HURRICANE IMPACTS ON KEY DEER IN THE FLORIDA KEYS

ROEL R. LOPEZ,1 Department of Wildlife end FISheries Sciences. Texas A&M University, College Station. TX 77843, USANOVA J. SILVY. Department of Wildlife and Fisheries Sciences. Texas A&M University, College Station. TX 77843. USARONALD F. LABISKY. Department of Wildlife Ecology and Conservation. University of Florida. Gainesville. FL 32611. USAPHILIP A. FRANK, U.S. Fish and Wildlife Service. National Key Deer Refuge. Big Pine Key. FL 33043. USA

A.bsb'OCt: The landing in the Florida Keys of Hurricanes Georges (Category 2) in 1998 and Irene (Category 1) in1999, in combination with an ongoing radiotelemetry stUdy of Florida Key deer (OdocoiJeus vi1ginanus dlJvium),offered a unique opportUnity to ~uate the impactS of natUral disturbances on Key deer. We relocated 53 deer(female, n = 29; male, n = 24) during Hurricane Georges and 45 deer (female, n = 27; male, n = 18) during Hurri-cane Irene. One adult male drowned due to Hurricane Georges «2% of radiomarked deer); no deaths wereattributed to Hurricane Irene. A comparison of productivity estimates betWeen years found a significant (P< 0.001)increase in fawn:doe estimates for post-hurricane years (1999-2000) as compared to pre-hurricane years(1995-1998). The mean fawn:doe ratio observed during 1995-1998 was 0.31. The mean fawn:doe ratio observedduring 1999-2000 was 0.64. We found no significant difference in mean daily distances moved by deer betWeen hur-ricane and non-hurricane years. However, we observed significantly larger ranges (95% probability area) and coreareas (500/0 probability area) for both males and females following Hurricane Georges. Fifteen water holes weremonitored monthly following Hurricane Georges, and due to the storm surge, 27% (4/15) of these water holeswere found to be unsuitable for deer use (salinity> 15 ppt). In some cases, water-hole suitability did not improveuntil several weeks or months later. Our stUdy suggestS that mild to moderate hurricanes (Category 1-2) have lit-tle direct impact on the survival of Key deer; however, stronger storms (>Category 3) might have a greater impactdue to stronger winds and greater storm surges (>3.5 m).

JOURNAL OF WlLDUFE MANAGEMENT 67(2):280-288

Key words: Florida Key deer, home ranges, hurricanes, mortality, movements, namral disturbances, Odocoileus vir-ginanus clavium, white-tailed deer.

The endangered Florida Key deer, the smallestsubspecies of white-tailed deer in the UnitedStates. are endemic to the Florida Keys on thesouthern end of peninsular Florida, USA (Fig. 1;Hardin et al. 1984). Key deer occupy 20-25 islandswithin the boundaries of the National Key DeerRefuge (NKDR), with approximately 75% of theestimated deer population on Big Pine Key(BPK) and No Name Key (NNK; Lopez 2001).Reports indicate that hurricanes or tropicalstorms might have a negative impact on the Keydeer population (Klimstra et al. 1974, Silvy 1975,Seal and Lacy 1990), which currently numbersabout 482 on BPK (406) and NNK (76; Lopez2001). Hurricanes have been a major natural dis-turbance affecting coastal areas in the UnitedStates, particularly in the Caribbean islands(Boose et al. 1994, Ross et al. 2000). The role ofhurricanes on the flora and fauna of the FloridaKeys is of interest due to the relatively high fre-quency of tropical stonn occurrences and the lowland elevations «3 m), that make these islandssusceptible to stonn surges (Folk 1991, Ross et al.2000). At the landscape level, hurricanes have thepotential to reshape shorelines, cause extensive

damage to vegetation in forested areas, andchange hydrological properties (Boose et al.1994, Ross et al. 2000). Potential impacts of hur-ricanes on deer populations include direct mor-tality and/or a reduction in herd productivity(Folk 1991, Labisky et al. 1999). Additionally,storm impacts also include changes to vegetationcommunities (windthrown trees, broken branches,

It,

Fig. 1. Direction of passage of Hurricanes Georges and Irenein the Florida Keys. USA.1 [-mail: roel@tamu.edu

280

HURRICANES A;~ DEER . Lopa" Gl. 281J. Wild!. Manage. 67(2):2003

mation (Dickson 1955). Typically, island areasnear sea level (maritime zones) are comprised ofred mangrove (Rhisophora mangle), black man-grove (Avicennia gmninans) , white mangrove(Lagunculmia racemosa), and buttonwood (Cono-carpus ~ta) forests. With increasing elevation,tidal areas transition into hardwood (e.g.,Gumbo limbo [Bursera simaruba],jamaican dog-wood [Piscidia piscipulaa]) and pineland (e.g.,slash pine [Pinus eUioUuJ, saw palmetto [Serenoa~]) forests whose vegetation are intolerant ofsalt water (Dickson 1955). From 1960 to 2000,approximately 24% of native areas have beendeveloped on BPK and NNK (Lopez 200 I ).

defoliation) and a decrease in freshwater avail-ability. In the case of the latter, freshwater is a lim-iting factor for Key deer, and a significant stormsurge might limit the amount of freshwater avail-able in the fonD of natural water holes (Folk 1991).

Since 1968, researchers and biologistS havestudied the population ecology of the federallyprotected and endangered Key deer herd. Hurri-canes, though a rather common occurrence inthe Keys, have rarely passed through the narrowrange «30 km) of the Key deer population(Lopez 2001). The last significant stonn thatlanded in the lower Keys was Hurricane Donna in1960 (Category 2, Saffir-Simpson scale; Ne~ann1991). In january 1998, a research project was ini-tiated to evaluate the current status of the Keydeer population. In September 1998, the centerof Hurricane Georges (Category 2) passedthrough the lower Florida Keys within the Keydeer range (Fig. 1; Guiney 1998, Lopez 2001). InOctober 1999, a less substantial stonn, HurricaneIrene (Category 1), also landed directly withinthe Key deer range (Fig. 1; Avila 1999, Lopez2001). The landing of these 2 hurricanes, in con-cert with the ongoing radiotelemetry study,offered a unique opportunity to evaluate the im-pactS of violent natural disturbances or caWU"Gophes to an island deer population. An under-standing of deer mortality and movements andimpacts to natural resources such as vegetationand freshwater would aid in developing strategiesto manage and recover this endangered deerpopulation. Furthermore, such information alsowould be useful in the implementation of a Popu-lation Viability Analysis (PVA) being proposed forthe Key deer (Boyce 1992, Ak~akaya 2000, Lopez2001). Our objectives were to (1) detennine post-hurricane deer mortality immediately followingthe stonn; (2) compare herd productivity in hurri-cane and non-hurricane years; (3) evaluate deermonthly ranges, core areas, and mean dailymovementS during hurricane and non-hurricaneyears; and (4) detennine changes in freshwateravailability due to the associated stonn surge.

METHODSRadiotelemetry

Key deer were radiomarked as part of a researchproject during January 1998-December 1999 onBPK and NNK Key deer were captured andradiomarked using portable drive nets (Silvy et al.1975), drop nets (Lopezetal.I998),andhandcap-ture (Silvy 1975, Lopez 2001). Captured deer wereradiomarked using plastic neck collars (8-cm wide)for females, leather antler collars (0.25<m wide)for yearling and adult males, and elastic expand-able neck collars (3<m wide) for fawns. A battery-powered, mortality-sensitive radiotransmitter(150-152 MHz, 100-110 g for plastic neck collars,10-20 g for antler transmitters and elastic collars;Advanced Telemetry Systems, Isanti, Minnesota,USA) was attached to collar material. In addition,each captured animal received an ear tattoo thatserved as a permanent marker (Silvy 1975).

Key deer were monitored via radiotelemetrybefore and after Hurricanes Georges and Irene.Radiomarked deer were monitored 6-7times/week at random intervals. Each 24-hr peri-od was divided into 6 4-hr segments. During Irandomly selected 4-hr segment, all deer werelocated (Silvy 1975). Deer locations were deter-mined primarily via homing (=98%) due to thehigh number of roads (~155 km roads/2,500 ha;Lopez 2001). In most cases, deer locations wereplaced within a l-ha block of habitat. Deer loca-tions were then recorded on geo-referencedmaps and transferred into a Geographical Infor-mation System (GIS) using ArcView (Version 3.2)and Microsoft Access (Version 97).

STUDY AREAThe Florida Keys are a chain of continuous

small islands that stretch approximately 200 kInsouthwest from peninsular Florida. Big Pine Key(2,548 ha) and NNK (461 ha) are within theboundaries of the NKDR. Monroe County, andsupport approximately 75% of the Key deer pop-ulation (Lopez 2001). Soils vary from marldeposits to bare rock of the oolitic limestone for-

Productivity Estimatesu.s. Fish and Wildlife Service (USFWS) biolo-

gists conducted fall spotlight counts (5-8 surveys

J. Wildl. Manage. 67(2):2003HURRICA.~ A.~ DEER. Lopez; et aL282

15 water holes was measured and recordedmonthly using a Yellow Springs Instrument (YSl)conductivity/salinity meter (Model 33, YSI, Yel-low Springs, Ohio, USA) until salinity returned toor below the historic average or the study ended.Water-hole salinity following Hurricane Irene wasnot measured due to the insignificant storm surge«0.5 m) associated with that storm (Avila 1999).

Water-hole locations (upland vs.lowlands) werecompared to determine water-hole susceptibilityto a storm surge based on elevation. We catego-rized water-hole locations by general vegetationtypes based on elevations: (I) Iowlandf--maritimezones (Sl m above mean sea level) comprised ofmangrove and buttonwood forests, and (2)uplandf--nontidal areas (> I m above mean sealevel) comprised of hammocks, pinelands, andfreshwater wetlands (Lopez 200 I ) .

annually, first week of October, 1995-2000) alonga standard route on BPK (56 km) and NNK (3km) beginning at 2000-2100 hr (Humphrey andBell 1986, Lopez 200 I). These surveys providedthe refuge with an index to population size andalso served as the official survey route for NKDR(Humphrey and Bell 1986, USFWS 1999). Withthe aid of spotlights, 2 observers in a vehicle(average travel speed, 16-24 km/hr) recordedthe number of deer observed along the route inaddition to sex and age (fawn, yearling, adult)estimates (Humphrey and Bell 1986). The start-ing and ending points for the survey route wereidentical each time. A fawn:adult doe ratio wascalculated from survey data. Data were pooledbetween islands due to the small contribution ofobserved animals on NNK «5% of roads sur-veyed were on this island).

Water AvailabilityFreshwater has long been believed to be an

important factor for Key deer survival (Folk 1991).Past studies report that Key deer can drink brack-ish water up to IS ppt (Klimstra et al. 1974, Folk1991). In January 1988-June 1990, Folk (1991)conducted a comprehensive study evaluating thedistribution and quality of freshwater resources(i.e., natural water holes) throughout the Keydeer range. Folk's (1991) data were used to iden-tify suitable freshwater resources for Key deerand served as water-hole salinity benchmarks dueto the absence of hurricanes during this period(1988-1990). The selection of water holes used inour study was based on (I) the availability of his-toric records (Folk 1991) from the NKDR, and(2) water holes that were suitable for Key deer«IS ppt) based on these historic averages. First.we identified water holes with a minimum of 6measurements during the period of interest(Sep-Feb). Because water salinity might be affect-ed by many factors such as precipitation, water-hole characteristics (i.e., size, depth), and season(Folk 1991), we also attempted to identify waterholes with long-term trend data that includednatural fluctuations in water salinity throughoutthe year. From this review, we identified waterholes that were considered suitable to Key deer«IS ppt) during the sampling period. Fifteenwater holes met our criteria, representing approx-imately 5% (15/276) of the total available waterholes (Lopez 2001) and included the followingislands: Big Pine (11 = 13), Middle Torch (11 = I),and Big Torch (11 = I). Following the landing ofHurricane Georges, water-hole salinity for these

Data AnalysisMortalilJ.-We determined percent direct mor-

tality due to each hurricane from radiomarkeddeer by dividing the number of radiomarked deerthat died in each storm (I week post-storm) by thetotal number of radiomarked deer alive prior tothe storm (I week pre-storm). We excluded deerdeaths that occurred during the I-week post-storm period that were non-hurricane related(e.g., deer killed on highway) from the analysis.

Movements.-We calculated Key deer ranges(95% probability area) and core areas (50% prob-ability area) using a fixed-kernel home-range esti-mator (Worton 1989; Seaman et al. 1998, 1999)with the animal movement extension in ArcView(Version 2.2; Hooge and Eichenlaub 1999). Cal-culation of the smoothing parameter (kernelwidth) as described by Silverman (1986) was usedin generating kernel range estimates.

Ranges (ha), core areas (ha), and mean dailymovements (m) were calculated by sex for a 3-week period following Hurricane Georges. Wepooled Key deer movements by age-class due tosmall sample sizes (White and Garrott 1990).Hurricane Georges occurred on 25 September1998; therefore, Key deer ranges, core areas, andmean daily distances moved were calculated from26 September to IS October 1998 and comparedto 26 September-IS October 1999 (non-hurri-cane year). Movement data could not be extend-ed beyond 15 October 1999 because of potentialbias in overlap with Hurricane Irene. Ranges andmovements for Hurricane Irene could not becompared due to the earlier occurrence of Hur-ricane Georges. Differences in Key deer ranges,

HURRICANES A.'"D DEER. Loptz It aL 283J. Wildl. Manage. 67(2):2003

1.00

0,15

t~~J!

0.50

I .t::8;

~185 1- 1997 I'" 1m ~

Y...

Fig. 2. Mean annual fawn:adult doe ratios from U.S. Fish andWiIdUfe Service October fall counts of Florida Key deer,1995-2000 (Florida Keys, USA). Vertical lines represent stan-dard deviation. Comparisons followed by the same letter arenot signifICantly different (P > 0.05).

core areas, and mean daily movements were test-ed using an ANOVA, and Tukey's Honestly Sig-nificantly Different (HSD) for multiple compar-isons was used to separate means when F-valueswere significant (P< 0.05; Ou 1993).

Productivity.-Mean productivity, derived fromfawn:adult doe ratio estimates in October, weredetermined by year 1995-2000. We defined pro-ductivity estimates between 1995-1998 as non-hurricane years and 1999-2000 as hurricaneyears. Impacts due to Hurricane Georges andIrene would not be measurable in the populationuntil 1999-2000. Differences in productivity esti-mates among years were tested using an ANOVA,and Tukey's HSD for multiple comparisons wasused to separate means when F-values were sig-nificant (P< 0.05; Ott 1993).

Water Availability.-Water salinity measure-ments following Hurricane Georges were com-Pared to available historic salinity measurements(1988-1990, non-hurricane years; Folk 1991). Foreach unique water hole, observed average salini-ty post-Hurricane Georges was compared to his-toric averages using a 2-sample ..test (Folk 1991,Ott 1993). We defined water-hole unsuitability asan increase in average salinity above 15 ppt. Wedetermined the percentage of fresh water holesthat were unsuitable for deer (>15 ppt) due tothe storm surge. Furthermore, water-hole loca-tions (upland vs.lowlands) also were evaluated todetermine water-hole susceptibility to a stormsurge based on elevation.

sustained 2-min winds of 127 kIn/hr, and a stannsurge of approximately 0.5 m above nonnal wererecorded for Hurricane Irene on BPK. Rainfallrecorded in Key West was 22 cm (Avila 1999).

MortalityWe monitored S3 (female. n = 29; male. n = 24)

radiomarked Key deer during Hurricane Georgesand 4S (female. n = 27; male. n=18) during Hur-ricane Irene. During Hurricane Georges. I adultmale (2% of radiomarked sample) drowned dueto the storm. No radiomarked deer died as aresult of Hurricane Irene.

ProductivityWe conducted 42 surveys during October

1995-2000, within a range of 5-8 surveys/year. Acomparison of October productivity estimatesamong years revealed a significant (P < 0.00 I) in-crease in fawn:adult doe estimates during1999-2000 (Fig. 2). The mean fawn:doe ratioobserved for 1995-1998 was 0.31. whereas themean fawn:doe ratio observed in post-hurricaneyears (1999-2000) was 0.64.

MovementsWe used 100 (females, n = 66; males, n = 34) Key

deer to calculate movements (ranges, core areas,daily distances) during Hurricane Georges (Table1). The mean number of locations used to calcu-late ranges, core areas, and mean daily move-ments was 15 (SD = 3, range = 10-21). As expect-ed, ranges, core areas, and daily distances differed(P < 0.05) between sexes for both storm events(Table 1). No significant difference (P> 0.076)

RESULTS

Hurricanes

Hurricane G«nges.-The center of HurricaneGeorges made landfall on 25 September 1998 nearKey West, Florida, with minimum central pressureof 981 mb and maximum sustained 2-min windsof 178 kIn/hr. Strongest recorded winds and stonndamage occurred betWeen Cudjoe Key and BPK(Fig. 1; Guiney 1998). The estimated stonn surgeon BPKwas approximately 1.75 m. Rainfall ass0-ciated with Georges was low; Key West recordedonly 21 cm of rainfall (Guiney 1998).

Humcane Jrme.-The center of then TropicalStorm Irene crossed Havana, Cuba on 14 October1999. Irene reached hurricane status over the flori-da Straits, with the center of the stann movingnortheast of Key West, Florida, on 15 October 1999.Hurricane Irene continued in a northeasterly track,crossing the northern half of BPK (Avila 1999). Aminimum central pressure of 986 mb, maximum

J. Wildl. Manage. 67(2):2003HURRICAI ES AND DEER. Loptz" oJ.284

Table 1. Key deer mean daily distances (m), and 95% and 50% ranges (he) br non-hUrrDne (NH) and hurricane (H) years in

the Aorida Keys, USA, 1998-1999.

Tests !-Yearb so Sex Years. In

Humcsne GeorgesDistance 99

277195230

A

B

A

B

21.94

0.000

A

B

A

B

19.77

O.CKXI

A

B

A

B

16.83

O.CKXI

AAAA

2.730.103

BBAA

6.880.010

BBAA

4.53

0.036

NHNHHH

24132521

268431310552

3Q

88

60

184

.9878

150

NHNHHH

24132521

95%ArN

6191424

NHNHHH

24132521

6161029

50% Area

FemaleMaleFemaleMale

F,.2PFemaleMaleFemaleMale

F,.2pFemaleMaleFemaleMale

F,.2P

b Comparisons followed by the same letter are not significantly different (P > 0.05).Hurrk:ane Georges occurred on 25 Sep 1998; therefore. deer movements were calculated 26 Sep-15 Octr 1998 (3 weeks.

hurricane year) and compared to 26 Sep-15 Oct1999 (non-hurricane year).

hurricane: winds, storm surge, and rain (Labiskyet al. 1999). Wmds and storm surges are of par-ticular concern in the Keys due to the low eleva-tion of these islands (highest point < 3 m), thus,storm surges can negatively impact flora and fau-nal communities. To illustrate, many areas occu-pied by Key deer were submerged due to thestorm surge of Hurricane Georges for severalhours (R. R Lopez, personal observation). Manysmaller islands that suppon small deer popula-tions, such as Munson Island or Annette Key,were completely inundated. Previous studieshave noted that Key deer are strong swimmers(Folk 1991), which suggests that Key deer mighthave been forced to tread water until the stormsurge subsided. Few studies on the impacts ofhurricanes on white-tailed deer populationsexist; however, Labisky et al. (1999) reponed1000/0 survival for radiomarked white-tailed deerin the Everglades during the aftermath of Hurri-cane Andrew (Category 4, 242 km/hr wind) in1992. Results from our study suggest that mild tomoderate hurricanes (Category 1-2) have littledirect impact on the Key deer population, but wepropose that stronger storms (>Category 3)would have a greater impact due to strongerwinds and greater storm surge (>3.5 m; Neu-mann 1991). For example, a moderate Category3 hurricane with a storm surge of 4 m would

was found in comparing mean daily distancesbefore and after for both storm events. However,significantly (P = 0.036-0.010) larger ranges andcore areas were observed for both males andfemales following Hurricane Georges (Table 1).

Water AvailabilityIn reviewing historic water-hole salinity data, all

15 water holes, representing approximately 5%(15/276) of the total available water holes (Lopez2001), were suitable «15 ppt) for Key deer prior toHurricane Georges (Table 2). Following HurricaneGeorges, water salinity measurements for 27%(4/15) of water holes were unsuitable for deer use(Table 2). Many water holes were not only unsuit-able immediately following the stonn but alsoremained so several weeks or months later (Fig. 3).Fifty percent (3/6) of monitored water holes foundwithin lowland areas were impacted by the stormsurge, whereas 11% (1/9) of upland water holeswere impacted. All water holes affected by the stormsurge eventUally returned to their historic average.

DISCUSSION

MortalityResults from the study revealed that Key deer

mortality due to hurricanes was low «2%). Threemajor forces contribute to damage caused by a

HURRlCA1'.'ES k'\'D DEER . Lopez. ttal. 285Wildl. Manage. 67(2):2003

Table 2. Water salinity (ppt) for 15 selected water holes within the Key deer range pre- and post-Hurricane Georves. FlorIda Keys.

USA, 1988-1990 and 1998.

PQSt.~Pr.~~

0.000"

0.0900.053"0.041"0.210"0.0710.0070.0490.2100.0100.0920.0680.9400.9200.002

24.0011.8819.0019.7015.4011.175.373.633.003.507.575.719.171.837.57

8.91

8.94

7.02

12.3

11.9

8.24

1.80

4.00

2.45

0.54

3.412.e91.470.742.84

. Pre-storm is defined as water salinity data from historical reCOfds (Folk 1991) collected in 1988-1990; post-storm data ~

cotIected bliowing Hurricane Georges W11998 and compared to historic averages.Number of water samples (monthly readings).

C Asterisk represents _ter holes that were suitable pre-storm (avelage salinity < 15 ppt) that became unsuitable (averagesalinity> 15 ppt) for Key deer following the storm. Pair-wise comparisons fA average salinities for water holes 3 and 8 did not d"1f-fer; ~r. some monthly readings during the 8-month post-storm period were unsuitable for Key deer.

Result! from our stUdy differed from Labisky etal. (1999), who reponed a decrease in the pro-ductivity of Everglades white-tailed deer (0. v.seminolus) following Hurricane Andrew. A possi-ble explanation for this difference might be thatvegetation types in the Florida Everglades consistprimarily of wet prairie with limited to no over-story (Miller 1993, Labisky et al. 1999). Converse-ly, dense hammocks and open pineland charac-terize upland vegetation types in the Florida Keys(Folk 1991, Lopez 2001). In the case of hardwoodhammocks, strong winds opened the dense forestcanopy, which in turn resulted in a flush of newunderstory growth (R R Lopez, personal obser-vation). Such results would not be expected inthe Everglades ecosystem due to the lack of a for-est canopy.

result in the complete submersion of BPK andNNK These 2 islands suppon most (approx.75%) of the Key deer population and providemost (approx. 51%) of the freshwater sources(Lopez 2001). Greater deer mortality and impactto deer resources (e.g.. vegetation and freshwa-ter) would be expected with more severe storms.

ProductivityWe found that Key deer productivity nearly

doubled following Hurricane Georges on BPKand NNK. Most of the overstory component wasreduced by approximately 50% due to strongwinds and windthrown trees (R. R. Lopez. per-sonal observation).. This reduction in the over-story may have resulted in both an immediateand long-term effect in the amount of food avail-able to Key deer. The reduction in the overstorymay have caused (1) a shon-term increase in theamount of forage available from windthrowntrees and/or broken branches, and (2) a long-term increase in re-growth or sprouting that pro-vided additional forage for Key deer. Other stud-ies have reponed an increase in understoryvegetation following tropical storms or hurri-canes (Ross et al. 1997, 1998; Wallace 1998). Per-haps the increase in herd productivity was due toan increase in the overall fitness of female Keydeer resulting from higher food availability priorto the breeding season (Verme and Ullrey 1984).

MovementsDeer movements (ranges, core areas, daily dis-

tances) were compared by year and sex to deter-mine effects of hurricanes on the Key deer pop-ulation. As expected, differences in ranges, coreareas, and daily distances were found betWeensexes. This difference is well documented(Marchinton and Hirth 1984, Beier and McCul-lough 1990, Demarais et aI. 2000) and attributedto behavioral differences betWeen sexes. In gen-eral, males have greater ranges and move greaterdistances. We found no significant difference in

J. Wildl. Manage. 67(2):2003HURRICANES AND DEER . LDpDo" at286

Uplands

'0~-OJ=Co

e-

2 ] . ,MODW folio";.."""

'C~

u.Coe

--0

z

t......~

I 4 S

Mums roa-iq I8Or8

6

Fig. 3. Salinity measurements for water holes found in the lowlands (figures on left) and uplands (figures on right) that _reimpacted (top figures) and not impacted (bottom figures) by the storm surge during Hurricane Georges, Big Pine Key, USA, 25September 1998. Horizontal lines represent maximum average salinity (long dashed line) and standaro deviation (short dashedline) for water holes in a given figure from historic data (1988-1990). Water-hoie suItabIity for Key deer use is 15 ppt (Folk 1991).

comparing mean daily distances among years forboth storm events; however, significantly largerranges and core areas were observed for bothmales and females following Hurricane Georges.Two reasons might explain the increase inranges for both sexes. First, heavy machineryUsed to dean up following the aftermath of Hur-ricane Georges might have disrupted or dis-turbed normal movement patterns. Second,freshwater availability was reduced in lower ele-vations, forcing some deer to move greater dis-tances to obtain needed freshwater. For exam-ple, freshwater in the southernmost area of BPKwas scarce and likely caused Key deer to movegreater distances than under non-storm circum-stances (R. R. Lopez, personal observation).However, these results should be viewed withcaution. Seaman et al. (1999) suggested that aminimum of 30 locations (>50 locations pre-ferred) were needed for unbiased kernel esti-mates. Despite the potential sample size bias, wepropose that this bias would be found in bothhurricane and non-hurricane years, and differ-ences in deer movements are relative.

Water AvailabilityFollowing the storm, water salinity measure-

ments for 27% (4/15) of monitored water holeswere found to be unsuitable for Key deer (>15ppt). Interestingly, some water holes were suit-able for Key deer immediately following thestorm, becoming unsuitable several weeks ormonths later. SaltWater has a greater density thanfreshwater causing the latter to separate andform an upper lens (Folk 1991). We hypothesizedthat. following Hurricane Georges, rainfall quick-ly recharged the upper portion of water holeswith freshwater. As the dry season approachedseveral months later (Nov-Feb), however, evapo-ration may have increased water salinity. In somecases, water holes became hypersaline (>30 ppt.salinity of ocean water) beyond normal water-hole fluctUations with the added saltWater fromthe storm surge (Folk 1991). These results sug-gest that freshwater might not be limiting to Keydeer until several weeks or months following ahurricane event.

Fifty percent (3/6) of monitored water holesfound in lowlands were impacted due to the

HURRICANES A."'D DEE;R . Lf!PD." aL 287Wildl. Manage. 67(2):2003

storm surge, which suggests that islands with lowelevation, such as Cudjoe or Sugarloaf, might notsustain deer populations following a hurricanedue to the limited availability of freshwater. H thelatter supposition proves valid, it strengthens thepoint that larger islands such as Big Pine Key arecritical to the Key deer due to the high numberof fresh water holes in upland areas (Lopez200 I). Lopez (200 I) reponed that approximately64% (91/142) of all permanent water holes onBig Pine, Big Torch, and Middle Torch keys werefound in upland areas.

MANAGEMENT IMPLICATIONSKey deer have survived in this island environ-

ment subject to frequent storm-related catastro-phes for over 10,000 years (Hardin et al. 1984).Despite the small negative impacts to Key deerobserved in our study, more severe stOrms (>Cat-egory 3) could be expected to have a greaterimpaCt to the deer population, as was reported byLabisky et al. (1999) following Hurricane Andrewin the Everglades in 1992. In our study, hurri-canes may have actually benefited the deer popu-lation by rejuvenating plant growth.

ACKNOWLEDGMENTSThanks to Texas A&M University student

interns who assisted in the collection of fielddata. Furthermore, we thank R Hudson, C. Faul-haber, N. Wilkins, W. Grant, T. Peterson, and 2anonymous reviewers for constrUctive criticism inthe preparation of this manuscript. Funding wasprovided by Texas A&M University System, Roband Bessie Welder Wildlife Foundation, FloridaFish and Wildlife Conservation Commission, andUSFWS (Special Use Permit No. 97-14). Specialthanks are extended to the staff of the NationalKey Deer Refuge, Monroe County, Florida. Thismanuscript is supported by the Welder WlldlifeFoundation, Contribution No. 582.

~, M. S. 1992. Population viability analysis. AnnualReviews Ecology and Systematics 23:481-506.

DE.\tARAIs, S., K. V. MILLER. A.'"D H. A. JACOBSON. 2000.White-tailed deer. Pages 6Ol~28 in S. Demarais andP. R. Krausman, editors. Ecology and management oflarge mammals in !'.'orth America. Prentice-Hall,Upper Saddle River, Kew Jersey, USA.

DICKSON,J. D., 111.1955. An ecological study of die Keydeer. Florida Game and Fresh Water Fish Commis-sion Technical Bulletin 3, Tallahassee, Florida, USA.

FoLK, M. L 1991. Habitat of die Key deer. Dissertation,Southern Illinois University, Carbondale, Illinois, USA.

GUIr."EY, J. L 1998. Preliminary repon of HurricaneGeorges. National Oceanic and Atmospheric Admin-istration Technical Repon. National Hurricane Cen-

ter, Miami, Florida, USA.Hardin,J. W., W. D. Klimstra, and N.J. Siny. 1984. Flori-

da Keys. Pages 381-390 in L K. Halls, editor.White-tailed deer: ecology and management. Stack-pole Books, Harrisburg, Pennsylvania, USA.

HOOGE, B. N., A.'iD B. EICHENLAUB. 1999. Animal move-ment extension to ArcView, version 1.1. Alaska Bio-

logical Center, U.S. Geological Survey, Anchorage,Alaska, USA.

HUMPHREY, S. R., A.'"D B. BELL. 1986. The Key deer popu-lation is declining. Wlldlife Society Bulletin 14:261-265.

KuMSTRA, W. D.,J. W. HARDI~, N.J. SILVY, B. N.JACOB-SO~, A.'"D V. A. TERPE~I~G. 1974. Key deer investiga-tions [mal report: December 1967-June 1973. U.s.Fish and Wildlife Serllice, Big Pine Key, Florida, USA.

LABISKY, R. F., K. E. MILUR, A."D c. S. I-l~. 1999.Effect of Hurricane Andrew on survi\-a1 and move-ments of white-tailed deer in die Everglades. JournalofWlldlife Management 63:872-879.

LoPEZ, R. R. 2001. Population ecology of Florida Keydeer. Dissertation, Texas A&:M University, College Sta-

tion, Texas, USA.-, N. J. Silvy, J. D. Sebesta, S. D. Higgs, and M.

Salazar. 1998. A portable drop net for captUringurban deer. Proceedings of die Soudieastern Ass0cia-tion of Fish and Wildlife Agencies 52:206-209.

Marchinton, R. L, and D. H. Hirth. 1984. Behavior.Pages 129-168 in L K. Halls, editor. White-taileddeer: ecology and management. Stackpole Books,Harrisburg, Pennsylvania, USA.

Miller, K. E. 1993. Habitat use ofwhite-tailed deer in dieEverglades: tree islands in a seasonally flooded land-scape. Thesis, University of Florida, Gainesville, Flori-da,USA.

Neumann, C.]. 1991. The national hurricane centerrisk analysis program (HURISK). National Oceanicand Atmospheric Administration Technical Memo-randum, NWS NHC 38, Coral Gables, Florida, USA.

Ott, R. L 1993. An introduction to statistical mediodsand data analysis. Fourth edition.Duxbury Press, Belmont, California, USA.

Ross, M. S., M. Carrington, L.J. Flynn, and P. L Ruiz.2000. Forest succession in tropical hardwood ham-mocks of die Florida Keys: effects of direct mort3licyfrom Hurricane Andrew. Everglades National ParkRepon. Soudieast En\ironmental Research Program,Florida International University, University Park,Miami, Florida, USA.

-,J. F. MEEDER,]. P. SAM, A. HER.'\:oo:\, P. L RUll,A..."D G. TELES:\ICKI. 1997. Winddirow in soudi Floridapine rockIands: pit-and-mound features and plant

LITERATURE CITEDAKl;:AKAVA, H. R. 2000. Population viability analyses with

demographically and spatially sU"Uctured models.Ecological Bulletin 48:23-38.

AVT1.A, L A 1999. Preliminary repon of HurricaneIrene. National Oceanic and Atmospheric Adminis-tration Technical Report, National Hurricane Center,Miami, Florida, USA.

BEIER, P., A...D D. R. McCl:LLOI:CH. 1990. Factors influ-encing white-r.ailed deer activity patterns and habitatuse. Wildlife Monographs 109:1-51.

BoosE, E. R., D. R. FOSTER, A...D M. FLL"F.T. 1994. Hurri-cane impacts to tropical and temperate landscapes.Ecoloy;ical Monographs 64:369-400.

J. Wildl. Manage. 67(2):2003HURRICA.'JES Ai'1D DEER. Lopt:; It al.288

habitat utilization of Key deer, 0d0c0iJeus vi1ginianusclavium. Dissertation, Southern Illinois University,Carbondale, Illinois, USA.

-,J. W. HARDIN, A."D W. D. Ku~STRA. 1975. Use ofa portable net to capture free-ranging deer. WlldlifeSociety Bulletin 3:27-29.

U.S. FISH A.'lD WILDUFE SERVICE (USFWS). 1999. SouthFlorida muiti-species recovery plan: Key deer recov-ery plan. USFWS, Adanta, Georgia, USA.

VER.\tE, L. J., A.,,'D D. E. ULLREY. 1984. Physiology andnutrition. Pages 91-118 in L. K. Halls, editor.White-tailed deer: ecology and management. Stack-pole Boob, Harrisburg, Pennsylvania, USA.

WA!.LACE, G. E. 1998. Demography of Cuban bird com-munities in the non breeding season: effects of foresttype, resources, and hunicane. Dissertation, Univer-sity of Missouri, Columbia, Missouri, USA.

WHITE, G. C., A."D R A. G.URon. 1990. Analysis ofwildlife radio-tracking data. Academic Press, SanDiego, California, USA.

WORTO~, B.J. 1989. Kernel methods for estimating theutilization disuibution in home-range studies. Ecolo-

gy70:164-168.

R«eived 2 Ffbnjary 2002-Acceptld 21 January 2003.Associate Editor: Hudson.

microhabitat associations following HurricaneAndrew. Everglades National Park Report, SoutheastEnvironmental Research Program, Florida Interna-tional University. University Park, Miami, Florida, USA.

--, P. L RUll, G. TELES~ICJ(J, A...DJ. F. MEEDER. 1998.Aboveground biomass and production in mangrovecommunities of Biscayne National Park, Florida(USA), following Hurricane Andrew. South FloridaWater Management District Report, Southeast Envi-ronmental Research Program, Florida InternationalUniversity, University Park, Miami, Florida, USA.

SEAL, U. S., A."D R. C. LAcy. 1990. Florida Key deer pop-ulation viability assessment. Captive Breeding Spe-cialist Group, International Union for Conservationof Nature and Natural Resources, Apple Valley, Min-nesota, USA.

SEAMA..., D. E., B. GRJrnTH, A."D R. A. PO\\"E1.L. 1998.KERNElliR: a program for estimating animal homeranges. Wildlife Society Bulletin 26:95-100.

-,J.J. MILLSPAL"GH, B.J. KEL--OHA..., G. C. BRUNDI-GE, K. J. RAED£K£, A.--D R. A. GITZE~. 1999. EffectS ofsample size on kernel home range estimates. Journalof Wildlife Management 63:739-747.

SILVERMA..., B. W. 1986. Density estimation for statisticsand data analysis. Chapman &: Hall, London, United

Kingdom.SILVY, N. J. 1975. Population density, movementS, and