Abundance and Density of Florida Black Bears in Okefenokee National Wildlife Refuge and Osceola National Forest
University of Tennessee, Knoxville TRACE: Tennessee Research and Creative Exchange
Masters Theses Graduate School
5-2002
Abundance and Density of Florida Black Bears in Okefenokee National Wildlife Refuge and Osceola National Forest
Steven T. Dobey University of Tennessee, Knoxville
Follow this and additional works at: https://trace.tennessee.edu/utk_gradthes
Part of the Animal Sciences Commons
Recommended Citation Dobey, Steven T., "Abundance and Density of Florida Black Bears in Okefenokee National Wildlife Refuge and Osceola National Forest. " Master's Thesis, University of Tennessee, 2002. https://trace.tennessee.edu/utk_gradthes/4533
This Thesis is brought to you for free and open access by the Graduate School at TRACE: Tennessee Research and Creative Exchange. It has been accepted for inclusion in Masters Theses by an authorized administrator of TRACE: Tennessee Research and Creative Exchange. For more information, please contact [email protected]. To the Graduate Council:
I am submitting herewith a thesis written by Steven T. Dobey entitled "Abundance and Density of Florida Black Bears in Okefenokee National Wildlife Refuge and Osceola National Forest." I have examined the final electronic copy of this thesis for form and content and recommend that it be accepted in partial fulfillment of the equirr ements for the degree of Master of Science, with a major in Wildlife and Fisheries Science.
Michael R. Pelton, Major Professor
We have read this thesis and recommend its acceptance:
Gary McCracken, Lisa Muller, Joseph Clark
Accepted for the Council: Carolyn R. Hodges
Vice Provost and Dean of the Graduate School
(Original signatures are on file with official studentecor r ds.) To the Graduate Council:
I am submitting herewith a thesis written by Steven T. Dobey entitled "Abundance and Density of Florida Black Bears in Okefenokee National Wildlife Refuge and Osceola National Forest". I have examined the finalpaper copy of this thesis for form, content, and quality and recommend that it be accepted in partial fulfillment of the requirements forthe degree of Master of Science, with a major in Wildlife and Fisheries Science.
Michael R. Pelton, Major Professor
We have read this thesis and recommend its acceptance:
Accepted forthe Council:
Vice Provost and Dean of Studies
ABUNDANCE AND DENSITY OF FLORIDA BLACK BEARS IN THE
OKEFENOKEE NATIONAL WILDLIFEREFUGE AND OSCEOLA
NATIONAL FOREST
A Thesis
Presentedfor the
Master of Science Degree
The Universityof Tennessee, Knoxville
Steven Thomas Dobey
May 2002 U.T. ARCHIVES �is ACKNOWLEDGEMENTS 2�2-, ,J)i,8
I amgrateful to Dr. Michael Pelton, my major advisor, forgiving me the opportunity to pursue a graduate degreeunder his direction. My fieldseasons in the
OkefenokeeSwamp provided me withsome of the most difficult, enduring, and most enjoyable experiences of my life. I also extend my sincere thanksto Dr. Joseph Clarkfor stepping in as fny second advisor followingDr. Pelton's semi-retirement. Your enthusiasm andwillingness to help has provided with somethifl:gthat I could never have learnedin a classroom.
I greatlyappreciate all of the assistance and advice provided by themembers of my graduate committee. They included Dr. Michael Pelton, Dr. Joseph Clark, Dr. Lisa
Muller, andDr. GaryMcCracken. Their input andsuggestions greatly improved the qualityof my thesis. I also thankDr. Frankvan Manen foralways talcing time to answer my questions andproviding invaluable computer assistance.
The staffat OkefenokeeNational WildlifeRefuge provided researchpersonnel withall of theresources that we could have ever needed throughoutthe course of this 5- yearstudy. This research wouldnot have been possible without your generosity. I also would like to thankthe Georgia Departmentof Natural Resources for all of the time and assistance thatwas so eagerly donated on their behalf.
I owe a special acknowledgementto thestaff of Osceola National Forest,
JeffersonSmurfit and Container Corporation, Gillman Paper Company, Champion
InternationalCorporation, Rayonier, and Superior Pine forallowing me to conduct researchon their lands. I amgrateful to the members of the Jamestown, Okefenokee
ii Sportsman,Big Swamp,Low Road, BearBay, andBanker's Trust hunting clubs for their cooperation.
I also thankthe employees of Stephen Foster State Parkfor their hospitality and theinvitation forThanksgiving dinner when I couldn't go home. I extend a special thanksto Refugeemployee's Stacey Williamsand Stiner Jones fornever being too busy to come by the house andsee how things were going.
I express much appreciation to Dr. Tim King and ShannonJulian for providing invaluable assistancewith hair preparations,DNA extractions, andanalysis of genetic data. Althoughyour patience andwillingness to teach made my 2-week visits to the lab knowledgeable,your hospitality made them most enjoyable.
I greatlyappreciate the help ofmy fieldtechnicians Jeremy Dixon andAmanda Jackson for theirassistance with data collection. Both of you worked hard under extremely difficultconditions, andfor that I amvery grateful.
I amespecially indebted to Rich Beausoleil forintroducing me to bearresearch andgraduate school at _UT. I won't forgetthe lunch breaksspent in the Tiger Truck Stop andall themoney we lost. I never suspected thattrapping bearsin Louisianafor $3 50 a monthwould lead to this. Thanks forall of your help along the way.
To all of my fellow graduatestudents I would like to say thank you formaking my education at UT more rewardingthan I could've ever imagined. These include
JenniferMurrow, Donny Martorello, Rich Beausoleil, Jay Clark,Thomas Eason, Daron
Barnes,David Brandenburg,Mark Boersen, BrandonWear, Meghan Carr, Andy
Edwards,Jun Yoshizaki,Jason Kindall, Dave andRebecca Telesco, Laura Thompson,
iii DanielMoss, andCarrie Schumacher. Thank you all for your support, advice, and friendship.
I would like to especially thankJennifer Murrow for her unwavering friendship andher ability to always say the rightthing when I needed it the most. Throughthe good times andthe not-so-good times, you always had a shoulder forme to leanon. Thank you foreverything, I amproud of you.
Lastly,I must thankall of my familyfor their encouragement andsupport throughoutmy stay here at UT. AlthoughI loved working in Okefenokee,you always made coming home worthwhile andI wouldn't trade that for anything. Thankyou.
iv ABSTRACT
The Florida black bear(Ursus americanusjloridanus) exists as 7 relatively disjunct populations in Alabama,Florida, Georgia,and possibly Mississippi. In197 4, the
Florida Fish and WildlifeConservation Commission listed the Florida black bearas threatened statewide because of habitat loss andillegal killing. Althoughthe species has not been affordedfederal protection, the U.S. Fish and WildlifeService is currently involved in a lawsuit over this issue. Althougha judge's decision is still pending in the case, the earlierruling by the USFWS could be reversed andthe Florida black bearwould be grantedfe deral protectionas a ''threatened" species.
I investigated population size anddensity of Florida black bearsin the
Okefenokee-Osceolaecosystem in southeast Georgia andnorthcentral Florida. I sampled bearsat the OkefenokeeNational WildlifeRefuge (ONWR) andOsceola National Forest
(ONF). This study provided a rareopportunity to compareestimates between a hunted
(ONWR) andunhunted (ONF) assemblage of bearswithin the samepopulation.
In addition to livetrapping,I also sampledeach study area using a non-obtrusive samplingtechnique of collecting hair samples fromfree -rangingbears using baited barbed-wireenclosures. Individualidentification was facilitatedby microsatellite analysisof DNA extractedfrom collected hairsamples.
From 1995-1998, I, along with other project personnel, live captured 123 individual bears208 times on the Okefenokee area and 79 bears13 2 times on the Osceola area. Duringthe 15 weeks of hairsampling, 43 5 and742 bearvisits resulted in the collection of374 and 637 hairsamples on the Okefenokee andOsceola study areas,
V respectively. A subsampleof 79 hair samples w�s randomlyselected foranalysis from the Okefenokeedata. Complete multi-locus genotypes were obtained for78 of those
samples,of which 39 individual bearswere identified. Eighty-eight hairsamples were
chosen foranalysis from the Osceola data; complete multi-locus genotypes from 37 bears
were obtained from84 samples. All sampleswere analyzed at the same8 microsatellite
loci.
The heterogeneity model Mh produced anestimate of71 (95% CI= 59-91) bears
2 forthe Okefenokeestudy area,corresponding to a density of0.14 bears/km • On the
Osceola study area, the null model Mo estimated population size at 44 (95% CI= 40-57)
2 bears,or 0.12 bears/km •
The hair-samplingtechnique is a promising new tool formark-recapture
experiments andbear research. Because largeareas can be sampledat one time, spatial
andtemporal variationin capture probabilities can be overcome. Likewise, trapresponse
biasis likely minimized because the "capture"involves no physical restraint or undo
stress. Based on the above, largesample sizes can be collected in a relatively short
period oftime. Thereby facilitatingthe use of closed models forestimating population
size.
vi TABLE OF CONTENTS
CHAPTER PAGE
I. INTRODUCTION ...... 1 General Problem Statement ...... 1 Justification...... ; ...... 4 Objectives ...... -...... 5
II. STUDY AREAS...... 7 General ...... 7 Location ...... 7 Topography and Geology ...... 10 Hydrology ...... 13 Climate ...... 14 Fauna ...... 14 Flora ...... 15 History and Land-Use ...... 15 ill. MATERIALS ANDMETHODS ...... 20 Study Design ...... 20 Study Area Delineation ...... 22 Trapping and Handling ...... 23 Hair Trapping ...... 27 Subsampling ...... 30 Microsatellite Analysis ...... 31 Probability of Identity············�························································· 31 Hardy Weinberg and Linkage Disequilibrium Tests...... 35 Estimation ofPopulation Size ...... 36 General ...... 36 Closed Models ...... 42 Open Models ...... 47 Population Density ...... _. � ...... 48
IV. RESULTS ...... 49 · Trapping ...... 49 Hair Trapping ...... 49 Micro satellite Analysis ...... 54 Probability ofldentity ...... 57 Hardy Weinberg and Linkage Disequilibrium ...... 61 Population Size and Density ...... ; ...... 61 Closed Models ...... 61 Open Models ...... 69 V. DISCUSSION...... 74
vii CHAPTER PAGE
Population Size ...... 74 Population Density ...... 81 Genetic Considerations ...... 84
VI. MANAGEMENTIMPLICA TIONS ...... 87 General ...... 87 Genetic Sampling as a Mark-recaptureMethod ...... 90
LITERA.TURE CITED ...... 95
APPENDICES...... -...... 106 Appendix A. Black bear captures on the Georgia study . area, 1995-1998 ...... ; ...... 107 Appendix B. B�ack bear captures on the Florida study area, 1996-1998 ...... 116 Appendix C. Laboratoryprotocol for microsatellite analysis ...... 122
VITA ...... 126
viii LIST OF TABLES
TABLE PAGE
1. Proportions of habitat typeswithin the Okefenokee and Osceola study areas, Georgiaand Florida, 1999 (Scheick 1999)...... 16
2. Closed multiple mark-recapturemodels used to estimate within year population size from1999 hair capture data (Otis et al. 1978 andChao 1987, 1988, 1989) ...... 39
3. Trapping summaries forthe Okefenokee(1995-1998) and Osceola (1996-1998} study areas, Georgiaand Florida...... 51
4. Summaries of black bearhair trapping on the Okefenokeeand Osceola study areas, Georgia andFlorida, 1999...... - 55
5. Observed allelesand frequenciesfor 3 9 black bears identifiedfrom barbed-wire hair traps forthe Okefenokee study area, Georgia, 1999...... 56
6. Observedalleles and frequenciesfor 3 7 black bears identified from barbed-wire hair traps forthe Osceola study area, Florida,1999 ...... 58
7. Probability of identity estimates for39 individual black bears identifiedfrom barbed-wire hair traps on the Okefenokeestudy area, Georgia, 1999 ...... � ...... 59
8. Probability of identity estimates for3 7 individual black bears identifiedfrom barbed-wire hair traps on the Osceola study area, Florida, 1999...... 60
9. Estimated black bear population size from hair captures using multiple mark-recapture models in the Okefenokeestudy area, Georgia, 1999...... 63
ix TABLE PAGE
10. Estimated black bear population size from hair captures using multiple mark-recapture models in the Osceola study area, Florida, 1999 ...... · 64
11. Observed capture frequenciesof bears identified at hair traps on the Okefenokee study area andthe zero-truncatedPoisson frequencies to be expected if catchabilityis constant (1999)...... 65
12. Observed capture frequenciesof bears identifiedat hair traps on the Osceola study area and the zero-truncatedPoisson frequenciesto be expected if catchability is constant (1999) ...... 68
13. Observed capture frequencies of bears identifiedat 'snares and hair traps on the Okefenokee study area and the zero-truncated Poisson frequenciesto be expected if catchability is constant (1995-1999)...... 70
14. Observed capture frequencies of bears identified at snares and hair traps on the Osceola study area and the zero-truncatedPoisson frequenciesto be expected if catchability is constant (1996-1999)...... 71
15. Estimated black bear population size fromthe Jolly-Seber models in Okefenokee (OKE) and Osceola (OSC) study areas using a combination of live-capture and hair-trappingdata, Georgia and Florida, 1995-1999...... - ...... 72
X LIST OF FIGURES
FIGURE PAGE
1. Current distribution of the American black bear (Ursus americanus) in the southeasternUnited States (fromPelton and van Manen 1997) and (B) current distribution of the 7 relatively disjunct Florida black bear (U. a. floridanus) populations...... 2
2. General area of the OkefenokeeNational Wildlife Refuge, Georgia, and the Osceola National Forest, Florida, 1999 ...... 8
3. Black bear study area, Okefenokee NationalWildlife Refuge, Georgia, 1995-1999...... 9
4. Black bear study area," Osceola National Forest, Florida, 1996-1999 ...... 11
5. Delineation of the Okefenokee and Osceola studyareas using home range data, Georgia and Florida, 1999 ...... 24
6. Diagramof a baited barbed wire enclosure used to collect hair samplesfrom black bears on the Okefenokee andOsceola study areas, Georgiaand Florida, 1999 ...... 28
7. Capture-recapture models used to estimate black bear population size on the Okefenokee and Osceola study areas, Georgiaand Florida, 1995-1999 ...... 38
8. Pooling configurationsof the hair trapping sessions considered for multiple mark-recapturemodels to estimate population size on the Okefenokeestudy area, Georgia,1999 ...... 40
9. Pooling configurationsof the hair trapping sessions considered for multiple mark-recapture models to estimate populationsize on the Osceola study area, Florida, 1999 ...... : ...... 41
xi FIGURE PAGE
10. Locations of snaretrapsites on the Okefenokee study area, Georgia, 1995-1998 ...... 50
11. Locations of snare trapsites on the Osceola study area, Florida, 1996-1998 ········································································································· 52
12. Locations of barbed-wire hair traps on the Okefenokee and Osceola study areas, Georgiaand Florida, 1999 ...... 53
xii CHAPTERI
INTRODUCTION
General Problem Statement
The Americanblack bear ( Ursus americanus) once occupied most of the forested regionsof North America. However, habitat loss and fragmentationby humans has significantlydecreased that range(Pelton andvan Manen 1994). In the southeastern
United States, black bearscurrently exist in the InteriorHighlands, the Appalachians, and theSoutheastern Coastal Plain (Fig. 1). Within the Southeastern Coastal Plain, 3 subspecies of black bears exist: the eastern black bear(U. a. americanus), the Louisiana black bear(U. a. luteolus), andthe Florida black bear(U. a.floridanus).
In addition to Florida, the Florida black bearoccurred in the coastal plain of
Georgia, Alabama, andMississippi (Hall 1981). Since the late 1800s, however, land clearingfor agriculture and urbanization significantly decreased availablehabitat in the southeasternU.S. More importantly,loss of those native forestshas resulted in severe forest fragmentation and bearpopulations that aregeographically isolated (Wooding and
Hardisky1992). Although black bearsonce occupied the entire Florida mainlandand most coastal islands(Brady andMaehr 1985), their historic rangehas been reduced by nearly83 % (Florida Fish andWildlife Conservation Commission [FFWCC] 1993). The current Floridablack bearpopulation exists as 7 relatively disjunct populations in_
Alabama,Florida, Georgia, andpossibly Mississippi (Fig. 1 ). The largestof these bear populations arefound in Apalachicola NationalForest (NF), Ocala NF,Big Cypress
National Preserve, andOkefenokee NWR-Osceola NF.
1 B.
Ocala
Mobile Bay, AL Apalachicola National Forest, FL
Chassahowitzka National Wildlife Refuge, FL
Big Cypress National 0 50 Kilometer Preserve, FL
Fig. 1. Current distribution of the American black bear(Ursus americanus) in the southeasternUnited States (fromPelton and vanManen 1997) and(B ) current distribution of the7 relatively disjunct Florida black bear(U. a.jloridanus)populations.
2 In 1974, the FFWCC listed the Florida black bear as threatened statewide because of habitat destruction andillegal killing. Black bearhunting seasons were closed except in Baker andColumbia counties, Apalachicola NFand Osceola NF, andTyndall Air
Force Base, where regulated harvestswere allowed to continue. In 1990, the U.S. Fish andWildlife Service (USFWS) was petitioned to list the Florida black bear as a federally threatened species under the EndangeredSpecies Act of 1973. The petition cited illegal hunting, loss andfragmentation of habitat, huntingpressure, androad mortality as the primaryjustifications for federal protection. In 1992, theUSFWS concluded that the status of the Florida black bearwas ''warrantedbut precluded" fromofficial designation as a protected species by higherpriority listing actions (Wooding 1992). Consequently,
ONF was closed to bearhunting in 1992, and all black bearhunting seasons in Florida were terminatedin 1994.
A subsequent reexaminationby the USFWS to federallylist this subspecies was mandatedin 1998 andit was ruled that, based on currentbiological data, theFlorida black beardid not warrantfederal protection. The USFWS reported that thelargest of the remaining Florida black bearpopulations (Apalachicola NF, Ocala NF, Big Cypress
National Preserve,and Okefenokee NWR-Osceola NF)were viable andthat habitat loss andfragmentation do not threatentheir persistence because theyare secure on public conservation lands(Bentzien 1998). It was concluded that, because thosepopulations are distributedover most of thehistorical range of the species, theFlorida black bearis not endangeredor likely to become so in the foreseeablefuture (Bentzien 1998). The 1998 ruling by theUSFWS, however, drew criticism fromsome conservationgroups and a lawsuit has since been filedin anattempt to overturnthe settlement. Althougha judge's
3 decision is still pending on the case, the ruling could be reversed andthe Florida black bearwould be granted federal protection under the ESA. The subspecies, however, is still classifiedas a threatened species by the State of Florida.
Justification
Given the geographicisolation of populations of Florida black bearsand the degreeof public concernover theirstatus, it is imperative thatmanagement decisions be based on the most recent andpertinent biologicaldata. Most research on Florida black bears, however, waspublished in the 1980s (Maehrand Brady 1982, 1984). Even during the 1990s, researchwas concentrated only on the Apalachicola NF (Seibert 1993), Eglin
AFB (Stratman 1998), andONF (Mykytka andPelton 1990, Wooding andHardinsky
1992, Wooding andHardinsky 1994, Scheick 1999) bearpopulations. Although those studies yieldedvaluable informationon beardenning ecology, food habits, habitat use, mortality, movements, andthe effectsof prescribed fireon bears, none provided estimates of population size.
Considering the degreeand rate of human development surroundingFlorida black bear populations, obtaining currentpopulation estimates is paramount for developing sound management plans, especially where bears arehunted. However, public interest and managementconcerns vary greatlythroughout the rangeof the Florida black bear.
Whereas preservationists strive forlegislation to protect the bearand its habitat, hunters believe the goal of management should be to restore the bear's status as a gamespecies.
Additionally, people suffering losses in revenue from agriculture or apiary damage typicallyview bears as a nuisanceand desire more stringent depredation guidelines.
4 Regardlessof managementobjectives, biologistscannot accurately predict the effectivenessof managementactions without reliable informationon population demographicsand abundance.
In 1994 theUSFWS contractedthis study to aid in the development and implementation of managementguidelines forblack bearsinhabiting the Okefenokee
Osceola ecosystem. Encompassing approximately 6,147 km2 (1.5 million acres), this areais one of the largestcontiguous habitats occupied by black bears in the Southeast.
Although habitatswithin the ecosystem arebiologically similar, managementobjectives aredifferent withinthe rangeof bears in this population. Whereas bears areprotected in
Florida, thereis a year-roundchase season anda limited bear hunting season forcounties surrounding portions of OkefenokeeNWR (ONWR) in Georgia. Therefore, one of the primaryobjectives of this 5-yearstudy was to document andcompare population dynamics andhabitat use of bears within the Okefenokee-Osceolaecosystem. We accomplished that goal by trapping andradio-collaring bears on 2 study areas;one consisting of landin andadjacent to ONWR andthe other in ONF. The primaryfocus of my researchin thisstudy was to analyzecapture data andprovide estimates of abundance foreach of the2 study areas.
Objectives
The Okefenokee-Osceolaecosystem is regardedas one of the last strongholdsof the Florida black bearwithin its range. Althoughthis ecosystem serves as one of the remaining core populations forthis species, humandevelopment continues to isolate thosepopulations. The designationof the Florida black bearas a ''threatened species" in
5 Florida, combined with the continued uncertainty of its federalstatus, identifythe need for reliable estimates of population size forbears in the Okefenokee-Osceola ecosystem.
Therefore, the objective ofmy study was to estimate population size and density of black bearsin the Okefenokeeand Osceola study areas.
6 CHAPTER II
STUDY AREAS
General
I conducted researchon 2 study areas within the Okefenokee-Osceolaecosystem in southeastGeorgia andnorth central Florida (Fig. 2). The center of thisecosystem is located at approximately 30 ° 40' north latitude and 82° 30' west longitude. The northern study area,which I will referto as the Okefenokeestudy area,is located in the northwestern comerof the OkefenokeeNational WildlifeRefuge (ONWR).
Approximately 40 kmto the south,the second study area(Osceola study area)is situated within thenorthern section ofOsceola National Forest (ONF) andincludes thewestern edge ofPinhook Swamp. Comprising over 6,147 km2, theOkefenokee-Osceola ecosystem contains one ofthe largestcontiguous blocks ofblack bearhabitat remaining in the Southeast.
Location
The OkefenokeeSwamp occupies partsof Charlton, Clinch, andWare counties,
Georgia andBaker County, Florida. The ONWRwas established in 1937 and encompasses approximately 1,580 km2 ofswamp and adjacent pinelands (Scheick 1999); therefuge includes over 90% ofthe swamp. The 511-km2 Okefenokeestudy area included theswamp and islandsof the OkefenokeeNWR andthe adjacent private lands to thenorthwest ofthe Refuge(Fig. 3). Interior portions ofthe swamp included in the
7 Waycross
GEORGIA - . - . - \ FLORIDA
I
\ DeepCreek '-·-·-./ Jacksonville
Olustee lake City • 0 20 Kilometers
Figure 2. General area of the OkefenokeeNational WildlifeRefuge, Georgia, andthe Osceola National Forest, Florida, 1995-1999.
8 To Waycross
To Manor and US Hwy 84 /
To Argyle and US Hwy 84 • OKEFENOKEE NATIONAL WILDLIFE To US Hwy 441 Hickoiy REFUGE \ Hammock Cravens Island Minnies Island Floyds Island "7. J _, I I: / • Billys Island
The Bugaboo Island Pasture ... _ , I .J HoneysJIsland
0 10 Kilometers
Figure3. Black bear studyarea, OkefenokeeNational Wildlife Refuge, Georgia, 1995- 1999.
9 study areawere Craven' s Hammock, Craven' s Island,Hickory Hammock, andPine
Island. Private landswithin the study areawere predominately managedpine plantations owned by JeffersonSmurfit and Container Corporationand Rayonier. The nearestmaj or roadways were US 84 to the north,US 1 and 23 to the east, US 2 to the south, andUS
441 to the west. Portions of Charlton, Clinch, andWare counties made up the
Okefenokeestudy area. Nearbypopulation centerswere the cities of Waycross, to the north of the refuge, and Folkston, to the east.
Osceola NF,comprised of 2 tracts of land, encompassed approximately 798 km2 in portions of Bakerand Columbia counties in north centralFlorida. The 366-km2
Osceola study area included the southwest portion of Pinhook Swamp, the northeast portion of Impassable Bay, andad jacent private lands(Fig. 4). Private landswithin the study areawere predominately managed pine plantationsowned by Bankers Trust,
JeffersonSmurfit and Container Corporation,and Rayon ier. The nearestma jor roadways were 2, 127, US 90, US 441, which surround ONF. The closest areasof urban development were Lake City, to the southwest, and McClenny, to the southeast.
Topography and Geology
The Okefenokee Swampwas created by a landscape basin that was shaped by a sea level changeapproximately 200,000 years ago (Parrishand Rykiel 1979). As surface water accumulatedthe basin becameincreasingly inundated andhydrophytic plant communities were established (Loftinet al. 2000). Eventually, the Okefenokee Swamp became a peat-forming bog characterizedby forested wetlands, marshes, and open water
10 \o Fargo
GEORGIA - . - ·-·- . - . - . FLORIDA - · - ·
Pinhook Swamp
Impassable
Bay
OSCEOLA NATIONAL I FOREST To Lake City 0 10
Kilometers
Fig. 4. Black bear study area, Osceola National Forest, Florida, 1996-1999.
11 maintained by periodic firesand drought (Meyers andOdum 1991). Currently, the topography of the OkefenokeeSwamp is characterizedby vast expansesof relatively flat landwith elevations rangingfrom 34-40 m above sea level (Yin andBrook 1992).
Occupying a portion of the AtlanticCoastal Plain, the Swampis located approximately
160 kmfrom the AtlanticOcean. The name "Okefenokee"arises fromthe Native
Americanterm meaning "land of the quakingearth". The swampacquired that name because of the floating peat layers that can be foundwithin its boundaries. Those layers areso deep in some areasthat treesare rooted entirely in the peat, creating free-floating islandsthat sway when walkedon or disturbed. The OkefenokeeSwamp is composed of a wide diversity of habitat types. Bay forests, blackgum (Nyssa sylvatica) forests, cypress forests, prairies, shrubswamps, and open water constitute the 1,360 km2 of wetlandhabitats in the ONWR. Predominantly dry areasare comprised of largeexpanses
of foresteduplands and remote islands. Within ONWR, approximately 134 km2 of
uplandhabitat aremanaged for the protection andrestoration of longleaf pine (Pinus palustris) andwiregrass (Aristida stricta) communities (U.S. Fish andWildlife Service
2001).
Thesurf ace geology of theupland watershed of theOkef enokee Swamp is
characterizedby intensively leached sandysoils that have a markedincrease in the
amountof clay with increasing depth (Laerm andFreeman 1986). Soils in this region,
formedfrom marine sediments deposited during the Pleistocene andHolocene epochs,
aretypically acidic andpoor in nutrients.
Elevations within ONF rangefrom 115-125 m above present sea level. Soils
were primarilyMascotte-Oscilla-Surrency associations (U.S. Departmentof Agriculture
12 1984). Althougha lack of topographicrelief made this area similar to theOkefenokee
Swamp,differences in habitat composition exist. Unlike the Okefenokee,ONF is composed of a largemosaic <;>fsmaller swamps,bayheads, andwet pine flatwoods interspersed in uplandpine plantations. The largestcontiguous swampin ONF is the
243-km2 Pinhook Swamp,often referred to as thesouthern extremity of theOkefenokee
Swamp.
Hydrology
The OkefenokeeSwamp is one of thelargest freshwater wetlandsin theUnited
States. Water input to the swampprimarily occurs throughprecipitation. Rainfalland streamrunoff from the northwesternportion of the swampcontribute to theSuwannee
River drainage. Approximately 85% of the exiting flowoccurs via theSuwannee River in the western portionof the swamp(Loftin et al. 2000). The remainder of the water flow out of the swampis carriedby the St. Mary'sRiver drainage to the south (Rykiel 1977).
Major tributariesin thestudy area are Alligator Creek, BarnumBranch, Bear Branch, Big
Branch,Cane Creek, Goose Branch,Greasy Branch,Mill Branch, SurveyorsCreek,
SuwanneeCreek, Turkey Branch,and Water OakCreek. Manysmaller creeks throughoutthe private landssurrounding the refuge also contribute to water input; these
ripariancorridors serveas importanttravel routes for bearsmoving into andout of the
refuge. Practically all thewater runoffoccurs via the Suwannee andSt. Mary'sRivers which exit theswamp to the southeastand south, respectively.
Like the Okefenokee, the primarysource of water into ONFoccurs through precipitation. However, because thereare few channeled or natural streamsin this
13 region, most flowoccurs as sheet water. The most well developed channelis where
SuwanneeCreek drops from120 m to 95 m above sea level, serving as Pinhook Swamp's primarydrainage (Scheick 1999). Little Creek, the only other creek on the Osceola study area, flowswest into the SuwanneeRiver.
Climate
The climate of southeasternGeorgia and north central Florida is subtropical.
Summers were characterizedas hot andwet andwinters as cool anddry. In Georgia, meansummer highand low temperatures were 32.2°C and17 .8°C, whereasmean winter temperatures were 23.3°C and 8.9°C, respectively. Annual precipitation amounts during
1998 and1999 were 86.6 cm withmost rainfalloccurring in July (2 3.6 cm) andleast in
May (0 .51 cm) (National Climatic Data Center 1998, 1999).
Meansummer highand low temperatures in Florida were 31. 7°C and19 .4 ° C, whereas meanwinter temperatureswere 23.3°C and11.1 °C, respectively. Annual
precipitation during 1998 and1999 was 114.2 cm with most rainfalloccurring in
September(2 8.5 cm) andleast in November (0.46 cm) (National Climatic Data Center
1998, 1999).
Fauna
The ONWRand ONF support a diversity of wildlifespecies, including over 233
bird, 64 reptile, 49 mammal,39 fish,and 37 amphibianspecies. Big gamespecies
include black bear(protected on the Osceola study area) andwhite-tailed deer
(Odocoileus virginianus). Small gamespecies include easterncottontail (Sylvilagus
14 floridanus), marshrabbit (Sylvilagus palustris), andgray squirrel (Sciurus carolinensis).
Other common mammalsinclude bobcat (Felis rufu s), grayfox ( Ur ocyon cinereoargenteus ), coyote ( Canis latrans ), opossum (Didelphis virginianus ), raccoon
(Procyon lotor), river otter (Lutra canadensis), andskunk (Me phitis mephitis). Common uplandgame birds are bobwhite quail( Colinus virginianus), mourningdove (Zenaida macroura), andwild turkey(M eleagris gallopavo). Resident species that are recognized as endangeredby theUSFWS include gopher tortoise ( Gopheruspol yphemus), Eastern indigo snake(Dry marchon corais), red cockheaded woodpecker (Picoides borealis), wood stork (Mycteria americana), andgray bat (Myotis grisescens).
Flora
The 2 study areaswere dominated by an interspersion of pine flatwoods and hardwood swamps. Bald cypress(T axodium distichum)," black gum, andfetterbush
(Lyonia lucida) dominated the largerswamps (Wooding andHardinsky 1994). Pine flatwoodswere dominated by slash pine (Pinus elliotti), saw palmetto (Sereona repens ), andgallberries (Rex coriacea andL glabra) (Avers andBracy 1973). Small cypress swampsand bays also were distributed throughoutthe pine flatwoodshabitats in both study areas. Althoughthe habitattypes within the study areas were distributed differently, their proportions were similar (Scheick 1999) (Table 1 ).
History and Land Use
Historically,the primary useof landin the Okefenokeehas been timber production. In 1889, the SuwanneeCanal Company purchased 964 km2 of the
15 Table 1. Proportions of habitat types within the Okefenokeeand Osceola studyareas, Georgiaand Florida, 1999 (Scheick 1999). Habitat Type Proportion of GA Area Proportion of FL Area
Bay/ Gum/Cypress Forest 4.9% 3.2% Loblolly Bay Forest 17.9% 18.7% Slash Pine Forest 18.7% 27.6% CypressForest 37.7% 28.5% Mixed HardwoodHammock 1.7% 1.5% Mixed Hardwood SwampForest 13.2% 12 .6% Shrubland 1.1% 2.3% Emergent Marsh/Open Water 3.3% 0.4% BareSoiV Urban 1.5% 5.2%
16 Okefenokee Swampfrom the State of Georgia (McQueen andMizell 1926). After a failed attempt at draining portions of the swampto facilitateagricultural production, the companybegan harvesting cypress fromthe swamp. Beforedeclaring bankruptcy in
1897, theSuwannee Canal Company harvested over 7 million boardfeet of cypress timber fromthe Okefenokee Swamp(Izlar 1984). Timber harvestingresumed in 1909 when theHebard Cypress Company constructed several railroad tramsthat allowed access to theinterior of the swamp. By thetime logging in theswamp ceased in 1927, approximately 42 5 million boardfeet of timber had been harvested(Hopkins · 1947). From 1890-1927, enough maturecypress were removed that approximately 40% of the forestedarea that was virgin old growthhas now been replacedby younger second growth(Hamilton 1982). In 1936, the U.S. Fish and Wildlife Service purchasedthe
HebardCypress Company's holdings and created theOkefenokee National Wildlife
Refuge. Although logging within therefuge has ceased, themajority of landad jacent to andsurrounding the swamp is now owned andintensively managedby largetimber companies. Those private holdings aretypically managed for slash pine production on a
20-25 yearrotation. Inthe more remote areasof theinterior refuge, virgincypress forestsare still present; the oldest treeshave been aged at over 600 years (Duever and
Riopelle 1983).
Excluding forestryuses, huntingis the most common use of private lands outside the ONWR. All private landholdings within the Okefenokeestudy area were leased to 3 hunt clubs. The bearhunting season consisted of 3, 2-day hunts occurring on thelast
Friday andSaturday of September andthe first 2 Fridays and Saturdaysin October.
Beginning in 1998, theGeorgia Department of NaturalResources permitted a 3-day hunt
17 in the Dixon Memorial Forest (adjacent to the northernedge ofONWR) during the first week of December. Hunting regulations stipulated that 1 bear may be harvested per licensed hunter per yearby still-hunting (firearmsand archery ) or with the aidof dogs,
andno baiting is permitted. Althoughhunting pressure is significantduring the relatively
short season, the majority of humanpresence on the Okefenokeestudy area was the result
of a year-round dog chase season wherebaiting was allowed. On the Okefenokeestudy
area, hunt club members chasedbears with dogs approximately3 days perweek from
April to mid-September. The only times that bearstypically were not being chased was
during thedenning season, the2-week period immediately prior to thehunt season, and
the weekdays between the 3 consecutive hunt weekends. Access into the ONWRwas
restrictedto designatedentrances andcanoe trails, andhunting of anykind is prohibited.
The Osceola area was originally occupied by the Timucuan Indians, as noted in
1535 by Hernandode Soto during his travels throughthe area that is now LakeCity,
Florida. During the early 1800's, Seminole Indians occupied much of the area until
moving to anarea further south (U.S. Department of Agriculture 1984).
Throughout the mid- to late-1800s much of the prosperity andgrowth in the
Osceola areawas associated withcotton production (U.S. Departmentof Agriculture
1984). As timber practices becamemore refined,however, the majority oflandwas
converted forforestry practices. As is the casein Georgia,the majority oflandwithin the
Osceola study area,including ONF, is currently managedfor slash pine production.
Access to the private landsin the Florida areawas regulated by timber companies
andis restricted to employees and members of leased hunt clubs. Unlike the ONWR,
18 however, ONFhad freeaccess via public roads. Although the bear season was closed in
Florida, white-tailed deer andhogs were hunted on the Osceola study area.
19 CHAPTER III
MATERIALS AND METHODS
StudyDesign
Examiningthe dynamics of wild animalpopulations oftenrequires estimates of population size (Pollocket al. 1990). Obtaining reliable population abundancedata, however, has long been a challenge to wildlifebiologists. Althoughmark-recapture analysesare the standard for estimating population size, models used for these analyses oftenrequire specificassumptions thatare difficult to meet (Mowat and Strobeck2000).
The assumption of demographicclosure, in which thereare no additions or permanent deletions during thestudy, canbe used to separate2 types of population models
(Menkens andAnderson 1988). The firstbeing closed models, which assume population
closure, andthe second being open models, which allow for additions and deletions.
Unfortunately, both open andclosed models require additional assumptions thatmay be
readily violated when applied to bears (Mace et al. 1994). The most importantof these
may be theassumption ofequal catchability, which is thatevery individual in a
population has anequal chanceof being capturedduring a trapping session. Pollock et
al. (1990) identified2 formsof capture variation that violate theequal catchability
assumption: (1) trapheterogeneity, whereby animalshave differentcapture probabilities
forreasons inherentto theindividual (e.g. social status, age, sex); and(2 ) trap response,
whereby anindividual's capture probability is dependent upon its capture history (trap
happy andtrap shy response). Lastly, both closed andopen models arebased on the
20 assumption thatmarks are not lost andall markedindividuals arerecorded correctly (Otis et al. 1978, White et al. .1982, Pollock et al. 1990).
Of primaryinterest to this study arethe effects of trap response andhow biases associated withthis response affectestimates of population size. Our original intent was to live trapblack bearsfor 5 yearsin theOkefenokee-Osceola ecosystem andprovide reliable estimates of population size forbears in the Okefenokee andOsceola study areas using mark-recapturemodels. Although researchers were successful in maintaining adequate capture andrecapture samples in the 1995-1998 trapping seasons, analysisof capture data indicated that a trappingbias mightbe adversely affecting our capture success. Inparticular, we were concernedthat our intensetrapping efforts were resulting in a learnedtrap response (avoidance)among bears. Consequently, we decided to employ a relatively new recapture technique during the finalyear of the study in an attempt to alleviate such samplingbias.
Recent effortsto develop non-invasive genetic samplingtechniques have resulted in promising new tools forbear research and management (Paetkau and Strobeck 1994,
Paetkauet al. 1995, Mowat and Strobeck2000). Of particularinterest is the use of DNA fingerprintingtechniques based on microsatellite loci to individually identifyanimals fromhair, scat, or tissue (Woods andMcLellan 1995). Consequently, microsatellite analysis canbe used to "mark"animals, andprovide recapturehistories forall individuals sampled. . Those genetic markersobtained frommicrosatellite loci canbe identifiedfrom minute quantitiesof DNA, arehighly variable, andeasily interpretedin terms of allele frequencies(Wright and Bentzen 1995, Parkeret al. 1998). When incorporatedinto mark-recapture models, recapture histories collected from non-obtrusivesampling
21 techniques may provide estimates of population size that help reduce biases such as those associated with live trapping. To investigate this, hair sampleswere collected from free ranging blackbears in the Okefenokee-Osceola ecosystem using baited barbedwire enclosures (Woods et al. 1996).
Consideringthe assumptions of open andclosed population models, non obtrusivesampling techniques andusing DNA fingerprintsas "marks"have several advantagesover traditional methodsof capturing andmarking bears. Becausebears do not have to be physically captured, restrained,or marked,there is no negative association with the samplingtechnique, thereby decreasing the likelihood fora learnedtrap response. Temporal and spatial biases in capture probabilities canalso be minimized by
because an entire study areacan be sampled at one time using the hair-trapping technique. In addition, genetic markers cannot be lost, as is oftenthe case with traditionalmarks. This is extremelyimportant for mark-recapture studies since loss of
marks always results in positively biased estimators (Pollock et al. 1990). Another
advantageis that samplesizes will likely increase because trap response is minimized and trap densities can �e higher. Lastly, there is no special expertise needed in capturinghai r
samples,so more effortcan given to trappingrather thantraining.
Study Area Delineation
I delineated each of the respective study areasby circumscribingeach of the 1999
hairtrap sites with a circle, thearea of which was equivalent to the average home range
estimate for fe male bears. Home rangeswere estimated using the 95% Minimum Convex
Polygon estimator in ArcView® GIS (ESRI, Redlands, California). I limited the
22 telemetrydata to bearswith �30 locations collected from1995-1999 and1996--1999 on
the Okefenokeeand Osceola study areas, respectively. Furthermore, telemetry data were
restrictedto locations with collection dates thatcoincided with the months of the hair
trappingseason (June to September, 1999). All home rangedata used in delineating
study areaswere collected during this study (J. D. Clark, U.S. Geological Survey,
unpublished data).
Based on my criteria, average home rangesizes forfemale black bears were 40 .0
and 54.5 km2 on the Okefenokeeand Osceola study areas, respectively. I bufferedeach
of the 1999 hair trapsites in the Okefenokeearea with theaverage radius of 3,198 m.
Each hair trap site in the Osceola area was bufferedwith the average radius of 4,330 m. · · Consequently, the Okefenokeestudy areawas 511 km2 andthe Osceola study area was
366 km2 in size (Fig. 5).
Trapping and Handling
Black bearswere primarilytrapped from early June through lateSeptember from
1995-1998 and1996--1998 on the Okefenokee andOsceola study areas, respectively.
Limited trappingsessions during late October andearly November of 1995, 1996, and
1997 focusedon more remote locations withinthe ONWR. All bears were captured
using Aldrich spring-activated footsnares (Aldrich Animal Trap Company, ClallamBay,
Washington). Because inaccessible and impenetrablehabitat precluded random trap
placement, trapsites wereestablished according to habitat type, knownbear travelroutes,
andbear sign.
23 Georgia
Florida
.,__ __. state and fe deral landholdings _ N CJstudy area bolllldary /\ / state line / \/ 20------0 20 40 Kilometers I
Fig. 5. Delineation ofthe Okef enokee andOsceola study areasusing home range data, Georgiaand Florida, 1999.
24 We primarily used 3 trapping techniquesto capturebears on both study areas.
Standardtrail sets baited withdry com placed in hangingplastic bottles were used as an initial attempt to catch bearsat trapsites(Clark 1991, Brandenburg 1996). Bearswere lured to thosetrapsites with artificial raspberry flavoring(Mother Murphy's, Greensboro,
North Carolina). Blind sets (i.e., traps without bait) were used to capture bearsthat had learnedto steal baitswithout being captured. Lastly, dirt-hole sets were used when blind sets fai led to capture''trapwise" bears. A dirt-hole set consisted of a snarewith the foot loop placed atop a freshlydug hole andcamouflaged with discarded trash. We used trees to secure snares,but we used mobile home anchors(12 3 cm long witha 10-cm auger) to secure snareswhen treeswere unavailable.
We checked traps1 to 2 times daily, depending on site conditions. Inwell-shaded areas, trapswere usually checked by 1100. I checked trapsites without the cover of shade or in close proximity to humanactivity by 0800. Those trapswere deactivated during the day to prevent bears frombeing captured in direct sunlightor fullview of thepublic.
Captured bearswere immobilized witha 2:1 mixture of ketaminehydrochloride
(Ketaset, Bums VeterinarySupply Incorporated,Farmers Branch, Texas) andxylazine hydrochloride (Rompun,Haver-Lockhart Incorporated,Shawnee, Kansas).
Immobilization drug was administered intramuscularlywith a push pole at a dosage of
4.4 mg (1 mV 22.7 kg) ofKetaset and2. 2 mg (1 mV 45.5 kg) ofRompun per kg of estimated body mass. Bearswith injuriesreceived aninjection of Liquamyacin(LA- 200,
Pfizer AnimalHealth, New York, New York) at a dosage of8.8 mg/kg body mass. After immobilization, a wetting agent (AkwaTears, Akom Incorporated, Abita Springs,
25 Louisiana)was applied to thebears' eyes to prevent desiccation. A blindfoldwas then placed over theeyes to protect them fromdebris andto minimize visual stimuli.
A permanentidentification number was tattooed on theinside upper lip of each bearusing 0.8-cm numeric digits (Nasco,Fort Atkinson, Wisconsin) andanimal tattoo ink(Ketchum Manufacturing,Ottawa, Canada). Numbered eartags, corresponding to individual tattoo identificationnumbers, were placed in both earsof each bear. Male bearsreceived a rectangularmetal tag (Hasco Tag Company,Dayton, Kentucky) in the rightear and a plastic round colored tag in theleft ear. Female bearsreceived the same types of tags but they were placed in opposite ears. This methodof tagging enabled hunters to identifymale and femalebears that were seen during theyear-round dog chase season.
All female bears> 1 year-oldreceived a motion-sensitive radio-collar(Telonics
Incorporated,Mesa, Arizona andLotek EngineeringIncorporated, Ontario, Canada). A select number of male bearson theOkefenokee study areas were radio-collaredas partof anotherstudy (M. R. Pelton, Universityof Tennessee, unpublished data). I equipped each collarwith a 12.5-cm by 0.4-cm leather spacer that servedas a breakawaydevice
(Hellgrenet al. 1988). All spacers were soakedin vegetable oil forat least 1 month beforebeing placed on a collar to prolong its durability. A firstupper premolartooth was extractedfor aging by cementum annulianalysis (Willey 1974). Sectioning, staining, and aging of teeth was conducted by Mattson Laboratories (Milltown, Montana). Bearswere weighedwith a spring scale andstandard morphologicalmeasurements were recorded.
Body temperature, pulse, andrespiration were monitored throughout each immobilization.
26 Information concerningthe general description, reproductive status, tooth wear, andphysical appearancewere recorded forall bears. I collected tissue andhair samples fromeach bear to be used formicrosatellite analysis. Lastly, yohimbine hydrochloride
(Lloyd Laboratories, Shenandoah,Iowa), anantagonist for xylazine hydrochloride, was administered throughthe sublingual vein at a dosage of 0.2 mg/kgof body mass.
Hair Trapping
I collected hair samples from free-rangingbears on both study areaswith bait surroundedby barbedwire. These baited"enclosures" consisted of a single strandof barbedwire (2-strandwire, 4 points, 7.5-cm spacing between barbs)attached to treesto form a polygon (Fig. 6). Wire was affixedto the outside of perimeter trees using2.5 -cm aluminum fence staplesand tensioned by handusing fencingtools. Eighty-eightand 90 hair trapswere maintainedbetween 12 June and 27 September 1998 on the Okefenokee and Osceola study areas,respectively. This resulted in anaverage density of 1 hair trap per 5.8 km2 on the Okefenokeestudy area, or approximately 7 and24 hair traps per femaleand male home range,respectively. On theOsceola study area, hair trapdensity
2 was 1 per 4.1 km , or approximately 13 trapsper femalehome range. No male bears were radio-collaredon the Osceola study area. Based on the average home rangesize for
Okefenokeemales, however, there were approximately 34 hair trapsper male home range on theOsceola study area. Those hair trap densities were within those suggested by Otis et al. (1978) thatpopulation studies be designedso that animalshave �4 trapsin their estimated home range.
27 �1 1 ,,, ·1111 � f;:. ' Com . '" J1 ------i . ·' \cr- ;I Jf ------J '----.Ji:
N 00
2.5cm aluminum fencingstaple
-45 cm T above groun1 12 gauge barbed wire with 4 points, 7.5-cm spacing
Fig. 6. Diagramof a baited barbedwire enclosureused to collect hair samples from black bearson the Okefenokeeand Osceola study areas, Georgiaand Florida, 1999. Three 20-ounce plastic bottles filledwith dried com were hung approximately 2 m above the groundwithin the enclosures froma wire affixedto 2 perimeter trees. I positioned the bait so that no point on the perimeter fence was less than3 m away to prevent bearsfrom reaching baits without crossing the barbed wire. Furthermore, I was carefulto ensurethat the barbed wire was approximately 45 cm fromthe ground. If necessary, terrainirregularities were blocked with debris to ensure uniform wire height above ground. I used 3 bottles of com to entice bearsto enter anenclosure multiple times, thereby increasing the likelihood that a hair sampleof sufficientsize (�5 hairs) would be collected. I refrainedfrom using scent attractantsso that bearswould not be enticed into enclosures after bait had already been taken. That was done to decrease the chanceof cross-contaminatingsamples with hair frommultiple bears.
I checked hair traps at intervalsof about 9 days; each of intervalwas considered a single samplingoccasion. From June-September, I sampledthe Okefenokeeand Osceola study areason 10 and11 occasions, respectively. The Okefenokee study areareceived 1 less samplingoccasion because it took longer to install hair trapson that area. Nine days was selected as the intervalbetween hair collections to minimize the risk of DNA degradation,to decrease the chanceof having individual samples contaminated by multiple bears, and dueto logisticalconstraints of travelling between 2 study areas. I examinedeach barb for hair by placing my handor a white card behind each barb and removing anyhairs. Groups of hairs collected fromindividual barbswere considered separate samples and placed in individual manila coin envelopes. Each envelope was labeled with the date, trap identification number, number of hairsin the sample, and number of roots in the sample. I placed all collected hair samples in airtightstorage
29 containing desiccants. Sampleswere then frozen until microsatellite analysiscould be performed.
As was thecase with trappingusing footsnares, inaccessible and impenetrable habitats within both study areasprecluded randomplacement of hair traps. Therefore,I
established hair trap locations according to habitat type, knownbear travel corridors, and bearsign. Every effort was made to ensure adequate coverage without �ajor gaps in the
trapping gridon each study area. All hair traps were mapped using a Global Positioning
System (GPS) unit andUniversal TransverseMercator {UTM) coordinateswere
recorded.
Subsampling
Bearvisits to hair trapstypically resulted in multiple hair samplesbeing lefton
barbs. Although all samples were individuallycollected, only samples with �5 hairs
were good candidates for microsatellite analysis(T. King, U.S. Geological Survey,
personal communication). Because of the largenumber of hair samplescollected,
however, a subsamplingdesign was implemented to reduce the number of samplesfor
analysis. To accomplish that, I randomlychose 8 hair samples fromeach of the 10 and
11 trapping periods on the Okefenokeeand Osceola study areas, respectively. I selected
8 samples for microsatellite analysis based on simulation results of the modifiedLincoln
Petersen estimator using live-capture data from1998. Those simulations suggested that8
samplesper samplingperiod should provide enough recaptures to provide population
estimates with coefficients of variation �0.25. The method of uniformrandom sampling
30 was chosen to ensure equal samplingeffort from every trapping period (K. Pollock, N. C.
State University, personal communication).
Microsatellite Analysis
Removal of hair roots andpreparation forDNA extraction was performedat the
University of Tennessee. I clipped approximately 0.6 cm of the root end of each hair and placed all roots fromeach sampleinto a 1.5-ml centrifugetube. DNA extractions, replication by polymerase chain reaction (PCR), and microsatellite analysiswere performedat the U.S. GeologicalSurvey Aquatic Ecology Laboratory at the Leetown
Science Center, Kearneysville, West Virginia(see Table Cl fordetails of genetic analyses).
Hair samplesselected for analysiswere identifiedbased on 8 individual microsatellite loci. Those loci included GIA, GlD, GlOB, andGlO L (Paetkau and
Strobeck 1994), andGlO C, GlOM, GlOP, andGlO X (Paetkau at al. 1995).
Probability of Identity
One of the primary assumptions of open and closed population models is that marksare not lost or overlooked (Pollock 1990). Furthermore, it imperative that individuals be correctlyidentified. Geneticmarks, in the form of microsatellite genotypes, have the potential to replaceconventional marksif the tags correctlyidentify individuals in trappingsessions (Woods et al. 1999). Although I assumed matching hair samplesrepresented a recaptured animal, it was possible that differentindividuals could shareidentical genotypes at the 8 loci examined(Woods et al. 1999, Mills et al. 2000).
31 Factors influencingthe likelihood of hair sampleshaving an identical genotypeinclude thenumber of loci examined(Woods et al. 1999) and the degreeof genetic variability present in thepopulation (Paetkau et al. 1998).
To assess the variability of the loci examined, I estimated the probability that2 individuals drawn at randomfrom a population would sharean observed genotype. That statistic, referredto as the probability of identity (Pl), is definedas the proportion of the population possessing genotypesthat cannotbe distinguished fromone other individual
(Mills .et al. 2000). In studies of this type, PI essentially provides a measure of how useful a suite of loci will be for individual identification. The PI for individual loci can be calculated as follows:
where p; andPi are thefrequencies of the ith and jthalleles (Paetkau andStrobeck 1994).
Small PI values (indicating a small proportion of matching genotypes) typically occurin the presence of many alleles of approximately equal frequency, whereas largePI values canbe expected in populations with low genetic variation (Mills et al. 2000). When calculating Pisingle locus, however, it is necessaryto assume thatallele genotypesare in
Hardy-Weinberg equilibrium (Taberlet and Luikart1999) . . In the Hardy-Weinberg law, the followingconditions are presumed (Klug and Cummings 1991 ):
1) largepopulation size,
2) random matingwithin the population,
3) there is no selective advantagefor any genotype,
4) there is anabsence of genetic mutations, and
32 5) gene flowis prohibited due to population isolation.
Assuming loci areindependent (Mills et al. 2000), a PI across all loci canthen be calculated as (Paetkau et al. 1995):
PI overall = TT(Pisingle locus )·
Assuming theconditions of Hardy-Weinberg proportions aremet, thecalculation of
Ploveran serves as a statistical basis forgenetic match declarationsamong individuals.
It is uncommon, however, fornatural populations to successfullymeet all of the assumptions that areassociated with Hardy-Weinberg equilibrium. Telemetry data from this study indicate that spatial organizationand movement by bearsare not restricted(J.
D. Clark, U.S. Geological Survey, unpublished data). Therefore, isolation does not appearto be a detriment to gene flowfor the Okefenokee-Osceola population. The number of bearsinhabiting the ecosystem, however, is unknown. Furthermore,research indicates that mating may not be randomwithin black bearpopulations (Rogers 1987).
Therefore,addressing assumption violations in relation to PI values is importantto identifypotential biases that canaffect the ability to correctlyidentify animals basedon genotypes. The most probable form of bias will arisein populations that are highly substructured(Taberlet andLuikart 1999), species that exhibit a highincidence of philopatry(A vise et al. 1995), andin populationscontaining manysiblings (Donnelly
1995). In all of these cases, it is likely thatsamples will come fromindividuals with sharedancestry, resulting in a PI that is biased low (Taberletand Luikart 1999). In the
Okefenokee-Osceolablack bearpopulation, survivalrates andreproduction were high during this study (J. D. Clark,U.S. GeologicalSurvey, unpublished data). Therefore,it is
33 possible that bearssampled at hair trapsin 1999 possessed shared ancestry, which could hinder my ability to correctly identifyindividuals.
Consequently, I used analternative computation forPI thatestimates a probability of identity amongrandomly sampled siblings (Plsibs)(Taberlet and Luikart1999). That statistic provides a more conservative meansof identifyinghow manyloci areneeded to obtain a sufficientlylow Pl, thereby increasing the likelihood that all individuals are correctly identified. A PI for randomly sampled siblings can be estimated by:
Plsibssinglctocus = 0.25 + (0.5'22p; )+ lo.5('22 p; J J- (0.25'22p: ). Additionally, a Plsibsacross all loci can thenbe calculated by:
Plsibs = TI(Pisibssinglclocus )·
Like Ploveran, this equation assumes randomsampling of individuals andindependence amongalleles within andbetween loci. However, by assuming that genotypes are arising
fromclosely related individuals,Plsibs representsan upper'limit on the rangeof PI
values across all observed genotypeswithin a population (Taberlet andLuikart 1999).
Two additional tests, developed by Woods et al. (1999), calculate the probabilities
that a parentor offspring of anindividual (Ppar-otrs) or theirsibling (Psib) would have the
samegenotype. As a result of the close genetic relations between siblings, the sibling
match test (Psib) is the most conservative of all tests described and canbe calculated by
(Woods et al. 1999):
= 1 + 2 P; + 2 �ib { P; )1 4 , forhomozyg .otes, and
�ib = (1 + P;+ pj + 2 P;Pj )14 , for heterozygotes.
34 I used the sibling match test to identify8-loci genotypes that are potentially shared between > 1 individual. Genotypes were accepted as unique bears when Psib < 0.05. Hair samples failing to meet that criterion were excluded from analysis (Yvoods et al. 1999,
Mowat and Strobeck 2000).
Hardy Weinberg and LinkageDisequi librium Tests
In order to make valid inferences from calculations of Plsmg1elocus, the assumptions of the Hardy-Weinberg law must be upheld. For studies of this type that are concerned with individual identification frommicrosatellite loci, the assumptions of random mating and an absence of genetic mutations are particularly important. Violations of those assumptions are often theresult of breeding between closely related animals.
Differentiating individuals with shared ancestries is complicated by the fact that these animals may be sharing identical genotypes at several loci (Taberlet and Luikart 1999).
Furthermore, there is potential for inbreeding to cause genetic mutations within the genome, which is a direct assumption violation of the Hardy-Weinberg law. To investigate the likelihood of inbreeding in the Okefenokee-Osceola population, I used the
Hardy-Weinberg probability test in ProgramGE NEPOP 3 .1 (Raymond and Rousset
1995). That analysis tests the null hypothesis that there is a random union of gametes and
homozygote genotypes occur at a frequencyexp ected fromthe overall allele frequency
distributio�. I performed individual tests at each locus forevery 8-loci genotype that was
identified (Paetkau et al. 1998, Boersen 2000). Rejection of Ho would imply a likelihood
of inbreeding or other formof non-random mating in the population.
35 I also used the linkagedisequilibrium test in ProgramGENEPOP 3.1 (Raymond and Rousset 1995) that tests thenull hypothesisthat genotypes at one locus are independent from genotypes at another. Rejection of thathypothesis would indicate
some non-randomassociation between alleles of differentloci (i.e. linkage
disequilibrium) (Avise 1994). Because the 8 microsatellite loci used in my analysis have
been foundto be independent(Paetkau and Strobeck1994, Paetkauet al. 1994), any
significantlinkage observedamong loci pairs may indicate samplingbias, non-random
mating withinthe population, or stochastic processes which affectpopulation genetics (T.
L. King, U.S. Geological Survey, personal communication).
Estimation of Population Size
General. Population models for estimating wildlife abundancecan be classified
as eitherclosed or open. Closed models are constrainedby the assumption of no births,
immigration, deaths,or emigrationduring and between samplingperiods (Otis et al 1978,
Pollock et al. 1990). Therefore, closed models are best suited for experimentsestimating
population size where samplingeff ort is maintainedfor relatively short periodsof time
(Pollock et al. 1990). Conversely, open models typically spanlonger time periods,
consist of 2!:3 samplingoccasions, and allowfor fluctuationsin population size between
sampling periods. Consequently, open models canbe used to estimate additional
parameterssuch as survival and recruitmentrates, as well as population size. The
assumptions thatmarks arenot lost or overlooked andall animalsin a population have an
equal probability of capture arerequired of both open andclosed population models.
36 I chose to use a combination of closed ap.d open models to estimate black bear population size on the Okefenokeeand Osceola study areas (Fig. 7). Estimates for1999 were obtained by dividing the hair-trappingseason into halves andusing the modified
Lincoln-Petersen model (Pollock et al. 1990). Within-yearestimates for1999 were also calculated using several closed multiple mark-recapture models described by Otis et al
(1978), White et al. (1982), and Chao (1987, 1988, 1999) {Table 2). Estimates from closed multiple mark-recapture models were obtained using Program CAPTURE
(Rexstad andBurnham 1992). Additionally, several pooling configurationsof the1999 hair trapsessions were considered for all models (Boersen 2001) (Fig. 8 and 9).
Collapsingcomplete capture history matrices allowed me to reduce the number of trappingoccasions and increase samplesizes within sampling sessions. Unfortunately, thedisadvantage of pooling samplingperiods is that some recaptures will be lost due to multiple obseivations withina session. Therefore a trade-offexists; identifythe pooling configurationthat maximizes captureprobabilities while minimizing the loss of data. . I used the criteria described by Otis et al. (1978) to select the model that was most appropriate formy data.
I also used the open Jolly-Seber models (Jolly 1965, Seber 1965, Pollock et al.
1990) provided in ProgramJOLLY (Pollock et al. 1990) to estimate population size by including the 1999 hair-capturedata as anadditional yearof captures forthe live-trapping data. Althoughpopulation closure was not a concernwhen using the open Jolly-Seber model, every effortwas made to identifypotential assumptionviolations and minimize potential biases. In this study sex ratios of captured bears did not differ from 1:1 on eitherstudy area(GA, FL: P= 0.015, 0.033). However, annualhome rangesizes of
37 Jolly-Seber : I
ModifiedLinco ln-Petersen
GA I 199s II 1996 ,, 1997 11 1998 I DJIJJ [I) [I)0 [I) [zJ [TI [2J (1Q] w 00 FL I 1996 11 1997 II 1998 1 DJ12] [I] 0[I] [I) [zJ [TI[1Q] [2] IT!] Live-capture Periods Ha ir-trapp ing Periods
Multiple mark-recapture models
Fig. 7. Capture-recapture models used to estimate black bear population size on theOkefenokee and Osceola study areas, Georgia andFlorida, 1995-1999. Table 2. Closed multiple mark-recapturemodels used to estimate within-yearpopulation size from1999 hair capture data (Otis et al. 1978 and Chao 1987, 1988, 1989). Model Description of CaptureProbabilities
Constant
Mt, Chao Mt Varywith time Mb Varyby behavioral response to capture Mb, Chao Mb Varyby individual animal
Mth, Chao Mth Varyby time andindividual animal Mbh Varyby behavioral response to capture and individual animal
Mtbh Varyby time, behavioral response to capture, andindividual animal
39 1999 Hair Trapping Sessions
A. 10 Sampling Periods, 9 days each
B. 5 Sampling Periods, 18 days each
C. 3 Sampling Periods, 36 or 27 days each
I D. 3 Sampling Periods, 27 or 36 days each
E. 3 Sampling Periods, 27 or 36 days each
Fig. 8. Pooling configurationsof the hair trappingsessions considered formultiple mark-recapture models to estimate population size on the Okefenokeestudy area,Georgia, 1999. 1999 Hair Trapping Sessions
A. 9 Sampling Periods, 9 days each
B. 5 Sampling Periods, 18 or 27 days each
C. 4 Sampling Periods, 18 or 27 days each
D. 3 Sampling Periods, 27 or 36 days each
Fig. 9. Pooling configurationsof the hair trapping sessions considered formultiple mark-recapture models to estimate population size on the Osceola study area, Florida,1999. male bearson the Okefenokeestudy areawere approximately 50% largerthan those of females (J. D. Clark, U.S. Geological Survey, unpublisheddata). Consequently, there is potential forcapture probabilities to be biased as the likelihood of encountering traps differs by sex. One way to minimize the effectsof heterogeneity in capture probabilities is to stratifycapture data into groupsexhibiting similarcapture probabilities andestimate populationsize foreach strata(Pollock et al. 1990). The advantageof not stratifying data, however, is that the increase in sample size as a result of pooling will oftenresult in higher capture probabilities (Pollock et al. 1990). Furthermore, an increase in capture probabilities will make mark-recapture experiments more robust to model violations. I tested thathypothesis by running the Jolly-Seber models using data pooled by sex and age classes andstratified by sex.
Closed Models. The modified Lincoln-Petersen model yields an estimate of
population size based on a mark-release period followedby a recapture period.
Assumptions of the modified Lincoln-Petersen model are (Otis et al. 1978, Pollock et al.
1990):
1) the population is geographicallyand demographically closed,
2) allanimals have the samecapture probability in each sample,
3) marksare not lost or overlooked and arerecorded correctlyupon capture, and
4) marking does not affect the catchability of the animal.
This model estimates population size as (Chapman1951):
42 where n1 is the number of animalscaptured in the firstsample, n 2 is the total number of animals captured in the second sample,m 2 is the number of markedanimals captured in the second sample, and Ne is the estimated population size.
Anapproximate unbiased estimate of variance (Seber 1982) is calculated as
n n varNe (11i + 1X 2 + 1X11i - m2 X 2 - m2 ) (m2 + lf (m2 + 2)
Using only the hair trap data·from 1999, the capture periods forthe modifiedLincoln
Petersen model were frommid-June to earlyAugust (1st period), andearly August to mid-September (2nd period).
I also used 9 of the closed models described by Otis et al. (1978) to provide estimates of population size using capture histories obtained from1999 hair trapping data. Those models were developed for use with mark-recaptureexperiments involving
>2 capture occasions. Therefore,the multiple pooling configurationsof my hair trap data were appropriate for within-year population estimates for1999.
The simplest, yet most restrictive, of the multiple mark-recapturemodels used in this analysiswas the equal catchability model (Mo). Like the Lincoln-Petersen estimator, this model is based on the assumption of population closure, equal capture probabilities, no lost marks,and constant capture probabilities. Consequently, model M0 is one in which there is no heterogeneityassociated with capture probabilities, no behavioral response to capture, andno temporal variationin the experimental situation (Otis et al.
1978). Itis unlikely, however, that all of these assumptions canbe met in most wild populations (White et al.1982, Pollock et al. 1990). .
43 Assumptions of model Mt, thetime variationmodel, are thesame as those of model Mo,with one exception. Although it is assumed that capture probabilities are constantwithin individual trappingsessions, these probabilities can change between sessions (Otis et al. 1978). The behavioral response model, Mb, allows forbehavioral responses (i.e., trap happy or trapshy) to influencecapture probabilities. The heterogeneity model, Mb, assumes that capture probabilitiesvary between individuals.
Models Mtb, Mbb, andM tbb were also used to provide estimates of population size. These
3 models allow for differentforms of heterogeneity to simultaneously affectcapture probabilities.
I also considered a suite of 3 additional multiple mark-recapturemodels described by Chao (1987, 1988, 1989) thatwere designed foruse with relatively sparse capture frequency data. Two of those models (Mb andM t) are modified formsof the heterogeneity andtime models described by Otis et al. (1978). However, when mean capture probabilities are relatively small (e.g. animalscaught 1-2 times during sampling), the Chao estimators areusually less biased thanthe heterogeneity andtime variation models presented by Otis et al. (1978) (Chao 1988).
The heterogeneity model developed by Chao (1987, 1988), Chao Mb, has a closed form estimator which calculates population size as:
2 Chao Nh = S + J; . 2J;
or,
44 2 f. 2J;]/[(1 -3j; )]} · 2 Chao. N,_ S + [ ]{[1- 1 > 2 > 31/2,and 3f./2 > 2J; = "f tfi f , tJ; , 2J; if. tJ; 2 where S= the number of individuals capturedin the experiment, /k= the number of animalscaptured exactly k times, andt = the number of samplingoccasions (Chao 1988).
To calculate confidenceintervals, variance estimators are calculated as:
and,
where
[(l -2/2) (l - 3 A = / /i)] (Chao 1988). tfi tJ;
Model assumptionsassociated with Chao Mh andChao Mt are identical to those of their Mh andM t counterparts. Estimators formodels Chao Mt andChao Mth are described by Chao (1988) andChao et al. (1992). Those models consider time variation in capture probabilities and time variationin the presence of heterogeneity.
Estimates of population size and associated standarderrors were calculated forall multiple mark-recapturemodels using Program CAPTURE(Rexstad andBurnham
1992). Because model Mtbh is useful only conceptually (Otis et al. 1978), an estimator for thismodel does not exist (White et al. 1982). I used chi-squaregoodness -of-fittests
45 within Program CAPTURE to identifyvariation in capture probabilities as a result of time, behavior, andindividual heterogeneity effects. ProgramCAPTURE incorporates the results of these chi-squaretests into a discriminant function procedure which constructs a model classificationfunction (Otis et al. 1978). This classification function was then used to aid in selecting the most appropriate model forthe different pooling configurations of the 1999 hair data. Additionally, I tested forequal catchability by comparingobserved capture frequenciesto a zero-truncatedPoisson distribution
(Caughley 1977).
One of the most critical assumptions of multiple-markrecapture models is that of population closure. Failure to meet that assumption indicates additions (birth or immigration) and/ordeletions ( death or emigration)to the population during the sampling period. In this study I was only concernedwith geographicclosure since bearswere sampledoutside of the birthing season. I attempted to minimize violations of the
geographic closure assumptionby delineating study area boundariesusing 3 and4 years of telemetry data collected during this study. I created a bufferaround each hair trap, the
area of which was equivalent to the averagehome rangeestimate forfe male bears on each study area. Therefore,I hoped to increase thelikelihood that my hair traps would be
samplingmore resident bears thantemporary immigrants. I used the procedure in
ProgramCAPTURE to test for population closure forall pooling configurations.of my
hair-trapping data. Simulation results fromOtis et al. (1972) indicated that the closure
test is oftennot valid when a behavioral response bias was present. However, the test
· does not appear to be affectedby individual heterogeneity or random time variation (Otis
et al. 1972).
46 Op en Models. I used the open Jolly-Seber model (Pollock et al. 1990), provided withProgram JOLLY, to calculate estimates of population size across yearsusing capture andhair -trappingdata. Assumptions of that model are(Pollock 1990):
1) the population is open to additions frombirths andimmigration, and deletions
from deaths andpermanent emigration,
2) individuals have anequal probability of capture in the ith sample,
3) every markedindividual in the population at the time of ;th samplehas the
sameprobability of survival until the ith + 1 sample,
4) marks arenot lost or overlooked, and
5) all samplesare instantaneousand each individual is released immediately after
capture.
ProgramJOLLY provides 5 models forestimating population size: model A, A',
B, D, and2. Model A is the general model in Program JOLLY andallows capture and survival probabilitiesto vary. The deaths only model, model A', allows for losses
(deaths andemigration) in a population but not additions . . Model B, the constantsurvival model, assumes survival probabilities areconstant throughout the experiment. Model D is the constantsurvival andcapture model, which assumes that probabilities of capture andsurvival remain constant over the entire study. Model 2, the temporarytrap response model, allows for a short-termeffect of markingon the survival and captureprobabilities.
Population size is estimated by:
and,
47 ,. ,. = n.z. = R.z. Mi m; +-'-' or M. m. +-'-' if R;< n; r; I I r; where n; is the number of animalscaught in the ;th sample, m; is the number of marked
animals caughtin the ith sample,z; is the number of animals caughtbefore the ith sample
that are not caught in the ith sample, but arecaptured in a later sample, r; is the number of
animalsreleased fromthe i th samplethat aresubsequently recaptured, R; is the number of
, animalsreleased with marksfrom the ith sample, N; represents the total number of
animalsin thepopulation just beforethe ith sample,and M; represents the total number of
. ·marked.animals in the population just beforethe ith sample.
. The samplingperiods for the Jolly-Seber estimates consisted of the1995-1998
and1996-1998 live-capture seasons combined with the1999 hair-capture session on the
Okefenokeeand Osceola study areas, respectively. Capture histories were generated
based on genotypesobtained fromhair and tissue samplescollected from live-captured
bearsand those identifiedat hair traps. ProgramJOLLY uses chi-squareanalyses to
identifyany variation in capture probabilities due to trapheterogeneity or trap response.
These goodness-of-fittests were used to aid in the selection of appropriate models.
Population Density
I calculated average density estimates for black bears on theOkefenokee and
Osceola study areasby dividing the population estimates by the sizes of the 1999 study
areas. Dividing the upperand lower limits of the population estimates by the sizes of the
study areasproduced 95% confidenceintervals for each density estimate.
48 CHAPTER IV
RESULTS
Trapping
Between June 1995 andSeptember 1998, 377 traps(Fig. 10) produced 6,357 trap nightsresulting in 208 captures of 123 individual black bears(75M: 48F) on the
Okefenokee study area (TableA. l). Overall, trapsuccess was 3.3%, or approximately 31 trapnights per capture (Table 3). The age of markedbears on the Okefenokeestudy area rangedfrom 1 to 12 years (Table A.1). The sex ratio of captured bears (75M: 48F) differedfrom 1: 1 ( Zios = 5 .93, 1 df, P =0.0 149).
On the Osceola study area,296 traps (Fig. 11) were set for5,12 0 trapnights from
June 1996 to September 1998. Project personnel captured 79 individual black bears
(49M: 30F) 132 times (Table B. l). Overall, trap success was 2.6%, or approximately 39 trapnights per capture(Table 3). The age of markedbears on the Osceola study area rangedfrom <1 to 13 years(Table B. l ). Like the Okefenokeestudy area, the sex ratio of captured bears(49M: 30F) differedfrom 1: 1 (z�.os = 4.57, 1 df, P= 0.0325).
Hair Trapping
From 12 June to 17 September 1999, 88 and94 barbedwire hair traps (Fig. 12) were maintained simultaneouslyon the Okefenokeeand Osceola study areas, respectively . . Eight hundred andeighty hair-trapsessions (88 trapsx 10 trapping occasions) were recorded on the Okefenokee study areain 1999. Overall, 435 bearvisits
49 GEORGIA
-- Okefenokee National Wildlife Refuge • Snaretrap locations N D studyarea boundary 10-- --0 10 10 Kilometers I Fig. 10. Locationsof snaretrapsites on theOkefenokee study area, Georgia, 1995- 1998.
50 Table3. Trapping summaries forthe Okef enokee (1995-1998) and Osceola(1 996-1998) study areas,__ Georgiaand Florida. GeorgiaStudy Area
Numberof Trap Numberof Bear % Rate of . # TrapNights Year Sites Nights Visits Escapes Captures Capture0 per Capture 1995 116 1,323 133 13 78 58.6 17.0 1996 93 1,581 57 2 32 56.1 49.4 1997 79 1,691 204 3 49 24.0 34.5 1998 89 1,762 410 7 49 12.0 36.0
Total 377 6,357 804 25 208 25.9 30.6
Florida Study Area
Numberof Trap .. Numberof Bear % Rate of # Trap Nights Year Sites Nights · Visits Escapes Captures Capture0 perCapture 1996 82 l,454 48 3 40 83.3 36.4 1997 101 1,829 107 3 47 43.9 38.9 1998 113 1,837 418 8 45 10.8 . 40.8
Total 296 5,120 573 14 132 23.0 38.8 • = % rateof capture definedas numberof bear captures dividedby numberof bear visits. FLORIDA
OkefeAokee National WildUfeRefuge
·--.. ..._. . ..._.. .._,. .
C
a Cl
Federal land holdings .._..__ __, , • Snare trap locations D study area bowidary / / Stateline / \, 10- 0- 10 20 Kilometers Fig. 11. Locations of snaretrapsites on the Osceola study area,Florida, 1996-1998.
52 -"'·
Georgia
...... � State and federal land holdings N • Hair trap locations D Study area boundary I' ,I \ / State line / \: t 10-- -0 10- 40 Kilometers Fig 12 . Locationsof barbed-wirehair trapson the Okefenokeeand Osceola study areas, Georgiaand Florida, 1999.
53 were documented, of which 374 (86%) produced �1 hair sampleper visit (Table 4). Of the 3 74 hair captures on the Okefenokeestudy area, 10 9 (29%) samplescontained �5 roots, making them candidatesfor microsatellite analysis. Nineteen bears (9M:10F) live captured from1995-1998 were detected at hair trapsduring the 1999 hair-trapping season. Hair trapping success was 42 .5% on the Okefenokeestudy area in 1999, averaging2.4 trap sessions per hair capture event.
On theOsceola study area,1, 034 hair trapsessions resulted in 742 bearvisits in
1999, of which 637 (86%) produced �1 hair sample (Table 4). Two hundred andseventy two (43%) of the 637 hair samplescollected contained �5 roots. Twenty three bears
(12M:11F) livecaptured from1996-1998 were detected at hairtraps during the1999 hair trappingseason. Hair trappingsuccess on the Osceola study site was 61.6%, averaging
1.6 trapsessions per capture event.
Microsatellite Analysis
Complete multi-locus microsatellite genotypes were obtained fromhair andtissue samplesfor 111 of 121 (92%) and 72 of79 (91 %) live-captured bears on the Okefenokee andOsceola study areas,respectively.
On the Okefenokeestudy area,79 hair sampleswere selected for microsatellite analysisfrom the 1999 hair-trappingseason. Complete multi-locus genotypes were obtained for78 (99%) of thosesamples, of which 39 individual bearswere identified. At each locus, 5-8 alleles were observed (Table 5) andaverage heterozygosity forthe 8 loci was 66.3% (n = 39) . . Microsatellite analysis resulted in 8, 3, 3, and5 bearsbeing
54 Table 4. Summariesof black bearhair trapping on theOkefenokee and Osceola s tudy areas,Georgia and Florida, 1999.
Georgia Study Area
Number of Trap Numberof Bear % Rate of Trap Sessions Trapline Sites Sessionsa Visits Captures Capture6 perCapture
Big Swamp 23 230 111 93 83.8 2.5 Ok. Sportsman 27 270 148 123 83.1 2.2 Jamestown 24 240 75 69 92.0 3.8 Craven' s Island 14 140 101 89 88.1 1.6 Total VI 88 880 435 . 374 86.0 2.4 VI
Florida Study Area
Numberof Trap Numberof Bear % Rate of Trap Sessions Trapline Sites Sessionsa Visits Captures Capture6 perCapture Banker's Trust 40 ·,, 440 354 295 0.83 1.5 BearBay 34 · · 374 233 205 0.88 1.8 Low Road 20 220 155 137 0.88 1.6
Total 94 1,034 742 637 0.86 ·1.6 • = Number of trap sessions refers to the number of hairtraps times the number ofperiods the hairtraps were activated. b = % rate of capture equals the number of bearca ptures divided by �e number of bearvisits. Table 5. Observedalleles andfrequencies for 39 black bears identifiedfrom barbed-wire hair traps for the Okefenokee study area, Georgia, 1999.
Locus Allele n Frequency Locus Allele n Frequency
GlOC 10 6a 0 0.000 GlOX 140 6 0.077 110 1 0.013 142 5 0.064 112 3 0.038 144 26 0.333 114 29 0.372 152 8 0.103 116 42 0.538 154 26 0.333 118 3 0.038 156 1 0.013 160 6 0.077
GIA 183 20 0.256 Gl0P 148 15 0.192 187 29 0.372 158a 0 0.000 189 5 0.064 160 4 0.051 191 9 0.115 162 47 0.603 193 15 0.192 166 7 0.09 168 5 0.064
GlOB 153 17 0.218 GlOL 133 2 0.026 155 9 0.115 135 1 0.013 157 10 0.128 137 29 0.372 159 5 0.064 143 1 0.013 161 30 0.385 149 12 0.154 165 5 0.064 151 8 0.103 167 2 0.026 153 22 0.282 155 3 0.038 165a 0 0.000
GlO 207 GlD M 7 0.09 176 38 0.487 211 4 0.051 178 5 0.064 213 23 0.295 180 8 0 0.000 215 26 0.333 184 15 0.192 217 5 0.064 186 10 0.128 219 13 0.167 188 8 0.103 190 2 0.026
a Indicatesalleles that were observedin bears from the Osceola (Florida) study area
56 identifiedat hair trapsin 1999 that were initially capturedin 1995, 1996, 1997, and 1998, respectively. On theOsceola study area, 88 hair sampleswere selected formicrosatellite analysis; complete genotypes were obtained for84 (96%) samples. Thirty-seven individual bearswere identifiedon the Osceola study area. At each locus, 4--8 alleles were observed (Table 6) andaverage heterozygosity was 67.9% (n = 37). Twelve, 5, and
6 bearsthat were initially captured in 1996, 1997, and1998 were identifiedat hair traps in 1999, respectively.
Probability of Identity
Based on thefrequency distribution of alleles at the8 microsatellite loci, the
Ploveran for39 individual bearssampled with hair trapson the Okefenokee study areawas
6.57 x 10-8 (Table 7). That correspondedto a chanceof 1 in 15,223,0 17 that 2 individuals drawn at randomfrom the Okefenokee population would sharean identical genotypeacross all loci. The overall Pisibswas estimated at 1.00 x 10-3 (n = 39) (Table
7), or 1 chancein 1,000 of encountering matching genotypes. Estimates of Pisibs for individual loci rangedfrom 0. 388 to 0.533. The sibling match test foreach 8-loci genotypewas Psib < 0.003. Cons�quently, all genotypesidentified from 1999 hair traps were included in the capture history data based on my criterion forinclusion (Psib � 0.05).
The Ploverall for37 individuals identifiedwith hair trapson the Osceola study area was 2.92 x 10-7 (Table 8), or 1 chancein 3,421,763 of randomlysampling 2 bears possessing identical genotypes. Pisibsacross thesame 8 loci was estimated at 2.00 x 10-3
(n = 37) (Table 8), approximating a l in 500 chanceof encounteringmatching genotypes in the Osceola bearpopulation. Pisibsestimates forindividual loci rangedfrom 0. 349 to
57 Table 6. Observedalleles andfrequencies for 37 black bearsidentified from barbed-wire hair trapsfor the Osceola study area,Florida, 1999.
Locus Allele n Frequency Locus Allele n Frequency
GlOC 10 6 1 0.014 Gl0X 140a 0 0.000 110a 0 0.000 142a 0 0.000 112 2 0.027 144 38 0.514 114 37 0.500 152 4 0.054 116 32 0.432 154 26 0.351 118 2 0.027 156 4 0.054 160 2 0.027
GlA 183 5 0.068 GlOP 148 11 0.149 187 31 0.419 158 1 0.014 189 4 0.054 160 5 0.068 191 20 0.270 162 57 0.770 193 14 0.189 166a 0 0.000 168a 0 0.000
GlOB 153 4 0.054 GlOL 133 11 0.149 155 12 0.162 135a 0 0.000 157 25 0.338 137 10 0.135 159 1 0.014 143 2 0.027 161 27 0.365 149 12 0.162 165 3 0.041 151 6 0.081 167 2 0.027 153 22 0.297 155 5 0.068 165 6 0.081
GlO GlD M 207 13 0.176 176 32 0.432 211 4 0.054 178 5 0.068 213 14 0.189 180 1 0.014 215 28 0.378 184 8 0.108 217 2 0.027 186 6 0.081 219 13 0.176 188 22 0.297 190a 0 0.000
a Indicatesalleles thatwere observedin bears fromthe Okefenokee(Georgia) study area
58 Table 7. Probability of identity estimates based on 39 individual black bears identified from barbed-wire hairtraps on the Okefenokeestudy area, Georgia, 1999. Number of Probability of Probability of Locus Alleles Identity Identity (siblings)
GlOC 5 0.269 0.533 GIA 5 0.109 0.406 GlOB 7 0.085 0.388 Gl0M 6 0.095 0.394 Gl0X 7 0.099 0.399 GlOP 5 0.211 0.510 GlOL 8 0.103 0.403 GlD 6 0.129 0.435 a 8 Overall 6.13 6.57 X 10- b 1.00 X 10-J b
· a Average number of alleles h Product of individual values
59 Table 8. Probability of identity estimates based on 37 individual black bears identified frombarbed-wire hair trapson the Osceola study area, Florida, 1999. Number of Probability of Probability of Locus Alleles Identity Identity (siblings)
GlOC 5 0.287 0.541 GIA 5 0.133 0.429 GlOB 7 0.124 0.421 Gl0M 6 0.096 0.396 Gl0X 5 0.225 0.50 3 GlOP 4 0.417 0.664 GlOL 8 0.051 0.349 GlD 6 0.135 0.433 a Overall 5.75 2.92 X 10-? b 2.00 X 10-J b a Average number of alleles b Product of individual values
60 0.664. The sibling match test foreach 8-loci genotype was Psib < 0.005, allowing me to include all observedgenotypes in the capture history data.
Hardy Weinberg and Linkage Disequilibrium
For the 39 individual bearsidentified at hair traps on the Okefenokeestudy areain
1999, the Hardy-Weinberg equilibrium test detected evidence of non-randommating for
2 loci (GlOM, P = 0.016; GlOP, P = 0.0003) at the 5% significancelevel. Only loci
G1 OP provided evidence of non-randommating, however, after applyingthe Bonferroni experimentwise error rate (Rice 1989, Sokal andRohlf 1995). On the Osceola study area, the Hardy-Weinbergtest detected no evidence of non-randommating amongthe 8 microsatellite loci examinedfor the 37 individual bearsthat were identifiedin 1999.
Linkage disequilibrium tests were used to identifypossible non-random associations between alleles of differentmicrosatellite loci. On the Okefenokeestudy area,4 loci pairs (GlOC vs. GlOB, P= 0.00014; GIA vs. GlOB, P= 0.00103; GIA vs.
GlD, P = 0.000; GlOX vs. GlD, P = 0.000) had probabilityvalues smaller thanthe comparison-wisesignificance level of 0.0018. Pairwisetests comparingthe 37 individual bearsidentified at hair traps on the Osceola study areain 1999 detected no associations between anyloci pairs.
Population Size and Density
Closed Models . . Capture histories were pooled for the firstand second halves of the1999 hair-trappingperiods. The modifiedLincoln-Petersen model produced a population estimate of 86 bearson the Okefenokeestudy area in 1999, fora density
61 estimate of 0.17 bears/km2 (Table 9). Because there was anodd number of total trapping periods on the Osceola study area(n = 11), 2 estimates were calculated using the Lincoln
Petersen model. The firstpooling configuration, 5 and 6 periods, resulted in anestimate of 53 bears. Conversely, dividing the 1999 hair-trappingseason into 6 and 5 periods produced anestimate of 47 bears. Resultantbear densities on the Osceola study area during1999 were 0.14 and0. 13 bears/km2 using the 5/6 and6/5 session pooling configurations, respectively(Table 10 ).
The equal catchability test described by Caughley (1977) indicated that observed capturefrequencies fromthe Okefenokeestudy areadiffered fromthe expected zero truncated Poisson distribution when all of thehair-captures from 1999 were considered
(z;.05 = 13.790, 1 df, P= 0.0002) (Table 11). Although the behavioral response test within ProgramCAPTURE produced nonsignificant results for the 10 session pooling arrangement, the probability value associated with this test was questionable ( z;05 =
3.533, 1 df, P= 0.060). Additional tests, however, detected individual heterogeneity amongcapture probabilities for all pooling configurations(n = 5) of the 1999 hair data.
Time variationwas not detected as a significantinfluence on captureprobabilities in any of the pooling arrangements. Based on the above, I gave further consideration only to models that allowed for variationin captureprobabilities as a result of behavioral response or individual heterogeneity.
To aid in selecting the most appropriate pooling configuration for my analysis, I used the population closure test in ProgramCAPTURE . . The tests for closure detected lack of closure when 10 (Z= -3.451, P= 0.00028) and 5 (Z= -2.485, P= 0.00649)
62 Table 9. Estimatedbl ack bearpopulation size from haircaptures using multiplemark-recapture models in the Okefenokee study area, Georgia, 1999. Population Coefficient of 95% Confidence Density 95% Model Confidence Size Estimate Variation (%) Interval (bears/km2) Interval
Mod. Lincoln-Petersen 86 28 39-133 0.17 0.08-0.26 .· Mo 84 26 57-151 0.16 0.1 1--0.30 Mh8 71 11 59-91 0.14 0.12--0.18 Mt 84 26 55-148 0.16 0.1 1--0.29 Mb 117 118 47-849 0.23 0.09-1.66 Mbh 117 119 47-851 0.23 0.09-1.67 b Mtb Chao Mb 175 48 84-452 0.34 0.16--0.88 Chao Mt 110 38 64-243 0.22 0.13--0.48 Chao Mth 292 88 88-1,357 0.57 0.17-2.66 a Indicates selectedmodel b Population size estimate and associated standard errors were impossibly large Table 10. Estimatedbla ck bearpopulation size from hair captures using multiple mark-recapture models in the Osceolastudy area, Florida, 1999. Population Coefficientof 95% Confidence Density 95% Confidence 2 Model Size Variation(%) Interval (bears/km ) Interval Estimate
Mod. Lincoln-Petersen (5-6)8 53 16 36-70 0.14 0.10---0.19 Mod. Lincoln-Petersen (6-5t 47 12 36-58 0.13 0.10---0.16
Moc 44 9 40-57 0.12 0.1 1--0. 16 Mb 50 12 43--66 0.14 0.12--0.18 Mt 44 9 40-56 0. 12 0. 1 1--0.15 Mb 48 21 40-87 0.13 0.1 1--0.24 Mbh 48 21 40-87 0.13 0.1 1--0.24 Mtb 47 35 39-130 0. 13 0.1 1--0.36 Chao Mb 48 15 41-71 0.13 0.1 1--0. 19 Chao Mt 45 12 40-63 0.12 0. 1 1--0.17 Chao Mth 47 16 40-73 0. 13 0. 1 1--0.20 a Two sampling sessions of 5 and6 periodseach b Two samplingsessions of 6 and 5 periodseach c Indicates selectedmodel Table 11. Observedcapture frequenciesof bearsidentified at hair traps on the Okefenokeestudy area andthe zero-truncatedPoisson frequenciesto be expected if catchability is constant(1999). Number of Number of Expected [f-E(r)f times captured individuals frequencies E(r) (i) (j) E(j) 1 27 15.785 7.967 2 5 12.628 4.608 3 3 6.735 4 1 2.694 5 2 0.862 6 0 0.230 7 0 0.053 8 0 0.011 9 0 0.002 10 0 7 0.000 10.586a 1.215 11 0 0.000 12 0 0.000 13 0 0.000 14 0 0.000 15 0 0.000 16 0 0.000 17 0 0.000 18 1 0.000 39 39.000 z2 =13.790 df= 1 P=0.0002
a Capture frequencies �3 were pooled so that expected frequencieswould be �5 (Caughley1977).
65 session pooling configurations were considered. Additionally, lack of closure was detected for1 of the 3-session (3-3-4) pooling configurations(Z= -1.732, P= 0.0416).
The population closure test failedto detect a lack of closure, however, forthe 4--3-3 (Z=
-1.414, P= 0.0787) or 3-4-3 (Z= -1.581, P= 0.0569) session pooling configurations.
Unfortunately, collapsing samplingoccasions results in the exclusion of hair samples fromanalysis due to individual bearsbeing observed multiple times at hair traps within sessions. The resultantdecrease in samplesize, however, tends to result in higheroverall capture probabilities, therebyimproving model performance. Although41 % (n = 32) of the79 samples were excluded frommy analysis using the 4--3-3 arrangement, capture probabilities were higherthan for anyother pooling configuration. Furthermore, estimates provided from the4--3-3 arrangementproduced thelowest standarderrors of all models analyzed. Therefore,I selected the3-session pooling configurationthat divided capture histories into samplingperiods of 36, 27, and27 days each (Fig. 8).
Multiple mark-recapture models producedpopulation estimates that rangedfrom
71-292 bearson the Okefenokeestudy areain 1999 (Table 9). The jackknife heterogeneity model Mh produced a population estimate of 71 bearsduring the 1999 hair
2 trapping season, corresponding to a density of 9.14 bears/km •.Model Chao Mh estimated
population size at 175 bears, or 0.34 bears/km2 on the Okefenokeestudy area.
Considering the possibility of a behavioral response bias in capture probabilities, model
Mb produced a population estimate of 117 bearsduring the 1999 hair-trapping season, or
2 0.23 bears/km • The goodness-of-fittest forthe individual heterogeneity models did not
66 indicate a poor fit( z�.os = o.o5k, 2 df, P= 0.973), whereas a poor fit was indicated for model Mb (z�.os = 7.356,.2 df, P= 0.0253).
On the Osceola study �a, the equal catchabilitytest (Caughley 1977) indicated that theobserved capture frequ�ncies from 1999 did not differfrom the expected zero- truncatedPoisson distributionwhen all hair-captureswere considered ( z�.os = 1.457, 2 df,P = 0.483) {Table 12 ). Furthermore,chi-square goodness-of -fittests indicated that . I captureprobabilities were not significantlyinfluenced by trap heterogeneity, behavioral response, or time variation for ty pooling configurations(n = 4). The population closure test in ProgramCAPTURE failedto detect a lack of closure forall pooling configurations (P= 0.06-0.60). The largestprobability value was associated with the pooling configurationthat colla�sed capturehistories into 5 samplingsessions (Z=
0.262, P= 0.60337). Furthermre,only 19 (2 3%) of the 84 hairsamples were excluded fromanalysis due to multiple observationswithin a session. Therefore, I selected the 5- session pooling configurationt�at divided capturehistories into 4 samplingperiods of 18 days with the fifthsession lasting 27 days (Fig. 9).
Population estimates rangedfrom 44--50 bears on the Osceola study area using the multiple mark-recapture models within ProgramCAPTURE (Table 10). The null model Mo produced a populati+ estimate of 44 bearsduring the 1999 hair-trapping
2 period, fora density estimate ofi0. 14 bears/km • The largest estimate was provided by model Mb, the heterogeneity mldel, which estimated population size on the Osceola study area at 50 bears,or 0.16 lears/kni2. The goodness-of-fittests did not indicate a
67 Table 12. Observedcapture frequenciesof bears identifiedat hair trapson the Osceola study area and the zero-truncatedPoisson frequenciesto be expected if catchability is constant(1999). Numberof Nwnber of Expected 1/-E(r)f times captured individuals frequencies E(r) (i) (j) E(j) 1 14 11.999 0.334 2 13 11.675 0.150 3 5 7.573 0.874 4 3 3.684 5 0 1.434 6 1 0.465 7 0 5 0.129 5.7533 0.098 8 0 0.031 9 0 0.007 10 0 0.001 11 1 0.000 37 37.000 z2 =1.457 df= 2 P=0.4 83 a Capturefrequencies �4 were pooled so that expected frequencieswould be �5 (Caughley1977).
68 poor fit forthe individual heterogeneity, behavior response, or time variation models.
Op en Models. Estimates produced fromJolly-Seber models using stratified data indicated that captureprobabilities were higher when capturedata were pooled. In addition, differencesin population size estimates were negligiblebe tween pooled and stratifiedmodels. If heterogeneity in capture probabilities was present, it appearsthat increasing sample sizesmade theJolly-Seber models more robust to thatviolation.
Therefore,I chose to pool sex and age classes forthe 1995-1998 livetrappingdata.
The equal catchability test described by Caughley (1977) indicated that the observed capture frequenciesdiffered from a zero-truncatedPoisson distribution ( z;_05 =
17.337, 2 df, P = 0.000129) on theOkefenokee study area, but not (z;_05 = 7.092, 3 df, P
= 0.06902) on theOsceola study area (Tables 13 and14). Models A, A', B, andD producedestimates of population size forboth study areas. However, chi-square
goodness-of-fittests indicated thatmodel A', the deaths only model, did not provide a
good fitfor the Okefenokee data (P = 0.0004). Inaddition, captureprobabilities for model A' (p= 0.28) were low in comparisonwith models A, B, andD (p= 0.48-0.52).
Therefore,only models A, B, andD were given furtherconsideration. Model A produced
a mean population estimate of 68 bears,corresponding to a density of 0.13 bears/km2 on
theOkefenokee study area(Table 15). Models B andD produced meanestimates of 73
2 and 77 bears,or 0.14 and0. 15 bears/km , respectively {Table 15).
On theOsceola study area,Jolly-Seber models produced meanpopulation
estimates that rangedfrom 90 -114 bears {Table 15). Individualtests between models
indicated that allowing survivaland capture probabilities to varydid not provide a better
69 Table. 13. Observed capture frequenciesof bears identifiedat snares andhair trapson theOkefenokee study area andthe zero-truncatedPoisson frequencies to be expected if catchability is constant(1995-1999). Number of Number of Expected [f-E(J)f times captured individuals frequencies E(r) (i) (j) E(j) 1 82 61.991 6.458 2 28 46.184 . 7.159 3 16 22.938 4 9 8.544 5 7 2.546 6 0 0.632 7 0 0.135 8 0 0.025 9 0 0.004 10 0 17 0.001 11.887a 2.119 11 0 0.000 12 0 0.000 13 0 0.000 14 0 0.000 15 0 0.000 16 0 0.000 17 0 0.000 18 1 0.000 143 143.000 z2 = 17.915 df= 2 P=0. 000129 a Capture frequencies�4 were pooled so thatexpected frequencieswould be �5 (Caughley1977).
70 Table 14. Observedcapture frequencies of bearsidentified at hair traps on the Osceola study area andthe zero-truncated Poisson frequencies to be expected if catchability is constant(1999). Number of Number of Expected [r - E(r)fl times captured individuals frequencies E(r) (i) (j) E(j) 1 35 28.638 1.413 2 32 28.996 0.311 3 11 19.572 3.754 4 7 9.908 0.854 5 2 4.013 6 2 1.354 7 2 0.392 a 8 0 8 0.099 5.886 0.759 9 1 0.022 10 0 0.005 11 0 0.001 12 1 0.000 93 93.00 z2 = 7.092 df= 3 P=0. 06902 a Capturefrequencies �5 were pooled so that expected frequencieswould be �5 (Caughley1977).
71 Table 15. Estimatedblack bearpopulation size from theJolly-Seber models in Okefenokee (OKE) and Osceola(OSC) study areasusing a combination of live-capture and hair-trapping data, Georgiaand Florida, 1995-1999. b c,r d Study Model Aa,e Model A' Model a Model D Area Year N (95% CI) N (95% CI) N (95% CI) N (95% CI) OKE 1995 ------304 (209--398) OKE 1996 58 (33-83) 153 (104-201) 79 (53-105) 63 (46-81) OKE 1997 82 (44-1 19) 137 (93-181) 84 (57-1 10) 79 (57-101) ' . OKE 1998 64 (41-87) 110 (72-148) 64 (48-79) ·:- 86 (62-1 11) OKE 1999 ------65 (40-89) 81 (55-107)
OKE Mean 68 (50-85) 176 (144-207) 73 (41-104) 77 (41-1 13)
osc 1996 ------117 (96-138) osc 1997 102 (54-149) 128 (92-164) 95 (64-125) 98 (66-131) osc 1998 77 (44-1 11) 95 (63-127) 92 (58-127) 89 (59--119) osc 1999 ------99 (51-141) 91 (55-127) osc Mean 90 (61-1 19) 114 (96-131) 95 (38-153) 93 (43-143) a Model A allows forsurvival and capture probabilitiesto vary b Model A' is forcases where death is allowedbut immigrationis not c Model B asswnes survival is constantover the entire study d Model D asswnesboth survival and capture probabilities are constant overthe entire study e Indicates selectedJolly-Seber model for the Okefenokee capture data (1995-1999) r Indicates selectedJolly-Seber model forthe Osceolacapture data (1996-1999) fitto my data. Consequently, only models B andD were givenfurther consideration.
Model B, which is based on theassumptions of constantsurvival and time-specific capture probabilities, produced a meanpopulation estimate of 95 bearson the Osceola study area, resulting in a density estimate of 0.26 bears/km2 {Table 15). Model D, the constantsurvival and capture probability model, provided a meanpopulation estimate of
93 bears, or 0.25 bears/km2 {Table 15). Meancapture probabilities for models D andB were 0.40 and0. 41, respectively.
73 CHAPTER V
DISCUSSION
PopulationSize
The accuracyof estimates provided by open andclosed models is largely dependent on whetherrequisite assumptions are met during sampling efforts.
Furthermore,failure to identifymodel violations tends to result in estimates that are biased. Samplingthe Okefenokeeand Osceola study areasin 1999 using the hair trappingtechnique had several advantagesthat enabled me to minimize model violations andreduce potential biases in the estimates of population size. First, I was able to trap each of the entire study areasat thesame time. Doing so decreased anytemporal or spatial variationin capture probabilities because a largeportion of the bearpopulation was concurrentlysampled. During the 1995-1998 livetrappingseasons, only a minimal number of snares(15- 25) could be employed at a time due to logistical constraints( e. g., traveltime, remote access, safetyrisks forcaptured bears). Furthermore, the non obtrusivenature of thehair-trapp ing techniqueprobably decreased thelikelihood of a trap response bias. Infact, the proportion of bearvisits thatresulted in the collection of�l useable hair sample(�5 hairs) was 25% and37% on the Okefenokee andOsceola study areas, respectively. During the 1998 live trapping season only 49 (12%) of 410 bear visits resulted in captures on the Okefenokee study area; likewise, capture success on the
Osceola study area was 11 % in 1998. Althoughthe observed increase in capturesuccess does not indicate the absence of a trapresponse bias, it does suggest that the effectsof trapavoidance were lessened using the hair-trapping technique.
74 Population closure also is animportant assumption of the closed models thatI used to provide within-yearestimates. Fortunately, the closure assumption can be met if the interval between samplesis short (Otis et al. 1978). In this study intervals between hair collections were approximately9 days; the entire samplingduration lasted only 4 months fromJune through September. Naturally, no reproduction occurred during that period. Furthermore, I documented only 5 bear mortalities during June through
September sampling periods from 1995-1998 (J. D. Clark, U.S. GeologicalSurvey, unpublisheddata). Therefore, the assumption of demographicclosure probably was not violated during the1999 hair-trappingperiod.
Whether therequirements of geographicclosure wereupheld forboth study areas, however, is questionable. Although the closure test in ProgramCAPTURE rejected the null hypothesisof population closure in all of thepooling configurations of the Osceola hair-capture data, only two 3-session pooling arrangements met theclosure assumption in the Okefenokee study area(P4-3-3 = 0.08 andP3-4-3 = 0.06). However, the performance of the closure test lacks power when capture probabilitiesare affectedby behavioral variation(Otis et al. 1978). Considering the significant results of Caughley's(1977) test forequal catchability, theclosure tests for the Okefenokee data were probably influenced by a behavioral response. If thatvariation was extreme, however, I would expect a significantresult fromProgram CAPTURE' goodness s -of-fittest, which did not occur.
Althoughcapture rates declined markedlyon both study areaswith each yearof live trapping,the number of markedand unmarked bears caught each yearwere relatively stable. Therefore,I suspect that influenceof anybehavioral bias on thewithin-year estimates for theOkefenokee study areawas negated by theintensive samplingdesign.
75 In addition to population closure, the mark-recapturemodels that I used to provide within-yearestimates of population size require that marks arenot lost andevery markedindividual is correctlyidentified. Based on the low average probability of identity estimates forbears in theOkefenokee-Osceola population, each observed genotypeshould represent a single individual. Consequently, it is highly unlikely that markedand unmarked bears were incorrectlyidentified as such in thecapture history
data.
Most mammal studies have detected variation in capture probabilities (Young et
al. 1952, Carothers 1973, Seber 1982) . . The statistical challenge using closed mark
recapture models, therefore, is not to achieve equal catchability, but to select the
appropriate model forthe data (Mowat and Strobeck2000). I used the model selection
procedure in ProgramCAPTURE to assist in my selection of themodel that best fitted
my data. Simulation studies of populations of known size have indicated that this
selection algorithmdoes not always select themost appropriate model (Menkins and
Anderson 1988, Manning et al. 1995). Consequently, I used the selection procedureonly
as an�d in selecting the most appropriatemodel formy data by identifyingsignificant
sources of variation in captureprobabilities.
On the Okefenokee study area, the population estimate of 86 bears produced by
themodified Lincoln-Petersen model fellwithin the 95% confidenceintervals of all the
multiple mark-recapturemodels. The Lincoln-Petersen estimator may be biased,
however, as suggested by thesmall number of recapturesin thesecond sample(n = 5).
Althoughthe equal catchability test (Caughley 1977) indicated a behavioral response bias
in the hair-trappingdata, I would not expect trap shy behavior using this sampling
76 technique. In addition, the uniformlyrandom subsampling designthat I employed appearsto have minimized time variationin capture probabilities. Therefore,I suspect that individual capture heterogeneity was the cause of the low recapturerate associated with theLincoln-Petersen model.
ProgramCAPTURE's chi-squaregoodness-of -fittests identifiedsignificant patternsof variation in captureprobabilities formany of the data pooling configurationsI considered. The most obvious patternwas a consistent detection of individual heterogeneity in capture probabilities forall pooling arrangements of theOkefenokee data. Therefore,models that allow forindividual captureheterogeneity were given primaryconsideration. The population estimate of 175 bearsproduced by model Chao
Mh was the second highestof all multiple mark-recapture models considered. However, simulation results indicate that model Chao Mh tends to overestimate in the presence of weakheterogeneity (Mowat andStrobeck 2000). The jackknife heterogeneity model Mh produced a population estimate of 71 bearson the Okefenokeestudy area. Although model Chao Mth also considered heterogeneity, I disregardedit andall other time variationmodels (Mt, Chao Mt, andMt b) because there was no indication of a temporal influenceon captureprobabilities forany pooling configurationsof the Okefenokee hair trappingdata.
Behavioral variationwas only detected by the equal catchability test (Caughley
1977) when the fullcapture history matrix from1999 was considered. Program
CAPTURE did not detect a behavioral response, however, when samplingoccasions
were �10. Furthermore,the probability of rejecting the null hypothesis(i.e., no
behavioral response exists) increased as the number of samplingoccasions decreased. It
77 appearsthat the increase in samplesize within each samplingoccasion andthe increase in capture probabilities as a result of data pooling minimized any potential behavioral
effects. Consequently, the behavioral models Mb andMbh were not suitable forthe 3- session pooling configurationthat was selected forthe Okefenokeehair data.
Considering all of the closed models, I conclude that the within-yearestimate of71 bears
(95% CI = 59-91) produced by the jackknife heterogeneity model Mh is the most appropriate forthe Okefenokeehair-trapping data.
The Jolly-Seber model A produced a meanpopulation estimate of 68 bears (95%
CI = 50-85) on the Okefenokeestudy area basedon 4 yearsof live-capture data and 1 yearof hair-trapping data. With the exception of the deaths only model A', all models
exhibited similarannual trends and produced meanestimates that differedby only 9
bears. Estimates frommodel A' were seriously elevated andlacked precision relative to
other models. The deaths only model also was probably biaseddue to bears moving out
of ONWRand onto the study area. Althoughthe assumptionsof model B (constant
survival) andmodel D (constant survival and capture probabilities) probably were not
met during this study, highcapture probabilities made these models robust to model
violations. However, 52 bear mortalities were documented during the 4 years oflive
trapping on the Okefenokeestudy area; 88% (n = 46) of mortalities were due to hunting.
Considering that 123 individuals were captured during that period, there appearsto be a
high rateof population turnoveron the Okefenokee study area. Therefore, the general
Jolly-Seber model A, which accountsfor variation in survivalprobabilities, was the more
appropriate open model.
78 Unlike the Okefenokeestudy area, capture probabilities for the Osceola hair trappingdata did not appearto be influencedby individual heterogeneity. Because time andbehavioral variation also were not detected, it is not surprising that Program
CAPTUREselected the null model M0 as most appropriate. Althoughthe assumption of equal catchability is typicallyconsidered unrealistic in most mark-recaptureexperiments
(Otis et al. 1978, White et al. 1982, Pollock et al. 1990), the 1999 samplingperiod on the
Osceola study area appears to be anuncommon situation where this assumption was upheld. Fortunately, selecting the most appropriate closed model for theOsceola data became less critical because the highest andlowest estimates differedby only 9 bears
(Table 10). Based on the absence of heterogeneity in capture probabilities, however, I selected theestimate of 44 bearsprovided by the null model Mo as most appropriate.
The Jolly-Seber models B andD produced meanpopulation estimates of 95 (95%
CI= 38-153) and93 (95% CI= 43-143) bearson the Osceola study area, respectively.
Because capture probabilities were differentbetween the 2 samplingtechniques, however, I selected against model D for the Osceola data. Therefore, I chose model B, which is based on the assumption of time-specificcapture probabilities, as the most appropriateopen model.
Althoughestimates provided by Jolly-Seber models B andD differed by only 2 bears, they were much higherthan the within-yearestimate of 44 provided by the closed model Mo. I suspect that the discrepancybetween the open andclosed estimators is the result of a combination of fa ctors. Firstly, it appears that the Jolly-Seber models using only the 1996-1998 live-trapping data performedpoorly. In addition to relatively low capture probabilities, estimates forall Jolly-Seber models were significantlyhigher ( N=
79 112-143) relative to thoseof themultiple mark-recapturemodels (N= 44-50).
Furthermore, all survivalestimates produced from 3 yearsof live-capture data exceeded
1.20, animpossibility. Unfortunately, thepoor performance of the Jolly-Seber models seems to be theresult of anunexplainable lack of recaptures fromthe 1997 trapping season. Of the 26 bearsthat were initially markedin 1997, only 5 were recapturedin
1998; 15 bears from1996 were caughtin 1998. Likewise, only 5 bearsthat were initially markedin 1997 were identifiedat hair traps in 1999, whereas 12 bearsfrom 1996 were observedin 1999. It appearsthat the observed positive bias is most likely due to an inability to recapturesome markedindividuals, which will always result in an overestimation of population size (Pollock et al. 1990). Althoughthe performanceof the
Jolly-Seber models was improved by including a fourth year of capturedata, estimates remained elevated.
Inaddition to model performance, I suspect thatthe discrepancybetween the closed andopen model estimates canalso be explained largelybased on several biologicalfactors. From 1996--1998, only 1 bearmortality was documented on the
Osceola study area;that bearwas founddead in a trap. In addition, 23 (2 9%) of the 79 individual bears that were live-trappedfrom 1996--1998 were identifiedat hair traps in
1999. The relatively high proportion of bearsremaining on the study areathroughout the
duration of the study suggests that,unlike the Okefenokee study area, population turnover
is low forOsceola bears. The low turnover is not surprisingconsidering there is no
significantmortality factor(i.e. hunting) to remove bears fromthis portion of the
Okefenokee-Osceolabear population. Consequently, I believe that the within-year estimates provided by multiple mark-recapturemodels, althoughlower thanthose
80 provided by the open Jolly-Seber models, are more appropriate forthe Osceola study area.
Considering all data pooling configurations andmodel types used in this analysis,
I conclude thatthe most reliable estimates of population size were obtained fromclosed models using the 1999 hair-trapping data. For the Okefenokeestudy area, I select the estimate of 71 bears(95% CI= 59-91) obtained fromthe individual heterogeneity model
Mh as the most appropriate formy data. I selected model Mh over the Jolly-Seber model
A because theclosed model required fewer assumptions andarrived at approximately the sameestimate using only 2 parameters. Because fewerparameters were involved, precision was likely increased as a result of individual model assumptions being met. I conclude that the estimate produced by the null model Mo of 44 bears(95% CI= 40--57) is the most appropriate forthe Osceola study area. Althoughall Jolly-Seber estimates appearedto be biased high,the degreeof that bias is unknown. Therefore,it is possible that the number of bearsthat were on the Osceola study in 1999 areamay have been closer to the upper rangeof the 95% confidenceinterval for model M0•
Population Density
The estimated densities of black bears on the Okefenokeeand Osceola study areas
2 were 0.14 and0. 12 bears/km , respectively. Those estimates fellwithin the rangeof densities reportedfor other black bear populations throughoutthe southeastern US (Table
16). Direct comparisonsof population densities between areas,however, should be made with caution because of differencesin spatial extent, samplingmethodology, andhabitat
81 Table 16. Population densitiesof black bearsin thesoutheastern United States.
Locality Bears /km2 Reference
OkefenokeeSwamp, Georgia 0.14 This study OsceolaNational Forest, Florida 0.12 This study White RiverNWR, Arkansas 0.29 Smith1985 White Rock, Arkansas 0.08 Clark1991 Dry Creek, Arkansas 0.09 Clark1991 Deltic, Tensas RiverBasin, Louisiana 1.43 Beausoleil 1999 Tensas RiverNWR, Louisiana 0.35 Boersen2001 Alligator River NWR, NorthCar olina 0.86 _Allen1999 Big Pocosin,North Carolina 0.53 Martorello1998 Gum Swamp,North Carolina 1.35 Martorello1998 Camp Lejeune, NorthCarolina 0.02 Brandenburg1996
Great Dismal Swamp,North Car olina-Virginia 0.47--0.68 Hellgrenand Vaughan 1989
GreatSmoky MountainsNP, Tennessee 0.87 J. Chadwick, Universityof Tennessee,unpu blishedreport quality. Although density estimates were similarbetween my study areas, significant differencesin habitat use andland use practices existed that influencedthose estimates.
The Okefenokeestudy areawas situated adjacent to andincorporated a portion of
1,580 km2 of federally protected habitat (ONWR)where hunting was prohibited and public access was virtually nonexistent. Consequently, ONWR may serve as an importantsource of bears to fuelthe high population turnovercaused by hunting mortality in the counties surroundingthe Refuge. Conversely, the protectedstatus of black bearson theOsceola study areamay be creating a situation in which dispersal is becoming anincreasingly importantfactor in population regulation. From 1997-1998,
18 (43%) of the 42 new bearsthat were live-capturedon theOsceola study areawere subadult males. However, only 3 of those subadults were ever capturedagain; 1 bear from1997 was recaptured in 1998 andthe other 2 were identifiedat hairtraps in 1999.
Those low recapture rates suggest thatsubadult males may be dispersing fromthe
Osceola area at a relatively fast pace. In fact,2 bears that were initially caught on the
Osceola study area in 1996 were harvestedon the Okefenokeestudy area in 1996 and
1999 (J. D. Clark, U.S. GeologicalSurvey, unpublished data). Ineach of those cases, bearswere subadult males that had traveledat least 50 kmfrom their last capture location in ONF.
Based on the above, the density estimates forthe Okefenokeeand Osceola study areasappear to be influencedby completely differentbiological factors. Population turnover appearsto be low on the Osceola study area, therefore competition forresources may be resulting in adult bears forcingyounger males fromthe area. Conversely, population size anddensity of bears on the Okefenokee study area appearto be primarily
83 balancedby hunting mortality andimmigration. Because density estimates arealmost the samefor each of the study areas, however, theaverage between the Okefenokee and
Osceola densities may provide a roughestimate of population size for bears in this area.
2 Basedon the average density of0.135 bears/km , approximately 830 bears (95% CI=
70 7-1,045) may inhabit the6,147-km 2 Okefenokee-Osceolaecosystem.
Genetic Considerations
Inmark-recapture studies it is imperative that individuals be correctlyclassified.
The Pisibsstatistic estimated theprobability of randomlydrawing 2 bearswith identical
3 genotypes fromthe Okefenokee population at 1.00 x 10- , or 1 chancein 1,000. The
Pisibs for Osceola bearsindicated a 1 in 500 chanceof encountering matching genotypes across all 8 loci. Therefore, there was a highprobability thatmultilocus genotypes identifiedat hair traps on �2 occasions were correctlyclassified as recaptures of an individual bear. That conclusion was supported by the results of the sibling match tests.
Those tests concluded that10 0% (n =162) of thehair samplesthat were analyzedpassed my threshold of Psib<0.05 andrepresented 76 individual black bears,_ 42 of which were identifiedfrom the 1995-1998 live-trapping data.
Eight-locusgenotypes were obtained for 39 and 37 black bears on theOkefenokee and Osceola study areas,respectively (Tables 5 and6). AlthoughOkefenokee and
Osceola bears seem to have roughly thesame average number of alleles (6.13 vs. 5.75), noticeable differenceswere observedfor 3 individual microsatellite loci. Bears from the
Okefenokeestudy areadisplayed 2 unique alleles at loci GlOX and a1· op on 11 and 12 separateoccasions. Likewise, 6 bears from the Osceola areaexhibited 1 allele at loci
84 G1 0L that was never observedamong Okefenokeebears. Despite therelatively infrequent occurrenceof thosealleles, theirpresence indicates genetic variabilitywithin this bear population. In addition, the significantresults of theHardy-Weinberg equilibrium test forlocus G1 OP amongOkefenokee bears appearsto be the result of a select fewindividual bearspossessing alleles at this locus that areuncommon to most other bearswithin the Okefenokee study area. It is possible that theOkefenokee-Osceola ecosystem is largeenough andsufficient genetic variationexists to have bearsthat are genetically distinct fromone another(T. L. King, U.S. GeologicalSurvey, personal communication). · Consideringthat population turnover was highon the Okefenokee study area,it is likely that some bearsidentified at hair trapsin 1999 were immigrants fromother areaswithin ONWR. If sufficientgenetic variationexisted between those bears andbears on the Okefenokee study area,that could explained the observationof relatively uncommonalleles at certain loci.
Average heterozygositywas calculated as a measureof variationand found to be
66.3% and 67.9% forthe Okefenokee andOsceola bearpopulations. Those estimates may be relatively highin comparisonto other black bearpopulations inthe southeastern
United States. Throughout much of the Southeast, habitat loss andfragmentation have been so extensive that gene flowmay be restricted, resulting in geographically and geneticallyisolated populations. For example,average heterozygosity of black bears inhabiting the largelyisolated Tensas RiverNational WildlifeRefuge in northeast
Louisianawas documented at 47. 7% across the same8 loci used in thisstudy (Boersen
2001). When genotypes were extended to 12 loci for36 individualsfrom the Tensas
NWR study, average heterozygosity increased to 57 .6%. That estimate, however,
85 remains approximately 10% lower thanthose obtained forbears in the Okefenokee
Osceola population using only 8 loci. Clearly,the effects of geographic isolation as a result of habitat loss andfragmentation play animportant role in population genetics. . In this study, population size anddegree of habitat isolation did not appearto be a detriment to gene flowor genetic variability. Consequently, microsatellite DNA analysis using8 multilocus genotypes was sufficient forindividual identification.
86 CHAPTER VI
MANAGEMENT IMPLICATIONS
General
Althoughthe estimated densities of bearswere similarbetween the Okefenokee and Osceola study areas, these 2 areasmay representopposite ends of the spectrumin relation to population turnover. Turnover appears to be quite high on theOkefenokee study areaas a result of losses fromhunting mortality andadditions by recruitmentand immigration. Although the number of bearmortalities (n = 72) that was documented from1995-1999 was high relative to thenumber of individual bearsidentified (n = 143), the proportion of unmarkedbears captured each yearwas always >50%. The constant influxof new bearssuggests thateither reproduction is very high for Okefenokeebears, or ONWRmay be serving as animportant source for areasoutside of refugeboundaries.
I suggest that thelatter is themore probable explanation. Black bearreproduction on the
Okefenokeestudy area was heavily dependent on the swamp'sblack gum crop in the fall.
Twelve of 13 radio-collaredfemales fai led to reproducein 1996 followinga failureof a black gum crop in 1995. Incontrast, researchers observed anunusually abundant crop of black gumthe followingyear that resulted in significantlyhigher reproductive success;
26 of31 radio-collaredfemales were observed with cubs in1997 (J. D. Clark, U.S.
Geological Survey,un published data). When fieldworkwas completed in 1999, the majority of female bears on the Okefenokeearea were still in reproductivesynchrony from1997.
87 Based on theabove, population regulation on the Okefenokee area appears to be primarily governedby hunting mortality, emigrationby bears out of ONWR, andblack gum productivity. Because biologists have little control over the 2 latter factors, most management decisions concerningbears surrounding the OkefenokeeSwamp involve harvest regulations. The Black BearManagement Plan for Georgia ( Georgia Department ofNatural Resources [GDNR] 1999) calls fora maximumharvest rate of20% with females comprising no more than50% of the harvest. Since 1984, the annualbear harvest for the 5 counties contiguous with the OkefenokeeSwamp has averaged 35 bears.
By extrapolatingmy density estimate of 0.14 bears/km2 across the 1,580 km2 ONWR,
20% of this population would be approximately 44 bears. Although actual harvest numbers arelower thanmy estimate, I would caution against increasing bearhunting opportunities to compensate fora differenceof 9 bears. Annual harvestnumbers will
fluctuatearound the 20% level, but there will undoubtedly be an occasion of black gum
fai lure anda significantly higher proportion of bearswill be outside of theswamp and
available forharvest. Because black bears have such low reproductive potential (Pelton
1982), an excessively largeharvest of femalescould depress bears numbers for yearsto
come.
Thesituation on theOsceola study areais a marked contrast fromthat mentioned
above, primarilyas a result of the protected status of bearsin the State of Florida. In the
absenceof a hunting season, bear numbers on the Osceola areamay be approaching some
biologicalor social thresholdthat is resulting in high rates of dispersalfor subadult
males. Unfortunately,identifying density-dependent factors that promote subadult
dispersalis a difficulttask. Nevertheless, futuremanagement effortsshould be oriented
88 towardsidentifying and protecting bearhabitat outside of ONF. In particular, special emphasis should be placed on maintainingand improving habitat connections between
ONF andONWR. We documented bearstraveling from Florida into ONWR on 2 occasions, so exchangebetween the 2 areasis possible andlikely occurs more oftenthan expected. Althoughthe 2 study areasare separated by approximately 50 km,the distance between thecenter of theOsceola study areaand the southernedge of the Refuge is only
18 km. Consequently, the scarcity of data supporting interchangefrom Florida to the
Okefenokee Swampmay exist because dispersing males are establishing territories in the southernhalf of ONWR. If that is the case, thenfuture research needs to involve radio collaringmale bears fromthe Osceola study area,especially subadults, to document dispersalrates, movement patterns,and survival. The resultantdata could thenbe used identifyhabitat thatis serving as linkage zones between the Osceola and Okefenokee areas.
My study provided thefirst estimates of population size and density forbears in the Okefenokee and Osceola ecosystem using a rigorous mark-recaptureexperiment.
Althoughthe Okefenokee-Osceola bearpopulation is relatively largeand does not appear to be threatened with extinction, the long-termpersistence of other Florida black bear populations is questionable. Habitat loss and fragmentationand human encroachment are resulting in populations that are becoming increasingly isolated fromother bear populations. Of the 7 recognized Florida blackbear populations, the USFWS has concluded thatonly theApalachicola NF, Ocala NF, Big CypressNational Preserve,and
Okefenokee-Osceola ecosystem populations areviable (Bentzien 1998). In contrast,the
Chassahowitzkabear population, located on the centralGulf Coast of Florida, may
89 contain <20 individuals (Bentzien 1998). For the smaller, more isolated populations to persist into the foreseeablefuture, it may be necessaryto augmentthem with bears from one of the larger populations. If augmentationwere to be considered as a management option, the donor population must be able to withstandthe loss of some bears, presumably adult females. Furthermore, the receiving population will need to have enoughsuitable resources to supply the habitat requirements of additional bears.
Consequently, the currentstatus of all Florida black bearpopulations, largeand small, needs to be determinedin order to make the most beneficialmanagement decisions in the future.
In addition to biologicalconcerns surrounding the status of theFlorida black bear, there also arelegal matters that have yet to be resolved. The possibility remains that the
1998 rulingmade by the USFWS that precluded the Florida black bearfrom receiving federalprotection could be reversed. If that decision were overturned, theFlorida black bearwould be classified as a threatenedspecies in accordancewith the ESA of 1973. As a result, currentmanagement guidelines for bears in the ONWR would have to be reconsidered because it is possible that theblack bear hunting season would be terminated. Unfortunately,if that action were taken it would have serious biological and social consequences for bears andhunters.
Genetic Sampling as a Mark-recapture Method
Individual identificationusing genetic markershas the potential to become a powerfultool for estimating population size. This is especially true when non-invasive sampling techniques can be incorporatedinto mark-recapturestudies. Before
90 implementing a mark-recaptureexperiment of this type, however, manyaspects of study designrequire special attention. Although inaccessible andimpenetrable habitat precluded systematic trap placement in my study, I would suggest the use of a grid system in which every cell is trappedwhenever possible. This would likely prevent individual heterogeneity from imposinga significantbias on captureprobabilities. If bait were to be used to lure animals into hair trapsI would also recommend using minimal portions. Based on visitation rates fromthis study, bearsexhibited no ''trap-shy" behavior whatsoever in response to hair traps. Consequently, using excessive amountsof bait has the potential to quickly initiate a trap-happyresponse. That typeof behavior is difficultto alleviate andtypically results in a negatively biased estimate of population size. Althoughtemporal variation in capture probabilities is often difficultto predict, serious biases canresult if undetected. For example,animals that exhibit seasonal movements among sessions would inherently have lower capture probabilities thanthose that remained in the areato be sampled. Inthat scenario, mean capture rates would differamong sampling occasions andthe time assumption would be violated.
Fortunately, biases resulting from temporal variation can often be easily prevented by having ana priori understandingof the study animal'sbiology.
Samplingeffort is ananother matter that should be decided upon beforetrapping begins. The primaryissue thatneeds to be addressed is approximately how manyhair sampleswill be required to produce reliable estimates. Of course withtoo few data the performanceof most models will be poor because of low capture probabilities as a result of smallsample sizes (Otis et al. 1978, White et al. 1982). That problem can be alleviated, however, by deployingmore traps, checking traps more frequently,or
91 lengtheningthe duration of the trapping season. Unfortunately, each of those options can produce negative effects as well. Althoughincreasing the density of traps within a study areamay result in more captures, that response could be due to heterogeneity or trap response, rather thantrapping effort. The samedilemma applies to increasing the frequencythat traps are checked and extending the sampling time. In addition, lengthening trappingseasons may violate the closure assumption, which canseriously bias estimates of closed models. The challenge of determiningsampling effort, therefore, lies in maximizing capture probabilities while minimizing model violations to reduce bias andincrease precision (Pollock et al. 1990). This challenge canbe met by performinganalyses on simulated data beforefieldwork even begins. I suggest using the simulation procedure provided within Program CAPTURE that allows users to compare the performance of estimators for populations of known size in the presence of heterogeneous capture probabilities.
Of all the assumptions associated with open andclosed models, the most importantis that marksare not lost andmarked animals are identified correctly. Thisis logicalconsidering thatmost estimators of population size are derived fromthe proportion of marked andunmarked animals in a population. Althoughloss of marksis not a concern whenusing genetic markers, themisidentification of genotypesis possible andthe result is seriously biased estimators. Fortunately, probability of identity (Pl)
statistics provide estimates of the probability that 2 individuals from a population will
sharean identical genotype. Conceptually that probability should be low since many microsatellites aretypically used; 8 loci were used in thisstudy. However, PI's will often
be high if heterozygosity within a population is low, andthis will increase the chance that
92 animalsare misidentified in capture history data. Average heterozygosity estimates in
my study fe ll withina range of 60-80%, qualifying them as good genetic markersfor
individual identification (Taberlet andLuikart 1999). For smaller, more isolated
populations, however, researchersmay have to contend with heterozygosity levels below
thatrange. As a consequence, additional microsatellites will have to be analyzedto
differentiate individuals thathave similargenotypes.
Based on theabove, I suggest conductinga pilot study to assess thesampling
technique andthe utility of themicrosatellites to be used beforeinitiating a mark
recapture experiment using geneticmarkers. The firststep in thepilot study is to collect
the biological material of interest (i.e., hair,scat, feathers)using thetechnique intended
forthe experiment. There is limited use in paying forgenetic analyses if sampling
. methodsare unproductive. Assuming that adequate sample sizesare obtainable, I would
then evaluate the performanceof theDNA extraction andamplification process. The
amountof DNA contained in hairroots, forexample, is so minute that microsatellite
analyseswould oftenbe impossible without amplificationby polymerase chainreaction
(PCR). Once it is known that collection and analysisof samplesis possible, I would then
estimate thePI in the population of study. That will indicate whether themic rosatellites
that arechosen for analysiswill adequately performthe task of correctlyidentifying
individual animals. Furthermore,identification of genetic markers withhigh
heterozygosity will permit the use of fewermicrosatellites to reach the desired PI, which
will reduce thecost of analysisand the risk of misclassifications(Taberlet andLuikart
1999).
93 Genetic samplinghas greatpotential as a new technique for mark-recapture experiments. For studies primarilyconcerned with estimating population size, this method of samplingis especially promising. Because trap response bias appearsto be minimized using this technique, manysamples canbe collected in an efficient manner over a relatively short time period. Consequently, those capture history data are best suited for closed models.
94 LITERATURE CITED
95 LITERATURE CITED
Allen, T. G. 1999. Black bearpopulation size andhabitat use on Alligator River National WildlifeRefuge, North Carolina. Thesis, University of Tennessee, Knoxville, Tennessee,USA.
Avers, P. E., andK. C. Bracy. 1973. Soils andphysiography of the Osceola National Forest. U.S. Departmentof Agriculture, Forest Service,Atlanta, Georgia, USA. 94 pp.
Avise, J. C. 1994. Molecular markers,natural history andevolution. Chapmanand Hall, New York, New York, USA.
___ S. M. Haig, 0. A. Ryder, M. Lynch, andC. J. Geyer. 1995. Descriptive genetic studies: applications in population managementand conservation biology. Pages 183-244 in J. D. Ballou, M. Gilpin, andT. J. Foose, editors. Population managementfor survival andrecovery. ColumbiaUniversity Press, New York, New York, USA.
Beausoleil, R. A. 1999. Population andspatial ·ecology of the Louisiana black bearin a fragmentedbottomland hardwoodforest. Thesis, University of Tennessee, Knoxville, Tennessee,USA.
Bentzien, M. M. 1998. Endangeredand threatened wildlifeand plants; new 12-month finding fora petition to list the Florida black bear. Federal Register 63(235):67613-67618.
96 Boersen, M. R. 2001. Abundance and density of Louisiana black bearson theTensas River National WildlifeRefuge. Thesis, University of Tennessee, Knoxville, Tennessee, USA.
Brady, J. R., andD. S. Maehr. 1985. Distribution of black bearsin Florida. Florida Field Naturalist 13(1):1-7.
Brandenburg,D. M. 1996. Effectsof roads on behavior andsurvival ofblack bearsin coastal North Carolina. Thesis, University of Tennessee, Knoxville, Tennessee, USA.
Carothers,A. D. 1973. Capture-recapturemethods applied to a population with known parameters. Journalof Animal Ecology 45:125-146.
Caughley, G. 1977. Analysis of vertebrate populations. John Wiley andSons. New York, New York, USA.
Chao, A. 1987. Estimating population size forcapture-recapture data with unequal catchability. Biometrics43 :783-791.
___. 1988. Estimating animal abundancewith capture frequencydata. Journalof WildlifeManagement 52:2 95-300.
___. 1989. Estimating population size forsparse data in capture-recapture experiments. Biometrics 45:427-438.
___ S. M. Lee, andS. L. Jeng. 1992. Estimating population size capture-recapture data when captureprobabilities vary by time andindividual animal. Biometrics 48:201-216.
97 Chapman,D. G. 1951. Some properties of the hypergeometricdistribution with applications to zoologicalcensuses. University of CaliforniaPublications in Statistics 1: 131-160.
Clark,J. D. 1991. Ecology of two black bear(Ursus americanus) populations in the Interior Highlandsof Arkansas. Dissertation, University of Arkansas, Fayetteville, Arkansas, USA.
Donnelly, P. 1995. Non-independence of matches at different loci in DNA profiles: quantifyingthe effectof close relatives on thematch probability. Heredity 75:26-- 34.
Duever, M. J., andL. A. Riopelle. 1983. Successional sequences and rates on tree islands in the Okefenokee Swamp. AmericanMidland Naturalist 110(1):186-- 193.
Florida Game andFreshwater Fish Commission. 1992. A review of black bearhunting in Florida. Bureau of WildlifeResearch, Division of Wildlife,Florida Gameand Fresh Water Fish Commission. Tallahassee, Florida, USA.
Georgia Department of Natural Resources. 1999. Black bearmanagement planfor Georgia. WildlifeResources Division, GameManagement Section, Georgia Departmentof Natural Resources. Folkston, Georgia, USA.
Hall, E. R. 1981. The mammalsof NorthAmerica. Volume 1 and2. Second edition. John Wiley and Sons Publishers, New York, New York, USA.
Hamilton,D. B. 1982. Plantsuccession and influence of disturbancein the Okefenokee Swamp,Georgia. Dissertation, University of Georgia,Athens, Georgia, USA.
98 Hellgren,E. C., D. W. Carney,N. P. Garner,and M. R. Vaughan. 1988. Use of beakawaycotton spacers on radio collars. WildlifeSociety Bulletin 16:216-218.
__ andM. R. Vaughan. 1989. Demographicanalysis of a black bear population in theGreat Dismal Swamp. Journalof Wildlife Management49 :668- 674.
Hopkins, J. M. 1947. Forty-fiveyears with the OkefenokeeSwamp: 194 0-1945. GeorgiaSociety of Naturalists, Bulletin ( 4). Atlanta,Georgia, USA. 97 pp.
Izlar,R. L. 1984. A history of Okefenokee logging operations: a bourbon and branch · water success story. Pages 5-17 in A. D. Cohen, D. J. Casagrande, M. J. Andrejko, and G. R. Best, editors. The OkefenokeeSwamp: its natural history, geology, andgeochemistry. WetlandSurveys, Los Alamos,New Mexico, USA.
Jolly, G. M. 1965. Explicit estimates fromcapture-recapture data with both death and immigration-stochastic model. Biometrika 52 :225-247.
Klug, W. S., andM. R. Cummings. 1991. Concepts of Genetics. Macmillan,New York, New York, USA.
Learm,J., andB. J. Freeman. 1986. Fishes of the Okefenokee Swamp. Universityof Georgia Press, Athens, Georgia, USA.
Loftin, C. S., W. Raspberry, andW. M. Kitchens. 2000. Development of a grid-cell topographicsurface for Okefenokee Swamp, Georgia. Wetlands20 (3):487-499.
Mace, R. D., S. C. Minta, T. L. Manley, andK. E. Aune. 1994. Estimating grizzlybear population size using camerasightings. Wildlife Society Bulletin 22 :74-83.
99 Maehr, D. S., andJ. R. Brady. 1982. Fall food habitsof black bears in Bakerand Columbia counties, Florida. Proceedings of the Annual Conferenceof the SoutheasternAssociation of Fish andWildlife Agencies 36:565-570.
___ and___ . 1984. Food habits ofFlorida black bears. Journalof WildlifeManagement 48(1 ):23�235.
Martorello,D. A. 1998. Ecology of black bearsin coastal North Carolina. Thesis, University of Tennessee, Knoxville, Tennessee, USA.
Manning,T. , W. D. Edge, andJ. 0. Wolf. 1995. Evaluating population-size estimators: anempirical approach. Journalof Mammalogy76: 1149-1158.
McQueen, A. S., andH. Mizell. 1926. History of Okefenokee Swamp. Press of Jacobs andCompany, Clinton, South Carolina, USA.
Menkens, G. E., andS. H. Anderson. 1988. Estimation of small-mammal population size. Ecology 69:1952-1959.
Meyers, J. M., andE. P. Odum. 1991. Breeding bird populations of the Okefenokee Swampin Georgia:baseline for assessing future avifaunal changes. Journal of Field Ornithology62(1 ):53-68.
Mills, L. S., J. C. Citta, K. P. Lair, M. K. Schwartz,and D. A. Tallmon. 2000. Estimating animalabundance using n�ninvasi�e DNA ampling:promise and pitfalls. · EcologicalApplications 10 0 ):283-2�4.
:..
Mo�at, G.-, and C. Strobeck. 2000. Estimating popul�tion size of grizzlybears using haircapture, DNA profiling,and mark-recap �e analysis. Journalof Wildlife Management64: 183-193.
100 Mykytka, J. M., andM. R. Pelton. 1990. Managementstrategies forFlorida black bears based on home rangecomposition. International Conferenceon BearResearch andManagement 8: 161-167.
Otis, D. L., K. P. Burnham,G. C. White, andD. R. Anderson. 1978. Statistical inference from data on closed animalpopulations. WildlifeMonographs 62 .
Paetkau, D., W. Calvert, I. Stirling, andC. Strobeck. 1995. Microsatellite analysisof population structure in Canadian polar bears. MolecularEcology 4:347-354.
--- G. F. Shields, and C. Strobeck. 1998. Gene flow between insular,coastal and interior populationsof brown bearsin Alaska. MolecularEcology 7:1283-1292.
___ andC. Strobeck. 1994. Microsatellite analysisof genetic variationin black bearpopulations. MolecularEcology 3:489-495.
Parker,P. G., A. A. Snow, M. D. Schug, G. C. Booton, andP. A. Fuerst. 1998. What molecules cantell us about populations: choosing andusing a molecularmarker. Ecology 79:361-382.
Parrish,F. K., andE. J. Rykiel. 1979. Okefenokee Swamporigin: review and reconsideration. Journalof the Elisha Mitchell Science Society 95:17-31.
Pelton, M. R. 1982. Black bear. Pages 504--514 in J. A. Chapmanand G. A. Feldhamer, editors. Wild Mammalsof NorthAmerica. Johns Hopkins University Press, Baltimore, Maryland,USA.
101 -- and F. T. van Manen. 1997. Status ofblack bearsin the southeasternUnited States. Pages 31-44 in A. L. Gaski andD. F. Williamson, editors. Proceedings of the Second International Symposiumon the Trade of BearParts. Traffic USA/World WildlifeFund, Washington D.C., USA.
Pollock, K. H., J. D. Nichols, C. Brownie, andJ. E. Hines. 1990. Statistical inference for capture-recapture experiments. WildlifeMonographs 10 7.
Rexstad, E., andK. P. Burnham. 1992. User's guide for interactive program CAPTURE. Colorado Cooperative Fish andWildlife Research Unit, Colorado State University, Fort Collins, Colorado, USA.
Rice,W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225.
Robson, D. S., and H. A.Reiger. 1964. Samplesize in Petersen mark-recapture experiments. Transactionsof the AmericanFisheries Society 93:215-216.
Rogers,L. L. 1987. Effectsof foodsupply and kinship on social behavior, movements, andpopulation growthof black bearsin northeastern Minnesota. Wildlife Monographs97.
Rykiel, E. J. 1977. The OkefenokeeSwamp watershed: water balance and nutrient budgets. Dissertation, University of Georgia, Athens, Georgia, USA.
Scheick, B. K. 1999. Black beardiet, movements, and habitat selection in northern Florida and south Georgia. Thesis, Universityof Florida, Gainesville, Florida, USA.
Seber, G. A. F. 1965. A note on themultiple-recapture census. Biometrika 52 :249-259.
102 ___. 1982. The estimation of animal abundance andrelated parameters. Second edition. Macmillan,New York, New York, USA.
Seibert, S. G. 1993. Status andmanagement of black bearsin Apalachicola National Forest. Final Report Number W-41-30. Florida Game andFresh Water Fish Commission, Tallahassee, Florida, USA. 29 pp.
Smith, T. R. 1985. Ecology of black bearsin the bottomlandhardwood forest in Arkansas. Dissertation, University of Tennessee, Knoxville, Tennessee, USA.
Soil ConservationService. 1984. Soil survey of Columbia county Florida. Soil ConservationService, U.S. Departmentof Agriculture, Washington D.C., USA.
Sokal, R. R., andF. L. Rohlf. 1995. Biometry: theprinciples andpractice of statistics in · biologicalresearch. Third edition. W. H. Freeman,New York, New York, USA.
Stratman,M. R. 1998. Habitat use andeffects of prescribed fireon black bearsin northwesternFlorida. Thesis, University of Tennessee, Knoxville, Tennessee, USA.
Taberlet, P., andG. Luikart. 1999. Non-invasive genetic samplingand individual identification. BiologicalJournal of the Linnean Society 68:41-55.
U.S. Fish andWildlife Service. 2001. OkefenokeeNational WildlifeRefuge Biological Review. Folkston, Georgia, USA.
Yin, Z., andG. A. Brook. 1992. The impact of theSuwannee River Sill on the surface hydrology of Okefenokee Swamp,USA. Journal of Hydrology 136:193-217.
103 White, G. C., D. R. Anderson, K. P. Burnham,and D. L. Otis. 1982. Capture-recapture andremoval methods for samplingclosed populations. Los AlamosNational Laboratory, LA 8787-NERP, Los Alamos,New Mexico, USA.
Willey, C. H. 1974. Agingblack bearsfrom firstpremolar tooth sections. Journalof Wildlife Management38 :97-100.
Wooding, J. B. 1992. The Florida status report. Proceedings of the EasternWorkshop on Black BearResearch and Management 11 :12-13.
___ andT. S. Hardinsky. 1992. Denning by black bearsin northcentralFlorida. Journal ofMammalogy73(4): 895-898.
___ and___ . 1994. Home range, habitat use, andmortality of black bears in north-centralFlorida. InternationalConference on BearResearch andManagement 9:349-356.
Woods, J. G., andB. L. Mclellan. 1995. West slopes bearresearch project firstprogress report. 29 pp.
______D. Paetkau, C. Strobeck,M. Proctor. 1996. DNA fingerprinting applied to mark-recapture bearstudies. InternationalBear News 5:9-10.
--- D. Paetkau, D. Lewis, B. L. Mclellan,M. Proctor, C. Strobeck. 1999. Genetic taggingfree-ranging black andbrown bears in mark-recapture experiments. Wildlife Society Bulletin 27:616-627.
Wright,J. M., andP. Bentzen. 1995. Microsatellites: genetic markers forthe future. Pages 117-121 in G. R. Carvalhoand T. J. Pitcher, editors. Moleculargenetics in fishes. Chapman and Hall, London, United Kingdom.
104 Young, H., J. Neess, andJ. T. Emlen. 1952. Heterogeneity of trap response in a population of house mice. Journalof Wildlife Management 16(2):169-180.
105 APPENDICES
106 Appendix A. Trapping results for the Georgia study area, 1995-1998
107 Table A. l. Black bearcaptures on the Georgia study area, 1995-1998.
Date Bear ID# Capture Type Sex Weight (kg) Age (yrs) 06-Jun-95 001 Initial Male 79.5 7 08-Jun-95 002 Initial Male 63.6 5 09-Jun-95 003 Initial Male 72.7 4 10-Jun-95 004 Initial Male 125.0 10 11-Jun-95 005 Initial Male 90 .9 5 13-Jun-95 003 Recapture Male 63.6 4 15-Jun-95 001 Recapture Male 90.9 7 26-Jun-95 007 Initial Female 54.5 5 26-Jun-95 008 · Initial Male 159.1 5 28-Jun-95 010 Initial Female 40 .9 3 30-Jun-95 011 Initial Male 54.5 3 01-Jul-95 020 Initial Male 63.6 3 03-Jul-95 012 Initial Female 34.1 2 04-Jul-95 021 Initial Female 45.5 10 05-Jul-95 022 Initial Male 136.4 7 05-Jul-95 020 Recapture Male 68.2 3 09-Jul-95 023 Initial Female 36.4 5 09-Jul-95 011 Recapture Male 54.5 3 13-Jul-95 021 Recapture Female 56.8 10 14-Jul-95 024 Initial Male 34.1 1 18-Jul-95 013 Initial Male 45.5 2 19-Jul-95 014 Initial Female 63.6 5 21-Jul-95 025 Initial Male 68.2 2 26-Jul-95 026 Initial Male 113.6 5 27-Jul-95 029 Initial Male 72.7 3 27-Jul-95 030 Initial Male 70 .5 2 29-Jul-95 031 Initial Female 47.7 2 30-Jul-95 032 Initial Male 131.8 6 30-Jul-95 033 Initial Male 43.2 1
108 Table A.1. ( continued)
Date ID# Capture Type Sex Weight (kg) Age (yrs) 02-Aug-95 034 Initial Male 68.2 4 04-Aug-95 - 013 Recapture Male 45.5 2 09-Aug-95 035 Initial Male 34.1 1 09-Aug-95 036 Initial Male 125.0 6 09-Aug-95 037 Initial Female 52.3 7 10-Aug-95 039 Initial Female 38.6 2 11-Aug-95 015 Initial Male 90.9 3 12-Aug-95 040 Initial Female 50 .0 10 13-Aug-95 016 Initial Male 79.5 3 14-Aug-95 017 Initial Male 81.8 8 16-Aug-95 018 Initial Male 81.8 4 17-Aug-95 019 Initial Male 50 .0 2 17-Aug-95 040 Recapture Female 50 .0 10 18-Aug-95 027 Initial Male 61.4 5 20-Aug-95 038 Initial Male 75.0 3 24-Aug-95 022 Recapture Male 90 .9 7 27-Aug-95 041 Initial Male 70 .5 3 28-Aug-95 042 Initial Male 131.8 6 29-Aug-95 043 Initial Male 56.8 2 30-Aug-95 044 Initial Male 79.5 6 31-Aug-95 015 Recapture Male 93.2 3 01-Sep-95 045 Initial Female 56.8 3 . 01-Sep-95 046 Initial Female 50 .0 4 03-Sep-95 047 Initial Female 59.1 2 05-Sep-95 048 Initial Female 61.4 5 06-Sep-95 049 Initial Male 113.6 9 07-Sep-95 050 Initial Male 72.7 4 12-Sep-95 051 Initial Female 52.3 6 12-Sep-95 043 Recapture Male 63.6 2
109 Table A.l. (continued) Date ID# Capture Type Sex Weight (kg) Age (yrs) 15-Sep-95 051 Recapture Female 52.3 6 18-Sep-95 028 Initial Male 56.8 2 19-Sep-95 052 Initial Female 50.0 8 23-Oct-95 053 Initial Female 45.5 3 24-Oct-95 054 Initial Female 50 .0 7 25-Oct-95 055 Initial Male 38.6 2 26-Oct-95 056 Initial Female 52 .3 5 30-Oct-95 057 Initial Male 56.8 3 30-Oct-95 058 Initial Male 50.0 2 31-Oct-95 059 Initial Female 50 .0 3 31-Oct-95 060 Initial Female 56.8 6 03-Nov-95 061 Initial Male 47.7 3 04-Nov-95 055 Recapture Male 37.7 2 06-Nov-95 062 Initial Male 61.4 2 07-Nov-95 063 Initial Female 47.7 5 25-Nov-95 064 Initial Female 45.5 6 26-Nov-95 065 Initial Female 50 .0 2 04-Dec-95 067 Initial Female 40.9 10 04-Dec-95 068 Initial Female 52.3 8 06-Dec-95 070 Initial Female 40.9 3 16-Jun-96 071 Initial Female 61.4 6 19-Jun-96 072 Initial Female 50 .0 7 24-Jun-96 073 Initial Female 45.5 5 24-Jun-96 004 Recapture Male 129.5 11 24-Jun-96 074 Initial Male 54.5 1 26-Jun-96 020 Recapture Male 61.4 4 26-Jun-96 023 Recapture Female 45.5 6 28-Jun-96 014 Recapture Female 75.0 6 28-Jun-96 075 Initial Female 50 .0 3
110 Table A.1. (continued)
Date ID# Type Sex Weight (kg) Age (yrs) 30-Jun-96 076 Initial Male 50.0 3 30-Jun-96 077 Initial Female 40.9 5 30-Jun-96 010 Recapture Female 47.7 4 02-Jul-96 020 Recapture Male 61.4 4 04-Jul-96 073 Recapture Female 45.5 6 18-Jul-96 031 Recapture Female 54.5 3 20-Jul-96 038 Recapture Male 106.8 4 21-Jul-96 078 Initial Female 50.0 6 22-Jul-96 . 074 Recapture Male 54.5 1 22-Jul-96 038 Recapture Male 106.8 4 25-Jul-96 079 Initial Female 40 .9 NIA 27-Jul-96 080 Initial Male 38.6 1 29-Jul-96 021 Recapture Female 52.3 11 31-Jul-96 081 Initial Male 120.5 9 15-Aug-96 003 Recapture Male 118.2 5 18-Aug-96 016 Recapture Male 84.1 4 22-Aug-96 022 Recapture Male 113.6 8 24-Aug-96 013 Recapture Male 68.2 3 25-Aug-96 082 Initial Female 77.3 Yng. adult 0l-Sep-96 083 Initial Male 61.4 2 25-Sep-96 004 Recapture Male 113.6 11 03-Oct-96 051 Recapture Female 68.2 7 20-Oct-96 091 Initial Female 47.7 3 22-Oct-96 092 Initial Female 79.5 NIA 26-May-97 103 Initial Male 65.9 2 01-Jun-97 084 Initial Male 56.8 1 09-Jun-97 038 Recapture Male 136.4 5 11-Jun-97 103 Recapture Male 61.4 2 13-Jun-97 085 Initial Male 65.9 1
111 Table A.1. ( continued)
Date ID# Type Sex Weight (kg) Age (yrs) 13-Jun-97 086 Initial Male 36.4 1 14-Jun-97 013 Recapture Male 136.4 4 15-Jun-97 043 Recapture Male 90 .9 4 15-Jun-97 104 Initial Male 72.7 2 16-Jun-97 087 Initial Male 43.2 1 17-Jun-97 085 Recapture Male 65.9 1 19-Jun-97 088 Initial Male 63.6 2 20-Jun-97 089 Initial Male 59.1 1 21-Jun-97 038 Recapture Male 147.7 5 03-Jul-97 045 Recapture Female 68.2 4.5 13-Jul-97 089 Recapture Male 72 .7 1 15-Jul-97 078 Recapture Female 50.0 7 20-Jul-97 10 4 Recapture Male 75.0 2 21-Jul-97 103 Recapture Male 70 .5 2 23-Jul-97 093 Initial Male 52.3 2 23-Jul-97 074 Recapture Male 102.3 2 23-Jul-97 088 Recapture Male 50.0 2 23-Jul-97 094 Initial Male 54.5 4 30-Jul-97 016 Recapture Male 115.9 5 30-Jul-97 093 Recapture Male 45.5 2 0l-Aug-97 084 Recapture Male 56.8 1 04-Sep-97 049 Recapture Male 181.8 11 05-Sep-97 095 Initial Male 77.3 1 06-Sep-97 020 Recapture Male 102.3 5 08-Sep-97 073 Recapture Female 56.8 6 10-Sep-97 096 Initial Male 127.3 5 11-Sep-97 075 Recapture Female 52.3 4 1 l-Sep-97 025 Recapture Male 127.3 4 11-Sep-97 002 Recapture Male 122 .7 7
112 Table A. l. (continued) Date ID# Type Sex Weight (kg) Age (yrs) 11-Sep-97 098 Initial Male 63.6 2 12-Sep-97 105 Initial Male 45.5 1 14-Sep-97 100 Initial Female 36.4 Yearling 15-Sep-97 101 Initial Male 145.5 5 22-Sep-97 102 Initial Female 56.8 4 24-Sep-97 093 Recapture Male 50.0 2 16-Sep-97 075 Recapture Female 52.3 4 17-Sep-97 095 Recapture Male 77.3 1 19-Sep-97 002 Recapture Male 122.7 7 25-Sep-97 107 Initial Female 56.8 5 04-Nov-97 108 Initial Female 59.1 4 08-Nov-97 109 Initial Female 45.5 2 08-Nov-97 037 Recapture Female 59.1 9 10-Nov-97 074 Recapture Male 113.6 2 11-Nov-97 110 Initial Male 68.2 1 13-Jun-98 110 Recapture Male 61.4 1.5 16-Jun-98 131 Initial Female 50.0 Adult 17-Jun-98 110 Recapture Male 61.4 1.5 17-Jun-98 133 Initial Male 72.7 Mid-adult 19-Jun-98 039 Recapture Female 45.5 5.5 21-Jun-98 134 Initial Female 25.0 Yearling 22-Jun-98 043 Recapture Male 106.8 4.5 24-Jun-98 093 Recapture Male 63.6 3 27-Jun-98 135 Initial Male 125.0 Yng. adult 27-Jun-98 136 Initial Male 95.5 Subadult 02-Jul-98 040 Recapture Female 63.6 12.5 04-Jul-98 045 Recapture Female 40.9 5.5 08-Jul-98 137 Initial Female 22.7 Yearling 08-Jul-98 138 Initial Female 27.3 Yearling
113 Table A.1. ( continued) Date ID# Type Sex Weight (kg) Age (yrs) 12-Jul-98 051 Recapture Female 56.8 8.5 13-Jul-98 9998 Initial Female 52.3 Mid-adult 24-Jul-98 140 Initial Female 36.4 Subadult 25-Jul-98 141 Initial Male 34.1 Yearling 26-Jul-98 085 Recapture Male 93.2 2 27-Jul-98 084 Recapture Male 68.2 2 29-Jul-98 094 Recapture Male 63.6 5 30-Jul-98 093 Recapture Male 61.4 3 30-Jul-98 143 Initial Female 52.3 Mid-adult 01-Aug-98 039 Recapture Female 54.5 5 10-Aug-98 144 Initial Male 22.7 Yearling 11-Aug-98 037 Recapture Female 63.6 10 15-Aug-98 003 Recapture Male 125.0 7 15-Aug-98 145 Initial Male 50.0 Yearling 17-Aug-98 146 Initial Male 45.5 Yearling 18-Aug-98 074 Recapture Male 104.5 3 18-Aug-98 094 Recapture Male 63.6 5 24-Aug-98 138 Recapture Female 29.5 Yearling 25-Aug-98 086 Recapture Male 54.5 2 30-Aug-98 147 Initial Male 54.5 Subadult 30-Aug-98 10 5 Recapture Male 61.4 2 01-Sep-98 148 Initial Male 10 9.1 Yng. adult 05-Sep-98 072 Recapture Female 61.4 9 08-Sep-98 049 Recapture Male 136.4 12 08-Sep-98 0908 Initial Male 88.6 Old adult 10 -Sep-98 076 Recapture Male 79.5 5 11-Sep-98 088 Recapture Male 68.2 3 12-Sep-98 149 Initial Male 75.0 Subadult 12-Sep-98 087 Recapture Male 54.5 2
114 Table A.1. (continued)
Date ID# Type Sex Weight (kg) Age (yrs) 13-Sep-98 020 Recapture Male 102.3 6 14-Sep-98 073 Recapture Female 56.8 7 14-Sep-98 150 Initial Male 54.5 Subadult 16-Sep-98 151 Initial Male 54.5 Yearling 17-Sep-98 096 Recapture Male 118.2 6 23-Sep-98 152 Initial Male 38.6 Yearling a Initial captures of nuisance bearsthat were relocated onto the study areain 1997.
115 Appendix B. Trapping results for the Florida study area, 1996-1998
116 Table B. l. Black bearcaptures on the Florida study area,1996-1998.
Date Bear ID# Capture Type Sex Weight (kg) Age (yrs) 07-Jun-96 201 Initial Male 147.7 4 08-Jun-96 202 Initial Female 38.6 3 11-Jun-96 203 Initial Male 75.0 3 14-Jun-96 205 Initial Female 68.2 4 14-Jun-96 207 Initial Male 79.5 2 15-Jun-96 204 Initial Male 88.6 3 16-Jun-96 209 Initial Male 145.5 3 17-Jun-96 206 Initial Female 52.3 3 17-Jun-96 208 Initial Female 45.5 2 20-Jun-96 210 Initial Male 63.6 3 24-Jun-96 211 Initial Female . 59.1 5 24-Jun-96 213 Initial Male 147.7 6 25-Jun-96 207 Recapture Male 79.5 2 26-Jun-96 215 Initial Female 59.1 7 29-Jun-96 212 Initial Male 25.0 1 29-Jun-96 214 Initial Male 90.9 3 30-Jun-96 216 Initial Male 61.4 2 08-Jul-96 217 Initial Female 25.0 1 11-Jul-96 219 Initial Female 56.8 3 12-Jul-96 218 Initial Male 54.5 3 13-Jul-96 220 Initial Male 40.9 1 14-Jul-96 22 1 Initial Male 43.2 1 15-Jul-96 222 Initial Male 97.7 2 20-Jul-96 223 Initial Male 47.7 3 21-Jul-96 224 Initial Female 63.6 3 24-Jul-96 225 Initial Male 63.6 1 24-Jul-96 226 Initial Female 59.1 5 26-Jul-96 227 Initial Female 31.8 1 26-Jul-96 209 Recapture Male 145.5 3
117 Table B. l. (continued)
Date Bear ID# Capture Type Sex Weight (kg) Age (yrs) 26-Jul-96 228 Initial Female 52 .3 3 29-Jul-96 229 Initial Male 147.7 6 30-Jul-96 230 Initial Male 45.5 1 01-Aug-96 231 Initial Male 150.0 6 10-Aug-96 232 Initial Female 75.0 3 17-Aug-96 233 Initial Male 43.2 2 18-Aug-96 225 Recapture Male 63.6 1 27-Aug-96 234 Initial Female 0.0 4 02-Sep-96 235 Initial Female 77.3 6 16-Dec-96 236 Initial Male 181.8 4 02-1a.n-91 216 Recapture Male 113.6 2.5 15-Jun-97 237 Initial Male 43.2 1 16-Jun-97 248 Initial Female 59.1 10 18-Jun-97 249 Initial Male 38.6 2 22-Jun-97 238 Initial Male 61.4 1 22-Jun-97 239 Initial Male 68.2 1 23-Jun-97 227 Recapture Female 52.3 2 24-Jun-97 218 Recapture Male 79.5 4 27-Jun-97 211 Recapture Female 84.1 6 28-Jun-97 207 Recapture Male 10 9.1 3 01-Jul-97 215 Recapture Female 54.5 8 02-Jul-97 240 Initial Male 93.2 5 09-Jul-97 241 Initial Male 65.9 1 10-Jul-97 242 Initial Male 111.4 2 15-Jul-97 241 Recapture Male 61.4 1 15-Jul-97 239 Recapture Male 63.6 1 17-Jul-97 243 Initial Female 43.2 2 17-Jul-97 214 Recapture Male 10 6.8 4 17-Jul-97 266 Initial Male 143.2 5
118 Table B.l. (continued) Date ID# Capture Type Sex Weight (kg) Age (yrs) 04-Aug-97 · 244 Initial Female 40 .9 1 05-Aug-97 245 Initial Male 147.7 7 06-Aug-97 246 Initial Female 97.7 7 14-Aug-97 247 Initial Male 129.5 11 17-Aug-97 216 Recapture Male 10 4.5 3.5 19-Aug-97 220 Recapture Male 65.9 2 21-Aug-97 225 Recapture Male 95.5 2 22-Aug-97 250 Initial Female 50.0 1 22-Aug-97 251 Initial Male 145.5 3 27-Aug-97 252 Initial Male 125.0 13 29-Aug-97 246 Recapture Female 79.5 7 02-Sep-97 233 Recapture Male 77.3 3 03-Sep-97 · 253 Initial Female 45.5 3 03-Sep-97 250 Recapture Female 47.7 1 04-Sep-97 254 Initial Female 68.2 9 04-Sep-97 255 Initial Male 100.0 3 05-Sep-97 260 Initial Female 45.5 1 06-Sep-97 219 Recapture Female 50.0 4 07-Sep-97 256 Initial Male 118.2 6 08-Sep-97 261 Initial Male 52.3 3 09-Sep-97 254 Recapture Female 68.2 9 11-Sep-97 257 Initial Female 68.2 1 l 1-Sep-97 236 Recapture Male 163.6 5 12-Sep-97 224 Recapture Female 54.5 4 14-Sep-97 253 Recapture Female 45.5 3 15-Sep-97 240 Recapture Male 125.0 6 16-Sep-97 263 Initial Male 120.5 5 18-Sep-97 220 Recapture Male 75.0 2 20-Sep-97 234 Recapture Female 86.4 5
119 Table B. l. ( continued) Date ID# Capture Type Sex Weight (kg) Age (yrs) 16-Jun-98 258 Initial Male 34.1 Yearling 16-Jun-98 259 Initial Female 25.0 Yearling 18-Jun-98 205 Recapture Female N/R 6 19-Jun-98 258 Recapture Male 34.1 Yearling 20-Jun-98 202 Recapture Female 45.5 5 21-Jun-98 206 Recapture Female 59.1 5 21-Jun-98 267 Initial Male 181.8 Yng. adult 22-Jun-98 264 Initial Male 38.6 Subadult 24-Jun-98 268 Initial Female 27.3 Yearling 25-Jun-98 269 Initial Male 31.8 Yearling 29-Jun-98 205 Recapture Female N/R 6 02-Jul-98 207 Recapture Male 131.8 4 02-Jul-98 270 Initial Male 59.1 Subadult 02-Jul-98 269 Recapture Male 31.8 Yearling 03-Jul-98 248 Recapture Female 63.6 11 08-Jul-98 271 Initial Female 43.2 Yng. adult 09-Jul-98 215 Recapture Female 65.9 9 10-Jul-98 209 Recapture Male 159.1 5 11-Jul-98 272 Initial Female 54.5 Yng. adult 14-Jul-98 211 Recapture Female 59.1 7 15-Jul-98 208 Recapture Female 56.8 4 16-Jul-98 273 Initial Male 29.5 Yearling 22-Jul-98 258 Recapture Male 34.1 Yearling 22-Jul-98 274 Initial Female 31.8 Yearling 23-Jul-98 275 Initial Female 27.3 Yearling 24-Jul-98 231 Recapture Male 147.7 8 01-Aug-98 232 Recapture Female 77.3 5 02-Aug-98 206 Recapture Female 59.1 5 04-Aug-98 274 Recapture Female 31.8 Yearling
120 Table B.1. (continued) Date ID# Type Sex Weight (kg) Age (yrs) 05-Aug-98 276 Initial Male 10 6.8 Subadult 06-Aug-98 277 Initial Male 25.0 Yearling 07-Aug-98 216 Recapture Male 102.3 4.5 08-Aug-98 226 Recapture Female 72.7 7 11-Aug-98 278 Initial Male 102.3 Subadult 14-Aug-98 220 Recapture Male 90.9 3 16-Aug-98 263 Recapture Male 136.4 6 16-Aug-98 245 Recapture Male 159.1 8 17-Aug-98 244 Recapture Female 36.4 2 19-Aug-98 279 Initial Male 104.5 Mid-adult 19-Aug-98 233 Recapture Male 84.1' 4 28-Aug-98 279 Recapture Male 104.5 Mid-adult 29-Aug-98 203 Recapture Male 136.4 5 16-Sep-98 216 Recapture Male 90.9 4.5 17-Sep-98 253 Recapture Female 59.1 4 21-Sep-98 283 Initial Male 27.3 Yearling
121 Appendix C. Laboratory protocol formicrosatellite analysis
122 MICROSATELLITE ANALYSIS
DNA Isolation
DNA was extractedfrom hairfollicles using theInstaGene Matrix (Bio-Rad
Laboratories, Hercules, California). Specifically, follicleswere incubated in the
InstaGene Matrix in the presence of Proteinase Kat 65°C overnight. This mixture was boiled (100 °C) for8-10 minutes, followed by centrifugationat 10,000-12,000 rpm. The resulting supernatant was used in PCR reactions.
First Stage
Microsatellite DNA amplificationwas performed in 2 stages. First Stage analysis consisted of theamplification of 8 microsatellite DNA loci using thePCR primers described in Paetkau and Strobeck (1994) andPaetkau et al. (1995). These loci are G lA,
GlD, GlOB, GlOC, GlOL, GlOM, GlOP, andGl 0X.
First Stage PCR
Each PCR reaction consisted of 1.5 µl of genomic DNA extract,0.8 75 X PCR buffer(59 mM Tris-HCl, pH 8.3; 15 mM (Nlii)2SO4; 9 mM J3-mercaptoethanol; 6 mM
EDTA), 2.25 mM MgCh, 0.2 mM dNTPs,0. 15-0.43 µM.of each primer (forward primer fluorescentlylabeled with TET, FAM, or HEX; Applied Biosystems (ABI), Foster City,
California),1.2 units ofTaq polymerase (ABI), anddeionized water added to achieve the finalvolume of 15 µI. The amplificationcycle consisted of aninitial denaturing at 94 °C for2 min followedby 35 cycles of94°C denaturing for30 sec, 56°C annealingfor 30
123 sec, and72 °C extension for 1 min. Cycling culminated with a 5-min extension at 72°C.
Thermal cyclingwas performedin anMJ DNA Engine PTC 200 (MJ Research,
Watertown, Massachusetts) configuredwith a heated lid.
Fragment Analysis
Generally, 1 µ1 of PCR product was diluted 1:1 with deionized water and thoroughlymixed. · One µI of this dilution was added to 12 µ1 of deionized formamide and0. 5 µ1 of the internal size standardGENE SCAN-500 (ABI). Alternatively, PCR products of separate multiplexed reactions (2-3 loci each) andmultiple separate reactions
(2--4)were combined andanalyzed without dilution. Loci were identifiedin these multiplexed samplesby virtueof theircharacteristic molecularmass and attached fluorescent label. The size standardcontained DNA fragmentsfluorescently labeled with thedye phosphoramidite TAMRA(red). This PCR product/size standard/fonnamide mixture was heat denaturated at 95°C for3 min andplaced immediately on ice for at least
5 min. The mixture was subjected to capillary electrophoresison an ABIPRISM 310
Genetic Analyzer (i.e., automated sequencer). Fluorescently labeled DNA fragments were analyzed, andgenotype data generated using GeneScansoftware (ABI).
GENOTYPER v. 2.0 (ABI) DNA fragmentanalysis software was used to score, bin, and outputallelic (and genotypic) designationsfor each bearsample.
Statistical Analyses
The multilocus genotypegenerated for each individual fromthe series of PCR amplificationswas analysedto determinethe uniquenessof each hair sample. Estimates
124 of individual pair-wise genetic distances, using the proportion of sharedalleles algorithm, was calculated using a 32-bit version of Microsat 1.5d (Eric Minch, Stanford University,
California).
Observedgenotype frequencies were tested forconsistency with Hardy-Weinberg and linkageequilibrium expectations using randomization tests implemented by
GENEPOP 3.1 (Raymond and Rousset 1995). The Hardy-Weinbergtest used the
Markovchain randomizationtest of Guo andThompson (1992) to estimate exact 2-tailed
P-values foreach locus. Bonferroniadjustments (Rice 1989) were used to determine statistical significance forthese tests. Linkagedisequilibrium tests used the randomizationmethod of Raymond and Rousset (1995) for all pairs ofloci. The amount of genetic variationin each samplewas summarizedby gene diversity ( average expected heterozygosity) and the average frequencyof unique alleles.
Literature Cited
Guo, S. W., andE. A. Thompson. 1992. Performingthe exact test of Hardy-Weinberg proportions formultiple alleles. Biometrics 48:361-372.
Raymond, M., andR. Rousset. 1995. GENEPOP (version 1.2): population genetics softwarefor exact tests and ecumenicism. Journalof Heredity 86:248-249.
Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225.
125 VITA
Steven Thomas Dobey was born on September 11, 1972 in Tupelo, Mississippi.
He graduatedfrom Charles. E. JordanHigh School in Durham, NorthCarolina in 1990.
He attended East CarolinaUniversity in Greenville, North Carolinawhere he received a
Bachelor of Science degreein Biology in December 1995. InAugust 1996, Steven was acceptedinto North Carolina State University as a non-degreestudent in the Department of Fisheries andWildlife Sciences. Steven was hired as a researchtechnician by the
University of Tennessee in May of 1997 andspent 7 months trappingblack bears in
Louisiana. One yearlater, Steven began graduateschool in the Departmentof Forestry,
Wildlifeand Fisheries at the University of Tennessee, Knoxvillestudying Florida black bears. He received his Master's of Science degreein Wildlife and Fisheries Science in
May 2001.
37842850 r 2 ('J IN7-a2 tm le
126