesssalamanders. Ecological Monographs42:95- frogRana catesbeiana Zeit. Vergl.Physiol. 73:84-104. 125.

Manso,P., and P. Ramirez.1987. El Nifio1986-7. BoletinMe- Vitt,L.J., J. P. Caldwell, H. M. Wilbur,and D. C. Smith.1990. teorologicalMensual (Costa Rica) 2. Amphibiansas harbingersof decay. Bioscience 40:418.

ConservationBiology Volume 8. No. 1. March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions Pounds& Crump AmphibianDeclines and Climate Disturbance 85

Wake, D. B. 1991. Declining amphibianpopulations. Science Wyman,R L., editor.1991. Global climatechange and lifeon 253:860. earth.Chapman and Hall, New York Walker,R. F., and W. G. Whitford.1970. Soil waterabsorption in selected of anurans. capabilities species Herpetologica Wyman,R. L., and J.Jancola. 1992. Degree and scale of terres- 26:411-418. trialacidification and amphibiancommunity structure. Journal of 4:392-401. Wyman,R. L. 1988. Soil acidityand moistureand the distribu- Herpetology tion of amphibiansin fiveforests of southcentralNew York. Copeia 1988:394-399. Zug, G. R., and P. B. Zug. 1979. The marinetoad, Bufo mari- Wyman,R. L. 1990. What'shappening to the amphibians?Con- nus: A naturalhistory resume of nativepopulations. Smithso- servationBiology 4:350-352. nian Contributionsin Zoology 284:1-58.

-9w

114

ConservationBiology Volume 8, No. 1, March 1994

This content downloaded from 163.118.172.206 on Thu, 20 Mar 2014 13:14:15 PM All use subject to JSTOR Terms and Conditions