Title: Modeling Florida panther movements to predict conservation strategies in north Florida
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00100698/00001
 Material Information
Title: Modeling Florida panther movements to predict conservation strategies in north Florida
Physical Description: Book
Language: English
Creator: Cramer, Patricia Catherine, 1960-
Publisher: State University System of Florida
Place of Publication: <Florida>
<Florida>
Publication Date: 1999
Copyright Date: 1999
 Subjects
Subject: Wildlife Ecology and Conservation thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Wildlife Ecology and Conservation -- UF   ( lcsh )
Genre: government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )
 Notes
Summary: ABSTRACT: The Florida panther (Puma concolor coryi) is one of the most endangered mammals in North America. Reintroducing the panther to portions of its former range has been deemed critical to the species' future existence. The north Florida-south Georgia region is a prime candidate site for such reintroductions. Modeling the movements of Florida panthers is used as a tool to identify specific regional landscape features and conservation strategies that would be most critical to panthers, other species, and the ecosystems upon which they depend. The spatially explicit model PANTHER was created based on results from a state sponsored reintroduction feasibility study and ongoing studies of south Florida panthers. It mimics panther behavior and movement over Geographic Information Systems (GIS) maps representing natural communities, roads, deer densities, human densities, and human attitudes. Potential future effects of human development were also modeled, based on data derived from county and regional comprehensive plans, population projections, and development patterns. The model was validated by comparing output estimates with those from previous Florida panther studies. The model identified high probability use locations within the 7,000 square kilometer study area. The majority of these locations are also places of high development pressure, especially along the Suwannee River. Model output indicates panthers used private property approximately 67 percent of all moves. Model outputs were compared with data from a public education program conducted earlier in the research process.
Summary: ABSTRACT (cont.): Over 70 percent of panther moves were in Hamilton and north Columbia Counties, areas of lowest public support for panther reintroductions. Landscape connections for panthers and specific areas of high panther use along Interstates I-75 and I-10 were also identified. These and other model results support conservation approaches that include a continued commitment to regional and county planning in environmentally sensitive areas, possible public purchase of environmentally sensitive lands, and financial incentives to owners of private properties deemed critical to panthers. Model results support targeting landowners and residents of Hamilton and Columbia counties for future education programs and inclusion in conservation processes.
Summary: KEYWORDS: Florida panther, Puma concolor coryi, north Florida, Suwannee River, reintroductions, spatially explicit model
Thesis: Thesis (Ph. D.)--University of Florida, 1999.
Bibliography: Includes bibliographical references (p. 212-219).
Statement of Responsibility: by Patricia Catherine Cramer.
General Note: Title from first page of PDF file.
General Note: Document formatted into pages; contains xi, 220 p.; also contains graphics.
General Note: Vita.
 Record Information
Bibliographic ID: UF00100698
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 45265727
alephbibnum - 002484336
notis - AMJ9950

Downloads

This item has the following downloads:

cramer ( PDF )


Full Text












MODELING FLORIDA PANTHER MOVEMENTS TO
PREDICT CONSERVATION STRATEGIES IN
NORTH FLORIDA









By

PATRICIA CATHERINE CRAMER


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1999



























Copyright 1999

By

Patricia Catherine Cramer












ACKNOWLEDGMENTS


I wish to express my sincere appreciation for all those who helped me through

the Ph.D. process. I would first and foremost like to thank my spouse, Robert Hamlin,

for all the intellectual, emotional, and financial support he has given me these years.

This dissertation is a quality piece of science thanks in large part to him. I also would

like to thank Dr. Ken Portier for the past several years of inspiration and computer

support. I wish to thank Dr. Larry Harris for helping me arrive at the University of

Florida, opening my horizons to landscape ecology, and supporting me. To the other

members of my committee Drs. Wiley Kitchens, Mel Sunquist, and Clay Montague -

who have all been there for me over the years and have provided much scientific

advice, I also extend my thanks. I would also like to thank Brad Stith (Sir Bradley),

for his enormous contributions to the development of the model. Without Brad, I

would not have graduated until the next millennium. I also want to extend a general

thank you to the graduate students I have shared my thoughts and work with.

Technical advice, ideas, laughter, and anguish were shared with many friends and

peers. Thank you all for the past seven years. I wish to only begin to show my

appreciation to Tom and Janet Cramer, my loving parents, who began my education

process 38 years ago. Where I am today is due in large part to the support and

inspiration they have given me since my first breath. Thank you.
















TABLE OF CONTENTS


page
ACKN OW LEDGM ENTS ............................................. iii

LIST OF TABLES ......... .................................... vi

LIST OF FIGURES ............................................ vii

ABSTRACT ........................................ ....... .. x

CHAPTERS

1 INTRODUCTION......... ............................... 1

Overview ............. .. ..... .... ............... ......... 1
Landscape Ecology and Multi-Disciplinary Study .............. . ... 2
Spatially Explicit M odels ............................... ..... 6
The N orth Florida Setting . ............. ..................... .. 10
Regional and Land Use Planning in Florida .................... 20
D issertation Structure . . ....... .............................. 32

2 M ODEL M ETH OD S.. . ....... ............................... 34

Introduction......... .. ........................ .. ....... 34
Movement Model Mimicking Florida Panther Movements. ........... 39
Quantification of the Landscape Images .................... 65
Landscape Scenario Development ........................ ...... 89
Model Evaluation and Sensitivity Analysis ................... 92









3 MODEL SIMULATION RESULTS .............................. 96

Introduction . . ....... ..................................... 96
Model Development .................... ................. 96
Establishing Model Settings: 1990's Conditions, Scenario II ........... 139
Predicting Effects of Human Pressure on Panthers:
Future Scenarios III and IV .............................. . 149
Summary......... ............................. ....... 162

4 DISCUSSION.......................................... 164

Panthers and Environmental and Human Variables ................... 165
Conclusions and Significance of PANTHER Results ................. 176
Significant Areas for Florida Panthers in North Florida ............... 181
Conservation Recommendations .......................... 188

5 CONCLUSIONS AND FUTURE APPLICATIONS ................. 197

APPENDIX
THE NORTHEAST FLORIDA PANTHER EDUCATION
PROGRAM EXECUTIVE SUMMARY ............................... 202

REFERENCES.......... ..................................... 212

BIOGRAPHICAL SKETCH . . ...... ............................. .. 220













LIST OF TABLES


Table Page

1.1 Human population estimates of north Florida counties (taken from
Bureau of Economic and Business Research 1996). 17

1.2 Approximate acreage of generalized land use within the
unincorporated areas of north Florida counties (taken from Data and
Analysis Plans of Baker (1996), Columbia (1996), Hamilton (1996),
Suwannee (1996), and Union (1996) Counties. 19

1.3 Protected lands of significant size (>1,000ha) in north Florida and
respective managing agencies. 21

1.4 Natural Resources of Regional Significance as identified by the North
Central and Northeast Florida Regional Planning Councils in their
Strategic Plans (1996). 24

2.1 Initial ranking of habitat variables of neighboring cells. 43

2.2 Devaluation amounts for cells recently visited. 52

2.3 Home range cell index value increases and decreases. 55

2.4 Land cover types used in Natural Community Landscape Image (taken
from Cox et al. 1994). 66

2.5 Natural community rankings for cell habitat index values. 68

2.6 Classification of Roads Landscape Image. 74

2.7 Final road evaluation by panthers and mortality probabilities. 77

2.8 Initial ranking of deer densities. 80

2.9 Human use intensity values assigned to individual parcels based on
Florida Department of Revenue land use codes. 84

2.10 Human density classes based on sum of parcels in each square mile. 84









2.11 Panther evaluation of human density classes. 86

2.12 Response by county to the telephone survey question: How would you
agree or disagree with the following statement? "I favor the
reintroduction of panthers in my county or surrounding counties." 89

3.1 Various settings of rankings of Landscape Images and home range
rules for resident females used in model calibration. 98

3.2 Baseline settings for each panther type for each Landscape Image. 103

3.3 Home range estimates in kilometers2 for all panthers under various
settings. 108

3.4 Percentage utilization of natural communities by simulated panthers. 127

3.5 Average rate of daily road crossings per panther for selected settings. 129

3.6 Mean home range size (km2) for panthers in Scenario II, setting 10. 140

3.7 Average percent utilization of natural communities by panthers. 142

3.8 Average number of road crossings per day per cat, based on gender. 144

3.9 Private and public land use by overall panther population, and among
three geographic groups. 145

3.10 Panther use of counties grouped according to resident attitudes toward
panthers. 146

3.11 Estimated average home range size (km2) per panther for all four
scenarios. 154

3.12 Average percent utilization of natural communities by simulated
panthers. 156

3.13 Average number of road crossings per day per panther, all scenarios. 158

3.14 Average private and public land use, all scenarios. 159

4.1 Panther use of counties in contrast to county support for panthers. 176













LIST OF FIGURES


Figure Page

1.1 The five county study area within the state of Florida. 11

1.2 North Florida major towns and conservation areas. 16

1.3 Flow diagram of research synthesis. 33

2.1 Flow diagram of PANTHER model. 38

2.2 Full flow diagram of PANTHER model (minor classes excluded). 41

2.3 Panther perception distance of neighboring cells. 48

2.4 Devaluation of cells based on previous moves. 50

2.5 Initial starting places for seven panthers simulated in PANTHER. 64

2.6 Natural Communities Landscape Image based on Florida Fish and
Wildlife Conservation Commission, Closing the Gaps Database. 69

2.7 Roads and public conservation lands in north Florida. 75

2.8 Deer Densities Landscape Image. 81

2.9 Human Density Landscape Image for north Florida, 1996. 83

2.10 Public Conservation Lands Landscape Image in north Florida. 88

2.11 Human Density Landscape Image for the year 2020 91

2.12 Human Density Landscape Image for Scenario III, 2020 92

3.1 Locations and home ranges of panthers simulated in setting 1. 114

3.2 Locations and home ranges of panthers simulated in setting 5. 117

3.3 Locations and home ranges of panthers simulated in setting 16. 118









3.4 Locations and home ranges of panthers in one simulation of setting
10. 121

3.5 Locations and home ranges of panthers simulated in setting 3. 123

3.6 Road associated mortality sites generated during model sensitivity
analysis. 133

3.7 Locations and home ranges of panthers simulated in setting 11. 136

3.8 Locations of panthers in a simulation of Scenario II, 1990's condition. 141

3.9 Areas of use by panthers during three different simulations of
Scenario II, present condition with humans. 148

3.10 Locations and home ranges of panthers simulated in Scenario III,
future with best conservation. 152

3.11 Locations and home ranges of panthers simulated in Scenario IV,
future, no added conservation. 153

4.1 Major existing and potential conservation areas in north Florida. 184
















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

MODELING FLORIDA PANTHER MOVEMENTS
TO PREDICT CONSERVATION STRATEGIES
IN NORTH FLORIDA

By

Patricia Catherine Cramer

August 1999


Chairman: Kenneth Portier
Major Department: Wildlife Ecology and Conservation

The Florida panther (Puma concolor coryi) is one of the most endangered

mammals in North America. Reintroducing the panther to portions of its former range

has been deemed critical to the species' future existence. The north Florida-south

Georgia region is a prime candidate site for such reintroductions. Modeling the

movements of Florida panthers is used as a tool to identify specific regional landscape

features and conservation strategies that would be most critical to panthers, other

species, and the ecosystems upon which they depend. The spatially explicit model

PANTHER was created based on results from a state sponsored reintroduction

feasibility study and ongoing studies of south Florida panthers. It mimics panther

behavior and movement over Geographic Information Systems (GIS) maps

representing natural communities, roads, deer densities, human densities, and human









attitudes. Potential future effects of human development were also modeled, based on

data derived from county and regional comprehensive plans, population projections,

and development patterns. The model was validated by comparing output estimates

with those from previous Florida panther studies. The model identified high

probability use locations within the 7,000 square kilometer study area. The majority

of these locations are also places of high development pressure, especially along the

Suwannee River. Model output indicates panthers used private property

approximately 67 percent of all moves. Model outputs were compared with data from

a public education program conducted earlier in the research process. Over 70 percent

of panther moves were in Hamilton and north Columbia Counties, areas of lowest

public support for panther reintroductions. Landscape connections for panthers and

specific areas of high panther use along Interstates 1-75 and 1-10 were also identified.

These and other model results support conservation approaches that include a

continued commitment to regional and county planning in environmentally sensitive

areas, possible public purchase of environmentally sensitive lands, and financial

incentives to owners of private properties deemed critical to panthers. Model results

support targeting landowners and residents of Hamilton and Columbia counties for

future education programs and inclusion in conservation processes.















CHAPTER 1
INTRODUCTION

Overview


Today's conservation challenges compel ecologists to approach scientific inquiry

in innovative, multidisciplinary ways. Landscape ecology, in particular, has become a

scientific approach that mixes several different elements of conservation over large-scale

areas to better understand and address conservation in an increasingly human dominated

world. One such approach to large-scale conservation challenges is the use of spatially

explicit models. Spatially explicit models are used to bring together large amounts of

data on populations and landscapes to make predictions over large temporal and spatial

scales that would otherwise be near impossible to address through traditional research.

This dissertation evolved from the development of The Upper Suwannee Vision, a

landscape ecology research endeavor of north Florida conservation alternatives,

supported by Occidental Chemical, Inc. (Cramer and Harris 1993). The Upper Suwannee

Vision revealed that traditional scientific inquiry and data management methods would

be insufficient to address large-scale conservation issues in the upper Suwannee River

basin of north Florida. This dissertation developed conservation strategies for north

Florida based, in part, on the potential reintroduction of a population of Florida panthers

(Puma concolor coryi) in the area. A spatially explicit model was developed as an

innovative approach to panther conservation. The objective of the model was to predict









what lands and conservation strategies would best serve this wide-ranging species.

Geographic Information Systems maps, along with model outputs were used to help

identify potential conservation, management, planning, and scientific strategies for the

upper Suwannee River basin of north Florida.

This introductory chapter provides a theoretical basis for this synthesis and

introduces the reader to the north Florida setting. The chapter concludes with a brief

synopsis and flow diagram of research synthesis.


Landscape Ecology and Multidisciplinary Study

The theory and applications of landscape ecology are logical choices for

addressing large-scale conservation efforts. Landscape ecology is a multi-disciplined

scientific approach, which has developed in response to the increased recognition of the

need to expand the scale of ecological investigations. Landscape ecology's roots can be

traced to Europe where it began as a science that merged geography and holistic ecology

with infusions from landscape architecture, land management, land use planning, and

sociology (Naveh 1991, Naveh and Lieberman 1984, Schreiber 1990, Wiens et al. 1993,

Zonneveld 1990). The main focus of landscape ecology has been on the spatially explicit

patterns of landscape mosaics and interactions among their elements (Risser et al. 1984,

Forman and Godron 1986).


Connectivity

Analysis of connectivity is one method to assess the potential for interactions

among landscape elements. Connectivity typically involves linking together once

contiguous natural systems. These natural systems can be at the species, community,









landscape, or ecological processes levels, and at multiple spatial and temporal scales

(Noss 1990). Connectivity at one level or for a specific taxon may not equate to

connectivity for other scales and species. Connectivity is a function of the abundance

and spatial patterning of landscape resources and the organism's scale of resource use

(Turner et al. 1995a, O'Neill et al. 1996). Landscape level connectivity is measured at a

large or coarse scale that covers areas measured in kilometers. Traditionally, wide

ranging species, such as the Florida panther, have been used to measure and promote

connectivity at such landscape and even regional scales.

Typically, landscape level connectivity involves linking vestiges of ecological

structure (ecosystems, communities, populations, or individuals) and processes that once

were the dominant features of the landscape. It is becoming increasingly complex to

efficiently preserve and maintain the integrity of such landscape features in a human

dominated world. This situation forces ecologists to address the influences of humans on

the species or system under study. Scientists may best ameliorate the effects of humans

on the natural world by incorporating the study of human processes into scientific

inquiry. Research based in part on human processes may be better suited for

incorporation into human land use changes and conservation planning processes than

traditional research that fails to include human influences.


A Broader Approach

Today, multi-disciplined study is becoming more common as scientists begin to

integrate a multitude of approaches and enhance the rigor of landscape ecology.

Integrated methods allow a greater generalized understanding of the overlying science of

landscape ecology and help to address specific problems in reserve design, habitat









fragmentation, maintenance of biological diversity, and sustainability. In addition, an

expanded approach to landscape ecology integrates questions at many biological levels

and spatial scales (Weins 1989. Wiens et al. 1993) and examines the role of processes

including human effects (Turner 1989, Turner et al. 1995b). While addressing these

challenges, landscape ecology has emerged as a problem-solving science that is important

for conservation assessment, planning, management, and restoration at relevant scales in

order to solve complex issues in all regions of the world (Caldwell 1990, Vos and Opdam

1993, Forman 1983, Naimen 1996). This application of science to real world challenges

has brought many landscape ecologists to embrace conservation biology. Conservation

biology, as defined by the Society for Conservation Biology, is concerned with the

scientific study of the phenomena that affect the maintenance, loss, and restoration of

biological diversity.


New Solutions to Complex Problems

In response to the challenge of applying science to real world situations, there is a

trend today for traditional and applied disciplines to come together to seek solutions to

complex problems of ecosystem sustainability and human welfare (Kessler et al. 1992).

This includes taking steps outside the academic world. Researchers around the world are

recognizing the need to include scientists and scientific information in the management,

planning, and policy arenas (Huenneke 1995, Underwood 1995, Van Der Ploeg and

Vlum 1978). Naiman (1996) urged landscape ecologists to embrace their responsibility

to actively integrate scientific information with decision making about fresh water

systems on different relevant scales. In California Beier (1993) demonstrated the need

for bringing scientific data on mountain lion (Puma concolor) movements in southern






5


California into regional planning. In Oregon, Swanson and Franklin (1992) called on

ecosystem researchers to assume roles in the human social processes for determining the

future course of management of natural resources. In Australia Hobbs et al. (1993)

argued for the need to integrate landscape research, planning, and management to protect

native habitat patches. Norton (1998) and Primm and Clark (1996) suggested more

effective and active roles for scientists in designing solutions to the problems of

landscape level conservation.

In Florida, with the human population growing at approximately 5.2 percent each

year, with a net gain of 2087 people for every day in 1998 (J. Nogle, Bureau of Economic

and Business Research, University of Florida, personal communication), there is an

urgent need for scientists to become involved in land use and conservation processes.

Scientists can continue to play an important role in the state's ambitious land acquisition

programs such as Florida Forever, Preservation 2000, Save Our Rivers, and Conservation

and Recreation Lands (CARL). They must also lend scientific knowledge to regional and

local land use planning. With suburban sprawl and comprehensive planning as the two

most important issues in many local political arenas, now is a prime opportunity for

Florida scientists to lend their knowledge to social processes that will have a direct

influence on the species and natural systems they strive to preserve. This research was

designed to generate results that can be utilized by these human processes.









Spatially Explicit Models


The Use of Spatially Explicit Models as Tools for Understanding

The spatially explicit model (SEM) is a tool that can further our understanding of

complex situations pertaining to ecological research and human systems. The use of

models to make or defend management decisions is becoming more common in

conservation biology (Bart 1995, DeAngelis and Gross 1992, Conroy et al. 1995,

Dunning et al. 1995, Holt et al. 1995, Turner et al. 1994, Turner et al. 1995b, Lorek and

Sonnenschein 1998). Spatially explicit models consider both species-habitat relations

and the arrangement of habitats in time and space often through the use of object-

orientated computer languages and Geographic Information Systems (GIS). SEMs have

a structure that specifies the location of each object of interest (such as an organism,

population, or habitat patch) within a heterogeneous landscape. Models help to define

the spatial relations between habitat patches and other features of the landscape such as

boundaries and corridors (Dunning et al. 1992). SEMs can help researchers and

managers understand the complex relations between landscape configuration and

population dynamics (Pulliam and Dunning 1995), and how proposed management

strategies or other land-use change scenarios might affect animal and plant populations

(Forman 1995, Pulliam and Dunning 1995, Turner et al. 1995a).


Spatially Explicit Models and Species Conservation

Spatially explicit models can be used to predict specific locations in the landscape

where conservation planning efforts should be concentrated in order to provide adequate

connectivity for wide-ranging species. The majority of efforts in landscape modeling









concentrate on the types of habitat used by radio-collared animals. These involved

statistical analyses of habitats frequented by large ranging species such as black bear

(Ursus americanus) (Clark et al. 1993), the grizzly bear (Ursus arctos) (Mace et al.

1999), gray wolf (Canis lupus) (Corsi et al. 1999, Haight et al. 1998, Lewis and Murray

1993, Mladenoff et al. 1995), mountain lion (Puma concolor) (Beier 1993), and Florida

panther (Maher and Cox 1995). These statistical models predicted areas critical for home

ranges and dispersal corridors. However, SEMs can develop outputs that provide much

finer detail of predicted habitat use and other factors. Through a series of rules, spatially

explicit models can describe the dynamics of every individual in a population. Simulated

animals going through the program move over a number of digitized and classified

landscapes in ways that mimic known movement patterns of actual wildlife studied in

nature. Landscape features can be represented in a number of different ways depending

on the objectives of the model. Landscape features such as natural communities, roads,

and human densities can be depicted in several different scenarios. The animal

movement part of a SEM can then be tested over different scenarios to better predict a

range of probabilities of animal paths and population sample reactions to changes in the

landscape.

Examples of ecologically based spatially explicit models include BACHMAP

(Pulliam et al. 1992) and ECOLogical-ECONomic (Liu 1993, Liu et al. 1995), both

models of Bachman's sparrow (Aimophila aestivalis) populations in southeastern United

States. OWL is a SEM developed for the northern and California subspecies of the

spotted owl (Strix occidentalis caurina) (Lamberson et al. 1992, McKelvey et al. 1992,

McKelvey et al. 1993, Murphy and Noon 1992, Noon and McKelvey 1992). NOYELD









is a SEM that analyzes the movements of Yellowstone National Park ungulates in

response to fire (Turner et al. 1994). One of the most ambitious projects utilizing SEMs

to date is ATLSS. ATLSS (Across Trophic Level Species Systems) is a set of integrated

models that simulate the hierarchy of ecosystem responses across all trophic levels and

spatial and temporal scales within the Florida Everglades region (Fleming et al. 1994).

ATLSS also incorporates Florida panther movements in south Florida (J. Comiskey,

University of Tennessee, personal communication). Although these simulations are

predictive and often can be tested only minimally, they serve to test our assumptions and

assist in understanding the important dynamics of species and landscape functions.

Predictions generated and understanding gained from spatially explicit models can then

support the conservation and protection of species and landscape processes.


The Florida Panther and Spatially Explicit Models

Spatially explicit modeling is instrumental in predicting movements and

conservation strategies for the wide-ranging Florida panther. Florida panthers have

extensive home ranges. Female home ranges average from 100 to 185 km2 and male

home ranges average 257 to 500 km2, with the largest recorded home range for a male at

1182 km2 (Land et al. 1998, Maehr et al. 1991, Maehr et al. 1992). A population of

Florida panthers requires thousands of square kilometers of undeveloped natural

ecosystems to survive (Maehr and Cox 1995). While these characteristics make it

difficult to study the panther through more traditional methods, they make the panther a

prime candidate as an umbrella species for natural resource conservation. As an umbrella

species, if important home range needs and landscape connections for the panther can be

identified and in turn protected, then other species and ecosystem processes may also be









preserved. Similar strategies have been used for conservation plans in Brazil, using the

jaguar (Panthera onca) (Quigley and Crawshaw 1992).

The only known existing population of Florida panthers (a subspecies of the

puma) is currently in southwest Florida, with approximately 50 individuals (Maehr et al.

1991, Land et al. 1998). A conservation strategy crucial to Florida panther survival is the

reintroduction of other Florida panther populations within portions of its historic range

(U. S. Fish and Wildlife Service 1987). The north Florida-south Georgia region is one

such possible site. Reintroducing a population of panthers there can serve to both help

perpetuate the Florida panther and promote conservation strategies for the north Florida

area.

A spatially explicit model can assist reintroduction efforts by developing

predictions about panther movements under current and future landscape conditions. In a

two-phase project, the Florida Fish and Wildlife Conservation Commission conducted the

Florida Panther Reintroduction Feasibility Study from 1988 to 1995. This study released

and later retrieved radio-collared Texas cougars (Puma concolor stanleyana) in the north

Florida-south Georgia region to determine if reintroducing Florida panthers was

biologically feasible. Information gathered from this study was used in the spatially

explicit model, PANTHER, developed during this research. PANTHER predicted a

range of possible panther movements over five counties in the north Florida landscape in

response to various development and conservation scenarios. The model can be used to

assist in the conservation and human land use planning processes. Model results reveal

areas that panthers and other species would use as suitable habitat, and how projected

human growth and development would affect their use of these areas. These results in









turn may support conservation and regional planners in the selection of public land

purchases, locating off site mitigation projects, providing landowner incentives for

conservation, the implementation of land management strategies, and the promulgation of

planning laws that would benefit panthers and north Florida ecosystems. Model results

may also be instrumental in on-the-ground management of a reintroduced population of

panthers.



The North Florida Setting


Introduction

The objective of this research is to develop and use PANTHER to identify lands

and conservation strategies in north Florida that would be most important to potential

reintroduced Florida panthers. The model incorporates natural and human settings in

north Florida in an effort to produce outputs that suggest conservation actions feasible in

today's world and twenty years into the future.

The study area for this research encompasses five north Florida counties: Baker,

Columbia, Hamilton, Suwannee, and Union (Figurel.1). North Florida is a rural region

that contains extensive tracts of pinelands interspersed with cypress swamps, rivers, and

areas of hardwood forests. The natural quality of much of the area has been affected by

commercial timber operations and agricultural uses throughout the past century, yet these

regenerated pinelands and forested wetlands still provide important habitat for several

wide ranging species such as bobcat (Lynx rufus) and Florida black bear (Ursus

americanusfloridanus). Ecological processes such as river flooding and fires (to a lesser

degree) have continued on a somewhat modified basis. These ecological processes and




























\ mitooBorder
Hamilton Pinhook Swamp


r *Osceola National Forest


Suwannee


Baker

Columbia

7 Union


0 0 30 60 Kilometers
I i


Figure 1.1. The five county study area within the state of Florida.


J~7~i13


N

E
S









native species are affected by human development, land use, and natural resource

management, and will be increasingly influenced and constrained as the region continues

to gain human inhabitants. It is necessary to look at human influences on these processes

when trying to predict animal movements and identify lands crucial for continued

ecosystem function.


Natural Setting

Climate

North Florida lies in an ecologically unique region. This area is located on the

Coastal Plain of the southeastern United States at the southern edge of the temperate

region and at the northern range of the sub-tropics. Warm summers and mild winters

characterize the climate. The average annual temperature is 20.5 degrees Celsius,

ranging from an average high of 34.0 degrees C in the summer to an average low of 4.4

degrees C in the winter. Average annual rainfall is 150 centimeters, with more than 50

percent falling from June through September (Northeast Florida Regional Planning

Council 1996).


Geologic formation

North Florida lies over two physiographic regions, the Ocala Uplift and the Sea

Island Districts. The Ocala Uplift District lies beneath Suwannee, Hamilton, and the

southern portion of Columbia counties. In this district, limestone occurs at or near the

surface, and soils have medium to high clay content (Brown et al. 1990). Soils are

mostly entisols and are dominated by excessively drained sands.









The Sea Island District encompasses the study area east of the Suwannee River,

including all of Baker and Union Counties and the northern two thirds of Columbia

County. Here, the underlying limestone is too deep below the overburden to influence

the landscape or drainage. Soils of the Sea Island District are predominantly spodosols,

which are poorly drained sandy soils with dark sandy subsoil layers (Brown et al. 1990).


Natural communities

Distinct ecosystems within the study area can be characterized as Natural

Communities (Florida Natural Areas Inventory and Department of Natural Resources

1990). Mesic Flatwoods, Wet Flatwoods, Dome Swamp, and Basin Swamp originally

dominated the Sea Island District. Mesic Flatwoods, Upland Pine Forest, and Upland

Mixed Forest originally dominated the uplands of the Ocala Uplift District, while

Bottomland Forest and Floodplain Swamp were found primarily along waterways.

Presently, humans have altered most of these Natural Communities to some degree.

Mesic Flatwoods Natural Community is characterized by periodic fire, which

probably burned every one to eight years. The most common association of plants in this

community include: long leaf pine (Pinuspalustris), wiregrass (Aristida berichiana) and

numerous other herbaceous species in more frequently burned areas (Myers 1990), runner

oak (Quercus pumila), slash pine (Pinus elliottii), gallberry (Ilex glabra), and saw

palmetto (Serenoa repens). The majority of Mesic Flatwoods and Wet Flatwoods in the

study area have been highly altered by commercial forestry practices. Fire return

intervals have been lengthened and fire seasons reversed. Lack of fire during the

lightning season and mechanical harvesting and site preparation have extirpated many









herbaceous species from these natural communities. Animal species requiring uneven

aged open stands of pine have also declined.

Dome Swamp and Basin Swamp Natural Communities occur in depressions in the

landscape and are characterized by extended hydroperiods. Common tree species include

pond cypress (Taxodium ascendens), slash pine, and swamp tupelo (Nyssa biflora). A

portion of these Natural Communities within the study area has been modified by forestry

practices and hydrologic alterations.

Upland Pine Forest Natural Community is characterized as a rolling forest of

widely spaced trees and frequent fire (every three to five years). The dominant species

include longleaf pine and wiregrass. Secondary tree species associated with this

community include loblolly pine (Pinus taeda), southern red oak (Quercusfalcata),

runner oak, and bluejack oak (Quercus incana). This is an endangered Natural

Community. Most tracts have been converted to agricultural and silvicultural use. Very

little (less than ten percent) of the original stands exist today.

Bottomland Forest Natural Community presently occurs within the study area on

low-lying flatlands that usually border streams with distinct banks, such that water rarely

overflows the stream's channel to inundate the forest (Florida Natural Areas Inventory

and Department of Natural Resources 1990). Bottomland Forest also occurs in isolated

low-lying areas in basins and depressions. These forests are inundated only during

extreme floods or after heavy rains. A diverse array of overstory tree species and

relatively few herbaceous species characterize these forests (Platt and Schwartz 1990).

Typical plants include live oak (Quercus virginana), water oak (Quercus nigra), cabbage

palm (Sabal palmetto), red maple (Acer rubrum), sweetgum (Liquidamber styraciflua),









magnolia (Magnolia grandiflora), spruce pine (Pinus glabra), and dogwood (Cornus

florida).

Floodplain Swamp Natural Community within the study area occurs along stream

channels and in low spots within river floodplains, and is flooded for most of the year,

approximately 250 days annually (Florida Natural Areas Inventory and Department of

Natural Resources 1990). Dominant trees such as bald cypress (Taxodium distichum) and

tupelo (Nyssa spp.) are usually buttressed. The understory and ground cover are typically

very sparse. Dominant species include ogeechee tupelo (N. ogeche), water tupelo (N.

aquatica), dahoon holly (Ilex cassine), and wax myrtle (Myrica cerifera). The process of

flooding maintains these forests.


Rivers

A major feature of north Florida is the Suwannee River, hence the name, the

Upper Suwannee Basin (Figure 1.2). The Suwannee is a Blackwater Stream Natural

Community, so named because of its tea-colored waters that are laden with tannins,

particulate, and dissolved organic matter and iron derived from waters that drain from the

swamps and flatwoods of south Georgia and north Florida. It is Florida's second largest

river in terms of length (394 kilometers, with 333 kilometers in Florida), average flow

(304.7 m3/second), and drainage basin (26,641 km2) (Bass 1983). The Suwannee arises

from the Okefenokee Swamp in southeast Georgia and discharges into the Gulf of

Mexico in Florida. The major tributaries of the Suwannee in north Florida include the

Santa Fe and Alapaha rivers. There are also numerous creeks and swamps that drain the

watershed. While the Suwannee drains the majority of north Florida waters to the Gulf of

Mexico, the St. Mary's River flows eastward and drains the remainder of the study area










to the Atlantic Ocean. The St. Mary's is also a Blackwater Stream, with its origins on the

eastern side of the Okefenokee Swamp. It runs a length of 193 kilometers, 161 of them

in Florida. Seasonal discharge from these rivers is greatest during the winter and spring

months (Nordlie 1990).












nFlorida-Georgia Border
[ ) -1 1 H 1 rPIH rn Co .. ..
.1 I I
Rive Alapaha Beehaven Bay '-i-. -- nlin
Alapaha ,ndlin p i,,:,,:,i : ,,l, .
El ,- River Jasper I. y..:,: ,:,1 .111 :i11 F.: t .-s
I ,' .lr, '* .. *.k Suwannee River
S.- -*.- Water Mgm't
S strict Lands

River b ." Osceola
'. National Forest
Live Oak Suwan ee River

rson Maccleny


Suwannee Co. Lake City Baker Co.


Lake Butler F :.e 1, ,:,:

Columbia Co. .
S : Union Co.

B ford rings S.P. Sa Riv er


White .-


Figure 1.2. North Florida major towns and conservation areas.









Human Setting

Human population

The Upper Suwannee Basin of north Florida is a relatively rural area. The five

counties in the study area encompass approximately 7,316 square kilometers, with a 1995

human resident estimate of 126,330 (Smith and Nogle 1996) (Table 1.1). The population

density in 1995 was approximately 17.3 people per square kilometer. The human

population in the Upper Suwannee basin is predicted to grow in the coming years. From

1995 to 2020, there will be an increase of approximately 53,670 people in the five county

area (Smith and Nogle 1996), a 42 percent increase in the population. This gain is greater

than the current population in Baker, Hamilton, and Union Counties combined. In 2020,

the majority of area residents will continue to be located in Columbia and Suwannee

counties.


Table 1.1. Human population estimates of north Florida counties

County 1990 1995 2000 2010 2020
Population Population Population Population Population
Estimate estimate estimate estimate estimate
Baker 18,486 20,275 22,005 24,797 27,600
Columbia 42,613 50,387 55,101 64,399 73,700
Hamilton 10,930 12,487 14,500 16,196 17,900
Suwannee 26,780 30,534 35,398 38,998 44,500
Union 10,252 12,647 13,398 14,899 16,300
Total 109,061 126,330 140,402 159,289 180,000
Number
residents /km2 14.9 17.3 19.2 21.8 24.6

From Smith and Nogle 1996.









Industries

Major land uses within the north Florida study area are agriculture and silviculture

(Table 1.2.). Suwannee and Hamilton counties, located in the Ocala Uplift District,

produce the majority of the row crops grown in the study area, while Union, Baker, and

northern Columbia counties, located in the Sea Island District, produce the majority of

the forestry products. Two mining sites are located in the region: the PCS Phosphate

mine and the DuPont Heavy Minerals mine. The PCS Phosphate mine encompasses

approximately 8,100 ha with mining rights that total 40,400 ha in Hamilton County. The

DuPont mine is approximately 3,000 ha and is situated in portions of Baker, Duval, Clay,

and Columbia counties. Approximately 1,025 ha are located in Baker County. Both

mines are expected to end operations by the year 2020.

There is pressure to extend development into current agricultural areas, especially

farms occupying well-drained uplands. The preservation of agricultural lands is a goal of

the State Comprehensive Plan (Florida Secretary of State 1996). Future agricultural land

uses are expected to diminish in direct correlation to the additional needs projected for

residential and, in some cases, commercial land uses (Hamilton County 1996).

Projections for decreases in agricultural lands by the year 2020 range from 0.2 percent in

Hamilton County (Hamilton County 1996) to 4.0 percent in Suwannee County

(Suwannee County 1996).









Table 1.2. Approximate acreage of generalized land use within the unincorporated areas
of north Florida counties


Existing Land
Use Category

Residential
Commercial
Industrial
Agriculture
Row crops
Forestry
products
Conservation
Recreation
Other
Total
unincorporated
hectares
Total hectares in
county


Baker
County
hectares
1,116
121
25

5,917

75,955
46,767
11
20,704


Columbia
County
hectares
7,410
842
101

41,035

110,353
35,843
1,080
4,828


Hamilton
County
hectares
1,080
120
49

33,135

88,905
3,448
453
1,538


150,618 201,501 128,763

151,500 206,400 133,400


Taken from County Comprehensive Plans of Baker (1996),
(1996), Suwannee (1996), and Union (1996) Counties.


Columbia (1996), Hamilton


Pine plantations, lands used for timber production, are under little development

pressure. Future trends suggest that expanding wood product markets and increased

demand will place an incentive on the forest products (timber) industry to sustain, if not

increase, future production (USDA 1996). Overall forest coverage is not expected to

decline appreciably. Silvicultural areas in north Florida are located predominantly on

poorly drained soils which are unsuitable for home and industry development, and are

usually not considered high priority for development.


Suwannee
County
hectares
6,030
486
243

81,232

75,579
2,104
1,012
3,080


169,207

178,200


Union
County
hectares
1,117
162
0

8,628

48,226
0
16
4,872


63,022

63,200









Conservation

Public agencies and private corporations own and manage land holdings of

significant size within the study area (Table 1.3, Figure 1.2). Most public agencies are

involved with land acquisition programs that target areas within north Florida. The one

exception, the Florida Fish and Wildlife Conservation Commission, does not own land in

the area, but works with private landowners to designate properties as Wildlife

Management Areas (WMA) for public hunting. The primary land use is continued,

which is typically timber production. The Florida Fish and Wildlife Conservation

Commission also manages certain public parcels as WMAs for hunting purposes.

The Osceola National Forest-Okefenokee Swamp complex of federal lands

encompasses 242,811 hectares, and is the core of all conservation strategies for the

region. The presence of the Osceola-Okefenokee complex was an important factor in

designating north Florida and south Georgia as a potential Florida panther reintroduction

site.


Regional Land Use Planning in Florida

Introduction

An additional part of this comprehensive analysis of the north Florida landscape

addresses regional human population growth and development. While the main objective

of this synthesis is to predict panther movements over the current landscape, a secondary

objective is to predict movements in response to human growth and development changes

over time. Three possible future scenarios for the north Florida landscape were









Table 1.3. Protected lands of significant size (>1,000ha) in north Florida and their
respective managing agencies

Name Hectares & Managing Agency Uses/Comments
County


Osceola National
Forest


Okefenokee
National Wildlife
Refuge

Big Shoals State
Preserve




State Parks:
River Rise
Oleno

SRWMD Lands


Wildlife
Management
Area Units

Public:
Raiford State
Prison
Holton Creek

Private:
Lake Butler


41,538 Baker
38,722 Columbia
Total 80,260

161,063 GA
1,488 Baker


1,416 Hamilton







1,655 Columbia
1,729 Columbia

4,796 Hamilton
2,025 Columbia
3,838 Suwannee


3,642 Union
1,012 Hamilton


12,587 Union


U. S. Forest Service


U. S. Fish and
Wildlife Service


Suwannee River
Water Management
District and Florida
Division of Forestry


Florida Department
of Environmental
Protection

Suwannee River
Water Management
District

Managed by Florida
Fish and Wildlife
Conservation
Commission


Predominantly
timber production,
multiple use

Wildlife and
hydrologic
function

Forestry,
Recreation,
Hunting,
Hydrologic
functions

Recreation,
hydrologic
function

Timber, recreation,
hydrologic
functions

Public Hunting


Prison Acreage
SRWMD


Private timber
lands, primarily
for timber
production, not
conservation
purposes









developed through analyses of comprehensive planning laws, regional and county

strategic plans, human population projections, and state listed potential conservation

lands. These settings were designed to incorporate the increase in projected human

population through the year 2020, predict where increased human infrastructures would

be placed on the landscape, and add listed potential conservation lands. To best create

these scenarios, it is important to understand and communicate the planning and

development process with respect to the conservation of natural resources.


Regional and County Comprehensive Planning

It is essential for scientists concerned with the protection of wide ranging

carnivores such as the Florida panther to understand policy-making processes in order to

best integrate scientific knowledge into such processes (Primm and Clark 1996). In

Florida, land-use policy is created at the state level in the Department of Community

Affairs. The comprehensive plan for the state is modified in each region by one of eleven

regional planning councils and adapted at the county level. Comprehensive plans have

evolved into strategic plans, which are developed through a collaborative process. These

plans are "long range guides for the physical, economic, and social development of a

planning region, which identify regional goals and policies" (Florida Secretary of State

1996). They serve as the basis for growth and development regulation by regional

planning councils, counties, and municipalities. Comprehensive and strategic plans can

conserve species and environmentally sensitive lands when development is executed in

accordance with regulatory decrees.

Strategic and comprehensive plans incorporate data on flora and fauna species'

locations, habitat requirements, and other natural resource requirements that come from









"professionally accepted sources such as the State University System of Florida, original

data, or special studies" (Florida Secretary of State 1996). Information on regional

conservation related subjects is needed in the development of these plans, and data from

the Florida Fish and Wildlife Conservation Commission is often the only source used. In

fact, the Florida Fish and Wildlife Conservation Commission identified Strategic Habitat

Conservation Areas that are crucial in identifying Natural Resources of Regional

Significance in regional strategic plans (Table 1.4).

While strategic plans appear to guide the development process, they must balance

scientific conservation concerns for natural areas with human desires. Once strategic

plans are adopted they are law, and restrict land development in natural areas that are

deemed environmentally sensitive. The difficulty associated with these plans concerning

natural resource protection is that these laws are loosely defined and easily amended by

county commissions. As a result, portions of county strategic plans, such as goals, are

written in ways that can not be quantified or validated, or have no regulatory authority.

For example, the Northeast Florida Regional Planning Council's Strategic Policy Plan

addresses the rights of private property owners. The policy states, "Land uses within

identified Natural Resources of Regional Significance which are compatible with the

habitat conservation needs of listed species shall be encouraged" (Northeast Florida

Regional Planning Council 1996). Compatible land uses are not defined. As a result,

there are liberal interpretations of these laws, subject to the will of the people enforcing

them in each county.









Table 1.4. Natural Resources of Regional Significance as identified by the North Central
and Northeast Florida Regional Planning Councils in their Strategic Plans


Name
Suwannee River

Santa Fe River


Olustee Creek

Beehaven Bay

Osceola National
Forest/Pinhook
Swamp
Okefenokee National
Wildlife Refuge
Water Management
District Lands
Water Management
District Easements
Stephen Foster State
Cultural Center
Big Shoals Tract
Nature Conservancy
Lands
Little River
Falling Creek
Strategic Habitat
Conservation Areas
(Sandlin Bay, Pinhook
Swamp & Impassable
Bay)
Strategic Habitat
Conservation Areas
(Moccasin Swamp,
Cross Branch, North
Prong of the St.
Mary's River)
Strategic Habitat
Conservation Areas
(Areas along Santa Fe
& Suwannee Rivers,
and Olustee Creek)
Jasper Ridge Trail


Hectares Counties
53,858 Columbia, Hamilton,
Suwannee
338 Union, Columbia


50 Union

2,883 Hamilton

80,260 Columbia, Baker


1,499 Baker

10,659 Columbia, Hamilton,
Suwannee
154 Suwannee,
Columbia, Hamilton
85 Columbia

1,416 Hamilton
1615 Columbia, Hamilton,
Suwannee, Baker/
8,865 Suwannee
6,261 Columbia
Columbia


Baker


Classification
River Corridor

Surface Water
Improvement Management
Waterbodies
Surface Wa. Improv. Mgt.
Waterbodies
Surf. Wa. Impr. Mgt.
Waterbodies
Fresh Water Wetlands
And Public Lands

Public Lands

Public Lands

Public Lands

Public Lands

Public Lands
Private Lands

Steam-to-Sink Watershed
Stream-to-Sink Watershed
Florida Fish and Wildlife
Conservation Commission



Florida Fish and Wildlife
Conservation Commission


Union, Suwannee,
Columbia and
Hamilton


Florida Greenway


Hamilton










Table 1.4.-contined.
Name Hectares Counties Classification
Upper Suwannee River Columbia, Hamilton, Florida Greenway
Greenway Suwannee
Osceola National Forest Columbia Florida Greenway
National Scenic Trail
Pinhook Columbia Florida Greenway
Swamp/Okefenokee
Greenway


Each county has a process of permitting development in environmentally sensitive

areas. State agencies and county commissioners grant final decisions on development

permits. These factors make the local constituency of a county quite crucial to the

creation of conservation goals and the policies necessary to enforce them. This situation

makes it critical to study regional and local development projections on a county by

county basis. Four of five counties in the north Florida study area are members of the

North Central Florida Regional Planning Council, while Baker County is a member of the

Northeast Florida Regional Planning Council. Each county is different in their

commitment to conservation related areas and how growth will be directed. These are

major factors in planning conservation strategies for these areas. The following individual

county descriptions address the existing publicly owned conservation areas within each

county, the county's commitment to natural resource protection, human population

projections, and how growth is predicted to occur. These issues are crucial to predicting

human growth and conservation strategies for the region.


Baker County

In Baker County 26.7 percent of the land is federally protected in either the

Osceola National Forest or the Okefenokee National Wildlife Refuge. There are also









approximately 2023 ha of Pinhook Swamp in Baker County that are scheduled to be

acquired and added to the Osceola National Forest. The St. Mary's River is the only

other listed Natural Resource of Significance in the county. Baker County supports

additional protection of areas along the river through state initiatives.

Between 1995 and 2020, Baker County is expected to gain approximately 7,325

new residents (Smith and Nogle 1996). Approximately 6192 ha or four percent of the

county's total land is under consideration for development to accommodate increased

residential land use. This growth is projected to occur within existing principal centers

such as the east-west Macclenny and Glenn St. Mary's corridor along US 10, the towns

of Sanderson and Olustee, areas north of Macclenny and Glen St. Mary's along State

Highways 23A and 125, and smaller sites at crossroads and earlier established

subdivisions (Baker County 1996). Much of the recent development occurring in Baker

County has been along the banks of the South Prong of the St. Marys River and its

tributaries. Although the soils in this area are poorly drained, continued development is

expected. Much of Baker County's soils poses severe limitations for septic tanks, and are

prone to flooding. Development appears to be limited due in part to these conditions and

the ability of public sewer systems to expand into more rural areas.

There will be some mining influence in the county, as the existing DuPont Heavy

Minerals mine is expected to encompass approximately 10.3 km2 within eastern Baker

County.


Columbia County

Columbia County contains the Suwannee River, the Santa Fe River, and the

Osceola National Forest. Designated conservation areas within the county include









Suwannee River Water Management District Lands, River Rise State Preserve, and the

Osceola National Forest. These areas encompass 35,843 ha, or 17 percent of the county.

Areas within the county designated by the Regional Strategic Plan as areas of Significant

Natural Resources include Falling Creek, Santa Fe River, Olustee Creek Corridor, and

the Florida Fish and Wildlife Conservation Commission's Strategic Habitat Conservation

Areas. A goal of the Columbia County Comprehensive Plan states, "Conserve, through

appropriate use and protection, the resources of the county to maintain the integrity of

natural functions." Policy V.2.3 states that "The county shall identify and make

recommendations, where appropriate, for the purchase of environmentally sensitive

lands, especially Alligator Lake, the shoreline and floodplains of Falling Creek, Deep

Creek and Robinson Branch, by the state of Florida, Water Management Districts, or the

U.S. Government. The county shall also apply for federal and state funds to purchase the

above environmentally sensitive lands" (Columbia County 1996).

From 1995 to 2020, the county is expected to gain approximately 23,313 residents

(Smith and Nogle 1996). Ninety-eight percent will settle outside of Lake City and Fort

White and in the unincorporated areas of the county (Columbia County 1996). The

population of unincorporated areas is projected to be 43,850 by the year 2020. Estimates

project it will take 12,367 ha of land to accommodate new residential development

projected to occur in the county (Columbia County 1996).

Four types of growth have occurred within the county in recent years that may

help indicate where future growth will occur. The first emerging pattern is growth

concentrated within existing public facility service areas immediately surrounding

municipal urban areas. The second form of growth has concentrated development around









unincorporated market centers, which despite the lack of public facilities have developed

into urbanizing settlements. The third pattern is residential lot development along the

Suwannee and Santa Fe Rivers. The final pattern is radial growth along major roadways

throughout the county (Columbia County 1996). The most efficient growth pattern for

the county as a whole, as described by the Columbia County Planning Agency, is growth

concentrated within the geographic service areas surrounding Lake City. These areas are

within the Lake City urban development area, are not affected by environmental

development constraints, and are targeted for future urban expansion. Development

constraints such as unsuitable soils will limit development expansion in the north-

northeastern portions of the county. An estimated 700 ha of land within the County are

undeveloped, previously platted lands within flood prone areas.


Hamilton County

Hamilton County has only two designated conservation lands. These include

Suwannee River Water Management District Lands that total 3,838 ha, and a portion of

Suwannee River State Park. Areas within the county designated by the North Central

Florida Regional Planning Council as Natural Resources of Regional Significance

include the Suwannee River, Rocky Creek, Beehaven Bay, Florida Fish and Wildlife

Conservation Commission Strategic Habitat Conservation Areas, the Suwannee River

Trail, the Jasper Trail, and potential Greenway Trails. Hamilton County planners have

not created strategic plan policies to support acquisition of additional environmentally

significant lands.

From 1995 to 2020, Hamilton County is projected to add 5,413 new residents

(Smith and Nogle 1996). The County predicts approximately 4734 ha will be needed in









the unincorporated areas of the county to accommodate the new residents (Hamilton

County 1996). The Hamilton County Comprehensive Plan does not indicate where

growth is predicted to occur. Approximately 550 ha of undeveloped platted lands are

located within flood prone areas. The vast majority of new development is directed by

the North Central Florida Regional Planning Council to occur in designated urban

development areas. These are the county's incorporated areas that include the cities of

Jennings, Jasper, and White Springs. Areas outside urban development areas are directed

by the Comprehensive Regional Plan to maintain their existing rural character. The

county strategic plan states development will occur, and does not direct its growth, yet

envisions rural areas to continue to maintain their rural character. This may make

protection of lands along the Suwannee difficult in part since the county strategic plan

states that areas along the Suwannee River and its tributaries are under the greatest

development pressure.


Suwannee County

The Suwannee River is a major feature of Suwannee County, constituting its

northern, southern, and western borders. Conservation Lands in Suwannee County

encompass 2,104 ha. The majority of these lands is located adjacent to the Suwannee

River and owned by the Suwannee River Water Management District. Areas deemed

environmentally sensitive include the Santa Fe and Suwannee River Corridors, the

Ichetucknee River and stream to sink recharge areas near these major rivers.

The Suwannee County Comprehensive Plan does not indicate a specific policy for

identifying and securing environmentally sensitive lands. Policy V.2.3. states that the

County "shall identify and make recommendations where appropriate for the purchase of









environmentally sensitive lands" by state and federal agencies, but it does not identify

particular areas. Other aspects of the Comprehensive Plan show a slight weakening of

the policies stated by other counties in the North Central Florida Regional Planning

Council. These policies (such as Policy V.2.14) exempt agricultural uses from standards

such as restrictions on development along the Suwannee River.

Between 1995 and 2020, the county is expected to grow by almost 14,000 people

(Smith and Nogle 1996). Live Oak and Branford are the two incorporated areas in the

county and will add approximately 508 people within their city boundaries. The rest (94

percent) will move into rural areas of the county. An estimated 10,651 total additional ha

of land will be needed between 1996 and the year 2011 to accommodate the residential

development projected to occur within the County (Suwannee County 1996). Mandated

densities of dwelling units for agricultural zoned lands in Suwannee County are reduced

from four categories to just one, which is the most highly populated density. This allows

greater development on agricultural lands. The Suwannee County Comprehensive Plan

indicates (Objective 1.1) that the total area of all the County's urban development shall be

limited to 10 percent of the total acreage in the County.

Growth projections for Suwannee County are similar to those for Columbia

County. The four types of growth patterns (within municipal areas, unincorporated

market areas, residential lots along the rivers, and radial growth) occur here, with

residential lots occurring along the Santa Fe, Ichetucknee, and Suwannee Rivers. The

vast majority of new development within the county "should" occur in designated urban

development centers, which comprise the County's incorporated areas and

unincorporated urban settlements (Suwannee County 1996).









A significant portion (87 percent) of undeveloped lands that were platted before the

County's entry into the National Flood Insurance Program is within flood prone areas. It

is estimated that 1,926 ha of flood prone areas have been platted for future subdivision

(Suwannee County 1996).


Union County

Union County has no designated conservation lands. There are two Wildlife

Management Areas, the Lake Butler and the Raiford tracts, which encompass 26,337 ha

within the county. The Lake Butler tract is owned and operated by wood processing

corporations and is managed for timber. The Raiford tract is home to the Florida State

Prison and public timber production lands. They consist predominantly of pine

plantations and have little resemblance to flatwoods natural communities. Protection of

these lands is partly dependent on the success of the wood processing industry in the

coming years. The only designated Environmentally Sensitive land is the 100 year flood

plain of the Santa Fe River. The Union County Comprehensive Plan does not indicate a

specific policy for identifying and securing environmentally sensitive lands. The only

area the county has made a conservation commitment to is in regulation of development

along the 100-year flood plain of the Santa Fe River. The county comprehensive plan

states that the total area of urban development shall be limited to 10 percent of the total

acreage within the county. Higher density development shall be restricted to areas

adjacent to arterial or collector roads.

Most of the past growth has occurred in unincorporated areas of the county, which

are still primarily rural. These areas have experienced the previously listed four types of

growth. Undeveloped platted lands within the flood prone areas encompass 161 ha. The









designated urban development areas will provide the primary location of new

development within the unincorporated areas of the County (Union County 1996). From

1995 to 2020, it is estimated that there will be an increase of 3,653 new residents (Smith

and Nogle 1996) who will need a total of 1,296 additional ha to accommodate residential

development within unincorporated areas of the County (Union County 1996).


Dissertation Structure

The objective of this research is to use the spatially explicit model PANTHER to

identify conservation strategies for north Florida through the movements of a wide

ranging carnivore and to understand the synergism of how different variables affect

panther movement and survival. Figure 1.3 is a flow diagram of my research synthesis.

The movement model combines data from Florida panthers in south Florida and an

experimental population of Texas cougars in north Florida. The model incorporates

Geographic Information Systems (GIS) computerized maps, representing natural

communities, roads, white-tail deer (Odocoileus virginianus) densities, human densities,

and human attitudes in the five county study area of north Florida. Landscape scenarios

are proposed and altered to predict a variety of changes involving human populations,

development and land-use over time, increased and decreased amounts of conservation

protection, volume and density of roads, and industrial uses of land. Model simulations

generated a range of predictions of how these factors will affect the area's ecosystems

and wildlife. Key landscape features and possible conservation approaches are identified

for protecting Florida panthers and the ecosystems they would rely on if reintroduced to

north Florida.









Chapter 2 of this dissertation describes the process of creating this research

synthesis. It details the development of the movement model, the creation of GIS layers,

and the methods used to represent future scenario conditions. Chapter 3 describes the

development of the model through calibrations and sensitivity analysis, the variables that

most affected panther movements, and the results of the simulations of the model.

Chapter 4 discusses the implications of research results. Chapter 5 summarizes the entire

synthesis.



Panther Human and
Movement Natural
Model Landscape
Setting







Synthesis





Predictions of panther movement
Recommendations for conservation,
management, planning
Maps of landscape linkage hotspots
Knowledge, understanding


Figure 1.3. Flow diagram of research synthesis















CHAPTER 2
MODEL METHODS


Introduction


Overview

One of the requirements of a successful spatially explicit model is a full

communication of the development of that model (Bart et al. 1995, Swartzman 1979,

Lorek and Sonnerschein 1998). Chapter 2 gives a full account of the development of the

PANTHER model, beginning with a detailed description of how spatially explicit models

are generally developed. The creation of this synthesis is then documented.

Documentation is grouped into three sections: the development of the movement model

in C++ programming language, the quantification of the landscape through GIS, and the

process of identifying two future scenarios of the north Florida setting. The chapter

concludes with a description of the methods used to quantify and analyze the results of

the model.


Spatially Explicit Models

The PANTHER model developed for this synthesis has a structure similar to

previous models. There is an animal movement component developed in C++ computer

language (Ellis and Stroustup 1990) and a landscape based system of computerized maps.

In this model, as in others, movements are based on population interactions, dispersal

behavior, movement rates, habitat selection, and prey preferences and presence (Conroy









et al. 1995, Dunning et al. 1995). The computerized landscapes over which animal

movements are simulated are composed of individual cells, scaled to the species of

concern and resolution accuracy of GIS databases. These cells are assigned numerical

attributes pertinent to the landscape, the species of concern, and the objectives of the

model. Values within cells relate information according to preferred and avoided natural

communities, intraspecific presence, food and water sources, and human factors such as

human densities and the presence of roads. The status of each individual is followed

through the entire simulation (Dunning et al. 1995) and every move recorded to an output

file. Multiple simulations (Monte Carlo) are run to generate a range of prediction

possibilities. These predictions are related to the objectives of the model and may

include how the organisms successfully or unsuccessfully move over the landscape, find

food and interact with other individuals. A unique characteristic of spatially explicit

models is that they can incorporate individual organisms' movements between specific

patches across the landscape and quantify how this movement may affect modeled

population dynamics. Individual attributes and rules of movement can be based on

appropriate empirical data, including a certain degree of randomness which allows for a

range of outputs.

Several authors have suggested a series of guidelines for model development and

testing (Swartzman 1979, Bart et al. 1995, Conroy et al. 1995). These include clearly

stated objectives of the model, a detailed description of the model's general structure and

organization, and the sequence of steps it carries out to make its predictions. The

parameter values used in the model should be identified by reference. The specific

functional mechanisms, such as the equations used and the underlying processes that are









assumed to control aspects of the model should be well communicated and calibrated.

Next, the model should undergo sensitivity analysis to assess the effect of uncertainty of

variables that have a substantial affect on model outputs, or that are poorly known. Final

validation is accomplished by comparing simulation results with on-the-ground data.

However, quantitative validation, when possible, however, only addresses how well the

model mimics historical situations. Validation is a process of exploring the limits of

model credibility (Clark et al.1979). What makes a model most useful is its ability to

explain phenomena. The best models explain low probability events at the extremes or

limits of situations, often events that could not be experimented for on the ground, such

as future conditions.


The Conceptual Model

The objective of this synthesis is to use a spatially explicit model to predict

panther movements in north Florida, which in turn can be used to develop conservation

strategies. The first goal in the development process was to construct a dynamic

hypothesis (Montague et al. 1982). The hypothesis predicts parameters of importance in

the general movement patterns of panthers observed in studies in south Florida (Land et

al. 1998, Maehr et al. 1991) and cougars placed in north Florida (Belden and McCown

1996). After this hypothesis was constructed it was tested with the aid of an individually

based spatially explicit model, which specified the location and movements of every

member of the population. The model was calibrated and tested through sensitivity

analysis, and validated conceptually with statistical estimates obtained from the studies

cited above.









The dynamic hypothesis for the synthesis is that panther movements are

influenced by landscape (habitat) variables, personal attributes, past history, and the

population of panthers in the area (Figure 2.1). The landscape is best represented by

quantifying the attributes of natural communities, deer (prey) densities, human densities,

and roads. Each landscape image is divided into 30 by 30-meter pixels (cells). Panthers

evaluate their surroundings based on a pixel by pixel decision process. Personal

attributes are best represented by the panther's gender, residency status (transient or

resident), how fast an animal moves, its preference for moving in a specific direction

(rather than randomly), and where it has been before (history). The population of

panthers affect an individual panther's movements through scent marking of territories

and actual presence of the specific panther. Each panther reacts to members of the

population based on their gender and residency status, and past interactions with one

another.

These parameters are combined to influence how each panther moves. The

spatially explicit model quantifies this conceptual hypothesis through a series of rules of

movement (discussed below). The model is the mechanism by which the dynamic

hypothesis was tested. Through calibration and sensitivity analysis of the model, this

hypothesis was continually adjusted to compensate for changes to initial assumptions.

The final model hypothesis was developed after validation of the model. This final

hypothesis is represented by model parameters that were used in simulations of several

different landscape scenarios.











C++ Code
Previous Directi
Position



Intra-Specific
Panther Interactions
Position


Landscape Images


Figure 2.1. Flow diagram of PANTHER model.











Movement Model Mimicking Florida Panther Movements


How the C++ Model is Structured

The PANTHER model is a decision-based movement model developed using

Microsoft Visual C++ computer programming language. C++ is an object-orientated

language, which allows the developer of the model to nest information into protected

classes (or objects), thus eliminating errors stemming from certain commands from one

section of the model interfering with other sections. This allows for control over how

information is shared between model components. The model was run on a Windows NT

personal computer. A PANTHER class was created, to represent all personal, internal

parameters considered important to panthers in this synthesis. The PANTHER class,

served as a template for four subclasses which represent combinations of genders

and two residency statuses: Resident female, Resident male, Transient female, and

Transient male. Three actions were associated with the PANTHER class: Look, Choose,

and Move. A simulated panther enters its PANTHER subclass and first Looks at the

landscape and the positions of all other panthers, evaluates its position in relation to past

moves, Chooses which neighboring cell it will move to next, and Moves to that cell.

These actions involve a series of complex interaction loops between different classes of

information and individual panthers. The LANDSCAPE class assists the PANTHER

class in analyzing information from the landscape images. It utilizes CERDASGIS class

to access and read in landscape image data. The MEMORY class is used by the

PANTHER classes to store, retrieve, and output past moves. Several other minor classes

are used to facilitate these class communications (Figure 2.2.).










Philosophy of Model Settings

In order to facilitate the model's evaluation of the landscape, a system of

quantifying landscape factors believed to be important to panthers was established. Each

landscape image was processed to produce an individual layer index value for each cell.

Individual cell values were summed across Landscape Images to produce an overall

habitat index score, with a maximum of 100 points per cell. Quantitative values allow the

computer to quickly evaluate each pixel under consideration. This habitat index value

was more heavily influenced by natural communities, and was also influenced by home

range values (discussed below) (Table 2.1). Each panther assessed these values against

rules of movement before making a move. Panthers did not actually learn over time, but

had increasing amounts of memory concerning where they had been as the model

simulation progressed. Panthers made final movement decisions based on the habitat

index values, past moves, and the presence or absence of other panthers. The neighboring

cell with the highest summed habitat index score was the next move choice for a panther

progressing through the model. In the event that more than one neighboring cell had the

same highest value, the panther chose randomly among the highest value cells.
















































Output Files


x, y coordinates


Stats


GIS Comments


Figure 2.2. Full flow diagram of PANTHER model (minor classes excluded).









Legend to Figure 2.2.


1. MAIN class (object) sends a message to Landscape class (object) to call up
CERDASGIS class (object). CERDASGIS class prompts user for input files, then
opens those files, and reads them into computer memory.

2. MAIN calls up each panther to cycle through each function. Movement decisions are
made within the three functions of the PANTHER sub-classes (inherited classes or
objects): Look, Choose, and Move.

3. Look function calls up LANDSCAPE class to download landscape attributes found in
landscape images into panther inherited class.

4. The Choose function calls up MEMORY class functions to look at past moves of
panther. MEMORY calls up class (object) Memrect to analyze memory rectangles
built around past coordinates.

5. The panther checks present position and past moves of all panthers in population, in
relation to its current coordinates it uses PANTHER class and MEMORY class
functions.

6. The panther Moves to new coordinates, and sends past coordinates to MEMORY to be
stored on move list.

7. After entire simulation is finished each panther calls up MEMORY to download move
information to the four output files.

8. MEMORY downloads x, y coordinates, information on mating, fighting, road crossing,
private and public land usage, number of times a GIS cell was used by panthers, and
other information to output files.









Table 2.1. Initial ranking of habitat variables of neighboring cells.


Factors (Parameters) Percentage of Total Value
Natural Community 40
Roads 7
Deer 13
Human Density 15
Home Range (Residents only) 25
Total 100


The model produces four output files. The GIS output file is a landscape image of

the entire study area with each pixel coded for the number of times a panther entered that

cell. This allows for a detailed analysis of landscape use. The x, y coordinate output file

lists every move made by each panther. A statistics file tallies the movement

characteristics of each individual and records events such as type of natural communities

selected, x, y coordinates, and road crossings. Lastly the comment file describes the

number of steps each panther took, private and public land selections, road crossings,

type of roads crossed, types of natural communities selected, panther interactions such as

matings and fights. The file also notes if and how the panther was killed during the

model simulation, or if it remained alive at the end of the program, and several other

factors.

Each subclass (Resident female, Resident male, Transient female, Transient male)

reacts differently to outside and internal parameters in different ways. For instance,

transient males move 40 percent more in a given day than resident females. Resident

males are programmed to seek females above all other priorities, and if they detect one,

give precedence over habitat variables to tracking her. Detailed descriptions concerning

differences between panther subclass reactions to model variables follow.









Rules of Movement

When animal movements are analyzed an animal's reasons for movement choices

are never entirely known. As a result, animal movements may appear random, but in fact

may be very discerning and distinct to the animal (Montgomery 1974). Pumas have

been shown to most often travel with some general overall direction to their path, cover

large areas, show some site fidelity to home ranges, and display individual variation in

their movement over the landscape (Beier 1995, 1996, Maehr et al. 1991, Maehr and Cox

1995, Seidensticker et al. 1973). In the PANTHER model movement decisions on how

an animal moves are based on assumptions about what is most important to a panther's

movement, and how those factors can be quantified. The model is not deterministic and

allows for some degree of randomness at specific decision points to mimic events and

reasons (parameters) not considered in the model, or that are unknown. The basic rules

of movement process is summarized below.

1. The MAIN class in the program reads in input files and ascertains each panther's

gender and residency, and then assigns the panther to a specific subclass within the

model, either resident female, resident male, transient female, or transient male.

2. As each panther on the input list is called up, it enters the Look function. The panther

must decide to take a step or skip a step (this creates different movement rates). The

model queries the random number generator for the current random number (a

number between one an one hundred, seeded with the time). If the number is above

the panther's assigned threshold for skipping steps it skips the step. If the number is

below the threshold, the panther proceeds.









3. The panther determines its x,y coordinates, and those of its 24 neighboring pixels. It

then queries the LANDSCAPE class for the habitat index values of each cell at a

particular x,y coordinate for each landscape image.

4. The panther stores the information about habitat index values of each neighboring cell

in the computer's temporary memory. The panther evaluates the landscape images

based its on residency and gender. The panther then leaves the Look function.

5. The panther enters the function Choose. Habitat index values from every landscape

image are summed for each neighboring cell.

6. The panther compares all neighboring cells against its last 10 moves (1000 moves for

males). If a cell was previously visited in the last 10 steps (1000 for males), it is

down graded in points. If it was a cell next to a cell visited in past 10 moves, it is also

down graded, although by only half as much as previously visited cells.

7. If the panther is a resident it checks its neighboring cells with a series of home range

questions. It compares its moves in the past 60 days with neighboring cells. If a cell

was visited in the past 60 days, it is given several more points to its habitat index

score, based on a series of rules pertaining to how long ago the cell was visited.

8. The panther checks its neighboring cells for signs of other panthers. The model builds

a ten-kilometer box around the panther. The panther then checks to see if any of the

other panthers are currently within the bounds of the box. If no other panthers are

detected, the panther skips the entire routine dealing with other members of the

population.

9. If a neighboring panther is detected within ten kilometers, the current panther queries

the memory of the neighboring panther's list of past moves, and finds out if the









detected panther has used any of the neighboring cells in the past 1000 moves. If the

other panther has not used these cells, the current panther skips further questions

concerning this neighbor panther, and looks for signs of the next panther on the list of

population individuals.

10. If the current panther detects sign of the nearby panther's path, it enters a series of

decision rules on how to interact with the neighboring panther. If it is a same gender

resident, the current panther will avoid it. If it is the opposite gender, the panther

will be attracted to it. If the current panther is female, it will ignore transient females.

If the current panther is male, it will pursue a transient male.

11. These rules of attraction and avoidance are enacted by adding or subtracting values

from the habitat index value of specific cells where other panthers' paths are detected,

or where the panthers presently reside.

12. Once a panther has finished stepping through the rules of interaction among panthers,

it chooses the neighboring cell with the highest habitat score as the place for its next

move. In the event more than one cell is of highest value, a random number is

summoned to finalize the choice.

13. Once this new cell is chosen, the model compares the x,y coordinates of this cell with

the landscape images to determine the status of this cell in relation to the types of

natural communities, roads, deer densities, human densities, and type of

landownership (public or private). Counters within the model tally the different

groupings of these types of factors.

14. The panther then exits Choose and enters Move, where its current coordinates are sent

to a memory list and the new x,y coordinates are assigned as the new position.









15. Once all panthers in the population have completed this process, the model takes

another time step, and the process is repeated until the maximum number of steps is

reached (typically 125,000 time steps).

16. Once all panthers have completed their steps, the model sends messages to each

panther for them to dump their x,y coordinates of every move to an output file. The

panthers are also instructed to dump the character strings (which are text describing

their movements) to another output file.

17. The model finishes by deleting all information held in the computer's temporary

memory.


Perception Distance

A major component of any model is the ability of the individuals to perceive their

environment. The distance a puma can see, smell, or hear environmental variables and

sign of other individuals' paths or presence can be quantified in the model by how far out

it can survey the neighboring cells of the landscape images. Distance perception in the

model is 60 meters or two 30 meter cells (Figure 2.3). This value was chosen because it

represents the assumed sight and sound distance for the immediate neighborhood of a

panther. In addition, 60 meters was the maximum value that allowed computer run time.

With perception distances of 90 meters or more, the number of neighboring cells a

specific panther must check for all landscape image values and against all other panther

locations becomes exponentially greater (24 versus 47 neighbors), and greatly increases

the time required for model computation.
























Figure 2.3. Panther perception distance of neighboring cells.


The average distance at which one Florida panther can perceive another panther is

not known. The maximum distance panthers can detect one another is based on the

ability to detect urine markings and vocalizations. Montgomery (1974) modeled red fox

(Vulpes vulpes) movements, with the foxes able to perceive one another at 800 meters.

While the length of time a scent remains in wolf urine is estimated to be 23 days under

winter conditions in Minnesota (Peters and Mech 1975), such estimates have not been

published for panthers.

As model development progressed, a second routine was included in the program

to handle actions of resident males seeking females. With the perception neighborhood

of 60 meters, resident males were not able to detect and visit with area females within the

same amounts of time as observed in wild populations. In wild populations of panthers

and other large felids a resident male will visit with all females in his territory within

every four to six weeks (Sunquist, personal communication). To best mimic this

behavior in the PANTHER model, a male's perception of the general direction a female

was located in was increased from 60 meters to 10 kilometers. Ten kilometers was


30m
60









chosen because at smaller distances, males would not perceive and locate females in the

majority of runs representing an eight-month time span, even when the females were

within two kilometers. With a perception of 10 kilometers, the resident male begins

moving in the general direction of a female. This action was not intended to mimic scent

perception, but to represent a male's knowledge of where specific females have

established home ranges, and occasional vocalizations of often young females which can

be heard for kilometers (M. Sunquist, personal communication).


Direction of travel and backtracking

Pumas are not known to directly backtrack unless confronted with a sea of

inhospitable habitat entered through a peninsula of more natural habitat (Beier 1993,

1996). In this model, inhospitable habitat are areas where the habitat index values of

cells are lower than others because of roads, high human densities, and less preferred

natural communities such as agriculture, urban and barren lands. The probability of

panthers entering these areas is low, with males having a slightly higher probability of

entering these areas than females. The PANTHER model is structured to discourage but

still allow backtracking. To prevent a simulated panther from easily backtracking over its

last ten steps, the habitat index values of each of these steps are devalued according to

how recently they were used (the most recent cell visited is devalued by a greater number

of points than a cell visited ten steps back). The model is also programmed to encourage

panthers to continue moving in the same general direction as previous moves. Turns of

90 degrees or more from the previous direction of travel are discouraged by devaluing the

eight cells which lie adjacent to the cell previously used (Figure 2.4).
















Current
position

Last
-10 position -10

Position
- 8 -10 2 steps

Positionf
- 8 3 steps -10
back


Figure 2.4. Devaluation of cells based on previous moves.





Methods for cell devaluation are :

For cells visited in the past 10 steps: Devalue cell index value by (21 minus number of

steps back)

For neighboring cells adjacent to cell visited in past 10 steps: Devalue cell index value by

(11 minus number of steps back)

For cells that were visited both ten steps back and are within the eight cell neighborhood

of past moves: Devalue cell index value by the greater of the two values.

Differentiation among panther types in movement patterns is achieved by varying the cell

devaluation values (Table 2.2). Sensitivity analysis indicated that these rankings play a









major role in making males cover large areas, while forcing females to stay in smaller

movement patterns.


Speed of travel

The speed with which Florida panthers travel the landscape depends on many

factors. In the model, speed translates into how many steps a panther takes in a day,

which is determined by a panther's gender and residency status. The maximum known

distance a transient male Florida panther has traveled in 24 hours is approximately 20

kilometers (Maehr et al. 1991, Maehr and Cox 1995). Other puma studies have recorded

similar long distance movements, especially during dispersal. Florida panther dispersal

distances range from 16 to over 200 km (Land et al. 1998, Maehr et al. 1991).

In the model transient males are programmed to average 15 kilometers a day

represented by 495 moves (30 meters per move, 33 moves per kilometer). Resident males

move less than transients in a day, averaging 13 kilometers (or 429 moves per day).

Female panthers are unlikely to move more than 10 kilometers per day (Sunquist,

personal communication), represented in the model by 330 moves per day. Most studies

of the movements of Florida panthers and pumas in general record the straight-line

distances panthers moved over the course of 24 hours. Beier (1996) found western

pumas moved an average of 5.4 km per day. This model tallies the full movement path,

resulting in longer recorded distances.











Table 2.2. Devaluation amounts for cells recently visited.


Number of moves Devaluation Number of moves Deval
back for females amount for back for males nation
females amou
nt for
males


Last cell visited
2 steps back
3 steps back
4 steps back
5 steps back
6 steps back
7 steps back
8 steps back
9 steps back
10 steps back

Cells adjacent to
the cell visited


Devaluation
amount for
females


Last step
2 steps back
3 steps back
4 steps back
5 steps back
6 steps back
7 steps back
8 steps back
9 steps back
10 steps back


Past 100 moves
101 to 200 moves
201 to 300 moves
301 to 400 moves
401 to 500 moves
501 to 600 moves
601 to 700 moves
701 to 800 moves
801 to 900 moves
901 to 1000 moves

Cells adjacent to
the cell visited


Last step
2 steps back
3 steps back
4 steps back
5 steps back
6 steps back
7 steps back
8 steps back
9 steps back
10 steps back


The differences in daily movement distances between different panthers are

resolved within the model by giving each type of panther subclass a different probability

of taking a step at each iteration. This assigns movement rates. When each panther is

called upon to take a step, it calls up a random number. If the number is above a certain


Deval
nation
amou
nt for
males









threshold, the panther skips taking a step for that iteration. If it is below a threshold the

panther procedes. Transient males typically moved on 95 to 100 percent of the iterations

(steps). Resident males were programmed to move 90 percent, resident females 85

percent, and transient females 87 percent of the iterations. This approach allows male

panthers to cover greater distances than females (on average) over the same time period.


Emigration off GIS map

The study area encompasses five northern Florida counties, which represents

approximately 7,316 kilometers2. While this is sufficient space to model the majority of

movements of a small population of panthers, experimental Texas cougars emigrated out

of the area (Belden and McCown 1996). Florida panthers will also probably emigrate out

of the area. When panthers in the model emigrate off the study area boundaries the

model stops tracking them.

The opportunity to leave the study area is achieved by creating "virtual" cells that

do not actually exist on the GIS maps. These off site cells are assigned the average index

value of all actual existing cells in the 24-cell neighborhood. This allows for the

possibility of "virtual" cells to be chosen in instances where they are of equal value to the

best of existing cells. If an off site cell is chosen for the panther's next move, the panther

stops being tracked by the model and in essence, no longer exists in the model. A

message indicating movement out of the study area is sent to the comment output file.


Home range

The home range of a puma can be defined in several ways. Seidensticker et al.

(1973) defined home area as the ground over which a puma roams. Belden and Hagedorn









(1993) defined home range as the area within which a Texas cougar in north Florida

restricted at least 95 percent of its movements in a predictable area for three or more

months. Movements out and back were considered excursions. The latter definition of

home range is used in this model. The PANTHER program mandates that residents do

not begin to set up home ranges until they have been moving on the landscape for two

months. This mimics the average 74 days taken by Texas cougars to establish home

ranges after being released into the area (Belden and McCown 1996). The model does

not pre-determine location or size of home ranges. For resident panthers, the tendency to

establish a home range is promoted by increasing index values on neighboring cells that

were previously visited some time in the past (Table 2.3). Montgomery (1974) used

similar movement rules in analyzing red fox movements.

For home range evaluation, male and female resident sub classes assign increased

cell index values differently. Female Florida panthers, and pumas in general, establish

smaller home ranges than males and cover less area over a given span of time (Belden

and Hagedorn 1993, Belden and McCown 1996, Land et al. 1998, Maehr 1987, Maehr et

al. 1991). In this program, to mimic these conditions, males were not tied as tightly to

past cells as females, and thus did not receive instructions to rate cells visited in the past

as highly as female panthers.










Table 2.3. Home range cell index value increases and decreases.


Index value
Number of days in the past the cell was visited Males Females
2 to 9 -10 1
10 to 14 -7 3
14 to 60 0 5
Greater than 60 5 10


General Panther Population Interactions

While a single simulaed panther may move across the landscape in fairly

predictable ways, it is the addition of other panthers to the area that makes for interesting

analyses of affects of behavior and resulting movement patterns. The addition of more

than one panther to the model required extensive modifications to several key classes and

function. The Panther, Memory and Main classes were allowed to access different

panther classes, signals and counters, as well as the individual lists of each panther's past

moves. Each individual, before it takes a step, must survey all 24 neighboring cells for

signs of every other panther in the population. If it detects the path or presence of

another panther, it immediately establishes the gender and residency of the detected

neighbor. From there, the program enters a series of loops that dictate how the panther

will react to the signs of others. Males strongly follow females, transient males can be

killed by resident males, and avoidance or attraction have the potential to greatly alter the

path of any individual. Pumas in general practice a land tenure system of hierarchy

(Seidensticker et al. 1973), where social intolerance by females and territoriality by males

not only influences puma movements, but may regulate male density as well (Beier

1993). Model code for these interactions is critical for model movement prediction

accuracy. Without these interactions, simulated females stay in preferred habitats, and









males wander far and wide over the landscape, often without establishing core home

ranges or establishing home ranges over 1000 km2, and of course, never finding females.

Later modifications of the model included code to mimic avoidance behavior,

which at the same time streamlined model runs. In these modifications, a panther only

checks the locations of other panthers in the population only if they are within 10

kilometers. Ten kilometers is used as the maximum distance potentially moved by a

female in one day, and the lower range of movement for males. Seidensticker et al.

(1973) reported that even though pumas mutually avoid one another, in situations when

individuals wanted to find each other, such as breeding periods or mothers looking for

kittens, contact was accomplished rapidly.


Female-male interactions

In general, male and female panthers are attracted to one another, but in different

ways. Males have been documented abandoning an area of abundant prey, little human

disturbance, and large home range to occupy a smaller area containing one resident

female and between two resident male home ranges (Maehr et al. 1991). In this model

males are attracted to the sign or presence of females. This is accomplished in a

quantitative method that makes the presence of females override other habitat variables.

When a male panther detects that a female has used a neighboring cell in her past 1000

moves, the cell is increased in value by adding 1000 minus the number of steps back the

female was in that cell. With this mechanism, the male can detect the exact path of a

female panther and thus choose the most recent cell she visited. This rating method

mimics odor markers. Females can detect males in the same way, but the added value of

a male's presence to the cell index value is only equivalent to 100-step number. This









ranking makes a female attracted to a male, but not as strongly as males are attracted to

females.

During the calibration process this method of tracking potential mates resulted in

few male-female pairings. Coding was added that streamlines the male's ability to find

females. Males are able to detect females up to 10 km, and choose cells in the general

direction of the female. With this new code, the three neighboring cells closest to the

direction of the female are given higher habitat index values, thus encouraging the male

to strike out in the general direction of the female. He randomly chooses one of the three

cells. Once the male is within one km of the female, the one neighboring cell that is

closest to her direction is heavily favored to be chosen by the male for the next move.

This allowed males to find females rapidly, similar to Seidensticker et al.'s (1973)

observations.

In the model transient males will pursue a female into a resident male's home

range until it senses a resident male within 60 meters, at which time it will head away

from the resident male and female. If it does not detect a resident male, it will enter a

male-female visit routine that is identical to interactions between resident males and a

female.

Once a male and female are within the same or adjacent cells, they have entered

the "honeymoon loop," and a counter goes off in the male's class, records the number of

steps the two panthers take that are within the 30 meter range of one another. The female

makes all the next move selections based on habitat variables. The male automatically

follows the female. With each step the female takes, she determines from the male how

many steps they have been together. Once a threshold number has been reached, the









number represents two and one-half days together (Maehr et al. 1991, Seidensticker et al.

1973), she turns off the honeymoon loop and turns on the "take-a-hike" signal. The male

and female are then automatically repelled from one another and each other's past paths.

The actual negative values added to the value of a cell to make the pair avoid one another

were not as important as the timing of the routine. A range of numbers will keep

panthers from being attracted to one another, and even just ignoring one another for a few

steps will usually keep panthers from coming back together. Females keep a record of

how long it has been since she began a visit with a specific male, and controls the

attraction signal. That way, a male does not have to keep track of how long he has been

with each female, a rather difficult situation because of the females to male ratio. The

recently visited female has the "take-a-hike" signal on for the equivalent of three weeks.

If the specific male she has just mated with encounters her trail at any step during this

period, he queries her class to see if the "take-a-hike" signal is still on. If it is, he

continues to ignore her and her path, while she in turn does the same for his. Once the

"take-a-hike" signal is turned off, the male and female are attracted to one another once

again.

There are several loops of information within the female-male interaction loops.

If a female moves away from a male, and due to random chance associated with step

taking he is not able to move with her, there is a catch mechanism whereby the male and

female are still associated with each other. For example, if the male and female are not in

the same or adjacent cell, but they are still within the "honeymoon loop," and if the male

is more than 30 meters from the female for five steps before he catches up with her, they

are still considered together within the honeymoon loop. It is when they have been apart









10 or more steps, that the honeymoon loop counter is zeroed out. If this happens, they

are still attracted to one another.

If during this time a transient male is detected within 60 meters of the pair of

panthers, the resident male will be more attracted to the transient male panther than the

female. He is attracted to this transient in order to fight it, and thus remove the

competition. Once the two males have fought, and the resident male has won (which

occurs 90 percent of the time), the resident male goes back to staying with the female. If

the transient male survives the altercation, the transient enters the "high tail" loop, where

he departs the area as efficiently as possible.

Through these rules male and female panthers are put on a schedule whereby the

male visits with the female for several days and then avoids her until it is time to check

her estrous status within three weeks. Within these rules, counters and comment lines are

fed to an output file to report these interactions. With this output, the user can assess the

number of times and duration of visits between specific panthers.


Male-male interactions

Male pumas in general do not associate with one another. Seidensticker et al.

(1973) recorded 72 instances of known puma interactions and did not find a single

instance of two adult males interacting. Communication between male panthers is

largely believed to be through olfactory signs, such as urine, feces and scent markings left

at scrape sights (Seidensticker et al. 1973). Maehr et al. (1991) found that male

"interactions" occurred at distances greater than 2 km apart. In the model, a panther can

check the recent path of all other panthers and find if their trails have passed through the









immediate area the current panther is in. This gives the model the ability to mimic scent

marking.

The model is structured so that if a resident and transient male are near one

another the resident male dominates the transient male's decisions. If a resident male

perceives a transient male or its path the resident pursues the transient male. To prevent

the resident male from wasting an extended amount of time chasing a transient or its path

there are counters within the Resident male subclass that record the number of time steps

the resident male has been chasing a particular transient. Once the counter has reached a

threshold, and the transient male is not within the neighborhood of 24 cells, the resident

male gives up the chase.

If a resident male does give chase on the path of the transient, he follows a series

of if rules. Cells of the transient panther's path are attractive to the resident, with the

most recently visited cell the most attractive. If the resident male is in a cell adjacent or

is in the same cell as the transient male, the two enter into a "rumble." There are a series

of rules to simulate the outcome of a fight between the panthers. The random number

generator is summoned several times through the process.

If the males fight, the random number generator is called, and 90 percent of the

time the resident panther wins, and 10 percent of the time the transient wins. In south

Florida, over a five year period of study, no resident males were displaced by dispersing

subadult or transient males (Maehr et al. 1991). If the transient panther wins, the random

number generator is summoned again, with a probability that 5 percent of the time, the

resident male is killed. The remaining 95 percent of the times the transient wins, the

resident male is chased off, and becomes a transient, while the transient becomes the









resident male. This involves invoking several flags, and syntax to create a new transient

male with all the qualities of the original resident male, while terminating the original

resident male.

When a resident male wins an altercation the random number generator is

summoned to determine whether the transient male dies or is chased off. The model

predicts the transient male has a 60 percent chance of being killed and a 40 percent

chance of breaking away from the fight site and entering a "high-tail" loop. This is an

estimate based on Florida panther studies that found resident males killed three transient

males over a period of five years (Maehr et al. 1991). When the transient does run away

a "runner" flag is invoked. All these actions take place in the resident male class. Later

in the transient male class the transient panther will query if this runner flag is on, and if

so, will enter into a "high tail" subroutine, where it will move in a straight line away from

the site of the altercation and the resident male's territory. This behavior is created by

doubling the negative value of cells visited in the past 10 moves. The negating of the

past cells makes it twice as unlikely that a panther will retrace its steps (which are closest

to the site of the confrontation with the resident male).

The model mimics avoidance and confrontation between male panthers.

Typically, simulated transient males will detect and avoid any resident male. This is

insured by subtracting an extra 100 points from the value of the cell which contains the

past path of a resident male. This negates any positive features of a cell, even the

presence of a female panther. Negative values lower than 100 allowed transient males to

come in contact with resident males, thus creating altercations especially when a female

was nearby.









Interactions between transient males are very different than between a resident

and transient males. In south Florida three young males (aged 16 to 36 months) had

ranges that overlapped considerably with each other (Maehr 1990). Young transients

have been documented to shift home ranges together in an area of overlap to be near

females (Maehr et al. 1990). In this model, transient males ignore one another, and are

not affected by each other's presence.


Female-female interactions

Female panthers are more tolerant of each other than male panthers are of one

another and have been documented with extensive home range overlap, probably among

related panthers (Maehr et al. 1991). Young females are more readily recruited into the

population and there are no documented cases of female intragender aggression (Maehr et

al. 1991). In this model, females are aware of one another's presence, but they do not

engage in fighting or chasing. All females have a slight avoidance tendency of areas

visited by a resident female. Resident females avoid each other, and ignore transient

females. This in part, mimics how mother pumas and Florida panthers allow their

daughters to occupy parts of their home range (Ross and Jalkotzy 1989).


Placement of Panthers

Seven panthers were included for all model simulations. Five of the initial

starting positions of these panthers were outside the Pinhook Swamp release site of the

Florida Panther Reintroduction Feasibility Study. The rationale for placement of

panthers outside of Pinhook Swamp was to demonstrate where panthers may move later

in reintroduction efforts. Initial sensitivity runs during the first phase of model









development started all panthers from the Pinhook area, which resulted in the majority of

the panthers finding their way to and remaining along the Suwannee River at the

Columbia-Hamilton county border, much like Texas cougars did in the Florida Fish and

Wildlife Conservation Commission study. For the second phase of model simulations

panthers were placed in five different conservation areas, with four of these sites on

publicly owned conservation lands. The one panther not placed on public lands was

Resident male 3, which was placed in Columbia County along Deep Creek, a candidate

site on the Florida list for potential conservation acquisition. Since females typically

found their way to the Suwannee, they were placed along the Suwannee during

simulations to advance the model along in the reintroduction process. Their movement

patterns were programmed to identify the best core habitat for a population of panthers.

Resident female 2 was placed on the eastern edge of Hamilton County, along the

Suwannee River in the Big Shoals State Forest, along with her transient daughter,

Transient female 12. A transient male (Transient male 7) and a transient female

(Transient female 8) were released in the Pinhook Swamp near the actual release site for

the Florida Panther Reintroduction Feasibility Study. Two panthers were placed in the

western edge of the study area, in Hamilton County. This included a transient male

placed in Twin Rivers State Forest, along the eastern border of the Northern

Withalacootchee River (Transient male 5), and a resident female (Resident female 4)

placed in Holton Creek Wildlife Management Area along the Suwannee River (Figure

2.5). These two western study area panthers were used







64






1 -1Florida--Georgia Border
l, .,. h Beehaven Bay :..--i-, I -.- T--7-7
A lapaha ,r Ilin i |,,,-,,-,I i ,, r
E | R river i :: : ,,:1 :, : 1- :r
H ..I SuwanneeRiver Trans Malp 7
--. Water Mgm't Trans Fen 8
,* .. -- strict Lands ,:
S-: p Suwannee "
River
SRes Male 3
Res Fern 4
la i es Fem 2
T ns Male 5 rans Fern 12
Suwannee Co | Baker Co



Columbia Co
.Union Co -

rings p _,S PFSa e River












Figure 2.5. Initial starting places for seven panthers simulated in PANTHER.




to predict movement, landscape use, and effects of humans on a portion of the study site

that did not host panthers during the reintroduction study. For analyses based on

geographic locations the panthers were grouped by area: Big Shoals-Suwannee River

panthers on the Hamilton-Columbia County border, the Pinhook panthers, and the

Western Hamilton County panthers.

Seven panthers were the maximum number feasible for a 12-hour computer

simulation, which represented an eight-months of time. When more panthers were added









to the population, simulation run times increased measurably. A population of seven

panthers is two more panthers than the average number of panthers that took up residence

in north Florida in the Florida Panther Reintroduction Study. No males other than the

resident male actually stayed in the study area during the reintroduction study. The

PANTHER model introduced two added transient males. These transients were crucial

for identifying potential landscape connections.


Quantification of the Landscape Images

Introduction

Five landscape images were developed to represent landscape components needed

for the PANTHER model. The north Florida landscape was quantified using best

available existing GIS databases as well as newly created ones. All landscape images

were created as ERDAS GIS images and read into the C++ model.


Natural Communities

The Natural Communities Landscape Image is derived from the Florida Fish and

Wildlife Conservation Commission's Florida Land Cover Map. This map was developed

from Landsat satellite data collected from 1985-1989 (Kautz et al. 1993). Landsat

Thematic Mapper data were collected at the 1:24,000 scale using a predefined grid-work

of pixels, 30 meters to a side. This pixel size represented the theoretical limits of

resolution of Landsat data (Cox et al. 1994). This pixel size was used as the base for all

pixels in all Landscape Images for this study. Natural communities were classified

identical to the Florida Fish and Wildlife Conservation Commission land cover map,

which was processed with 22 land cover types (Kautz et al. 1993) (Table 2.4, Figure 2.6).









Natural community rankings for Florida panthers are central to how they choose to move

in the landscape (Table 2.5.). Natural communities were given a maximum value of 40

out of 100 points in the habitat index scoring. This was the greatest number of points

assigned to any Landscape Image or other variable.


Table 2.4. Land cover types used in Natural Community Landscape Image.

Natural Upland Communities
Coastal Strand
Dry Prairie
Pineland
Sandpine Scrub
Sandhill/OakScrub
Mixed Hardwood-Pine
Upland Hardwood Forest/ Hardwood Hammock
Tropical Harwood Hammock
Natural Wetland Communities
Coastal Salt Marsh
Fresh Water Marsh
Cypress Swamp
Hardwood Swamp
Bay Swamp
Shrub Swamp
Mangrove Swamp
Bottomland Hardwood Forest
Disturbed Land Cover
Agriculture and Grassland
Shrub and Brush
Barren and Urban
Exotic Plants
Open Water

Taken from Cox et al. 1994.


Natural communities received a total of 40 percent of the total habitat index

ratings to represent the importance natural communities play in panther movement

decisions, especially for females. The distribution of points between the classes was

calibrated based on sensitivity analysis. When there was greater difference of points









between classes, panthers became more selective in their natural community choice,

especially females who were limited to areas of preferred natural communities, and were

not as willing to venture out to less preferred areas as they were under baseline rankings.

This was also observed when natural communities accounted for more than 50 percent of

the total ranking. When there was a smaller difference in scoring between classes,

panthers in the Pinhook Swamp became more accepting of less preferred natural

communities and stayed in the Pinhook area, not venturing to the Suwannee the way the

reintroduction study cougars did. Males were programmed to be more accepting of

different natural communities in order to encourage wide ranging movements and to

more easily find females. The smaller range of ranking among natural communities types

for both male subclasses allows for males to move over the landscape more readily than

females.

Hardwood hammocks have been listed as the most important or one of the preferred

natural communities of Florida panthers in numerous studies (Belden et al. 1988, Maehr

1990, Maehr et al. 1990, Maehr and Cox 1995). Hardwood hammocks, as defined by

Maehr et al. (1991), are a natural community with well to poorly drained soils and

dominated by broad leaved deciduous oaks in association with cabbage palms and many

temperate and tropical shrubs. Hardwood hammocks would be comparable to the

Bottomland Forest and Upland Mixed Forest Natural Communities (Florida Natural

Areas Inventory and Florida Department of Natural Resources 1990). A more

ambiguous, undefined natural community type, forested wetlands/swamps, was ranked as

a preferred natural community of cougars in the Florida Panther Reintroduction

Feasiblity Study (Belden and McCown 1996). This community is not defined by Belden









and McCown, but is assumed to closely resemble Floodplain Swamp Natural Community

and the Hardwood swamp landcover type listed in Table 2.4, for this analysis. In both

phases of the Florida Panther Reintroduction Study, the majority of all radio locations

were in forested wetlands and pine flatwoods (Belden and Hagedorn 1993, Belden and

McCown 1996). Hardwood hammocks and floodplain swamps were ranked as the most

preferred natural communities of the Florida panthers in this model.


Table 2.5. Natural community rankings for cell habitat index values.

Habitat index value
Natural Communities Females Resident Transient
Males Males


First Choice
Hardwood hammocks
Hardwood swamp
Bottomland hardwoods
Second Choice
Pinelands
Third Choice
Cypress
Fourth Choice
All Other Natural Communities
Dry prairie
Mixed hardwood/pine
Shrub swamp
Bay swamp
Marsh
Shrub/brush land, replanted
Clearcuts
Fifth Choice
Avoided Lands
Agriculture
Barren land (clear cuts)
Urban
Sixth Choice
Least Desirable
Open water


+40

+38

+36




+34





+30




+25












N
1k






Natural Communities
Background
M pine
W sandhill
LI]mixed hard-pine
M hariwood hamn-.
mars h
~ ~~~ a. .*...m rs h
presss stvamp
.. I hard wood z'uamp
,L bay sw amp
,L sh rub sniamp
Open water
Wa
E shrub-brush cc
I. I urb3 ran
W No Data






30 0 30 60 Kilometers



Figure 2.6. Natural Communities Landscape Image of north Florida study area. Taken from Florida Fish and Wildlife Conservation
Commission, Closing the Gaps database.









Pine flatwoods has been ranked as a secondary natural community for Florida

panthers (Maehrl990, Maehr et al. 1990, Maehr et al. 1991, Maehr and Cox 1995). Pine

flatwoods as defined by Maehr et al. (1991) are a natural community dominated by slash

pine growing in open forests on moderately well drained soils. Saw palmetto is found as

a common understory shrub. A comparable landcover type for north Florida would be

Pineland. The original Florida Natural Areas Inventory Natural Community for these

areas is Mesic Flatwoods. Use of pine flatwoods by Texas cougars in north Florida was

found to be significantly higher than the availability of the community in the study area

(Belden and McCown 1996). In this model pine flatwoods were ranked second in natural

community preferences. Although pine flatwoods is named as a natural community, it

rarely exists in large (greater than 100 ha) patches. Today in north Florida, pine

flatwoods most often refers to replanted pine plantations.

Cypress swamps have been recognized as playing a significant role in Florida

panther radio-collared locations (Maehr 1990, Maehr et al. 1990, Maehr and Cox 1995).

Maehr et al. (1991) ranked cypress as a third choice for Florida panther daytime radio-

collar locations. Cypress as defined by Maehr et al. (1991) is a seasonally flooded forest

composed of tall cypress trees with few or no hardwoods. This description includes

cypress domes, strands, dwarf cypress, and cypress swamps with scattered slash pine in

better-drained areas. This description would encompass both Dome Swamp and Basin

Swamp in the Florida Natural Areas Inventory Natural Community types. It would

qualify as Cypress swamp in the Florida Fish and Wildlife Conservation Commission

landcover classification. Cypress swamps were found to be one of three communities

where 60 percent of all south Florida panther locations were located (Maehr and Cox









1995). This was due in part to the distribution of the preferred natural community,

hardwood hammocks, which occur in small (<20 ha) pockets mixed within mosaics of

cypress, mixed swamps and pine flatwoods.

In the Florida Panther Reintroduction Feasibility Study all cypress natural

communities were clumped into an undescribed forested wetland class. This forested

wetland class is estimated to be used by the Texas cougars significantly more than the

availability of the natural communities included in this class would predict (Belden and

McCown 1996). In the model, cypress swamps were initially grouped as a preferred

natural community, but later changed to a third choice. This change in ranking was due

to initial sensitivity analysis, which demonstrated simulated panther use of cypress

communities much heavier than observed in the Florida Panther Reintroduction

Feasibility Study. A second reason for this drop in preference ranking for the cypress

type natural community was due to the method used to classify natural communities in

the Florida Fish and Wildlife Conservation Commission landcover Landscape Image.

Landcover classification from satellite imagery of cypress wetlands in north Florida was

found to be inaccurate to a higher degree than other communities (Kautz et al. 1993).

All other natural communities within this analysis were ranked as a fourth

preference class. This is based on telementry results of both Florida panther studies

(Maehr et al. 1991, Maehr and Cox 1995), and the Florida Panther Reintroduction

Feasibility Study (Belden and Haegdorn 1993, Belden and McCown 1996), which found

panther and Texas cougar use of these other natural communities less than their

availability.









Natural community types that were in essence disturbed lands, such as

agricultural lands, barren, and urban were considered avoided communities in the model

and in other studies (Maehr and Cox 1995, Belden and McCown 1996). Open water was

also part of the natural community classification, and was ranked as the least preferred of

all natural community types (Maehr and Cox 1995).

In initial model calibration efforts, natural community rankings were varied

among the six classes and for different panther types. Natural community choices were

considered crucial to where panthers moved, so many variations with subtle changes

between them were simulated. Females are known to be more closely tied to the favored

natural communities (Maehr et al. 1991, Maehr and Cox 1995), so in the model, less

preferred communities were ranked lower values for females than for males. Female

avoidance of agricultural and urban lands initially was mimicked by rating these

"communities" in the single digits, and even negative numbers. This severely restricted

the movements of females and sizes of their home ranges, and boxed them into areas of

preferred natural communities. Agricultural and urban lands were then ranked as 25

points for females, only five points less than other less preferred natural communities.

This allowed for greater flexibility of movement choices for females. This was the final

assigned value for these areas used in the baseline runs. Males on the other hand, are

more willing to use sub-optimal habitat (Belen and McCown 1996, Maehr et al. 1991,

Maehr and Cox 1995). The model instructed males to rank less preferred natural

community types (agriculture and urban) only two points different than other natural

communities. This setting, along with movement rules allowed males to move about the

landscape more freely than females, and in ways similar to the male Texas cougars









released in the study area. While males tended to avoid going directly into areas of

avoided natural communities such as urban and agriculture, ranking these areas just two

points below more natural areas allowed males to often follow the edge of such human

dominated areas such as agricultural lands.

Earlier simulations of the model also ranked communities differently than the

final baseline rankings. This affected the tendency of panthers to move to and from

certain areas. When cypress-type natural community was ranked as a high preference

(highest preferred class or second highest as reported by Maehr et al. 1991), panthers had

a tendency to stay in the Pinhook Swamp release area. This resulted in panthers not

traveling to the Suwannee River where the preferred natural communities occur and

where the original study cougars set up home ranges. When panthers ranked cypress

lower than pinelands they began traveling out of the Pinhook area, finding the preferred

Suwannee habitat, and one another. These types of interacting variables were key in

many of the analyses performed on different natural community rankings.


Roads

The Roads Landscape Image was created by obtaining coverages from

transportation data included in U.S. Geologic Service (USGS) Digital Line Graphs

(DLG) (Figure 2.7). These coverages were derived from 1:24,000 scale maps (McEwen

1985). Road classification specific to this study was based on the USGS DLG

transportation information (Table 2.6.). DLG databases were chosen because they have a

high degree of accuracy and give ample information about roads such as ownership,

number of lanes, and whether it is paved or unpaved.









Table 2.6. Classification of Roads Landscape Image.


Class Description
1 Major interstate highways 4 to 6 lanes, heavy
traffic loads

2 State and Federal roads, 2 lanes, moderate to
heavy traffic

3 State roads, 2 lanes, moderate to light traffic

4 County and city roads, moderate to light traffic

5 Logging roads, light duty dirt roads

6 Trails


Examples
1-75, 1-10
Cloverleaf interchanges

U.S. 441, U.S. 41
S.R. 100

S.R. 51, S.R. 143

C.R. 6, C.R. 135

Rural residential and
industry roads
Paths in the Osceola
National Forest


Roads represent significant barriers to some pumas' movements. Pumas (Beier

1995), Florida panthers (Maehr et al. 1991), and the experimental population of Texas

cougars in north Florida (Belden and McCown 1996) have been documented avoiding

interstate highways. High volume traffic interstates are not the only roads panthers

avoid. In north Florida, Texas cougars tended to avoid crossing primary highways,

secondary hard surfaced roads, and light duty roads (Belden and Hagedorn 1993).

Gender appears to play a role in the willingness of a panther to cross a road. Male

Florida panthers and experimental population male Texas cougars were found to be more

tolerant of roads than females, even crossing Interstate Highways 75 (1-75) and 95 (1-95)

on occasion










































Figure 2.7. Roads and public conservation lands in north Florida study area. Blue lands are federal government properties, green are
Suwannee River Water Management District lands, and red are state parks and forests.









(Maehr et al. 1991, Belden and Hagedorn 1993). Belden and McCown (1996) report that

1-75 did not appear to hinder dispersing males. Females are much more wary of 1-75.

Within the population of Florida panthers in south Florida (Maehr et al. 1991) no radio-

collared female panthers were known to cross 1-75. One female in the reintroduction

study was killed on Interstate 95. No female Texas cougars established home ranges that

straddled interstates. Florida panthers avoid interstates and other roads in their home

range (Maehr et al. 1991). While the busiest of roads are avoided, western pumas have

been documented using dirt roads and hiking trails (Beier 1995).

Roads in the landscape image were reclassified into the classes represented in

Table 2.7. Roads were ranked with only one point difference from one class to the next,

except for interstates, which were ranked in the negative numbers. Ranking was finalized

through calibration and sensitivity analysis. When there was a greater range of values

between road classes, panthers became boxed into roadless areas, and continually

backtracked, thus having restricted home ranges. This was especially true for females.

Thus, the ranking range was condensed to the smallest difference possible on this scale.

Interstates devalued to keep panthers from freely crossing them. Through calibration, it

took a final value of -10 to restrict female movements across the interstates, while still

allowing occasional crossings (less than .05 percent of all female moves). The model

ranked interstates as negative two for males. This allowed males to cross interstates

anywhere from zero to 45 times in an eight-month period. These movements facilitated

establishment of large home ranges by males similar to those estimated in the

reintroduction study. These wide ranging movements helped males to find females, and

transient males to leave resident male territories.










Table 2.7. Final road evaluation by panthers and mortality probabilities
associated with those roads.

Road Type Female Male Probability of
mortality
No roads 7 7 0
Hiking trail 6 6 1 in 12,500
Dirt road 5 5 1 in 1785
Local Road 4 4 1 in 1339
County Road 3 3 1 in 937
State Road 2 2 1 in 667
Interstate Highway -10 -2 1 in 10


Different volumes of traffic, speed of vehicles, and number of lanes all affect the

way a panther will perceive a road, and the potential for the panther to be killed while

near or crossing the road. Even dirt roads and paths have a greater potential for panther

mortality than roadless areas, due to human intrusion and the increased opportunity for

poaching of the panthers. The model quantified avoidance tendencies of panthers for

different road types and introduced the possibility of the panther getting killed if it

crossed a road (Table 2.7). Road mortalities were estimated solely from model

simulations. No published estimates of the chance of mortality associated with specific

roads could be found. Since the road classes were based on number of lanes and volume

of traffic, it was assumed that the lower value a road had in this classification system the

greater the chance of mortality. Mortality values were calibrated with the three types of

roads rated as least desirable to panthers: interstates, state, and county roads. When

mortality on interstates was 25 to 15 percent, males were always killed before the end of

the simulation. A final value of ten percent mortality on interstates was based on

personal assumptions and simulations where anywhere from none to all three males were

killed in an eight month period. In retrospect, this value still may be too high, since









existing underpasses were not modeled, thus allowing for an over exaggeration of

mortality when in fact panthers can cross under interstates, rather than over them.

Chances of mortality associated with other roads were greatly reduced compared to of

mortality associated with interstates. Again, these probabilities were assigned based on

personal assumptions, and calibrated through simulations. When mortality was greater

than those values in Table 2.7, one to all panthers per simulation were killed. Through

calibration, mortality probabilities were considered acceptable if zero to two panthers

were killed per simulation in association with roads other than interstates. These values

allowed for overall population mortality rates of zero to 43 percent. In the reintroduction

study seven cougars were killed, a mortality rate of approximately one-third. Road

associated mortality, while somewhat subjective, was one factor in the synthesis and was

not an important component to the overall objective.


Deer Density

The Deer Density Landscape Image for this study was created by compiling

information from several sources. Four classes of deer density were created: very low,

low, areas with no data, and high densities. Very low areas were defined as

approximately 1 deer per square kilometer, and were delineated as the dog hunt areas of

the Osceola National Forest (J. Norment, personal communication, Florida Fish and

Wildlife Conservation Commission unpublished data). Low densities were defined as the

still hunt areas of the Osceola National Forest, and were estimated to average 2.8 deer per

square kilometer (J. Norment, personal communication). Areas of no information

included private lands in the study area that are outside the flood plain regions of the

Suwannee and Santa Fe rivers. These areas encompassed the majority of the study area,









and all natural community types, especially the agricultural lands of Suwannee County.

High density areas were defined by Florida Fish and Wildlife Conservation Commission

data sources (unpublished) as 6 to 12 deer per square kilometer (J. Norment, personal

communication), and were typically found on public land along the Suwannee and Santa

Fe River floodplains (Figure 2.8).

Estimates of deer densities are based on averages recorded from track-count

surveys from 1981 to 1997 (Florida Fish and Wildlife Conservation Commission

unpublished data), and may differ slightly from separate studies conducted in specific

areas (Fritzen et al. 1995, Belden and McCown 1995, R. Labisky, personal

communication). The purpose of this landscape image is to approximate relative

densities, not specific amounts, since estimates of populations of deer vary from year to

year and between studies.

Deer are the major prey base of Florida panther (Maehr et al. 1990) and were for

the Texas cougars released in north Florida (Belden and Hagedorn 1993, Belden an

McCown 1996). In the beginning of the Florida Panther Reintroduction Study, 90

percent of all prey found killed by cougars was deer. During the winter months, deer

intake decreased, and hog (Sus scrofa) increased as a prey species from five to 22 percent

(Belden and Hagedorn 1993, Belden an McCown 1996). In this model, deer densities

within cells were ranked according to the four classes (Table 2.8). These rankings were

initially based on assumptions of the importance of deer and possible differences in deer

densities. Through sensitivity analysis, the range of points between these four classes was

finalized. When the very low and low deer density class rankings were decreased,

panthers did not travel into these areas. When there was less of a difference between









these two classes and unknown deer density areas or highest area rankings, panthers that

did venture into those low areas had more of a tendency to stay in them, which produced

results unlike those reported for the reintroduction study. The deer ranking value most

important to panther movement, especially for females, is the value of unknown deer

density areas. Sensitivity analysis results found subtle differences in home range sizes if

this area was ranked a value of four rather than zero. Those results are reported in the

next chapter.


Table 2.8. Initial ranking of deer densities.


Deer Density Approximate Resident Transient Resident Transient
Number of Female Female Male Male
Deer/km2
Very Low 1 -6 -3 0 0
Low 3 -5 -2 1 1
Unknown/
No Data ? 0 0 3 3
High 6-12 5 5 5 4



Human Densities

The Human Densities Landscape Image was developed from several different

data sources (Figure 2.9). Florida Department of Revenue (DOR) Property Tax Data

Records were combined with the Township, Range, and Section data available from the

Public Land Survey System (PLSS). The PLSS uses the square mile as the unit of
















4W*E


























Figure 2.8. Deer Densities Landscape Image. Red (darkest color)=highest, green (lighest
color)=low, brown (medium color darkness)=lowest. No color=unknown densities.









measurement. The DOR tax tables store individual parcel information, which includes

the Township, Range, and Section the parcel occurs in, and a land use code, which

represents how the land is used. For this analysis, each parcel was assigned a human use

intensity value, based on the land use code(Table 2.9.).

The human use intensity values were summed for all parcels within each square mile

section, resulting in a single value of human use intensity per section. This was

accomplished through a C++ program specifically created for the task. Sum values

ranged from one to 5,000. In the computer software program, ArcView (ArcView GIS

1996), these values were then added to the PLSS section information to create a GIS

coverage of human density. This section coverage was then converted to a grid coverage

and then to an ERDAS GIS coverage, with the standard 30 m pixel. Each 30 m pixel in

turn was assigned the numeric rating of human density based on the section it was within.

This method quantified human densities according to a rough approximate of human

housing units and businesses per square mile section, which was then converted to square

kilometers. Within the PANTHER model, the panther clumped the range of human

density values into six human density classes, which it then evaluated. This rating was

heavily favored toward lower densities of humans, grouping total sum values below 200

into four classes, and all values from 200 to 5000 into two classes (Table 2.10.).

















-1 N



s Human Density 1990's
No Human
Minimal
Very low
Low
Low-Moderate
SModerate
Moderate-Intense
Intense
Very Intense
Highest density
No human(water)






30 0 30 60 90 Kilometers


Figure. 2.9. Human Density Landscape Image for north Florida study area, 1996.












Table 2.9. Human use intensity values assigned to individual parcels based on Florida
Department of Revenue land use codes.

Department of Revenue land use type Human use intensity
value

All residential uses 4
Businesses:
Stores, restaurants 5
Airports, Public transportation centers, greenhouses,
theaters, tourist attractions, manufacturing,
lumber yards, mills, hospitals, schools, churches,
Military, feed lots, dairies
roads, recreational areas
Mining, Mineral processing, Subsurface rights 3
Improved agriculture 3
Camps 1
Timberland 1
Grazing lands 1
Cropland 1
Forest, Park, and Recreational Areas 1
Rivers and water bodies 1



Table 2.10. Human density classes based on sum of parcels in each square mile.

Human density Initial density Approximate range of Approximate
class values number dwellings per range of number
km2 ha per dwelling
No Human 1-3 No dwellings No dwellings
Minimal 4-17 .4-1.5 256-64
Low 18-60 1.5 -5.8 64- 17
Moderate 61-200 5.8 19.3 17 5
Intense 201-500 19.3 -48.3 5-2
Highest 501-5000 48.3 1930 2 .2


The six human density classes were based on the assumption that there is a

threshold of human density above which the probability of panthers moving into that area

is very low. The exact threshold is not known, but has been analyzed by Beier (1995) and









Belden and McCown (1996). Beier found dispersing pumas would move through low

density housing areas (about 1 dwelling per 16 hectares) and that pumas saw dense

housing places (greater than 20 dwellings per hectare) as impassable. In north Columbia

County where the small population of Texas cougars in the Florida Panther

Reintroduction Feasibility Study established themselves, Belden and McCown (1996)

estimated the overall human density to be less than one dwelling per 243 hectares. They

also found that housing/human densities were much less than this in other parts of the

study area where the cougars established home ranges. Both sets of researchers observed

dispersing pumas, on occasion, traveling through relatively dense housing areas, and

coming within 100 meters of urban areas, typically at night (Beier 1995, Belden and

McCown 1996).

Panthers in the model evaluated human presence based on these six human

density classes (Table 2.11). The range of values for the human density classes was

determined through sensitivity analysis. Humans density was given the maximum value

of seven points in part because of the coarseness of the data base. Square mile sections

are at a rather crude scale for a model based on 30 meter pixels, and do not adequately

represent conditions on the ground. Square mile sections represent an overall

classification. Sensitivity analysis revealed that when the overall value of human density

is increased from seven points to 15, and the spread of points between the classes is

increased from one to two points, to four points between classes, panthers become very

restricted in their movements, particularly females. When the most dense classes of

humans are ranked higher than in the baseline values, males are found to enter small

towns and areas of human settlement.










Table 2.11. Panther evaluation of human density classes.
Habitat Index Ranking per Panther Type
Human density class ResFemale TransFem ResMale TransMale
No Human 7 6 7 6
Minimal 6 5 6 5
Low 4 3 4 3
Moderate 2 2 2 2
Intense -15 -10 -12 -5
Highest -20 -15 -15 -10


Public Conservation Lands

The Public Conservation Lands Landscape Image was developed from the

University of Florida Geoplan Center Conservation Lands 1997 data set (University of

Florida Geoplan Center 1997). This coverage was developed from maps on the 1:24,000

scale. The Conservation Lands database was compiled and standardized by each Florida

Water Management District. Conservation Lands were classified according to

ownership. For PANTHER, conservation lands were grouped into the following

categories: Federal, State, Water Management District, and Local. All lands included as

conservation lands were actually in public possession as of November 1998 (Figure

2.10). Panthers in the model did not evaluate this landscape image.


County Grouping According to Public Attitudes

The Public Attitudes Landscape Image was specifically created for PANTHER.

The image was based on results and experiences from the Northeast Florida Panther

Education Program (Cramer 1995), and political realities revealed during a series of

workshops sponsored by the Florida Conflict Resolution Consortium (Taylor and

Pedersen 1998). Counties in the image were assigned public attitude ratings based in part

on questions asked during a random telephone survey of 300 area residents conducted









during the North East Florida Panther Education Program (Cramer 1995). Other factors

influencing the rating of a county included pre- and post- slide presentation

questionnaires gathered, and events and experiences that evolved over the course of the

Northeast Florida Panther Education Program (Appendix). Pixels in specific counties

were given one of three ratings based on support for panthers: high, ambivalent, and low.

Areas of low support were Hamilton and northern Columbia Counties. In Hamilton

County telephone survey results revealed the lowest levels of support for panthers, and

County Commissioners publicly declared opposition to panther reintroductions. Local

landowners in north Columbia County created an opposition group to the Florida Fish

and Wildlife Conservation Commission panther study and was home to participants of a

series of conflict resolution workshops conducted in Columbia County in 1998 (Taylor

and Pedersen 1998). North Columbia County was designated in the model by drawing an

east-west boundary just north of Deep Creek, and a north-south boundary west of

Pinhook Swamp. Any coordinates a panther used that occurred inside those boundaries

were considered north Columbia. These ratings were not factored into panther decisions

in the movement model. The remainder of Columbia County, Baker County and Union

County were all ranked as areas of highest support for panthers. Suwannee County was

ranked as ambivalent. Table 2.12 presents the findings and rankings concerning human

attitudes taken from the Florida Panther Reintroduction Feasibility Study.











Hamilton orer
TVA n- RPinhool Swamp
'i nForest Suwannee River(- Osceola Nationa
SBlue Springs Water Manageme.t Forest
District Lands
\ _


K Holton Creek
'annee River
e Park

Suwannee


River


Big Shoals


National
Forest


Baker

ColumbiRaifor State-

Ichetucknee Union /
Spr. St. Park /
Oleno St.Pk anta Fe v


Figure 2.10. Public Conservation Lands Landscape Image


I


L











Table 2.12. Response by county to the telephone survey question: How would you agree
or disagree with the following statement? "I favor the reintroduction of panthers in my
county or surrounding counties."

County Agree Don't Disagree Ranking
know/Neither
Columbia 77.7% 7.2% 15.1% 3
Baker 72.9% 3.4% 23.7% 3
Union 76.9% 2.6% 20.5% 3
Suwannee 65.5% 17.2% 17.2% 2
Hamilton* 55.9% 5.9% 38.2% 1

* Statistically significantly different (at the .05 level, p=.034).




Landscape Scenario Development

The PANTHER model was run over GIS layers that depict four landscape

scenarios, each representing various levels of human influence on natural features. The

scenarios involved modifications to and exclusions and inclusions of the Human

Densities Landscape Image. In Scenario I (no humans) panther movement decisions

were based on information from the Natural Communities and Deer Density Landscape

Images but did not incorporate the Human Densities or Road Landscape Images.

However, the Natural Communities Landscape Image was based on satellite imagery,

which represented some moderate-to-large human settlements, and locations of major

hard surfaced roads. This allowed simulated panthers to detect and avoid parts of human

dominated areas and major roads, but did not facilitate detection of portions of small

human settlement and less heavily traveled roads. Scenario II was created to mimic the

current (1990's) human setting in north Florida. It included the Natural Communities,

Deer Density, Human Densities, and Roads Landscape Images, which were incorporated




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs