I!TRODUCTIO!

Large, catastrophic ecological disturbances areintegral, yet little-understood, events in forestecosystems. Commonly associated withinfrequent disturbances such as catastrophicwildfires, windstorms, floods, or winter weathersystems, these events often producetransformative and long-lasting impacts on foreststructure, succession, and composition (Lorimer1989; Foster et al. 1998; Turner et al. 1998).Windstorms such as hurricanes and tornadoes aresome of the most common large disturbancespresent in forest ecosystems across thesoutheastern United States (Greenberg andMcNab 1998; Peterson 2000; Duryea et al.2007). Although the timing of these events isultimately stochastic and typically infrequent,they are nonetheless ubiquitous in thedisturbance history of individual ecosystems.For example, while over 900 strong (EnhancedFujita Scale 3 or higher) tornadic systems haveimpacted the United States since 1880, the returnintervals of these systems are highly variable,ranging from 10 years to several centuries,depending on location (Broyles and Crosbie2004). Severe thunderstorm activity thatgenerates these extreme wind disturbance events,however, is expected to increase in frequencyand severity in coming decades, given current

climatological trends (Garinger and Knupp1993; Trapp et al. 2007).

The effects of such large, infrequent winddisturbances on amphibian populations havereceived little focus and, when studied, oftenshow variable impacts. Greenberg (2001), forexample, found no significant effect ofdownburst-produced canopy gaps on amphibianabundance, species richness, and diversity in asouthern Appalachian hardwood forest.However, Woolbright (1991) observedsignificant shifts in population density for forestfrogs in Puerto Rico following wind disturbancefrom Hurricane Hugo. The stochastic nature ofsuch events has forced many researchers to relyon surrogate, ‘undisturbed’ reference habitats forthe purposes of comparative analyses. Explicitcomparisons of pre- and post-disturbance datafrom the same sites are uncommon, yet representa more robust approach to evaluating the effectsof a given disturbance (see Woolbright 1991).Such before-and-after study designs have beenpreviously used to examine the responses ofamphibians to large tropical systems (Schrieveret al. 2009; Gunzburger and Hughes 2010), butsuch studies are lacking for extreme wind eventsassociated with tornadic systems, due to thesesystems’ highly stochastic nature and relativelysmall size.

On the evening of 27 April 2011, an unusually

Herpetological Conservation and Biology

AMPHIBIA!S A!D LARGE, I!FREQUE!T FOREST DISTURBA!CES: A!EXTREME WI!D EVE!T FACILITATES HABITAT CREATIO! A!D A!URA!

BREEDI!G

WALTER H. SMITH

Department of !atural Sciences, The University of Virginia’s College at Wise, One College Avenue, Wise, Virginia 24293, USA,e-mail: whs2q@uvawise.edu

Abstract.—Large, catastrophic wind disturbances are historically common across the Coastal Plain of the southeasternUnited States, a region which harbors one of the planet’s most speciose regions for amphibian taxa. Links between suchdisturbances and the amphibians they potentially impact, however, have received little focus, and studies addressing theseimpacts are often limited by an inability to compare post-disturbance survey data to pre-disturbance conditions at the samesites. I conducted habitat and anuran surveys following an intense tornadic disturbance in 2011 within a series of transectsin the upper Coastal Plain of central Alabama that has been studied continuously since 2009, with the purpose of examiningthe effects of catastrophic wind events. These surveys revealed that several Southern Toads (Anaxyrus terrestris) appearedto have been killed due to storm impacts, while Cope’s Gray Treefrogs (Hyla chrysoscelis) began breeding in the storm-damaged area immediately following the storm. This was the first instance of anuran breeding activity; none had occurredduring the preceding three years in this transect. Breeding activity in H. chrysoscelis was facilitated by the creation ofmultiple, isolated pools via debris scouring as a direct result of the tornado. Although extreme wind events are oftenstochastic and infrequent, these results illustrate their importance for habitat creation within storm-damaged areas, andprovide potentially useful information for managers charged with monitoring amphibian populations following extremewind disturbance.

Key Words.—amphibian; anuran; disturbance; Cope’s Gray Treefrog; Hyla chrysoscelis; tornado; wetland; wind

Copyright © 2013. Walter Smith. All Rights Reserved. 732

Herpetological Conservation and Biology 8(3):732−740. Submitted: 15 March 2013; Accepted: 29 July 2013; Published: 31 December 2013.

large and violent EF–3 (Enhanced Fujita Scale)tornado moved through a portion of theTalladega National Forest, Oakmulgee District,in Bibb and Hale Counties of west-centralAlabama. Containing surface winds estimatedat up to 230 km per hour, this system impacted a1.6-km wide and 115 km long region within anexisting mixed pine-hardwood forest (NOAA2011a). This event removed approximately100,000 m3 of forest cover across a 600-haportion of the Oakmulgee District (Ragland2011). The southwestern portion of this storm’spath directly impacted an existing transect in thePine Flat area of the Oakmulgee District that hasbeen used to study anuran populations andassociated wetland habitats continuously since2009. This portion of the storm track includedsome of the most intense and widespread winddamage present throughout the affected area.

The coincidence of a severe tornado occurringon the aforementioned transect presents a rareopportunity to directly examine the effects of alarge, infrequent forest disturbance on anuranpopulations and to compare post-disturbanceobservations to multiple years of pre-disturbancedata. In addition, high-resolution orthoimagerycollected immediately following this storm(NOAA 2011b) provides the means to analyzespatial patterns of breeding habitat alteration.Specifically, I couple geospatial analyses withpost-disturbance field surveys to analyzepatterns of tornado induced formation of scourpools and subsequent colonization of pools bybreeding anurans. I discuss potentialimplications of these observations foramphibians and forest management within thecontext of intense, infrequent disturbances.

MATERIALS A!D METHODS

Anuran surveys.—I commenced anuransurveys in the Oakmulgee District of Bibb andHale Counties of west-central Alabama inFebruary 2009 as part of a larger studyinvestigating the response of breeding anuranassemblages to forest management in the upperGulf Coastal Plain (Smith 2011). I establishedthree transects (each approximately 1 km inlength) within midslope and bottomland foresthabitat directly adjacent to an unnamed, third-order tributary of Fivemile Creek (Table 1).Forest types within this area ranged fromextensive Longleaf Pine (Pinus palustris) forestin midslope and ridgetop areas to large stands of

Black Tupelo (!yssa sylvatica) along creekbottoms. Two of these transects were locatedadjacent to actively-maintained beaverimpoundments, while the third was locatedadjacent to a former, drained beaverimpoundment transitioning back to foresthabitat.

I surveyed each transect approximately twiceper month, from February to July, for breedinganurans and the presence/condition of aquatichabitat. I inventoried aquatic habitat (defined asephemeral pools of standing water separate fromthe main channels of permanent streams) bywalking each transect and documenting thelocation of each habitat feature withinapproximately 20 meters of the transectcenterline (a gated forest road). Anuran surveysconsisted of nocturnal, auditory call surveysfollowing the guidelines and abundance indicesrecommended by Weir and Mossman (2005).When calling anurans were detected, Iperformed diurnal surveys of the associatedaquatic habitats for the presence of anuran eggmasses to confirm breeding activity. I performedthe final anuran call survey preceding thedisturbance event on 11 April 2011.

Post-disturbance surveys.—I performed aninitial visual survey of each transect on 30 April2011. Surveys consisted of walking eachtransect and documenting the location and typeof aquatic habitat present within each transect,similar to pre-disturbance surveys. Followingthese initial surveys, I performed standardsurveys (as described above) weekly to monitoraquatic habitat for evidence of successful anuranbreeding. I continued these surveys throughSeptember 2011. Newly created pools had driedby then and the U.S. Department of AgricultureForest Service began salvage logging activities,preventing site access.

Only one of the three transects had stormdamage. This transect was located directly in thepath of the tornado, in an area estimated to haveexperienced EF-3 scale wind damage (NOAA2011a). The other two transects did not receivesignificant habitat disturbance at any pointthroughout the three-year study period,effectively creating a before-after control-impact(BACI) study design. I used linear mixed-effectsmodels in Minitab v16 (Minitab, Inc., StateCollege, Pennsylvania, USA) to examine theeffects of disturbance on species richness andpool count, with disturbance (impacted and

Smith—Tornadic forest disturbance and anurans.

733

control) and year (2009, 2010, and 2011) as fixedeffects. I included transect as a random effect,and ln-transformed counts (species and pools)prior to model construction. Emphasis wasplaced on tests of the interaction term (year ×disturbance) due to this term’s value in a BACIdesign (Underwood 1994; McDonald et al.2000); tests of main effects are therefore notreported hereafter. I used a significance level ofα = 0.05 for all statistical analyses.

Geospatial analyses.—I used a combination ofdata collected in the field and digital datasetsprepared by the National Oceanic andAtmospheric Administration (NOAA) toexamine spatial patterns of newly-createdaquatic habitat within the storm-damagedtransect. I mapped the location of each newly-created habitat feature within this transect in

ArcGIS v.10.1 (Esri, Redlands, California, USA)using geographic coordinates collected using aGarmin 60CS handheld GPS unit (approx. 3–4meters accuracy; Garmin International Inc.,Olathe, Kansas, USA). I overlaid these pointswith a digital elevation model (10 m resolution)of the study area produced by the U.S. GeologicSurvey (USGS) and high-resolution (25-cmground sample distance) digital orthoimageryproduced during post-disturbance aerial surveysby the National Geodetic Survey of NOAA(NOAA 2011b).

I obtained wind and terrain vectors at thelocation of each newly-created breeding siteusing this dataset in order to test the hypothesisthat interactions between wind and terrainpromote debris scouring of topsoil and,subsequently, pool creation. I estimateddirection (azimuth) of wind vectors by

Herpetological Conservation and Biology

734

TABLE 1. Transect location, pool counts, and species observed from 2009 to 2011 in the Oakmulgee District.Geographic coordinates refer to latitude/longitude for each transect’s starting point (western terminus) and ending point(eastern terminus) along an unnamed tributary of Fivemile Creek. The transect labeled “disturbed” refers to the areaimpacted by tornadic winds in 2011.

Transect StartingCoordinates

EndingCoordinates Year # pools Species Observed

Undisturbed 32.90671, -87.39526

32.90885, -87.38944 2009 4

Acris crepitans, Anaxyrus terrestris, Hylaavivoca, Hyla chrysoscelis, Hyla cinerea,

Lithobates catesbeianus, Lithobates clamitans

2010 4Acris crepitans, Hyla avivoca, Hyla

chrysoscelis, Hyla cinerea, Lithobatescatesbeianus, Lithobates clamitans

2011 (Post-disturbance)* 4

Acris crepitans, Hyla avivoca, Hylachrysoscelis, Hyla cinerea, Lithobates

catesbeianus, Lithobates clamitans

Undisturbed 32.90370, -87.40700

32.90329, -87.39991 2009 5

Acris crepitans, Anaxyrus terrestris, Hylaavivoca, Hyla chrysoscelis, Hyla cinerea,

Lithobates catesbeianus, Lithobates clamitans

2010 5Acris crepitans, Hyla avivoca, Hyla

chrysoscelis, Hyla cinerea, Lithobatescatesbeianus, Lithobates clamitans

2011 (Post-disturbance)* 5

Acris crepitans, Hyla avivoca, Hylachrysoscelis, Hyla cinerea, Lithobates

catesbeianus, Lithobates clamitans

Disturbed 32.91096, -87.44071

32.91163, -87.43043 2009 1 Lithobates clamitans

2010 1 Lithobates clamitans

2011 (Post-disturbance)* 16 Acris crepitans, Anaxyrus terrestris, Hyla

chrysoscelis, Lithobates clamitans

*visits performed as weekly surveys, rather than twice-monthly pre-disturbance survey visits

examining the pattern of treefall and direction oftopsoil scouring (indicators of wind direction) ateach pool via field visits and examination of thedigital orthoimagery. I obtained the direction(azimuth) of terrain vectors at each pool locationby performing an aspect analysis on the USGSdigital elevation model in the Spatial Analystextension of ArcGIS v.10.1. I verified terrainaspect in the field.

I overlaid a set of 16 random points (equivalentto the number of actual aquatic habitat featurespresent following disturbance) across the samespatial extent of the storm-damaged transect inArcGIS v.10.1 to determine if the relationshipbetween wind and terrain vectors at newly-created pools was significantly different fromthat expected due to random chance. I recordedthe azimuth of wind and terrain vectors at eachof these random points, and calculated theabsolute difference (degrees) between vectors forboth newly-created pools and random points. Icompared differences between these groupsusing a Student’s t-test (α = 0.05) in Minitabv.16. A significant difference between wind andterrain vector angles from scour pools andrandom points indicates nonrandom interactionsbetween wind and terrain in creating pools. Dueto a low sample size of newly-formed pools, Iperformed 100 bootstrap samples of this datasetand compared the observed t statistic for theactual dataset to the resulting distribution of tstatistics from resampling to further assesssignificance.

RESULTS

The tornadic system produced significant

physical changes in forest cover within thestorm-impacted transect, removing all canopycover and, in most cases, removing vegetationdown to the ground layer. By contrast, the othertwo transects exhibited no change in foreststructure (Fig. 1). Coarse woody debrismobilized by the wind event collected in largepiles within the storm-damaged transect,particularly at the base of embankments andstanding but heavily-damaged trees. Three deadSouthern Toads (Anaxyrus terrestris) were foundwithin these debris piles during immediate post-disturbance surveys (Fig. 2a). All had significantphysical trauma to the head and neck. I observedno mortality in the two transects not impacted bythe storm.

Debris piles also resulted in the creation ofnumerous isolated pools within the storm-damaged transect via topsoil scouring (Fig. 2b).The interaction term (year × disturbance) wassignificant for pool count (F1,8) = 9.449, P =0.028), indicating an effect of the tornado on theabundance of aquatic habitat. Specifically, Irecorded 16 pools immediately following thedisturbance event in the tornado-impactedtransect, fifteen of which were not present duringthe preceding two years of habitat surveys (Table1). By contrast, I detected no new pools or otherforms of aquatic habitat following thedisturbance event in the other two transectsoutside of the storm’s path. Four (25%) of thenew pools occurred close enough to the forestroad bisecting the transect to experience someminor anthropogenic modification (widening) asa result of road clearing in preparation forsalvage logging in summer 2011. The other

Smith—Tornadic forest disturbance and anurans.

735

FIGURE 1. Forest habitat (A) in an undisturbed transect and (B) in the disturbed transect at the same time period inspring 2011. (Photographed by Walter H. Smith).

A B

pools, however, occurred far from this roadbedand did not experience changes beyond theinitial debris scouring.

Similarly, the interaction term was significantfor species count (F1,8 = 12.050, P = 0.018),indicating an effect of the tornado on anuranpopulations. The most significant change inspecies count occurred in the tornado-disturbedtransect, in which three species were recordedfollowing disturbance that had not been recordedin the years prior. These species were theNorthern Cricket Frog (Acris crepitans),Southern Toad (Anaxyrus terrestris), and Cope’sGray Treefrog (Hyla chrysoscelis; Table 1).Within two days of the disturbance event, Iencountered H. chrysoscelis egg masses withinnewly-created pools in this transect andeventually found breeding activity in 10 of 16pools. These were the first observations ofanuran breeding within this transect over thethree years of survey efforts (Fig. 3).

Egg mass counts from this transect totaled 53clutches. Average clutch size for H. chrysoscelisis 30–40 eggs per clutch (Wright 1932; Greer2008), so total egg deposition at this site mayhave summed to 1,590–2,120 individual eggs.Larvae exhibited typical development for H.

chrysoscelis throughout the survey period, andall pools were empty of larvae by 8 September2011. Pools subsequently dried, with all poolsdry by 25 September 2011 (Fig. 4).

Spatial analyses indicated a non-randompattern of pool creation across the storm-damaged transect. Specifically, angles ofdifference between wind and terrain vectors atpools created by debris scours were significantlygreater than random differences (t = -3.53, df =30, P < 0.001; resampling P = 0.01, 100resamples). Data for this analysis were normallydistributed (P = 0.571 Shapiro-Wilk NormalityTest). This significant difference indicates thatinteractions between wind direction and terrainat the location of scour pools exhibited greaterresistance than those present at random pointsacross the same landscape.

DISCUSSIO!

Large, infrequent forest disturbances produceintense and transformative impacts on forestcomposition and health through the large-scaleremoval of forest cover. In the OakmulgeeDistrict, impacts from an intense mesocycloneresulted in the removal of forest cover down to

Herpetological Conservation and Biology

736

FIGURE 2. Images of (A) one of three dead SouthernToads (Anaxyrus terrestris) encountered during immediatepost-disturbance surveys, and (B) a newly-created scourpool following wind disturbance. Wind direction in thisphoto traveled from behind the photographer, generatingthe pile of woody debris located just beyond the picturedpool. Herbaceous plant growth in and around the pooloriginated following disturbance (photograph taken 8September 2011). (Photographed by Walter H. Smith).

A B

bare topsoil, including not only mature trees butalso smaller woody and herbaceous vegetation.This resulted in mobilization of large amounts oforganic matter which, while airborne for only amatter of seconds, produced large piles of woodydebris and associated debris scour paths acrossthe ground surface. This immediate damage andtopsoil gouging by debris is consistent withknown impacts from EF-3 scale and highertornadic systems, as evidenced from previoussurveys and studies (Marshall 2002; Karstens etal. 2010; NOAA 2012). Post-disturbancesurveys found evidence of direct effects of thetornado on anuran mortality and indirect effectsthrough the creation of breeding pools in scouredareas.

Although anuran mortality was not a primaryfocus of this study, examination of the dead toads(A. terrestris) found lodged in woody debrisindicated direct lethal effects by trauma.

Smith—Tornadic forest disturbance and anurans.

737

FIGURE 3. Location and condition (eggs present, larvae present, water only) for 16 ephemeral pools following winddisturbance in a transect studied continuously since 2009. The light-colored band in the center of the photo indicatesdeforestation from wind disturbance in 2011. Blue cross-hatched areas represent the location of abandoned,unmaintained beaver ponds within the study region. Inset image denotes the location (starred symbol) of the studysite with respect to additional tornado locations on 27 April 2011. Tornado track data and site orthoimagery weresourced from the National Oceanic and Atmospheric Administration (NOAA 2011b).

FIGURE 4. Proportion of ephemeral pools (n = 16)containing water (dotted line) and Hyla chrysoscelis eggsor larvae (solid line) over time following tornadicdisturbance in the Oakmulgee District in 2011. Opencircles denote individual sampling occasions.

Mortality probably did not cause demographicchanges in populations of A. terrestris due to thestudy site’s small size and location far from anyestablished breeding sites of A. terrestris.However, storm-induced mortality was almostcertainly underestimated because most debrispiles were either inaccessible due to safetyconsiderations or could not be surveyed withoutthe use of heavy equipment.

In contrast to mortality, indirect impactsinvolving the creation of anuran breeding habitatcaused the largest changes between pre- andpost-disturbance conditions. Specifically, thecreation of potholes and debris scours bytornadic winds and associated debrismobilization formed ephemeral, yet highlyabundant, pools of standing water across theimpacted landscape. Cope’s Gray Treefrogsbred in these new pools, in a transect which hadpreviously lacked anuran breeding activity otherthan two observations of calling Bronze Frogs(Lithobates clamitans) throughout 32 hours ofobservation during the preceding two years.

H. chrysoscelis will breed in isolated pools inhighly disturbed habitats. Hillis et al. (1987), forexample, found this species to use floodedfurrows of cornfields in Kansas as sites of eggdeposition, and roadside ditches or otheranthropogenically-created pools are acommonly-cited location for both calling andegg deposition for this species (Mount 1975;Greer 2008). Experimental work, in fact, showsa preference for open-canopy breeding sites inGray Treefrogs (H. versicolor) subjected toanthropogenic forest disturbance (Hocking andSemlitsch 2007). The observations of H.chrysoscelis colonizing and breeding in isolatedpools within a recently-disturbed habitat, asreported here, are not surprising. However, theseresults are notable in that they highlight the roleof large, catastrophic wind disturbance events infacilitating the establishment of amphibianbreeding sites within a forest ecosystem.

These data additionally show that anurans notonly used scour pools for egg deposition, buttadpoles were able to successfully complete theirdevelopment prior to pool desiccation. Theextent of new recruitment in response to tornado-altered habitat may be substantial, consideringthat the transect used in this study representedless than one percent of the area of theOakmulgee District directly impacted by thistornado. Opportunistic surveys of additionalportions of the storm track showed similar

evidence of anuran breeding, even on the crestsof ridgelines > 1 km from any known source ofstanding water. These data suggest that suchlarge, catastrophic forest disturbances maycontribute substantially to the establishment andsuccess of anuran breeding sites across the largerlandscape.

More importantly, the creation of breedinghabitat across the impacted landscape was not anentirely stochastic process. Instead, scour poolswere located more frequently in areas ofmaximum resistance between wind and terrainvectors than would be expected due to randomchance. Mesocyclonic systems such as thoseimpacting the Oakmulgee District are known forproducing tornadoes, which in many cases canbe intense. As air is pulled into these stormsystems, debris is often mobilized and can gougeinto topsoil when surface winds drive debris intothe ground surface (Karstens et al. 2010). Large,wind-borne objects such as trees and othercoarse woody debris are more likely to gougeinto topsoil if traveling perpendicular to – ratherthan parallel with – the aspect of the groundsurface. This arrangement of wind and localmicrotopography would produce higher amountsof force on topsoil and thus increase theprobability that debris would gouge an area largeand deep enough to retain rainfall during andimmediately after the wind disturbance itself.

While large, catastrophic forest disturbancesare ultimately stochastic events with respect towhen and where they will occur, these dataindicate that there may be some predictability indetermining where potential breeding habitatmay be created within a disturbed area. Usingknowledge of local terrain and storm movement,forest managers may be able to pinpoint areaswithin a storm-damaged region that have a highlikelihood of containing newly-created breedinghabitat for anurans. Additional factors, such asdebris size, soil type, and slope of terrain, mayalso influence pool formation beyond terrainaspect and wind direction. I did not assess thesevariables because large machinery would havebeen required to excavate most debris piles toaccurately measure tree type and size. Salvagelogging activities in summer 2011 preventedfollow-up studies that could have addressedthese variables in detail. Future investigationinto patterns of pool creation following winddisturbance should take these complicatingfactors into account, while using the predictivevalue of terrain aspect and wind direction as a

Herpetological Conservation and Biology

738

foundation.To build upon the observations reported here

and facilitate pre- and post-disturbancecomparisons, there is a need to establish andmaintain access to more long-term monitoringsites for anurans in forests prone to winddisturbance and subsequent salvage logging.Moisture availability and temperature in thewestern Gulf Coastal Plain contribute to thisregion’s status as both one of the most activeregions in the world for tornadic wind events anda temperate hotspot of anuran diversity (Buckleyand Jetz 2007; Trapp et al. 2007). Legacies ofwind disturbance in this region undoubtedlyinfluence local anuran populations, particularlywith respect to spatial patterns of breedingactivity, but the extent to which storm effectsmay impact anuran populations is just beginningto be understood. Although salvage logging iscommonly used in wind-disturbed areas toremove damaged timber and reduce fire risk,these activities may take place within areascontaining newly-created aquatic habitat andbreeding anurans. Timber harvesting may eitherput a developing cohort at risk by eliminatingnew aquatic habitat or increase habitatavailability by expanding pool number anddepth, and these impacts should be of interest tofuture studies of post-disturbance managementstrategies. Expanding the study of pool creationby storms, not just with discrete tornadic windevents but to other broad-scale winddisturbances initiated by tropical systemsthroughout the Gulf Coastal Plain region, wouldgreatly improve our knowledge of how winddisturbance affects anurans, how different taxarespond to this form of habitat change, and therole of post-storm forest management.

Acknowledgments.—Cynthia Raglandprovided access to the Talladega National Forest(Oakmulgee District) for the establishment ofmultiyear anuran surveys. J.J. Apodaca, HeatherCunningham, and Reid Downer assisted in pre-disturbance nocturnal call surveys. Katie Dunnand Sam Smith provided valuable fieldassistance with both diurnal and nocturnal post-disturbance surveys. Milton Ward providedvaluable conceptual support and comments onoriginal anuran survey design. Research for thisstudy was performed using funds provided bythe Birmingham Audubon Society’s Walter F.Coxe Research Fund.

LITERATURE CITED

Broyles, C., and C. Crosbie. 2004. Evidence ofsmaller tornado alleys across the United Statesbased on a long track F3 to F5 tornadoclimatology study from 1880 to 2003.Proceedings of the 22nd Conference on LocalSevere Storms. American MeteorologicalSociety, Hyannis, Massachusetts, USA.

Buckley, L.B., and W. Jetz. 2007. Environmentaland historical constraints on global patterns ofamphibian richness. Proceedings of the RoyalSociety B 274:1167–1173.

Duryea, M.L., E. Kampf, and R.C. Littell. 2007.Hurricanes and the urban forest: I. Effects onsoutheastern United States Coastal Plain treespecies. Arboriculture and Urban Forestry33:83–97.

Foster, D.R., D.H. Knight, and J.F. Franklin.1998. Landscape patterns and legaciesresulting from large, infrequent forestdisturbances. Ecosystems 1:497–510.

Garinger, L.P., and K.R. Knupp. 1993. Seasonaltornado climatology for the southeasternUnited States. Geophysical Monograph Series79:445–452.

Greenberg, C.H. 2001. Response of reptile andamphibian communities to canopy gapscreated by wind disturbance in the southernAppalachians. Forest Ecology andManagement 148:135–144.

Greenberg, C.H., and W.H. McNab. 1998. Forestdisturbance in hurricane-related downbursts inthe Appalachian mountains of North Carolina.Forest Ecology and Management 104:179–191.

Greer, G.C. 2008. Cope’s Gray Treefrog (Hylachrysoscelis). Pp. 56–58 In Amphibians andReptiles of Georgia. Jensen, J.B., C.D. Camp,W. Gibbons, and M.J. Elliott (Eds.). TheUniversity of Georgia Press, Athens, Georgia,USA.

Gunzburger, M., and W. Hughes. 2010.Hurricane storm surge and amphibiancommunities in coastal wetlands ofnorthwestern Florida. Wetlands Ecology andManagement 18:651–663.

Hillis, D.M., J.T. Collins, and J.P. Bogart. 1987.Distribution of diploid and tetraploid speciesof Gray Treefrogs (Hyla chrysoscelis and Hylaversicolor) in Kansas. American MidlandNaturalist 117:214–217.

Hocking, D.J., and R.D. Semlitsch. 2007. Effectsof timber harvest on breeding-site selection by

Smith—Tornadic forest disturbance and anurans.

739

Gray Treefrogs (Hyla versicolor). BiologicalConservation 138:506–513.

Karstens, C.D., T.M. Samaras, W.A. Gallus,C.A. Finley, and B.D. Lee. 2010. Analysis ofnear-surface wind flow in close proximity totornadoes. Proceedings of the 25th Conferenceon Severe Local Storms. AmericanMeteorological Society. Denver, Colorado,USA.

Lorimer, C.G. 1989. Relative effects of smalland large disturbances on temperate hardwoodforest structure. Ecology 70:565–567.

McDonald, T.L., W.P. Erickson, and L.L.McDonald. 2000. Analysis of count data frombefore-after control-impact studies. Journal ofAgricultural, Biological, and EnvironmentalStatistics 5:262–279.

Marshall, T.P. 2002. Tornado damage survey atMoore, Oklahoma. Weather and Forecasting17:582–598.

Mount, R.H. 1975. The Reptiles and Amphibiansof Alabama. The University of Alabama Press.Tuscaloosa, Alabama, USA.

NOAA (National Oceanic and AtmosphericAdministration). 2011a. Sawyerville-Eoline(Greene, Bibb, and Hale Counties) EF–3Tornado Summary. National Weather ServiceForecast Office, Birmingham, Alabama, USA.

NOAA (National Oceanic and AtmosphericAdministration). 2011b. Aerial photography:tornado response imagery of the surroundingregions. Silver Spring, Maryland, USA.

NOAA (National Oceanic and AtmosphericAdministration). 2012. Tornadoes of March 2,2012 – Storm Survey. National WeatherService Forecast Office, Louisville, Kentucky,USA.

Peterson, C.J. 2000. Catastrophic wind damageto North American forests and the potentialimpact of climate change. Science of the TotalEnvironment 262:287–311.

Ragland, C. 2011. 2011 Tornado damage – PineFlat salvage. Review of proposed action.USDA Forest Service, Talladega NationalForest, Oakmulgee District. Brent, Alabama,USA. 11 pp.

Schriever, T., J. Ramspott, B. Crother, and C.Fontenot. 2009. Effects of hurricanes Ivan,Katrina, and Rita on a southeastern Louisianaherpetofauna. Wetlands 29:112–122.

Smith, W.H. 2011. Ecological determinants ofherpetofaunal assemblage structure in theLongleaf Pine ecosystem. Ph.D. Dissertation,University of Alabama, Tuscaloosa, Alabama,

USA. 137 p.Trapp, R.J., N.S. Diffenbaugh, H.E. Brooks,

M.E. Baldwin, E.D. Robinson, and J.S. Pal.2007. Changes in severe thunderstormenvironment frequency during the 21st centurycaused by anthropogenically enhanced globalradiative forcing. Proceedings of the NationalAcademy of Sciences 104:19719–19723.

Turner, M.G., W.L. Baker, C.J. Peterson, andR.K. Peet. 1998. Factors influencingsuccession: lessons from large, infrequentnatural disturbances. Ecosystems 1:511–523.

Underwood, A.J. 1994. On beyond BACI:Sampling designs that might reliably detectenvironmental disturbances. EcologicalApplications 4:3–15.

Weir, L.J., and M.J. Mossman. 2005. NorthAmerican Amphibian Monitoring Program(NAAMP). Pp. 307–313 In AmphibianDeclines: Conservation Status of United StatesAmphibians. Lannoo, M. (Ed.). University ofCalifornia Press, Berkeley, California, USA.

Woolbright, L.L. 1991. The impact of HurricaneHugo on forest frogs in Puerto Rico.Biotropica. 23:462–467.

Wright, A.H. 1932. Life Histories of the Frogsof the Okefinokee Swamp, Georgia. NorthAmerican Salientia (Anura) Number 2.Macmillan Press, New York, New York, USA.

Herpetological Conservation and Biology

740

WALTER H. SMITH earned his Ph.D. at the University ofAlabama in 2011, during which time he performedresearch on the impacts of ecological disturbance onherpetofauna within the Longleaf Pine ecosystem of theGulf Coastal Plain. He is now an Assistant Professor ofBiology at the University of Virginia’s College at Wise,where his research interests center around the broaderimplications of habitat disturbance for both wildlife andhuman populations within the southern AppalachianMountains. (Photographed by Katie Dunn).