weather, hydroregime, and breeding effort influence ... hydroregime, and breeding effort inﬂuence...

Weather, hydroregime, and breeding effort influence juvenilerecruitment of anurans: implications for climate change

C. H. GREENBERG,1,� S. J. ZARNOCH,2 AND J. D. AUSTIN3

1Bent Creek Experimental Forest, USDA Forest Service, Southern Research Station, Upland Hardwood Ecology and ManagementRWU-4157, 1577 Brevard Road, Asheville, North Carolina 28807 USA

2USDA, Forest Service, Southern Research Station, 481 Grand Oak Way, Moore, South Carolina 29369 USA3Department of Wildlife Ecology and Conservation, University of Florida, 110 Newins Ziegler Hall, Gainesville, Florida 32611 USA

Citation: Greenberg, C. H., S. J. Zarnoch, and J. D. Austin. 2017. Weather, hydroregime, and breeding effort influencejuvenile recruitment of anurans: implications for climate change. Ecosphere 8(5):e01789. 10.1002/ecs2.1789

Abstract. Amphibians that primarily breed in ephemeral wetlands are especially vulnerable to climatechange because they rely on rainfall or temperature to initiate breeding and create suitable hydroregimes(water duration, timing, frequency, depth) for reproductive success. Hydroregime effects on reproductivesuccess are likely to differ among species because of differences in reproductive strategies: the length andtiming of breeding period, rate of larval development, and timing of metamorphosis. We applied an infor-mation-theoretic approach to 22 consecutive years of continuous amphibian trapping data at eight ephem-eral wetlands to test hypotheses regarding environmental (hydroregime, weather) and biological (adultbreeding effort) factors affecting juvenile recruitment (JR) by six focal species representing four reproduc-tive strategies. We hypothesized that (1) JR by species with similar reproductive strategies would be influ-enced by similar variables; (2) JR would be higher for all species when models encompassed the maximumtime span of potential tadpole occurrence and development; and (3) JR rates within individual wetlandsand breeding cycles would correlate most closely between species with similar breeding strategies. Thebest model for all focal species (except Scaphiopus holbrookii) encompassed the maximum time span andindicated that ≥1 hydroregime variable, total precipitation, or both were important drivers of reproductivesuccess; average air temperature was not. Continuous hydroperiod through peak juvenile emigration wasan important predictor of JR for species with prolonged breeding periods, slow larval development, and a“fixed” late spring start date for juvenile emigration (regardless of when oviposition occurred, or cohortage; Lithobates capito, Lithobates sphenocephalus), but not for species with rapid larval development and con-tinual emigration as cohorts complete metamorphosis (Anaxyrus terrestris, Anaxyrus quercicus, Gastrophrynecarolinensis, S. holbrookii). Total rainfall was positively associated with recruitment for most species; depthcharacteristics affected species differently. Annual JR was positively correlated among species with similarreproductive strategies. Our results indicate that weather and hydroregime characteristics interact withreproductive strategies that differ among amphibian species and influence reproductive plasticity, oppor-tunity, and success. Effects of altered weather patterns associated with climate change on amphibian repro-ductive success may correspond more closely among species having similar reproductive strategies, withcritical implications for population trends and assemblages.

Key words: amphibian recruitment; Anaxyrus quercicus; Anaxyrus terrestris; anuran reproduction; breeding strategy;climate and amphibians; ephemeral wetland; Gastrophryne carolinensis; hydroregime; Lithobates capito; Lithobatessphenocephalus; Scaphiopus holbrookii.

Received 14 March 2017; accepted 20 March 2017. Corresponding Editor: Debra P. C. Peters.Copyright: © 2017 Greenberg et al. This is an open access article under the terms of the Creative Commons AttributionLicense, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.� E-mail: [email protected]

❖ www.esajournals.org 1 May 2017 ❖ Volume 8(5) ❖ Article e01789

info:doi/10.1002/ecs2.1789

http://creativecommons.org/licenses/by/3.0/

INTRODUCTION

Climate change is already evident. Since the1970s, global mean average temperature hasrisen by nearly 1°C, with the greatest seasonalincrease during winter (Vose and Klepzig 2014).Temporal precipitation pattern changes, includ-ing severe drought, have affected much of thesoutheastern United States over the past threedecades. For example, mean summer precipita-tion has decreased, whereas mean fall precipita-tion has increased by 30% since 1901; severityand patterns of storms are changing, with moreheavy downpours in many parts of the southeastand more powerful Atlantic hurricanes (Karlet al. 2009). Climate models project a 2–10°Cincrease in temperature by 2100 (IPCC 2007) inthe southeastern United States. Historically, cli-mate change has been an important driver inrange shifts, extinction, colonization, and evolu-tion through adaptation of species (Greenberget al. 2014). Amphibians have evolved diverselife history strategies allowing them to persistunder stochastic and changing environmentalconditions. Yet, it remains unknown whether cli-mate change will outpace their adaptive poten-tial (Walls et al. 2013a).

Amphibians that primarily breed in ephemeralwetlands are especially vulnerable to alteredweather patterns resulting from climate changebecause of their reliance on precipitation-drivenhydroregime characteristics such as the timing,depth, frequency, and duration of water in wet-lands (Brooks 2009, Greenberg et al. 2014).Changes in weather patterns including tempera-ture, and the amount, frequency, timing, andseverity of precipitation will almost certainlyalter wetland hydroregimes (Greenberg et al.2015), with critical implications for populationviability of wetland-breeding amphibians.Hydroregime effects on reproductive success(i.e., juvenile recruitment [JR]) are likely to differamong species because of differences in repro-ductive strategies: the length and timing ofbreeding period, rate of larval development, andtiming of metamorphosis (Snodgrass et al. 2000,Babbitt et al. 2003). Dry wetlands are unavailablefor breeding; likewise, wetlands that dry beforelarval development is complete will result in eggor tadpole mortality and reproductive failure(Walls et al. 2013b).

Many amphibian species depend on ephem-eral wetlands completely or facultatively forreproduction because they are generally fish-free,and host lower densities of predatory aquaticmacroinvertebrates (Moler and Franz 1987).Several studies report high stochastic variation inamphibian reproductive success related tohydroregime or unknown biotic factors such aslarval competition and predation on eggs andlarvae (Pechmann et al. 1989, Dodd 1994,Semlitsch et al. 1996, Richter et al. 2003, Green-berg and Tanner 2005a). Widespread reproduc-tive failure for even one year can severely reducepopulations of short-lived species and alteramphibian assemblages (Stewart 1995, McMena-min et al. 2008, Cole et al. 2016). Given theirlimited dispersal ability and reliance on weatherpatterns for reproductive success, ephemeralwetland-breeding amphibians represent an idealsystem to examine vulnerabilities of specieswith different reproductive strategies to climatechange.In this paper, we used 22 yr (March 1994–

December 2015) of continuous, concurrentamphibian trapping data at eight ephemeral wet-lands to examine the influence of weather,hydroregime, and breeding effort (adult capturesduring defined breeding periods) on reproduc-tive success of six anuran species representingfour reproductive strategies (Table 1). Reproduc-tive strategies and representative species were (1)prolonged, continuous, or intermittent breedingperiod, slow larval development (minimum2.5 months), and a “fixed” late spring start datefor juvenile emigration, regardless of whenoviposition occurred (cohort age), represented byfall-through-spring breeding Lithobates capito andLithobates sphenocephalus; (2) prolonged breedingperiod, rapid larval development (4–8 weeks),and juvenile emigration continual as cohortscomplete metamorphosis, represented by latewinter-through-fall breeding Anaxyrus terrestris;(3) season-limited breeding period, rapid larvaldevelopment (4–8 weeks), and juvenile emigra-tion continual as cohorts complete metamorpho-sis, represented by summer-breeding Anaxyrusquercicus and Gastrophryne carolinensis; and (4)aseasonal, explosive breeding, very rapid larvaldevelopment (2–3 weeks), and most juvenileemigration immediately after completing meta-morphosis, represented by Scaphiopus holbrookii.


GREENBERG ET AL.

We used the information-theoretic (IT) app-roach (Burnham and Anderson 2002) to modelthe reproductive success of each species in rela-tion to (1) hydroregime variables only; (2) weathervariables only; (3) hydroregime and weather vari-ables together; (4) model sets 1–3 (above) repeatedwith the addition of the number of breedingadults as a factor influencing reproductive suc-cess. The environmental and hydrologic variablesin these models represented conditions over

specified time intervals, defined as the period(s)between potential onset, mid-, or late breedingseason and peak juvenile emigration (PJE), toassess their relative importance during shorter orlonger time spans prior to tadpole metamorpho-sis. We also examined whether JR rates were mostclosely correlated among species with similarbreeding strategies. We hypothesized that (1) JRby species with similar reproductive strategieswould be influenced by the same environmental

Table 1. Life history attributes† of Lithobates capito (LCAP), L. sphenocephalus (LSPH), Anaxyrus terrestris (ATER),A. quercicus (AQUE), Gastrophryne carolinensis (GCAR), and Scaphiopus holbrookii (SHOL) related to reproduc-tive strategy (breeding, rate of larval development, and timing of metamorphosis and juvenile emigration) andexplanations of select variables and time spans used in statistical analyses.

SpeciesReproductiveStrategy‡

Breedingperiod

Eggs/Female

Larvaldevelopment

(days)Breeding

adult variable§

Juvenilerecruits

(JR variable)#

Time spansanalyzed (breedingthrough peak Jemigration)‖

LCAP PB-SLD-SE Fall–Winter–Spring

1200–2200 85–225 May Wk1 Yr1-Apr Wk4 Yr2

Apr Wk1-Oct Wk4(J if <50 mm SVL)

Sept Wk1 Yr1-May Wk4 Yr2(35 weeks)

Jan Wk1-May Wk4(19 weeks)

Mar Wk1-MayWk4(11 weeks)

LSPH PB-SLD-SE Fall–Winter–Spring

>1000 67–230 May Wk1 Yr1-Apr Wk4 Yr2

Apr Wk1-Dec Wk4(Juv <52 mm SVL)

Sept Wk1-JunWk1 (36 weeks)

Jan Wk1-Jun Wk1(20 weeks)

Mar Wk1-Jun Wk1(12 weeks)

ATER PB-RLD-CE Latewinter–Fall

Severalthousand

28–55 Feb Wk1 Yr1-Oct Wk4 Yr1

Mar Wk4-Dec Wk4(J < 25 mm SVL)

Feb Wk1-MayWk2 (13 weeks)

Mar Wk4-MayWk2 (6 weeks)

AQUE SB-RLD-CE Summer 420 max 33–44 May Wk1 Yr1-Oct Wk4 Yr1

Jun Wk1-Dec Wk4(J < 20 mm SVL)

May Wk1-Jul Wk4(11 weeks)

Jun Wk2-Jul Wk4(6 weeks)

GCAR SB-RLD-CE Summer 800 23–67 May Wk1 Yr1-Oct Wk4 Yr1

Jun Wk1-Dec Wk4(J < 23 mm SVL)

May Wk1-AugWk4 (15 weeks)

Jun Wk2-AugWk4 (10 weeks)

SHOL EB-VRLD-IE Explosiveaseasonal

Av 2872 14–30 Wk of explosivebreeding �1 week(3 weeks in total)¶

4 weeks followingbreeding event(J < 18 mm SVL)

Breeding event+2 weeks

Breeding event+3 weeks

† Ashton and Ashton (1988), Dodd (1995), Phillips (1995), Semlitsch et al. (1995), Palis (1998), Bridges and Dorcas (2000), Greenberg(2001), Wright (2002), Greenberg and Tanner (2004), McCallum et al. (2004), Greenberg and Tanner (2005a, b), Richter et al. (2009).

‡ PB-SLD-SE = prolonged breeding period, slow larval development, juvenile emigration not initiated until late spring, regardlessof when oviposition occurred (cohort age); PB-RLD-CE = prolonged breeding period, rapid larval development, juvenile emigrationcontinual as cohorts complete metamorphosis; SB-RLD-CE = season-limited breeding period; rapid larval development; juvenile emi-gration continual as cohorts complete metamorphosis; EB-VRLD-IE = explosive aseasonal breeding, very rapid larval development,most juvenile emigration immediately after completing metamorphosis.

§ All first-captures, summed across specified time periods; most breeding initiated later, ended earlier than dates here, but definedconservatively to include “stragglers.”

¶ Explosive breeding defined as ≥30 adults captured at an individual wetland within a one-week period.# All exiting (emigrating) first-captured juveniles, summed across specified time intervals; most emigration initiated later, ended

earlier than dates here, but defined conservatively to include “stragglers.”‖ Week of peak emigration (see Fig. 4) bolded.


GREENBERG ET AL.

(hydroregime and weather variables) and biologi-cal (adult breeding effort) variables; (2) JR wouldbe higher for all species when models encom-passed the maximum period (time span) of poten-tial tadpole occurrence and development withinwetlands; and (3) JR rates within individual wet-lands and breeding cycles would correlate mostclosely between species with similar breedingstrategies.

METHODS

Study areaOur eight study wetlands were a representa-

tive selection of small (0.1–0.37 ha), ephemeral,groundwater-driven sinkhole wetlands, embed-ded within xeric longleaf pine-wiregrass uplandsof the Floridan Aquifer System region, OcalaNational Forest, Marion and Putnam Counties,Florida (Fig. 1; Greenberg et al. 2015). Averageweekly temperatures (February 1997–December2015) ranged from 13.9°C in December to 28.6°Cin August (Fig. 2a). Average annual precipitation(March 1994–2015) at our study areas was132 cm, with more than half occurring duringlate spring and summer (Fig. 2a). Heavy precipi-tation providing groundwater recharge wasassociated with thunderstorms and tropical sys-tems in summer and fall, and wet autumn, win-ter, or spring frontal systems (Fig. 2a; Winsberg1990). Wetland depths were generally highest inwinter and lowest in summer due to rainfall pat-terns and groundwater recharge, as well as lowevapotranspiration in winter (Figs. 2b and 3;Knowles et al. 2002).

Amphibian samplingWe installed drift fences 7.6 m long and spaced

7.6 m apart around the perimeter of each wetlandnear the high water line, such that 50% of eachwetland was fenced, with fences and spacesequally distributed and encircling wetlands. Pit-fall traps (19-L buckets) were positioned insideand outside of each end of each fence (four perfence), and a double- or single-ended funnel trap(one of each per fence) was positioned at the mid-point of each fence on both sides, to detect direc-tional movement by amphibians to and fromwetlands. We placed a sponge in each pitfall trap,and moistened as needed during trap checks toreduce the likelihood of desiccation. All traps

were checked approximately three times weeklyfrom late March 1994 through December 2015.We identified, sexed, and measured snout-ventlength (SVL) and weight of captured frogs,and assigned age (adult or recent metamorph/first-year juvenile) based on a specified SVL cutofffor each species (Table 1). All individuals weremarked by wetland number and year of captureby toe clipping. Exceptions were newly metamor-phosed A. quercicus and S. holbrookii that were notmarked, as they were too small to toe-clip.Because all wetlands were sampled continuouslyand in proportion to basin size, we did not furtheradjust for trapnights. However, we discardeddata from one or more wetlands during someyears when most traps were flooded for pro-longed periods during defined adult breeding orjuvenile emigration periods, which differedamong species (Table 1).

Weather and water depth measurementsWe measured local precipitation using a rain

gauge placed in the open near Wetland 1, dailyfrom 28 March 1994 through mid-2006, andapproximately three times weekly thereafter.Maximum and minimum air temperatures weremeasured using a max-min thermometer (in theshade, also near Wetland 1) daily beginning 22February 1997 through mid-2006 and approxi-mately three times weekly through 2015. Wealso measured water depth approximatelyweekly at polyvinyl chloride pipe (PVC) staffgauges that were permanently established at theapproximate lowest point within each of theeight wetland basins. Our statistical modelsrequired complete environmental data for theexplanatory variables, which for some timeperiods or wetlands were missing. To avoid dis-carding entire breeding cycles for analyses andhypothesis testing, we estimated missing depthdata (22 total weeks for seven wetlands; 42 totalweeks for one wetland) using an existing model(developed using data from our study wetlands)that predicts wetland depth based on knowndepth and rainfall the prior week (Greenberget al. 2015). Missing rainfall data (54 totalweeks) were estimated by reversing the modelto predict rainfall the prior week based onchange in wetland depth from the prior week(Greenberg et al. 2015). We discarded entirebreeding cycles (which differed among species)


GREENBERG ET AL.

from statistical analyses requiring temperaturedata, if temperature data were missing. Wecoupled weekly weather data and wetlanddepth measurements with corresponding anuran

capture totals at four, approximately weeklyintervals (7.6 � 0.06 d per “week”) per month,using dates when all environmental variableswere sampled simultaneously.

Fig. 1. Locations of study wetlands in the Ocala National Forest, Putnam and Marion Counties, Florida (cour-tesy of Dale Johnson). Double-dashed and heavy lines represent sand and paved roads, respectively.


GREENBERG ET AL.

Determining reproductive cyclesWe used long-term drift fence capture data of

adults and emigrating juveniles (Figs. 4 and 5),tadpole sampling data from study wetlands(2000–2001; Greenberg et al., in press), and pub-lished literature to determine approximate breed-ing periods, rates of larval development to

metamorphosis, and juvenile emigration periodsfor each species (Table 1, Fig. 4), recognizing thateach can vary somewhat among years and wet-lands (Fig. 5). We defined a single breeding cycleas the period between the potential onsetof breeding through the completion of juvenileemigration from wetlands resulting from that

Fig. 2. Mean (�SE) (a) weekly temperature (minimum, maximum, average; February 1997–December 2015)and precipitation (March 1994–December 2015) and (b) approximate maximum depth (March 1994–December2015) of eight study wetlands, Ocala National Forest, Florida.


GREENBERG ET AL.

breeding effort (Table 1, Fig. 4), and used thatdefinition for all years. We defined breedingadults as all first-captures entering or exitingwetlands within their defined breeding period(Table 1, Fig. 4), recognizing that adult migrationto wetlands (Greenberg et al., in press), or evencalling activity (D. Saenz, personal communica-tion), is not always indicative of breeding. Weconservatively defined juvenile recruits as totalfirst-captures exiting (only) wetlands within theirdefined emigration period (Table 1, Fig. 4).Breeding and juvenile emigration periods par-tially overlapped within a breeding cycle, for allspecies except S. holbrookii (Table 1, Fig. 4).

For some species, the breeding cycle was com-plete within a given calendar year. In our studyarea, A. terrestris bred February through the endof October, and juveniles emerged beginning thelast week of March through the end of December;both A. quercucis and G. carolinensis bred Maythrough October, and juveniles emerged Junethrough December (Table 1, Fig. 4). For L. capitoand L. sphenocephalus, breeding extended fromone calendar year into the next (Table 1, Fig. 4).Breeding season for both species was difficult to

determine with confidence because adult capturerates were low, and breeding can occur continu-ously or intermittently during fall through spring(Palis 1998) or even summer (Godley 1992); how-ever, juveniles of both species emigrate in latespring through early fall within our study area(Greenberg 2001; Table 1, Fig. 4). Thus, wedefined breeding adults as all first-captures May(calendar year 1) through the following April (cal-endar year 2) as potentially contributing to JR forthat breeding cycle, and we defined juvenilerecruits within that breeding cycle as all first-captured emigrating juveniles April throughOctober for L. capito, or April through Decemberfor L. sphenocephalus, calendar year 2 (Table 1,Fig. 4). Scaphiopus holbrookii are explosive breedersthat can breed nearly any time of year after heavyrainfall, and sometimes go several years withoutbreeding; larval development is rapid, and up tothousands of juveniles metamorphose and emi-grate from wetlands two to three weeks later(Greenberg and Tanner 2004). Because they donot have a distinctive breeding season, weincluded only wetlands and years when breedingevents occurred in our analyses. We defined a

Year

Pon

d de

pth

(cm

)

0

50

100

150

200

250

300

Rai

nfal

l (cm

)

0

5

10

15

20

25

30

35Pond 1Pond 2Pond 3Pond 4Pond 5Pond 6Pond 7Pond 8Rainfall

Janu

ary 19

95

Janu

ary 19

97

Janu

ary 19

96

Janu

ary 19

99

Janu

ary 20

00

Janu

ary 20

01

Janu

ary 20

02

Janu

ary 20

03

Janu

ary 20

04

Janu

ary 20

05

Janu

ary 20

06

Janu

ary 20

07

Janu

ary 20

08

Janu

ary 20

09

Janu

ary 20

10

Janu

ary 20

11

Janu

ary 20

12

Janu

ary 20

13

Janu

ary 20

14

Janu

ary 20

15

Janu

ary 19

98

Fig. 3. Weekly (March 1994–2015) total precipitation and depths of eight study wetlands, Ocala National For-est, Florida.


GREENBERG ET AL.

breeding event as ≥30 adults captured at a parti-cular wetland within a single week, but includedadult first-captures during the week prior andafter that event as “breeding adults.” We definedS. holbrookii juvenile recruits to include allemigrating juveniles within four weeks after abreeding event to include peak emigration andpotential “stragglers” exiting wetlands (Table 1,Fig. 4).

Defining time spans of larval vulnerability toweather and hydroregime

We focused our analyses on periods when eggsand tadpoles were potentially present and devel-oping to metamorphosis, assuming that larval

vulnerability to weather or hydroregime vari-ables would occur within the aquatic stage (eggsthrough metamorphosis and juvenile emigra-tion). We defined two time periods (“timespans”) for focal species with rapid larval devel-opment (A. terrestris, A. quercicus, G. carolinensis,S. holbrookii), or three for those with protractedlarval development (L. capito and L. spheno-cephalus). We determined the average PJE weekfor each species (Fig. 4) and then defined thepotential periods for larval development asbeginning at the onset, mid-, or late breedingperiod (or presumed oviposition), extendingthrough PJE (Table 1, Fig. 4). Because S. hol-brookii can breed explosively during any month

Scaphiopus holbrookii

Week of year

050

100150200250300350400450500

Lithobates sphenocephala

0

5

10

15

20Lithobates capito

0

5

10

15

20

25AdultsJuveniles

Anaxyrus terrestris

0

10

20

30

40

50Anaxyrus quercicus

0

10

20

30

40

50

60

70

Janu

ary

Februa

ryMarc

hApri

lMay

June Ju

ly

Augus

t

Septem

ber

Octobe

r

Novem

ber

Decem

ber

Janu

ary

Februa

ryMarc

hApri

lMay

June Ju

ly

Augus

t

Septem

ber

Octobe

r

Novem

ber

Decem

ber

Gastrophryne carolinensis

Mea

n (+

SE

) to

tal w

eekl

y fir

st-c

aptu

res

(199

4–20

15)

0

10

20

30

40

50

60

70

Week of year

61 7± 612 605 ± 605 348 ± 346

Fig. 4. Mean (+SE) (1994–2015) number of adult and juvenile first-captures (all wetlands combined) per weekof year (approximately four weeks per month; 48 weeks per year) of six anuran species at isolated ephemeralwetlands in the Ocala National Forest, Florida.


GREENBERG ET AL.

of the year, with metamorphs emerging a fewweeks later, we defined the time spans for thisspecies as (1) two weeks or (2) three weeks aftera breeding event, to encompass the period whentadpoles would be developing to metamorpho-sis. The dates and time spans encompassed dif-fered among our focal species because ofdifferences in the timing and lengths of breedingperiods, larval developmental rates, and juvenileemigration periods (Table 1, Fig. 4).

Statistical analysesSelection of weather and hydroregime variables.—

For each species, we performed initial exploratoryanalyses using simple and stepwise regressionsto determine the most important weather andhydroregime variables influencing JR, and to

narrow down pertinent variables for developinghypotheses and models. For the purpose of thispaper, we considered each breeding cycle withina wetland to be an independent sample point. Ini-tial hydroregime variables explored for all species,within each defined time span, were maximumhydroperiod (duration of continuous water, evenif interrupted before PJE); the number of weekswith continuous water (>0 cm), including andpreceding PJE week; total number of hydroperi-ods (>1 if water was intermittent); total weeks dry(even if intermittent); maximum number of weekscontinuously dry (we re-tested these variablesusing alternative definitions of “dry” [e.g., <5 cmand <10 cm depth] to examine whether very shal-low water might be functionally dry, regardinginfluence on tadpoles); water depth (average,

Scaphiopus holbrookii

Wetland number

0

200

400

600

800

Lithobates capito

0

10

20

30

40

AdultsJuvenile Recruits

Lithobates sphenocephala

0

5

10

15

20

25

Anaxyrus terrestris

Ave

rage

(+S

E) a

nnua

l firs

t-cap

ture

s (1

994–

2015

)

0

20

40

60

80

100Anaxyrus quercicus

0

20

40

60

80

100

Gastrophryne carolinensis

Wetland number1 2 3 4 5 6 7 81 2 3 4 5 6 7 8

0

40

80

120

160

Fig. 5. Mean (+SE) annual (1994–2015) number of first-captured adults and juvenile recruits per wetland of sixanuran species, Ocala National Forest, Florida.


GREENBERG ET AL.

minimum, maximum); and depth fluctuations(maximum increase, decrease, or maximumchange). The observed precipitation variable wasweekly total; these were used to create four pre-dictor variables, defined as minimum, maximum,average, and total precipitation over the course ofa given time span. The observed temperaturevariables were weekly minimum, maximum, andaverage (more representative of the entire weekthan maximum or minimum extremes); thesewere used to create three predictor variablesdefined as minimum, maximum, and averagetemperature over the course of a given time span.

Initial, preliminary analyses consisted of uni-variate graphical plots, correlations, simple linearregressions, and stepwise multiple regressions(PROC GLMSELECT SAS 2011) of square-root-transformed juvenile recruits on theenvironmental and weather variables within spe-cies-specific time spans. Our goal was to narrowdown the most important weather and hydrore-gime variables to use in model development,minimize the use of strongly correlated variables,and assess whether some variables were sharedor differed among species.

Preliminary analyses were followed by using ageneralized linear mixed model to explain JR asa function of environmental and climatic vari-ables. The response variable (JR) for all six spe-cies exhibited a negative binomial distribution;variance greatly exceeded the mean, plots of thedistribution showed more observations at lowand high values than expected based on random-ness, and analyses revealed that the dispersionparameter was not zero. To further ensure thatthe distribution was negative binomial, we alsofitted zero-inflated Poisson and zero-inflatednegative binomial distributions; comparison oftheir corrected Akaike information criterion(AICc) values to the negative binomial did notjustify using them in further modeling. The finalmodel assumed the response variable (JR) wasnegative binomial distributed and data were ana-lyzed with wetlands as a fixed factor with norepeated-measures factor for years using PROCGLIMMIX (SAS 2011).

We used preliminary stepwise regression toliberally select independent variables for inclu-sion and deletion in the models based on the 0.25significance level. In addition, we consideredknown biological relationships to help develop

meaningful models for further consideration.Second-order interactions between the mainindependent variables were entertained in modelfits, and used if they were significant and mean-ingful. Resulting models were examined by plot-ting observed over predicted values, andinspecting plots of the residuals over the inde-pendent variables.Hypothesis testing and model development.—We

used the IT approach (Burnham and Anderson2002) to test hypotheses designed to examine rel-ative importance of individual wetlands, hydro-regime, weather, time span prior to PJE, and thenumber of breeding adults on JR by our six focalspecies with different breeding strategies. Includ-ing individual wetland in our models likelyreduced effects of some hydroregime variables,but also reduced variability associated with wet-land-specific characteristics or potential philopa-try. We elected to use the number of breedingadults, rather than females only because thenumber of females was often zero, or very low,when split out by wetland and breeding cycle,which was the basis for our models. Preliminaryanalyses indicated that the female sex ratio was<0.5 for most species (range 0.34 for A. terrestris,to 0.46 for L. sphenocephalus) when pooled overall wetlands and breeding cycles, but deviatedabove and below 0.5 across the breeding cycles.The somewhat lower level of captures of femaleadults may be due to an unknown differentialcapture bias between females and males. More-over, 24% of the 6676 adults captured over thecourse of the study could not be sexed accuratelyand were classified as “unknown”; these couldrepresent a disproportionate number of females.Thus, we determined that total first-capturedadults, both male and female, was a more stablemetric of the adult female breeding populationthan females only.For each species, numerous models were devel-

oped based on a priori hypotheses and then eval-uated to determine which ones were mostplausible. For a given species, we included thebase model which consisted of the effect of theindividual wetland alone on JR. We then evalu-ated initial sets of three models for each speciesfor their power to predict JR. Each model setincluded individual wetland and (1) two to threeselected hydroregime variables only, (2) one tothree selected weather variables only, or (3) the


GREENBERG ET AL.

selected hydroregime and weather variables andnone or one interaction between hydroregime andweather variables. Each model set was repeatedto encompass two (onset and mid-breeding per-iod through PJE for A. terrestris, A. quercicus,G. carolinensis) or three time spans (onset, mid-,and late breeding period through PJE for L. capitoand L. sphenocephalus); Scaphiopus holbrookii wasthe exception, with time spans encompassing theperiod between explosive breeding events and thesubsequent two or three weeks, as describedabove. Finally, all model sets and time spans wererepeated, to include the number of breedingadults as a potential influence on JR. Thus, wetested a total of 13 models for A. terrestris, A. quer-cicus, G. carolinensis, and S. holbrookii, and 19models for L. capito, L. sphenocephalus.

Model sets for all species (except S. holbrookii)and defined time spans contained the same twobasic hydroregime variables—(1) the number ofweeks with continuous water (>0 cm), includingand preceding PJE week, and (2) average waterdepth—and the same two basic weather vari-ables: (1) total precipitation and (2) average tem-perature (maximum + minimum temperature/2),selected a priori based on preliminary exploratoryanalysis and biological considerations. For somespecies, model sets included one additionalhydroregime, weather, or hydroregime–weatherinteraction variable, if preliminary analyses sug-gested its particular importance for that species.We did not include weeks with continuous waterin S. holbrookii model sets due to virtual absenceof variability (all wetlands continuously heldwater for two or three weeks after n = 34 explo-sive breeding events used in analyses, with twoexceptions). Similarly, we did not use averagetemperature in S. holbrookii models because miss-ing temperature data would have (1) reduced thesample size too much and (2) produced variousmodels with different sample sizes which cannotbe validly compared with the AICc criterion. Weassumed that S. holbrookii tadpoles are less likelyto be temperature sensitive at our Florida studysites than some other species, given their poten-tial for explosive breeding events at any time ofyear. For each set of models, the AICcs andAkaike weights were computed and parameterestimates, standard errors, and 95% confidenceintervals obtained using PROC GLIMMIX (SAS2011).

Is recruitment success correlated between specieswith similar breeding strategies?.—We used Ken-dall’s tau rank correlations to test the hypothesisthat JR would be most similar among species withsimilar breeding strategies, using each wetland-breeding cycle as one data point. Kendall’s wasselected over Pearson’s tests because the responsevariable, JR, exhibited a negative binomial distri-bution. Scaphiopus holbrookii was not includedbecause it breeds infrequently (e.g., missing val-ues for many years) and has no defined breedingor JR period for comparison with other species.We also used Kendall’s tau rank correlations toexamine the relationship between breeding effortand JR within species, using each wetland-breedingcycle as one data point.

RESULTS

Our results indicated that one or more precipi-tation-driven hydroregime variable was a stronginfluence on JR for all tested species exceptA. terrestris, but variables differed among spe-cies. In contrast, temperature played a relativelyminor role, within the range of variation tested.Continuous hydroperiod through PJE was animportant predictor of recruitment for specieswith prolonged breeding periods, slow larvaldevelopment, and a “fixed” late spring start datefor juvenile emigration—regardless of whenoviposition occurred, or cohort age (L. capito andL. sphenocephalus)—but not for species with rapidlarval development and continual emigration ascohorts complete metamorphosis (A. terrestris,A. quercicus, G. carolinensis, S. holbrookii). Thebest-fitting models for all species included themaximum time span tested, from onset of breed-ing season through PJE. Total rainfall was posi-tively associated with recruitment for mostspecies, and depth characteristics (maximum,average, or maximum change) affected speciesdifferently. Total breeding adults was a signifi-cant, important predictor of JR for A. terrestris,G. carolinensis, and S. holbrookii and improvedmodel strength for A. quercicus, but was not sig-nificant or important for Lithobates spp. AnnualJR was correlated among species with similarreproductive strategies.Our best-fitting model for predicting JR by

L. capito (model 10; w = 0.467) was a function ofindividual wetland, hydroregime, and weather


GREENBERG ET AL.

variables, and interactions between them duringthe maximum time span tested (onset of breed-ing period through PJE; September week 1through May week 4; Table 2). Among hydrore-gime variables, continuous water (>0 cm) andmaximum depth change were significantly posi-tively correlated with JR, while average depthwas not. Among weather variables, only totalprecipitation was significant and positively cor-related with JR. Average temperature alone wasnot a significant predictor, but its interactionwith continuous water was significantly nega-tively associated with JR. Weather variablesalone during the same time span (model 9;w = 0.030) had a greater influence than hydrore-gime variables alone (model 8; w = 0.000). Addi-tion of total breeding adults was not significant,but slightly weakened model strength (see model19; w = 0.429 vs. model 10; Table 2).

Our best-supported model for predicting JR byL. sphenocephalus (model 10; w = 0.362) was also

a function of individual wetland, hydroregime,and weather variables, and an interactionbetween them during the maximum time spantested (onset of breeding period through PJE;September week 1 through June week 1; Table 3).Among significant hydroregime variables, aver-age wetland depth was negatively associated,and continuous water (>0 cm) was positivelyassociated with JR. No weather variables alonewere significant, but an interaction between totalrainfall and maximum depth change was posi-tively associated with JR. Weather variablesalone, including total rainfall (positively associ-ated) and average temperature (nonsignificant)during the same time span (model 9; w = 0.011),had a greater influence on JR than hydroregimevariables alone, including maximum depthchange and continuous water (both positivelyassociated) (model 8; w = 0.000). Addition oftotal breeding adults reduced model strengthconsiderably (see model 19; w = 0.104 vs. model

Table 2. Information-theoretic model (M) results† for Lithobates capito juvenile recruitment (1995–2015; n = 118)from eight isolated, ephemeral wetlands, Ocala National Forest, Florida.

MTime span toMay Wk4 W MCD Cont0 DAv PptTot TAv Cont0xTAv A AICc Akaike Wt

1 x 660.0 0.000002 Mar Wk1 x* � + +* 639.8 0.000003 Mar Wk1 x +* �* 644.1 0.000004 Mar Wk1 x* + +* +* +* + �* 624.6 0.000195 Jan Wk1 x* + +* + 642.9 0.000006 Jan Wk1 x* +* � 614.2 0.034227 Jan Wk1 x* + + � +* + � 619.2 0.002738 Sept Wk1 x* +* +* � 643.8 0.000009 Sept Wk1 x* +* �* 614.4 0.0300410 Sept Wk1 x* +* +* � +* � �* 608.9 0.4667411 Mar Wk1 x* � + +* + 640.4 0.0000012 Mar Wk1 x +* �* + 645.3 0.0000013 Mar Wk1 x + +* +* +* + �* + 625.9 0.0001014 Jan Wk1 x* + +* + + 643.7 0.0000015 Jan Wk1 x +* � + 615.1 0.0217016 Jan Wk1 x + +* � +* + � + 619.3 0.0026317 Sept Wk1 x* +* +* � + 644.0 0.0000018 Sept Wk1 x +* �* + 616.2 0.0126019 Sept Wk1 x* + +* � +* � �* + 609.1 0.42906

Notes: Models were run for each of three time spans “pegged” on peak week of juvenile emigration (May week 4), to thepreceding onset, mid-, and late breeding period (breeding cycle) and include hydroregime variables (MCD, Cont0, DAv)‡ only;weather variables (PptTot, TAv)§ only; both hydroregime and weather variables with an interaction effect (Cont0xTAv); allmodels repeated with adult breeding effort (A) included. Wetland (W) was considered a fixed effect and included in all models.

† Models with the smallest AICc and highest Akaike weights have more support. The + or � indicates positive or negativerelationship with juvenile recruitment. Significant (where confidence intervals did not include zero) variables are denoted bythe � symbol, and models with the smallest AICc and highest Akaike weights are bolded.

‡ Maximum change in depth (MCD); weeks with continuous water (>0 cm) (Cont0); average depth (DAv); within specifiedtime span.

§ Total precipitation (PptTot); mean average temperature (TAv); interaction Cont0xTAv; within specified time span.


GREENBERG ET AL.

10). The next best two models (model 7;w = 0.178 and model 13; w = 0.185) includedboth hydroregime and weather variables withinshorter time spans (mid- or late breeding seasonthrough PJE, respectively).

Our best-fitting model for predicting JR byA. terrestris (model 12; w = 0.389) was a functionof individual wetland, total precipitation (posi-tively associated), and total breeding adults dur-ing the maximum time span tested (onset ofbreeding period through PJE; February week 1through May week 2; Table 4). Total breedingadults was significantly positively associatedwith JR and substantially improved modelstrength (see model 6; w = 0.000 vs. model 12),whereas addition of hydroperiod variables didnot (see model 13; w = 0.316 vs. model 12).

Our best-fitting model for predicting JR byA. quercicus (model 11; w = 0.392) was a function

of hydroregime variables and total breeding adultsduring the maximum time span tested (onset ofbreeding period through PJE; May week 1 throughJuly week 4; Table 5); individual wetland was nota significant predictor of JR. Among hydroregimevariables tested, only maximum (positively associ-ated) and average depth (negatively associated)were significant predictors of JR. Total breedingadults was not a significant predictor, but nonethe-less slightly improved model strength (see model5; w = 0.275 vs. model 11). Similarly, addition ofweather variables did not improve model strength(see model 13; w = 0.162 vs. model 11).Our best-fitting model for predicting JR by

G. carolinensis (model 13; w = 0.994) was a func-tion of individual wetland, hydroregime, andweather variables, and total breeding adults,during the maximum time span tested (onset ofbreeding period through PJE; May week 1 through

Table 3. Information-theoretic model (M) results† for Lithobates sphenocephalus juvenile recruitment (1995–2015;n = 118) from eight isolated, ephemeral wetlands, Ocala National Forest, Florida.

MTime span toJun Wk1 W MCD DAv Cont0 PptTot TAv MCDxPptTot A AICc Akaike Wt

1 x 618.1 0.000002 Mar Wk1 x* + +* � 598.6 0.000053 Mar Wk1 x + � 617.6 0.000004 Mar Wk1 x +* +* �* +* +* �* 585.0 0.049085 Jan Wk1 x + +* � 601.9 0.000016 Jan Wk1 x +* + 586.0 0.030217 Jan Wk1 x* � + �* + +* + 582.5 0.178178 Sept Wk1 x* +* � +* 602.9 0.000019 Sept Wk1 x* +* � 588.0 0.0112210 Sept Wk1 x* � �* +* + � +* 581.9 0.3622311 Mar Wk1 x* + +* � � 601.1 0.0000212 Mar Wk1 x + � + 717.4 0.0000013 Mar Wk1 x* +* +* �* +* +* �* �* 582.4 0.1846114 Jan Wk1 x* + +* � � 604.1 0.0000015 Jan Wk1 x +* + + 587.6 0.0140216 Jan Wk1 x* � + �* + +* + � 584.5 0.0634817 Sept Wk1 x +* � +* + 605.3 0.0000018 Sept Wk1 x* +* � + 590.5 0.0032719 Sept Wk1 x* � �* +* + � +* + 583.6 0.10361

Notes: Models were run for each of three time spans “pegged” on peak week of juvenile emigration, to the preceding onset,mid-, and late breeding period (breeding cycle) and include hydroregime variables (MCD, DAv, Cont0)‡ only; weather vari-ables (PptTot, TAv)§ only; both hydroregime and weather variables with an interaction effect (MCDxPptTot); all modelsrepeated with adult breeding effort (A) included. Wetland (W) was considered a fixed effect and included in all models.


‡ Maximum change in depth (MCD); average depth (DAv); weeks with continuous water (>0 cm) through peak juvenileemigration week (Cont0); within specified time span.

§ Total precipitation (PptTot); mean average temperature (TAv); interaction MCD 9 PptTot; within specified time span.


GREENBERG ET AL.

Table 4. Information-theoretic model (M) results† for Anaxyrus terrestris juvenile recruitment (1994–2015;n = 118) from eight isolated, ephemeral wetlands, Ocala National Forest, Florida.

MTime span toJun Wk1 W D5 DAv Cont0 PptTot TAv A AICc Akaike Wt

1 x* 477.7 0.000032 Mar Wk4 x* �* � � 471.5 0.000593 Mar Wk4 x* + + 479.4 0.000014 Mar Wk4 x* �* � � + + 472.5 0.000365 Feb Wk1 x* � � � 473.9 0.000186 Feb Wk1 x* +* + 472.3 0.000407 Feb Wk1 x* � �* � +* + 469.5 0.001638 Mar Wk4 x* � � � +* 460.5 0.146899 Mar Wk4 x* + + +* 463.5 0.0324410 Mar Wk4 x* �* � � + + +* 462.6 0.0501911 Feb Wk1 x* � �* + +* 462.2 0.0622112 Feb Wk1 x* +* + +* 458.5 0.3894413 Feb Wk1 x* � �* � +* + +* 459.0 0.31563

Notes: Models were run for each of two time spans “pegged” on peak week of juvenile emigration, to the preceding onsetand mid-breeding period (breeding cycle) and include hydroregime variables (D5, DAv, Cont0)‡ only; weather variables(PptTot, TAv)§ only; both hydroregime and weather variables; all models repeated with adult breeding effort (A) included. Wet-land (W) was considered a fixed effect and included in all models.


‡ Total number of weeks (not necessarily continuous) when depth ≤5 cm (D5); weeks with continuous water (>0 cm)through peak juvenile emigration week (Cont0); average depth (DAv); within specified time span.

§ Total precipitation (PptTot); mean average temperature (TAv); within specified time span.

Table 5. Information-theoretic model (M) results† for Anaxyrus quercicus juvenile recruitment (1994–2015;n = 118) from eight isolated, ephemeral wetlands, Ocala National Forest, Florida.

MTime span toJun Wk1 W DMax DAv Cont0 PptTot TAv A AICc Akaike Wt

1 x 218.4 0.002132 Jun Wk2 x + �* + 219.5 0.001213 Jun Wk2 x + + 220.5 0.000744 Jun Wk2 x + � + + + 224.4 0.000115 May Wk1 x +* �* � 208.7 0.275266 May Wk1 x +* + 217.2 0.003767 May Wk1 x +* �* � + + 210.8 0.096048 Jun Wk2 x + � + + 217.2 0.003949 Jun Wk2 x + � + 215.5 0.0092310 Jun Wk2 x + � + + � + 222.0 0.0003611 May Wk1 x +* �* � + 208.0 0.3919712 May Wk1 x +* + + 212.0 0.0528913 May Wk1 x +* �* � + � + 209.7 0.16236

Notes: Models were run for each of two time spans “pegged” on peak week of juvenile emigration, to the preceding onsetand mid-breeding period (breeding cycle) and include hydroregime variables (DMax, DAv, Cont0)‡ only; weather variables(PptTot, TAv)§ only; both hydroregime and weather variables; all models repeated with adult breeding effort (A) included. Wet-land (W) was considered a fixed effect and included in all models.


‡ Maximum depth (DMax); weeks with continuous water (>0 cm) (Cont0); average depth (DAv); within specified timespan.

§ Total precipitation (PptTot); mean average temperature (TAv).


GREENBERG ET AL.

August week 4; Table 6). Among hydroregimevariables, only average depth was significantlynegatively correlated with JR. Among weathervariables, only total precipitation was significantlypositively associated with JR. Total breedingadults was significantly positively associated withJR and substantially improved model strength (seemodel 7; w = 0.001 vs. model 13).

Our best-fitting model for predicting JR byS. holbrookii (model 8; w = 0.670) was a functionof individual wetland, total precipitation (theonly weather variable tested; negatively associ-ated), and total breeding adults during the timespan between explosive breeding events the sub-sequent two weeks (Table 7). Hydroregime vari-ables alone (see model 9; w = 0.004 vs. model 8)or in addition to weather variables (see model 10;w = 0.050 vs. model 8) did not improve modelstrength.

Kendall’s tau rank correlations indicated thatadult breeding effort was significantly positivelycorrelated with reproductive success for all spe-cies, when tested without other model variables;tau values ranged from 0.22842 (A. terrestris) to0.47972 (S. holbrookii; Fig. 6).

Annual JR was significantly positively associ-ated between both species having a prolongedbreeding period, slow larval development, and a“fixed” late spring start date for juvenile emigra-tion, regardless of when oviposition occurred, orcohort age (L. capito and L. sphenocephalus; tau =0.41069). In most cases, associations of JR betweenboth Lithobates species, and other species havingdifferent reproductive strategies, were negative ornonsignificant; L. sphenocephalus and A. quercicusJR was weakly positively associated. Annual JRwas also positively associated between A. ter-restris (prolonged breeding period, rapid larvaldevelopment, and juvenile emigration continualas cohorts complete metamorphosis) and bothspecies having a season-limited breeding period,rapid larval development, and continuous juve-nile emigration as cohorts complete metamorpho-sis: A. quercicus, tau = 0.23850; and G. carolinensis,tau = 0.16302. Annual JR was also positively asso-ciated between A. quercicus and G. carolinensis,both species having a season-limited breedingperiod, rapid larval development, and juvenileemigration continual as cohorts complete meta-morphosis (tau = 0.45937; Table 8).

Table 6. Information-theoretic model (M) results† for Gastrophryne carolinensis juvenile recruitment (1994–2015;n = 126) from eight isolated, ephemeral wetlands, Ocala National Forest, Florida.

MTime span toJun Wk1 W DMin DAv Cont0 PptTot TMaxAv TAv A AICc Akaike Wt

1 x* 387.5 0.000002 Jun Wk2 x* �* � + 352.5 0.000003 Jun Wk2 x +* +* +* 367.4 0.000004 Jun Wk2 x* + �* + +* +* + 336.4 0.000055 May Wk1 x* �* � � 342.7 0.000006 May Wk1 x* +* +* � 372.7 0.000007 May Wk1 x* � �* + +* + � 331.1 0.000668 Jun Wk2 x* �* + � +* 343.8 0.000009 Jun Wk2 x* +* +* + +* 352.6 0.0000010 Jun Wk2 x* + � � +* +* + +* 327.3 0.0044811 May Wk1 x* �* + � +* 332.3 0.0003612 May Wk1 x* +* +* � +* 353.8 0.0000013 May Wk1 x* + �* � +* + + +* 316.5 0.99445

Notes: Models were run for each of two time spans “pegged” on peak week of juvenile emigration, to the preceding onsetand mid-breeding period (breeding cycle) and include hydroregime variables (DMin, DAv, Cont0)‡ only; weather variables(PptTot, TMaxAvTav)§ only; both hydroregime and weather variables; all models repeated with adult breeding effort (A)included. Wetland (W) was considered a fixed effect and included in all models.


‡ Minimum depth (DMin); weeks with continuous water (>0 cm) (Cont0); average depth (DAv); within specified time span.§ Total precipitation (PptTot); mean maximum temperature (TMaxAv); mean average temperature (TAv); within specified

time span.


GREENBERG ET AL.

DISCUSSION

Our models indicated that environmental vari-ables affecting reproductive success differedamong species, but were most similar amongthose with similar reproductive strategies, char-acterized by length and timing of breeding per-iod, rates of larval development, and timing ofmetamorphosis. Strong correlations of annual JRbetween species with similar reproductive strate-gies further corroborated those results. This sug-gests that altered weather patterns associatedwith climate change will differentially affectreproductive success among amphibian species,but could correspond more closely among thosehaving similar reproductive strategies.

The best-supported models for all focal species(except explosive breeder S. holbrookii) encom-passed the maximum tested time span (onset ofbreeding period through PJE) when tadpoles, ifpresent, would be developing to metamorphosiswithin wetlands. Models encompassing shortertime spans were somewhat supported only forL. sphenocephalus. Based on known breeding andtadpole phenology (footnotes, Table 1), all ofour focal species could potentially reproduce

successfully even if suitable wetland characteris-tics did not occur until mid- or late breeding sea-son. However, given the extreme variation inreproductive success related to hydroregime,weather, and other, seemingly stochastic factors(Pechmann et al. 1989, Semlitsch et al. 1996,Richter et al. 2003), a protracted time span pro-vides the maximum opportunity for continuousor intermittent breeding and introduction of newegg and tadpole cohorts (Greenberg et al., inpress), likely increasing chances that some willsurvive to metamorphosis.Our strongest models indicated that different

hydroregime characteristics, total precipitation,or both were important drivers of reproductivesuccess for most species. Continuous waterthrough PJE was an important predictor of JR forfocal species with prolonged breeding seasonsand slow larval development (L. capito andL. sphenocephalus), but not for those with rapidlarval development (A. terrestris, A. quercicus,G. carolinensis, S. holbrookii). A continuoushydroperiod during the onset of breeding periodthrough PJE time span (the maximum time spantested) would potentially allow L. capito orL. sphenocephalus tadpoles originating from

Table 7. Information-Theoretic model† (M) results for Scaphiopus holbrookii juvenile recruitment (1994–2015;n = 34) from eight isolated, ephemeral wetlands, Ocala National Forest, Florida.

MTime span toJun Wk1 W DAv DMax PptTot A AICc Akaike Wt

1 x 319.8 0.000682 Jun Wk2 x � 322.8 0.000153 Jun Wk2 x � + 327.1 0.000024 Jun Wk2 x �* +* �* 328.1 0.000015 May Wk1 x �* 320.9 0.000396 May Wk1 x � � 325.7 0.000047 May Wk1 x + � � 329.6 0.000018 Jun Wk2 x* �* +* 306.0 0.670079 Jun Wk2 x + � +* 316.5 0.0035010 Jun Wk2 x � + �* +* 311.2 0.0503311 May Wk1 x �* +* 308.6 0.1804712 May Wk1 x + �* +* 310.5 0.0699313 May Wk1 x � � � +* 312.6 0.02441

Notes: Models were run for each of three time spans “pegged” on week of explosive breeding (≥30 individual adults cap-tured at a given wetland within a single week) and the subsequent two or three weeks and include hydroregime variables(DAv, DMax)‡ only; weather variables (PptTot)§ only; both hydroregime and weather variables; all models repeated with adultbreeding effort included. Wetland was considered a fixed effect and included in all models.


‡ Average depth (DAv); maximum depth (DMax).§ Total precipitation (PptTot).


GREENBERG ET AL.

continuous or intermittent fall, winter, or springbreeding to cumulatively survive to metamor-phosis, which does not occur until late spring.Alternatively, shorter hydroperiods lasting win-ter—or spring—through peak metamorphosiswould eliminate tadpoles resulting from earlierbreeding, providing opportunity only for cohortsfrom later breeding to survive; wetlands dryingprior to spring metamorphosis would eliminateall cohorts within a breeding cycle. In contrast,species with rapid larval development and theability to metamorphose continually may notrequire continuous water through PJE becausetadpoles could complete metamorphosis in oneor multiple shorter hydroperiods within thebreeding cycle. A positive association between

total rainfall and A. terrestris and G. carolinensisJR suggests that higher rainfall may provideenough suitable hydroperiods within breedingcycles for successful JR, even if water is not con-tinuously in wetlands through PJE. Thus, bothtiming and duration of hydroperiods are likelyto affect amphibian species differently, in relationto their reproductive strategies.Influential water depth variables also differed

among species and generally correspondedamong species with similar breeding strategies.juvenile recruitment rates of A. quercicus andG. carolinensis were negatively associated withaverage depth, suggesting that shallow watermay increase tadpole survival to metamorphosisfor these species. Average depth was negatively

Lithobates capitoτ = 0.36060P < 0.0001n = 150

0 2 4 6 8 10

020406080

100120140160180200

Lithobates sphenocephalaτ = 0.34116P = 0.0001n = 150

0 5 10 15 20 25

010203040506070

Anaxyrus terrestrisτ = 0.22842P = 0.0091n = 150

0 20 40 60 80 100 120 140 160

Tota

l em

igra

ting

juve

nile

s

0

200

400

600

800Anaxyrus quercicusτ = 0.41701P < 0.0001n = 158

0 100 200 300 400 500

0

10

20

30

40

50

Gastrophryne carolinensisτ = 0.41963

P < 0.0001n = 158

Total breeding adults

0 100 200 300 400 500

0

10

20

30

40

50

60Scaphiopus holbrookiiτ = 0.47972P = 0.0002n = 34

0 100 200 300 400 500

0

1000

2000

3000

4000

5000

6000

Total breeding adults

Fig. 6. Kendall’s tau rank correlations of adult breeding effort with juvenile recruitment rates of six anuranspecies from eight isolated, ephemeral wetlands, Ocala National Forest, Florida (1994–2015). Each data pointrepresents adult breeding and juvenile recruitment from one wetland within one breeding cycle (Table 1). ForScaphiopus holbrookii only explosive breeding events (n ≥ 30 adults per wetland within a one-week period) andresulting juvenile recruits were included.


GREENBERG ET AL.

associated with L. sphenocephalus recruitment inthe best model encompassing the maximum timespan (model 10), but not in other good modelsencompassing shorter time spans (models 7 and13), suggesting that effect of average depth mayvary with timing. Maximum change in wetlanddepth and total precipitation, or an interactionbetween those variables, were important predictorsof L. capito and L. sphenocephalus JR, respectively,suggesting that large, depth fluctuations may havean important influence on L. capito and L. spheno-cephalus recruitment. J. H. K. Pechmann (personalcommunication) observed L. sevosa “waiting” untilwater levels increased to about 55 cm at Glenn’sPond before initiating breeding activity; deeperwater could increase the probability of waterremaining in wetlands through metamorphosis.Maximum water depth was positively correlatedwith A. quercicus JR, suggesting that deeper waterat some point during the breeding cycle is associ-ated with hydroperiod duration adequate fortadpole survival to metamorphosis. Other studiesindicate that wetland dry-downs immediatelyprior to breeding may enhance JR forG. carolinensis(Pechmann et al. 1989, Semlitsch et al. 1996), byreducing populations of predatory aquaticmacroinvertebrates and salamander larvae.

Our strongest model for S. holbrookii indicateda strong, positive association between breedingeffort and recruitment, and a negative associationbetween total precipitation and recruitmentwithin two weeks after explosive breeding events.Explosive breeding events are triggered by heavyrainfalls, usually associated with tropical stormsor hurricanes that rapidly refill dry wetlands(Greenberg and Tanner 2004). Thus, whereasheavy rainfall is critical in eliciting breedingevents, our results suggest that substantial rainfallwithin the two to three weeks after breedingcould negatively affect tadpole survival to meta-morphosis. Our results indicate that two weekspost-breeding is the most critical time span affect-ing JR, as tadpole development is very rapid.Air temperature variables had little or no influ-

ence on JR for most species, suggesting that tad-poles can tolerate a range of temperatures withinthe seasons they occur in wetlands. Lithobatescapito was an exception, where a negative inter-action effect between continuous water and tem-perature, in conjunction with a (nonsignificant)association with temperature, suggested thatrecruitment may be negatively affected by highertemperatures, even when hydroperiod timingand duration are suitable. Clearly, tadpoles of

Table 8. Kendall’s tau rank correlations of juvenile recruitment rates between all pairs of five anuran species fromeight isolated, ephemeral wetlands, Ocala National Forest, Florida (1994–2015).

SpeciesPB-SLD-SE†L. capito

PB-SLD-SE†L. sphenocephalus

PB-RLD-C‡A. terrestris

SB-RLD-CE§A. quercicus

SB-RLD-CE§G. carolinensis

Lithobates capito . . . tau = 0.41069 tau = 0.04635 tau = �0.11155 tau = �0.29418P < 0.0001 P = 0.4634 P = 0.0982 P < 0.0001n = 161 n = 161 n = 155 n = 155

L. sphenocephalus . . . . . . tau = 0.08248 tau = 0.15466 tau = �0.04391P = 0.1869 P = 0.0206 P = 0.4924n = 161 n = 155 n = 155

Anaxyrus terrestris . . . . . . . . . tau = 0.23850 tau = 0.16302P = 0.0006 P = 0.0147n = 155 n = 155

A. quercicus . . . . . . . . . . . . tau = 0.45937P < 0.0001n = 169

Notes: Species exhibit three reproductive strategies defined by length and timing of breeding period, rate of larval develop-ment, and timing of metamorphosis and juvenile emigration. Scaphiopus holbrookii exhibits a fourth reproductive strategy (ex-plosive breeding), but was not included because breeding and juvenile recruitment are unpredictable and aseasonal. Each datapoint represents juvenile recruitment from one wetland within the same year.

† Prolonged breeding period, slow larval development, spring emigration regardless of when oviposition occurred (cohortage).

‡ Prolonged breeding period, rapid larval development, juvenile emigration continual as cohorts complete metamorphosis.§ Season-limited breeding period, rapid larval development, juvenile emigration continual as cohorts complete metamor-

phosis.


GREENBERG ET AL.

species with prolonged, multi-season breedingperiods (e.g., L. capito, L. sphenocephalus, andA. terrestris), or explosive breeding at any time ofyear (S. holbrookii), would require wider temper-ature tolerance than summer breeders such asA. quercicus and G. carolinensis. Our models onlyincluded temperatures measured within speci-fied time spans relevant to each species’ breedingphenology. Thus, we could not address howreproductive success might be affected if breed-ing phenology (seasonal timing) was substan-tially altered due to changing temperature orprecipitation (Saenz et al. 2006) associated withclimate change, thereby exposing eggs and tad-poles to a wider range of temperatures. Addi-tionally, altered changes in hydroregimeassociated with climate change could increaseeffects of air temperature on tadpole survival tometamorphosis, as water temperature varieswith both air temperature and water depth (Ber-ven 1982, Harkey and Semlitsch 1988). Nonethe-less, our results suggest that altered rainfallpatterns and hydroregimes associated with cli-mate change are likely to have a much biggerimpact on amphibians than temperature alone.

In our models, individual wetland alone(model 1) was not associated with JR for any spe-cies except A. terrestris and G. carolinensis, butwas significant in the strongest models for allspecies except A. quercicus. The influence of indi-vidual wetland could be a function of (hence cor-related with) characteristics such as basin size orshape that, in turn, could affect hydroregimecharacteristics (Figs. 2b and 3), wetland vegeta-tion, or substrate characteristics important forbreeding adults, oviposition sites, or tadpoles.Although all wetlands varied annually in hydro-regime characteristics, some contained watermore of the time, increasing chances of suitabletiming and duration of hydroperiods for multi-ple species. Philopatry could also influence wet-land selection by breeding adults over multiplegenerations (Pechmann et al. 2001, Semlitsch2008). Juveniles of many species were producedfrom wetlands that were dry much of the time,and recruitment failure or low average juvenileproduction often occurred in wetlands that heldwater much of the time during our 22-yr studyperiod (Figs. 2b, 3 and 5), highlighting theimportance of maintaining multiple ephemeralwetlands within landscapes.

Rank correlations of breeding adults to juve-nile recruits indicated that breeding effort was asignificant predictor of JR for all species. In con-trast, our multivariate models indicated thatbreeding effort was not an important predictor ofJR for L. capito, L. sphenocephalus, or A. quercicus.This seeming inconsistency was likely due to cor-relation between adult captures and other vari-ables in our models, such as continuoushydroperiod or total rainfall; a strong associationbetween breeding effort and JR for A. terrestris,G. carolinensis, and S. holbrookii suggests thatbreeding effort was an important predictor of JR,above and beyond the influence of weather and/or hydroregime. The influence of weather andhydroregime on breeding effort was not a focusof our study. However, these factors can stronglyinfluence adult movement (Greenberg and Tan-ner 2004, 2005a), calling activity (Saenz et al.2006), and (likely) breeding effort and thus havean important, indirect influence on JR.Sampling bias associated with drift fence trap-

ping (Dodd 1991) potentially affected resultsassociated with influence of breeding effort, aswell. For example, low adult capture rates forL. capito and L. sphenocephalus could be due topitfall trap escape by jumping, or trap avoidanceby going around or over fences; Pechmann et al.(1989) also reported that large Lithobates adults(L. catesbeianus and L. sphenocephalus) were likelyable to trespass fences and pitfall traps, althoughjuveniles were not. Alternatively or in addition,adult populations of L. capito and L. spheno-cephalus may be small compared to some otheranuran species, and relatively few adults may beable to produce a comparatively large number ofjuvenile recruits. Similarly, relatively low capturerates of G. carolinensis and A. quercicus could bedue to trespass, or because juveniles may remainin wetlands and grow for prolonged periodspost-metamorphosis.Our results suggest that annual variation in

reproductive success was most similar amongspecies having similar reproductive strategies. Forexample, JR was strongly associated between bothspecies having a prolonged breeding period, slowlarval development, and juvenile emigration notinitiated until late spring (L. capito and L. spheno-cephalus); both require prolonged hydroperiodsthrough late spring, so that slow-developing tad-poles can complete metamorphosis and emigrate


GREENBERG ET AL.

from wetlands. Similarly, JR was significantly pos-itively associated among all species pairs withrapid larval development (A. quercicus, A. ter-restris, and G. carolinensis), regardless of breedingperiod duration. Thus, hydroregime and weatherconditions could provide reproductive opportu-nity for species having a similar reproductivestrategy in some years or wetlands, but for specieswith a different strategy in others. Weather andhydroregime characteristics interact with life his-tory traits, or reproductive strategies, that differamong amphibian species and influence repro-ductive plasticity, opportunity, and success.

Clearly, reproductive opportunity in ephem-eral wetlands is higher for species with rapid lar-val development than for species with slowlarval development because of their ability toexploit shorter hydroperiods. Among specieswith rapid larval development, a protractedbreeding period offers greater reproductiveopportunity than a shorter, season-limited breed-ing season, by increasing chances of one or moresuitable hydroperiods. For example, A. terrestriscould potentially exploit multiple, short-durationhydroperiods during their extended (Februarythrough October) breeding season, whereas sum-mer-breeding A. quercicus or G. carolinensis (33–44 d or 23–67 d, respectively) would likelyencounter fewer suitable hydroperiods duringtheir shorter, summer breeding season whenephemeral wetlands are more often dry. Forrapidly developing species with a protractedbreeding period, either multiple short-duration(4–5 weeks) or prolonged hydroperiods withinrespective breeding seasons could allow continu-ous or intermittent breeding and multiplecohorts of juvenile recruits. The explosive breed-ing strategy and extremely rapid rate of larvaldevelopment (14 d) confer S. holbrookii with highreproductive plasticity. Nonetheless, S. holbrookiiare constrained by their requirement for specificweather cues (heavy rainfall, usually associatedwith late summer and fall tropical storms andhurricanes) and wetland conditions (dry wet-lands that refill rapidly; Greenberg and Tanner2004). In contrast, reproductive opportunity inephemeral wetlands is relatively lower for spe-cies with slow larval development, requiringprolonged, continuous hydroperiods for success-ful reproduction, such as Lithobates spp. Repro-ductive opportunity is further constrained for

L. capito and L. sphenocephalus, as tadpoles donot metamorphose or emigrate until late spring,relative to species that can metamorphose year-round (e.g., L. catesbeianus and L. grylio; C. H.Greenberg, unpublished data). Wetlands dryingbefore metamorphosis in May or June result inreproductive failure for L. capito and L. spheno-cephalus, regardless of prior breeding effort andhydroregime.Our results illustrate the complexity and

uncertainty involved in determining effects ofweather and hydroregime on amphibian recruit-ment. Although our models clearly implicatedprecipitation and specific hydroregime character-istics as important predictors of anuran JR, muchof the variability in JR remained unexplained.Our IT modeling approach used weather andhydroregime variables selected a priori based oninitial exploratory analysis and biological consid-erations. Nonetheless, correlation among multi-ple variables (both selected and unused) addedambiguity, and rendered it difficult to assesswhat specific variables or combination of vari-ables was most important in predicting repro-ductive success. For example, individual wetlandbasin shape and area, coupled with the amountand timing of precipitation, are likely correlatedwith hydroregime characteristics such ashydroperiod duration, frequency, timing, andfluctuations in water depth and surface area. Inturn, water temperature varies in relation to bothair temperature and depth or volume. Precipita-tion, temperature, and hydroregime characteris-tics are, in turn, closely correlated with breedingeffort by adults (Saenz et al. 2006). Thus, manyor all of these correlated variables are likely toinfluence amphibian reproductive success.Whereas unsuitable hydroperiod duration

during tadpole development can be clearly,causally linked to recruitment failure, manyunknown biotic factors, such as competition andpredation on eggs and larvae, likely contribute tohigh stochastic variation in amphibian reproduc-tive success.Competition within and among species, preda-

tion by aquatic macroinvertebrates on eggs andlarvae, larval density, and hydroregime can allaffect survival, developmental rate, and body sizeat metamorphosis (Wilbur 1980, 1982, Morin1983, Semlitsch et al. 1996, Richter et al. 2009,McMenamin and Hadly 2010). These factors are


GREENBERG ET AL.

confounded by temporal dynamics, includinginterspecific differences in the timing of oviposi-tion, rates of larval development, and apparentlyhigh mortality rates and reintroduction of neweggs and cohorts (Greenberg et al., in press),resulting in continual changes in larval assem-blages. Ongoing fluctuation of wetland surfacearea and depth, driven by precipitation, basinmorphology, and groundwater levels (Greenberget al. 2015), and continually changing competitionand predation pressure (Gascon and Travis 1992,Pearman 1993, 1995, Azevedo-Ramos and Mag-nusson 1999) further confound factors affectingsurvival to metamorphosis. Water temperature, adynamic interaction between changing air tem-perature and fluctuating water depth and surfacearea (Berven 1982, Harkey and Semlitsch 1988), inturn affects the growth and development rate oftadpoles (e.g., Blaustein et al. 2001, 2010, Corn2005, Todd et al. 2011). This is further confoundedby season and complicates prediction of tadpolesurvival to metamorphosis because effects couldvary with developmental stage. Additionally,amphibian breeding phenology is influenced byboth air temperature and precipitation (e.g., Corn2005, Saenz et al. 2006, Blaustein et al. 2010, Toddet al. 2011, Greenberg et al. 2014).

Our results suggest that moderate climatewarming may not directly affect anuran popula-tions, but associated changes with weather pat-terns including the amount, frequency, timing,and severity of precipitation will almost certainlyalter wetland hydroregimes (Greenberg et al.2014, 2015), breeding activity, and reproductivesuccess among amphibian species. Climate sce-narios with substantially reduced summer andfall precipitation would likely result in reducedgroundwater recharge and shorter, shallowerhydroperiods during the critical fall–winter–spring breeding cycle for L. capito and L. spheno-cephalus. These species may be especially vulnera-ble because their tadpoles do not metamorphoseuntil late spring; thus, even a prolonged hydrope-riod will result in reproductive failure if it driesbefore then. Alternatively, substantially reducedwinter precipitation would likely result in initialgroundwater levels too far belowground for sum-mer rains to raise aboveground during spring andsummer breeding seasons for species such asA. terrestris, A. quercicus, and G. carolinensis. Incontrast, increased frequency of tropical storms or

hurricanes (Karl et al. 2009) could provide morefrequent reproductive opportunity for S. hol-brookii, as high-severity storms prompt explosivebreeding events and often create suitable wetlandconditions for their very rapidly developing lar-vae. Clearly, altered weather patterns associatedwith climate warming will differentially affectreproductive success among amphibian specieswith different breeding strategies, with criticalimplications for amphibian population trends andassemblages (Greenberg et al. 2014, 2015). Ourstudy highlights the importance of maintainingmultiple ephemeral wetlands within landscapesto increase chances for reproductive success byspecies with different strategies, in at least someyears and wetlands.

ACKNOWLEDGMENTS

Funding was provided by the USDA Forest ServiceOcala National Forest, Longleaf Pine EcosystemRestoration Program, Southern Research Station, andSouthern Region (R8); the Department of Energy-Savannah River Operations (IA Agreement DE-AI09-76SR00056); and the Florida Fish and WildlifeConservation Commission (contracts NG99-014 andC1195). We thank G.W. Tanner, J. Beach, S. Wazny,D. Wooten, R. Ashton, M. Welker, J. Smith, J. Staiger,J. Barichivich, R. Owen, D. Johnson, S. Johnson,J. Wiebe, K. Garren, E. Roznik, T. Sheltra, R. Ashton,S. Doucette-Riisse, C.J. Kovach, C. Bugbee, R. Lara,L. Tirado, I. Luque, A. Heh, C. Hartmann, andG. Kamener for fieldwork, and the many volunteerswho assisted them. We also thank L. Lowery,R. Lowery, C. Sekerak, J. Hinchee, J. Clutts, J. Marr,M. Clere, M. Herrin, and the fire crew for assistance.We thank two anonymous reviewers who helped toimprove an earlier version of this manuscript.

LITERATURE CITED

Ashton Jr., R. E., and P. S. Ashton. 1988. Handbook ofreptiles and amphibians of Florida, part three: theamphibians. Windward Publishing Inc, Miami,Florida, USA.

Azevedo-Ramos, C., and W. E. Magnusson. 1999.Tropical tadpole vulnerability to predation: associ-ation between laboratory results and prey distribu-tion in an Amazonian savanna. Copeia 1999:58–67.

Babbitt, K. J., M. J. Baber, and T. L. Tarr. 2003. Patternsof larval amphibian distribution along a wetlandhydroperiod gradient. Canadian Journal of Zool-ogy 81:1539–1552.


GREENBERG ET AL.

Berven, K. A. 1982. The genetic basis of altitudinalvariation in the wood frog Rana sylvatica II. Anexperimental analysis of larval development.Oecologia 52:360–369.

Blaustein, A. R., L. K. Belden, D. H. Olson, D. M.Green, T. L. Root, and J. M. Kiesecker. 2001.Amphibian breeding and climate change. Conser-vation Biology 15:1804–1809.

Blaustein, A. R., S. C. Walls, B. A. Bancroft, J. J. Lawler,C. L. Searle, and S. S. Gervasi. 2010. Direct andindirect effects of climate change on amphibianpopulations. Diversity 2010:281–313.

Bridges, A. S., and M. E. Dorcas. 2000. Temporalvariation in anuran calling behavior: implicationsof surveys and monitoring programs. Copeia 2000:587–592.

Brooks, R. T. 2009. Potential impacts of global climatechange on the hydrology and ecology of ephemeralfreshwater systems of the forests of the northeast-ern United States. Climatic Change 95:469–483.

Burnham, K. P., and D. R. Anderson. 2002. Modelselection and multi-modal inference: a practicalinformation-theoretic approach. Second edition.Springer-Verlag, New York, New York, USA.

Cole, E. M., R. Hartman, and M. P. North. 2016.Hydroperiod and cattle use associated with lowerrecruitment in an r-selected amphibian with adeclining population trend in the Klamath Moun-tains, California. Journal of Herpetology 50:37–43.

Corn, P. S. 2005. Climate change and amphibians. Ani-mal Biodiversity and Conservation 28:59–67.

Dodd Jr., C. K.. 1991. Drift fence-associated samplingbias of amphibians at a Florida sandhills tempo-rary pond. Journal of Herpetology 25:1991.

Dodd Jr., C. K.. 1994. The effects of drought onpopulation structure, activity, and orientation oftoads (Bufo quercicus and B. terrestris) at a temporarypond. Ethology Ecology and Evolution 6:331–349.

Dodd Jr., C. K.. 1995. The ecology of a sandhillspopulation of the eastern narrow-mouthed toad,Gastrophryne carolinensis, during a drought. BulletinFlorida Museum of Natural History and BiologicalScience 38:11–41.

Gascon, C., and J. Travis. 1992. Does the spatial scaleof experimentation matter? A test with tadpolesand dragonflies. Ecology 73:2237–2243.

Godley, J. S. 1992. Threatened: gopher frog Rana capitoLe Conte. Pages 15–19 in P. E. Moler, editor. Rareand endangered biota of Florida, Volume 3.Amphibians and reptiles. University Press ofFlorida, Gainesville, Florida, USA.

Greenberg, C. H. 2001. Spatio-temporal dynamics ofpond use and recruitment in Florida gopherfrogs (Rana capito aesopus). Journal of Herpetology35:74–85.

Greenberg, C. H., S. Goodrick, J. D. Austin, and B. R.Parresol. 2015. Hydroregime prediction models forephemeral groundwater-driven sinkhole wetlands:a planning tool for climate change and amphibianconservation. Wetlands 35:899–911.

Greenberg, C. H., S. E. Johnson, R. Owen, and A. Stor-fer. In press. Amphibian breeding phenology andreproductive success at isolated ephemeral wet-lands: an examination using concurrent terrestrialand aquatic sampling. Canadian Journal of Zoology.

Greenberg, C. H., R. W. Perry, K. E. Franzreb, S. C.Loeb, D. Saenz, and D. C. Rudolph. 2014. Climatechange and wildlife in the southern United States:potential effects and management options. Pages379–420 in J. M. Vose and K. D. Klepzig, editors. Cli-mate change adaptation and mitigation managementoptions. CRC Press, New York, New York, USA.

Greenberg, C. H., and G. W. Tanner. 2004. Breedingpond selection and movement patterns by easternspadefoot toads (Scaphiopus holbrookii) in relationto weather and edaphic conditions. Journal ofHerpetology 38:569–577.

Greenberg, C. H., and G. W. Tanner. 2005a. Spatial andtemporal ecology of oak toads (Bufo quercicus) on aFlorida landscape. Herpetologica 61:422–434.

Greenberg, C. H., and G. W. Tanner. 2005b. Spatial andtemporal ecology of eastern spadefoot toads on aFlorida landscape. Herpetologica 61:20–28.

Harkey, G. A., and R. D. Semlitsch. 1988. Effects oftemperature on growth, development, and colorpolymorphism in the ornate chorus frog Pseudacrisornata. Copeia 1988:1001–1007.

IPCC. 2007. Climate Change 2007: Synthesis Report.52 pp. http://www.ipcc.ch/pdf/assessment-report/ar4/syr/ar4/syr.pdf

Karl, T. R., J. M. Melillo, and T. C. Peterson, editors.2009. Global climate change impacts in the UnitedStates. Cambridge University Press, UK. http://www.globalchange.gov/usimpacts

Knowles Jr., L., A. M. O’Reilly, and J. C. Adamski.2002. Hydrogeology and simulated effects of theground-water withdrawals from the Floridan aqui-fer system in Lake County and in the OcalaNational Forest and vicinity, north-central Florida.Water-Resources Investigations Report 02-4207. USGeological Society, Tallahassee, Florida, USA.

McCallum, M. L., S. E. Trauth, M. N. Mary, C. McDowell,and B. A. Wheeler. 2004. Fall breeding of the south-ern leopard frog (Rana sphenocephala) in northeast-ern Arkansas. Southeastern Naturalist 3:401–408.

McMenamin, S. K., and E. A. Hadly. 2010. Develop-mental dynamics of Ambystoma tigrinum in achanging landscape. BMC Ecology 10:10.

McMenamin, S. K., E. A. Hadly, and C. K. Wright.2008. Climatic change and wetland desiccation


GREENBERG ET AL.

http://www.ipcc.ch/pdf/assessment-report/ar4/syr/ar4/syr.pdf

http://www.ipcc.ch/pdf/assessment-report/ar4/syr/ar4/syr.pdf

http://www.globalchange.gov/usimpacts

http://www.globalchange.gov/usimpacts

cause amphibian decline in Yellowstone NationalPark. Proceedings of the National Academy ofSciences 105:16988–16993.

Moler, P. E., and R. Franz. 1987. Wildlife values ofsmall, isolated wetlands in the southeastern coastalplain. Pages 234–241 in R. R. Odum, K. A. Riddle-berger, and J. C. Ozier, editors. Proceedings of theThird Southeastern Nongame and EndangeredWildlife Symposium. Georgia Department of Natu-ral Resources, Atlanta, Georgia, USA.

Morin, P. J. 1983. Predation, competition, and the com-position of larval anuran guilds. Ecological Mono-graphs 53:119–138.

Palis, J. G. 1998. Breeding biology of the gopher frog,Rana capito, in western Florida. Journal of Herpetol-ogy 32:217–223.

Pearman, P. B. 1993. Effects of habitat size on tadpolepopulations. Ecology 73:1982–1991.

Pearman, P. B. 1995. Effects of pond size and conse-quent predator density on two species of tadpoles.Oecologia 102:1–8.

Pechmann, J. H. K., R. A. Estes, D. E. Scott, and J. W.Gibbons. 2001. Amphibian colonization and use ofponds created for trial mitigation of wetland loss.Wetlands 21:93–111.

Pechmann, J. H. K., D. E. Scott, J. W. Gibbons, andR. D. Semlitsch. 1989. Influence of wetland andhydroperiod on diversity and abundance of meta-morphosing juvenile amphibians. Wetlands Eco-logy and Management 1:3–11.

Phillips, K. M. 1995. Rana capito capito, the Carolinagopher frog, in southeast Georgia: reproduction,early growth, adult movement patterns, and tad-pole fright response. M.S. Thesis. Georgia SouthernUniversity, Statesboro, Georgia, USA.

Richter, J., L. Martin, and C. K. Beachy. 2009. Increasedlarval density induces accelerated metamorphosisindependently of growth rate in the frog Rana sphe-nocephala. Journal of Herpetology 43:551–554.

Richter, S. C., J. E. Young, G. N. Johnson, and R. A. Sei-gel. 2003. Stochastic variation in reproductive suc-cess of a rare frog, Rana sevosa: implications forconservation and for monitoring amphibian popu-lations. Biological Conservation 111:171–177.

Saenz, D., L. A. Fitzgerald, K. A. Baum, and R. N. Con-ner. 2006. Abiotic correlates of anuran calling phe-nology: the importance of rain, temperature, andseason. Herpetological Monographs 20:64–82.

SAS Institute Inc. 2011. SAS/STAT� 9.3 user’s guide.SAS Institute Inc., Cary, North Carolina, USA.

Semlitsch, R. D. 2008. Differentiating migration anddispersal processes for pond-breeding amphibians.Journal of Wildlife Management 72:260–267.

Semlitsch, R. D., J. W. Gibbons, and T. S. Tuberville.1995. Timing of reproduction and metamorphosisin the Carolina gopher frog (Rana capito capito)in South Carolina. Journal of Herpetology 29:612–614.

Semlitsch, R. D., D. E. Scott, J. H. K. Pechmann, andJ. W. Gibbons. 1996. Structure and dynamics of anamphibian community: evidence from a 16-yearstudy of a natural pond. Pages 217–248 in M. L.Cody and J. A. Smallwood, editors. Long-termstudies of vertebrate communities. Academic Press,San Diego, California, USA.

Snodgrass, J. W., M. J. Komoroski, A. L. Bryan Jr., andJ. Burger. 2000. Relationships among isolated wet-land size, hydroperiod, and amphibian speciesrichness: implications for wetland regulations.Conservation Biology 14:414–419.

Stewart, M. M. 1995. Climate driven population fluctu-ations in rain forest frogs. Journal of Herpetology29:437–446.

Todd, B. D., D. E. Scott, J. H. Pechmann, and J. W. Gib-bons. 2011. Climate change correlates with rapiddelays and advancements in reproductive timingin an amphibian community. Proceedings of theRoyal Society of London B: Biological Sciences278:2191–2197.

Vose, J. M., and K. D. Klepzig. 2014. Introduction toclimate change adaptation and mitigation manage-ment options. Pages 1–8 in J. M. Vose and K. D.Klepzig, editors. Climate change adaptation andmitigation management options. CRC Press, NewYork, New York, USA.

Walls, S. C., W. J. Barichivich, M. E. Brown, D. E. Scott,and B. T. Hossack. 2013a. Drought, deluge anddeclines: the impact of precipitation extremes onamphibians in a changing climate. Biology2013:399–418.

Walls, S. C., W. J. Barichivich, M. E. Brown, D. E. Scott,and B. R. Hossack. 2013b. Influence of drought onsalamander occupancy of isolated wetlands on thesoutheastern Coastal Plain of the United States.Wetlands 33:345–354.

Wilbur, H. M. 1980. Complex life cycles. AnnualReview of Ecology and Systematics 11:67–93.

Wilbur, H. M. 1982. Competition between tadpoles ofHyla femoralis and Hyla gratiosa in laboratory exper-iments. Ecology 63:278–282.

Winsberg, M. D. 1990. Florida weather. University ofCentral Florida Press, Orlando, Florida, USA.

Wright, A. H. 2002. Life-histories of the frogs of Okefi-nokee swamp, Georgia. North American Salientia(Anura) No. 2. Cornell University Press, Ithaca,New York, USA.


GREENBERG ET AL.

weather, hydroregime, and breeding effort influence ... hydroregime, and breeding effort inﬂuence...

Documents