<%BANNER%>

Estimation of Demographic Parameters and Population Viability Analysis for the Florida Panther

Permanent Link: http://ufdc.ufl.edu/UFE0042566/00001

Material Information

Title: Estimation of Demographic Parameters and Population Viability Analysis for the Florida Panther
Physical Description: 1 online resource (228 p.)
Language: english
Creator: Hostetler, Jeffrey
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: Wildlife Ecology and Conservation -- Dissertations, Academic -- UF
Genre: Wildlife Ecology and Conservation thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Inbreeding and low genetic diversity can cause reductions in individual fitness and increase extinction risk in animal populations. Introgression, achieved by releasing genetically diverse individuals into inbred populations, has been used as a conservation tool to improve demographic performance in endangered populations. By the 1980s, Florida panthers (Puma concolor coryi) had been reduced to a small, inbred population that appeared to be on the brink of extinction. In 1995, female pumas from Texas (P. c. stanleyana) were released in occupied panther range as part of an intentional introgression program to restore genetic variability and improve demographic performance of panthers. We used 25 years (1981-2006) of continuous radio-telemetry and genetic data to estimate and model sub-adult and adult panther survival and cause-specific mortality to provide rigorous sex and age class-specific survival estimates and evaluate the effect of the introgression program on these parameters. Genetic ancestry strongly influenced annual survival of sub-adults and adults after introgression, as F1 generation admixed panthers (s = 0.98) survived better than pre-introgression type panthers (s = 0.77) and other admixed individuals (s = 0.82). Furthermore, heterozygosity was higher for admixed panthers relative to pre-introgression type panthers and positively influenced survival. We used multiple sources of data collected during 1982-2008 and a live-recapture dead-recovery modeling framework to estimate and model survival of Florida panther kittens (age 0 1 year). Overall, annual survival of Florida panther kittens was 0.323 +- 0.065 (SE), which was lower than estimates used in previous population models. We found that kitten survival generally increased with degree of admixture: F1 admixed and backcrossed to Texas kittens survived better than canonical Florida panther and backcrossed to canonical kittens. Average heterozygosity positively influenced kitten and older panther survival, whereas index of panther abundance negatively influenced kitten survival. Our approach to integrate data from multiple sources was effective at improving robustness as well as precision of estimates of Florida panther kitten survival. We used long-term reproductive data (1995-2008) collected from 61 female Florida panthers to estimate and model breeding probability (probability of producing a litter) and litter size, and to investigate the influence of intentional introgression on these parameters. Overall, 6-month probability of breeding (+- 1 SE) was 0.232 +- 0.021 and average litter size was 2.60 +- 0.09. Although F1 admixed females had lower breeding probability than females with other ancestries, this was most likely because kittens born to F1 females survive better; consequently, these females are not available for breeding until kittens are independent. There was no evidence for the effect of ancestry on litter size, or of heterozygosity on probability of breeding or litter size. Previous population viability analyses (PVAs) for the Florida panthers have been critiqued for using unreliable inputs and for unrealistic model structures. The numbers of Florida panthers have increased recently; this increase has been attributed to the genetic introgression, but this continues to be debated. We developed a female-only, birth-flow, age-structured matrix PVA for the Florida panther that incorporated robust parameter estimates, environmental and demographic stochasticity, density-dependence, and model and parameter uncertainty; this model was used to study the dynamics and persistence of the panther population and to explore what would have happened if there had been no genetic introgression. The stochastic estimate of the overall population growth was 1.05 (2.5 and 97.5 percentiles: 0.95 1.15), suggestive of a growing population. We estimated probability of extinction within the next 100 years at 3.7% (2.5 and 97.5 percentiles: 0 55%), but a probability of dropping below 30 females at some point within the next 100 years at 73.2% (2.5 and 97.5 percentiles: 1 100%). These estimates were largely based on data collected during a period of population increase (1995-2008), and do not consider the potential effects of catastrophic events or future changes in anthropogenic influences and habitat conditions; thus, the risk faced by the panther population is likely to be greater than those indicated by our results. The deterministic estimate for population growth without introgression was 0.95 (2.5 and 97.5 percentiles: 0.82 1.07), suggestive of a shrinking population, but effects of introgression on population persistence were mixed. We cannot conclude that observed population growth would not have happened without introgression. Nonetheless, our results along with those reporting improvements in biomedical indicators of inbreeding depression, do provide strong evidence that genetic introgression contributed substantially to the observed increases in Florida panther population.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jeffrey Hostetler.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Oli, Madan K.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042566:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042566/00001

Material Information

Title: Estimation of Demographic Parameters and Population Viability Analysis for the Florida Panther
Physical Description: 1 online resource (228 p.)
Language: english
Creator: Hostetler, Jeffrey
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: Wildlife Ecology and Conservation -- Dissertations, Academic -- UF
Genre: Wildlife Ecology and Conservation thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Inbreeding and low genetic diversity can cause reductions in individual fitness and increase extinction risk in animal populations. Introgression, achieved by releasing genetically diverse individuals into inbred populations, has been used as a conservation tool to improve demographic performance in endangered populations. By the 1980s, Florida panthers (Puma concolor coryi) had been reduced to a small, inbred population that appeared to be on the brink of extinction. In 1995, female pumas from Texas (P. c. stanleyana) were released in occupied panther range as part of an intentional introgression program to restore genetic variability and improve demographic performance of panthers. We used 25 years (1981-2006) of continuous radio-telemetry and genetic data to estimate and model sub-adult and adult panther survival and cause-specific mortality to provide rigorous sex and age class-specific survival estimates and evaluate the effect of the introgression program on these parameters. Genetic ancestry strongly influenced annual survival of sub-adults and adults after introgression, as F1 generation admixed panthers (s = 0.98) survived better than pre-introgression type panthers (s = 0.77) and other admixed individuals (s = 0.82). Furthermore, heterozygosity was higher for admixed panthers relative to pre-introgression type panthers and positively influenced survival. We used multiple sources of data collected during 1982-2008 and a live-recapture dead-recovery modeling framework to estimate and model survival of Florida panther kittens (age 0 1 year). Overall, annual survival of Florida panther kittens was 0.323 +- 0.065 (SE), which was lower than estimates used in previous population models. We found that kitten survival generally increased with degree of admixture: F1 admixed and backcrossed to Texas kittens survived better than canonical Florida panther and backcrossed to canonical kittens. Average heterozygosity positively influenced kitten and older panther survival, whereas index of panther abundance negatively influenced kitten survival. Our approach to integrate data from multiple sources was effective at improving robustness as well as precision of estimates of Florida panther kitten survival. We used long-term reproductive data (1995-2008) collected from 61 female Florida panthers to estimate and model breeding probability (probability of producing a litter) and litter size, and to investigate the influence of intentional introgression on these parameters. Overall, 6-month probability of breeding (+- 1 SE) was 0.232 +- 0.021 and average litter size was 2.60 +- 0.09. Although F1 admixed females had lower breeding probability than females with other ancestries, this was most likely because kittens born to F1 females survive better; consequently, these females are not available for breeding until kittens are independent. There was no evidence for the effect of ancestry on litter size, or of heterozygosity on probability of breeding or litter size. Previous population viability analyses (PVAs) for the Florida panthers have been critiqued for using unreliable inputs and for unrealistic model structures. The numbers of Florida panthers have increased recently; this increase has been attributed to the genetic introgression, but this continues to be debated. We developed a female-only, birth-flow, age-structured matrix PVA for the Florida panther that incorporated robust parameter estimates, environmental and demographic stochasticity, density-dependence, and model and parameter uncertainty; this model was used to study the dynamics and persistence of the panther population and to explore what would have happened if there had been no genetic introgression. The stochastic estimate of the overall population growth was 1.05 (2.5 and 97.5 percentiles: 0.95 1.15), suggestive of a growing population. We estimated probability of extinction within the next 100 years at 3.7% (2.5 and 97.5 percentiles: 0 55%), but a probability of dropping below 30 females at some point within the next 100 years at 73.2% (2.5 and 97.5 percentiles: 1 100%). These estimates were largely based on data collected during a period of population increase (1995-2008), and do not consider the potential effects of catastrophic events or future changes in anthropogenic influences and habitat conditions; thus, the risk faced by the panther population is likely to be greater than those indicated by our results. The deterministic estimate for population growth without introgression was 0.95 (2.5 and 97.5 percentiles: 0.82 1.07), suggestive of a shrinking population, but effects of introgression on population persistence were mixed. We cannot conclude that observed population growth would not have happened without introgression. Nonetheless, our results along with those reporting improvements in biomedical indicators of inbreeding depression, do provide strong evidence that genetic introgression contributed substantially to the observed increases in Florida panther population.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jeffrey Hostetler.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Oli, Madan K.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042566:00001


This item has the following downloads:


Full Text

PAGE 1

1 ESTIMATION OF DEMOGRAPHIC PARAMETERS AND POPULATION VIABILITY ANALYSIS FOR THE FLORIDA PANTHER By JEFFREY ALLAN HOSTETLER 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 2010

PAGE 2

2 2010 Jeffrey A. Hostetler

PAGE 3

3 To my family and Jennifer Russell

PAGE 4

4 ACKNOWLEDGMENTS I thank my committee M. Oli, J. Nichols, M. Sunquist, B. Bolker, and D. Onorato for their tremendous help. I thank for valuable advice throughout the project. All genetic and ancestry analyses were performed at the National Cancer Institute by I am great ly indebted to them. I thank S. Bass, J. Benson, M. Cunningham, D. Giardina, D. J a nsen, A. Johnson, D. Land, M. Lotz, C. McBride Rocky McBride Rowdy McBride Roy McBride, L. Oberhofer, S. Schulze, D. Shindle, staffs at Everglades National Park and Big Cy press National Preserve, and others for assistance with fieldwork I also thank J. Benson, B. Bolker, M. Conroy, J. Hines, J. Laake, S. Mills, J. Nichols, M. Oli, D. Onorato, M. Sunquist D. Valle, G. White, and J. White for advice and assistance regardin g data analysis and panther biology. This work was funded through the Florida Panther Research and Management Trust Fund, National Park Service, University of Florida, and grant Agreement No 401816G091 from the United Sta tes Fish and Wildlife Service.

PAGE 5

5 T ABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 12 ABSTRACT ................................ ................................ ................................ ................... 14 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 17 2 INTENTIONAL GENETIC INTROGRESSION IMPROVES SURVIVAL OF F 1 A DULTS AND SUB ADULTS IN A SMALL, INBRED FELID POPULATION .......... 20 Introduction ................................ ................................ ................................ ............. 20 Materials and Methods ................................ ................................ ............................ 22 Field Methods ................................ ................................ ................................ ... 22 The Introgression Program ................................ ................................ ............... 23 Estimating Age Classes ................................ ................................ .................... 23 Genetic Variables ................................ ................................ ............................. 24 Index of Abundance ................................ ................................ ......................... 27 Survival Analyses ................................ ................................ ............................. 27 Results ................................ ................................ ................................ .................... 32 Age, Sex, Index of Abundance, and Year ................................ ........................ 32 Genetic Ancestry and Heterozygosity ................................ .............................. 32 Cause Specific Mortality ................................ ................................ ................... 34 Discussion ................................ ................................ ................................ .............. 34 3 GENETIC INTROGRESSION AND THE SURVIVAL OF FLORIDA PAN THER KITTENS ................................ ................................ ................................ ................. 46 Introduction ................................ ................................ ................................ ............. 46 Materials and Methods ................................ ................................ ............................ 49 Field Methods ................................ ................................ ................................ ... 49 Heterozygosity and Genetic Ancestry ................................ .............................. 50 Index of Abundance ................................ ................................ ......................... 52 Data Prepar ation and Analysis ................................ ................................ ......... 52 Three Month Time Step Data Preparation and Analysis ................................ .. 57 Results ................................ ................................ ................................ .................... 57 Discussion ................................ ................................ ................................ .............. 59 4 DOES GENETIC INTROGRESSION IMPROVE REPRODUCTIVE PERFORMANCE? A TEST ON THE ENDANGERED FLORIDA PANTHER ......... 71

PAGE 6

6 Introduction ................................ ................................ ................................ ............. 71 Materials and Methods ................................ ................................ ............................ 73 Study Area ................................ ................................ ................................ ........ 73 Field Methods ................................ ................................ ................................ ... 73 Genetic Ancestry and Heterozygosity ................................ .............................. 74 Index of Abundance ................................ ................................ ......................... 75 Estimating Age Classes ................................ ................................ .................... 76 Modeling Probability of Breeding ................................ ................................ ...... 76 Modeling Litter Size ................................ ................................ .......................... 78 Modeling Approach ................................ ................................ .......................... 79 Results ................................ ................................ ................................ .................... 81 Probability of Breeding ................................ ................................ ..................... 82 Litter Size ................................ ................................ ................................ ......... 82 Discussion ................................ ................................ ................................ .............. 83 5 THE EFFECT OF INTROGRESSION ON THE DYNAMICS AND PERSISTENCE OF THE FLORIDA PANTHER POPULATION .............................. 93 Introduction ................................ ................................ ................................ ............. 93 Methods ................................ ................................ ................................ .................. 95 Field and Genetic Methods ................................ ................................ ............... 95 Model Structure ................................ ................................ ................................ 96 Parameter Values ................................ ................................ ............................. 98 Deterministic Demographic Analyses ................................ ............................. 100 Incorporating Parameter Uncertainty ................................ .............................. 100 Incorporating Model Selection Uncertainty ................................ ..................... 101 Incorporating Env ironmental Stochasticity ................................ ...................... 101 Estimating Stochastic Growth Parameters ................................ ..................... 104 Incorporating Density Dependence ................................ ................................ 104 Incorporating Demographic Stochasticity ................................ ....................... 106 Estimating Extinction Parameters ................................ ................................ ... 108 What If the Intr ogression Never Happened? ................................ .................. 108 Results ................................ ................................ ................................ .................. 109 Density Independent Growth Rate and Perturbation Analysis ........................ 109 Extinction Parameters ................................ ................................ .................... 110 What If the Introgression Never Happened? ................................ .................. 110 Discussion ................................ ................................ ................................ ............ 112 6 SUMMARY AND CONCLUSIONS ................................ ................................ ........ 131 APPENDIX A AGE ERROR ESTIMATION ................................ ................................ ................. 135 Methods ................................ ................................ ................................ ................ 135 Results and Discussion ................................ ................................ ......................... 136

PAGE 7

7 B FINE SCALE ANCESTRY EFFECTS ON SUB ADULT AND ADULT SURVIVAL 139 Methods ................................ ................................ ................................ ................ 139 Results ................................ ................................ ................................ .................. 140 C CAUSE SPECIFIC MORTALITY ANALYSIS ................................ ........................ 146 Materials and Methods ................................ ................................ .......................... 146 R esults ................................ ................................ ................................ .................. 147 D INFERENCE RESULTS FOR ANCESTRY AND HETEROZYGOSITY ................ 150 Materials and Methods ................................ ................................ .......................... 150 Results ................................ ................................ ................................ .................. 150 E ANALYSIS OF FLORIDA PANTHER KITTEN SURVIVAL BASED ON 3 ANCESTRY CATEGORIE S ................................ ................................ .................. 154 Materials and Methods ................................ ................................ .......................... 154 Results and Discussion ................................ ................................ ......................... 155 F DATA PREPARA TION ................................ ................................ .......................... 159 G ANALYSIS OF SURVIVAL OF FLORIDA PANTHER KITTENS USING DATA ORGANIZED IN 3 MONTH TIME STEPS ................................ ............................ 161 Materials and Methods ................................ ................................ .......................... 161 Results and Discussion ................................ ................................ ......................... 163 H INCLUDING ANOTHER BASE MODEL ................................ ............................... 171 Methods ................................ ................................ ................................ ................ 171 Results ................................ ................................ ................................ .................. 171 I ANALYSIS OF THE INFLUENCE OF ANCESTRY USING FINE SCALE ANCESTRY TESTING ................................ ................................ .......................... 180 Methods ................................ ................................ ................................ ................ 180 Results and Discussion ................................ ................................ ......................... 181 J ANALYSIS OF REPRODUCTIVE PARAMETERS EXCLUDING OLDER ................................ ................................ ........................ 187 Methods ................................ ................................ ................................ ................ 187 Results ................................ ................................ ................................ .................. 187 K ANNUAL BREEDING PARAMETERS FOR PVA ................................ ................. 193 Methods ................................ ................................ ................................ ................ 193

PAGE 8

8 Results ................................ ................................ ................................ .................. 195 L DEMOGRAPHIC PARAMETERS FOR CANONICAL PANTHERS ...................... 201 Materials and Methods ................................ ................................ .......................... 201 Results ................................ ................................ ................................ .................. 201 LIST OF REFERENCES ................................ ................................ ............................. 214 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 228

PAGE 9

9 LIST OF TABLES Table page 2 1 Model comparison results for the effects of sex, age class, abundance index, and year on Florida panther survival ................................ ................................ .. 41 2 2 Model averaged annual survival rates ( ), standard errors ( ), and number of Florida panthers tracked ( n ) in sex and age class categories ......................... 42 2 3 Model comparison results showing top ranked models Information Criterion [ AIC] < 4) for the effects of ancestry, heterozygosity, sex, and age class (sub adults and prime adults only) on Florida panther survival ................................ ................................ ................................ ............... 43 2 4 Estimated annual survival rates (with and number of panthers) by ancestry, sex, and age class ................................ ................................ .............. 44 3 1 List of kitten survival models ................................ ................................ ............... 65 3 2 Model comparison t able to estimate annual kitten survival and select base model ................................ ................................ ................................ .................. 66 3 3 Model selection table for effects of covariates on survival ................................ .. 67 4 1 Model selection table for probability of breeding ( p ) ................................ ........... 88 4 2 Model selection table for litter size ( ) ................................ ................................ 89 5 1 Variable names used in th is manuscript, with definitions ................................ .. 119 5 2 Models for sub adult and adult survival ................................ ............................ 120 5 3 Models for kitten survival ................................ ................................ .................. 121 5 4 Models for sub adult and adult probability of breeding ................................ ..... 122 5 5 Models for the average number of kittens produced by sub adult and adult female s (for females that bred in a year) ................................ .......................... 123 5 6 Summary statistics for extinction and quasi extinction times (number of years until first dropping below critical thresholds) ................................ ..................... 124 A 1 Distribution of lengths of birthdate ranges (in months) ................................ ..... 137 A 2 Resampling results examining error from using estimated age at capture to estimate surv ival by age class ................................ ................................ .......... 138

PAGE 10

10 B 1 Definitions of ancestry models, an d the hypotheses they represent ................. 142 B 2 ) model comparison results showing all models for the effects of ancestry, heterozygosity, sex, and age class on adult Florida panther survival ................................ ................................ ............ 143 D 1 Full AIC table for ancestry and heterozygos ity. ................................ ................ 152 D 2 Genetic coefficients from all models including sex and age class and at least one genetic covariate ................................ ................................ ....................... 153 E 1 Definitions o f simple ancestry models, an d the hypotheses they represent ..... 156 E 2 Model selection table for testing hypotheses regarding the effect of genetic ancestry on survival of kittens and older panthers based on 3 ancestry categories ................................ ................................ ................................ ......... 157 G 1 A priori models used for the 3 month time step analysis ................................ .. 165 G 2 Kitten survival models to exam ine the effects of age (3 month time step) ........ 166 G 3 Estimates of annual sub adult and adult survival from the 3 month kitten survival analysis and the adult analysis of Chapter 2 ................................ ....... 168 G 4 Model comparison table for examining the effects of season on kitten survival ................................ ................................ ................................ ............ 169 H 1 Model selection table for sex and litter size effects on kitten survival, using Base2 model ................................ ................................ ................................ ..... 173 H 2 Model selection table for ancestry, heterozygosity, and abundance index effects on kitten survival, using Base2 model ................................ ................... 174 H 3 Model selection table for three category ancestry effects on kitten and older survival, using the Base2 model ................................ ................................ ....... 176 I 1 Definitions of complex ancestry models, a nd the hypotheses they represent. 183 I 2 Model comparison table testing for the effects of complex ancestry (see Table I 1) on probability of breeding ( p ) ................................ ............................ 184 I 3 Model comparison table testing for the effects of complex ancestry (see Table I 1) on litter size ( ) ................................ ................................ ................. 185 J 1 Model selection table evaluating the effects of factors on probability of breeding ( p ), excluding older adults ................................ ................................ 188 J 2 Model selection table evaluating the effects of factors on litter size ( ), excluding older adults ................................ ................................ ....................... 189

PAGE 11

11 K 1 Model selection table evaluating the effects of factors on annual probability of breeding ( q ) ................................ ................................ ................................ ...... 196 K 2 Model selection table evaluating the effects of factors on annual cumulative number of kittens ( ) ................................ ................................ ......................... 198 K 3 Model selection table testing for the random effect of year (rand(Year)), with and without abundance index as a covariate, on annual probability of breeding ( q ) ................................ ................................ ................................ ...... 199 K 4 Model selection table testing for the random effect of year (rand(Year)) on annual cumulative number of kittens ( ) ................................ ........................... 200 L 1 Models for canonical sub adult and adult survival ................................ ............ 203 L 2 Models for (density dependent) canonical kitten survival ................................ 205 L 3 Models for density independent canon ical kitten survival ................................ 208 L 4 Models for sub adult and adult canonical probability of breeding ..................... 211 L 5 Models for sub adult and adult a verage number of kittens produced (for canonical females that bred in a year) ................................ .............................. 213

PAGE 12

12 LIST OF FIGURES Figure page 2 1 Model averaged effects of ancestry a nd heterozygosity on annual survival of female prime adult Florida panthers from 1997 2006 ................................ ......... 45 3 1 The effect of genetic variables on model averaged kitten survival estimates ..... 68 3 2 The effect of a panther abundance index on model averaged kitten survival estimates ................................ ................................ ................................ ............ 69 3 3 Model averaged kitten survival by 3 month age interval wi thin the first year ...... 70 4 1 Frequency distribution of litter sizes produced by radio collared female panthers (1995 2008) ................................ ................................ ...................... 90 4 2 Mode l averaged reproductive rates ( unconditional standard errors) by age class, season, and abundance index ................................ ................................ .. 91 4 3 Model averaged reproductive rates ( unconditional standard errors) by ancestry ca tegory and heterozygosity ................................ ................................ 92 5 1 Pseudo code for estimating probability of extinction. ................................ ........ 125 5 2 Perturbation analysis deterministic pop ulation growth rate ( ) to lower level parameters ................................ ................................ ................................ ....... 126 5 3 Perturbation analysis results for stochastic population growth rate ( s ) to lower level parameters ................................ ................................ ..................... 127 5 4 Simulated cumulative probabilities of extinction and quasi extinction over 500 years ................................ ................................ ................................ ................. 128 5 5 Contributions of lower level parameters (see Table 5 1) to differences between the deterministic population growth rates for the whole population ( ) and the canonical population ( (c) ) ................................ ............................... 129 5 6 Results of simulations starting with conjectured 1995 population levels (total 15 females) and run for 500 years ................................ ................................ .... 130 B 1 Model averaged effects of complex ancestry and heterozygosity on annual survival of female prime adult Florida panthers from 1997 2006 ...................... 145 C 1 Cause specific mor tality rates (with standard errors) ................................ ........ 149 E 1 Model averaged estimates of annual survival of Florida panther kittens based on 3 ancestry categories ................................ ................................ .................. 158

PAGE 13

13 G 1 Individual model estimates of annual kitten survival ................................ ......... 170 H 1 The effect of genetic variables on model averaged kitten survival estimates, including two different base models ................................ ................................ .. 177 H 2 The effect of a panther abundance index on model averaged kitten survival estimates, including two different base models ................................ ................ 178 H 3 Model averaged estimates of annual survival of Florida panther kittens based on 3 ancestry categories, including two different base models ......................... 179 I 1 Model averaged reproductive rates by complex a ncestry category ( unconditional standard errors) ................................ ................................ .......... 186 J 1 Model averaged reproductive rates ( unconditional standard errors) by age class ................................ ................................ ................................ ................. 190 J 2 Model averaged reproductive rates ( unconditional standard errors) by season ................................ ................................ ................................ .............. 191 J 3 Model averaged reproductive rates ( unconditional standard errors) by ancestry category ................................ ................................ ............................. 192

PAGE 14

14 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 ESTIMATION OF DEMOGRAPHIC PARAMETERS AND POPULATION VIABILITY ANALYSIS FOR THE FLORIDA PANTHER By Jeffrey A. Hostetler December 2010 Chair: Madan K. Oli Major: Wildlife Ecology and Conservation Inbreeding and low genetic diversity can cause reductions in individual fitness and increase ex tinction risk in animal populations. Introgression, achieved by releasing genetically diverse individuals into inbred populations, has been used as a conservation tool to improve demograph ic performance in endangered populations. By the 1980s, Florida pa nthers ( Puma concolor coryi ) had been reduced to a small, inbred population that appeared to be on the brink of extinction. In 1995, female pumas from Texas ( P. c. stanleyana ) were released in occupied panther range as part of an intentional introgression program to restore genetic variability and improve demographic performance of panthers. We used 25 years ( 1981 2006 ) of continuous radio telemetry and genetic data to estimate and model sub adult and adult panther survival and cause specific mortality to provide rigorous sex and age class specific survival estimates and evaluate the effect of the introgression program on these parameters. Genetic ancestry strongly influenced annual survival of sub adults and adults after introgression, as F 1 generation ad mixed panthers ( = 0.98) survived better than pre introgression type panthers ( = 0.77) and other admixed individuals ( = 0.82). Furthermore,

PAGE 15

15 heterozygosity was higher for admix ed panthers relative to pre introgression type panthers and positively influenced survival. We used multiple sources of data collected during 1982 2008 and a live recapture dead recovery modeling framework to estimate and model survival of Florida panthe r kittens (age 0 1 year). Overall, annual survival of Florida panther kittens was 0.323 0.0 65 (SE), which was lower than estimates used in previous population models. We found that kitten survival generally increased with degree of admixture: F 1 admi xed and backcrossed to Texas kittens survived better than canonical Florida panther and backcrossed to canonical kittens. Average heterozygosity positively influenced kitten and older panther survival, whereas index of panther abundance negatively influen ced kitten survival. Our approach to integrate data from multiple sources was effective at improving robustness as well as precision of estimates of Florida panther kitten survival. We used long term reproductive data (1995 2008) collected from 61 femal e Florida panthers to estimate and model breeding probability (probability of producing a litter) and litter size, and to investigate the influence of intentional introgression on these parameters. Overall, 6 month probability of breeding (1SE) was 0.23 2 0.021 and average litter size was 2.60 0.09. Although F 1 admixed females had lower breeding probability than females with other ancestries, this was most likely because kittens born to F 1 females survive better; consequently, these females are not a vailable for breeding until kittens are independent. There was no evidence for the effect of ancestry on litter size, or of heterozygosity on probability of breeding or litter size. Previous population viability analyse s (PVAs ) for the Florida panther s have been critiqued for using unreliable inputs and for unrealistic model structures. The numbers

PAGE 16

16 of Florida panthers have increased recently; this increase has been attributed to the genetic introgression but this continue s to be debated. We developed a female only, birth flow, age structured matrix PVA for the Florida panther that incorporated robust parameter estimates, environmental and demographic stochasticity, density dependence, and model and parameter uncertainty; this model was used to study th e dynamics and persistence of the panther population and to explore what would have happened if there had been no genetic introgression. T he stochastic estimate of the overall population growth was 1.05 (2.5 and 97.5 percentiles: 0.95 1.15), suggestive of a growing population. We estimated probability of extinction within the next 100 years at 3.7% (2.5 and 97.5 percentiles: 0 55%), but a probability of dropping below 30 females at some point within the next 100 years at 73.2% (2.5 and 97.5 percentile s: 1 100%). These estimates were largely based on data collected during a period of population increase (1995 2008), and do not consider the potential effects of catastrophic events or future changes in anthropogenic influences and habitat conditions; t hus, the risk faced by the panther population is likely to be greater than those indicated by our results. The deterministic estimate for population growth without introgression was 0.95 (2.5 and 97.5 percentiles: 0.82 1.07), suggestive of a shrinking p opulation, but effects of introgression on population persistence were mixed. W e cannot conclude that observed population growth would not have happened without introgression. Nonetheless, our results along with those reporting improvements in biomedical indicators of inbreeding depression, d o provide strong evidence that genetic introgression contributed substantially to the observed increases in Florida panther population.

PAGE 17

17 CHAPTER 1 INTRODUCTION The Florida panther ( Puma concolor coryi ), an endangered puma subspecies, is a textbook example of a small and isolated population facing a mu ltitude of threats to its persistence Once distributed across the southeastern United States, their number was the (McBride et al., 2008) The population was thought to be severely inbred and was characterized by extremely low genetic diversity (Culver et al., 2000; Roelke et al., 1993b) Many phenotypic traits thought to be characteristics of inbreeding were reported from this population, including very l ow sperm quality, and high incidences of cryptorchidism, cowlicks, and kinked tails (Barone et al., 1994; Mansfield and Land, 2002; Roelke et al., 1993b) Due to concerns about inbreedin g depression, 8 female Texas pumas ( P. c. stanleyana ) were brought into the Florida panther population in 1995. Since this genetic introgression, the population has increased (McBride et al., 2008) although the processes that hav e led to this increase continue to be debated (Creel, 2006; Maehr et al., 2006; Pimm et al., 2006b) Coordinated research on Florida panther bega n in 1981 to learn more about the ir biology and to support conservation efforts. This body of works has not only improved our understanding of panther ecology and behavior (Belden et al., 1988; Maehr et al., 1990; Maehr et al., 1989; Roelke et al., 1993a) but also has contributed substantially to our understanding of the risks faced by small populations and challenges inherent in managing large carnivores in an increasingly human dominat ed landscape (Comiskey et al., 2002; Dees et al., 2001; Janis and Clark, 2002; Maehr et al., 2002b; O'Brien et al., 1990)

PAGE 18

18 In 2003, the U.S. Fish and Wildlife Service (USFWS) and the Florida Fish and Wildlife Conservation Commission (FWC) convened the Scientific Review Team (SRT) to review the status of Florida panther science, and to make research and management recommendations. The SRT, composed of highly qualified scientists with e xpertise in carnivore ecology, population dynamics, and conservation biolog y thoroughly reviewed existing scientific literature on the biology of the Florida panther, and found that rigorous estimates of several demographic parameters were not available. The SRT also noted that previously reported estimates of some demographic variables and those used in earlier population viability analyses (Maehr et al., 2002a) were unreasonable and led to unreliable inferences regarding the persistence of the panther population (Beier et al., 2003; Beier et al., 2006; Gross, 2005) The team recommended reanalysis of existing data to rigorously estimate age specific demographic variables and to test hypotheses regarding fa ctors influencing those variables (Beier et al., 2003) Furthermore, the SRT recommended development and analysis of population models that reflect biology of panthers and do not excessively rely on aspects of panther biology that are unknown. In accordance with SRT recommendations (Beier et al., 2003) our goals were to provide rigorous estimates of demographic parameters of Florida panthers, and to evaluate the influences of various factors on those parameters W e analyzed long term radio telemetry (1981 2008) and passive integrated t ransponder (PIT) tagging (1995 2008) data to estimate and model survival rates, and reproductive data to estimate and model probability of successful reproduction (breeding probabilit y) and litter size. One of the most important management actions undertaken for the conservation of Florida panthers was the genetic introgression. This was and remains a controversial

PAGE 19

19 decision (Maehr and Caddick, 1995; Maehr et al., 2006; Pimm et al., 2006b) However, the potential influence of genet genetic structure on the demography of Florida panthers is not yet fully understood. Thus, we also evaluated the influence of genetic introgression on aforementioned demographic parameters. These results are expected to help guide future management efforts.

PAGE 20

20 CHAPTER 2 INTENTI ONAL GENETIC INTROGRESSION IMPROVES SURVIVAL OF F 1 ADULTS AND SUB ADULTS IN A SMALL, INBRED FELID POPULATION Introduction Hybridization, particularly between species, can have serious conservation implications if one of the parental types is rare or endang ered because such populations may decline or become extinct due to hybridization (Allendorf et al., 2001; Levin et al., 1996; Rhymer and Simberloff, 1 996; Wolf et al., 2001) However, intentional introgression, achieved by releasing individuals from different subspecies or populations into small, inbred populations, has been used as a conservation tool to restore genetic variability and improve demogr aphic performance of endangered populations (Madsen et al., 1999; Westemeier et al., 1998) Despite the apparent success of some of these genetic manipulations (e.g., Westemeier et al., 1998) they remain controversi al due to concerns regarding outbreeding depression (Edmands, 2007; Greig, 1979) potential loss of genetic integrity of endangered populations (Allendorf et al., 2001) and because few studies have convincingly quantified the effects of intentional introgr ession on demographic parameters (Beier et al., 2006) When estimating the effect of hybridization on survival for populations that have experienced introgression, it is important to determine genetic ancestry of individuals because hybrid fitness may differ relative to parental types (Arn old and Hodges, 1995; Burke and Arnold, 2001) Hybrid superiority can be primarily due to exogenous selection, when hybrids are favored under specific environmental conditions that vary across time (Grant and Grant, 1992) or space (Good et al., 2000) Alternatively, hybrid superiority can be due to intrinsic qualities of outbred individuals as predicted by hybrid vigor theory (Crow, 1948; Shull, 1908) Hybrid vigor, or heterosis, is a phenomenon

PAGE 21

21 whereby hybrids exhibit higher fitness relative to inbred parental types due to increased heterozygosity in the F 1 generation (Burke and Arnold, 2001) Florida panthers ( Puma concolor coryi ) were once distributed across much of the southeastern United States, but currently exist only in 1 small (~100 individuals) breeding p opulation in south Florida (McBride et al., 2008) This population represents t he only breeding population of pumas east of the Mississippi River and is listed as endangered under the United States Endangered Species Act (Sullivan, 2004) By the late 1980s and early 1990s the pop ulation appeared to be at imminent risk of extinction (McBride et al., 2008) and was severely inbred and lacking in genetic diversity (Roelke et al., 1993b) In 1995, 8 femal e pumas from Texas ( P. c. stanleyana ) were released into south Florida as part of an intentional introgression program to restore genetic diversity (Johnson e t al., 1997) The population has increased since the initiation of this program (McBride et al., 2008) but the effects of genetic introgression on specific demographic parameters, such as survival and sources of mortality, remain unclear. Rigorous evaluation of the effects of genetic admixture on survival rates following intention al introgression would have broad implications for the relevance of applied genetic management to the conservation of small, isolated populations (Beier et al., 2006) We estimated survival and caus e specific mortality of radio collared sub adult and adult panthers from 1981 2006 and investigated the influence of multiple intrinsic variables with 2 main objectives. First, we sought to provide sex and age class specific estimates of annual survival and cause specific mortality for panthers from 1981 2006 in order to provide a basic understanding of panther survival within which we could explore

PAGE 22

22 the influence of intentional introgression on these parameters. Second, we investigated the influence of: a) genetic ancestry, with respect to hybridization with Texas pumas released during the introgression program, and b) genetic diversity, as quantified by individual heterozygosity, on survival and cause specific mortality of sub adult and adult panthers fr om 1997 2006. We hypothesized that survival rates of admixed panthers would be higher than for pre introgression type panthers and that increased survival would be most evident in the F 1 generation, as predicted by hybrid vigor theory (Burke and Arnold, 2001) Materials and Methods Field Methods Florida panthers and Texas pumas were captured and monitored by radio telemetry from 1981 2006 across the range of the subspecies in south Florida, USA (Kautz et al., 2006) by biologists from the Florida Fish and Wildlife Conservation Commission (FWC) and National Park Service (NPS) using methods described by Belden et al. (1988) and Land et al. (2008) Kittens (<35 days old) were marked at natal dens with transponder chips (Benson et al., 2008; Land et al., 1998) which allowed us to more accurately estimate ages of marked kittens that were subsequently recaptured and radio collared. Beginning in 1985, captured panthers were v accinated against several diseases (for details see Cunningham et al., 2008) which p otentially improved their survival. When radio collared panthers died, a cursory examination of the carcass and surrounding area was conducted in the field, and a suspected cause of death was assigned if evidence allowed. Eighty four of 92 (91.3%) carca sses of radio collared panthers were also necropsied by experienced veterinarians or pathologists who

PAGE 23

23 attempted to determine cause of death. We estimated date of death from telemetry data and by assessing the condition of the carcasses found in the field. The Introgression Program Eight female pumas from west Texas were released into south Florida at 5 sites between 29 March and 26 July 1995 (Johnson et al., 1 997) Two were released in Everglades National Park (ENP), 4 in Big Cypress National Preserve (BCNP), and 2 in Fakahatchee Strand Preserve State Park (FSSP). The introgression program was undertaken to restore genetic variability to the panther populati on by mimicking natural gene flow that occurred between Florida panthers and other puma populations before the extirpation of pumas from most of the southeastern United States. Five of the released Texas pumas reproduced and produced a total of 12 litters Estimating Age Classes We knew the ages (within a few days) for panthers that were handled as kittens at natal dens and subsequently recaptured and radio collared ( n = 54, 36.7% of total). For panthers not marked as kittens, their age w as estimated in the field at time of first capture using tooth wear and pelage characteristics (Ashma n and Greer, 1976) We examined potential error in survival estimates due to error in age estimates and found it to be small (Appendix A ). Using both types of estimates, we separated panthers into 4 age classes: kittens (0 1 years old), sub adults (1 2.5 and 1 3.5 years old for females and males, respectively), prime adults (2.5 10 and 3.5 10 years old for females and males, respectively), and older combined prime adults and older adults into a single adult category for some of our analyses. Since panther kittens were not radio collared until at least 4 6 months of age,

PAGE 24

24 2001), a different analytical approach would be needed to properly estimate survival of kittens. Thus, we limited our current analyses to sub adults and adults. The sub adult approximate ages when females and males first establish home ranges. We estimated age of independence at just over 1 year (mean = 397 days, SD = 75 days, n = 35; FWC & NPS, unpublished data) from known age kittens that were radio tracked simultaneously with their mothers (before and after independenc e). We standardized the beginning of the sub adult age class to 1 year for simplicity. Genetic Variables We extracted total genomic DNA from blood and tissue samples obtained from wild caught panthers and captive pumas from south Florida and west Texas during 1981 2006. We amplified and scored 23 microsatellite loci (F37, F42, FCA43, FCA57, FCA75, FCA90, FCA91, FCA94, FCA95, FCA98, FCA124, FCA133, FCA161, FCA193, FCA 243, FCA249, FCA293, FCA310, FCA369, FCA441, FCA559, FCA566, FCA668) following previou sly described PCR amplification conditions (Menotti Raymond et al., 1999; Menotti Raymond et al., 1997) A Bayesian proce dure, implemented in the program STRUCTURE (Pritchard et al., 2000) was used to identify populations or genetic clusters and to estimate the genetic origin of individuals based on microsatellite allele frequencies. The STRUCTURE approach assumes departures from both Hardy Weinberg and complete linkage equilibriums to be indications of population substructure (Pritchard et al., 2000) In addition to assigning individuals to various lineages based upon composite microsatellite genotype, the analysis also allows for estimation of prop ortion of genetic contribution from each group for individuals of mixed origin. We used results from the

PAGE 25

25 STRUCTURE analysis, along with pedigree results and field evidence, to assign panthers to 3 groups that reflected the genetic makeup of the south Flor ida population (pre introgression type panthers, F 1 admixed panthers, and other admixed panthers). Pre introgression type panthers represent the genotypes present on the landscape prior to the introgression program and this group is composed mostly of non admixed Florida panthers, which had no direct non Florida relatives or < 10% non Florida genetic contribution based on STRUCTURE analyses. Additionally, 1 pre introgression type panther in the sample of individuals used to investigate the influence of ge netic ancestry and heterozygosity on survival ( n = 98) was a descendent of a panther from the Everglades genetic population with documented genetic links to Central America (Culver et al., 2000; Johnson et al., 2010; O'Brien et al., 1990; Roelke et al., 1993b) Admixed panthers were mostly th e descendents of the introduced female Texas pumas released in 1995, except for 5 radio collared individuals that shared genotypic similarities with pumas maintained in a large enclosure on the Seminole Indian Reservation (SIR) adjacent to BCNP. Radio col lared panthers were documented entering the enclosure and captive pumas were documented to have left the enclosure during our study. This captive population contained animals of unknown origins that had a genetic affinity with North American pumas (FWC, N PS, National Cancer Institute, unpublished data). Differences in fitness between classes of hybrids are possible and hybrid vigor is predicted to be strongest in the F 1 generation (Arnold and Hodges, 1995; Burke and Arnold, 2001) so we included 2 classes of admixed panthers in our analyses: F 1 and other admixed. We defined F 1 admixed panth ers as any offspring produced by matings

PAGE 26

26 between a Texas female and a pre introgression type male. As such, F 1 panthers in our sample were products of matings between Texas females and males that were either non admixed Florida panthers or Everglades pant hers. Of the 20 known kittens produced by Texas pumas, 15 were F 1 admixed panthers (the remaining 5 were sired by F 1 males). Ten of the 12 natal dens of the Texas pumas were visited shortly (<35 days) after parturition and all kittens were marked at thes e dens with transponder chips. Two F 1 kittens (1 from each litter) were captured from litters produced at the 2 dens not visited while they were still dependent offspring. Of the 15 known F 1 kittens, 8 were captured or recaptured and radio collared, wher eas t he fates of the other 7 F 1 kittens are unknown. In summary, all individuals in our radio collared sample were placed into 1 of 3 categories for our survival analyses: pre introgression type panthers ( n = 41), F 1 admixed panthers ( n = 8), and other ad mixed panthers ( n = 49). We were also able to further classify most individuals from the other admixed category into 3 finer categories: backcrossed to Florida admixed panthers ( n = 16), backcrossed to Everglades admixed panthers ( n = 4), and backcrossed to Texas admixed panthers ( n = 14). We considered potential differences in survival between pre introgression type, F 1 admixed, and these finer admixture distinctions (Appendix B ), but because admixed panthers beyond the F 1 generation did not differ subst antially in their survival, we combined all non F 1 admixed panthers into the other admixed category for our main analyses. We estimated microsatellite based average individual heterozygosity with the program MICROSAT (Minch et al., 1995) to investigate the influence of heterozygosity on survival during the same time period as the ancestry analysis. We also compared

PAGE 27

27 the distribution of heterozygosity values between panthers of our 3 ancestry categories usin g permutation tests (Efron and Tibshirani, 1998) Index of A bundance To examine the potential effects of panther density on sub adult and adult survival, we used minimum population counts as an index of abundance (McBride et al., 2008) These counts were based on radio tracking and field evidence of sub adult and adult panthers (tracks, scat, kills, and scrapes; McBride et al., 2008) The period 1981 1984 was characterized as a discovery period, when the researchers were still finding the panthers on the landscap e Because the counts from 1985 1995 were fairly constant, we applied the count from 1985 as the abundance index for 1981 1984. In 1994 and 2004, the panther surveys were incomplete, so complete counts were not possible We followed McBride et al. (2008) and extrapolated these missing values by using the mid point b etween counts from the years immediately prior to and after the years in question (1994 and 2004). Survival Analyses We estimated annual survival and examined the effects of covariates using a daily time scale and Cox proportional hazard regression (Cox 1972; Therneau and Grambsch, 2000) We right censored panthers that lost their collars or whose collars failed on the last day that an active signal was heard. An important assumption of survival analyses is that radio failures are independent of mort ality (Therneau and Grambsch, 2000) We are confident that this assumption was met in our study as we were able to confirm radio failures by subsequent recapture or recovery in 22 of 32 cases.

PAGE 28

28 We organized the data with rec ords for each panther year combination (Fieberg and DelGiudice, 2009) Time in the Cox model (the baseline) was defined as da y within the year. When a panther changed age class within a year, we created two records: one record with the younger age class right censored on the transition day and another record with the older age class left truncated (staggered entry) on the same day. To clustered by individual (except when including random effects; Fieberg and DelGiudice, 2009; Therneau and Grambsch, 2000) which were generally extremely close to uncorrected standard errors. We used the Fleming Harr ington method to generate survival estimates from the Cox analysis (Therneau and Grambsch, 2000) All survival analyses were performed in R version 2.8.1 (R Development Core Team, 2010) using the survival (version 2.34 1) and kinship (version 1.1.0 22) packages and additional code that we developed for our analyses (available on request). For our s urvival analyses, we used an information Information Criterion; AIC) for model selection and statistical inference (Burnham and Anderson, 2002) Information theoretic approaches allow comparison of non ne sted models, selection of models that should best predict future data (from the same statistical population), and model averaged estimates to address model selection uncertainty (Burnham and Anderson, 2002) We calculated AIC values, Akaike i difference between AIC value of the i th model and the top ranked model) and Akaike weights ( w i the weight of evidence that the i th model is actually the best model of the models being considered given the data) as in Burnham an d

PAGE 29

29 4 to be plausible models with less empirical the sum of the weights of models including a given variable can be interpreted as a measure of its importance, relative to other variables (Burnham and Anderson, 2002) We calculated model averaged estimates of annual survival and u nconditional variance for each level of categorical variables and across continuous covariates (Burnham and Anderson, 2002; Tinker et al., 2006) To generate these model averaged estimates of survival we used all models included in the model selection process for a given analysis, weighted by their Akaike weights. Models with no effects of the covariates being presented were included in the averages (as having the same survival for all values of the covariate) ; therefore, the model averages represent unconditional estimates of survival (Burnham and Anderson, 2002) The unconditional standard errors estimated around the model averages are generally larger than standard errors estimated for parameters of single models and should not be interpreted as measures of statistica l significance. We conducted 2 separate survival analyses: a main analysis to estimate survival from 1981 2006 with the entire dataset and a subset analysis using data from sub adults and prime adults from 1997 2006 to investigate the effects of the intro gression program on survival. In the first set of analyses we investigated the influence of sex, age class, index of abundance, and year, and included 144 radio collared sub adults and adults from 1981 2006 (1 panther whose death was capture related wa s excluded). We started by

PAGE 30

30 Because the sub adult and prime adult age classes were defined differently for males and females, and because we had a priori reason to beli eve that the patterns of mortality would be different across ages for the two sexes, we only considered the interactive effect of sex and age for the difference between sub adults and adults. We considered difference s in survival between prime adult and o lder adult age classes both additively and interactively with sex, which resulted in 5 sex and age class models. We used the best model or models (i.e. lowest AIC) from the preceding analysis as base models to test for the effect of abundance o n panther su rvival. We also investigated potential temporal variation in survival using Gaussian shared frailty models with year as a random effect (Therneau and Grambsch, 2000) Each year was treated as a separate category, except for 1981 1986, which w ere combined as one category due to small sample sizes. We estimated temporal variance (variance of the random effect) for each model using a Laplace approximation under a maximum likelihood approach (Pankratz et al., 2005) Because our intent was to estimate the temporal variance rather than produce an estimate of survival for each year, we counted the random effect as a single parameter for purpos es of model comparison (Vaida and Blanchard, 2005) In a second set of analyses, we investigated potential effects of genetic introgression, in terms of gen etic ancestry and genetic diversity, on survival rates. Given that the first offspring produced by a Texas female did not appear in our sample of sub adults and adults until 1997, we conducted this subset analysis using data only from 1997 until the end o f this study on December 31, 2006. Only 2 admixed panthers (born during 1995 and 1996) reached 10 years of age by the end of 2006, and both for

PAGE 31

31 periods of <1 year ; thus, we excluded older adult panthers from these analyses to account for possible differen ces in survival for older animals. In addition to including an ancestry model where the 3 ancestry classes were separated (Ancest1), we included ancestry models where other admixed panthers were combined with either pre introgression type (to test specifi cally for F 1 hybrid vigor; Ancest2) or F 1 panthers (to test for general hybrid superiority; Ancest3). We combined these 3 models with the effects of sex and age class (sub adult and prime adult; Age3), and the effect of heterozygosity (Het). Some radio collared panthers were removed from the wild and held in captivity for various reasons, and some received medical treatment. We right censored panthers that were permanently removed from the wild due to livestock depredation incidents ( n = 2) on the date of removal. We also right censored panthers that were temporarily removed for treatment and rehabilitation of injuries judged not to have been fatal ( n = 3), capture related injuries ( n = 3), or for reproductive evaluation ( n = 1) on date of removal and t hen re entered them into the analyses upon release to the wild. Panthers removed for various reasons as kittens (<1 year old) and later released into the wild ( n = 7) were entered into our analyses either on the day of release (if adults or sub adults) or upon reaching the sub adult age class after being released as kittens. Panthers removed from the wild due to injuries or illness judged to be fatal ( n = 4) were treated as mortalities and the injury or illness was assigned as cause of death. To estimat e and model the importance of different mortality agents on rates and patterns of mortality for sub adult and adult Florida panthers, we performed cause

PAGE 32

32 specific mortality analyses. Details of the methods used for these analyses can be found in Appendix C Results Age, Sex, Index of A bundance, and Year Models that incorporated the interactive effects of sex and age class (either 2 or 3 age classes per sex) had substantially higher empirical support than models with neither or only the effect of sex ( Table 2 1). Survival rates were higher for females than for males, but these varied among age classes ( Table 2 2). For males, prime adults had the highest survival, whereas for females, sub adults had the highest survival ( Table 2 2). The highest ranked model included an additive effect of sex and old age, indicating that survival for this age class differed from other age classes, but there was no support fo r an interactive effect of sex and old age ( Table 2 1). Survival estimates were lower for older adults of both sexes ( Table 2 2). Models including the effect of abundance always had less empirical support than the equivalent models with no such effect ( Table 2 1), indicating no evidence of density dependence in sub adult and adult survival. The models wi th a random effect of year had marginally lower support than the equivalent models with no random effect ( Table 2 1), and the estimated temporal variance of the hazard rate for model 2 was small (0.115), suggesting that the temporal variance in survival ra tes was also small. Genetic Ancestry and Heterozygosity There was strong evidence that genetic ancestry influenced survival as all models with substantial emp the top ranked model containing neither ancestry nor heterozygosity was 7.61, and the sum of the weights of models including ancestry variables was 0.928 ( Table 2 3).

PAGE 33

33 Ancestry variables included in the top ranked models were all in agreement that the survival of F 1 admixed panthers differed from that of other ancestry categories (sum of weights of such models was 0.890); however, evidence for a difference between other admixed panthers an d pre introgression type panthers was weaker (sum of weights = 0.315; Table 2 3). Model specific and model averaged survival was higher for F 1 admixed panthers than for other ancestry classes ( Table 2 4; Fig ure 2 1). There was also evidence that average heterozygosity influenced survival of sub adult and prime adult panthers after genetic introgression (1997 2006), as the sum of the weights of models including heterozygosity was 0.537 ( Table 2 3) and model averaged annual survival probability increased w ith heterozygosity (Fig ure 2 1). Average heterozygosity was highest for F 1 (mean = 0.337, SD = 0.035, n = 8), intermediate for other admixed (mean = 0.261, SD = 0.065, n = 49), and lowest for pre introgression type panthers (mean = 0.161, SD = 0.063, n = 41). Each ancestry group differed significantly in average heterozygosity from both other groups in pairwise comparisons (all p heterozygosity positively influenced survival even within ancestry groups, as the top ranked model included both factors ( Table 2 3). Although model averaged survival was similar for pre introgression type and other admixed panthers at a given level of heterozygosity, heterozygosity ranged higher for other admixed indivi duals, leading to an increased probability of survival (Fig ure 2 1; Table 2 4). The full AIC table and selected Cox model coefficients for t he genetic comparisons are presented in App endix D Robust z tests on individual coefficients (in the co ntext of sex and age class) also indicated that F 1 admixed panthers survived better than

PAGE 34

34 panthers in other ancestry classes and that heterozygosity positively influenced survival (Appendix D ). Cause Specific Mortality The greatest cause of mortality for radio colla red Florida panthers was intraspecific aggression, followed by mortality from unknown causes, vehicles, and other (Appendix C ). When sexes were combined, pre introgression type panthers had a higher level of mortality due to intraspecific aggression than admixed panthers ( z = 2.4 0 4, p = 0.01 6 risk ratio = 3.06). Increasing heterozygosity also significantly decreased the risk due to intraspecific aggression ( z = 2. 9 4 3 p = 0.00 3 risk ratio = 0.480). Additional details of the results of these analyses a re available in Appendix C Discussion We provide evidence of hybrid vigor in the panther population following intentional introgression, as genetic ancestry and heterozygosity strongly influenced survival of sub adult and prime adult panthers. Our findin gs correspond very closely to the outcomes of hybridization on an inbred population predicted by theory in that survival and heterozygosity were higher for F 1 admixed individuals compared with pre introgression type Florida panthers, but evidence for highe r survival in admixed generations beyond the F 1 was far weaker (Burke and Arnold, 2001; Crow, 1948) It should be noted that we did not compare survival between F 1 panthe rs and the released Texas females; thus we investigated hybrid vigor by comparing survival between F 1 panthers and only 1 of the parental populations. Hybrid vigor was first recognized by crossing divergent lines of agricultural plants and extensive resea rch in agricultural genetics has confirmed the phenomenon (Birchler et al., 2003; Crow, 1948; Whitlock et al., 2000) Much less is known about the fitness

PAGE 35

35 consequences of hybridization for plants and animals in natur e (Campbell and Waser, 2001; Grant a nd Grant, 1992) For animals, hybrid vigor has been invoked by previous studies for a variety of species including insects (Ebert et al., 2002) fish (Rosenfield et al., 2004) salamanders (Fitzpatrick and Shaffer, 2007) and corals (Slattery et al., 2008) We are unaware of previous studies of free ranging mammals demonstrating superior fitness of admixed individuals (or superiority of a component of fitness, as in our example) that was clearl y due to intrinsic hybrid vigor rather than hybrids being favored under specific environmental conditions. Broadly speaking, hybrid fitness may be influenced by endogenous or exogenous selection, and hybrid superiority is often assumed to be primarily due to the latter (Burke and Arnold, 2001) as hybrids may be favored under environmental conditions that vary across time (Grant and Grant, 1992) or space (Good et al., 2000; Moore, 1977) Our demonstration of higher survival in F 1 admixed panthers relative to an inbred parental type is consistent with an intrinsic hybrid vigor effect (Burke and Arnold, 2001) a nd temporal variation in environmental conditions were minimized in our study by limiting the analysis to years during which admixed and pre introgression type panthers co existed on the south Florida landscape. However, environmental conditions could hav e varied spatially across the range of the panther and potentially contributed to the higher survival of F 1 admixed panthers in some instances We suggest that variation in at least 2 extrinsic factors, panther density and habitat conditions, could have i nfluenced panther survival across ancestry categories. First, survey results and capture efforts suggested that adult males were at very low density in ENP and portions of BCNP (FWC & NPS, unpublished data). Low local density of adult males could have pro vided a survival advantage to sub adult and adult

PAGE 36

36 F 1 panthers in these areas ( n = 4) by decreasing the risk of death by intraspecific aggression. This possibility is consistent with our results as pre introgression type panthers were more frequently kille d by intraspecific aggression than admixed panthers. Furthermore, ENP was likely devoid of females when the Texas females were released, which could have contributed to higher survival of female F 1 panthers in ENP ( n = 3) through decreased competition and reducing the need for dispersal. However, half of the radio collared F 1 panthers inhabited areas known to be occupied by both adult male and female panthers, so even if low density of adult panthers influenced our results in some portions of the range, i t would only offer a partial potential explanation for the higher survival of F 1 panthers. A second extrinsic factor that may have influenced our results is variation in habitat conditions across the range of panthers, and we recognize the potential for interactions between habitat quality, genetic ancestry, and survival. Theoretical (Moore, 1977; Moore and Price, 1993) and empirical work (Good et al., 2000; Rand and Harrison, 1989) suggest that hybrids sometimes thrive in different habitat types than parental types, leading to the production of relatively fit hybrids and to the establishment and maintenance of hybrid zones. Novel phenotypes produced through hybridization can allow for niche differentiation between hybrid and parental types when some of the available habitat is not suitable for the parental types (Buerkle et al., 2000; Lewontin an d Birch, 1966) It has been suggested that since introgression, panthers have moved into areas in south Florida that were not occupied during years of lower population size just prior to introgression (Pimm et al., 2006b) meaning t hat admixed individuals might have occupied and survived in areas that were not used by pre introgression type

PAGE 37

37 panthers. However, Creel (2006) pointed out that such an expansion into new areas could have simply been driven by demographics, as growing populations will often expand into new habitats. An intriguing possibility to consider with future analysis is whether admixed panthers (especially F 1 s) used different habitat types than pre introgression typ e panthers and whether genotype specific habitat selection patterns influenced survival (and other components of fitness). An ideal test of the hybrid vigor theory would involve a carefully designed experiment with large sample sizes for robust statistical inferences. However, such experiments and sample sizes are rarely possible for elusive and highly endangered species of large carnivores that typically occur at low density. Populations small enough to experience inbreeding often result in smaller sampl e sizes than would be preferred for more robust statistical inferences Nonetheless, analyzing these data is critically important in terms of increasing understanding of the dynamics of highly endangered populations Although our sample of radio collared F 1 individuals is numerically small ( n = 8), it represents a substantial proportion of the total F 1 offspring produced by the introgression program (38 53%, depending on the range of possible litter sizes for the 2 dens not visited) ; therefore the surviva l of the F 1 individuals we studied should be representative Pim m et al. (2006) evaluated the effects of the introgression program on panther demographic parameters, including survival, but their results with respect to adult survival were equivocal (Cree l 2006). Our dataset, analyses, and results differed from those of Pimm et al. (2006) in several important ways. Pimm et al. (2006) excluded 15 panthers because they did not know the genetic ancestry of these individuals, whereas

PAGE 38

38 we determined ancestry f or the entire dataset with updated genetic analyses completed in 2009. We also addressed aspects of the analysis by Pimm et al. (2006) following recommendations made by Creel (2006). First, we separated adults from sub adults to consider possible differe nces in age structure between pre introgression type and admixed categories, whereas Pimm et al. (2006) categorized all panthers independent analysis because these individu als, which survived poorly relative to other age classes, were underrepresented in the admixed classes. Second, Pimm et al. (2006) compared demographic parameters of some panthers from years prior to introgression to those of admixed individuals following introgression, thus failing to control for potential temporal variation in panther demographic variables. We limited our comparison of survival between admixed and pre introgression type genotypes to years when both co existed on the landscape (i.e., 199 7 2006). It should be noted that this restriction by itself has little effect on inference, except to weaken the evidence for an effect of heterozygosity (J. Hostetler, unpublished data). Perhaps the most important difference between our analysis and tha t of Pimm et al. (2006) is that we separated F 1 generation offspring from other admixed panthers, as recommended by previous researchers investigating fitness consequences of hybridization (reviewed by Arnold and Hodges, 1995; Burke and Arnold, 2001) Sub adult and adult survival improved most dramatically for F 1 admixed panthers after introgression and this effect would be diluted by pooling all admixed panthers into a single category. Finally, we included data on heterozygosity to determine whether differences in survival across genotypes were associated with di fferences in genetic diversity. Thus, we believe our approach has provided additional

PAGE 39

39 insight into the effect of the introgression program on sub adult and adult panther survival and represents a well documented example of the utility of genetic introgres sion in improving a demographic parameter of an inbred population. Age and sex influenced survival of panthers as survival rates for females were higher than for males in each age class. The sub adult age class is the period when male panthers are dispers ing and attempting to locate and establish home ranges (Maehr et al., 2002b) and our results indicate this is a dangerous period for male panthers. Converse ly, female survival rates were highest for the sub adult age class. Female pumas are often philopatric, dispersing less frequently and for shorter distances than males (Maehr et al., 2002b; Sweanor et al., 2000) consistent with most species of polygynous mammals (Greenwood, 1980) Our results also strongly suggest that older small sample size of older adults and the potential bias against detecting survival senescence due to heterogeneity in individ ual survival (Cam et al., 2002) Our results are an important first step to determining the demographic mechanisms that led to the nu merical increase of panthers after introgression and indicate that intentional introgression can be a valuable tool for conserving small, inbred populations. However, potential variation in the response of populations to admixture, the possibility of outb reeding depression through loss of co adapted gene combinations or adaptation to local environmental conditions (Edmands, 2007; Greig, 1979; Templeton, 1999) and the problem of losing genetically unique populations thr ough swamping (Allendorf et al., 2001; Creel, 2006) suggest that intentional admixture of wild populations should be undertaken only when ext inction appears imminent, as in the

PAGE 40

40 panther example. Despite the success of the introgression program for panthers (at least in the short term), the problems that led to a small population size and inbreeding, habitat loss and isolation from other populat ions, have not been corrected and will likely be exacerbated as development and human population growth are projected to increase in south Florida (Kautz et al., 2006) Therefore, investigating the influence of introgression on other demographic parameters (e.g., kitten survival and fecundity) and population growth and determining the longevity of any demographic benefits associated with intentional introgression will be important next steps for evaluating whether genetic aug mentation will be an effective long term management tool for panthers.

PAGE 41

41 Table 2 1. Model comparison results for the effects of sex, age class abundance index, and year on Florida panther survival. Abundance was included as a linear trend and year was i ncluded as a random effect using the top ranked sex and age model as a base. For each model, we present the number of Akaike weight ( w i ) A) Model comparison results for the effects of sex and age class on survival. B) Model comparison results for the effects of abundance and year on survival. Model Parameters w i A. Sex and age class models 1. Sex Age1 a + Older b 4 0 0.658 2. Sex Age2 c 5 1.98 0.245 3. Sex Age1 3 4.69 0.063 4. Sex 1 5.95 0.034 5. Constant d 0 15.25 0.000 B. Additive effects of abundance (trend) and year (random) 1. Sex Age1 + Older 4 0.00 0.292 2. Sex Age1 + Older + rand(Y ear) e 5 0.18 0.267 3. Sex Age1 + Older + Abundance f 5 0.26 0.256 4. Sex Age1 + Older + Abundance + rand(Year) 6 0.92 0.185 a Age1 divides panthers into sub adults (1 2.5 and 1 3.5 yrs for females and males respectively) and b Older refers to older adult effec t for both sexes whereas model 2 allows for different older adult effects between sexes. c Age2 divides the panthers into sub adults (same as Age1), prime adults (2.5 10 and 3.5 10 yrs for females and males respectively) and older d No predictor variables e rand(Year) refers to a random effect of year as a categorical variable, with 1981 1986 grouped together (temporal variance). f Abundance refers to a linear trend in survival by abundance index.

PAGE 42

42 Table 2 2. Model averaged annual survival rates ( ), standard errors ( ), and number of Florida panthers tracked ( n ) in sex and age class categories. Models 1 5 (Table 2 1) were used for model averages. Females Males Category n n Sub adult a 0.951 0.034 40 0.713 0.049 54 Prime adult b 0.872 0.023 64 0.799 0.036 44 Older adult c 0.760 0.056 12 0.635 0.083 11 a 1 2.5 and 1 3.5 years old (estimated) for males and females, respectively b 2.5 10 and 3.5 10 years old (estimated) for males and females, respectively c

PAGE 43

43 Table 2 3. Model comparison results showing top Information Criterion [ AIC] < 4) for the effects of ancestry, heterozygosity, sex, and age class (sub adults and prime adults only) on Florida panther survival during 1997 2006. For each model, we present the number of parameters, w i ) The full table is presented in Appendix D. Model Parameters w i 1. Sex Age3 a + Ancest2 b + Het c 5 0.00 0.244 2. Sex Age3 + Ancest2 4 0.23 0.217 3. Sex Age3 + Ancest1 d 5 1.35 0.124 4. Sex Age3 + Ancest1 + Het 6 1.98 0.090 5. Sex + Ancest2 + Het 3 2.77 0.061 6. Sex + Ancest2 2 2.98 0.055 7. Sex Age3 + Het 4 3.09 0.052 a Age3 differentiates sub adult (age 1 2.5 for females and 1 3.5 for males) and prime adult (ages 2.5 10 for females and 3.5 10 for males) panthers. b Ancest2 divides panthers into two ancestry categories: F 1 admixed, and other admixed and pre introgression ty pe combined. c Het refers to individual average heterozygosity. d Ancest1 divides panthers into three ancestry categories: F 1 admixed, other admixed, and pre introgression type.

PAGE 44

44 Table 2 4. Estimated annual survival rates (with and number of panthers) by ancestry, sex, and age class. Group Pre Introgression Type Other Admixed F 1 Admixed All 0.775 (0.039; 41) a 0.821 (0.036; 49) a 0.978 (0.021; 8) a Females 0.837 (0.041; 20) b 0.864 (0.035; 22) b 0.982 (0.018; 6) b Males 0.713 ( 0.055; 21) b 0.758 (0.056; 27) b 0.967 (0.032; 2) b Female Sub Adults c 0.953 (0.049; 9) d 0.964 (0.036; 14) d 0.995 (0.007; 4) d Female Prime Adults e 0.803 (0.047; 19) d 0.848 (0.041; 22) d 0.977 (0.022; 6) d Male Sub Adults f 0.606 (0.088; 14) d 0.686 (0.073; 22) d 0.949 (0.050; 2) d Male Prime Adults g 0.789 (0.059; 12) d 0.836 (0.056; 13) d 0.976 (0.024; 2) d a Estimated using Ancest1 model (divides panthers into pre introgression type, other admixed, and F 1 admixed). b Estimated using Sex + Ancest1 model. c Female sub adults were 1 2.5 years old. d Estimated using Sex Age3 (sub adult vs. prime adult) + Ancest1 model. e Female prime adults were 2.5 10 years old. f Male sub adults were 1 3.5 years old. g Male prime adults were 3.5 10 years old.

PAGE 45

45 Figure 2 1. Mod el averaged effects of ancestry and heterozygosity on annual survival of female prime adult Florida panthers from 1997 2006. The range of heterozygosity values presented for each ancestry category approximately corresponds to those found in the dataset (a ll sub adult and prime adult panthers from 1997 2006 with that ancestry), except for one outlier with high heterozygosity excluded from the range for the pre introgression type. Error bars represent unconditional standard errors.

PAGE 46

46 CHAPTER 3 GENETIC INTROGRESSION AND THE SURVIVAL OF FLORIDA PANTHER KITTENS Introduction Rigorous estimate s of survival rates and understanding factors influencing those rates are critical for discerning proximate and ultimate causes of dynamics, regulation, and persistence of populations (Newton, 2004; Ozgul et al., 2006) and for devising and implementing management interventions for species conservation (Davis et al., 2007) Population dynamics of many vertebrate species are strongly influenced by changes in survival rates (Heppell et al., 2000; Oli and Dobson, 2003; Stahl and Oli, 2006) Estimates of survival also are necessary for the parameterization of population models, which are essential tools in conservation biology for evaluating population performance, diagnosing the causes of low numbers or population declines, developing solutions to those problems, and determining research priorities (Caswell, 2001; Caug hley and Gunn, 1996; Fujiwara and Caswell, 2001) These population models can often be improved by including the estimated effects of factors such as population density or genetic diversity on survival. The Florida panther ( Puma concolor coryi ) is a subs pecies of puma that once ranged throughout the southeast United States. Florida panthers give birth to 1 4 kittens, which remain in the den for approximately 7 8 weeks (Maehr et al., 1990) Average age of independence is just over 1 year (mean = 397 days, SD = 74 days, n = 3 2 ; FWC & NPS, unpublished data). The Florida panther is now restricted to <5% of its historic range in south Flo rida (Maehr et al., 2002a) where it occurs in a single population with ca. 100 sub adults and adults (McBride et al., 2008) It has been federally listed as endangered since 1967. Small population size, isolation, loss and

PAGE 47

47 fragmentation of habitat, road rel ated mortalities, and other anthropogenic influences continue to threaten the long term persistence of Florida panthers (Kautz et al., 2006) Small populations are expected to suffer from inbreeding depression (Frankham et al., 2002) ; earlier studies of Florida panthers found evidence of inbreeding and recommended genetic introgression via the introduction of a subspecies of puma whose range historically abutted with the range of the Florida panther (Barone et al., 1994; Roelke et al., 1993b) Based on this recommendation, 8 female Texas pumas ( P. c. stanleyana ) were released into the Florida panther population in 1995. Although the panther populat ion has experienced an almost fourfold increase since 1995 (McBride et al., 2008) the success of the introgression continues to be a topic of debate (Maehr et al., 2006; Onorato et al., 2010; Pimm et al., 2006a) Th ere have also been non management releases of captive pumas into south Florida; some of these (between 1957 and 1967) resulted in a genetic population with partial Central American ancestry in the Everglades (O'Brien et al., 1990) In 2003, the U.S. Fish and Wildlife Service (USFWS) and the Florida Fish and Wildlife Conservation Commission (FWC) convened the Scientific Review Team (SRT) to review the status of Florida panther science, and to make research and management recommendations. The SRT thoroughly reviewed existing literature on demography and population ecology of the panther, and found t hat rigorous estimates of age or stage specific survival rates and cause specific mortality rates were not available. Furthermore, SRT found that previously reported estimates of the survival rate of Florida panther kittens (Mae hr and Caddick, 1995) and those used in population viability analyses (Maehr et al., 2002a) were not based on reliable inference methods and raised

PAGE 48

48 doubt on predictions regarding the persistence of the panther population (Beier et al., 2003; Beier et al., 2006; Gross, 2005) The SRT recommended reanalysis of existing data to rigorously estimate age specific survival rates and test hypotheses regarding factors influencing survival rates ( Beier et al., 2003) In accordance with SRT recommendations (Beier et al., 2003) our goals were to provide rigorous estimates of survival of Florida panther kittens (defined here as age 0 1 year), and to evaluate factors influencin g kitten survival. We analyzed long term radio telemetry (1982 2008) and Passive Integrated Transponder (PIT) tagging (1995 2008) data within a live recapture, dead recovery modeling framework (Burnham, 1993; Williams et al., 2002) We tested the hypotheses that kitten survival: (1) would not differ between sexes, due to little or no sexual dimorphism during this stage of life; (2) would decrease with increasing litter size due to large litters potentially being more energetically demanding to dams and more difficult to protect from p redators; (3) would be higher during the wet season (June November) than during the dry season (December May), due to potential higher prey availability during the wet season; and (4) would be negatively influenced by population abundance due to densit y dependent effects. In many species of carnivores, survival of neonates is generally lower compared to older juveniles (Garrison et al., 2007; Logan and Sweanor, 2001) Thus, we also hypothesized that kitten survival (5) would increase with age during their first year. Results of rigorous analyses evaluating the effects of genetic introg ression on kitten survival could assist in setting future management and research priorities. Thus, we tested these additional hypotheses: (6) kitten survival probability would increase with

PAGE 49

49 average heterozygosity, because loss of genetic variation (espec ially due to inbreeding) has been shown to negatively influence fitness and its components (Crnokrak and Roff, 1999; Ralls and Ballou, 1983) ; and (7) su rvival probability would be higher for admixed kittens than for canonical (i.e., pure Florida panther) kittens as outcrossing has been shown to increase various measures of fitness for small, inbred populations (Heschel and Paige, 1995; Madsen et al., 1999; Vrijenhoek, 1994) Materials and M ethods Field M ethods Florida panthers and Texas pumas were captured and monitored by radio telemetry from 1982 2008 across the range of the subspecies in south Florida, USA (Kautz et al. 2006) by biologists from the FWC and National Park Service (NPS) using methods described by Belden et al. (1988) and Land et al. (2008). Age of captured panthers (if unknown) was estimated using a combin ation of toothwear, pelage characteristics, size, and developmental stage of teats and reproductive organs. Blood and tissue samples were collected for health assessment and genetic analysis. Radio collars were affixed on captured adults, sub adults, and older kittens (occasionally as young as 5 months old; age and size qualifications for collaring varied over time). Successive locations of females were continually assessed to determine the commencement of denning behavior; lack of movement between 3 4 fixes was taken as evidence of possible denning (Land et al., 1998) We visited dens 4 35 days post partum (approximate median = 14 days). Kittens in dens were counted, sexed, sampled for genetic mate rial, and implanted with passive integrated transponder (PIT) tags. Occasionally, dens were checked after dams vacated the area for evidence of dead kittens.

PAGE 50

50 classified as having lost that previous litter because 1) females are unlikely to copulate while they have dependent kittens, 2) the gestation period is about 3 months (Maehr, 1992) and 3) the minimum age of independence recorded is 9 months (D. Onorato, unpublished data). Similarly, litters whose d classified as having failed. These two types of litter failure data generally do not provide an estimated death date for individual kittens, only an upper limit on the date when litters could have failed. Heterozyg osity and Genetic A ncestry We extracted total genomic DNA from blood and tissue samples obtained from wild caught panthers and captive pumas from south Florida and west Texas during 1982 2007. We amplified and scored 23 microsatellite loci following prev iously described PCR amplification conditions (Menotti Raymond et al., 1999; Menotti Raymond et al., 1997) Derived genot ypes were used to calculate average individual heterozygosity using the program MICROSAT (Minch et al., 1995) A Bayesian procedure, implemented in the program STRUCTURE (Pritchard et al., 2000) was used to identify populations or genetic clusters (run without preset groups/ training individuals) and to estimate the genetic origin of individua ls. The STRUCTURE approach assumes departures from both Hardy Weinberg and complete linkage equilibriums to be indications of population substructure (Pritchard et al., 2000) We used results from the STRUCTURE analysis along with pedigree information and field evidence to assign panthers to groups that reflect the genetic makeup of the southern Florida population: canonical panthers (92 to tal panthers; 50 kittens), backcrossed to canonical admixed panthers (92 total panthers; 80 kittens), backcrossed

PAGE 51

51 to Everglades admixed panthers (19 total panthers; 18 kittens), backcrossed to Texas admixed panthers (42 total panthers; 38 kittens), and F 1 admixed panthers (15 total panthers; 13 kittens). Canonical Florida panthers showed no evidence of non Florida genetic admixture (no direct non Florida relatives or < 10% non Florida genetic contribution based on STRUCTURE analyses). Admixed panthers wer e mostly the descendents of the introduced female Texas pumas released in 1995, except for 6 radio collared individuals (0 kittens) that were genotypically similar to pumas maintained in a large enclosure on the Seminole Indian Reservation (SIR) adjacent t o Big Cypress National Preserve. These SIR panthers, panthers of unknown ancestry (4 kittens), and admixed panthers whose ancestry could not be accurately determined (total = 54, of which 43 were kittens) were excluded from the survival analysis that inco rporated ancestry. Backcrossed to canonical panthers were admixed panthers with predominantly (>50%) canonical heritage resulting from admixed panthers breeding with canonical panthers. Backcrossed to Everglades were admixed panthers with predominantly (> 50%) Everglades heritage resulting from breeding events between admixed and Everglades panthers. Backcrossed to Texas admixed panthers were admixed panthers of predominantly (>50%) Texas heritage resulting from admixed panthers breeding with introduced Te xas pumas. We defined F 1 admixed panthers as any offspring produced by matings between a Texas female and a pre introgression type male (either canonical Florida panther or Everglades panther). In theory, Everglades panthers (with documented genetic link s to Central American pumas) represent a sixth genetic

PAGE 52

52 population; however, there were no recorded kittens of this group during the post introgression study period (June 1995 May 2008). We also repeated our analyses based on a simpler ancestry classifi cation (canonical, F 1 admixed, and other admixed) that allowed the inclusion of admixed panthers with unknown ancestry; results of these analyses are presented in Appendix E Index of A bundance To examine the potential effects of panther density on kitte n survival, we used minimum population counts as an index of abundance (McBride e t al., 2008) These minimum counts were based on radio tracking and field evidence of sub adult and adult panthers (tracks, scat, kills, and scrapes) and collected for calendar years (e.g., total number of sub adult and adult panthers known to have been alive in the wild over the period January 2005 December 2005). In contrast, we analyzed annual kitten survival for periods extending from June May. We applied the sub adult/adult calendar year count for year x to the kitten June x May x + 1 year, because puma kitten mortality is generally highest in the first three months of life (see Results), so adult abundance during that period seemed most likely to affect survival. In 2004, the panther survey was incomplete, so the minimum count is unknown. We followed McBride et al. (2008) and extrapolated it halfway between the 2003 and 2005 counts. Data Preparation and A nalysis We used two datasets for the analysis of kitten survival. The first contained the capture (of live panthers) and recovery (of dea d panthers) histories, observations of litter failures (i.e., loss of entire litters), and radio tracking data for panthers that were initially PIT tagged in the den between June 1995 and May 2008 (13 years). The

PAGE 53

53 second contained the capture, recovery, an d radio tracking histories of all other panthers between June 1982 and May 2008 (26 years). Details of data preparation can be found in Appendix F We analyzed data using the R (R Development Core Team, 2010) package RMark (Laake and Rexstad, 2010) as an interface for program MARK (Wh ite and Burnham, 1999) recapture dead recovery modeling framework (Burnham, 1993; Williams et al., 2002) The Burnham model has four parameters: survival probability ( S ), recapture probability ( p ), recovery probability ( r ), and fidelity ( F ). For all models, we fix ed F =1 for all panthers, as the recapture and recovery areas are the same and encompass the entire range of the Florida panther. We set p and r for radio collared animals to 1, because we could count on knowing their status each year. Although the focus of our study was kitten survival, we also used data from sub adult and adult panthers to efficiently parameterize the likelihood functions. For example, the likelihood of a kitten being PIT tagged in the den and then recaptured as a 2 year old panther de pends on the probability of it surviving as a kitten, the probability of it surviving the first year of being a sub adult, and the probability of recapture as a sub adult. Thus, to estimate the kitten survival probability from data such as these, we also need to estimate the other aforementioned probabilities. We used a model with an interaction of sex and age class (sub adult or adult), a reasonably well supported model (Benson et al., in revision) that could be applied to all individuals, to estimate the surv ival of sub adults and adults. The boundary between sub adult and adult panthers was originally defined at age 2.5 for females and age 3.5 for males (Benson et al., in revision) Transitions at these ages were impossible with a one year time step;

PAGE 54

54 therefore, w e fitted survival models with those ages rounded up and down to the nearest integer (to 2 and 3 for females and 3 and 4 for males). We used an information sample size; QAIC c ) for model selection and statistical inference (Burnham and Anderson, 2002) We calculated QAIC c c i difference between QAIC c value of the i th model and the top ranked model) and Akaike weights (the weight of evidence that a model is the best model of the models being considered given the data) as in Burnham and Anderson (2002). We calculated model averaged estimates of annual survival and unconditional variance for each level of categorical va riables and across continuous covariates (Burnham and Anderson, 2002; Tinker et al., 2006) To generate these model averaged estimates of survival we used all models included in the model selection process for a given analysis, with model specific estimates weighted by their Akaike weights. Models with no effects of the covariates being presented were included in the averages. We commenced by testing all a priori models for recapture and recovery combined with a priori models for survival of kittens, sub adults, and adults with various divisi ons of age classes. Because kittens could only be recaptured as sub adults the following year, there was no capture probability parameter for kittens (e.g., Pollock, 1981) ; for sub adult and adult panthers we estimated a single capture probability. We used models that estimated separate recovery rate for kittens and older panthers as well as those that estimated a single recovery rate for all panthers. We used a model that allowed kitten survival to vary between sexes and recapture probability to be different for kittens (the most parameterized model that applies to the

PAGE 55

55 full dataset) to test for overdispersion of the dat a due to lack of independence within litters ( ) using the data bootstrap approach developed by Bishop et al. (2008) ; a > 1 indicates overdispersion. We estimated as the ratio of the replication based estimate of th e variance of bootstrapped survival estimates (this variance incorporates overdispersion associated with lack of independence) to the model based estimate of variance of survival (estimate d from the original dataset). For the bootstrap approach, w e sample d litters from the dataset with replacement. Panthers first captured as sub adults or adults were treated as litters of size 1 for the bootstrapping. If the model failed to estimate kitten survival parameters, the iteration was discarded. We repeated th e resampling and estimation until we had 1,000 estimates of female and male kitten survival and estimated the mean as in Bishop et al. (2008). The estimate of overdispersion was 1.37 for females and 1.60 for males, with a mean of 1.48, indicating slight overdispersion of data. We used this value as a variance inflation factor and to calculate QAIC c We selected the model with the lowest QAIC c from preceding analyses as the base model to test hypothes es regarding factors affecting kitten survival ( Table 3 1). The models for ancestry ( Table 3 1A, models 3 6, Table 3 1B) and heterozygosity ( Table 3 1A, models 7 10) were run on data collected between 1995 and 2008 only due to the absence of admixed panth ers in Florida before 1995. We tested for the effect of ancestry and heterozygosity on survival of (1) kittens only; (2) sub adult and adult panthers only; (3) kittens and older panthers considered separately; and (4) panthers of all ages. Based on previo us results (Benson et al., in revision) however, we only considered 2 ancestry categories for sub adult and adult panthers (F1Adv; Table 3 1B).

PAGE 56

56 We tested for the effect of abundance on kitten survival ( Table 3 1A, model 11) using data collected between 1995 an d 2008. Data on kittens collected prior to 1995 were insufficient to test for the effect of abundance on kitten survival. Finally, we estimated process variance (random effect of year; a critical parameter for stochastic population models) in kitten sur vival using the Markov Chain Monte Carlo (MCMC) option in MARK (Lukacs et al., 2009; White et al., 2009) We used the Bayesian MCMC approach instead of a method of moments approach because some missing kitten survival estimates for specific year s (fixed effects categorical model used as an input for method of moments) made the method of moments approach unreliable (G. White pers onal communication ). We estimated process standard deviation in logit survival from the mean and 95% credible interval of the posterior distribution. To estimate the temporal variation beyond that explained by abundance, we also estimated the temporal standard deviation associated with year in the model that included abundance index as a covariate. Temporal standard deviations were converted from the logit scale using the mean kitten survival ( to ; without abundance index model) and the kitten survival computed for the mean abundance ( to ; abundance index model). We estimated percentage of the temporal variance explained by abundance as (Loison et al., 2002) We used the same n on informative priors as White et al. (2009); estimates from the fixed effects models were used as starting values where appropriate. We tested for lack of convergence using 10 Markov chains for each model (Gelman, 1996) and found no evidence of lack of convergence. For each chain we sampled the MCMC for 50,000 iterations, after 4,000 tuning samples and 1,000 burn in samples.

PAGE 57

57 Three M onth T ime Step D ata P reparation and A nalysis Using the data organized on an annual time scale, it was not possible to test for seasonal variation in kitten survival and age spec ific variation in survival within the first year of life. Therefore, we recoded data into 3 month time intervals. This necessitated leaving out litter failure data, because information about the 3 month interval in which actual kitten deaths occurred was generally not available from these data. We also did not include the recapture/recovery histories of panthers not PIT tagged in the den (without working radio collars), for computation efficiency (Appendix G ). Analysis of the data with a 3 month time s tep was similar to the annual time step analysis, except that there was now a possibility of recapture as a kitten. We commenced by testing all a priori models for recapture and recovery combined with a priori models for survival of kittens based on age ( Appendix G ). We report model specific and model averaged estimates of annual kitten survival probability and model averaged estimates of 3 month kitten and annual sub adult and adult survival probabilities (Burnham and Anderson 2002). Results The most pa rsimonious model (model 1, Table 3 2) indicated that survival ( S ) and recapture ( r ) probabilities differed between kittens and older panthers, and that S differed between sex and age classes among older panthers, with the sub adult to adult transition at t he upper age limit for both sexes. The four models with lowest QAIC c provided similar estimate of kitten survival ( Table 3 2); the model averaged estimate of annual kitten survival was 0.323 0.065 ( 1 SE). We selected the most parsimonious model from the initial model set (model 1, Table 3 2) as a base model for testing the effects of covariates on kitten survival. There

PAGE 58

58 was no evidence of an effect of sex ( Table 3 c of litter size model = 2.04) on kitten survival. Two of the ancestry models (base + k:CanDisF1Adv and base + k:F1Adv) failed to estimate a kitten survival parameter and were discarded. There was considerable evidence that ances try influenced kitten survival; the 6 most parsimonious models all included ancestry effects on kitten survival (models 1 6, Table 3 3A). The top 2 models indicated that survival of kittens differed substantially between two ancestry categories: 1) canonic al and backcrossed to canonical kittens and 2) F 1 backcrossed to Texas, and backcrossed to Everglades kittens ( Table 3 3A). Model averaged kitten survival was lowest for canonical kittens and highest for F 1 kittens ( Figure 3 1A). The most parsimonious model including the effects of heterozygosity on survival indicated that survival of panthers of all ages (including survival of kittens) was positively affected by average heterozygosity ( Table 3 3B, model 1). Although this model differed from the base mo c <2, the 95% confidence interval for the slope parameter did not include zero ( = 2.89; 95% CI: 0.16 5.63) suggesting a positive effect of average heterozygosity on panther survival. Model averaged kitten survival increased with heterozygo sity ( Figure 3 1B). There was considerable evidence that abundance negatively influenced kitten c of 8.79 ( Table 3 3C). Kitten survival declined with increasing population size ( Figure 3 2; = 0 .034, 95% CI = 0.053 0.016). We estimated temporal standard deviation of kitten survival ( ) at 0.228 (95% credible interval = 0.019 0.431) and 0.098 (95% credible interval = 0.004 0.286)

PAGE 59

59 using models without and with the abundance index, res pectively. Abundance index explained 81.5% of the temporal variation in kitten survival. There was considerable uncertainty in the base model selection, with 8 models in Table 3 c < 1. We reran all covariate tests using model 5 from Table 3 2 as a base model (model 5 was the highest ranked model with a different estimate of kitten survival than model 1). Model selection and statistical inference were barely affected by the switch of base models, and model averaged estimates that included bo th sets of models were somewhat lower, but generally had similar precision (Appendix H ). There was strong evidence that kitten survival was lowest in the first 3 months of life ( Figure 3 3). There was no evidence for seasonal variation in kitten survival (Appendix G ). Details of results based on the analysis of data organized using 3 month time steps are presented in Appendix G Discussion Models developed for population conservation and management have historically focused on either population dynamics (Morris and Doak, 2002) or threshold numbers required to avoid genetic problems (Franklin, 1980) The population ecology and population genetics of a species are frequently not integrated in these models. When they are, the demographic effects of inbreeding are often based on data from captive animals and/or program defaults (e.g., Brito, 2009; Maehr et al., 2002a; Nilsson, 2004) An important first step towards developing integrated models is determining the functional relationship between measures of inbreeding and juvenile survival in the wild, especially for managed populations.

PAGE 60

60 Overall, the model averaged estimate of annual survival probability of Florida panther kittens was 0.323 0.065 ; this estimate is lower than those reported f or western North American populations of pumas (range: 0.44 to 0. 72; Lambert et al., 2006; Laundr et al., 2007; Logan and Sweanor, 2001; Robinson et al., 2008) Our estimate of kitten survival also is substantially lower than those used in earlier demographic analyses of the Florida panther (Maehr and Caddick, 1995: 0.84 0.87; Maehr et al., 2002a: 0.735 1.0; Root, 2004: 0.62) Although our estimates of kitten survival varied slightly depending on the model (0.30 to 0.34; Table 3 2), they were consistently lower than those used in previous analyses. Furthermore, kittens were not tagged immediately after birth, and we could have missed kittens that had died before they were PIT tagged at the den site; thus, true survival may be slightly lower. However, since most of our data on kitten survival came from a period of population growth (McBride et al., 2008) we have no reason to believe that the kitten survival rate for this population is unsustainable. Ongoing work developing upda ted population models for Florida panthers may shed additional light on the role of kitten survival in determining population dynamics. Survival rates did not differ between male and female kittens. This result was as expected due to the lack of sexual dim orphism among kittens, although some authors have suggested sex specific differences in survival of puma kittens (Logan and Sweanor, 2001) There was no evidence that litter size influenced survival of kittens. This may refl ect the possibility that females most able to successfully raise large litters are those most likely to produce them.

PAGE 61

61 Perhaps the most dramatic panther conservation measure undertaken to date is the release of 8 female Texas pumas into the Florida panthe r population in 1995 (Seal 1994). Genetic introgression has been highly controversial (e.g., Maehr and Caddick, 1995; Maehr et al., 2006; Pimm et al., 2006b) Pimm et al. (2006) reviewed the arguments for and against genetic introgression in general, and in the Florida panther popu lation in particular. They reported that admixed kittens are more than three times based solely on the proportion of kittens PIT tagged in the den that were subsequent ly captured and (generally) radio collared. These kittens were recaptured at ages ranging from 6.5 months to 3.5 years and detection probability was not taken into account, so it is unclear how well these ratios represent differences in kitten survival. I n the present study, we used multiple sources of field data, a thorough determination of ancestry of nearly all kittens captured since genetic introgression occurred in 1995, and a statistically rigorous live recapture dead recovery modeling approach that takes detection probabilities into account to test for the effect of genetic introgression on the survival of Florida panther kittens. As recommended by Creel (2006), we limited this portion of the analysis to a period when both canonical and admixed kitt ens were on the landscape. We also were able to examine the effects of heterozygosity, which differed dramatically between canonical Florida panthers (0.161) and Texas pumas (0.318; Driscoll et al., 2002; Johnson et al., 2010) Furthermore, we were able to determine ancestry of panther kittens at a much finer scale than Pimm et al. (2006). We found that admixed kittens generally survived better than kitt ens born to canonical Florida panthers. Furthermore, F 1 admixed kittens had the highest survival,

PAGE 62

62 followed by backcrossed to Texas kittens (Fig. 1A). Finally, we found that average heterozygosity positively influenced survival of Florida panther kittens ( Fig. 1B). These results provide unambiguous evidence for the positive, population level impact of genetic introgression. Index of panther abundance negatively influenced kitten survival, which is suggestive of a density dependent effect. This could resu lt from infanticide by sub adult and adult males during territorial disputes or for mating opportunities (Garrison et al., 2007; Logan and Sweanor, 2001; Packer et al., 2009) from deaths of the dependent resources. We note, however, that our results are based on an index of abundance, and thus may not be sufficient to conclusively demonstrate density dependent influence on kitten survival. Interestingly, a substantial proportion of temporal variation in kitten survival was explained by temporal variance in index of abundance. Analysis of data organized with 3 month time intervals revealed no evidence for seasonal (i.e., wet vs. dry) variation in survival of kittens. However, we found that survival was lowest during the fir st 3 months of life, which is consistent with other large carnivore populations (Garris on et al., 2007; Logan and Sweanor, 2001) These results point to the fact that young kittens are particularly vulnerable to various mortality factors. Estimating survival of young can be challenging in elusive carnivores because they are rarely observed Additionally, in the case of panthers, kittens are typically too small to be fitted with radio transmitters when sampled at the den site a method that would be effective for monitoring their fate during the first year of life. Furthermore, the small p opulation size of endangered species and invariably small sample sizes may not

PAGE 63

63 be adequate for rigorous estimates of survival of the young. For example, estimates of kitten survival that did not include litter failure data lacked robustness as well as prec ision; models that were similarly ranked based on QAIC c provided very different estimates of kitten survival (Appendix G ). We addressed this challenge by using a modeling framework that can utilize multiple data sources. Resulting estimates of kitten survi val were more robust and precise, and also allowed us to test various hypotheses regarding factors influencing survival of Florida panther kittens. Similar approaches can be used to estimate survival of young in other elusive species that occur in low numb ers. Our study is the first to provide rigorous estimates of survival of Florida panther kittens. We found that kitten survival in our study population was lower than reported for western cougars, and substantially lower than values used in previous demogr aphic analyses of Florida panthers. These results are consistent with an earlier suggestion that results of a previous PVA (Maehr et al., 2002a) that assumed kitten survival of up to 87% may be unreliable (Beier et al., 2003) Finally, we found strong evidence that survival of admixed (particularly, F 1 ) kittens was substantially higher than that of purebred kittens, and that heterozygosity positively influenced survival of Florida panther kittens. These results are consistent with hybrid vigor theory (Crow, 1948; Frankham et al., 2002) and suggest that genetic introgression had positive population level effect s in our study population. Although there are concerns that genetic introgression can have undesirable or even negative impacts on populations (Edmands, 2007; Maehr and Caddick, 1995; Shields, 1993) our analyses revealed beneficial results of deliberate introgression for this population. A combination of these results

PAGE 64

64 with other recently published or on going assessments of demography pre and post introgression shoul d assist in directing management and research priorities that will ensure the continued persistence the critically endangered Florida panther.

PAGE 65

65 Table 3 1. List of kitten survival models. A) All a priori models that were used to test hypotheses regard ing the effect of covariates on kitten survival probability ( S ) and B) definitions of ancestry models. All models are based on the additive effects of covariates on survival of kittens (unless otherwise noted) from the base model(s) (base; see Table 3 2). s&a indicates an effect on sub adult and adult survival and k indicates an effect on kitten survival. A. All Models Model Description 1. S (base + k:sex) Additive sex effect 2. S (base + k:LS) Additive linear litter size effect 3. S (base + k:Ancestry) A dditive ancestry effect (see below) on kitten survival 4. S (base + Ancestry) Additive ancestry effect (see below) on survival of panthers of all ages 5. S (base + s&a:F1Adv) Additive F 1 ancestry effect (see below) on survival of sub adult and adult panthe rs only 6. S (base + k:Ancestry + s&a:F1Adv) Separate additive ancestry effects (see below) on kitten survival and sub adult and adult survival (F 1 only) 7. S (base + k:Het) Additive linear heterozygosity effect on kitten survival 8. S (base + Het) Additiv e linear heterozygosity effect on survival of panthers of all ages 9. S (base + s&a:Het) Additive linear heterozygosity effect on survival of sub adults and adults only 10. S (base + k:Het + s&a:Het) Separate additive linear heterozygosity effects on surv ival of kittens and survival of sub adults and adults 11. S (base + k:Abundance) Additive linear effect of abundance index B. Ancestry Models Term Definition Hypothesis IntAdv 2 ancestry categories: 1) canonical; 2) backcrossed to canonical, backcrossed to Everglades, b ackcrossed to Texas, and F 1 General introgression advantage (or disadvantage). AdmAdv 2 ancestry categories: 1) canonical and backcrossed to canonical; 2) backcrossed to Everglades, b ackcrossed to Texas, and F 1 General admixture adva nta ge (or disadvantage), but with b ackcrossed to canonical reverting to c anonical. 50TexAdv 2 ancestry categories: 1) canonical, backcrossed to canonical, and backcrossed to Everglades; 2) b ackcrossed to Texas, and F 1 Survival advantage (or disadvantag 50% Texas. CanDisF1Adv 3 ancestry categories: 1) canonical; 2) b ackcrossed to c anonical, b ackcrossed to Everglades, and b ackcrossed to Texas; and 3) F 1 Differences in survival between pre introgression types, backcrosses, and F 1 CanD is50TexAdv 3 ancestry categories: 1) c anonical; 2 ) b ackcrossed to c anonical and backcrossed to Everglades; 3) b ackcrossed to Texas and F 1 Differences in survival between pre introgression intermediate types. F1Adv 2 ancestry cat egories: 1) canonical, backcrossed to c anonical, backcrossed to Everglades, and b ackcrossed to Texas; and 2) F 1 Survival advantage (or disadvantage) for F 1

PAGE 66

66 Table 3 2. Model comparison table to estimate annual kitten survival and select base model. For each model, we present the number of parameters (K), the overdispersion and small sample size ( c ), the model QAIC c weight, and the estimate of annual kitten survival with standard error. The base model selected is in the first row (representing survival differing between kittens, females age 1 and 2, females 3 and older, males age 1 3, and mal es 4 and older; recapture the same for all uncollared panthers; and recovery differing between uncollared kittens and uncollared older panthers). Model K c Weight Annual Kitten Survival Estimate (SE) s&a ) 8 0 .00 0.1 47 0.343 (0.070 ) s&a ) 8 0.19 0.1 34 0.342 (0.070 ) s&a ) 8 0.24 0.13 0 0.345 (0.071 ) s&a ) 8 0.43 0.1 19 0.344 (0.070 ) 7 0.47 0.1 16 0.296 (0.049 ) 7 0.62 0.1 08 0.296 (0.0 4 9 ) 7 0.76 0.10 0 0.297 (0.049 ) 7 0.92 0.09 3 0.297 (0.0 4 9 ) r(k, s&a ) 9 2.02 0. 053 0.349 (0.084) a 0.335 (0.089) b a Female kitten survival. b Male kitten survival.

PAGE 67

67 Table 3 3. M odel selection table for effects of covariates on survival. A) A ncestry effects B) H e terozygosity effects C) A bundance ind ex effects. These analyses were performed on subsets of the data (excluding panthers before 1995 for all analyses and panthers of unknown ancestry for A, and panthers before their heterozygosity wa s sampled for B). For each model, we present Information Criterion adjusted for overdispersion and sample size ( c ), and the weight. See Table 3 1 for descriptions of the covariate models and Table 3 2 for the base model. All models estimate constant recapture rates and recovery rates that differ between kittens and older panthers. A. Ancestry models Surviva l Model K c Weight 1. base + k:AdmAdv + s&a:F1Adv 10 0.00 0.242 2. base + AdmAdv 9 1.34 0.124 3. base + F1Adv 9 1.44 0.118 4. base + CanDisF1Adv 10 1.99 0.090 5. base + k:50TexAdv + s&a:F1Adv 10 2.38 0.074 6. base + k:AdmAdv 9 3.27 0.047 7. bas e + s&a:F1Adv 9 3.35 0.045 8. base + 50TexAdv 9 3.43 0.044 9. base + k:F1Adv + s&a:F1Adv 10 3.46 0.043 10. base + k:IntAdv + s&a:F1Adv 10 3.49 0.042 11. base + k:CanDis50TexAdv + s&a:F1Adv 11 3.98 0.033 12. base + k:CanDisF1Adv + s&a:F1Adv 11 4.22 0.0 29 13. base + CanDis50TexAdv 10 4.92 0.021 14. base + IntAdv 9 5.56 0.015 15. base + k:50TexAdv 9 5.80 0.013 16. base + k:IntAdv 9 7.34 0.006 17. base + k:CanDis50TexAdv 10 7.36 0.006 18. base 8 7.43 0.006 B. Heterozygosity models Survival Model K Weight 1. base + Het 9 0.00 0.325 2. base + s&a:Het 9 0.75 0.223 3. base 8 0.91 0.206 4. base + k:Het 9 1.88 0.127 5. base + k:Het + s&a:Het 10 2.00 0.119 C. Abundance index models Survival Model K Weight 1. base + k:Abundance 9 0.00 0.988 2. base 8 8.79 0.012

PAGE 68

68 Figure 3 1. The effect of genetic variables on model averaged kitten survival estimates. A) Annual survival plotted against ancestry category. B) Annual survival plotted against heterozygosity. Error bars represen t unconditional standard errors.

PAGE 69

69 Figure 3 2. The effect of a panther abundance index on model averaged kitten survival estimates. Error bars represent unconditional standard errors.

PAGE 70

70 Figure 3 3. Model averaged kitten survival by 3 month age inter val within the first year. Error bars represent unconditional standard errors.

PAGE 71

71 CHAPTER 4 DOES GENETIC INTROGRESSION IMPROVE REPRODUCTIVE PERFORMANCE? A TEST ON THE ENDANGERED FLORIDA PANTHER Introduction The small population paradigm of conservation biology postulates that inbreeding (mating between close relatives) is inevitable in small, isolated populations, and that inbred individuals have reduced fitness (Caughley, 1994; Slate et al., 2000) This phenomenon, commonly referred to as inbreeding depression, is though t to be a major challenge in manag ing s uch i mperil ed populations. Inbreeding and loss of genetic diversity have been shown to influence many fitness traits, including age of first reproduction (Charpentier et al., 2008) litter or clutch size (Ortego et al., 2007) survival (Hostetler et al., 2010; Keller et al., 1994) lifetime reproductive success and other measures of individual fitness (Ferreira and Amos, 2006; Slate et al., 2000) and population persistence (Brook et al., 2002b; Ebert et al., 2002) However, outbreeding depression (reduction in fitness due to the matin g of individuals that are genetically too dissimilar) is also a risk when intentional hybridization is undertaken to mitigate the effects of inbreeding depression (Edmands, 2007) The Florida panther ( Puma concolor coryi ), an endangered puma subspecies, is a textbook example of a small and isolated population. Once distributed across the southeastern United S (McBride et al., 2008) The population was thought to be severely inbred and characterized by extremely low genetic diversity (Culver et al., 2000; Johnson et al., 2010; Roelke et a l., 1993b) Many phenotypic traits thought to be characteristics of inbreeding were reported from this population, including very low sperm quality, and high incidences of cryptorchidism, cowlicks, and kinked tails (Barone et al., 1994;

PAGE 72

72 Mansfield and Land, 2002; Roelke et al., 1993b) Due to concerns about inbreeding depression, 8 female Texas pumas ( P. c. stanleyana ) were brought into the Florida panther population in 1995. Since this genetic introgression, the population has increased (McBride et al., 2008) alth ough the processes that have led to this increase continues to be debated (Creel, 2006; Maehr et al., 2006; Pimm et al., 2006b) The influence of genetic introgression on age specific survival of Florida panthers has been explored (Benson et al., in revision; Hostetler e t al., 2010; Pimm et al., 2006b) but the unknown (but see Pimm et al., 2006b fo r a qualitative comparison) Furthermore, the Scientific Review Team (SRT), commissioned by the U.S. Fish and Wildlife Service (USFWS) and the Florida Fish and Wildlife Conservation Commission (FWC) partly in order to make recommendations for continuing Florida panther research, recommended reanalysis of existing data to rigorously estimate reproductive parameters, and to evaluate the potential influence of introgression on demographic parameters of Florida panthers (Beier et al., 2003 ) We used long term field and genetic data (1995 2008) to address the following questions: (1) Do probability of breeding or litter size differ among age classes? (2) Do probability of breeding or litter size vary seasonally? (3) Are probability of bre eding or litter size affected by population size (through negative or positive density dependent effects)? (4) Are probability of breeding or litter size affected by genetic diversity (as measured by heterozygosity)? (5) Are reproductive parameters positi vely influenced by genetic introgression as predicted by the hybrid vigor theory?

PAGE 73

73 Materials and Methods Study A rea Our study area was the entire current breeding range of the Florida panther, and encompassed Big Cypress National Preserve, Everglades Nati onal Park, Florida Panther National Wildlife Refuge, Okaloacoochee Slough Wildlife Management Area and State Forest, Fakahatchee Strand Preserve State Park, Picayune Strand State Forest, Big Cypress Seminole Indian Reservation, and adjoining private lands in south Florida. Although transient males tend to roam widely, no females have been verified north of Caloosahatchee River since 1972 (Land et al., 2008) The climate of south Florida is tropical (Henry et al., 199 4) with a wet season extending roughly from June to November and a dry season from December to May. Field M ethods We located and treed Florida panthers using trained hounds and houndsmen (supplied by Livestock Protection Company, Alpine, Texas) Immobi lization and capture protocols are described in detail elsewhere (FWC, 2008; Land et al., 2008; McCown et al., 1990; Taylor et al., 2002) Age of captured panthers (if unknown) was estimated using a combination of toothwear, pelage characteristics, size, and developmental stage of teats and reproductive organs. Blood and tissue samples were collected for health assessment and genetic anal ysis. We affixed radio collars on captured adults, sub adults, and older kittens. Collars contained mortality switches that altered the pulse rate of transmitters when a panther did not move for a period of time. We located radio collared panthers usin g aerial telemetry, generally three times per week. When mortality signals were detected, the locations were investigated on the ground to confirm the death of the panther.

PAGE 74

74 Successive locations of females were continually assessed to determine the commenc ement of denning behavior; lack of movement between 3 4 fixes was taken as evidence of possible denning (Land et al., 1998) We visited dens 4 35 days post partum (approximate median = 14 days). Kitte ns in dens were counted, sexed, sampled for genetic material, and implanted with passive integrated transponder (PIT) tags. Florida panther gestation length averages at 3 months and litter sizes range from 1 to 4 (Maehr, 1992) Genetic A ncestry and H eterozygosity We estimated the heterozygosity and genetic origins of individuals from genetic samples using techniques detailed elsewhere ( Benson et al., in revision; Hostetler et al., 2010; Johnson et al., 2010) We used results from the genetic origins analysis along with pedigree information and field evidence to assign panthers to 3 groups that reflect the genetic makeup of the southern Florida population (pre introgression type panthers, F1 admixed panthers, and other admixed panthers; Benson et al., in revision) Pre introgression type panthe rs represent the genotypes present on the landscape prior to the introgression program and this group is composed mostly of canonical Florida panthers, which showed no evidence of non Florida genetic admixture (no direct non Florida relatives or < 10% non Florida genetic contribution). Additionally, 2 pre introgression type panthers in the sample of individuals used to investigate the influence of genetic ancestry on reproductive rates ( n = 58) were descendents of panthers from the Everglades genetic popul ation with documented genetic links to Central America (Culver et al., 2000; O'Brien et al., 1990) Admixed panthers were mostly the descendents of the introduced female Texas pumas released in 1995 (Benson et al., in revision; Johnson et al., 2010)

PAGE 75

75 Differences in fitness between classes of admixed panthers are possible and hybrid vigor is predicted to be strongest in the F 1 generation (Arnold and Hodges, 1995; Burke and Arnold, 2001) so we included 2 classes of admixed panthers in our analyses: F 1 and other admixed. We defined F 1 admixed panthers as any offspring produced by matings between a Texas female and a pre introgression type male. In summary, all individuals in our radio collared sample were placed into 1 of 3 categories: pre introgression type panthers ( n = 24), F 1 admixed panthers ( n = 6), and other admixed panthers ( n = 28). We were also able to further classify most individuals from the other admixed category into 3 f iner categories: backcrossed to Florida admixed panthers ( n = 6), backcrossed to Everglades admixed panthers ( n = 3), and backcrossed to Texas admixed panthers ( n = 11). We considered potential differences in reproductive rates between pre introgression t ype, F 1 admixed, and these finer admixture distinctions (results available in Appendix I ). Index of A bundance To examine the potential effects of panther density on reproductive rates, we used the minimum population counts as an index of abundance (McBride et al., 2008) These minimum counts were based on radio tracking and field evidence of sub adult and adult panthers (tracks, scat, kills, and scrapes) collected for calendar years. In contrast, we analyzed reproductive rates based on years from June May. We applied the calendar year count for year t to the June t May t + 1 year. In 2004, the panther survey was incomplete, so the minimum count is unknown. We followed McBride et al. (2008) and extrapolated it halfway between the 2003 and 2005 counts.

PAGE 76

76 Estimating Age C lasses We knew the ages (to within a few days) for panthers that were handled as kittens at natal dens and then subsequentl y recaptured and radio collared ( n = 29, 46.8% of total). These kittens were handled <35 days after birth as estimated by the date when the mother was first located at the den site and by the physical characteristics of the kittens. Age of panthers not m arked as kittens was estimated in the field at time of first capture using tooth wear and pelage characteristics. We separated female panthers into 4 age classes: kittens (0 1 years old), sub adults (1 2.5 years old), prime adults (2.5 10 years old), and o lder limited analyses to sub adults, prime adults, and older adults. The sub adult age class approximate ag es when females and males first establish home ranges. We estimated age of independence at just over 1 year (mean = 397 days, SD = 7 4 days, n = 3 2 ; FWC and NPS, unpublished data) from known age kittens that were radio tracked simultaneously with their mot hers (before and after independence). We standardized the beginning of the sub adult age class to 1 year for simplicity. Modeling P robability of B reeding We used radio tracking and reproductive histories of female panthers monitor ed between June 1995 and May 2008 (13 years) in the present analyses. We organized the data into 6 month intervals (June November and December May). W e recorded the following information for each panther in each 6 month interval: number of months tracked, number of kittens produced, number of kittens produced that survived to be PIT tagged in the den, age class, heterozygosity, and ancestry category. If a female

PAGE 77

77 advanced to an older age class within a 6 month interval, two records were created, each with one of the age clas ses and the appropriate number of months tracked. We modeled probability of a female panther giving birth at least once during a season ( p ) using binomial regression. Since no female panther was recorded giving birth twice within 1 season, p can also b e interpreted as the probability that a panther gives birth once during a season. The parameter p can be modeled as a function of covariates. We modeled p using a complementary log log link: ( 4 1 ) where Z i is the matrix of covariates for panther i and is the vector of binomial Often, panthers were not tracked for entire seasons (e.g., due to radio failure before the end of a season or radio collaring after the start of a season). To account for the effect of the number of months a panther was tracked in a season on the probability that it gives bir th while being tracked, we used log( m ) as an offset, where m is the number of months a panther was tracked during a season. The combination of this offset and the complementary log log link for p were chosen to generate the property: ( 4 2 ) where b is an binary variable indicating whether a panther bred or not in an interval. In other words, the probability that a panther does not breed in m months is the same as the probability that a panther does not breed in 1 month, taken to the m power. We used the glm function in the stats library in R (version 2.11.1; R Development Core Team, 2010) to fit the model.

PAGE 78

78 Modeling L itter S ize We estimated and m odeled litter sizes of female panthers, conditional on their giving birth, using cumulative logit regression. Although this model is traditionally used for ordered categorical data (Agresti, 2007) it can also be used for count data (Min and Agresti, 2005) The model includes J categories for number of offspring, with N j being the number of offspring represented by category j (because panthers can have 1 to 4 offspring, J = 4 and N j = j in our case). T he probability that the litter size will be at most N j for individual i ( Y i ) is represented by ij ( 4 3 ) where j is the intercept for litter size being at most N j X i is the matrix of covariates for panther i and In E quation 4 3, has a subscript for j but does not. This model therefore represents the underlying structure of the data i n a flexible fashion but as sumes the effects of X are identical for all J 1 cumulative logits. When >0, the covariate positively influences the probability of a larger litter size. The probability th at Y i will be exactly N j is ( 4 4 ) The expected average litter size for individual i is ( 4 5 ) We implemented the cumulative logit model using the clm function in the ordinal library in R (version 2.10.1).

PAGE 79

79 Modeling A ppro ach We used an information for model selection and statistical inference (Burnham and Anderson, 2002) We i difference between A IC value of the i th model and the top ranked model) and Akaike weights ( w i the weight of evidence that the i th model is actually the best model of the models being considered given the data) as in Burnham and Anderson (2002). Generally, we considered mod els w 4 to be plausible models support. In addition, the sum of the weights of models including a given v ariable can be interpreted as a measure of its importance, relative to other variables (Burnham and Anderson 2002). We calculated model averaged estimates of reproductive rates and unconditional variances for each level of categorical variables and across continuous covariates (Burnham and Anderson 2002). To generate these model averaged estimates we used all models included in the model selection process for a given analysis, weighted by their Akaike weights. Models with no effects of the covariates bei ng presented were included in the averages (as having the same reproductive parameter values for all values of the covariate), the model averages therefore represent unconditional estimates of reproductive parameters (Burnham & Anderson 2002). We used a step wise approach to test for the effects of various covariates on reproductive parameters. First, we examined the effects of age class on reproductive parameters (i.e., breeding probability and litter size). We considered 3 age structures: Age1, whi ch s eparates female panthers in to 3 age classes (sub adults, prime adults,

PAGE 80

80 and older adults); Age2, which separates females into 2 age classes: older adult and young adult (sub adults and prime adults combined) panthers; and A ge3, which also divides female pan thers into 2 age classes: sub adult and adult (prime adults and older adults combined) age classes (note that these model labels differ from those used in Chapter 2 ) We tested for the effect of all 3 age structures on p but did not test for the effect o f Age1 or Age2 on because those separate out older adults and sample size was not adequate for a separate analysis of litters produced by older adult panthers. We chose the minimum AIC model from each age class analysis as a base model to test for the additive effect of season (wet: June November vs. dry: December May) on p and The most parsimonious models from these analyses were used as base models for all subsequent analyses. We next tested for the effects of age class and season on reproducti ve rates, excluding older adults. We did this to check the robustness of our estimates of reproductive parameters for sub adults and prime adults. Excluding the older adults had very little effect on estimates (Appendix J ). We used the minimum AIC model s from the season al analyses as base models, and tested for linear and quadratic effect of abundance index on reproductive parameters; the quadratic term was used to test for the possibility of both negative and positive density dependence in components of reproduction. Effect of abundance on both p and were tested additively with the base models. Next, we tested for the additive effect of ancestry on p and We excluded 1 panther with unknown ancestry from these analyses. We tested three models of ancestry: Ancest1, which divides females into 3 ancestry categories: pre introgression

PAGE 81

81 type, other admixed, and F1 admixed; Ancest2, which combines the other admixed with the pre introgression type females; and Ancest3, which combines the other admixed wit h the F1 admixed females. There were no older adult female admixed panthers in our dataset, so we wanted to check the robustness of our ancestry statistical inference and estimates to the inclusion of the pre introgression type older adults. Excluding th e older adults had very little effect on estimates or inference (Appendix J ). Finally, we tested for the effects of heterozygosity on reproductive parameters. We excluded 1 panther with unknown heterozygosity from these analyses. We tested for the addit ive effect of heterozygosity on both p and Appendix K contains results of analyses performed on an annual time scale, rather than 6 month, for comparison with other studies and for use in population models. For these models, kittens were only counted (included as breeding events and in litter c ounts) if they received PIT tags in the den, and estimates of temporal variance in reproductive parameters were generated (Appendix K ). Results We tracked 61 female panthers for a total of 2,414 panther months, and recorded 94 litters. Estimated ages of d ams when they produced litters ranged from 21 months to 11.45 years. There was only one recorded litter produced by 9 older years) females, which were tracked for a total of 313 months. Mean litter size was 2.60 and the modal litter size was 3 ( Figure 4 1). A nave estimate of probability of breeding during a 6 month period for all female panthers (number of litters/number of months tracked 6) was 0.234. These correspond closely to the model based estimates for all female panthers (mean 1SE litter size = 2.60 0.09, probability of breeding = 0.232 0.021).

PAGE 82

82 Probability of B reeding There was strong evidence that breeding probability varied among age classes; Table 4 1A). Older adults had a lower probability of breeding than younger females, but there was no evidence that breeding probabili ties differed between sub adult and prime adult females ( Figure 4 2A). Furthermore, there was evidence that probability of breeding varied seasonally (Model 1, Table 4 1B) with m odel av eraged probability of breeding being higher during the dry season ( Fi gure 4 2B ). Index of abundance positively influenced breeding probability (Model 1, Table 4 1C ; Figure 4 2C ). The most parsimonious model evaluating the effect of ancestry on reproductive parameters revealed that probability of breeding differed betwe en F 1 panthers and those with other ancestry (Model 1, Table 4 1D). Model averaged probability of breeding was lower for F 1 introgression type and marginally higher for other admixed than pre introgression type ( Figure 4 3 A). The most parsimon ious heterozygosity model suggested that heterozygosity had no effect on probability of breeding (Model 1, Table 4 1E). Model averaged probability of breeding declined somewhat with increasing heterozygosity ( Figure 4 3B ). Litter S ize Sub adult females produced somewhat larger litters than females of other age classes; there was evidence for age specific variation in litter size ( Table 4 2A; Figure 4 2 D ). Older adult females produced only 1 litter during this study, and we were unable to test for differe nces in litter size between older adults and younger females. There was some evidence for seasonal variation in litter size (Model 1, Table 4 2B) w ith larger

PAGE 83

83 litters during the dry season ( Figure 4 2E ). However, t here was no evidence that litter size wa s affected by abundance (Model 1, Table 4 2C ; Figure 4 2F ). The most parsimonious model evaluating the effect of ancestry showed no effect on litter size (Model 1, Table 4 2D) ; admixed females produced slightly larger litters than pre introgression type females ( Figure 4 3C ). The most parsimonious heterozygosity model suggested that heterozygosity had no effect on litter size (Model 1, Table 4 2E ; Figure 4 3F ). Discussion The case of the Florida panther demonstrates how loss and degradation of habitat, p ersecution, rapid urbanization and other anthropogenic influences can threaten biodiversity. Indeed, the plight of the panther has attracted the attention of ecologists, conservation organizati ons, state and Federal agencies, and various interest groups (http://www.florid apanthernet.org/, Fergus, 1998; Gross, 2005) Various conservation measures have been undertaken to ensure long term persistence of this population; perhaps the most drastic among these was the genetic project implemented in 1995 (Seal, 1994) Although the prevalence of phenotypic traits indicative of inbreeding (e.g., heart defects, susceptibility to heavy parasite loads, poor sperm quality and cryptorchidism) have declined (Man sfield and Land, 2002; Onorato et al., 2010) and the panther population has increased (McBride et al., 2008) the role of the introgression in the panther population increase continues to be debated (Gross, 2005; Maehr and Caddick, 1995; Maehr et al., 2006; Pimm et al., 2006b) Also, some aspects of the reproductive ecology of panthers remain poorly understood, especially the e ffect of genetic introgression.

PAGE 84

84 The overall 6 month probability of b reeding was 0.232 0.021, which translates into annual cumulative probability of breeding of 0.410 0.034 (Appendix K ). This annual estimate is within the range reported for western North America puma populations (0.24 0.45; Cooley et al., 2009; Stoner et al., 2006) However, our annual e stimate of breeding probability was lower than that used by previous demographic analyses of Florida panthers (0.5; Maehr et al., 2002a; Root, 2 004) Breeding probability probability of breeding compared to younger females but the breeding probability of sub adult females (1 2.5 years) did not differ from that of pri me adults (2.5 10 years). We expected probability of breeding to be higher for prime adults than for the other two age classes because they should have a defined home range that allows them to maximize their energetic allocations towards reproduction. Our results provide strong evidence of reproductive senescence (manifested as a lower probability of breeding) in older Florida panthers, but suggest that age specific differences in reproduction prior to senescence are minor. The overall average litter si ze was 2.60 0.09; this estimate, although slightly higher than previous estimates for Florida panthers (Lotz et al., 2005: 2.4; Maehr and Caddick, 1995: 2.25) is within the range of litter size estimates reported for western populations of pumas (1.7 3.0; Cooley et al., 2009; La mbert et al., 2006; Laundr et al., 2007; Logan and Sweanor, 2001; Robinson et al., 2008; Stoner et al., 2006) Sub adult females produced somewhat larger litters. This result was to some extent surprising because we expected prime aged females to produc e larger litters than sub adult or older females. Effects of age and/or breeding experience on litter size have previously

PAGE 85

85 been shown for other large carnivores (Garrison et al., 2007; Zedrosser et al., 2009) although not for pumas (Logan and Sweanor, 2001) P robability of breeding was higher in the dry season than in the wet, but litter sizes were similar. Previous examinations of raw number of litters per month suggest that the actual birth peak is March June (Lotz et al., 2005) which is supported by our results. Dry season coincides with the peak white tailed deer ( Odocoileus virginianus ) fawning season in south Florida (Fleming, 1994; February March; Richter and Labisky, 1985; Schortemeyer et al., 1991) and peak feral hog ( Sus scrofa ) farrowing (parturition) season (Belden and Frankenberger, 1990) which may allow females to make caloric gains necessary to achieve a successful pregnancy by preying on vulnerable fawns and piglets during their first months of life. We found that abundance index positively influenced breeding probability. This may be due to a mate finding Allee effect (Gascoigne et al., 2009) Howe ver, there was also a decrease in kitten survival observed with increasing abundance index possibly due to infanticide by male panthers (Hostetler et al., 2010) Panther females are unlike ly to copulate while they have dependent offspring, and, in fact, a female panther that the first litter failed (Hostetler et al., 2010) Therefore, one possibility is that the increase in probability of breeding is caused by the decrease in kitten survival. Abundance index also increased fairly steadily over time during our stu dy period (1995 2008; McBride et al., 2008) It is possible that there was an increase in probability of breeding over time that was due to something besides or in addition to, population size. One intriguing (although speculative) possibility is that reproductive vigor of the male

PAGE 86

86 panthers has increased since 1995 possibly due to the reduction in inbreeding depression in male reproductive traits (Johnson et al., 2010) In earlier studies, we found that F 1 admixed panthers of all ages survived substantially better than panthers in ancestry categories (Benson et al., in revision; Hostetler et al., 2010) Based on these results and predictions of hybrid vigor theory (Crow, 1948) we expected higher probability of breeding (and litter siz e) for F 1 female panthers Thus, our results that breeding probabilities were lower for F 1 (and backcrossed to Texas; Appendix I ) female panthers than for other admixed and pre introgression type female panthers were unexpected This may be a sign of out breeding depression in female reproduction, due to loss of local adaptations and/or co adapted gene sequences (Edmands, 2007; Hallerman, 2003) Although outbreeding depression cannot be ruled out as a possib ility the relative lack of genetic differentiation within North American pumas (Culver et al., 2000) and the former adjacency and presumed interbreeding between P. c. coryi and P. c. stanleyana make outbre eding depression less likely An alternative, and perha ps a more plausible, explanation is that the F 1 panthers were breeding less often because their offspring were more likely to survive to independence (Hostetler et al., 2010) Finally, the s ample of radio collared F 1 female Florida panthers was small ( n = 6), and it is possible that these individuals bred less often than other pa nthers due to chance alone (or for reasons other than genetic ancestry). Thus, although our results provide strong evidence that that genetic introgression did not positively influence probability of breeding for Florida panthers, we cannot conclude that parameters.

PAGE 87

87 Interestingly, litter size did not vary among females with various ancestry categories. Additionally, we found no evidence that heterozygosity influenced breeding probability or litter si ze. These results were surprising because heterozygosity has been shown to influence litter or clutch size (Ortego et al., 2007) lifetime reproductive success and other measures of individual fitness (Ferreira and Amos, 2006; Slate et al., 2000) and also because heterozygosity positively influenced age specific survival probabilities of Florida panthers (Benson et al., in revision; Hostetler et al., 2010) Our study extends the work of Pim m et al. (2006b) who graphically examined the lifetime kitten production of 24 female panthers and found no evidence of a difference between pre introgression type and admixed panthers. W e used a larger sample size more detailed ancestry categories and separated two components of reproduction (breeding probability and litter size) using robust statistical methods. These results clearly demonstrate that one intrinsic or extrinsic environmental factor can different ially influence demographic rates or fitness components. For example, prior to this study we had concluded that genetic introgression positively influences demographic traits, whereas population density can negatively influence them (Benson et al., in revision; Hostetler et al., 2010) ; neither of these conclusions is supported by the present study. Th ese contrasts highlight the importance of simultaneously examining the effect of management actions on all componen ts of fitness

PAGE 88

88 Table 4 1. Model selection table for probability of breeding ( p ). A) A ge class effects. B) Season effects. C) Abundance index effects. D) Ancestry effects. E) H eterozygosity effects The top ranked model for age class (part A, model 1, in bold) was used as a base model to test for the effects of season; the top ranked model for season (part B, model 1, in bold) was used as a base model to test for the effects of abundance, ancestry, and heterozygosity. For each model, we present the Information Criterion ( w i ) The ancestry and heterozygosity analyses were performed on a subset of the data (excluding 1 panther with unknown heterozygosity). Model p K IC w i A) Age Models 1 Age2 a 2 0.00 0.562 2 Age1 b 3 0.50 0.438 3 Constant c 1 19.94 0.000 4 Age3 d 2 21.58 0.000 B) Season Models 1 Age2 + Season 3 0.00 0.743 2 Age2 2 2.12 0.257 C) Abundance Index Models 1 Age2 + Season + Abundan ce 4 0.00 0.433 2 Age2 + Season + Abundance + Abundance 2 5 0.78 0.293 3 Age2 + Season 3 0.91 0.274 D) Ancestry Models 1 Age2 + Season + Ancest2 e 4 0.00 0.627 2 Age2 + Season + Ancest1 f 5 1.58 0.284 3 Age2 + Season 3 4.54 0.065 4 Age2 + Season + Ancest3 g 4 6.48 0.025 E) Heterozygosity Models 1 Age2 + Season 3 0.00 0.610 2 Age2 + Season + Het h 4 0.89 0.390 a Age2 divides panthers into young adults and older adults. b Age1 divides panthers into sub adults, prime adults, and older adults. c No predictor variables. d Age3 divides panthers into sub adults and adults. e Ancest2 divides panthers into 2 ancestry categories: F 1 admixed, and other admixed and pre introgression type combined. f Ancest1 divides panthers into 3 ancestry categories: p re introgression type, other admixed, and F 1 admixed. g Ancest3 divides panthers into 2 ancestry categories: pre introgression type and admixed. Heterozygosity.

PAGE 89

89 Table 4 2. Model selection table for litter size ( ). A) A ge class effects. B) Season effects. C) Abundance index effects. D) Ancestry effects. E) H eterozygosity effects The top ranked model for age class (part A, model 1, in bold) was used as a base model to test for the effects of season; the top ranked model for season (part B, model 1, in bold) was used as a base model to test for the effects of abundance, ancestry, and heterozygosity. For each model, we Criterion ( C), and the Akaike weight ( w i ) The ancestry and heterozygosity analyses were performed on a subset of the data (excluding 1 panther with unknown heterozygosity). Model K w i A) Age Models 1 Age3 a 4 0.00 0.794 2 Constant b 3 2.70 0.206 B) Season Models 1 Age3 + Season 5 0.00 0.551 2 Age3 4 0.41 0.449 C) Abundance Index Models 1 Age3 + Season 5 0.00 0.619 2 Age3 + Season + Abundance 6 1.74 0.260 3 Age3 + Season + Abundance + Abundance 2 7 3.26 0.121 D) Ancestry Models 1 Age3 + Season 5 0.00 0.443 2 Age3 + Season + Ancest3 c 6 0.87 0.286 3 Age3 + Season + Ancest2 d 6 1.98 0.165 4 Age3 + Season + Ancest1 e 7 2.86 0.106 E) Heterozygosity Models 1 Age3 + Season 5 0.00 0.729 2 Age3 + Season + Het f 6 1.98 0.271 a Age 3 divides panthers into sub adults and adults. b No predictor variables. c Ancest3 divides panthers into 2 ancestry categories: pre introgression type and admixed. d Ancest2 divides panthers into 2 ancestry categories: F 1 admixed, and other admixed and pre introgression type combined. e Ancest1 divides panthers into 3 ancestry categories: pre introgression type, other admixed, and F 1 admixed. f Heterozygosity.

PAGE 90

90 Figure 4 1. Frequency distribution of litter sizes produced by radio collared female panthers (1995 2008). Numbers inside bars indicate number of litters of that size.

PAGE 91

91 Figure 4 2. Model averaged reproductive rates ( unconditional standard errors) by age class season, and abundance index. A) Six month probability of breeding by age cl ass. B) Same by season, for young adult panthers (<10 years old) C) Same by abundance index, for young adult panthers in the dry season D) Average litter size by age class. E) Same by season, for F) Same by abundance index, f o r adult panthers in the dry season Sub adults are defined as ages 1 2.5, prime adults ages 2.5 10, and older adults ages 10 and older.

PAGE 92

92 Figure 4 3. Model averaged reproductive r ates ( unconditional standard errors) by ancestry category and heterozygosity A) S ix month probability of breeding for young adult panthers (<10 years old) in the dry season by ancestry. B) Same by heterozygosity. C) A verage litter size, for adult pan s old) in the dry season, by ancestry D) Same by heterozygosity.

PAGE 93

93 CHAPTER 5 THE EFFECT OF INTROGRESSION ON THE DYNAMICS AND PERSISTENCE OF THE FLORIDA PANTHER POPULATION Introduction Population viability analyses (PVA) are a widely used set of tools in conservation ecology, but have been critiqued for often ignoring unce rtainty in their inputs, outputs, and model structure (Ellner et al., 2002; Reed et al., 2002) Recent studies have highlighted the need to incorporate parametric uncertainty as well as model selection uncertainty to quantify uncertainty in PVA predictions (Bakker et al., 2009; Ellner and Fieberg, 2003; White, 2000) Despite these uncertainties, PVAs can help understand the dynamics and persistence of biological populations, especially when adequate data are available to paramet erize biologically meaningful models (Brook et al., 2002a; Brook et al., 2000; Lindenmayer et al., 2003) Population projection models in general and PVA in particular tend to be forward looking. However, there can be advantages to using them to look backwards. These advantages include model validation (Ball et al., 2003; Brook et al., 2000; Lindenmayer et al., 2003) and evaluating the impacts of past manage ment actions (Kauffman et al., 2003) Ask can be at least as bedeviled by known and unknown biases and uncertainties as trying to forecast the future. However, if we acknowledge these biases and uncertainties, both processes can be valuable: evaluating the effects of past management actions on population dynamics and modeling potential future population changes and risks are critical tools for choosing current and future management actions. The Florida panther ( Puma concolor coryi ) i s an endangered large carnivore found in a single small population in south Florida (McBride et al. 2008). Monitoring, research

PAGE 94

94 and management actions implemented to prevent the extinction of the panther have been extensive (Beier et al. 2003). Perhaps t he most important (and controversial) management action has been the release of 8 female Texas pumas ( P. c. stanleyana ) into south Florida in 1995. This genetic introgression was undertaken to reduce apparent inbreeding depression and mimic historic gene flow between these subspecies. Since 1995, the number of panthers has grown dramatically (McBride et al. 2008). Previous analyses have shown that admixed panthers resulting from genetic introgression have higher survival than canonical or pre introgressi on type panthers (Benson et al., in revision; Hostetler et al., 2010) Additionally, admixed panthers have a lower breeding probability than pre introgression type panthers; litter sizes were not different between these groups (Chapter 4). How this combination of demographic responses to genetic introgression contributed to the observed population growth remains unanswered. Concern about the future of the panther population has lead to the development of several p opulation viability analyses (PVA) since the 1980s (Maehr et al. 2002, Root 2004). The Scientific Review Team (SRT), commissioned by the Florida Fish and Wildlife Conservation Commission and US Fish and Wildlife Service to review the scientific literature on the panther and to make recommendations for future research reviewed results of these PVA and other demographic studies. The SRT noted that previous PVA were based on unreliable estimates of demographic parameters, required too many arbitrary assumpti ons, performed insufficient sensitivity analysis, and also value s and model structure (Beier et al., 2003; Beier et al., 2006; Gross, 2005) The team recommended rigorous

PAGE 95

95 analyses of available data to estimate demographic paramete rs, and to perform demographic analyses using those empirically estimated parameters. Now that statistically rigorous estimates of demographic parameters for the Florida panther are available (Chapters 2 4), the time is appropriate for revisiting the dynam ics and persistence of the panther population. In accordance with SRT recommendations, our objectives for this study were to: 1) estimate deterministic and stochastic population growth rates for Florida panthers, 2) perform sensitivity analyse s to explore the sensitivity and elasticity of deterministic and stochastic population growth rate s to model parameters 3) estimate probabilities of extinction and distribution of extinction time under various stochastic models, 4 ) estimate deterministic population g rowth rate probability of extinction, and effects of vital rates on the population growth rate for a hypothetical population of pre introgression type panthe rs, and 5) to adequately address the uncertainty in all of the model inputs and outputs Methods F ield and Genetic Methods Florida panthers were treed, captured, radio collared, and tracked starting in 1981 by biologists from the Florida Fish and Wildlife Conservation Commission (FWC) and National Park Service (NPS) using methods described in Chapter 2, by Belden et al. (1988) and by Land et al. (2008) Litters of Florida panther kittens were detected, counted, PIT tagged, recaptured, and recovered starting in 1995 as described in Chapters 3 and 4 and by Land et al. (1998) We divide d female panthers into four age cate gories: kittens (age 0 1 years), sub adults (age 1 2.5 years), prime adults (age 2.5 10 years), and older adults (age

PAGE 96

96 adult to prime adult occurred at age 3.5 years. Age of capt ured panthers was known or estimated from time since previous capture, tooth wear, and pelage characteristics (Chapter 2; Ashman and Greer, 1976) and error in parameter estimates due to age estimation error was found (for sub adult and adult survival) or suspected (for other parameters) to be small (Chapter 2). Genetic sample s were collected and analyzed as described in Johnson et al. (2010) For the purposes of estimating parameters for canonical panthers (see below), w e followed Benson et al. (in revision) and Chapter 4 and assigned panthers to three ancestry cat egories for the analysis of sub adult and adult survival and reproductive parameters (pre introgression type, other admixed, and F 1 admixed). We followed Hostetler et al. (2010) and assigned panthers to five ancestry categories for the analysis of kitten survival (canonical, backcrossed to canonical, backcrossed to Everglades, backcrossed to Texas, and F 1 admixed). Since the pre introgression type panther s in these analyses were mostly canonical individuals (with a few Everglades individuals; Johnson et al., 2010) the pre introgression type and canonical categories were treated as equivalent (and hereafter referred to as canonical). Model Structure We modeled population dynamics of the Florida panther using age structured, birth flow matrix models focusing on the female segment of the population (Caswell 2001). We used birth flow approximation s to age specific surviva l and fertility rates (Caswell, 2001) because panthers can give birth in any month of the year (Lotz et al. 2005, Chapter 4 of this dissertation). We assume d that female panthers can live and reproduce until age 18.5 which corresponds to t he age of the oldest radio colla red

PAGE 97

97 female in our study ( 18.6 years) Therefore, the population projection matrix has dimensions 19x19. Each annual population projection matrix was of the form : ( 5 1 ) w here F and P indicate survival and fertility rates, respectively; subscripts indicate age classes and t stands for time specificity ( the superscript does no t apply to deterministic demographic analyses; Table 5 1) In a birth flow population, individuals within an age class can have a range of ages. We approximated age specific survival rate as : ( 5 2 ) where S i t is the survival rate from age i 1 to age i between time steps t and t + 1 (equivalent to equation 2.25 in Caswell 200 1). Because the boundary between the sub adult and prime adult age categories was at age 2.5 (and there was no estimated survival rate between ages 2 and 3), we simplified the equations for two of the age specific survival rates : ( 5 3 ) where S sa t is the survival rate of sub adults between time steps t and t + 1 and S pa t is t he survival rate of prime adults between time steps t and t + 1 ( Table 5 1) We approximated age specific fertility rate as: ( 5 4 )

PAGE 98

98 where S k t is the survival rate of kittens between time steps t and t + 1 and m i t is the fecundity (expected number of female offspring produced per surviving female of age class i ) for individual s of age i during time step t (Caswell 2001). Similar to age specific survival, we simplified the equations for two of the age specific fertility rates : ( 5 5 ) where m sa t and m pa t are the age specific fecundity rates for sub adult and prime adult female panther s during time step t The age specific fecundity rate was estim ated as a function of 3 underlying parameters as: ( 5 6 ) where q i t is th e probability of breeding for a female in age class i during time step t i t is the cumulative number of kittens produced by a panther that does breed in age class i during time step t (cumulative because panthers can give birth more than once in a year) and f k is the proportion of kittens that are female ( Table 5 1) Using logistic regression, we found no evidence that f k d eviated from 0.5 (observed proportion = 0.46, p = 0.22) ; thus, we used 0.5 for f k in all cases. Parameter Values We estimated survi val probabilities of sub adult, prime adult, and older adult females ( S sa S pa and S oa ) using Cox proportional hazard model s as described in Benson et al. (in revision) We ran models evaluating the effects of sex, age class, and abundance on s urvival. Survival (Burnham and Anderson, 2002) and estimates of S sa S pa S oa (at abundance 0), and effects of abundance on log hazard based on each of these models are presented in Table 5 2

PAGE 99

99 We estimated kitten survival probability ( S k ) using mark recapture recovery modeling techniques (Burnham, 1993) described in Hostetler et al. (2010) We ran models with different recover y ( r ) probabilities for kittens and older panthers, an d evaluat ed effect s of sex and abundance on kitten survival, using model presented in Hostetler et al. (2010) Survival c < 10 (compared to the top model and/or the top density independent model) Akaike weights, and estimates of S k (at abundance 0) and the effect of abundance on logit kitten survival based on each of these models are presented in Table 5 3 We estimated annual breeding probabilities (probabilities of successful reproduction) of sub adult, prime adult, and older adult females ( q sa q pa and q oa ) using complementary log log regression as described in Chapter 4 We ran models evaluating th e effects of age class and abundance in all combinations on probability of breeding Probability of breeding estimates of q sa q pa q oa and the effects of abundance and abundance 2 on complementary log log monthly probability of breeding based on each of these models are presented in Table 5 4 We estimated the number of kittens produced by sub adult and adult females ( sa and a respectively ) using the cumulative logit methods described in Chapter 4. We ran models evaluating the effects of age class of the dam and panther abundance on annual kitten production. M ode estimates of sa and a and the effect of abundance on cumulative logit number of kittens based on each of these models are presented in Table 5 5

PAGE 100

100 Deterministic Demographic Analyses The asymptoti c deterministic annual population growth rate ( ) was estimated as the dominant eigenvalue of A (Caswell 2001). Sensitivity and elasticity of to lower level parameters ( Table 5 1 ) were calculated following Caswell (2001). Incorporating Parameter Uncerta inty We incorporated parameter or sampling uncertainty into model outputs via a parametric bootstrap method (Efron and Tibshirani, 1998) We estimated a v ariance covariance matrix for each set of parameters : 1) kitten survival (on logit scale), 2) sub adult and adult survival (baseline survival converted to logit scale and log hazard effect sizes), 3) probability of breeding (on complementary log log scale) and 4) number of kittens produced in the year from female panthers that did breed (on cumulative logit scale). W e assumed no overall covariance between the four categories but used the estima ted covariance within categories. We ran multiple parametric bootstraps (50,000 for deterministic analyses), selecting random multivariate normal values using the parameter means (on the appropriate scales) and variance covariance matri ces and then converting the results to the real scale, for each bootstrap run. In this way, we selected random values from the empirical distributions of the demographic parameters. results obtained from the point estimates of parameters as the parameter valu es. I f the mean and/or median of the bootstrap outputs differ substantially from the deterministic output based on point estimates, we present those a s well. We also report the 2.5 and 97.5 percentiles of bootstrap outputs as a measure of uncertainty (these should not be inter preted as 95% confidence intervals when the point estimate differs substantially from the median; Efron and Tibshirani, 1998)

PAGE 101

101 Incorporating Model Selection Uncertainty Another source of statistical uncertainty is model selection uncertainty incertit ude about whether the minimum AIC (or QAIC c ) model for a parameter is actually the best model in the model set (Burnham and Anderson, 2002) We incorporated this form of uncertainty by randomly selecting statistical models for each s et of parameters, based on their AIC (or QAIC c ) weights (Bakker et al., 2009) For deterministic and stochastic demographic analyses, we restricted model sets to density independent mo dels only (Tables 2 5 ). In all cases, we di c ) > 10. When one or more models were excluded, model weights were recalculated to sum to 1. For each analysis for which model selection uncertainty was incorporated (we ran deterministic demographic analyses with and without model uncertainty, and did not include model uncertainty in stochastic demographic analyses) parameter uncertainty was also incorporated by selecting parameter values from the ir distribution s (as discussed in Incorporating Parameter Uncertaint y) for t he model selected, and then elasticities). This process was repeated multiple times (50,000 for the deterministic demographic analyses). As there is no single point estimate f multiple models, the means of these outputs were reported (if the median differed substantially from the mean, both were repor ted). Again, we report the 2.5 and 97.5 percentiles (Efron and Tibshirani, 1998) Incorporating Environmental Stochasticity Our estimate of temporal variance (random effect of year) for was essentially 0 (chapter 4); therefore, we only consider e d enviro nmental stochasticity in sub adult and

PAGE 102

102 adult survival, kitten survival, and probability of breeding, where there was evidence for environmental stochasticity (Benson et al., in revision; Hostetler et al., 2010; chapter 4) Because factors such as abundance can affect estimates of temporal variance, we estimated temporal variance in those probabilities for each statistical model, and present those estimates in Tables 2 4. For all stochas tic analyses, we used the temporal variances from the selected models. However, we assumed that the fixed effect model weights and the fixed effect estimates and covariances were unchanged by the inclusion of random effects. W e assumed that temporal corr elation between S and q was 0 because we know of no reasonable method for estimat ing this correlation Our method of estimating a single temporal variance for S sa S pa and S oa implicitly assumes that the temporal correlation between those rates is 1 (as did the estimate of a single temporal variance for q sa q pa and q oa ). To estimate the temporal correlation between S k and other survival parameters, we used the Markov Chain Monte Carlo (MCMC) option in MARK (as we did for estimating temporal variance in kitten survi val; Lukacs et al., 2009; White et al., 2009) Because the kitten survival model also estimates sub adult and adult survival for both sexes (Hostetler et al., 2010; it does not distinguish between prime adult and older adult survival) the MCMC option can be used with it to estimate temporal variance in logit sub adult and adult survival s and correlation of th ese survivals with logit kitten survival. We used the same non informative priors as White et al. (2009); estimates from the fixed effects models were used as starting values where appropriate. We sampled the MCMC for 50,000 iterat ions, after 4,000 tuning samples and 1,000 burn in samples.

PAGE 103

103 Estimates of the temporal correlation between S k and other survival parameters were consistently around 0.341 in models where neither kitten survival nor sub adult and adult survival depended on abundance. In models where kitten survival and/or sub adult and adult survival depended on abundance, the estimates of correlation were close to 0. Therefore we assumed that the correlation between logit S k and logit of other survival parameters was 0. 341 in model sets where all survival parameters were density independent and 0 otherwise. Although temporal variance in S sa S pa and S oa was estimated and applied on the log hazard rather than the logit scale (Benson et al., in revision; Therneau and Grambsch, 2000) we found that simpl y reversing the sign (to 0.341 ) for correlation between logit kitten survival and sub adult and adult log hazard generated the correct correlation between logit survivals. When incorporating environmental stochasticity into a simulation, we selected th ree random numbers from multivariate normal distributions each time step, using means of 0 and the temporal variances (and correlations) described above. We added the first random number to the intercept for kitten survival (on the logit scale) and the th ird to the intercept for probability of breeding (on the complementary log log scale). These could then be converted to S k q sa q pa and q oa (on the probability scale). The second random number ( 2 ) represent ed temporal change in log hazard for sub adults and adults, and was applied to S sa S pa and S oa as : ( 5 7 ) In this way, we generated random values for survival and breeding probabilities based on the estimated temporal variances in the same ways in which they were estimated.

PAGE 104

104 Estimating Stochastic Growth Paramete rs We estimated the stochastic version of instantaneous growth rate (log( s )) via simulation as the arithmetic mean of the log ratio of population sizes in successive years (Caswell, 2001; Morris and Doak, 2002) We exponentiated this to obtain estimates of finite stochastic growth rate ( s ). We applied parametric uncertainty but limit ed inference to the minimum AIC (and QAIC c ) density independent models (we did not include model selection uncertainty in this analysis) We ran 10,000 parametric bootstraps, each for 50,000 years. We incl uded environmental stochasticity (see above) but not demographic stochasticity (see below). Which parameters impact the stochastic growth rate the most and whether this is primarily due to the sensitivity of s to their means or their standard deviatio ns is often of considerable interest for determining research and management priorities. We estimated overall stochastic sensitivities, overall stochastic elasticities, stochastic elasticities to the parameter means, and stochastic elasticities to the par ameter standard deviations of s to lower level parameters (Caswell, 2005; Tuljapurkar et al., 2003) Incorporating Density Dependence There was strong evidence for negative density depend ence in kitten survival, moderate evidence for positive density depend ence in probability of breeding (with weak evidence for a quadratic trend), weak evidence for negative density depend ence in sub adult and adult survival, and weak evidence for positive density dependence in number of kittens produced by females (Benson et al., in revision; Hostetler et al., 2010; chapter 4) The presence and form of density dependence in population models can affect the estimated probability of extinction (Henle et al., 2004) Therefore, we included

PAGE 105

105 models that did and did not include the effect of panther abundance on vital rates in our simulations. The annual index of abundance (Benson et al., in revision; Hostetler et al., 2010; chapter 4) was the minimum count of older kittens (out of the den but still dependent on their dams) sub adult and adult panthers of both sexes. Since age class 1 in our birth flow model can be roughly viewed as representing ages 0.5 1.5 years, we included all modeled age classes in com puting abundance. Because our model was female only but the index of abundance was based on both sexes, we had to convert modeled female population size into total population size. We used the following formulae to estimate the proportion of female s (Miller and Lacy, 2005) : ( 5 8 ) ( 5 9 ) where w i f is the proportion of total panthers that are females in age class i w i m is the proportion of total panthers that are males in age class i l i f is the female survivorship to age class i l i m is the male survivorship to age class i and f is the proportion of female s ( Table 5 1) This formula was parameterized using demographic parameters obtained from the top density independent models (Tables 2 5 ; male sub adult, prime adult, and older adult survivals were 0.712, 0.802, and 0.624, respectively). Using t his approach, w e estimated f at 0.609. P reliminary simulations showed that total equilibrium population size of both sexes (back computed from the same f used for density dependence) varied less than one panther across the range of possible f values (a lthough the equilibrium population size in terms of number of females did vary),

PAGE 106

106 suggesting that model outputs are generally robust to violations of the assumption of constant f. At each time step of a density dependent simulation (after applying enviro nmental stochasticity but before applying demographic stochasticity; see Box 1 below) the abundance index was computed by dividing total female population size by f .. We recomputed the vital rates affected by abundance using this index. These vital rates could be used to compute A t or for individual panthers under demographic stochasticity (see below). Not all model sets included negative density dependence, and simulations with those that did not could experience extremely large populations. For biol ogical plausibility and computational efficiency, we imposed a ceiling density limit of 1000 female panthers for all simulations. When the population exceeded 1000 females, we removed panthers proportionally from all age classes to reach a total populatio n size of 1000. Incorporating Demographic Stochasticity At small population sizes demographic stochasticity (fluctuations in population growth caused by the random fates of individuals) can increase probability of extinction and otherwise influence popu lation dynamics. We simulated demographic stochasticity by applying the time specific vital rates (influenced by environmental stochasticity and/or density dependence) to the fates of individual female panthers. The number of females that survived a year in each age class was determined by drawing a binomial random sample using the matrix entry survival probabilities ( P i t ). The birth flow model complicated determining how many females reproduced in a year. We constructed a birth flow breeding probabili ty from equations 5.4 and 5.6:

PAGE 107

107 ( 5 10 ) We could then determine the numbe r of females that bred in a year by drawing binomial random samples using PB i t We thus modeled P i t into breeding probability separate from its effect on number of survivors. We used the cumulative logit formula to randomly determine the number of offspr ing for each female that bred in a year: ( 5 11 ) where n r is the number of offspring for individual r r is a uniform random number generated for individual r and rs is the probability that individual r will have at most s offspring (chapter 4 ; Table 5 1 ). We drew random binomial samples to determine the number of these ki ttens that were female (using f k ) and survived the first year (using ). When large numbers of individuals were simulated for one time step using this algorithm, the numbers in each age class the next year closely matched the resu lts from matrix multiplication as they should When number of females in every age class exceeded 20, we used matrix multiplication instead of demographic stochasticity to project the population forward, but rounded the results so they could be used with demographic stochasticity (which operates on individuals) the next year (Morris and Doak, 2002) This was done to r educed as population size increases (Morris and Doak, 2002)

PAGE 108

108 Estimating Extinction Parameters We simulated populations starting with the current estimate of popul ation size (100 total panthers) W e determined number of females in each age class by multiplying 100 by each w i f (see density dependence section and Table 5 1), and then rounding (so demographic stochasticity could be applied). After rounding, the estimated total number of females was 5 9. We simulated the population forward 500 years. If a simulated population fell below a critical threshold (set at 1, 10, 20 and 30 females) it was considered extinct or quasi extinct from that point forward. In addition to probabilities of extinction and quasi extinction, we were interested in the timing of these events. Therefore we also kept track of the year each simulation first fell below each critical threshold (extinction and quasi extinction times). We ran simulations with the equivalent of fo ur nested loops (Bakker et al., 2009; Ellner and Fieberg, 2003; White, 2000) The process used is detailed in Figure 5 1. We present mean, median, and 2.5 and 97.5 percentiles of extincti on and quasi extinction probabilities at each time step, and minimum, maximum, mean, median, and 2.5 and 97.5 percentiles of extinction and quasi extinction times. What If the Introgression Never Happened? An interesting question in the management of the Florida panther is: what might have happened to the panther population if genetic introgression had not occurred? To address this question, we estimated demographic parameter for canonical panthers only ( Appendix L ). However, we used estimates of t from the overall population as described previously. We assumed that demographic parameters based on canonical panthers after introgression were representative of a hypothetical post 1995 canonical

PAGE 109

109 panther population. Using these estimates, we parameterized a p rojection matrix and estimated population growth rate for a canonical population ( (c) ), as well as error in (c) To determine how lower level parameters drove differences between and (c) we performed a fixed effect, one way life table response experiment analysis (LTRE; Caswell, 2001) In addition to population growth, we wanted to investigate how introgression affected proba bility of extinction. We ran the overall population and canonical population models both starting at the abundance index for 1995 (26, or 15 females; McBride et al., 2008) We ran 1000 runs of 1000 simulations, each for 500 years using procedures described previously. Since the starting female po pulation size was <20, we only estimated probability of quasi (extinction) and extinction times for critical thresholds of 1 and 10 panthers. Results Density Independent Growth Rate and Perturbation Analysis The estimate of deterministic from the top se t of statistical models was 1.06 ( 2.5 and 97.5 percentiles : 0.95 1.18). When model selection uncertainty was incorporated, dropped to 1.04 ( both mean and median; 2.5 and 97.5 percentiles : 0.94 1.16). The had the highest sensitivity to kitten surv ival and the highest elasticity to prime adult survival; model selection uncertainty had little effect on sensitivity or elasticity (Figure 5 2 ). The estimate of s was 1.05 ( mean and median = 1.04; 2.5 and 97.5 percentiles : 0.95 1.1 5 ). Stochastic ove rall sensitivities and elasticities were similar to deterministic ones, with s having the highest sensitivity to kitten survival and the highest elasticity to prime adult survival (Figure 5 3 A & B). Most of the overall stochastic elasticity was

PAGE 110

110 contribut ed by elasticity to the means of parameters, not their temporal standard deviations (Figure 5 3 C & D) ; s had much higher magnitude elasticity to the standard deviation of S k than th ose of other vital rates (Figure 5 3 D; sa and a had temporal standard de viation of 0). Extinction Parameters We estimated probability of extinction within the next 100 years at 0.037 (median = 0, 2.5 and 97.5 percentiles: 0 0.55; Figure 5 4 A). The probabilities of quasi extinction within the next 100 years were 0.083 (media n = 0, 2.5 and 97.5 percentiles: 0 0.88), 0.280 (median = 0.12, 2.5 and 97.5 percentiles: 0 1), and 0.732 (median = 0.92, 2.5 and 97.5 percentiles: 0.01 1) for critical threshold s of 10, 20, and 30 females, respectively (Figure 5 4 B D). At 500 years 2.5 and 97.5 percentiles stretched from 0 to 1 for all critical limits except 30 females, for which they were 0.05 1. Distributions of extinction and quasi extinction times are shown in Table 5 6. The median time to extinction (among populations that went extinct within 500 years) was 128 years. Median time to (quasi )extinction dropped with increasing critical numbers, and mean times of extinction and quasi extinction were higher than the medians ( Table 5 6). What If the Introgression Never Happene d? The estimate of (c) for the hypothetical canonical population based on the top set of statistical models was 0.95 (mean and median = 0.94; 2.5 and 97.5 percentiles: 0.82 1.07). When model selection uncertainty was included, the estimate of (c) was still 0.95 (mean and median; 2.5 and 97.5 percentiles: 0.80 1.12). Patterns of sensitivities and elasticities (not shown) were similar to those for the overall population (Figure 5 2 ).

PAGE 111

111 The largest contributor to the difference between and (c) was kit ten survival probabil ity (Figure 5 5 ). This was due to both a large difference in kitten survival probabilities (0.133; Table 5 3, Appendix L ) and a large sensitivity of and (c) to S k (Figure 5 2 A). Prime adult survival probability also contributed to the lower for the canonical population, while prime adult probability of breeding, sub adult probability of breeding, and sub adult survival probability had slightly compensatory effects on for the canonical population (Figure 5 5 ). Although canonica l panthers in general had lower survival than the overall panther population, the estimate for S sa (c) was slightly higher in the top models ( Appendix L and Table 5 2), because the canonical estimates came from a subset of the data (1997 2006) from which female sub adult survival estimates were higher in general. The predicted (c) from the LTRE (calculated as the sum of the contributions from all vital rates) was also 0.95, indicating that the LTRE approximation was reasonable. A sampling of density dependent simulations run from 1995 population levels (whole population vital r ates and canonical population vital rates scenarios) showed both scenarios tended to increase towards some equilibrium density but that equilibrium densities were, on average, higher for the whole population and the fluctuations greater (Figure 5 6 A and B) This may be due to an average stronger estimated density dependent effect on kitten survival for the canonical population than for the whole population and/or lower average t for q and S sa oa parameters for the canonical population than for the whole population (Tables 2 4 and Appendix L ). In turn, this could have lead to the higher estimated probability of extinction for the whole population (Figure 5 6 C; at 100 years = 0.074, median = 0, 2.5 and 97.5 percentiles: 0 0.963)

PAGE 112

112 than for the canonical population (Figure 5 6 D; at 100 years = 0.039, median = 0, 2.5 and 97.5 percentiles: 0 0.513), but a somewhat lower probability of dropping below 10 females for the whole popu lation (Figure 5 6 E; at 100 years = 0.136, median = 0.015, 2.5 and 97.5 percentiles: 0 0.989) than for the canonical population (Figure 5 6 F; at 100 years = 0.154; median = 0.009, 2.5 and 97.5 percentiles: 0 0.994). Both extinction and quasi extinctio n times were lower, on average, for the whole population ( Table 5 6). Discussion The situation for the Florida panther has undoubtedly improved since the early morphological abnormalities and other biomedical conditions indicative of inbreeding depression have decreased (Benson et al., in revision; Hostetler et al., 2010; Johnson et al., 2010; McBride et al., 2008; Onorato et al., 2010) However, the population still faces numerous threats, including isolation, habitat loss and fragmentation, road kills, other interactions with humans, disease, and the potential future loss of genetic diversity (Onorato et al., 2010) In this context, a quantitative assessment of risk face d by the Florida panther and an objective evaluation of the role of genetic introgression on observed increases in population performance is of paramount importance. The stochastic growth rate was 1.05 ( 2.5 and 97.5 percentiles : 0.95 1.1 5 ) This is cons istent with the evidence for a growing population between 1995 and 2007 (McBride et al., 2008) the period over which the bulk of the data used to estimate parameters was collected. Both and s showed highest elasticity to survival parameters. This is consistent with other studies for long lived species (especially for lower level elasticities; Heppell et al., 2000; Morris and Doak, 2005; Stahl and Oli, 2006) Although fewer studies have examined elasticity patterns for birth flow

PAGE 113

113 populations, t he same patterns seem to hold there (e.g., Alberts and Altmann, 2003; Federico and Canziani, 2005; Fisher et al., 2000) The lower elasticities to sub adult vital rates than to prime adult vital rates is probably an artifact of age class duration (1.5 and 7.5 years, respectively), as matrix entry elasticities (not s hown) generally declined with age. Our estimate of probability of extinction within 100 years (baseline scenario) was 3.7% (2.5 and 97.5 percentiles: 0 55%). Earlier PVA for the Florida panther had, under their baseline scenarios, estimates for this of 0% (Maehr et al., 2002a) and 2%, 5%, or 78.5% (depending on parameter values used; Root, 2004) Root (2004) estimated the probability of dropping from 40 to 20 females within 100 years at 9%, 20%, or 94.1% (depending on parameter values used); our estimate of th e probability that the population falls below 20 and 30 females (approximately by half) within 100 years were 28.0% (2.5 and 97.5 percentiles: 0 100%) and 73.2% (2.5 and 97.5 percentiles: 1 100%), respectively. We believe our PVA results are more robu st than those from earlier PVA efforts, because our results are based on robust estimates of demographic parameters, and a more realistic birth flow population model, and also because we have appropriately incorporated the potential effects of environmenta l and demographic stochasticities. Furthermore, we have considered the influence of parametric and model selection uncertainty. Fieberg and Ellner (2001) and Ellner et al. (2002) demonstrat ed that long term estimates of extinction risk can be highly imprecise, even with considerable quantities of data. Our results for Florida panthers support these predictions. Despite this, we disagree with the notion that the apparently inherent impreci sion of PVA results makes them useless to science or management.

PAGE 114

114 PVA are generally the best tools science has to offer for assessing extinction risk, and it is better to make use of what science has to tell us (taking into account the uncertainty) than to ignore it in favor of subjective assessments (Brook et al., 2002a) An estimate for probability of extinction with wide confidence intervals still represents the most likely value for that probability (under the modeling assumptions that lea d to it). Our inclusion of density dependent effects undoubtedly had a significant impact on our estimated probabilities of extinction and quasi extinction, and may have had different effects at different quasi extinction threshold s (Henle et al., 2004) There are reasons that these results should be interpreted care fully. The abundance index we used (McBride et al., 2008) was not a true estima te of population size and may be biased low in some years. We used different functional forms for each set of vital rates (Benson et al., in revision; Hostetler et al., 2010; Chapter 4) but these functional forms were chosen more for convenience than from biological intuition. We only tested quadratic effects for reproductive parameters (and could not fit those effects for ) and did not test for lagged/delayed density dependence. The abundance index may have stabilized at around 100 panthers in recent years (D. Land, personal communication), but most of our density dependent models (when run deterministically; not shown) l ead to an equilibrium population size (converted to total panthers) lower than that. This may result from error in the abundance index or how we modeled density dependence. We note that the estimated growth and extinction parameters were based on demograp hic data mostly collected during a period of rapid population growth. Field data suggest that the population growth may have ceased. Current and future conditions may therefore make long term population growth lower and probability of

PAGE 115

115 extinction higher t han reported here. Genetic stochasticity (not modeled) may also increase overall probability of extinction. The estimated deterministic growth rate for a canonical population was (c) = 0.95 (2.5 and 97.5 percentiles: 0.82 1.07). Although percentile ra nges for and (c) overlapped, a 10% difference in population growth rate is biologically meaningful. The most important demographic cause of this difference was the lower survival rate for canonical panther kittens. Given the evidence that a canonical p opulation would have declined (or continued to not increase), the concerns of conservationists and managers warranted (Fergus, 1991; Roelke et al., 1993b; Seal, 1994) However, these growth rates were estimated from models without stochasticit y or density dependence. When these were included, the canonical population actually had a lower probability of extinction, but a higher probability of dropping below 10 females. As pointed out in the results, this is probably due to stronger estimated d ensity dependence and/or lower assigned stochasticity for the models including ancestry. Because all of our statistical models had additive effects of abundance and ancestry (when both were included) we did not specifically look at how canonical vital rat es tracked abundance, but assumed that vital rates of all panthers were affected by abundance similarly. Density dependent effects may also have little effect on population dynamics at 1995 population levels. Future analyses could estimate temporal stand ard deviation separately for ancestry models, simulate short term extinction probabilities without density dependence, estimate demographic parameters for the canonical subset of the data separately

PAGE 116

116 (instead of using ancestry models), and/or examine intera ctive effects between abundance index and ancestry. Our models also did not include genetic stochasticity; potential further loss of genetic diversity could have further increased the probability of extinction for a canonical population after 1995. On t he other hand, all vital rates for canonical panthers were estimated over a period when they were sharing the landscape with admixed panthers. Without the increased competition and intraspecific aggression from these panthers, survival of canonical panthe rs (and (c) ) might have been higher. Although our results are suggestive, we cannot therefore state with certainty that the observed increase in panther numbers would not have happened without the genetic introgression (Creel, 2006) Currently we make no allowances for potential catastrophes such as disease outbreaks, which can have large impacts on population viability but which may require guesses as to the frequency and severity of catastrophes to implement (Coulson et al., 2001; de Castro and Bolker, 2005; Gerber et al., 2005) We could expand perturbation analysis for population growth to probability of extinction, using methods such as squared semi partial correlation coefficients (Bakker et al., 2009) We could take into account the impact of random effects on the model weights or fixed effect estimates, or develop true confidence interval ranges based on bias correction techniques for bootstrap percentiles (Caswell, 2001; Efron and Tibshirani, 1998; Ellner and Fieberg, 2003) Finally, a spatially explicit, individual based PVA for the Florida panther would be very useful for answering questions about the effects of habitat changes and behavioral interactions on panther population growth and persistence

PAGE 117

117 Nonetheless the PVA we developed can be used to ask several further management questions, and we plan to do so in the near future Road mortality is one of the leading causes of mortality for the Florida panther (Benson et al., in revision) and 2009 had the largest number of reported panther road kills. Management agencies are curious to know the effect of road mortality (and changes in it) affects population viability. There is no estimate of the effect of road kills on kitten survival; analyses would have to make some assumptions about how kitten survival changes with changes in sub adult and adult road kill m ortality. Although young male Florida panthers frequently roam out of south Florida, no female panthers have been recorded north of the Caloosahatchee River since 1972 (McBride, 2002) Starting a breeding population of Florida panthers in central Florida or elsewhere within their former range might be important for the long term recovery of the subspecies (Thatcher et al., 2009; USFWS, 2006) Against this, managers must weigh the impacts of removing female panthers from south Florida on the viability of the source population. We can predict the impacts of different numbers, ages, and timing of female panther removals on south Florida panther viability quite easily with the current model. In addition, we co uld model the viability of the new population or the joint viability of the two populations, although this would require making several (possibly unwarranted) assumptions about vital rates and stochasticity in the new location. The ideal age class to remo ve for behavioral reasons might be the sub adult females that have not yet set up home ranges. This age class also has the highest reproductive value (at least for the top deterministic density independent model; J. Hostetler,

PAGE 118

118 unpublished results) and so might also be ideal for contributing to population growth of the new population (Maguire and Servheen, 1992) The panther population size was at best stable (at low numbers) before 1995 but grew rapidly afterwards (McBride et al., 2008) We found evidence that the difference in survival probabilities between canonical and admixed panthers (particularly for kittens; Benson et al., in revision; Hostetler et al., 2010) contributed to this difference in population growth rates. If the Florida panther continues to exist in one small population in south Florida in the futu re, inbreeding depression may again lead to a population declin e Our results suggest that continued genetic introgressions may be a useful strategy for combating this risk.

PAGE 119

119 Table 5 1. Variable names used in this manuscript, with definitions. Variable D efinition abundance Modeled effects of abundance index on vital rates abundance2 Modeled effects of squared abundance index on vital rates rs Probability that individual r will have at most s offspring (given that she breeds) 2 (Multivariate) random normal nu mber generated for sub adult and adult log hazard r Uniform random number generated for individual r Deterministic asymptotic population growth rate s Stochastic asymptotic population growth rate a Average number of kittens produced by an adult f emale in a year, given that she breeds sa Average number of kittens produced by an sub adult female in a year, given that she breeds t Temporal standard deviation in a group of vital rates A t Time specific population projection matrix (c) Parameter estimated for canonical panthers f. (Stable) proportion of entire population that is female f k Proportion of newborn kittens that are female F i t Fertility (number of female kittens produced each year per female) for female of age class i in year t l i f Survivorship of a female to age i l i m Survivorship of a male to age i m i t Fecundity (number of female kittens produced each year per surviving female) for female of age class i in year t n r Number of kittens produced by individual r in a year, given tha t she breeds P i t Survival probability for female of age class i in year t PB i t Probability that a female of age class i breeds in year t q oa Probability that an older adult female breeds in a year, given that she survives q pa Probability that an prime adult female breeds in a year, given that she survives q sa Probability that an sub adult female breeds in a year, given that she survives S k Annual survival probability for kittens (age 0 1 years) S oa Annual survival probability for older adults (age 1 0 years and up) S pa Annual survival probability for prime adults (ages 2.5 10 years) S sa Annual survival probability for sub adults (ages 1 2.5 years) w i f Stable proportion of total panthers that are females in age class i w i m Stable proportion of tota l panthers that are males in age class i

PAGE 120

120 Table 5 2. Models for sub adult and adult survival, with Akaike model weights, estimates of sub adult female survival ( S sa ), prime adult female survival ( S pa ), older adult female survival ( S oa ), the effect of abu ndance index on log hazard ( abundance ), and temporal standard deviation ( t ). Survival estimates for density dependent models are for abundance index 0. Age category model names are different than in Benson et al. (in revision) Model Weight S sa S pa S oa abundance t Sex Age3 a + Older b 0.357 0.954 0.032 0.873 0.022 0.748 0.048 0 0.34 Sex Age3 + Older + Abundance c 0.313 0.966 0.025 0.906 0.029 0.795 0.057 0.005 0.004 0.31 Sex Age1 d 0.133 0.954 0.032 0.872 0.024 0.754 0.055 0 0.34 Sex Age1 + Abundance 0.118 0.966 0.025 0.905 0.030 0.803 0.061 0.005 0.004 0.30 Sex Age3 0.034 0.954 0.032 0.852 0.022 0.852 0.022 0 0.25 Sex 0.018 0.868 0.019 0.868 0.019 0.868 0.019 0 0.22 Sex Age3 + Abundanc e 0.018 0.961 0.028 0.875 0.034 0.875 0.034 0.003 0.004 0.24 Sex + Abundance 0.010 0.889 0.030 0.889 0.030 0.889 0.030 0.003 0.004 0.21 a Age 3 divides panthers into sub adults (1 2.5 and 1 3.5 years old for females and males respectively) and males respectively). b Older refers to older c Abundance refers to a linear trend in log hazard by abundance index. d Age 1 divides the panthers into sub adults (same as Age3), prime adults (2.5 10 and 3.5 10 years old for females and males respectively), and older

PAGE 121

121 Table 5 3. Models for kitten survival, with Akaike model weights, estimates of female kitten survival ( S k ), the effect of abundance index on logi t kitten survival ( abundance ) and temporal standard deviation ( t ). Survival estimates for density dependent models are for abundance index 0. For more details, see Hostetler et al. (2010) Model Weight S k abundance t S (Base a + k :Abundance b ) r (.) c 0.379 0.806 0.119 0.034 0.011 0.37 S (Base + k :Abundance) r ( k s&a ) d 0.337 0.829 0.115 0.034 0.011 0.46 S (Base + k :Abundance + k :sex e ) r (.) 0.148 0.824 0.119 0.035 0.011 0.36 S (Base + k :Abunda nce + k :sex) r ( k s&a ) 0.126 0.843 0.116 0.034 0.012 0.36 S (Base) r ( k s&a ) 0.004 0.345 0.076 0 1.03 S (Base) r (.) 0.003 0.291 0.050 0 1.01 S (Base + k :sex) r ( k s&a ) 0.001 0.350 0.090 0 S (Base + k :sex) r (.) 0.001 0.305 0.069 0 a The ba se survival model represents survival differing between kittens, females age 1 and 2, females 3 and older, males age 1 3, and males 4 and older (Hostetler et al., 2010) b k : Abundance refers to a linear trend in logit kitten survival by abundance index. c The same recovery probability for all panthers. d Recovery probabilities differing between kittens and older panthers. e Kitten survival differing by sex. These models are > 10 QAIC c units from the top model, but are < 10 QAIC c units from the top density independent model, and were therefore included in estimation of but not in estimation of probability of extinction.

PAGE 122

122 Table 5 4. Models for sub adult and adult prob ability of breeding, with Akaike model weights, estimates of sub adult probability of breeding ( q sa ), prime adult probability of breeding ( q pa ), older adult probability of breeding ( q oa ), the effects of abundance index and abundance index squared on monthl y complementary log log probability of breeding ( abundance and abundance2 respectively) and temporal standard deviation ( t ). Probability of breeding estimates for density dependent models are for abundance index 0. Model Weight q sa q pa q oa abundance abundance2 t Age2 a + Abundance b 0.304 0.278 0.095 0.278 0.095 0.024 0.025 0.009 0.005 0 0.38 Age2 0.209 0.459 0.037 0.459 0.037 0.038 0.037 0 0 0.42 Age2 + Abundance + Abundance 2 0.152 0.462 0.293 0.462 0.293 0.044 0.055 0.011 0.025 0.0001 0.0002 0.38 Age1 c + Abundance 0. 141 0.252 0.096 0.301 0.105 0.025 0.026 0.008 0.005 0 0.37 Age1 0.128 0.382 0.083 0.477 0.041 0.038 0.037 0 0 0.41 Age1 + Abundance + Abundance 2 0.066 0.407 0.287 0.465 0.292 0.043 0.054 0.009 0.025 0.0001 0.0002 0.37 a Age 2 div ides panthers into young adults (1 10 years old) and older b Abundance refers to a linear trend in monthly complementary log log probability of breeding by abundance index. c Age 1 divides the female panthers into sub adults (1 2.5 years old), prime adults (2.5 10 years old) and older

PAGE 123

123 Table 5 5. Models for the average number of kittens produced by sub adult and adult females (for females that bred in a year), with Akaike model weights, estimates of sub adul t kittens ( sa ), adult kittens ( a ), and the effects of abundance index on cumulative logit number of kittens ( abundance ). Number of kittens estimates for density dependent models are for abundance index 0. Estimated temporal standard deviation ( t ) was 0. Model Weight sa a abundance Age3 a 0.368 3.13 0.28 2.69 0.12 0 Constant b 0.318 2.76 0.12 2.76 0.12 0 Age3 + Abundance c 0.177 2.87 0.44 2.41 0.38 0.006 0.009 Abundance 0.137 2.55 0.39 2.55 0.39 0.005 0.009 a Age 3 divides female panthers into sub adults (1 b No predictor variables. c Abundance refers to a linear effect of abundance index on cumulative logit number of kittens produced by female panthers.

PAGE 124

124 Table 5 6. Summary st atistics for extinction and quasi extinction times (number of years until first dropping below critical thresholds). The starting size refers to the total number of females at the beginning of each simulation and the population refers to the group of pant hers parameters were estimated for. Therefore, the first four rows are for the baseline scenario, the next two rows are for the baseline parameter values but starting at 1995 population levels happened )extinction time was always 500, the number of years all simulations were run. Population Starting Size Critical Threshold Minimum Mean Median 2.5 and 97.5 Percentiles Whole 59 1 12 171.1 12 8 29 468 Whole 59 10 6 155.5 108 16 465 Whole 59 20 2 132.2 80 10 456 Whole 59 30 1 62.2 25 5 358 Whole 15 1 5 120.4 60 15 450 Whole 15 10 1 104.6 36 2 454 Canonical 15 1 6 167.6 123 20 471 Canonical 15 10 1 137.4 86 4 461

PAGE 125

125 Figure 5 1. Pseudo c ode for estimating probability of extinction. PARAMETRIC BOOTSTRAP LOOP (ru ns; 10,000 times) Select models randomly by AIC weights Select parameters from the distributions for those models SIMULATION LOOP (simulations; 100 times) TIME STEP LOOP (years; 500 years) INDIVIDUAL LOOP (number of individuals) Determine fate of i ndividual using demographic stochasticity END INDIVIDUAL LOOP Determine number of individuals the next year END TIME STEP LOOP END SIMULATION LOOP Estimate probabilities of extinction and quasi extinction as the proportion of simulations that go e xtinct/quasi extinct END PARAMETRIC BOOTSTRAP LOOP Summarize distributions of (quasi )extinction times and probabilities

PAGE 126

126 Figure 5 2. Perturbation analysis deterministic population growth rate ( ) to lower level parameters (see Table 5 1). A) Sensitivity for the set of top ranked density independent statistical models and for all density independent statistical models. B) P roportional sensitivity (elasticity) for the s ame Solid bars represent point estimates for top models (means and median of parametric bootstraps were very sim ilar) and medians of parametric bootstraps for all models (means were very similar). Error bars represent 2.5 and 97.5 percentiles of parametric bootstrap runs in both cases.

PAGE 127

127 Figure 5 3. Perturbation analysis results for stochastic population gro wth rate ( s ) to lower level parameters (see Table 5 1) for the set of top ranked density independent statistical models. A) Stochastic sensitivities. B) Overall stochastic elasticities. C) Stochastic elasticities to means. D) Stochastic elasticities t o standard deviations. Solid bars represent point estimates, solid circles represent medians of parametric bootstrap runs, and open circles represent means. Error bars represent 2.5 and 97.5 percentiles of parametric bootstrap runs.

PAGE 128

128 Figure 5 4. S imulated cumulative probabilities of extinction and quasi extinction over 500 years. Panels depict estimated probabilities of reaching A) <1 female, B) <10 females, C) <20 females, and D) <30 females. The thick dashed lines represent simulation means, th e solid lines simulation medians, and the dotted lines 2.5 and 97.5 simulation percentiles.

PAGE 129

129 Figure 5 5. Contributions of lower level parameters (see Table 5 1) to differences between the deterministic population growth rates for the whole population ( ) and the canonical population ( (c) ). Negative contributions lead to (c) being lower than

PAGE 130

130 Figure 5 6. Results of simulations starting with conjectured 1995 population levels (total 15 females) and run for 500 years. Panels A and B) depi ct random selections of population sizes for different scenario runs; C and D) depict cumulative probabilities of reaching <1 female; and E and F) depict cumulative probabilities of reaching <10 females. A, C, and E) are for the whole population and B, D, and F) are for a canonical population. The thick dashed lines represent simulation means, the solid lines simulation medians, and the dotted lines 2.5 and 97.5 simulation percentiles.

PAGE 131

131 CHAPTER 6 SUMMARY AND CONCLUSI ONS Our study achieved 3 major object ives. First, we have used all a ppropriate data an d robust statistical methods to provide rigorous estimates of demographic parameters for Florida panthers. Second we have provided estimates of population growth and persistence parameters for our study pop ulation. Finally we have objectively tested for the potential influence of genetic introgression, one of the most important and controversial management action s undertaken for the conservation of the panther, on demographic parameters and population grow th We estimated overall annual kitten (age 0 1 year) survival at 0.328 0.0 65 (SE) and sub adult and adult panther survival s ranging from 0.951 0.03 4 for female sub adults (1 2.5 years) to 0.6 35 0.08 3 for male older adults The estima te of kitten survival is lower than previously reported. There was evidence that sub adult and adult female s had higher survival than male sub adults and adults but no evidence of a difference in survival by sex for kittens. Survival within th e first ye ar increases with age, whereas female survival declines from sub adults to prime adults (2.5 10 years) to older adult stage, which was defined as 3.5 10 years. Overall female 6 month probability o f breeding was 0.232 0.021 and average litter size was 2. 6 0 0. 09. Sub adult dams had larger litters on average than adult (ages 2.5 10 years) dams, but we were unable to detect a difference in probability of breeding between sub adult and prime adul t female panthers. However, older adult females did have a lower probability of breeding ( 0.019 0.019 ). Females were more

PAGE 132

132 likely to give birth in the dry season (December May) than in the wet season (June November) and also had larger litters in t he dry season. We used annual minimum panther population counts (McBride et al., 2008) as an index of abundance to estimate the effects of density on vital rates. We found that abundance negatively affected the probability of survival for kittens and positively affected the female probability of breeding (which may have been an artif act of its effect on kitten survival). However, we found little evidence for an effect of abundance on litter size or sub adult and adult survival. Deterministic annual population growth rate ( ) for the overall population was 1.06 (2.5 and 97.5 percentil es: 0.95 1.18) and stocha stic annual growth rate ( s ) was 1.05 (2.5 and 97.5 percentiles: 0.95 1.15), which suggest that the population has been growing and corroborates recent trends in minimum population counts Elasticity analyses showed that s wa s proportionately most sensitive to changes in survival, particularly of prime adult and sub adult females. Our stochastic simulations incorporating demographic and environmental stochasticity parameter and model selection uncertainty, and density depende nce provided an estimate of the probability of extinction within 100 years of 3.7% (2.5 and 97.5 percentiles: 0 55%) We estimated the probability of dropping below 30 females at some point within the next 100 years at 73.2% (2.5 and 97.5 percentiles: 1 100%). Among populations that went extinct within 500 years, the median time to extinction was 128 years (2.5 and 97.5 percentiles: 29 468). These estimates may be optimistic, as most of the data used for our models were collected during a period of population increase, and our model did not include

PAGE 133

133 catastrophic events, genetic stochasticity, or potential future changes in habitat or anthropogenic influences. We found considerable evidence that the genetic introgression in 1995 had effects on both de mographic rates and population dynamics. Florida panther survival was higher for F 1 admixed (0.4 94 0.1 56 ) and backcrossed to Texas (0. 400 0.1 24 ) kittens than for backcrossed to canonical (0.2 67 0.0 78 ) and canonical Florida panther (0.2 45 0.0 73 ) ki ttens For sub adults and adults, survival was higher for F 1 admixed (0.978 0.021) panthers than for other admixed (0.821 0.036) or pre introgression type (canonical and Everglades) panthers (0.775 0.039). Conversely, 6 month probability of breedin g was lower for F 1 admixed sub adult and adult female panthers than for other admixed or pre introgression type female panthers (although this may have been an artifact of higher kitten survival), and there was no evidence that ancestry affected litter siz e. A population matrix model parameterized for a hypothetical population of all canonical panthers yielded a of 0.95 (2.5 and 97.5 percentiles: 0.82 1.07), suggestive of a shrinking population. A life table response experiment (LTRE) demonstrated that the vital rate with the largest effect on the difference between for the overall population and for the ca nonical population was kitten survival. T hese results can not conclusively prove that the observed post introgression increase in the panther population would not have occurred without genetic introgression or that factors other than introgression did not c ontribute to th e observed hat outside the reach of science. They do, however, strongly support the hypothesis that

PAGE 134

134 genetic introgression had a positive population level effect on Florida panthers, primarily via improved survival of admixed panthers.

PAGE 135

135 APPENDIX A AGE ERROR ESTIMATION Methods To examine the error in survival estimates associated with using the estimated age at capture to estimate the age category and age category chan ges of each animal, we created a range of plausible birthdates for each panther, centered at the estimated birthdate. When the field estimate of age included a range (e.g., 3 4 years old), we used that range. Otherwise, we used a range of 1 year (6 months on either side of the estimated birthdate). Birthdates were known (within a few days) for 54 panthers; birthdate ranges for other panthers were as low as 1 month or as high as 2 years (Table A 1). We used a resampling approach, randomly selecting (from a uniform distribution) a birthdate for each panther whose age was unknown within its birthdate range. We then assigned age category and age category changes for each panther based on the new birthdates. We ran the survival analysis for the top ranked age and sex category model (Sex Age1 + Older). We repeated these steps 1000 times, and recorded the mean and standard deviations of the survival estimates. We used the standard deviation as a measure of the error due to choosing the midpoint of the range. This assumes that the true birthdates were somewhere within the ranges. We also assumed that these errors and the standard errors of the original survival estimates were independent, and estimated the total error as:

PAGE 136

136 where is the survival rate estimated for category i using the Cox proportional hazard method and the midpoints of the birthdate ranges, SE( ) is the standard error of that estimate, is the mean of the resampling survival estimates for category i and SD( ) is the standard deviation of the resampling survival estimates for category i Results and Discussion The mean survival estimates from the resampling analysis were generally quite close to the original survival estimates (Table A 2). The errors due to birthdate selection were smaller than the errors due to other causes. The total errors were very similar to the errors due to other causes. The relatively high difference between and and relatively high SD( ) for female sub adults are probably due to a single female panther that died a few days less than a year after her estimated bir thdate. This panther was not included as a sub adult in the survival analyses in the body of the paper (i.e., because she was classified as a kitten) but was included in some resampling runs, bringing down the female sub adult survival estimates in those runs. We did not test the effects of birthdate error on model selection or multi model inference, but suspect that those effects would be quite small as well.

PAGE 137

137 Table A 1. Distribution of lengths of birthdate ranges (in months). A birthdate range of 0 i ndicates that the birthdate of the panther was known within a few days. The default birthdate range length used (when only a point estimate for birthdate was given in the field) was 12 months. Length of Birthdate Range (Months) Number of Panthers 0 55 1 5 2 15 3 1 6 7 12 58 24 5

PAGE 138

138 Table A 2. Resampling results examining error from using estimated age at capture to estimate survival by age class. For each sex and age class, we present estimated survival ( ), standa rd error of that estimate ( SE ( ) ), mean survival from the resampling approach ( ), standard deviation of that mean ( SD( ) ), and the total error from both sources ( TE( ) ). All results are using the top ranked sex and age model (Sex Age1 + Older). Sex Age Class SE ( ) SD( ) TE( ) Female Sub Adult 0.95 4 0.03 2 0.937 0.017 0.036 Female Prime Adult 0.87 3 0.02 2 0.875 0.003 0.022 Female Older Adult 0.7 48 0.0 48 0.747 0.011 0.049 Male Sub Adult 0.71 2 0.0 49 0.714 0.011 0.050 Male Prime Adult 0.802 0.0 34 0.802 0.008 0.034 Male Older Adult 0.6 24 0. 073 0.617 0.015 0.074

PAGE 139

139 APPENDIX B FINE SCALE ANCESTRY EFFECTS ON SUB ADULT AND ADULT SURVIVAL Methods In order to investigate potential variation in survival across ancestry categories based on finer scale genotypic distinctions t han those used in the body of the paper, we placed individuals from our radio collared sample into 7 categories: canonical Florida panthers ( n = 40), Everglades panther ( n = 1), F 1 admixed panthers ( n = 8), backcrossed to canonical admixed panthers ( n = 16 ), backcrossed to Everglades admixed panthers ( n = 4), backcrossed to Texas admixed panthers ( n = 14), and unknown admixed panthers ( n = 15). We assigned individuals to these categories using pedigree analysis and results of Structure analysis of microsat ellite allele distributions. Canonical Florida panthers showed no evidence of non Florida genetic admixture (no direct non canonical relatives or < 10% non canonical genetic contribution based on Structure analyses). Everglades panthers were mostly capt ured within or adjacent to ENP and had documented genetic links to Central America (O'Brien et al 1990, Culver et al. 2000). Note that canonical and Everglades panthers were combined into 1 category (pre introgression type panthers) for the ancestry anal ysis in body of the paper which represents the known genotypes present on the south Florida landscape prior to the introgression program. F 1 admixed panthers were defined previously in body of the paper. Backcrossed to Texas admixed panthers were admixed panthers of predominantly (>50%) Texas heritage resulting from admixed panthers breeding with introduced Texas cougars. Backcrossed to canonical panthers were admixed panthers with predominantly (>50%) canonical heritage. Backcrossed to Everglades were admixed panthers with predominantly (>50%) Everglades heritage resulting from

PAGE 140

140 breeding events between Texas admixed and Everglades panthers. Unknown admixed panthers were excluded because we were missing information that precluded placing them into the fi ner admixture categories (e.g., 1 or both parents were not identified). Note that backcrossed to Texas, backcrossed to canonical, backcrossed to Everglades, and unknown admixed were all combined into the other admixed category in the ancestry analysis in the body of the paper. We used ancestry models (GenAnc1 GenAnc6; Table B 1 ) with different combinations of ancestry categories to test specific hypotheses. We combined these 6 models with the effects of sex, age class (sub adult and prime adult; Age3), and heterozygosity. Other than the unknown admixed exclusions and the different ancestry models, this analysis was identical to the second analysis described in the methods of the main text. Results There was again strong evidence that genetic ancestry influenced survival as the top 3 models included an ancestry variable and the sum of the weights of models including ancestry variables was 0.854 (Table B 2 ) As in our main ancestry analysis, the evidence indicated that ancestry influence d survival through higher F 1 survival (ancestry models GenAnc4 and GenAnc6; sum of weights = 0.653), whereas models suggesting differences between survival of canonical and backcrossed panthers (GenAnc1 5) or between different classes of backcrossed panthe rs (GenAnc2, GenAnc3, and GenAnc5) had less empirical support (sum of weights = 0.394 and 0.113, respectively; Table B 2 ; Figure B 1 ). The evidence that higher heterozygosity positively influenced sub adult and prime adult survival remained strong (Table B 2 ; Figure B 1 ). Although model averaged survival was similar for canonical, backcrossed to canonical,

PAGE 141

141 and backcrossed to Texas panthers at a given level of heterozygosity, mean heterozygosity was highest for backcrossed to Texas individuals, leading to an increased probability of survival, and lowest for canonical panthers, leading to a decreased probability of survival ( Figure B 1 ).

PAGE 142

142 Table B 1. Definitions of ancestry models, and the hypotheses they represent. Term Definition Hypothesis GenAnc1 2 an cestry categories: 1) Canonical and Everglades; 2) Backcrossed to Canonical, Backcrossed to Everglades, Backcrossed to Texas, and F 1 General introgression advantage (or disadvantage). GenAnc2 2 ancestry categories: 1) Canonical and Backcrossed to Canon ical; 2) Everglades, Backcrossed to Everglades, Backcrossed to Texas, and F 1 General admixture advantage (or disadvantage), but with Backcrossed to Canonical reverting to Canonical. GenAnc3 2 ancestry categories: 1) Canonical, Everglades, Backcrossed t o Canonical, and Backcrossed to Everglades; 2) Backcrossed to Texas, and F 1 Survival advantage (or 50% Texas. GenAnc4 3 ancestry categories: 1) Canonical and Everglades; 2) Backcrossed to Canonical, Backcrossed to Everglades, and Backcrossed to Texas; and 3) F 1 Differences in survival between pre introgression types, backcrosses, and F 1 GenAnc5 3 ancestry categories: 1) Canonical and Everglades; 2) Backcrossed to Canonical and Backcrossed to Everglades; 3) Backcrossed to Texas and F 1 Differences in survival between pre introgression intermediate types. GenAnc6 2 ancestry categories: 1) Canonical, Everglades, Backcrossed to Canonical, Backcrossed to Everglades, and Backcrossed to Texas; and 2) F 1 Survival advantage (or disadvantage) for F 1

PAGE 143

143 Table B models for the effects of ancestry, heterozygosity, sex, and age class on adult Florida panther survival. For each model, we show the number of parameters, Akaike d ifferences ( w i ) The ancestry categories (GenAnc1 GenAnc6) are defined in Table B 1. This analysis was restricted to sub adults and prime adults that were of known ancestry and to the years 1997 2006. Model Parameters w i Sex Age3 a + GenAnc6 + Het b 5 0.00 0.148 Sex + GenAnc6 + Het 3 0.54 0.113 GenAnc6 + Het 2 0.94 0.093 Sex Age3 + Het 4 1.77 0.061 Sex Age3 + GenAnc4 + Het 6 2.00 0.055 Sex Age3 + GenAnc6 4 2.01 0.054 Sex + Het 2 2.42 0.044 Sex + GenAnc4 + Het 4 2 .49 0.043 Het 1 2.68 0.039 Sex + GenAnc6 2 2.75 0.037 GenAnc4 + Het 3 2.93 0.034 Sex Age3 + GenAnc4 5 3.05 0.032 Sex Age3 + GenAnc3 + Het 5 3.73 0.023 Sex Age3 + GenAnc1 + Het 5 3.74 0.023 Sex Age3 + GenAnc2 + Het 5 3.77 0.023 Sex + GenAnc4 3 4.01 0.020 Sex + GenAnc3 + Het 3 4.39 0.016 Sex + GenAnc2 + Het 3 4.42 0.016 Sex + GenAnc1 + Het 3 4.42 0.016 GenAnc3 + Het 2 4.66 0.014 GenAnc1 + Het 2 4.66 0.014 GenAnc2 + Het 2 4.68 0.014 GenAnc6 1 4.80 0.013 GenAnc4 2 5.53 0.009 Sex Age3 + GenAnc5 + Het 6 5.65 0.009 Sex + GenAnc5 + Het 4 6.39 0.006 GenAnc5 + Het 3 6.61 0.005 Sex Age3 + GenAnc2 4 7.13 0.004 Sex Age3 + GenAnc1 4 7.33 0.004 Sex + GenAnc2 2 8.43 0.002 Sex Age3 + GenAnc3 4 8.67 0.002 Sex + GenAnc1 2 9.07 0.002 Sex Age3 + GenAnc5 5 9.12 0.002 GenAnc2 1 9.43 0.001 Sex Age3 3 9.54 0.001

PAGE 144

144 Sex + GenAnc3 2 9.59 0.001 Sex + GenAnc5 3 10.58 0.001 GenAnc1 1 10.81 0.001 Sex 1 10.97 0.001 GenAnc3 1 11.38 0.001 GenAnc5 2 12.12 0.000 Constant 0 14.09 0.000 a Age3 d ifferentiates sub adult and prime adult panthers. b The predictor variable Het refers to heterozygosity.

PAGE 145

145 Figure B 1. Model averaged effects of complex ancestry and heterozygosity on annual survival of female prime adult Florida panthers from 1997 20 06. Panthers with Everglades and backcrossed to Everglades ancestries were included in the analysis but are not presented here; unknown admixed panthers were excluded from this analysis. The range of heterozygosity values presented for each ancestry cate gory roughly corresponds to those found in the dataset (all panthers with that ancestry in the genetics analysis). Error bars represent unconditional standard errors.

PAGE 146

146 APPENDIX C CAUSE SPECIFIC MORTALITY ANALYSIS Materials and Methods We attributed mortal ity of radio collared panthers to 1 of 4 causes: 1) hit by vehicle, 2) intraspecific aggression, 3) other (included known causes of death such as disease, heart failure, and infections unrelated to intraspecific aggression), and 4) unknown (mortalities for which evidence from field and necropsy examinations was insufficient to assign cause of death). We estimated cause specific annual mortality rates for Florida panthers overall and within categories (sex, age class within sex, and ancestry within sex), us ing the nonparametric cumulative incidence function estimator (NPCIFE; Heisey and Patterson 2006). The NPCIFE is a generalization of the staggered entry Kaplan Meier method of survival estimation (Pollock et al., 1989) All cause specific mortality ana lyses were performed using SPLUS c ode from Heisey and Patterson (2006) which we modified for use in R. We tested for effects of covariates (sex, age class, ancestry, and average heterozygosity) on cause specific mortality rates, using the same subsets of the data used for survival analyses We ran a Cox proportional hazard regression for each covariate, stratified by the cause of death (Heisey and Patterson, 2006) We identified effects on cause specific mortality rates. We compared cause specific mortality of sub adults to that of prime adults and cause specific mortality of older adults to that of prime adults for each sex, as well as older adults to prime adults for the sexes combined. For ancestry, we compared cause specific mortality of all admixed panthers (F 1 and other adm ixed were combined due to small sample sizes) to that of pre

PAGE 147

147 introgression types within each sex and for the sexes combined. We also present risk ratios for significant effects of covariates where cause specific mortality rates were not estimated. For he terozygosity (the only continuous covariate), these were scaled as the proportional increases in instantaneous hazard from a cause with an increase of 0.1 in the covariate level (e.g., a change from heterozygosity of 0.2 to 0.3). R esults The greatest cau se of mortality for radio collared Florida panthers was intraspecific aggression (0.078, SE = 0.013, 39 deaths), followed by mortality from unknown causes (0.043, SE = 0.010, 22 deaths), vehicles (0.03 6 SE = 0.008, 1 8 deaths), and other (0.027, SE = 0.008 14 deaths). Mortality was higher from all causes for males than for females ( Figure C 1 A), but only significantly so for intraspecific aggression ( z = 2. 823 p = 0.00 5 ). We were unable to compare mortality rates for sub adult and prime adult females d ue to the low number of deaths in the former. Older adult males had significantly higher ( z = 2. 867, p = 0.004 ) and sub adult males had marginally higher ( z = 1.8 66 p = 0.062 ) mortality due to intraspecific aggression than prime adult males ( Figure C 1 C) Older adult males had no recorded deaths from 25.430, p < 2e 16). Other differences in mortality rates among age classes within sexes were not significant ( Figure C 1 B and C). When sexes were combi ned, older adults had higher mortality due to intraspecific aggression than prime adults ( z = 2. 990 p = 0.003 risk ratio = 2 99 ). Female pre introgression type panthers had a higher level of mortality due to all causes compared to female admixed panthers but not significantly for any cause ( Figure C 1 D). Male pre introgression type panthers had no mortality due to other which was highly significant ( z = 21.861 p < 2e 16 Figure C 1 E). When sexes were

PAGE 148

148 combined, pre introgression type panthers had a h igher level of mortality due to intraspecific aggression than admixed panthers ( z = 2. 404 p = 0.0 16 risk ratio = 3.06). The effect of heterozygosity was significant for intraspecific aggression ( z = 2. 943 p = 0.003, risk ratio = 0.4 8 0 ), with the risk declining with increased heterozygosity.

PAGE 149

149 Female Male Figure C 1 Cause specific mortality rates (with standard errors) A) By sex. B) By age class within females. C) By age class within males. D) By ancestry within fem ales. E) By ancestry with males. For age class within each sex, older adult mortality rates for each cause were compared against prime adult rates. In addition sub adult male rates were compared against prime adult male rates. For ancestry within each sex, pre introgression panthers were compared with all admixed panthers. Sig

PAGE 150

150 Appendix D INFERENCE RESULTS FOR ANCESTRY AND HETEROZYGOSITY Materials and Methods We present the full AIC table for ancestry and heterozygosity (this was lim ited to top models in the main text due to space limitations) and selected Cox model coefficients for the genetic comparisons Along with the Cox model coefficient estimates, we present uncorrected and robust standard error estimates, test results for dif ferences between the coefficients and 0 (Wald/Z tests calculated using the robust standard errors), and the associated p values. Testing whether a categorical mostly equivalent to testing whe ther the survival rates for the two categories are different from each other but allows a single test of that covariate in a model that contains other covariates (e.g., sex and age). Results The full AIC table for the genetic comparisons is found in Tab le D 1 Cox model genetic coefficients for all models containing the sex and age class interaction and at least one genetic covariate are found in Table D 2. The F 1 effect was statistically significant when considered in isolation of heterozygosity ( p = 0.048 for Sex Age3 + Ancest1 model [difference between F 1 and other admixed panthers] and p = 0.031 for Sex Age3 + Ancest 2 model [difference between F 1 and all other panthers] ) The heterozygosity effect was statistically significant when considered i n isolation of ancestry ( p = 0.022 for Sex Age3 + Het model ). In contrast, the pre introgression type effect was not significant (when compared with either other admixed panthers or all other panthers). These results indicate that F 1 admixed panthers s urvived significantly

PAGE 151

151 better than either pre introgression type or other admixed panthers, but that there were not significant differences in survival between pre introgression type and other admixed panthers. When sex, age class, ancestry, and heterozygo sity were included in the same model, neither ancestry nor heterozygosity was significant, likely due to the se 2 factors confounding one another (Table D 2)

PAGE 152

152 Table D 1. Full AIC table for ancestry and heterozygosity. Model comparison results showing all models for the effects of ancestry, heterozygosity, sex, and age class (sub adults and prime adults only) on Florida panther survival during 1997 2006. For each model, we present the number of parameters, on) and the Akaike weight ( w i ) Model Parameters w i 1. Sex Age3 a + Ancest2 b + Het c 5 0.00 0.244 2. Sex Age3 + Ancest2 4 0.23 0.217 3. Sex Age3 + Ancest1 d 5 1.35 0.124 4. Sex Age3 + Ancest1 + Het 6 1.98 0.090 5. Sex + Ancest2 + Het 3 2. 77 0.061 6. Sex + Ancest2 2 2.98 0.055 7. Sex Age3 + Het 4 3.09 0.052 8. Sex + Ancest1 3 4.54 0.025 9. Ancest2 + Het 2 4.69 0.023 10. Sex + Ancest1 + Het 4 4.73 0.023 11. Sex Age3 + Ancest3 e + Het 5 4.99 0.020 12. Ancest2 1 5.99 0.012 13. Sex Age3 + Ancest3 4 6.13 0.011 14. Sex + Het 2 6.38 0.010 15. Ancest1 + Het 3 6.68 0.009 16. Ancest1 2 7.23 0.007 17. Sex Age3 3 7.61 0.005 18. Het 1 8.36 0.004 19. Sex + Ancest3 + Het 3 8.38 0.004 20. Ancest3 + Het 2 10.35 0.001 21. Sex + Ancest3 2 10.72 0.001 22. Sex 1 11.50 0.001 23. Ancest3 1 14.03 0.000 24. Constant 0 15.84 0.000 a Age3 differentiates sub adult (age 1 2.5 for females and 1 3.5 for males) and prime adult (ages 2.5 10 for females and 3.5 10 for males) panthers. b Ancest2 di vides panthers into two ancestry categories: F 1 admixed, and other admixed and pre introgression type combined. c Het refers to individual average heterozygosity. d Ancest1 divides panthers into three ancestry categories: F 1 admixed, other admixed, and pre introgression type. e Ancest3 divides panthers into two ancestry categories: F 1 admixed and other admixed combined, and pre introgression type.

PAGE 153

153 Table D 2. Genetic coefficients from a ll models including sex and age class and at least one genetic covaria te. We present the survival model, the genetic covariate, the regression coefficient associated with that covariate ( ), the uncorrected score, and the associated p value. values are on the log hazard scale; a negative indicates a reduction in hazard (and increase in survival) for an increase in a continuous covariate or the given level of a categorical covariate. Model Covariate Uncorr SE Robust SE z p Sex Age3 a + Ancest1 b Ancest1: PIT c 0.28 0.30 0.29 0.96 0.337 Sex Age3 + Ancest1 Ancest1: F1 d 1.98 1.03 1.00 1.98 0.048 Sex Age3 + A ncest2 e Ancest2: F1 2.14 1.01 0.99 2.16 0.031 Sex Age3 + Ancest3 f Ancest3: PIT 0.55 0.30 0.30 1.84 0.065 Sex Age3 + Het g Het 4.40 1.74 1.92 2.29 0.022 Sex Age3 + Ancest1 + Het Ancest1: PIT 0.04 0.37 0.34 0.13 0.895 Sex Age3 + Ancest1 + Het Ancest1: F1 1.80 1.04 1.01 1.78 0.075 Sex Age3 + Ancest1 + Het Het 2.61 2.25 2.33 1.12 0.263 Sex Age3 + Ancest2 + Het Ancest2: F1 1.81 1.04 1.01 1.79 0.074 Sex Age3 + Ancest2 + Het Het 2.76 1.86 2.01 1.38 0.168 Sex Age3 + Ancest3 + He t Ancest3: PIT 0.12 0.38 0.36 0.33 0.740 Sex Age3 + Ancest3 + Het Het 3.96 2.25 2.34 1.69 0.091 a Age3 differentiates sub adult (age 1 2.5 for females and 1 3.5 for males) and prime adult (ages 2.5 10 for females and 3.5 10 for males) panthers. b A ncest1 divides panthers into three ancestry categories: F 1 admixed, other admixed, and pre introgression type. c The covariate PIT indicates a difference in log hazard for pre introgression type panthers, compared with other admixed (Ancest1) or all other panthers (Ancest3). d The covariate F1 indicates a difference in log hazard for F 1 admixed panthers, compared with other admixed (Ancest1) or all other panthers (Ancest2). e Ancest2 divides panthers into two ancestry categories: F 1 admixed, and other admix ed and pre introgression type combined. f Ancest3 divides panthers into two ancestry categories: F 1 admixed and other admixed combined, and pre introgression type. g Het refers to individual average heterozygosity.

PAGE 154

154 Appendix E ANALYSIS OF FLORIDA PANTHER K ITTEN SURVIVAL BASED ON 3 ANCESTRY CATEGORIES Materials and Methods We used results from the STRUCTURE analysis along with pedigree results and field evidence to assign panthers to 3 groups that reflect the genetic makeup of the southern Florida population (canonical panthers, F 1 admixed panthers, and other admixed panthers). Canonical and F 1 admixed ancestry categories are defined in the main text. The other admixed category includes: backcrossed to canonical (admixed panthers with predominantly (>50%) c anonical heritage); backcrossed to Everglades (admixed panthers with predominantly (>50%) Everglades heritage resulting from breeding events between Texas admixed and Everglades panthers); backcrossed to Texas (admixed panthers of predominantly (>50%) Texa s heritage resulting from admixed panthers breeding with introduced Texas cougars); unknown admixed; and Seminole admixed panthers (these last two categories were not included in the analyses presented in the text). Based on these 3 ancestry categories, we formulated alternative hypotheses regarding the effect of ancestry on kitten survival ( Table E 1). We selected model 1 in Table 4 as the base model because this model had the lowest QAIC c among candidate models (Table 4), and used model selection to dra w inferences for hypotheses about the effects of ancestry on kitten survival ( Table E 1). For these analyses, we used a subset of data collected between 1995 and 2008, due to the absence of admixed panthers in Florida before 1995. We tested for the effec t of ancestry on survival of (1) kittens only; (2) sub adult and adult panthers only; (3) kittens and older panthers considered separately; and (4) panthers of all ages. Based on previous results (Benson et al., in revision) however, we only considered 2 ancest ry

PAGE 155

155 categories for sub adult and adult panthers (Ancest2; Table E 1). We report model averaged annual estimates of kitten survival probability (Burnham and Anderson, 2002) Results and Discussion Three models ( Base + k :Ancest1, Base + k :Ancest1 + s&a :Ancest2, and Base + k :Ancest2) failed to estimate a kitten survival parameter and were discarded. The most parsimonious model testing for the effect of simple ancestry on survival indicated that survival of panthers of all ages (includin g survival of kittens) was higher for F 1 admixed ( Table E 2, model 1). Although this model differed from the model where ancestry only affects the survival of sub adults and adults (model 3) c <2, the 95% confidence interval for the slope parameter did not include zero ( = 1.55; 95% CI: 0.46 2.65) providing strong evidence of a difference between the survival of F 1 admixed and other panthers. Model averaged estimated kitten surviv al increased from canonical to other admixed to F 1 admixed panthers (Figure E 1). These results suggest that although there are subtle differences in survival of kittens with various genetic ancestries, the main difference is that between F1 admixed kitten s and those of all other ancestry categories. F1 admixed kittens survived substantially better than kittens with other ancestry categories. This F1 admixed kitten survival advantage was also found in the ancestry analysis presented in the text. One dif ference in that analysis is that we found that some of the other admixed categories (specifically backcrossed to Texas and possibly backcrossed to Everglades) also had higher survivals than the canonical kittens. We believe that those results represent re al differences within the other admixed category, and are not an artifact of the exclusions required for that analysis.

PAGE 156

156 Table E 1. Definitions of simple ancestry models, and the hypotheses they represent. Term Definition Hypothesis Ancest1 3 ancestry c ategories: 1) Canonical; 2) Other Admixed; and 3) F 1 Admixed. Different survival probability for each ancestry group. Ancest2 2 ancestry categories: 1) Canonical and Other Admixed; and 2) F 1 Admixed. First generation survival advantage (or disadvantag e). Ancest3 2 ancestry categories: 1) Canonical; and 2) Other Admixed and F 1 Admixed. General admixed survival advantage (or disadvantage).

PAGE 157

157 Table E 2. Model selection table for testing hypotheses regarding the effect of genetic ancestry on survival of k ittens and older panthers based on 3 ancestry categories. These analyses were performed on a subset of the data (excluding panthers before 1995 and panthers of unknown ancestry). For each model, we present the number of parameters (K), the difference fro m and overdispersion ( c ), and the model QAIC c weight. All models estimate constant recapture rates and recovery rates that differ between kittens and older panthers ( p (.) r ( k s&a )). Model 1 in Table 2 was used as the base model for these analyses. Ancestry categories ar e defined in Table E 1. Survival Model K c Weight 1. Base + Ancest2 9 0.00 0.367 2. Base + Ancest1 10 1.15 0.207 3. Base + s&a :Ancest2 9 1.86 0.145 4. Base + k :Ancest2 + s&a :Ancest2 10 2.04 0.132 5. Base + k :Ancest3 + s&a :Ancest2 10 2.52 0.104 6 Base + Ancest3 9 5.81 0.020 7. Base 8 6.46 0.015 8. Base + k :Ancest3 9 6.91 0.012

PAGE 158

158 Figure E 1. Model averaged estimates of annual survival of Florida panther kittens based on 3 ancestry categories ( Table E 1). Error bars represent unconditional estimated standard errors.

PAGE 159

159 APPENDIX F DATA PREPARATION We used two datasets for the analysis of kitten survival. The first contained the capture (of live panthers) and recovery (of dead panthers) histories, observations of litter failures (i.e., loss of entire litters), and radio tracking data for panthers that were initially PIT tagged in the den between June 1995 and May 2008 (13 years). The second contained the capture, recovery, and radio tracking histories of all other panthers between June 1982 and May 2008 (26 years). We used a live dead data input format (Cooch and White, 2008) We coded the data with a 1 year time step. All litter failures were treated as recoveries within the first year, because all known litter failures occurred within the first tagged in the den, captures and radio failures were coded as occurring on the closest June 1, and known deaths were c oded as occurring in the June May interval in which they occurred. PIT tagging events of kittens in the den were also coded as occurring on the closest June 1. Subsequent captures, radio failures, and known deaths of that group were coded to happen at th e correct age, rather than the correct year, on the occasions when those were in conflict. Panthers that would have died if we had not removed them to captivity were treated as having died on the date of removal. One exception was a panther whose injuries were apparently capture related; it was right censored on the capture date and added again to the analysis on the date it was returned to the wild. Kittens that were found dead during the initial den visit (n=7) were not included in the survival analysis but were included in the litter size counts of PIT tagged kittens in their litters.

PAGE 160

160 The ages of 91 radio collared panthers not handled as kittens in the den were estimated as previously described; 11 were estimated to be < 1 year of age. When ages were estimated as a range in the field (e.g., 5 6 years old), we used the average of the range for assigning age class transitions. The influence of estimated age on the estimates of survival of adult panthers was insubstantial (Appendix A) ; this effect is lik ely even smaller for survival of kittens. The sex of two kittens from a single litter was not recorded in the field; since their fates were the same and the overall numbers of kittens PIT tagged for each sex were similar (119 females and 131 males), we randomly assigned one as a female and the other male. We did not have covariate data for all panthers included in the analyses. Thus, we tested for the effect of litter size and ancestry on survival separately after excluding kittens or all panthers, resp ectively, with missing values. For average heterozygosity, we ran a separate analysis excluding panthers that had not been genotyped, as well as captures of panthers that occurred before they were successfully genotyped. This approach removed the potenti al bias in survival estimates by heterozygosity that would be caused by only excluding those that were never genotyped (Nichols et al., 2004)

PAGE 161

161 APPENDIX G ANALYSIS OF SURVIVAL OF FLORIDA PANTHER K ITTENS USING DATA ORGANIZED IN 3 MONTH TIME STEPS Materials and Methods Using the data organized on an annual time scale, it was not possible to test for seasonal variation in kitten survival and changes in survival within the first year of life. Therefore, we recoded the data with 3 month time steps. This necessitated leaving out litter failure data, because th e 3 month interval in which actual kitten deaths occurred was generally not available from these data. The large number of capture occasions (26*4 = 104) made it computationally necessary to divide the sampling period in half. All data from panthers PIT tagged in the den were in the second half and could be handled normally, but all other panthers were only followed while radio collared to accommodate the half transition. The analyses of data organized in 3 month time step also allowed us to compare sub adult and adult survival with the results of Benson et al. (in revision) because we were now able to put the transition from sub adult to adult at the same age as in their analyses (2.5 years old for females and 3.5 years old for males). Analy sis of the data with a 3 month time step was similar to the annual time step analysis, except that there was now a possibility of recapture as a kitten. We set recapture probability ( p ) and recovery probability ( r ) for radio collared animals to 1, and p f or 3 month old panthers to 0, because none of the few panthers captured and radio collared at <6 months of age were PIT tagged first. The FWC and National Park Service (NPS) instituted a policy in November 2001 restricting the capturing and radio collarin g of kittens < 12 months in age. Therefore we fixed p at 0 for 6 month old kittens starting at that date.

PAGE 162

162 Data analysis proceeded as described in the text. We commenced by testing all a priori models for recapture and recovery combined with a priori model s for survival of kittens based on age (Table G 1). The kittens of Texas cougars may have been more likely to be captured than other kittens, due to greater efforts expended to capture those kittens (model 1, Table G 1A). The justification for the separa te recovery rate for kittens Table G 1B) is that a different series of events leads to recovery of kittens in the den (approximately the first 7 8 weeks, Maehr et al., 1990) compared to older panthers. For a kitten in the den to be recovered, it has to be PIT tagged, die, and then be found in a subsequ ent visit to the den. For an older (non radio collared) panther that was handled and marked with a PIT tag as a kitten to be recovered, it has to die (most likely due to a vehicle collision), and subsequently be reported to panther researchers. A cause s pecific mortality analysis of radio collared sub due to vehicular collision is independent of sex or age class (Benson et al., in revision) We tested all combinations of surviva l, recapture, and recovery models from Table G 1 and selected the most general model that converged successfully (if there was a tie, we selected the model with the lowest AIC c ). We used that model to test for overdispersion of the data due to lack of ind ependence within litters using a data bootstrap (Bishop et al., 2008) We sampled litters from the dataset with replacement. Panthers first captured as sub adults or adults were treated as litters of size 1 for the bootstrapping. We estimated annual survival of kittens as a derived parameter of the model. If the model failed to estimate any parameters, the iteration was discarded. We repeated the resampling and estimation until we had 1,000 estimates of kitten survival.

PAGE 163

163 We calculated the mean and standard deviation of the estimates, and estimated the ove rdispersion parameter ( ) as [(bootstrap standard deviation of estimates of kitten survival) / (standard error of kitten survival obtained from the original model run)] 2 An estimated > 1 indicates overd ispersion. We used the estimated to calculate the QAICc for each model and as a variance inflation factor. We selected the model or set of models with the lowest QAIC c as the base model(s) to test our hypothesis that kitten su rvival varies seasonally. We report model specific and model averaged 3 month and annual estimates of kitten survival probability and model averaged estimates of recapture, recovery, and sub adult and adult survival probabilities (Bur nham and Anderson, 2002) Results and Discussion Of the 72 models run (all permutations from Table G 1), 14 failed to estimate all parameters and were discarded, including all models using survival model 7 ( Table G 1C). From the most general models remai ning (with 13 parameters), the model S ( k <3, k <9, k p ( k :DO, s&a ) r ( k <3, k s&a ) had the lowest AIC c and was used in the litter bootstrap to estimate We estimated at 2.20, which is only slightly less than the average litter size of PIT tagged kittens used in the analysis (mean litter size = 2.47). There was strong support for kitten survival being lowest in the first 3 months of life ( Table G 2; main text Figure 1). Several closely ranked models yielded very different estimates of annu al kitten survival, and variances of most estimates were substantial ( Table G 2, Figure G 1). These results suggest that estimates of kitten survival obtained from data organized in 3 month time steps lacked precision and robustness, most likely due to sam ple size limitations. Among the top models, the main

PAGE 164

164 factor that affected estimated kitten survival was whether recovery of dead panthers differed between kittens in the first 3 months of life and other panthers (low estimates of survival, about 0.33) or not (high estimates of survival, about 0.50). The model averaged estimate of annual kitten survival from the 3 month analysis was 0.404 (95% CI: 0.169 0.693). The uncertainty of this estimate was high due to 1) small sample sizes and uncertainty about the fates of individual kittens leading to large standard errors for estimates from individual models; 2) further increase of these standard errors by the variance inflation factor, to account for overdispersion; and 3) different models often producing ver y different estimates of kitten survival, and the unconditional confidence interval presented here incorporates that uncertainty. Estimated annual survival rates of sub adults and adults were fairly consistent among models and the model averaged estimates were similar to those from the corresponding model in Benson et al. (in revision; Table G 3) The slightly lower estimate of sub adult female survival may be in part due to a death of a female panther near her estimat ed 1 st birthday that was categorized as a sub adult in the kitten analysis and as a kitten in the adult analysis. We selected two highly ranked models with differing estimates of kitten survival (models 1 & 2, Table G 2) as base models for testing the effe cts of season on kitten survival. There was no evidence of an effect of season on kitten survival ( Table G 4).

PAGE 165

165 Table G 1. A priori models used for the 3 month time step analysis A) Models for recapture probability ( p ). B) Models for recovery probabi lity ( r ) C) Models for survival probability (age models only; S ). All combinations of p r and S models in this table were run. We also tested for the effect of season (wet vs. dry) on S using the top model(s) from this analysis. For all models, p a nd r were fixed at 1 for radio collared panthers. We fixed p at 0 for kittens younger than 6 months old and for kittens younger than 9 months old after November 2001 for all models. All models estimated non kitten survival as the interaction of sex and a ge class (sub adult or adult); therefore, we only describe how kitten survival was modeled. A) Model for p Description 1. p ( k :DO, s&a ) Recapture probability differs between kittens and older panthers, and between kittens with dam origin of Florida and T exas. 2. p ( k s&a ) Recapture probability differs between kittens and older panthers. 3. p (.) Recapture probability is the same for all panthers. B) Model for r Description 4. r ( k <3, k s&a ) Recovery probability differs between kittens in their first three months, older kittens, and non kittens. 5. r ( k Recovery probability differs between kittens in their first three months and older panthers. 6. r (.) Recovery probability the same for all panthers. C) Model for S Description 7. S ( k k <6, k <9, k Age effect (survival differs between each of 4 3 month age intervals) 8. S ( k k <9, k Age effect (survival same for first two age intervals) 9. S ( k k <9, k Age effect (survival same for second two age intervals) 10. S ( k <6, k Age effect (survival same for first two age intervals, same for second two age intervals) 11. S ( k <3, k Age effect (survival same for last three age intervals) 12. S (AT) Age trend (survival over the first 4 3 month intervals as a linear trend) 13. S ( .) Survival same for all kittens

PAGE 166

166 Table G 2. Kitten survival models to examine the effects of age (3 month time step). For each model, we present the number of parameters (K), the difference from the top model in second order Quasi Criterion ( c ), the model QAIC c weight, and the estimate of annual kitten survival with standard error. See Table G 1 for descriptions of models. The base models selected are in bold. Model K c Weight Annual Kitten Survival Estimate (SE) 1. S ( k <3, k p ( k s&a ) r ( k 10 0 0.170 0.323 (0.090) 2. S ( k <3, k p ( k s&a ) r (.) 9 1.170 0.095 0.508 (0.136) 3. S ( k k <9, k p ( k s&a ) r ( k 11 1.361 0.086 0.326 (0.091) 4. S ( k <3, k p ( k s&a ) r ( k <3, k s&a ) 11 1.518 0.079 0.326 (0.092) 5. S (AT) p ( k s& a ) r (.) 9 1.914 0.065 0.484 (0.133) 6. S ( k <3, k p ( k :DO, s&a ) r ( k ) 11 1.991 0.063 0.326 (0.093) 7. S ( k k <9, k p ( k s&a ) r (.) 10 2.057 0.061 0.487 (0.133) 8. S ( k <3, k p ( k :DO, s&a ) r (.) 10 2.935 0.039 0.517 (0.136) 9. S ( k k <9, k p ( k s &a ) r ( k <3, k s&a ) 12 3.188 0.034 0.327 (0.093) 10. S ( k k <9, k p ( k :DO, s&a ) r ( k <3, 12 3.354 0.032 0.329 (0.095) 11. S (AT) p ( k s&a ) r ( k 10 3.454 0.030 0.399 (0.137) 12. S ( k <3, k p ( k :DO, s&a ) r ( k <3, k s&a ) 12 3.512 0.029 0.329 (0.096) 13. S (AT) p ( k :DO, s&a ) r (.) 10 3.717 0.026 0.493 (0.134) 14. S ( k k <9, k p ( k :DO, s&a ) r (.) 11 3.863 0.025 0.496 (0.134) 15. S (.) p ( k s&a ) r ( k ) 9 4.42 0.019 0.770 (0.137) 16. S (AT) p ( k s&a ) r ( k <3, k s&a ) 11 4.58 0.018 0.355 (0.111) 17. S ( k <3 k <9, k p ( k :DO, s&a ) r ( k <3, k s&a ) 13 5.18 0.013 0.330 (0.097) 18. S (AT) p ( k :DO, s&a ) r ( k ) 11 5.36 0.012 0.411 (0.148) 19. S ( k <6, k p ( k s&a ) r (.) 9 5.63 0.010 0.498 (0.138) 20. S ( k k <9, k p ( k s&a ) r (.) 10 5.67 0.010 0.481 (0.133) 21 S (.) p ( k :DO, s&a ) r ( k ) 10 5.98 0.009 0.775 (0.135) 22. S (.) p ( k s&a ) r ( k <3, k s&a ) 10 6.43 0.007 0.781 (0.267) 23. S (AT) p ( k :DO, s&a ) r ( k <3, k s&a ) 12 6.54 0.007 0.361 (0.118) 24. S (.) p ( k s&a ) r ( ) 8 7.12 0.005 0.590 (0.129) 25. S ( k k <9, k p ( k s&a ) r ( k <3, k s&a ) 12 7.22 0.005 0.400 (0.154) 26. S ( k <6, k p ( k :DO, s&a ) r (.) 10 7.40 0.004 0.507 (0.139) 27. S ( k k <9, k p ( k :DO, s&a ) r (.) 11 7.49 0.004 0.490 (0.135) 28. S ( k <6, k p ( k s&a ) r ( k <3, k s&a ) 11 7.74 0.004 0.404 (0.163) 29. S ( k <3, k p (.) r ( k 9 8.13 0.003 0.307 (0.076) 30. S (.) p ( k :DO, s&a ) r (.) 9 8.79 0.002 0.595 (0.128) 31. S ( k k <9, k p ( k :DO, s&a ) r ( k <3, k s&a ) 13 9.13 0.002 0.414 (0.172) 32. S ( k k <9, k p (.) r ( k 10 9.62 0.001 0.308 (0.077 )

PAGE 167

167 33. S ( k <6, k p ( k :DO, s&a ) r ( k <3, k s&a ) 12 9.62 0.001 0.422 (0.189) 34. S ( k <3, k p (.) r (.) 8 9.91 0.001 0.437 (0.118) 35. S ( k <3, k p (.) r ( k <3, k s&a ) 10 9.95 0.001 0.308 (0.077) 36. S (AT) p (.) r (.) 8 10.77 0.001 0.420 (0.113) 37. S ( k k <9 k p (.) r (.) 9 10.98 0.001 0.423 (0.113) 38. S ( k k <9, k p (.) r ( k <3, k s&a ) 11 11.57 0.001 0.308 (0.077) 39. S (AT) p (.) r ( k 9 11.97 0.000 0.355 (0.101) 40. S (AT) p (.) r ( k <3, k s&a ) 10 12.59 0.000 0.326 (0.086) 41. S ( k <6, k p (.) r (.) 8 14.13 0.000 0.428 (0.117) 42. S (.) p (.) r ( k 8 14.18 0.000 0.640 (0.184) 43. S ( k k <9, k p (.) r (.) 9 14.39 0.000 0.417 (0.112) 44. S (.) p (.) r ( k <3, k s&a ) 9 14.57 0.000 0.426 (0.144) 45. S ( k k <9, k p (.) r ( k <3, k s&a ) 11 14.99 0.000 0 .350 (0.099) 46. S (.) p (.) r (.) 7 15.09 0.000 0.487 (0.125) 47. S ( k <6, k p (.) r ( k <3, k s&a ) 10 15.21 0.000 0.350 (0.101) 48. S ( k <6, k p (.) r ( k 9 15.87 0.000 0.491 (0.218) 49. S ( k k <9, k p (.) r ( k 10 16.01 0.000 0.488 (0.211)

PAGE 168

168 Tab le G 3. Estimates of annual sub adult and adult survival from the 3 month kitten survival analysis and the adult analysis of Chapter 2 Sub adult is defined as ages 1 2.5 for females and 1 3.5 for males; adult is defined as ages 2.5 and up for females and 3.5 and up for males. The results from the kitten analysis are the model averaged survivals from all models in the 3 month analysis; the results from the adult analysis are from the Sex Age1 model (Chapter 2) Annual Survival Estimates (SE) Sex Age Class Kitten Analysis Adult Analysis Female Sub Adult 0.929 (0.051) 0.954 (0.032) Female Adult 0.858 (0.03) 0.852 (0.023) Male Sub Adult 0.723 (0.066) 0.712 (0.05) Male Adult 0.774 (0.049) 0.782 (0.035)

PAGE 169

169 Table G 4. Model comparison table for examini ng the effects of season on kitten survival. For each model, we present the number of parameters (K), the difference from the top model in second order Quasi Criterion ( c ), the model QAIC c weight, and the estimate of seasonal kitten survival with standard error. The same recapture model ( p ( k s&a ) ) was used in all models. See Table G 1 for descriptions of models. 6 Month Kitten Survival Model K c Weight Wet Season Dry Season 1. S ( k <3, k ) r ( k ) 10 0.00 0.469 0.568 (0.079) 2. S ( k <3, k ) r (.) 9 1.17 0.261 0.713 (0.096) 15. S ( k <3, k + season) r ( k <3, ) 11 2.01 0.172 0.564 (0.102) 0.572 (0.093) 16. S ( k <3, k + season) r (.) 10 3.14 0.098 0.736 (0 .144) 0.703 (0.107)

PAGE 170

170 A. B. Figure G 1. Individual model estimates of annual kitten survival A) Estimates f rom the 1 year time step analysis B) Estimates f rom the 3 month time step analysis. All models are shown in A, whereas only top model s (the difference from the top model in second order Quasi c < c value on the x axis. Error bars represent standard errors (adjusted by the variance inflation fa ctor).

PAGE 171

171 APPENDIX H INCLUDING ANOTHER BA SE MODEL Methods To examine the effects of base model uncertainty on covariate testing, we ran all covariates with the highest ranked model from Table 3 2 that had a different estimate of kitten survival than the base model (this was model 5). This model (hereafter called Base 2) modeled survival identically to the base model but estimated one recovery rate for all panthers instead of dividing it between kittens and older panthers. All covariate analyses were performed the same way as in the main text. We report full model average estimates based on all models run with both base models Results All covariate models using Base 2 could estimate all parameters. There was still no evidence that sex or litter size affecte d kitten survival (Table H 1). There was still considerable evidence that ancestry influenced kitten survival; the 7 most parsimonious models all included ancestry effects on kitten survival (models 1 7, Table H 2A). The top 2 models indicated that surviv al of kittens differed substantially between two ancestry categories: 1) canonical and backcrossed to canonical kittens and 2) F 1 backcrossed to Texas, and backcrossed to Everglades kittens (Table H 2A). Full model averaged kitten survival was lowest for canonical kittens and highest for F 1 kittens (Figure H 1A), and were slightly lower for all ancestry classes than model averages from the base model alone (Figure 3 1A). The most parsimonious model testing for the effect of heterozygosity on survival in dicated that survival of panthers of all ages (including survival of kittens) was

PAGE 172

172 positively affected by average heterozygosity (Table H 2B, model 1). Full model averaged estimated kitten survival increased with heterozygosity (Figure H 1B). The evidence f or the effect of abundance index on kitten survival was even stronger using Base 2 (Table H 2C). Full model averaged estimated kitten survival declined with abundance index (Figure H 2). We estimated temporal standard deviation of kitten survival ( ) at 0.199 (95% credible interval = 0.011 0.380) and 0.077 (95% credible interval = 0.003 0.234) using models with out and with abundance index, respectively. The most parsimonious model testing for the effect of simple ancestry on survival indicated that survival of panthers of all ages (including survival of kittens) was higher for F 1 admixed (Table H 3, model 1). Full model averaged estimated kitten survivals increased from canonical to other admixed to F 1 admixed (Figure H 3), and were slightly low er for all ancestry classes than model averages from the base model alone (Figure E 1). Interestingly, unconditional standard errors estimated for the full model set tended to be lower than those estimated for the base model set alone, even when the full m odel averaged estimate was closer to 0.5.

PAGE 173

173 Table H 1. Model selection table for sex and litter size effect s on kitten survival using Base2 model. A) Sex models. B ) L itter size models For each model, we present the number of parameters (K), the diffe rence from the top model in size ( c ), the model QAIC c weight, and the estimates of annual kitten survival with standard error (for both sexes and the highest and lowest litter sizes). The QAIC c values of the top models are 857.23 and 845.66, res pectively (compare to QAIC c values for the base models run with the same datasets of 856.76 and 845.64 respectively) See Tab le 3 1 for descriptions of the covariate models and Table 3 2 (model 5) for the Base 2 model. The litter size analysis was performed on a subset of the data (excluding kittens from litters of unknown size). All models estimate constant recapture and recovery rates ( p (.) r (.)). A. Sex models Annual Kitten Survival Estimate (SE) Survival Model K c Weight Female Male 1. Base 2 7 0.00 0.725 0.296 (0.054) 2. Base 2 + k :sex 8 1.94 0.275 0.311 (0.068) 0.282 (0.065) B. Litter size models Annual Kitten Survival Estimate (SE) Survival Model K c Weight 1 kitten 4 kittens 1. Base 2 7 0.00 0 .735 0.280 (0.050) 2. Base 2 + k :LS 8 2.04 0.265 0.279 (0.109) 0.280 (0.081)

PAGE 174

174 Table H 2. Model selection table for ancestry, heterozygosity, and abundance index effect s on kitten survival using Base2 model. A) A ncestry effec ts on kitten and older sur vival. B) H eterozygosity effec ts on kitten and older survival. C ) A bundance index effects on kitten survival. These analyses were performed on subsets of the data (excluding panthers before 1995 for all analyses and panthers of unknown ancestry for A, a nd panthers before their heterozygosity was successfully sampled for B). For each model, we present the number of Criterion adjusted for overdispersion and small sample size ( c ), and the model QAIC c weight. The QAIC c values of the top models are 623.98, 577.80, and 716.22, respectively (compare to QAIC c values for the top models of Table 3 of 624.41, 577.37, and 716.46 respectively) See Table 1 for descriptions of the covaria te models and Table 2 (model 5) for the Base 2 model. All models estimate constant recapture and recovery rates ( p (.) r (.)). A. Ancestry models Survival Model K c Weight 1. Base 2 + k :GenAnc2 + s&a :GenAnc6 9 0.00 0.232 2. Base 2 + GenAnc2 8 1.18 0.128 3. B ase2 + GenAnc6 8 1.35 0.118 4. B ase2 + GenAnc4 9 1.83 0.093 5. Base 2 + k :GenAnc3 + s&a :GenAnc6 9 2.07 0.082 6. Base 2 + k :GenAnc6 + s&a :GenAnc6 9 3.20 0. 047 7. Base 2 + GenAnc3 8 3.21 0.047 8. Base 2 + s&a :GenAnc6 8 3.23 0.046 9. Base 2 + k :GenAnc1 + s&a :GenAnc6 9 3.71 0.036 10. Base 2 + k :GenAnc5 + s&a :GenAnc6 10 3.84 0.034 11. Base 2 + k :GenAnc2 8 3.85 0.034 12. Base 2 + k :GenAnc4 + s&a :GenAnc6 10 4.29 0 .027 13. Base 2 + GenAnc5 9 4.69 0.022 14. Base 2 + GenAnc1 8 5.55 0.014 15. Base 2 + k :GenAnc3 8 5.94 0.012 16. Base 2 + k :GenAnc6 8 6.44 0.009 17. Base 2 + k :GenAnc4 9 7.53 0.005 18. Base 2 7 7.54 0.005 19. Base 2 + k :GenAnc5 9 7.72 0.005 20. Base 2 + k : GenAnc1 8 7.93 0.004 B. Heterozygosity models Survival Model K Weight 1. Base 2 + Het 8 0.00 0.320 2. Base 2 + s&a :Het 8 0.53 0.245 3. Base 2 7 0.78 0.216 4. Base 2 + k :Het + s&a :Het 9 2.07 0.114

PAGE 175

175 5. Base 2 + k :Het 8 2.22 0.105 C. Abundance in dex models Survival Model K Weight 1. Base 2 + k :Abundance 8 0.00 0.992 2. Base 2 7 9.65 0.008

PAGE 176

176 Table H 3. Model selection table for three category ancestry effects on kitten and older survival using the Base2 model These analyses were per formed on subsets of the data (excluding panthers before 1995 and panthers of unknown ancestry). For each model, we present the number of parameters (K), the overdispersion and s mall sample size ( c ), and the model QAIC c weight. The QAIC c value of the top model is 716.21 (compare to QAIC c value for the top model of Table A2 of 715.74) See Table A1 for descriptions of the covariate models and Table 2 (model 5) for the Base 2 model. All models estimate constant recapture and recovery rates ( p (.) r (.)). Survival Model K Weight 1. Base 2 + Ancest2 8 0.00 0.321 2. Base 2 + Ancest1 9 1.04 0.191 3. Base 2 + s&a :Ancest2 8 1.68 0.138 4. Base 2 + k :Ancest2 + s&a :Ancest2 9 1.77 0.133 5. Base 2 + k :Ancest3 + s&a :Ancest2 9 2.68 0.084 6. Base 2 + k :Ancest1 + s&a :Ancest2 10 3.19 0.065 7. Base 2 + k :Ancest2 8 5.50 0.020 8. Base 2 + Ancest3 8 5.74 0.018 9. Base 2 7 6.58 0.012 10. Base 2 + k :Ancest1 9 6.92 0.010 11. Base 2 + k :Ancest3 8 7.50 0.008

PAGE 177

177 A B. Figure H 1. The effect of genetic variables on model ave raged kitten survival estimates, including two different base models (Base and Base2) A) Annual survival plotted against ancestry category. B) Annual survival plotted against heterozygosi ty. Error bars represent unconditional estimated standard errors.

PAGE 178

178 Figure H 2. The effect of a panther abundance index on model averaged kitten survival estimates including two different base models (Base and Base2) Error bars represent unconditional estimated standard errors.

PAGE 179

179 Figure H 3. Model averaged estimates of annual survival of Florida panther kittens based on 3 ancestry categories including two different base models (Base and Base2) Error bars represent unconditional estimated standar d errors.

PAGE 180

180 APPENDIX I ANALYSIS OF THE INFLUENCE OF ANCESTRY USING FINE SCALE ANCESTRY TESTING Methods In order to investigate potential variation in reproductive rates across ancestry categories based on finer scale genotypic distinctions than those used in the body of the paper, we placed female s from our radio collared sample into 7 categories: canonical Florida panthers ( n = 22), Everglades panthers ( n = 2), F 1 admixed panthers ( n = 6), backcrossed to canonical admixed panthers ( n = 6), backcrossed to Eve rglades admixed panthers ( n = 3), backcrossed to Texas admixed panthers ( n = 11), and unknown admixed panthers ( n = 8). Canonical Florida panthers showed no evidence of non Florida genetic admixture (no direct non canonical relatives or < 10% non canonic al genetic contribution based on Structure analyses). Everglades panthers were mostly captured within or adjacent to Everglades National Park and had documented genetic links to Central America (Culver et al., 2000; O'Brien et al., 1990) Note that canonical and Everglades panthers were combine d into 1 category (pre introgression type panthers) for the ancestry analysis in the body of the paper which represents the known genotypes present on the south Florida landscape prior to the introgression program. We defined F 1 admixed panthers (as befor e) as any offspring produced by a mating between a Texas female and a pre introgression type male. Backcrossed to Texas admixed panthers were admixed panthers whose admixture proportions from the Structure analysis were >50% Texas. Backcrossed to canonic al panthers were admixed panthers whose admixture proportions were >50% but <90% canonical. Backcrossed to Everglades were admixed panthers were panthers with documented Texas admixture and admixture proportions of >50% Everglades. Unknown admixed

PAGE 181

181 panthe rs were excluded (in addition to 1 unknown ancestry panther) because we were missing information needed to place them into the finer admixture categories (e.g., 1 or both parents were not identified). Note that backcrossed to Texas, backcrossed to Canoni cal, backcrossed to Everglades, and unknown admixed were all combined into the other admixed category in the ancestry analysis in the body of the paper. We used ancestry models (GenAnc1 GenAnc6; Table I 1) with different combinations of ancestry categori es to test specific hypotheses. We used the minimum AIC model from the age and season analysis as a base model to test for the effects of ancestry on reproductive rates. We tested for the effects of ancestry on both p and additively with the base models. Although panthers in six ancestry categories were used in the analysis, the sample sizes for Everglades and backcrossed to Everglades were quite small, so we only present model averaged estimates for the other four ance stry categories. Results and Discussion The most parsimonious models for ancestry separated F 1 and backcrossed to Texas panthers from others for both probability of breeding and litter size (Model 1, Table I 2; Model 1, Table I 3). Model averaged probabil ity of breeding was highest for backcrossed to canonical panthers, intermediate for canonical panthers, and lowest for backcrossed to Texas and F 1 panthers (Figure I 1A). Model averaged litter size was lowest for canonical panthers and highest for backcro ssed to Texas and F 1 panthers, although these differences are less dramatic (Figure I 1B). These results corroborate and extend the ancestry results from the main text. In addition to F 1 panthers, backcrossed to Texas panthers have lower probabilities of reproduction than canonical panthers. Two of the proposed explanations in the main

PAGE 182

182 text discussion (outbreeding depression and increased kitten survival) are compatible with these results as well. The other explanation (chance) seems less likely, as the combined F 1 and backcrossed to Texas sample size ( n = 17 panthers) is more substantial than the F 1 sample size alone ( n = 6 panthers).

PAGE 183

183 Table I 1. Definitions of complex ancestry models, and the hypotheses they represent. Term Definition Hypothesis Ge nAnc1 2 ancestry categories: 1) Canonical and Everglades; 2) Backcrossed to Canonical, Backcrossed to Everglades, Backcrossed to Texas, and F 1 General introgression advantage (or disadvantage). GenAnc2 2 ancestry categories: 1) Canonical and Backcrosse d to Canonical; 2) Everglades, Backcrossed to Everglades, Backcrossed to Texas, and F 1 General admixture advantage (or disadvantage), but with Backcrossed to Canonical reverting to Canonical. GenAnc3 2 ancestry categories: 1) Canonical, Everglades, Bac kcrossed to Canonical, and Backcrossed to Everglades; 2) Backcrossed to Texas, and F 1 Reproduction advantage (or disadvantage) for those GenAnc4 3 ancestry categories: 1) Canonical and Everglades; 2) Backcrossed to Canonical, Backcrossed to Everglades, and Backcrossed to Texas; and 3) F 1 Differences in reproduction between pre introgression types, backcrosses, and F 1 GenAnc5 3 ancestry categories: 1) Canonical and Everglades; 2) Backcrossed to Canonical and Backcrossed to Everglades; 3) Backcrossed to Texas and F 1 Differences in reproduction between pre introgression intermediate types. GenAnc6 2 ancestry categories: 1) Canonical, Everglades, Backcrossed to Canonical, Backcrossed to Everglades, and Backcrossed to Texas; and 2) F 1 Reproduction advantage (or disadvantage) for F 1

PAGE 184

184 Table I 2 Model comparison table testing for the effects of complex ancestry (see Table I 1) on probability of breeding ( p ). The base model used was p (Age2 + Season) (Age2 distinguishes between young adults and older adults). For each model, we present the numb er of parameters (K), the difference in w i ) This analysis was performed on a subset of the data (excluding 1 panther with unknown ancestry and unknown admixed panthers). Model p K w i 1 Age2 + Season + GenAnc3 4 0.00 0.308 2 Age2 + Season + GenAnc5 5 0.51 0.239 3 Age2 + Season + GenAnc6 4 1.03 0.185 4 Age2 + Season + GenAnc2 4 1.43 0.151 5 Age2 + Season + GenAnc4 5 2.98 0.070 6 Age2 + Season 3 4.50 0.032 7 Age2 + Season + GenAnc1 4 6.05 0.015

PAGE 185

185 Table I 3. Model comparison table testing for t he effects of complex ancestry (see Table I 1) on litter size ( ). The base model used was (Age3 + Season) (Age3 divides panthers into sub adults and adults). For each model, we Criterion ( w i ) This analysis was performed on a subset of the data (excluding unknown admixed panthers). Model K w i 1 Age3 + Season + GenAnc3 6 0.00 0.252 2 Age3 + Season 5 0.29 0.218 3 Age3 + Season + GenAnc1 6 0.60 0.187 4 Age3 + Season + GenAnc5 7 1.66 0.110 5 Age3 + Season + GenAnc6 6 2.29 0.080 6 Age3 + Season + GenAnc2 6 2.29 0.080 7 Age3 + Season + GenAnc4 7 2.50 0.072

PAGE 186

186 Figure I 1. Model averaged reproductive rates by complex ancestry category ( unconditional standard errors). A) Probability of a young adult female panther giving birth within the 6 month dry season and B) average litter size for an adult female panther during the dry season. Although panthers in six ancestry categories were used in the analysis, the sample sizes for Everglades a nd backcrossed to Everglades were quite small, so we only present model averaged estimates for the other four ancestry categories

PAGE 187

187 APPENDIX J ANALYSIS OF REPRODUC TIVE PARAMETERS EXCL UDING OLDER YEARS) Methods We tested for the effects of age c lass and season on reproductive rates, excluding older adults. We did this to check the robustness of our estimates of reproductive parameters for sub adults and prime adults to distortion from older adults, which only bred once (with a litter size of 1). There were no older adult female admixed panthers in our dataset, so we also wanted to check the robustness of our ancestry statistical inference and estimates to the inclusion of the pre introgression type older adults. To consider the effects of age c lass on p and we created a new age class model: Age4, which divides sub adult (ages 1 2.5) and prime adult (ages 2.5 10) panthers. Otherwise, analyses proceeded as in the main text. Results Excluding older adults had little effect on inference (Ta ble J 1) or mod el averaged estimates (Figures J 1A, J 2A, and J 3A) for breeding probability of sub adults and prime adults. Excluding the one older adult that bred (instead of combining her with prime adults in the adult category) had some sma ll effects on inference (Table J 2) and mod el averaged estimates (Figures J 1B, J 2B, and J 3B) for including moving season from a marginally supported model (Table 4 2B) to a marginally un supported model (Table J 2B).

PAGE 188

188 Table J 1 Model selection table evaluating the effects of factors on probability of breeding ( p ), excluding older adults A) A ge class effects. B) S eason effects. C) A ncestry effects The top ranked model for age class (part A, model 1, in bold) was used as a base model to test for the effects of season; the top ranked model for season (part B, model 1, in bold) was used as a bas e model to test for the effects of ancestry. For each model, we present the number of the Akaike weight ( w i ) These analyses were performed on subsets of the data (excluding older adults for all analyses and 1 panther with unknown ancestry for ancestry analysis). Model p K w i A) Age Models 1 Constant a 1 0.00 0. 562 2 Age 4 b 3 0.50 0. 4 38 B) Season Models 1 Season 2 0.00 0. 785 2 Constant 1 2. 59 0. 215 C ) Ancestry Models 1 Season + Ancest2 c 3 0.00 0. 628 2 Season + Ancest1 d 4 1.5 9 0. 284 3 Season 2 4.5 6 0.0 64 4 Season + Ancest3 e 3 6.4 9 0.0 24

PAGE 189

189 Table J 2. Model selection table evaluating the effects of factors on litter size ( ), excluding older adults. A) A ge class effects. B) S eason effects. C) A ncestry effects The top ranked model for age class (part A, model 1, in bold) was used as a base model to test for the effects of season; the top ranked model for season (part B, model 1, in bold) was used as a base model to test for the effects of ancestry. For each model, we present the number of parameters weight ( w i ) These analyses were performed o n subsets of the data (excluding older adults for all analyses and 1 panther with unknown ancestry for ancestry). Model K w i A) Age Models 1 Age 4 a 4 0.00 0. 777 2 Constant b 3 2.50 0. 223 B) Season Models 1 Age 4 4 0.00 0. 519 2 Age 4 + Season 5 0 16 0. 481 C ) Ancestry Models 1 Age4 4 0 00 0. 478 2 Age 4 + Ancest3 c 5 1 29 0. 251 3 Age4 + Ancest2 d 5 2 00 0. 176 4 Age 4 + Ancest1 e 6 3.24 0. 095

PAGE 190

190 Figure J 1. Model averaged reproductive rates ( unconditional standard errors) by age class. Sub adults are defined as ages 1 2.5 and prime adults ages 2.5 10. We present A) six month probability of breeding and B) average litter size. The estimates are from an analysis excluding older panthers.

PAGE 191

191 Figure J 2. Model averaged reproductive rates ( unconditional standard errors) by season. We present A) six month probability of breeding for young adult pan thers (<10 years old) and B) average litter size for prime adult panthers (ages 2.5 10). The estimates are from an analysis excluding older adult (age

PAGE 192

192 Figure J 3. Model averaged reproductive rates ( unconditional standard erro rs) by ancestry category. We present A) six month probability of breeding for young adult panthers (<10 years old) in the dry season and B) average litter size, for prime adult panthers (ages 2.5 10). The estimates are from an analysis excluding older ad

PAGE 193

193 APPENDIX K ANNUAL BREEDING PARA METERS FOR PVA Methods For use in population viability analyses (PVA) we estimated and modeled reproductive parameters similar to the methods detailed in the main text, with five key differences : 1) we organized data with an annual time step and estimated annual reproductive rates; 2) we did not include kittens that were not PIT tagged in the den in reproductive counts; 3) we used all age models for examining the effects of abundance index, ances try, and heterozygosity on reproductive parameters instead of choosing base models; and 4) we estimated random effects of year on reproductive rates as measures of temporal variance (environmental stochasticity). For use in PVA with annual time steps, we e stimated and modeled annual probability of breeding ( q ) and annual cumulative number of kittens ( ; see below). We organized the data with years extending from June May. This precluded examining the effects of season on reproductive rates. Because the re were instances of female panthers giving birth more than once in a year, q is the probability of giving birth at least once during a year and is the total predicted number of kittens produced from all breeding attempts within a year. The range of obs erved was 1 6, but = 5 was not observed. We therefore modeled with J = 5 categories, with N j = j for j = 1 4 and N 5 = 6 Otherwise, q was modeled similarly to p, and was modeled similarly to Most kittens included in the reproductive analy ses in the main text were found and PIT tagged during den visits soon after parturition. However, some were found dead during den checks and a few were missed in initial den checks and documented later. Florida panther kitten survival estimates for use i n PVA were made conditional on

PAGE 194

194 kittens being PIT tagged in the den (Hostetler et al., 2010) To avoid bias in population growth rates from PVA, we defined reproduction as giving birth to kittens that were PIT tagged in the den and excluded kittens from analyses in this appendix that were not PIT tagged in the den. However, we also present overall annual reproductive rates without this exclusion, for comparison with other studies. Stepwise model selection (choosing a base model from testing one covariate to test the next covariate, as used in the main text) can be a useful technique for reducing the size of model sets used with information theoretic inference. However, the re are trade offs involved with this approach. When there are two or more closely ranked models in a model set and only one of these is chosen as a base model, subsequent model averages will not include this (possibly important) source of uncertainty. To averaged estimates that also included abundance index, ancestry, or heterozygosity, we included all age models in these model sets (in all combinations with the other models). F inally, we examined the random effects of year (June May, 1 13) on q and as a measure of environmental stochasticity. The minimum AIC models from the age analysis for each parameter were used as base models, with and without abundance index as a cova riate. We modeled temporal variance in q by using the logistic model including random effects for year, which is a type of generalized linear mixed model (GLMM; Bolker et al., 2009) Again, we used the complementary log log link (equation 1.4) and log( m ) as an offset. We implemented the mixed logistic model using the R function lmer in the package lme4. We modeled temporal variance by using the cumulative logit including random effects (Min and Agresti, 2005) We implemented the

PAGE 195

195 mixed cumulative logistic model using the R function clmm in the package ordinal. Because our intent was to estimate the temporal variance rather than produce an estimate of probability of breeding for each year, we counted the random effect as a single parameter for model comparison (Vaida and Blanchard, 2005) Results The overall annual probability of breeding was 0.406 0.034, or 0.410 0.034 including reproduction that did not lead to kittens being PIT tagged in the de n. The overall annual cumulative number of kittens was 2.76 0.12, or 2.84 0.11 including kittens not PIT tagged in the den. The model selection results (Tables K 1 and K 2) were similar to those in the main text. Model averaged annual probabilities of breeding were 0.430 0.066 for sub adults, 0.466 0.039 for prime adults, and 0.038 0.037 for older adults. Model averaged annual cumulative numbers of kittens were 2.96 0.28 for sub adults and 2.72 0.13 for adults. Models including the quadratic effect of abundance on failed to converge and were discarded. There was strong evidence for temporal variance in annual breeding probability, with or without abundance index (estimated at 0.123 and 0.138, respectively; Table K 3). There was no evidence for temporal variance in annual cumulative number of kittens (without abundance index) ( Table K 4). The model with both temporal variance and abundance index effects on annual cumulative number of kittens failed to converge and was discarded.

PAGE 196

196 Table K 1 Model selection table evaluating the effects of factors on annual probability of breeding ( q ). A) A ge class effects. B) A bundance index effects. C) Ancestry effects. D) H eterozygosity effects For each model, we present the number of parameters (K) ( w i ) The ancestry and heterozygosity analyses were performed on subsets of the data (excluding one panther with unknown heterozygosity). Model q K w i A) Age Models 1 Age2 a 2 0.00 0.620 2 Age1 b 3 0.98 0.380 3 Constant c 1 19.77 0.000 4 Age3 d 2 21.67 0.000 B) Abundance Index Models 1 Age2 + Abundance 3 0.00 0.304 2 Age2 2 0.75 0.209 3 Age2 + Abundance + Abundance 2 4 1.39 0.152 4 Age1 + Abundance 4 1.53 0.141 5 Age1 3 1.73 0.128 6 Age1 + Abundance + Abundance 2 5 3.06 0.066 7 Abundance 2 14.22 0.000 8 Abundance + Abundance 2 3 16.09 0.000 9 Age3 + Abundance 3 16.22 0.000 10 Age3 + Abundance + Abundance 2 4 18.08 0.000 11 Constant 1 20.53 0.000 12 Age3 2 22.43 0.000 C) Ancestry Models 1 Age2 + Ancest2 e 3 0.00 0.295 2 Age2 + Ancest1 f 4 0.48 0.232 3 Age1 + Ancest2 4 0.92 0.186 4 Age1 + Ancest1 5 1.13 0.168 5 Age2 2 3.44 0.053 6 Age1 3 4.40 0.033 7 Age2 + Ancest3 g 3 5.32 0.021 8 Age1 + Ancest3 4 6.22 0.013 9 Ancest1 3 16.21 0.000 10 Age3 + Ancest1 4 17.61 0.000 11 Ancest3 2 21.05 0.000 12 Ancest2 2 22.51 0.000 13 Age3 + Ancest3 3 22.62 0.000 14 Constant 1 23.26 0.000

PAGE 197

197 15 Age3 + Ancest2 3 24.43 0.000 16 Age3 2 25.15 0.000 D) Heterozygosity Models 1 Age2 + Het h 3 0.00 0.315 2 Age2 2 0.03 0.310 3 Age1 3 1.00 0.192 4 Age1 + Het 4 1.09 0.183 5 Constant 1 19.85 0.000 6 Age3 2 21.75 0.000 7 Het 2 21.82 0.000 8 Age3 + Het 3 23.73 0.000 a. Age 2 divides panthers into young adults and older adults. b. Age1 divides panthers into sub adults, prime adults, and older adults. c. No predictor variables. d. Age3 divides panthers into sub adults and adults. e. Ancest2 divides panthers into 2 ancestry categories: F 1 admixed, and other admixed and pre introgression type combined. f. Ancest1 divides panthers into 3 ancestry categories: pre introgression type, other admixed, and F 1 admixed. g. Ancest3 divides panthers into 2 ancestry categories: pre introgression type and admi xed. h. Heterozygosity.

PAGE 198

198 Table K 2. Model selection table evaluating the effects of factors on annual cumulative number of kittens ( ) A) A ge class effects. B) A bundance index effects. C) A ncestry effects. D) H eterozygosity effects For each model, we present the ( w i ) Model K w i A) Age Class Models 1 Age3 a 5 0.00 0.537 2 Constant b 4 0.30 0.463 B) Abundance Index Models 1 Age3 5 0.00 0.368 2 Constant 4 0.30 0.318 3 Age3 + Abundance 6 1.46 0.177 4 Abundance 5 1.98 0.137 C) Ancestry Models 1 Ag e3 5 0.00 0.270 2 Constant 4 0.30 0.233 3 Age3 + Ancest2 c 6 1.64 0.119 4 Ancest2 5 1.92 0.103 5 Age3 + Ancest3 d 6 1.98 0.100 6 Ancest3 5 2.23 0.089 7 Age3 + Ancest1 e 7 3.55 0.046 8 Ancest1 6 3.74 0.042 D) Heterozygosity Models 1 Age3 5 0.00 0.374 2 Constant 4 0.30 0.323 3 Age3 + Het f 6 1.75 0.156 4 Het 5 1.88 0.146 a Age3 divides panthers into sub adults and adults. b No predictor variables. c Ancest2 divides panthers into 2 ancestry categories: F 1 admixed, and other admixed and pre introgression type combined. d Ancest3 divides panthers into 2 ancestry categories: pre introgression type and admixed. e Ancest1 divides panthers into 3 ancestry categories: pre introgression type, other admixed, and F 1 admixed. f Heterozygosity.

PAGE 199

199 T able K 3. Model sel ection table testing for the random effect of year (rand(Year)), with and without abundance index as a covariate, on annual probability of breeding ( q ) Age2 divides panthers into young adults and older adults. The top ranked model fo r age (Table K 1A, model 1, in bold) was used as a base model. For each model, we present the number of parameters (K), the w i ), and temporal standard deviation on the complementary l og log scale and real scale Standard deviation was converted to the real scale for young adults (model 1), and young adults with mean abundance (model 2). Temporal SD Model p K w i Complementary Log Log Scale Real Scale 1 Age2 + rand(Year) 3 0.00 0. 449 0. 420 0. 138 2 Age2 + Abundance + rand(Year) 4 0. 63 0. 328 0. 3 7 8 0.123 3 Age2 + Abundance 3 2 .4 5 0. 132 4 Age2 2 3 .2 1 0. 090

PAGE 200

200 Table K 4. Model sel ection table testing for the random effect of year (rand(Year)) on annual cumulative nu mber of kittens ( ) Age3 divides panthers into sub adults and adults. The top ranked model for age (Table K 2A, model 1, in bold) was used as a base model. For each model, we present the number of the Akaike weight ( w i ), and temporal standard deviation on the cumulative logit scale and real scale Standard deviation was converted to the real scale for adults (model 2). Temporal SD Model K w i C umulative Logit Scale Real Scale 1 Age 3 5 0.00 0. 731 2 Age 3 + rand(Year) 6 2 00 0. 269 0. 0006 0.0004

PAGE 201

201 APPENDIX L DEMOGRAPHIC PARAMETERS FOR CANON ICAL PANTHERS Materials and Methods To address questions regarding the dynamics and persistence of a hypothetical post 1995 canonical population we needed estimates of demographic parameters for canonical panthers. Model fitting procedures followed those from the original ancestry analyses with a few differences. We ran additional ancestry models for each parameter to account for factors left ou t of the ancestry analyses (Benson et al., in revision; Hostetler et al., 2010; chapter 4) such as abundance index. In addition, we included older adults in the new sub adult and adult survival analyses (unlike in Benson et al., in revision) Originally, they were left out because there were few older adult admixed panthers in the period covered (1997 2006). For this analysis, we needed model specific estimates, v ariances, and covariances for older adult survival, so we included them, but applied ancestry effects to young adults only. From each of these models, we obtained estimates of parameter values for canonical panthers. However, we used estimates of t from the equivalent non c ) < 10. Only density independent models were used in estimation of (c) c ) < 10 were used in estimation of extinction parameters. Resul ts The top ancestry model for sub adult and adult survival contained terms for Ancest1 (survival varying by three ancestry categories) and abundance (Table L 1), even though the original top ancestry model contained Ancest2 instead (the equivalent was the second ranked model here) and the top model from the original abundance

PAGE 202

202 analysis did not contain abundance (Benson et al., in revision) Similarly, the inclusion of abundance index changed the top kitten survival model from having the AdmAdv model (backcrossed to canonical and canonical have lower survivals; Hostetler et al., 2010) to having the IntAd v model (only canonical kittens have lower survival; Table L 2). Because no density independent kitten survival models were within 10 QAIC c of the top model, we present those models separately (for estimation of (c) ; Table L 3). The top model for prob ability of breeding included Ancest2 and abundance (Table L 4). The top model for annual kitten production contained neither ancestry nor abundance effects (Table L 5).

PAGE 203

203 Table L 1. Models for canonical sub adult and adult survival, with Akaike model weigh ts ( w i ) estimates of sub adult female canonical survival ( S sa (c) ), prime adult female canonical survival ( S pa (c) ), older adult female canonical survival ( S oa (c) ), the effect of abundance index on log hazard ( abundance ), and temporal standard deviation ( t ). Density independent models are in bold. Survival estimates for density dependent models are for abundance index 0. Age category model names are different than in Benson et al. (in revision) Model w i S sa (c) S pa (c) S oa (c) abundance t Sex Age3 a + Older b + Young c :Ancest1 d + Abundance e 0.195 0.983 0.018 0.937 0.039 0.868 0.076 0.017 0.008 0.306 Sex Age3 + Older + Young:Ancest2 f 0.155 0.960 0.040 0.832 0.035 0.668 0.067 0 0.3 40 Sex Age3 + Older + Young:Ancest2 + Abundance 0.105 0.978 0.024 0.909 0.054 0.786 0.100 0.009 0.007 0.306 Sex Age1 g + Young:Ancest1 + Abundance 0.091 0.983 0.018 0.934 0.041 0.880 0.072 0.017 0.009 0.305 Sex Age3 + Older + Young: Ancest1 0.085 0.953 0.048 0.812 0.044 0.666 0.068 0 0.340 Sex Age1 + Young:Ancest2 0.069 0.960 0.040 0.825 0.037 0.691 0.067 0 0.339 Sex Age3 + Young:Ancest2 0.052 0.960 0.040 0.798 0.034 0.798 0.034 0 0.253 Sex Age1 + Young:Ance st2 + Abundance 0.046 0.978 0.024 0.905 0.057 0.801 0.096 0.008 0.008 0.305 Sex Age3 + Older + Young:Ancest3 h + Abundance 0.044 0.986 0.015 0.951 0.029 0.891 0.062 0.021 0.008 0.306 Sex Age1 + Young:Ancest1 0.039 0.953 0.049 0.803 0.047 0.691 0.068 0 0.339 Sex Age3 + Young:Ancest2 + Abundance 0.021 0.968 0.034 0.837 0.079 0.837 0.079 0.003 0.007 0.239 Sex Age1 + Young:Ancest3 + Abundance 0.021 0.986 0.015 0.949 0.031 0.901 0.058 0.020 0.008 0.305 Sex Age3 + Young:Ancest1 0.019 0.959 0.042 0.795 0.045 0.799 0.039 0 0.253 Sex + Young:Ancest2 0.011 0.824 0.030 0.824 0.030 0.824 0.03 0 0.224 Sex Age3 + Young:Ancest1 + Abundance 0.008 0.969 0.033 0.842 0.078 0.856 0.085 0.005 0.008 0.239 Sex Age3 + Older + Young:Ancest3 0.008 0.954 0.048 0.818 0.045 0.665 0.068 0 0.340 Sex + Young:Ancest2 + Abundance 0.005 0.865 0.067 0.865 0.067 0.865 0.067 0.004 0.007 0.210 Sex + Young:Ancest1 0.004 0.826 0.040 0.826 0.040 0.822 0.035 0 0.224 Sex Age3 + Older 0.004 0.967 0.033 0.861 0.029 0.669 0.067 0 0.340 Young:Ancest2 0.004 0.780 0.026 0.780 0.026 0.780 0.026 0 0.177 Sex Age1 + Young:Ancest3 0.004 0.953 0.048 0.810 0.048 0.691 0.068 0 0.339 Sex Age 3 + Older + Abundance 0.003 0.982 0.019 0.928 0.043 0.792 0.099 0.009 0.008 0.306 Sex + Young:Ancest1 + Abundance 0.002 0.866 0.067 0.866 0.067 0.872 0.077 0.005 0.008 0.210 Sex Age1 0.002 0.967 0.033 0.855 0.031 0.691 0.067 0 0.3 39 Young:Ancest2 + Abundance 0.002 0.831 0.085 0.831 0.085 0.831 0.085 0.004 0.007 0.167 Young:Ancest1 0.002 0.774 0.040 0.774 0.040 0.784 0.035 0 0.177

PAGE 204

204 a Age 3 divides panthers into sub adults (1 2.5 and 1 3.5 years old for females and mal and males respectively). b Older refers to older c Young refers to young adults (1 10 years old). d Ancest1 divides panthers into three ancestry categories: F 1 admixed, o ther admixed, and pre introgression type. e Abundance refers to a linear trend in log hazard by abundance index. f Ancest2 divides panthers into two ancestry categories: F 1 admixed, and other admixed and pre introgression type combined. g Age 1 divides the panthers into sub adults (same as Age3), prime adults (2.5 10 and 3.5 10 years old for females and males respecti vely), and older h Ancest3 divides panthers into two ancestry categories: pre introgression type, and other admixed and F 1 admixed combined.

PAGE 205

205 Table L 2. Models for (density dependent) canonical kitten survival, with Akaike model w eights ( w i ) estimates of female canonical kitten survival ( S k (c) ), the effect of abundance index on logit kitten survival ( abundance ) and temporal standard deviation ( t ). All models shown here are density dependent; s urvival estimates are for abundanc e index 0. For more details, see Hostetler et al. (2010) Model w i S k (c) abundance t S(Base a + k :IntAdv b + s&a :F1Adv c + k :Abundance d )r(.) e 0.149 0.782 0.145 0.055 0.016 0.368 S(Base + k :IntAdv + s&a :F1Adv + k :Abundance)r( k s&a ) f 0.080 0.796 0.145 0.055 0.016 0.457 S(Base + k :CanDisF1Adv g + s&a :F1Adv + k :Abund ance)r(.) 0.067 0.814 0.136 0.06 0.018 0.368 S(Base + k :IntAdv + s&a :F1Adv + k :Abundance + k :Sex h )r(.) 0.057 0.803 0.146 0.055 0.016 0.357 S(Base + k :CanDis50TexAdv i + s&a :F1Adv + k :Abundance)r(.) 0.053 0.779 0.15 0.055 0.016 0.368 S(B ase + k :CanDisF1Adv + s&a :F1Adv + k :Abundance)r( k s&a ) 0.035 0.822 0.135 0.06 0.018 0.457 S(Base + k :(IntAdv + Abundance))r(.) 0.034 0.805 0.137 0.059 0.017 0.368 S(Base + k :IntAdv + s&a :F1Adv + k :Abundance + k :Sex)r( k s&a ) 0.030 0.813 0.14 7 0.055 0.017 0.363 S(Base + k :CanDis50TexAdv + s&a :F1Adv + k :Abundance)r( k s&a ) 0.028 0.794 0.148 0.055 0.017 0.457 S(Base + k :CanDisF1Adv + s&a :F1Adv + k :Abundance + k :Sex)r(.) 0.026 0.834 0.134 0.06 0.018 0.357 S(Base + k :(IntAdv + Ab undance))r( k s&a ) 0.024 0.824 0.133 0.06 0.017 0.457 S(Base + k :AdmAdv j + s&a :F1Adv + k :Abundance)r(.) 0.023 0.743 0.16 0.037 0.013 0.368 S(Base + IntAdv + k :Abundance)r(.) 0.020 0.814 0.124 0.046 0.013 0.368 S(Base + k :CanDis50TexAdv + s&a :F1Adv + k :Abundance + k :Sex)r(.) 0.020 0.801 0.151 0.055 0.016 0.357 S(Base + CanDisF1Adv + k :Abundance)r(.) 0.018 0.779 0.146 0.042 0.013 0.368 S(Base + AdmAdv + k :Abundance)r(.) 0.017 0.771 0.146 0.038 0.013 0.368 S(Base + k :Ca nDisF1Adv + s&a :F1Adv + k :Abundance + k :Sex)r( k s&a ) 0.013 0.839 0.135 0.06 0.018 0.363 S(Base + k :(IntAdv + Abundance) + k :Sex)r(.) 0.013 0.825 0.137 0.059 0.017 0.357 S(Base + k :CanDis50TexAdv + k :Abundance)r(.) 0.013 0.799 0.143 0.058 0.017 0.368 S(Base + k :AdmAdv + s&a :F1Adv + k :Abundance)r( k s&a ) 0.012 0.757 0.16 0.037 0.013 0.457 S(Base + k :CanDisF1Adv + k :Abundance)r(.) 0.012 0.813 0.137 0.061 0.018 0.368 S(Base + CanDis50TexAdv + k :Abundance)r(.) 0.011 0.796 0.13 5 0.044 0.013 0.368 S(Base + IntAdv + k :Abundance)r( k s&a ) 0.011 0.829 0.122 0.046 0.013 0.457 S(Base + k :50TexAdv k + s&a :F1Adv + k :Abundance)r(.) 0.010 0.771 0.146 0.037 0.012 0.368 S(Base + k :CanDis50TexAdv + s&a :F1Adv + k :Abundance + k :Sex)r( k s&a ) 0.010 0.811 0.151 0.055 0.017 0.363 S(Base + CanDisF1Adv + k :Abundance)r( k s&a ) 0.010 0.799 0.143 0.041 0.014 0.457 S(Base + AdmAdv + k :Abundance)r( k s&a ) 0.010 0.791 0.144 0.038 0.013 0.457 S(Base + s&a :F1Adv + k :Abunda nce)r(.) 0.009 0.825 0.116 0.038 0.012 0.368 S(Base + IntAdv + k :Abundance + k :Sex)r(.) 0.009 0.845 0.118 0.046 0.013 0.357 S(Base + k :AdmAdv + s&a :F1Adv + k :Abundance + k :Sex)r(.) 0.009 0.77 0.162 0.038 0.013 0.357 S(Base + k :CanDis50 TexAdv + k :Abundance)r( k s&a ) 0.009 0.819 0.138 0.059 0.018 0.457

PAGE 206

206 S(Base + k :(IntAdv + Abundance) + k :Sex)r( k s&a ) 0.009 0.839 0.135 0.06 0.017 0.363 S(Base + k :CanDisF1Adv + k :Abundance)r( k s&a ) 0.008 0.824 0.136 0.06 0.018 0.457 S(Ba se + CanDisF1Adv + k :Abundance + k :Sex)r(.) 0.008 0.813 0.141 0.042 0.014 0.357 S(Base + 50TexAdv + k :Abundance)r(.) 0.008 0.788 0.138 0.038 0.013 0.368 S(Base + AdmAdv + k :Abundance + k :Sex)r(.) 0.007 0.802 0.144 0.039 0.013 0.357 S(B ase + F1Adv + k :Abundance)r(.) 0.006 0.774 0.147 0.035 0.013 0.368 S(Base + CanDis50TexAdv + k :Abundance)r( k s&a ) 0.006 0.813 0.132 0.044 0.013 0.457 S(Base + k :50TexAdv + s&a :F1Adv + k :Abundance)r( k s&a ) 0.005 0.786 0.145 0.037 0.013 0. 457 S(Base + s&a :F1Adv + k :Abundance)r( k s&a ) 0.005 0.837 0.114 0.038 0.013 0.457 S(Base + s&a :F1Adv + k :Abundance + k :Sex)r(.) 0.005 0.858 0.107 0.039 0.013 0.357 S(Base + k :CanDisF1Adv + k :Abundance + k :Sex)r(.) 0.005 0.833 0.136 0.06 0.018 0.357 S(Base + k :CanDis50TexAdv + k :Abundance + k :Sex)r(.) 0.005 0.819 0.144 0.058 0.017 0.357 S(Base + CanDis50TexAdv + k :Abundance + k :Sex)r(.) 0.005 0.827 0.131 0.044 0.013 0.357 S(Base + IntAdv + k :Abundance + k :Sex)r( k s&a ) 0.0 05 0.858 0.115 0.046 0.014 0.363 S(Base + k :(AdmAdv + Abundance))r(.) 0.005 0.756 0.156 0.039 0.013 0.368 S(Base + k :AdmAdv + s&a :F1Adv + k :Abundance + k :Sex)r( k s&a ) 0.005 0.779 0.164 0.037 0.013 0.363 S(Base + k :50TexAdv + s&a :F1Adv + k :Abundance + k :Sex)r(.) 0.004 0.804 0.143 0.038 0.013 0.357 S(Base + 50TexAdv + k :Abundance)r( k s&a ) 0.004 0.808 0.135 0.039 0.013 0.457 S(Base + CanDisF1Adv + k :Abundance + k :Sex)r( k s&a ) 0.004 0.831 0.138 0.042 0.014 0.363 S(Base + AdmAdv + k :Abundance + k :Sex)r( k s&a ) 0.004 0.819 0.142 0.039 0.013 0.363 S(Base + k :F1Adv + s&a :F1Adv + k :Abundance)r(.) 0.004 0.805 0.135 0.037 0.013 0.368 S(Base + F1Adv + k :Abundance)r( k s&a ) 0.004 0.798 0.144 0.035 0.013 0.457 S (Base + 50TexAdv + k :Abundance + k :Sex)r(.) 0.003 0.82 0.134 0.039 0.013 0.357 S(Base + k :CanDis50TexAdv + k :Abundance + k :Sex)r( k s&a ) 0.003 0.833 0.141 0.059 0.018 0.363 S(Base + k :(AdmAdv + Abundance))r( k s&a ) 0.003 0.776 0.154 0.039 0.013 0.457 S(Base + k :CanDisF1Adv + k :Abundance + k :Sex)r( k s&a ) 0.003 0.839 0.137 0.06 0.018 0.363 S(Base + F1Adv + k :Abundance + k :Sex)r(.) 0.003 0.814 0.14 0.035 0.013 0.357 S(Base + s&a :F1Adv + k :Abundance + k :Sex)r( k s&a ) 0.002 0.867 0.105 0.039 0.013 0.363 S(Base + CanDis50TexAdv + k :Abundance + k :Sex)r( k s&a ) 0.002 0.842 0.128 0.044 0.014 0.363 S(Base + k :50TexAdv + s&a :F1Adv + k :Abundance + k :Sex)r( k s&a ) 0.002 0.815 0.143 0.038 0.013 0.363 S(Base + k :(50TexAdv + Abundance))r(.) 0.002 0.782 0.143 0.039 0.013 0.368 S(Base + k :F1Adv + s&a :F1Adv + k :Abundance)r( k s&a ) 0.002 0.819 0.132 0.037 0.013 0.457 S(Base + 50TexAdv + k :Abundance + k :Sex)r( k s&a ) 0.002 0.837 0.131 0.039 0.013 0.363 S(Base + k :F1Adv + s&a :F1Adv + k :Abundance + k :Sex)r(.) 0.002 0.843 0.124 0.038 0.013 0.357 S(Base + k :(AdmAdv + Abundance) + k :Sex)r(.) 0.002 0.78 0.159 0.039 0.013 0.357 S(Base + F1Adv + k :Abundance + k :Sex)r( k s&a ) 0.002 0.835 0.137 0.036 0.0 13 0.363

PAGE 207

207 S(Base + k :Abundance)r(.) 0.001 0.832 0.114 0.039 0.012 0.368 S(Base + k :(50TexAdv + Abundance))r( k s&a ) 0.001 0.803 0.14 0.039 0.013 0.457 S(Base + k :(AdmAdv + Abundance) + k :Sex)r( k s&a ) 0.001 0.796 0.159 0.039 0.013 0.363 a The base survival model represents survival differing between kittens, females age 1 and 2, females 3 and older, males age 1 3, and males 4 and older (Hostetler et al., 2010) b 2 ancestry categories: 1) canonical; 2) backcrossed to canonical, backcrossed to Everglades, backcrossed to Texas, and F 1 c 2 ancestry categories: 1) canonical, backcrossed to canonical, backcrossed to Everglades, and backcrossed to Texas; and 2) F 1 d k : Abundance refers to a linear trend in logit kitten survival by abundance index. e The same recovery probability for all panthers. f Recovery probabilities differing between kittens and older panthers. g 3 ancestry categories: 1) canonical; 2) b ackcrossed to canonical, backcrossed to Everglades, and backcrossed to Texas; and 3) F 1 h Kitten survival differing by sex. i 3 ancestry categories: 1) canonical; 2) backcrossed to canonical and backcrossed to Everglades; 3) backcrossed to Texas and F 1 j 2 ancestry categories: 1) canonical and backcrossed to canonical; 2) backcrossed to Everglades, backcrossed to Texas, and F 1 k 2 ancestry categories: 1) canonical, backcrossed to canonical, and backcrossed to Everglades; 2) backcrossed to Texas, and F 1

PAGE 208

208 Table L 3 Models for density independent canonical kitten surv ival, with Akaike model weights ( w i ) and estimates of female canonical kitten survival ( S k (c) ) These models are > 10 QAIC c units from the top model, but are < 10 QAIC c units from the top d ensity independent model, and were therefore included in estimation of (c) but not in estimation of probability of extinction. For more details, see Hostetler et al. (2010) Model w i S k (c) S(Base a + k :AdmAdv b + s&a :F1Adv c )r(.) d 0.093 0.21 0.056 S(Base + k :AdmAdv + s&a :F1Adv)r( k s&a ) e 0.075 0.243 0.074 S(Base + AdmAdv)r(.) 0.051 0.229 0.052 S(Base + F1Adv)r(.) 0.047 0.266 0.053 S(Base + AdmAdv)r( k s&a ) 0.039 0.265 0.072 S(Base + CanDisF1Adv f )r(.) 0.037 0.221 0.059 S(Base + F1Adv)r( k s&a ) 0.037 0.308 0.079 S(Base + k :AdmAdv + s&a :F1Adv + k :Sex g )r(.) 0.036 0.23 0.078 S(Base + k :50TexAdv h + s&a :F1Adv)r(.) 0.033 0.234 0.057 S(Base + CanDisF1Adv)r( k s&a ) 0.028 0.261 0.083 S(Base + k :AdmAdv + s&a :F1Adv + k :Sex)r( k s&a ) 0.027 0.256 0.093 S(Base + k :50TexAdv + s&a :F1Adv)r( k s&a ) 0.023 0.265 0.074 S(Base + F1Adv + k :Sex)r(.) 0.021 0.299 0.076 S(Base + AdmAdv + k :Sex)r(.) 0.021 0.253 0.073 S(Base + k :F1Adv + s&a :F1Adv)r(.) 0.019 0.27 0.054 S(Base + 50TexAdv)r(.) 0.019 0.242 0.052 S(Base + s&a :F1Adv)r(.) 0.018 0.292 0.053 S(Base + CanDisF1Adv + k :Sex)r(.) 0.016 0.249 0.079 S(Base + F1Adv + k :Sex)r( k s&a ) 0.015 0.336 0.099 S(Base + AdmAdv + k :Sex)r( k s&a ) 0.015 0 .284 0.091 S(Base + k :AdmAdv)r( k s&a ) 0.015 0.246 0.076 S(Base + k :IntAdv i + s&a :F1Adv)r(.) 0.015 0.201 0.079 S(Base + s&a :F1Adv)r( k s&a ) 0.014 0.33 0.073 S(Base + k :CanDis50TexAdv j + s&a :F1Adv)r(.) 0.014 0.201 0.079 S(Base + k :50TexAdv + s&a :F1Adv + k :Sex)r(.) 0.014 0.262 0.08 S(Base + k :AdmAdv)r(.) 0.014 0.207 0.056 S(Base + 50TexAdv)r( k s&a ) 0.014 0.277 0.072 S(Base + k :F1Adv + s&a :F1Adv)r( k s&a ) 0.013 0.307 0.076 S(Base + k :IntAdv + s&a :F1Adv)r( k s&a ) 0.013 0.227 0. 093 S(Base + CanDisF1Adv + k :Sex)r( k s&a ) 0.011 0.285 0.101 S(Base + k :CanDisF1Adv + s&a :F1Adv)r(.) 0.011 0.201 0.079 S(Base + k :CanDis50TexAdv + s&a :F1Adv)r( k s&a ) 0.010 0.223 0.09 S(Base + s&a :F1Adv + k :Sex)r(.) 0.009 0.334 0.077 S(Base + k :CanDisF1Adv + s&a :F1Adv)r( k s&a ) 0.009 0.229 0.095

PAGE 209

209 S(Base + CanDis50TexAdv)r(.) 0.009 0.219 0.057 S(Base + k :50TexAdv + s&a :F1Adv + k :Sex)r( k s&a ) 0.009 0.288 0.094 S(Base + k :F1Adv + s&a :F1Adv + k :Sex)r(.) 0.008 0.305 0.077 S(Base + 5 0TexAdv + k :Sex)r(.) 0.008 0.269 0.074 S(Base + k :IntAdv + s&a :F1Adv + k :Sex)r(.) 0.007 0.234 0.101 S(Base + CanDis50TexAdv)r( k s&a ) 0.006 0.252 0.076 S(Base + s&a :F1Adv + k :Sex)r( k s&a ) 0.006 0.365 0.093 S(Base + IntAdv)r(.) 0.006 0.216 0.057 S(Base + k :F1Adv + s&a :F1Adv + k :Sex)r( k s&a ) 0.006 0.336 0.095 S(Base + k :CanDis50TexAdv + s&a :F1Adv + k :Sex)r(.) 0.005 0.226 0.099 S(Base + 50TexAdv + k :Sex)r( k s&a ) 0.005 0.3 0.091 S(Base + k :IntAdv + s&a :F1Adv + k :Sex)r( k s&a ) 0.005 0.253 0.112 S(Base + k :(AdmAdv) + k :Sex)r( k s&a ) 0.005 0.255 0.095 S(Base + k :(AdmAdv) + k :Sex)r(.) 0.005 0.224 0.077 S(Base + k :50TexAdv)r(.) 0.005 0.231 0.056 S(Base + IntAdv)r( k s&a ) 0.005 0.252 0.076 S(Base + k :CanDisF1Adv + s&a :F1 Adv + k :Sex)r(.) 0.005 0.23 0.1 S(Base + k :50TexAdv)r( k s&a ) 0.004 0.266 0.075 S(Base + k :CanDis50TexAdv + s&a :F1Adv + k :Sex)r( k s&a ) 0.004 0.242 0.109 S(Base + CanDis50TexAdv + k :Sex)r(.) 0.004 0.244 0.077 S(Base + k :F1Adv)r(.) 0.004 0.267 0.053 S(Base + k :CanDisF1Adv + s&a :F1Adv + k :Sex)r( k s&a ) 0.003 0.25 0.113 S(Base + IntAdv + k :Sex)r(.) 0.003 0.247 0.077 S(Base + CanDis50TexAdv + k :Sex)r( k s&a ) 0.003 0.274 0.093 S(Base + k :(F1Adv + IntAdv))r(.) 0.002 0.199 0.079 S(B ase)r(.) 0.002 0.29 0.053 S(Base + IntAdv + k :Sex)r( k s&a ) 0.002 0.279 0.094 S(Base + k :CanDis50TexAdv)r(.) 0.002 0.198 0.078 S(Base + k :(50TexAdv) + k :Sex)r(.) 0.002 0.256 0.079 S(Base + k :IntAdv)r( k s&a ) 0.002 0.226 0.093 S(Base + k :C anDis50TexAdv)r( k s&a ) 0.002 0.223 0.091 S(Base)r( k s&a ) 0.002 0.33 0.074 S(Base + k :IntAdv)r(.) 0.002 0.199 0.078 S(Base + k :(F1Adv) + k :Sex)r(.) 0.002 0.299 0.077 S(Base + k :(50TexAdv) + k :Sex)r( k s&a ) 0.002 0.285 0.095 S(Base + k :Sex )r(.) 0.001 0.331 0.077 S(Base + k :(F1Adv + IntAdv) + k :Sex)r(.) 0.001 0.226 0.099 S(Base + k :Sex)r( k s&a ) 0.001 0.365 0.094 S(Base + k :(IntAdv) + k :Sex)r(.) 0.001 0.23 0.099 S(Base + k :CanDis50TexAdv + k :Sex)r(.) 0.001 0.221 0.097 S(Ba se + k :(IntAdv) + k :Sex)r( k s&a ) 0.001 0.25 0.111

PAGE 210

210 S(Base + k :CanDis50TexAdv + k :Sex)r( k s&a ) 0.001 0.239 0.109 a The base survival model represents survival differing between kittens, females age 1 and 2, females 3 and older, males age 1 3, and mal es 4 and older (Hostetler et al., 2010) b 2 ancestry categories: 1) canonical and backcrossed to canonical; 2) backcrossed to Everglades, backcrossed to Texas, and F 1 c 2 ancestry categories: 1) canonical, backcrossed to canonical, backcrossed to Everglades, and backcrossed to Texas; and 2) F 1 d The same recovery probability for all panthers. e Recovery probabilities differing between kittens and older panthers. f 3 an cestry categories: 1) canonical; 2) backcrossed to canonical, backcrossed to Everglades, and backcrossed to Texas; and 3) F 1 g Kitten survival differing by sex. h 2 ancestry categories: 1) canonical, backcrossed to canonical, and backcrossed to Everglades ; 2) backcrossed to Texas, and F 1 i 2 ancestry categories: 1) canonical; 2) backcrossed to canonical, backcrossed to Everglades, backcrossed to Texas, and F 1 j 3 ancestry categories: 1) canonical; 2) backcrossed to canonical and backcrossed to Everglades ; 3) backcrossed to Texas and F 1

PAGE 211

211 Table L 4. Models for sub adult and adult canonical pro bability of breeding. Includes Akaike model weights ( w i ) estimates of sub adult female canonical probability of breeding ( q sa (c) ), prime adult female canonical pr obability of breeding ( q pa (c) ), older adult female canonical probability of breeding ( q oa (c) ), the effects of abundance index and abundance index squared on monthly complementary log log probability of breeding ( abundance and abundance2 respectively) and temporal standard deviation ( t ). Density independent models are in bold. Probability of breeding estimates for density dependent models are for abundance index 0. Model w i q sa (c) q pa (c) q oa (c) abundance abundance2 t Age2 a + Ancest2 b + Abundance c 0.183 0.301 0.100 0.301 0.100 0.024 0.025 0.009 0.005 0 0.378 Age2 + Ancest2 0.107 0.498 0.040 0.498 0.040 0.038 0.037 0 0 0.420 Age1 d + Ancest2 + Abundance 0.088 0.271 0.102 0.326 0.110 0.025 0.026 0.008 0.005 0 0.373 Age2 + Ance st1 e 0.084 0.443 0.060 0.443 0.060 0.038 0.037 0 0 0.420 Age2 + Ancest2 + Abundance + Abundance 2 0.076 0.412 0.275 0.412 0.275 0.035 0.044 0.003 0.025 0.0001 0.0002 0.378 Age2 + Ancest1 + Abundance 0.076 0.305 0.102 0.305 0.102 0.026 0.027 0.008 0.006 0 0.378 Age1 + Ancest2 0.067 0.416 0.089 0.518 0.044 0.038 0.037 0 0 0.407 Age1 + Ancest1 0.061 0.353 0.092 0.458 0.062 0.038 0.037 0 0 0.407 Age1 + Ancest1 + Abundance 0.039 0.272 0.103 0.337 0.115 0.027 0.028 0 .006 0.006 0 0.373 Age1 + Ancest2 + Abundance + Abundance 2 0.035 0.350 0.262 0.410 0.273 0.033 0.042 0.001 0.025 0.0001 0.0002 0.373 Age2 + Ancest1 + Abundance + Abundance 2 0.033 0.444 0.292 0.444 0.292 0.040 0.052 0.007 0.026 0.000 1 0.0002 0.378 Age2 + Abundance 0.030 0.274 0.094 0.274 0.094 0.024 0.025 0.009 0.005 0 0.378 Age2 0.019 0.459 0.037 0.459 0.037 0.038 0.037 0 0 0.420 Age1 + Ancest1 + Abundance + Abundance 2 0.016 0.376 0.277 0.451 0.293 0.040 0.05 1 0.005 0.026 0.0001 0.0002 0.373 Age2 + Abundance + Abundance 2 0.016 0.473 0.297 0.473 0.297 0.046 0.057 0.012 0.025 0.0001 0.0002 0.378 Age1 + Abundance 0.014 0.249 0.096 0.297 0.104 0.025 0.026 0.008 0.005 0 0.373 Age2 + Ances t3 + Abundance 0.012 0.273 0.094 0.273 0.094 0.023 0.024 0.010 0.006 0 0.378 Age1 0.012 0.382 0.083 0.477 0.041 0.038 0.037 0 0 0.407 Age2 + Ancest3 f 0.007 0.443 0.060 0.443 0.060 0.038 0.037 0 0 0.420 Age1 + Abundance + Abundance 2 0 .007 0.418 0.292 0.476 0.296 0.044 0.055 0.011 0.025 0.0001 0.0002 0.373 Age2 + Ancest3 + Abundance + Abundance 2 0.006 0.458 0.297 0.458 0.297 0.043 0.055 0.010 0.026 0.0001 0.0002 0.378 Age1 + Ancest3 + Abundance 0.005 0.250 0.09 6 0.295 0.104 0.024 0.025 0.009 0.006 0 0.373

PAGE 212

212 Age1 + Ancest3 0.005 0.361 0.093 0.457 0.062 0.038 0.037 0 0 0.407 Age1 + Ancest3 + Abundance + Abundance 2 0.003 0.410 0.292 0.465 0.298 0.042 0.054 0.009 0.026 0.0001 0.0002 0.373 a A ge 2 divides panthers into young adults (1 10 years old) and older b Ancest2 divides panthers into two ancestry categories: F 1 admixed, and other admixed and pre introgression type combined. c Abundance refers to a linear trend in mon thly complementary log log probability of breeding by abundance index. d Age 1 divides the female panthers into sub adults (1 2.5 years old), prime adults (2.5 10 years old) and older e Ancest1 divides panthers into three ancestry ca tegories: F 1 admixed, other admixed, and pre introgression type. f Ancest3 divides panthers into two ancestry categories: pre introgression type, and other admixed and F 1 admixed combined.

PAGE 213

213 Table L 5. Models for sub adult and adult average number of kitt ens produced (for canonical females that bred in a year). Includes Akaike model weights ( w i ) estimates of sub adult canonical dam kittens ( sa (c) ), adult canonical dam kittens ( a (c) ), and the effects of abundance index on cumulative logit number of kittens ( abundance ). Density independent models are in bold. Number of kittens estimates for density dependent models are for abundance index 0. Model w i sa (c) a (c) abundance Age3 a 0.186 3.13 0.28 2.69 0.12 0 Constant b 0.160 2.76 0.12 2.76 0.12 0 Age3 + Abundance c 0.089 2.87 0.44 2.41 0.38 0.006 0.009 Age3 + Ancest2 d 0.082 3.15 0.28 2.71 0.13 0 Ancest2 0.071 2.79 0.12 2.79 0.12 0 Age3 + Ancest3 e 0.069 3.11 0.33 2.67 0.19 0 Abundance 0.069 2.55 0.39 2.55 0.39 0.005 0.009 Ancest3 0.061 2.73 0.19 2.73 0.19 0 Age3 + Ancest2 + Abundance 0.038 2.90 0.45 2.45 0.39 0.006 0.009 Age3 + Ancest3 + Abunda nce 0.034 2.86 0.44 2.40 0.39 0.008 0.010 Age3 + Ancest1 f 0.031 3.10 0.33 2.67 0.19 0 Ancest2 + Abundance 0.030 2.59 0.40 2.59 0.40 0.005 0.009 Ancest1 0.029 2.73 0.19 2.73 0.19 0 Ancest3 + Abundance 0.025 2.55 0.39 2.55 0.39 0. 005 0.010 Age3 + Ancest1 + Abundance 0.014 2.90 0.45 2.44 0.40 0.007 0.011 Ancest1 + Abundance 0.011 2.60 0.40 2.60 0.40 0.004 0.010 a Age 3 divides female panthers into sub adults (1 b No predi ctor variables. c Abundance refers to a linear trend in log hazard by abundance index. d Ancest2 divides panthers into two ancestry categories: F 1 admixed, and other admixed and pre introgression type combined. e Ancest3 divides panthers into two ancestry categories: pre introgression type, and other admixed and F 1 admixed combined. f Ancest1 divides panthers into three ancestry categories: F 1 admixed, other admixed, and pre introgression type.

PAGE 214

214 LIST OF REFERENCES Agresti, A., 2007. An introduction to categorical data analysis, 2nd edn. John Wiley & Sons, Hoboken, NJ, USA. Alberts, S.C., Altmann, J., 2003. Matrix models for primate life history analysis, in: Kappeler, P.M., Pereira, M.E. (Eds.), Primate life histories and socioecology. U niversity of Chicago Press, Chicago, IL, USA, pp. 66 102. Allendorf, F.W., Leary, R.F., Spruell, P., Wenburg, J.K., 2001. The problems with hybrids: Setting conservation guidelines. Trends in Ecology & Evolution 16, 613 622. Arnold, M.L., Hodges, S.A., 199 5. Are natural hybrids fit or unfit relative to their parents? Trends in Ecology & Evolution 10, 67 71. Ashman, D.L., Greer, K., 1976. Age techniques, in: Christensen, G.C., Fisher, R.J. (Eds.), Transactions of the Mountain Lion Workshop. U.S. Fish and Wil dlife Service, Portland, Oregon. Bakker, V.J., Doak, D.F., Roemer, G.W., Garcelon, D.K., Coonan, T.J., Morrison, S.A., Lynch, C., Ralls, K., Shaw, R., 2009. Incorporating ecological drivers and uncertainty into a demographic population viability analysis f or the island fox. Ecological Monographs 79, 77 108. Ball, S.J., Lindenmayer, D.B., Possingham, H.P., 2003. The predictive accuracy of population viability analysis: a test using data from two small mammal species in a fragmented landscape. Biodiversity an d Conservation 12, 2393 2413. Barone, M.A., Roelke, M.E., Howard, J., Brown, J.L., Anderson, A.E., Wildt, D.E., 1994. Reproductive characteristics of male Florida panthers comparative studies from Florida, Texas, Colorado, Latin America, and North Americ an zoos. Journal of Mammalogy 75, 150 162. Beier, P., Vaughan, M.R., Conroy, M.J., Quigley, H., 2003. An analysis of scientific literature related to the Florida panther. Florida Fish and Wildlife Conservation Commission, Tallahassee. Beier, P., Vaughan, M .R., Conroy, M.J., Quigley, H., 2006. Evaluating scientific inferences about the Florida panther. Journal of Wildlife Management 70, 236 245. Belden, R.C., Frankenberger, W.B., 1990. Biology of a feral hog population in south central Florida. Proceedings o f Annual Conference of Southeastern Association of Fish and Wildlife Agencies 44, 231 242. Belden, R.C., Frankenberger, W.B., McBride, R.T., Schwikert, S.T., 1988. Panther habitat use in southern Florida. Journal of Wildlife Management 52, 660 663.

PAGE 215

215 Benson, J.F., Hostetler, J.A., Onorato, D.P., Johnson, W.E., Roelke, M.E., O'Brien, S.J., Jansen, D., Oli, M.K., in revision. Intentional genetic introgression improves survival of F 1 adults and sub adults in a small, inbred felid population. Journal of Animal Ec ology. Benson, J.F., Lotz, M.A., Jansen, D., 2008. Natal den selection by Florida panthers. Journal of Wildlife Management 72, 405 410. Birchler, J.A., Auger, D.L., Riddle, N.C., 2003. In search of the molecular basis of heterosis. Plant Cell 15, 2236 2239 Bishop, C.J., White, G.C., Lukacs, P.M., 2008. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Journal of Wildlife Management 72, 1085 1093. Bolker, B.M., Brooks, M.E., Clark, C.J., Geange, S.W., Poulsen, J.R., Ste vens, M.H.H., White, J. S.S., 2009. Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology & Evolution 24, 127 135. Brito, D., 2009. Genetic consequences of population subdivision: the marsupial Micoureus paraguayan us (Mammalia: Didelphimorphia) as a case study. Zoologia 26, 684 693. Brook, B.W., Burgman, M.A., Akakaya, H.R., O'Grady, J.J., Frankham, R., 2002a. Critiques of PVA ask the wrong questions: Throwing the heuristic baby out with the numerical bath water. C onservation Biology 16, 262 263. Brook, B.W., O'Grady, J.J., Chapman, A.P., Burgman, M.A., Akakaya, H.R., Frankham, R., 2000. Predictive accuracy of population viability analysis in conservation biology. Nature 404, 385 387. Brook, B.W., Tonkyn, D.W., O'G rady, J.J., Frankham, R., 2002b. Contribution of inbreeding to extinction risk in threatened species. Conservation Ecology 6, 16. Buerkle, C.A., Morris, R.J., Asmussen, M.A., Rieseberg, L.H., 2000. The likelihood of homoploid hybrid speciation. Heredity 84 441 451. Burke, J.M., Arnold, M.L., 2001. Genetics and the fitness of hybrids. Annual Review of Genetics 35, 31 52. Burnham, K.P., 1993. A theory for combined analysis of ring recovery and recapture data, in: Lebreton, J. D., North, P.M. (Eds.), Marked i ndividuals in the study of bird population. Birkhuser Verlag, Boston, MA, pp. 199 213. Burnham, K.P., Anderson, D.R., 2002. Model selection and multimodel inference, 2nd edn. Springer, New York.

PAGE 216

216 Cam, E., Link, W.A., Cooch, E.G., Monnat, J.Y., Danchin, E., 2002. Individual covariation in life history traits: Seeing the trees despite the forest. American Naturalist 159, 96 105. Campbell, D.R., Waser, N.M., 2001. Genotype by environment interaction and the fitness of plant hybrids in the wild. Evolution 55, 6 69 676. Caswell, H., 2001. Matrix population models, Second edn. Sinauer Associates, Sunderland, MA. Caswell, H., 2005. Sensitivity analysis of the stochastic growth rate: Three extensions. Australian & New Zealand Journal of Statistics 47, 75 85. Caughley G., 1994. Directions in conservation biology. Journal of Animal Ecology 63, 215 244. Caughley, G., Gunn, A., 1996. Conservation biology in theory and practice. Blackwell Science, Cambridge, MA, USA. Charpentier, M.J.E., Tung, J., Altmann, J., Alberts, S. C., 2008. Age at maturity in wild baboons: genetic, environmental and demographic influences. Molecular Ecology 17, 2026 2040. Comiskey, E.J., Bass, O.L., Gross, L.J., McBride, R.T., Salinas, R., 2002. Panthers and forests in South Florida: an ecological p erspective. Conservation Ecology 6. Cooch, E., White, G.C. eds., 2008. Program MARK 'a gentle introduction' 7th Edition. http://www.phidot.org/software/mark/docs/book/ Cooley, H.S., Wielgus, R .B., Koehler, G., Maletzke, B., 2009. Source populations in carnivore management: cougar demography and emigration in a lightly hunted population. Animal Conservation 12, 321 328. Coulson, T., Mace, G.M., Hudson, E., Possingham, H.P., 2001. The use and abu se of population viability analysis. Trends In Ecology & Evolution 16, 219 221. Cox, D.R., 1972. Regression models and life tables. Journal of the Royal Statistical Society Series B Statistical Methodology 34, 187 220. Creel, S., 2006. Recovery of the Flor ida panther genetic rescue, demographic rescue, or both? Response to Pimm et al. (2006). Animal Conservation 9, 125 126. Crnokrak, P., Roff, D.A., 1999. Inbreeding depression in the wild. Heredity 83, 260 270. Crow, J.F., 1948. Alternative hypotheses of hybrid vigor. Genetics 33, 477 487. Culver, M., Johnson, W.E., Pecon Slattery, J., O'Brien, S.J., 2000. Genomic ancestry of the American puma (Puma concolor). Journal of Heredity 91, 186 197.

PAGE 217

217 Cunningham, M.W., Brown, M.A., Shindle, D.B., Terrell, S.P., Hay es, K.A., Ferree, B.C., McBride, R.T., Blankenship, E.L., Jansen, D., Citino, S.B., Roelke, M.E., Kiltie, R.A., Troyer, J.L., O'Brien, S.J., 2008. Epizootiology and management of feline leukemia virus in the Florida puma. Journal of Wildlife Diseases 44, 5 37 552. Davis, J.B., Cox, R.R., Kaminski, R.M., Leopold, B.D., 2007. Survival of wood duck ducklings and broods in Mississippi and Alabama. Journal of Wildlife Management 71, 507 517. de Castro, F., Bolker, B.M., 2005. Mechanisms of disease induced extinct ion. Ecology Letters 8, 117 126. Dees, C.S., Clark, J.D., Van Manen, F.T., 2001. Florida panther habitat use in response to prescribed fire. Journal of Wildlife Management 65, 141 147. Driscoll, C.A., Menotti Raymond, M., Nelson, G., Goldstein, D., O'Brien S.J., 2002. Genomic microsatellites as evolutionary chronometers: A test in wild cats. Genome Research 12, 414 423. Ebert, D., Haag, C., Kirkpatrick, M., Riek, M., Hottinger, J.W., Pajunen, V.I., 2002. A selective advantage to immigrant genes in a Daphni a metapopulation. Science 295, 485 488. Edmands, S., 2007. Between a rock and a hard place: evaluating the relative risk of inbreeding and outbreeding for conservation and management. Molecular Ecology 16, 463 475. Efron, B., Tibshirani, R.J., 1998. An int roduction to the bootstrap. Chapman and Hall/CRC Press, Boca Raton, Florida. Ellner, S.P., Fieberg, J., 2003. Using PVA for management despite uncertainty: Effects of habitat, hatcheries, and harvest on salmon. Ecology 84, 1359 1369. Ellner, S.P., Fieberg, J., Ludwig, D., Wilcox, C., 2002. Precision of population viability analysis. Conservation Biology 16, 258 261. Federico, P., Canziani, G.A., 2005. Modeling the population dynamics of capybara Hydrochaeris hydrochaeris : a first step towards a management p lan. Ecological Modelling 186, 111 121. Fergus, C., 1991. The Florida panther verges on extinction. Science 251, 1178 1180. Fergus, C., 1998. Swamp screamer: at large with the Florida panther. University Press of Florida, Gainesville, Florida USA. Ferreira .G.A., Amos, W., 2006. Inbreeding depression and multiple regions showing heterozygote advantage in Drosophila melanogaster exposed to stress. Molecular Ecology 15, 3885 3893.

PAGE 218

218 Fieberg, J., DelGiudice, G.D., 2009. What time is it? Choice of time origin and scale in extended proportional hazards models. Ecology 90, 1687 1697. Fieberg, J., Ellner, S.P., 2001. Stochastic matrix models for conservation and management: a comparative review of methods. Ecology Letters 4, 244 266. Fisher, D.O., Hoyle, S.D., Blo mberg, S.P., 2000. Population dynamics and survival of an endangered wallaby: A comparison of four methods. Ecological Applications 10, 901 910. Fitzpatrick, B.M., Shaffer, H.B., 2007. Hybrid vigor between native and introduced salamanders raises new chall enges for conservation. Proceedings of the National Academy of Sciences of the United States of America 104, 15793 15798. Fleming, D.M., 1994. Distribution, abundance, and demography of white tailed deer in the Everglades, In Proceedings of the Florida Pan ther Conference. ed. Jordan, D. U.S. Fish and Wildlife Service. Frankham, R., Ballou, J.D., Briscoe, D.A., 2002. Introduction to conservation genetics. Cambridge University Press, Cambridge, UK. Franklin, I.R., 1980. Evolutionary change in small population s, in: Soule, M.E., Wilcox, B.A. (Eds.), Conservation Biology: An Evolutionary Ecological Perspective. Sinauer Associates, Sunderland, Massachusetts, USA, pp. 135 149. Fujiwara, M., Caswell, H., 2001. Demography of the endangered North Atlantic right whale Nature 414, 537 541. FWC, 2008. Annual report on the research and management of Florida panthers: 2007 2008. Fish and Wildlife Research Institute and Division of Habitat and Species Conservation, Naples, Florida, USA. Garrison, E.P., McCown, J.W., Oli, M .K., 2007. Reproductive ecology and cub survival of Florida black bears. Journal of Wildlife Management 71, 720 727. Gascoigne, J., Berec, L., Gregory, S., Courchamp, F., 2009. Dangerously few liaisons: a review of mate finding Allee effects. Population Ec ology 51, 355 372. Gelman, A., 1996. Inference and monitoring convergence, in: Gilks, W.R., Richardson, S., Spiegelhalter, D.J. (Eds.), Markov Chain Monte Carlo in Practice. Chapman and Hall/CRC, Boca Raton, Florida, USA, pp. 131 143. Gerber, L.R., McCallu m, H., Lafferty, K.D., Sabo, J.L., Dobson, A., 2005. Exposing extinction risk analysis to pathogens: Is disease just another form of density dependence? Ecological Applications 15, 1402 1414.

PAGE 219

219 Good, T.P., Ellis, J.C., Annett, C.A., Pierotti, R., 2000. Bound ed hybrid superiority in an avian hybrid zone: Effects of mate, diet, and habitat choice. Evolution 54, 1774 1783. Grant, P.R., Grant, B.R., 1992. Hybridization of bird species. Science 256, 193 197. Greenwood, P.J., 1980. Mating Systems, Philopatry And Di spersal In Birds And Mammals. Animal Behaviour 28, 1140 1162. Greig, J.C., 1979. Principles of genetic conservation in relation to wildlife management in southern Africa. South African Journal of Wildlife Research 9, 57 78. Gross, L., 2005. Why not the bes t? How science failed the Florida panther. PloS Biology 3, 1525 1531. Hallerman, E.M., 2003. Coadaptation and outbreeding depression, in: Hallerman, E.M. (Ed.), Population Genetics: Principles and Applications for Fisheries Scientists. American Fisheries S ociety, Bethesda, Maryland, USA, pp. 239 259. Heisey, D.M., Patterson, B.R., 2006. A review of methods to estimate cause specific mortality in presence of competing risks. Journal of Wildlife Management 70, 1544 1555. Henle, K., Sarre, S., Wiegand, K., 200 4. The role of density regulation in extinction processes and population viability analysis. Biodiversity and Conservation 13, 9 52. Henry, J.A., Portier, K.M., Coyne, J., 1994. The climate and weather of Florida. Pineapple Press, Sarasota, FL. Heppell, S. S., Caswell, H., Crowder, L.B., 2000. Life histories and elasticity patterns: Perturbation analysis for species with minimal demographic data. Ecology 81, 654 665. Heschel, M.S., Paige, K.N., 1995. Inbreeding depression, environmental stress, and populatio n size variation in scarlet gilia ( Ipomopsis aggregata ). Conservation Biology 9, 126 133. Hostetler, J.A., Onorato, D.P., Nichols, J.D., Johnson, W.E., Roelke, M.E., O'Brien, S.J., Jansen, D., Oli, M.K., 2010. Genetic introgression and the survival of Flor ida panther kittens. Biological Conservation 143, 2789 2796. Janis, M.W., Clark, J.D., 2002. Responses of Florida panthers to recreational deer and hog hunting. Journal of Wildlife Management 66, 839 848. Johnson, K.G., Land, E.D., Lotz, M.A., 1997. Flor ida panther genetic restoration and management. Annual Performance Report. Florida Fish and Wildlife Conservation Commission, Tallahassee, Florida.

PAGE 220

220 Johnson, W.E., Onorato, D.P., Roelke, M.E., Land, E.D., Cunningham, M., Belden, C., McBride, R., Jansen, D. Lotz, M.A., Shindle, D.B., Howard, J., Wildt, D.E., Penfold, L.M., Hostetler, J.A., Oli, M.K., O'Brien, S.J., 2010. Genetic restoration of the Florida panther. Science 329, 1641 1645. Kauffman, M.J., Frick, W.F., Linthicum, J., 2003. Estimation of habita t specific demography and population growth for peregrine falcons in California. Ecological Applications 13, 1802 1816. Kautz, R., Kawula, R., Hoctor, T., Comiskey, J., Jansen, D., Jennings, D., Kasbohm, J., Mazzotti, F., McBride, R., Richardson, L., Root, K., 2006. How much is enough? Landscape scale conservation for the Florida panther. Biological Conservation 130, 118 133. Keller, L.F., Arcese, P., Smith, J.N.M., Hochachka, W.M., Stearns, S.C., 1994. Selection against inbred song sparrows during a natura l population bottleneck. Nature 372, 303 388. Laake, J., Rexstad, E., 2010. RMark an alternative approach to building linear models in MARK, in: Cooch, E., White, G.C. (Eds.), Program MARK 'a gentle introduction' 9th Edition. http://www.phidot.org/software/mark/docs/book/ pp. C1 C108. Lambert, C.M.S., Wielgus, R.B., Robinson, H.S., Katnik, D.D., Cruickshank, H.S., Clarke, R., Almack, J., 2006. Cougar population dynamics and viability in the P acific Northwest. Journal of Wildlife Management 70, 246 254. Land, E.D., Garman, D.R., Holt, G.A., 1998. Monitoring female Florida panthers via cellular telephone. Wildlife Society Bulletin 26, 29 31. Land, E.D., Shindle, D.B., Kawula, R.J., Benson, J.F., Lotz, M.A., Onorato, D.P., 2008. Florida panther habitat selection analysis of concurrent GPS and VHF telemetry data. Journal of Wildlife Management 72, 633 639. Laundr, J.W., Hernndez, L., Clark, S.G., 2007. Numerical and demographic responses of pumas to changes in prey abundance: testing current predictions. Journal of Wildlife Management 71, 345 355. Levin, D.A., Francisco Ortega, J., Jansen, R.K., 1996. Hybridization and the extinction of rare plant species. Conservation Biology 10, 10 16. Lewontin, R.C., Birch, L.C., 1966. Hybridization as a source of variation for adaptation to new environments. Evolution 20, 315 336. Lindenmayer, D.B., Possingham, H.P., Lacy, R.C., McCarthy, M.A., Pope, M.L., 2003. How accurate are population models? Lessons from landscape scale tests in a fragmented system. Ecology Letters 6, 41 47.

PAGE 221

221 Logan, K.A., Sweanor, L.L., 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, DC. Loison, A., Sther, B. E., Jerstad, K., R stad, O.W., 2002. Disentangling the sources of variation in the survival of the European dipper. Journal of Applied Statistics 29, 289 304. Lotz, M.A., Land, E.D., Cunningham, M.W., Ferree, B.C., 2005. Florida panther annual report 2004 2005, p. 74 pp. Flo rida Fish and Wildlife Conservation Commission, Tallahassee, Florida, USA. Lukacs, P.M., White, G.C., Watkins, B.E., Kahn, R.H., Banulis, B.A., Finley, D.J., Holland, A.A., Martens, J.A., Vayhinger, J., 2009. Separating components of variation in survival of mule deer in Colorado. Journal of Wildlife Management 73, 817 826. Madsen, T., Shine, R., Olsson, M., Wittzell, H., 1999. Restoration of an inbred adder population. Nature 402, 34 35. Maehr, D.S., 1992. Florida panther, in: Humphrey, S.R. (Ed.), Rare an d endangered biota of Florida, volume I. mammals. University Press of Florida, Gainesville, FL, pp. 176 189. Maehr, D.S., Caddick, G.B., 1995. Demographics and genetic introgression in the Florida panther. Conservation Biology 9, 1295 1298. Maehr, D.S., Cr owley, P., Cox, J.J., Lacki, M.J., Larkin, J.L., Hoctor, T.S., Harris, L.D., Hall, P.M., 2006. Of cats and Haruspices: genetic intervention in the Florida panther. Response to Pimm et al. (2006). Animal Conservation 9, 127 132. Maehr, D.S., Lacy, R.C., Lan d, E.D., Bass, O.L., Hoctor, T., 2002a. Evolution of population viability assessments for the Florida panther: a multiperspective approach, in: Beissinger, S.R., McCullough, D.R. (Eds.), Population viability analysis. University of Chicago Press, Chicago, pp. 284 311. Maehr, D.S., Land, E.D., Roof, J.C., McCown, J.W., 1990. Day beds, natal dens, and activity of Florida panthers, In Annual Conference of Southeast Fish and Wildlife Agencies. pp. 310 318. Maehr, D.S., Land, E.D., Shindle, D.B., Bass, O.L., Hoc tor, T.S., 2002b. Florida panther dispersal and conservation. Biological Conservation 106, 187 197. Maehr, D.S., Roof, J.C., Land, E.D., McCown, J.W., 1989. First reproduction of a panther (Felis concolor coryi) in southwestern Florida, USA. Mammalia 53, 1 29 131.

PAGE 222

222 Maguire, L.A., Servheen, C., 1992. Integrating biological and sociological concerns in endangered species management: Augmentation of grizzly bear populations. Conservation Biology 6, 426 434. Mansfield, K.G., Land, E.D., 2002. Cryptorchidism in Fl orida panthers: Prevalence, features, and influence of genetic restoration. Journal of Wildlife Diseases 38, 693 698. McBride, R., 2002. Florida panther current verified population, distribution and highlights of field work: fall 2001 winter 2002, p. 25. Prepared for Florida Panther SubTeam of MERIT, U.S. Fish and Wildlife Service, Vero Beach, FL, USA. McBride, R.T., McBride, R.T., McBride, R.M., McBride, C.E., 2008. Counting pumas by categorizing physical evidence. Southeastern Naturalist 7, 381 400. McC own, J.W., Maehr, D.S., Roboski, J., 1990. A portable cushion as a wildlife capture aid. Wildlife Society Bulletin 18, 34 36. Menotti Raymond, M., David, V.A., Lyons, L.A., Schffer, A.A., Tomlin, J.F., Hutton, M.K., O'Brien, S.J., 1999. A genetic linkage map of microsatellites in the domestic cat ( Felis catus ). Genomics 57, 9 23. Menotti Raymond, M., David, V.A., Stephens, J.C., Lyons, L.A., O'Brien, S.J., 1997. Genetic individualization of domestic cats using feline STR loci for forensic applications. Jou rnal of Forensic Science 42, 1039 1051. Miller, P.S., Lacy, R.C., 2005. VORTEX: a stochastic simulation of the extinction process. Version 9.50 user's manual. Conservation Breeding Specialist Group (SSC/IUCN), Apple Valley, MN, USA. Min, Y., Agresti, A., 2005. Random effect models for repeated measures of zero inflated count data. Statistical Modelling 5, 1 19. Minch, E., Ruiz Linares, A., Goldstein, D.B., 1995. MICROSAT. http://hpgl.stanford.edu/pro jects/microsat/ Moore, W.S., 1977. Evaluation of narrow hybrid zones in vertebrates. Quarterly Review of Biology 52, 263 277. Moore, W.S., Price, J.T., 1993. Nature of selection in the northern flicker hybrid zone and its implications for speciation theo ry, in: Harrison, R.G. (Ed.), Hybrid Zones and the Evolutionary Process. Oxford University Press, New York, New York, USA, pp. 196 225. Morris, W.F., Doak, D.F., 2002. Quantitative conservation biology: theory and practice of population viability analysis. Sinauer Associates, Sunderland, MA.

PAGE 223

223 Morris, W.F., Doak, D.F., 2005. How general are the determinants of the stochastic population growth rate across nearby sites? Ecological Monographs 75, 119 137. Newton, I., 2004. The recent declines of farmland bird po pulations in Britain: an appraisal of causal factors and conservation actions. Ibis 146, 579 600. Nichols, J.D., Kendall, W.L., Hines, J.E., Spendlow, J.A., 2004. Estimation of sex specific survival from capture recapture data when sex is not always known. Ecology 85, 3192 3201. Nilsson, T., 2004. Integrating effects of hunting policy, catastrophic events, and inbreeding depression, in PVA simulation: the Scandinavian wolf population as an example. Biological Conservation 115, 227 239. O'Brien, S.J., Roelke M.E., Yuhki, N., Richards, K.W., Johnson, W.E., Franklin, W.L., Anderson, A.E., Bass, O.L., Belden, R.C., Martenson, J.S., 1990. Genetic introgression within the Florida panther Felis concolor coryi National Geographic Research 6, 485 494. Oli, M.K., Do bson, F.S., 2003. The relative importance of life history variables to population growth rate in mammals: Cole's prediction revisited. American Naturalist 161, 422 440. Onorato, D.P., Belden, R.C., Cunningham, M.W., Land, E.D., McBride, R.T., Roelke, M.E., 2010. Long term research on the Florida panther ( Puma concolor coryi ): historical findings and future obstacles to population persistence, in: MacDonald, D.W., Loveridge, A.J. (Eds.), Biology and conservation of wild felids. Oxford University Press, Oxfor d, UK, pp. 452 469. Ortego, J., Calabuig, G., Cordero, P.J., Aparicio, J.M., 2007. Egg production and individual genetic diversity in lesser kestrels. Molecular Ecology 16, 2383 2392. Ozgul, A., Armitage, K.B., Blumstein, D.T., Oli, M.K., 2006. Spatiotempo ral variation in survival rates: Implications for population dynamics of yellow bellied marmots. Ecology 87, 1027 1037. Packer, C., Kosmala, M., Cooley, H.S., Brink, H., Pintea, L., Garshelis, D., Purchase, G., Strauss, M., Swanson, A., Balme, G., Hunter, L., Nowell, K., 2009. Sport hunting, predator control and conservation of large carnivores. PLoS One 4, e5941. Pankratz, V.S., de Andrade, M., Therneau, T.M., 2005. Random effects cox proportional hazards model: General variance components methods for time to event data. Genetic Epidemiology 28, 97 109. Pimm, S.L., Bass, O.L., Dollar, L., 2006a. Of cats and Haruspices: genetic intervention in the Florida panther. Ockham and Garp. Reply to Maehr et al.'s (2006) response to Pimm et al. (2006). Animal Conserva tion 9, 133 134.

PAGE 224

224 Pimm, S.L., Dollar, L., Bass, O.L., 2006b. The genetic rescue of the Florida panther. Animal Conservation 9, 115 122. Pollock, K.H., 1981. Capture recapture models allowing for age dependent survival and capture rates. Biometrics 37, 521 5 29. Pollock, K.H., Winterstein, S.R., Bunck, C.M., Curtis, P.D., 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53, 7 15. Pritchard, J.K., Stephens, M., Donnelly, P., 2000. Inference of population s tructure using multilocus genotype data. Genetics 155, 945 959. R Development Core Team, 2010. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Ralls, K., Ballou, J., 1983. Extinction: lesson s from zoos, in: Schonewald Cox, C.M., Chambers, S.M., MacBryde, B., Thomas, W.L. (Eds.), Genetics and Conservation: A Reference for Managing Wild Animal and Plant Populations. Benjamin/Cummings, Menlo Park, California, pp. 164 184. Rand, D.M., Harrison, R .G., 1989. Ecological genetics of a mosaic hybrid zone mitochondrial, nuclear, and reproductive differentiation of crickets by soil type. Evolution 43, 432 449. Reed, J.M., Mills, L.S., Dunning, J.B., Menges, E.S., McKelvey, K.S., Frye, R., Beissinger, S .R., Anstett, M. C., Miller, P., 2002. Emerging issues in population viability analysis. Conservation Biology 16, 7 19. Rhymer, J.M., Simberloff, D., 1996. Extinction by hybridization and introgression. Annual Review of Ecology and Systematics 27, 83 109. Richter, A.R., Labisky, R.F., 1985. Reproductive dynamics among disjunct white tailed deer herds in Florida. Journal of Wildlife Management 49, 964 971. Robinson, H.S., Wielgus, R.B., Cooley, H.S., Cooley, S.W., 2008. Sink populations in carnivore manageme nt: Cougar demography and immigration in a hunted population. Ecological Applications 18, 1028 1037. Roelke, M.E., Forrester, D.J., Jacobson, E.R., Kollias, G.V., Scott, F.W., Barr, M.C., Evermann, J.F., Pirtle, E.C., 1993a. Seroprevalence of infectious di sease agents in free ranging Florida panthers ( Felis concolor coryi ). Journal of Wildlife Diseases 29, 36 49. Roelke, M.E., Martenson, J.S., O'Brien, S.J., 1993b. The consequences of demographic reduction and genetic depletion in the endangered Florida pan ther. Current Biology 3, 340 350.

PAGE 225

225 Root, K.V., 2004. Florida panther ( Puma concolor coryi ): using models to guide recovery efforts, in: Akakaya, H.R., Burgman, M.A., Kindvall, O., Wood, C.C., Sjogren Gulve, P., Hatfield, J.S., McCarthy, M.A. (Eds.), Specie s Conservation and Management: Case Studies. Oxford University Press, New York, New York, USA, pp. 491 504. Rosenfield, J.A., Nolasco, S., Lindauer, S., Sandoval, C., Kodric Brown, A., 2004. The role of hybrid vigor in the replacement of Pecos pupfish by i ts hybrids with sheepshead minnow. Conservation Biology 18, 1589 1598. Schortemeyer, J.L., Maehr, D.S., McCown, J.W., Land, E.D., Manor, P.D., 1991. Prey management for the Florida panther: A unique role for wildlife managers, In Transactions of the 56th N orth American Wildlife and Natural Resource Conference. pp. 512 526. Seal, U.S., 1994. A plan for genetic restoration and management of the Florida panther ( Felis concolor coryi ). Conservation Breeding Specialist Group, Apple Valley, Minnesota, USA. Shield s, W.M., 1993. The natural and unnatural history of inbreeding and outbreeding, in: Thornhill, N.W. (Ed.), The natural history of inbreeding and outbreeding: theoretical and empirical perspectives. The University of Chicago Press, Chicago, Illinois, USA, p p. 143 169. Shull, G.H., 1908. The composition of a field of maize. Reports of the American Breeders Association 4, 296 301. Slate, J., Kruuk, L.E.B., Marshall, T.C., Pemberton, J.M., Clutton Brock, T.H., 2000. Inbreeding depression influences lifetime bre eding success in a wild population of red deer ( Cervus elaphus ). Proceedings of he Royal Society Of London Series B Biological Sciences 267, 1657 1662. Slattery, M., Kamel, H.N., Ankisetty, S., Gochfeld, D.J., Hoover, C.A., Thacker, R.W., 2008. Hybrid vigo r in a tropical Pacific soft coral community. Ecological Monographs 78, 423 443. Stahl, J.T., Oli, M.K., 2006. Relative importance of avian life history variables to population growth rate. Ecological Modelling 198, 23 39. Stoner, D.C., Wolfe, M.L., Choate D.M., 2006. Cougar exploitation levels in Utah: Implications for demographic structure, population recovery, and metapopulation dynamics. Journal of Wildlife Management 70, 1588 1600. Sullivan, J.D., Jr., 2004. Florida's endangered species, threatened sp ecies, and species of special concern. Florida Fish and Wildlife Conservation Commission, Tallahassee, Florida, USA.

PAGE 226

226 Sweanor, L.L., Logan, K.A., Hornocker, M.G., 2000. Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biolo gy 14, 798 808. Taylor, S.K., Buergelt, C.D., Roelke Parker, M.E., Homer, B.L., Rotstein, D.S., 2002. Causes of mortality of free ranging Florida panthers. Journal of Wildlife Diseases 38, 107 114. Templeton, A.R., 1999. Coadaptation, local adaptation, and outbreeding depression, in: Meffe, G.K., Carroll, C.R. (Eds.), Principles of Conservation Biology. Sinauer Associates, Sunderland, Massachusetts, USA, pp. 171 172. Thatcher, C.A., van Manen, F.T., Clark, J.D., 2009. A habitat assessment for Florida panther population expansion into central Florida. Journal of Mammalogy 90, 918 925. Therneau, T.M., Grambsch, P.M., 2000. Modeling survival data: extending the Cox model. Springer, New York. Tinker, M.T., Doak, D.F., Estes, J.A., Hatfield, B.B., Hatfield, B.B., Staedler, M.M., Bodkin, J.L., 2006. Incorporating diverse data and realistic complexity into demographic estimation procedures for sea otters. Ecological Applicatio ns 16, 2293 2312. Tuljapurkar, S.D., Horvitz, C.C., Pascarella, J.B., 2003. The many growth rates and elasticities of populations in random environments. American Naturalist 162, 489 502. USFWS, 2006. Florida panther recovery plan ( Puma concolor coryi ), 3r d revision, p. 216. United States Fish and Wildlife Service, Atlanta, GA, USA. Vaida, F., Blanchard, S., 2005. Conditional Akaike information for mixed effects models. Biometrika 92, 351 370. Vrijenhoek, R.C., 1994. Genetic diversity and fitness in small p opulations, in: Loeschcke, V., Tomiuk, J., Jain, S.K. (Eds.), Conservation genetics. Birkhuser Verlag, Boston, pp. 37 53. Westemeier, R.L., Brawn, J.D., Simpson, S.A., Esker, T.L., Jansen, R.W., Walk, J.W., Kershner, E.L., Bouzat, J.L., Paige, K.N., 1998. Tracking the long term decline and recovery of an isolated population. Science 282, 1695 1698. White, G.C., 2000. Population viability analysis: data requirements and essential analyses, in: Boitani, L., Fuller, T.K. (Eds.), Research techniques in animal ecology: controversies and consequences. Columbia University Press, New York, NY USA, pp. 288 331.

PAGE 227

227 White, G.C., Burnam, K.P., Barker, R.J., 2009. Evaluation of a Bayesian MCMC random effects inference methodology for capture mark recapture data, in: Thomso n, D.L., Cooch, E.G., Conroy, M.J. (Eds.), Modeling demographic processes in marked populations. Springer, New York, NY, pp. 1119 1127. White, G.C., Burnham, K.P., 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46, S 120 S139. Whitlock, M.C., Ingvarsson, P.K., Hatfield, T., 2000. Local drift load and the heterosis of interconnected populations. Heredity 84, 452 457. Williams, B.K., Nichols, J.D., Conroy, M.J., 2002. Analysis and management of animal populations. Academ ic Press, San Diego, CA, USA. Wolf, D.E., Takebayashi, N., Rieseberg, L.H., 2001. Predicting the risk of extinction through hybridization. Conservation Biology 15, 1039 1053. Zedrosser, A., Dahle, B., Stoen, O.G., Swenson, J.E., 2009. The effects of primip arity on reproductive performance in the brown bear. Oecologia 160, 847 854.

PAGE 228

228 BIOGRAPHICAL SKETCH Jeffrey Hostetler was born in Cheverly, Maryland in 1974. He received a B. A. in b iology from Oberlin College in 1996 (with minors in e nvironmental s tud ies and c omputer s cience). He worked at Patuxent Wildlife Research Center for several years develop ing web applications and databases for the Breeding Bird Survey and other monitoring programs. He ha s done field work in Olympic National Forest, Washingt on, Tongass National Forest, Alaska, the Dolomite Mountains of Italy, and the outer banks of North Carolina among other places. He e arned an M.S. from North Carolina State University in 2004 in z oology (with minors in s tatistics and b iomathematics), wher e his modeling the spring migration of semipalmated sandpipers. From 2004 through 2006, he worked at the University of Florida for Madan Oli and Franklin Percival, modeling the survival and population dynamics of northern bobwhite and Florida black bears. In 2006, he started as a Ph.D. student at the University of Florida under the direction of Madan Oli. He received his Ph.D. in wildlife ecology and conservation from the Univ ersity of Florida in the fall of 2010.