BALANCING WHITE-TAILED DEER ECOLOGY WITH MICHIGAN NATIONAL GUARD TRAINING AT FORT CUSTER TRAINING CENTER IN AUGUSTA, MI By Joel T. Humphries A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTERS OF SCIENCE Fisheries and Wildlife 2011 ABSTRACT BALANCING WHITE-TAILED DEER ECOLOGY WITH MICHIGAN NATIONAL GUARD TRAINING AT FORT CUSTER TRAINING CENTER IN AUGUSTA, MI By Joel T. Humphries The Michigan Department of Military and Veteran Affairs (MDMVA) manage an array of natural resources at Ft. Custer Training Center (FCTC) in Augusta, Michigan, and their Integrated Natural Resources Management Plan (INRMP) promulgates management goals of ecosystem restoration and rehabilitation. White-tailed deer (Odocoileus virginianus) herbivory can influence the forest structure and composition. The hunter harvest period of white-tailed deer of approximately 75 days (the length of the season is subject to some minor annual variations) cannot take place since the FCTC functions as a military installation and its‟ needs dictate limited access to hunters, confounding the MDMVA‟s ability to meet their management goals. I evaluated the effectiveness of the current 5-day hunter harvest period by quantifying deer herbivory effects on structure and composition of forest types, and developed a suite of deer population indices. I captured, aged, ear-tagged and radio-collared 66 deer during winter from 2004 to 2008, and 14 neonatal fawns during spring in 2006 and 2007. The annual survival rate varied among the groups (adult females = 0.756, adult males = 0.493, yearling females = 0.443, yearling males = 0.379, fawns = 0.289). The short hunter harvest period is an effective and integral component of the ecosystem restoration and rehabilitation efforts of FCTC. In memory of my loving father, Robert Earl Humphries iii ACKNOWLEDGEMENTS I am grateful for the funding provided by the Michigan Department of Military (MDMVA) and Veteran Affairs that made this research project possible. I would like to thank the MDMVA personnel for their assistance with logistics and additional support in the field. Jonathan Edgerly, Michele Richards, Gregg Huntington, Jim Langeveld, and John Mitchell assisted in my endeavors at Ft. Custer in multiple ways as I dragged them out into the field to sample an exclosure, catch a deer, or move clover traps around. I am appreciative of how the military personnel at Ft. Custer allowed me access to the base and informed me of military operations that could affect my field research. In the Department of Fisheries of Wildlife, I would like to thank my lab mates and fellow students Dan Linden, Dan Walsh, Ali Felix, Jordan Burroughs, Kristen Schmitt, Katie Button, Tim Hiller, Jeanette McGuire, and Jen White for providing laughs, support, and stories of their own trials and tribulations. The efforts of my field assistants Kim Borland, Dan Traynor, Autumn Foutch, Justin Stahl, Ayaka Asada, Richard Whitfield, and Aaron Young made the capture of deer and vegetation sampling possible. I would like to thank Darian Muzo for sharing his experience trapping deer. I am grateful to Dr. Scott Winterstein for his support, mentoring and patience throughout my graduate school experience. I owe thanks to Dr. Rique Campa for assistance with the sampling protocol and study design of the deer iv exclosures. I like to say thank you to Dr. Jim Sikarskie for being a valuable committee member. I appreciate Dr. Kelly Millenbah‟s assistance with GIS and compositional analysis, and Dr. Shawn Riley for assistance with hunter surveys. I am thankful for the guidance that Dr. Bill Taylor provided. I would like to say thank you to Michigan Department of Natural Resources and Environment personnel Mark Bishop, Christine Hanaburgh, Tom Cooley, Sarah Mayhew, John Lerg, and Steve Schmitt. My family has provided never-ending support and encouragement, for which I am extremely grateful. So, thank you Michelle, Mark, Jenny, Becky, Bob, and Roberta for everything. v TABLE OF CONTENTS LIST OF TABLES…………………………………..………………………………...VIII LIST OF FIGURES…………………………………..…………………………….….XV LIST OF ABBREVIATIONS...……...………………………………….….…………XXI Chapter 1: INTRODUCTION………………………………………………………….1 Brief History and Site Description of Ft. Custer……………………………...1 Historical Look at the FCTC Deer Herd………………………………………6 The Impetus for Natural Resource Management at FCTC…………………8 Objectives………………………………………………………………………14 Chapter 2: METHODOLGY..……………………….……………………………......15 Spring Capture…………………………………………………………………15 Winter Capture…………………………………………………………………16 Monitoring Radio-Collared Deer……………………………………………..20 Habitat Use and Resource Selection………………………………………..22 Estimation of Population Parameters………………………………………..24 Population Size Estimate……………………………………...……..25 Forward Looking Infrared……………………………………...……..26 White-Tailed Deer Exclosures………………………………………………..28 General Overview of Vegetation Sampling…………………………………31 Chapter 3: RESULTS...……..…………...…………………………………………..35 Winter Capture…………………………………………………………………35 Spring Capture…………………………………………………………………37 Survival………………………………………………………………………….39 Spring Captured Neonates……………..…………………………….39 Winter Captured Deer…………………..…………………………….40 Home Range Estimates……………………………………………………….46 Habitat Use……………………………………………………………….…….47 Population Estimates…………………….……………………………………50 Spotlight Survey…………………………………..……………………50 Sex-Age-Kill Model………………...………………………………….51 Forward Looking Infrared…………...………………………………..54 Hunt Biodata……………………………………………………………………59 Vegetation Sampling……………………………………………………..……61 Chapter 4: DISCUSSION…..……...…………………………………………………79 Winter and Spring Capture……………………………………………………79 Survival and Fate Assessment………………………………………………81 SAK Model……………………………………………………………………...83 vi Forward Looking Infrared (FLIR)……………………………………………..84 Spotlight Survey…………………………..………………………....………...86 Hunt Biodata……………………………………………………………………86 Resource Utilization and Habitat Use……………………………………….88 Vegetation Sampling…………………………………………………………..91 Relative Deer Density…………………………………………………………92 Hunter Participation……………………………………………………………95 Management Implications…………………………………………………….97 Management Recommendations…………………………………………...108 APPENDICES APPENDIX A Historical Biodata of the deer herd at FCTC………...……..……..111 APPENDIX B Coordinates of the exclosure sites and paired control units and vegetation sampling data sheets………………………………115 APPENDIX C Home Range Estimates and Locations of Radio-Collared Deer………………………………………………………………....…120 LITERATURE CITED.……………………………………………………….………110 vii LIST OF TABLES Table 3.1. The number of deer captured at FCTC during the winters of 2004 to 2008………......……………….….36 Table 3.2. Capture success rate of winter captured deer at FCTC based on number of trap days and total number of captures.…………………..………….……37 Table 3.3. The average age and weights of spring captured neonatal fawns at FCTC in 2006 and 2007.…..……37 Table 3.4. Capture success based on neonates observed and number of personnel hours per capture during the spring capture at FCTC.…………………...………….37 Table 3.5. Fate of neonatal captured fawns at FCTC from 2006 to 2007………………………………………………….40 Table 3.6. Annual and daily survival rate estimates of radio-collared deer at FCTC based on the Mayfield Method………………………………………………….43 Table 3.7. Fate of female winter captured radio-collared deer at FCTC……………………………………………………44 Table 3.8. Fate of male winter captured radio-collared deer at FCTC…………………………………….…………..….45 Table 3.9. The mean core home range size of radio-collared deer in hectares by age and gender. .………………47 Table 3.10. The mean overall home range size of radio-collared deer in hectares by age and gender. The overall home range represents the area where radio-collared deer have 95% chance of ranging……..…..………..47 Table 3.11. Rankings of the preferred vegetation types used by radio-collared deer at FCTC………………………….49 Table 3.12. SAK model population estimates……..……………..53 viii Table 3.13 SAK Pym model annual populations estimates of age classes by gender with variance……………..………53 Table 3.14 The number of deer per square mile and kilometer with the SAK model………………….………………...54 Table 3.15. The number of deer sighted during the March 2007 FLIR survey………………………………….…..….…56 Table 3.16. The number of deer sighted from the November 2007 FLIR survey………………………….…….……56 Table 3.17. The deer density at FCTC and FCRA based on both FLIR surveys……………………………….…….……56 Table 3.18. The deer density estimate from the two forward looking infrared surveys conducted in 2007. There are low and high adjustments in the deer density to account for a slightly diminished detection rate associated with FLIR in deciduous cover types…....55 Table 3.19. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means……….………………………65 Table 3.20. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means……………………….………65 Table 3.21. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means……….………………………65 Table 3.22. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means…………………………………..………………66 ix Table 3.23. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means………………………………66 Table 3.24. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means………………………………66 Table 3.25. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means…………….…………………67 Table 3.26. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means…………………….…………67 Table 3.27. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means…….…………………………67 Table 3.28. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means……………………….………68 Table 3.29. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means……….………………………68 Table 3.30. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. x A standard t-test was used to test for differences between the two means………….……………………68 Table 3.31. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means………………………………69 Table 3.32. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means………………………………69 Table 3.33. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means………………………………69 Table 3.34. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means……………….………………70 Table 3.35. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means………………………………70 Table 3.36. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means………………………………70 Table 3.37. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means………………………………71 Table 3.38. Comparison of means between the paired control xi unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means.………………………………71 Table 3.39. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means………………………………71 Table 3.40. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means………………………………72 Table 3.41 Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means………………………………72 Table 3.42. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means………………………………72 Table 3.43. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………73 Table 3.44. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………73 Table 3.45. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………73 Table 3.46. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………74 Table 3.47. The annual mean of stem density, with a standard xii t-test to compare differences between the exclosure and its paired control site……………………………74 Table 3.48. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………74 Table 3.49. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………75 Table 3.50. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………75 Table 3.51. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………75 Table 3.52. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………76 Table 3.53. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………76 Table 3.54. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………76 Table 3.55. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………77 Table 3.56. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………77 Table 3.57. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………77 Table 3.58. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………78 xiii Table 3.59. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site……………………………78 Table A.1. The results from a Tukey studentized range test comparison of yearling bucks total point counts between FCTC and surrounding counties from 20012004 showing the difference of the means between areas with a 95% confidence intervals. FCTC was significantly less than the four other areas………...113 Table A.2. The results from a Tukey studentized range test comparison of yearling bucks mean beam diameters between FCTC and surrounding counties from 20012004 showing the difference of the means between areas with a 95% confidence intervals. FCTC was significantly less than the four other areas………...113 Table B.1. Locations and forest type of each exclosure and paired control site. The locations listed are based on the NAD 83 datum and projected in UTM 16T. Each site's forest type is listed as a mixed hardwood or as oak hardwood………………………………………..116 xiv LIST OF FIGURES Figure 1.1. Location of Ft. Custer Training Center in southwest Michigan. ……….……………………………..…………………2 Figure 1.2. Layout of the 9 training areas at FCTC. The majority of buildings are shown in the cantonment zone north of area 1. Area 9 is off limits to the general public and hunting at all times because of the potential for UXO…..……………………………………………………..……5 Figure 1.3. A chronological summary of events that shaped FCTC…...12 Figure 2.1. Clover trap distribution during winter capture seasons........18 Figure 2.2. A map of the exclosure site locations at FCTC……….…....33 Figure 2.3. A schematic layout of either an exclosure or a paired control site for vegetation sampling. There are five transect lines marked as T1, T2 … T5 to use for the vegetation density in the understory, midstory, and overstory. The rectangles marked A, B, and C are the nested plots that estimate species composition and frequency in the understory and midstory. The midstory nested plot is 1m x 5m, while the understory plot is a subplot of the midstory and measures 1m x 2m…………………………………………………………34 Figure 3.1. The population trend of deer from the SAK Pym model……52 Figure 3.2. The locations of deer from the March 2007 forward looking infrared flight……………………………………………………57 Figure 3.3. Locations of deer groups from the forward looking infrared survey on the evening of 13 November and morning of 14 November 2007. An asterisk indicates the locations from the 13th, while the locations from the 14th are indicated by a dot. Fog interrupted the flight after completion of approximately half of the survey………..……………………………..………58 Figure 3.4. The annual trend in hunt biodata of the mean antler point counts by age…………………………………………………..60 Figure 3.5. The annual trend of means with respect to beam diameter by age. No four-year-old deer were harvested in 2008 or 2009…………………………………………….…..61 xv Figure 4.1. Relative deer density (RDD) displayed on McCullough's (1984) graph. DeCalasta and Stout (1997) approximated the location of RDDS, the level associated with sustaining biodiversity, RDDT, the level at which timber productivity is sustained, and RDDI, the level associated with maximum sustained yield of deer numbers for harvest………………..93 Figure 4.2. The annual trend of total hunter days in the public hunt at FCTC and number of deer harvested (from data provided by the MDMVA)……………………………………………………96 Figure 4.3. The hunter efficiency at FCTC based on the total number of hunter days and deer harvested. No hunter efficiency estimate is available for 2007 because the total number of hunter days is unknown……………………………………….97 Figure 4.4. The ecological and perceived impacts of deer on stakeholders at high and low abundance levels with respect to stakeholder satisfaction and deer density. The stakeholder satisfaction is relative to the perception that individuals or groups of individuals might have about the deer herd (e.g. too few deer, too many deer because they are causing damage to the ecosystem). Whereas the relative deer density is in terms of biological levels being low or high with respect to resource availability acting as a constraint on population growth. Other stakeholders may have different views than those expressed by the deer hunters………..………………………………………………101 Figure 4.5. Deer seen as a function of deer density in five Midwestern data sets; the figure is a reproduction from Van Deelen and Etter (2003)……………………………………..……….103 Figure 4.6. Deer killed as a function of deer density in five Midwestern data sets; the figure is a reproduction from Van Deelen and Etter (2003)……………………………………..……....104 Figure 4.7. Hypothetical relationship between hunter effort, perceptions, and deer density. If hunter perception of deer density is based on an assumed linear relationship but is in reality hyperbolic, then for any given level of effort hunters will perceive fewer deer than there actually are. Moreover, this bias will accelerate as deer density decreases. The figure is a reproduction from Van Deelen and Etter (2003)……….105 xvi Figure A.1. The temporal trend among the five areas of interest in regards to average number of points for bucks age = 1.5 years…………………………………………………………...114 Figure A.2. The temporal trend of average beam diameter among the five areas of interest for bucks age=1.5 years…………….114 Figure B.1. The vertical cover data sheet used during vegetation sampling……………………………………………………….117 Figure B. 2. The understory stem density data sheet…………………..118 Figure B.3. The stem density data sheets for the midstory and overstory….…………………………………..……………….119 Figure C.1. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.004………………………………………..………………121 Figure C.2. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.010………………………………………………………..121 Figure C.3. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.020……………………………………………..…………122 Figure C.4. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.029……………………………………………..…………122 Figure C.5. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.039..………………………………………………………123 Figure C.6. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.049……………………………………………………..…123 Figure C.7. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.060……………………………………………………..…124 Figure C.8. The fixed kernel home range estimate with the 95% xvii and 50% probability of occupancy for radio frequency 150.069……………………………………………..…………124 Figure C.9. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.090………………………………………..………………125 Figure C.10. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.173………………………………………………………..125 Figure C.11. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.273……………………………………………..…………126 Figure C.12. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.280……………………………………………..…………126 Figure C.13. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.039..………………………………………………………127 Figure C.14. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.290-1…………………………………………………..…127 Figure C.15. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.300……………………………………………………..…128 Figure C.16. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.310……………………………………………..…………128 Figure C.17. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.330………………………………………………………..129 Figure C.18. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.334……………………………………………..…………129 Figure C.19. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.340……………………………………………..…………130 xviii Figure C.20. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.353……………………………………………..…………130 Figure C.21. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.370..………………………………………………………131 Figure C.22. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.380……………………………………………………..…131 Figure C.23. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.390……………………………………………………..…132 Figure C.24. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.400……………………………………………..…………132 Figure C.25. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.420-2…………………………………………..…………133 Figure C.26. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.430………………………………………………………..133 Figure C.27. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.440……………………………………………..…………134 Figure C.28. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.470……………………………………………..…………134 Figure C.29. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.480-1…………………………………………..…………135 Figure C.30. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.480-2……………………………………………………..135 Figure C.31. The fixed kernel home range estimate with the 95% xix and 50% probability of occupancy for radio frequency 150.490……………………………………………………..…136 Figure C.32. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.519……………………………………………………..…136 Figure C.33. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.530……………………………………………..…………137 Figure C.34. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.540……………………………………………..…………137 Figure C.35. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.560……………………………………………..…………138 Figure C.36. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.580-1……………………………………………………..138 Figure C.37. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.580-2…………………………………………………..…139 Figure C.38. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.590-2…………………………………………………..…139 Figure C.39. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.630……………………………………………..…………140 Figure C.40. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.710……………………………………………..…………140 xx LIST OF ABBREVIATIONS CEQ Council for environmental quality CFR Code of federal regulations FCRA Fort Custer Recreation Area FCTC Fort Custer Training Center INRMP Integrated natural resources management plan MNFI Michigan National Features Inventory MDMVA Michigan Department of Military and Veteran‟s Affairs MDNR Michigan Department of Natural Resources NEPA National Environmental Policy Act RDD Relative deer density USC United States code UXO Unexploded ordinance xxi CHAPTER 1: INTRODUCTION The Michigan Army National Guard base, Fort Custer Training Center (FCTC), in southwest Michigan functions as a training site for the Michigan National Guard, and other reserve components of the armed forces (Figure 1.1). Training exercises at FCTC include a variety of small arms, land navigation, light armor, and urban assault procedures, and an assortment of other tasks related to military preparedness. The Michigan Department of Military and Veterans Affairs (MDMVA) under license from the United States Army Corps of Engineers (USACE) is responsible for management of natural resources at FCTC which sits on 3,064 hectares. In accordance with a wide array of environmental laws and regulations, MDMVA has implemented the Integrated Natural Resources Management Plan and Environmental Assessment (INRMP) at FCTC. INRMP allows the Michigan National Guard (MING) to achieve the training standards set to ascertain the highest levels of military preparedness. Although it is not explicitly outlined in the INRMP, the white-tailed deer (Odocoileus virginianus) herd must be managed to allow the MDMVA to reach goals put forth in the INRMP. Brief History and Site Description of Ft. Custer Settlement of the land around FCTC began in the 1830‟s, converting the landscape to an agricultural landscape from oak-hickory (Quercus spp. and Carya spp.) hardwood forests, and prairie communities (Comer et al. 1995). With the onset of direct U.S. involvement in WWI the American government 1 sought the establishment of an area suitable for training large numbers of soldiers. The U.S. government secured 3,359 hectares from private land holders to found FCTC and trained 36,000 “doughboys” (INRMP 2001). In 1940 the U.S. government expanded the FCTC to 5,827 hectares. After WWII the role of FCTC diminished, and in the late 1960‟s and 1970‟s portions of land were sold, decreasing its size to 3,064 hectares (INRMP 2001). However, the conversion from small farm homesteads to a military installation allowed large portions of the landscape to proceed through successional stages. Figure 1.1. Location of Ft. Custer Training Center in southwest Michigan. As of a 1998/1999 forest inventory at FCTC, 77% of the land occurs in some seral stage in one of the following forest types: mixed hardwoods, oak hardwoods, and mixed oak community (INRMP 2001). An additional 15% of the area at FCTC has been classified as wetlands and aquatic communities (Legge 2 et al. 1995). The Michigan Natural Features Inventory (MNFI) conducted an intensive investigation of FCTC‟s flora and fauna. That resulted in the identification of 815 species of plants, which is roughly a third of the species found in Michigan (Herman et al. 1996). A total of 13 state listed or special concern plants were collected, and an additional 5 state listed or special concern plants were deemed likely to occur at FCTC (Legge et al. 1995). Interspersed across FCTC are 15 natural plant community types, 7 of which are identified by the MNFI as high quality rare communities because there is strong evidence of a high degree of native species richness and diversity (Legge et al. 1995). These high quality areas have been determined to also play a crucial role in preserving biotic diversity at regional and local scales, and four of the communities are considered rare globally (INRMP 2001). For training purposes, FCTC has been subdivided into 9 sequentially numbered training areas (Figure 1.2). An additional 32-hectare cantonment area lies in the northern portion of the installation and contains the majority of the buildings. Areas 8 and 9 (combined = 1,059 hectares) are used for firing live rounds of ammunition of various small caliber weapons on fourteen different ranges. Areas 8 and 9 are also referred to as the impact area because live ammunition is fired in those areas. Area 9 is distinct from other areas because of the potential for UXO (unexploded ordinance), which limits access by FCTC personnel and members of the public. Area 6 has a clear-cut of 95 hectares referred to as the tank range that provides a place to conduct M1-A1 Abrams tank and other mechanized infantry training maneuvers (Figure 1.2). Natural 3 resource management goals for the tank range include planting a variety of native prairie grasses and forbs that includes big bluestem (Andropogon gerardii) and switchgrass (Panicum virgatum). The north section of area 3 has several small clearings utilized as landing zones for helicopters. The remaining areas are used primarily by foot soldiers for other assorted operations, and bivouacking (INRMP, 2001). The areas surrounding FCTC vary in landscape composition, but a large portion offers suitable habitat for white-tailed deer. The Fort Custer Recreation Area (FCRA) managed by the Michigan Department of Natural Resources and Environment (MDNRE) lies on FCTC‟s northwest perimeter, and was originally a part of FCTC until it was deeded to the State of Michigan in 1971. The 1,227 hectares of land that makes up FCRA is similar to FCTC with large contiguous tracts of closed canopy forests consisting of oak-hickory hardwoods, mixed hardwoods, and remnant prairie communities. On FCTC‟s eastern perimeter is the Ft. Custer Industrial Park, also once a part of FCTC. The industrial park encompasses 1,190 hectares of land that varies in levels of development. Approximately 243 hectares of the industrial park serves as a buffer to FCTC. It will not be developed, and has a combination of oak forest types, wetlands and aquatic areas. The remainder of the industrial park has a mixture of light 4 Figure 1.2. Layout of the 9 training areas at FCTC. The majority of buildings are shown in the cantonment zone north of area 1. Area 9 is off limits to the general public and hunting at all times because of the potential for UXO. 5 Industrial sites, wooded areas, oak openings, and wetlands. The south end of FCTC is bounded by I-94 and agricultural fields. The west border consists of a fragmented landscape with a low density residential area, small scale agricultural fields, small patches of woods, and the Kalamazoo River which has a corridor of lowland habitat. Immediately to the north of FCTC is the 253 hectares Fort Custer National Cemetery and Veterans Administration complex, which has a mixture of small patches of oak hardwoods (Quercus spp.) and large manicured grass openings. Battle Creek is the nearest developed city with its boundary 1.1 miles to the northeast of FCTC and a population of 53,364 (2000 U.S. Census). The Kalamazoo-Portage area has a population of 143,717 and lies 12 miles to the west of FCTC. Historical Look at the FCTC Deer Herd Deer hunting ceased in the early 1970s when FCTC implemented a limited public access policy for security. With no public access and no deer hunting on the grounds of FCTC, the deer herd reached or exceeded carrying capacity by the mid 1980s according to biologists with the Michigan Department of Natural Resources (in January 2010 the MDNR became the MDNRE). During a wetland survey in the mid 1980s at neighboring FCRA, numerous over-winter deaths of deer were observed around the area of Eagle Lake (Figure 1.2). Biologists with the MDNR concluded that the deer herd at FCTC was overabundant, and the over-winter deaths were a spillover effect. A hunt for National Guard service members was established in 1985 at FCTC in which field 6 dressed half-year-old deer frequently weighed less than 50 pounds. This prompted the MDNR to further assess the situation. MDNR biologist John Lerg and several others noted an observable browse line in the hardwoods understory, deer trails that were rutted from a very high level of use and a dense population of deer. Lerg also observed a large number of emaciated deer at FCTC. These factors, along with pellet group counts led to an estimate of 60 deer per square mile on the grounds of FCTC (Lerg personal communication). To address the issue of overabundance in the deer herd a public hunt was established in 1986 as a means to manage the herd. However, the number of days the public could access FCTC, since it is a military installation, was limited to five days. With the resumption of hunting the MDNR felt is was necessary to document the reduced weight in the young-of-the-year deer, and other body indices that are considered reliable indicators for the health of a deer herd (Lerg personal communication). All deer harvested at FCTC were subject to a mandatory deer check in which body weight, and hind leg length were recorded, in addition to the typical age, sex, beam diameter, and total points biodata gathered by the MDNR. Antler characteristics such as beam diameter and number of points are correlated with the physical condition of a deer population (McCullough 1982, Severinghaus and Moen 1983, Rasmussen 1985). I compared the historical biodata from (1987-2004, the year prior to the onset the study) from FCTC and the counties that surround it (Kalamazoo and Calhoun), and other nearby counties with large amounts of closed canopy forest with similar cover types 7 (Barry and Allegan) to test for significant differences in indices of herd condition between FCTC and these four areas. I focused on the yearling age class to examine indices of herd condition because of the correlation of antler development and the added burden of body growth that requires sufficient food intake and nutrition (Moen and Severinghaus 1981, Harder and Kirkpatrick 1993). If resources are limited, this will be reflected in antler development and is especially apparent in yearlings (Severinghaus et al. 1950, Riney 1955, Rasmussen 1985). Yearlings with relatively large antler characteristics are indicative of a deer in good physical condition with resources readily available. The results of the deer biodata analysis prior to the onset of this study indicate the physical characteristics of yearlings at FCTC are significantly poorer than in the aforementioned areas. A summary of the results is provided in Appendix A. The Impetus for Natural Management at FCTC The foundation of the INRMP stems primarily from The Natural Resources Management Lands Act, Title 16 of the United States Code Section 670, commonly referred to as the Sikes Act; The National Environmental Policy Act of 1969 (NEPA) (42 U.S.C 4321 et seq.); and Natural Resource Management, Army Regulation 200-3. The Sikes Act dictates that military installations must initiate management plans to conserve and rehabilitate natural resources on military installations. The INRMP adheres to the regulations set forth in NEPA, in which the Council on Environmental Quality (CEQ) has established the key requirements in the Code of Federal Regulations (40 CFR, Parts 1500-1508). 8 The goal of NEPA is to ensure the restoration or enhancement of the environment through well informed decisions and lays out the basic principles to protect the countries‟ environment. As a result of being in line with NEPA and the Sikes Act, the INRMP natural resource objectives include rehabilitation and restorations of ecosystems to a state that would have been observed before Euro-American settlement and agriculture practices that altered the landscape. Southwest Michigan‟s vegetation prior to Euro-American settlement in the 1800s differed considerably from the land-use matrix of cities and farmland prevalent today. Accounts of early European settlers depict forests dominated by large contiguous stands of oak and hickory, oak savanna, and prairie (Kenoyer 1930, Brewer 1984, Comer 1998). After Euro-American settlement FCTC‟s landscape was used for farming, except for hillsides with a slope greater than 40% (Stoynoff 1983). The establishment of FCTC resulted in the cessation of agricultural practices, and the landscape shifted from agriculture fields to a majority of closed canopy forest (Chapman 1984). FCTC offers a rare opportunity for restoration of the ecosystems that once dominated the landscape of Southwest Michigan and is relatively rare in terms of size, species richness, and species diversity. FCTC is one of three places in southwest Michigan with large amounts of contiguous closed canopy forest (Michigan Natural Features Inventory, Lansing, Michigan, USA) INRMP goals include management for ecosystem restoration and rehabilitation. However, the effects of deer herbivory can drastically change the structure and composition of a forest ecosystem (Alverson 1988, Tilghman 1989, 9 Stomayer and Warren 1997, Strole 1988, Potvin et al. 2003). Strole and Anderson (1992) concluded that deer had a browse preference for white oak (Quercus rubra) and shagbark hickory (Carya ovata) and this preferential browse will cause greater scarcity in the overall abundance of oaks and hickorys in an ecosystem. Healy (1997), conducted a study in contiguous oak forest in central Massachusetts and when deer density in a refuge area exceeded 10 deer/km2 he found few saplings and no oak seedlings >100cm tall. When deer densities range from 3 to 6 deer/km2 he noted many small trees and seedlings distributed throughout his study site. Deer affect recruitment of tree species into the canopy and they can induce major changes in the forest understory plant communities (Strole 1988, Putnam et al. 1989, Waller and Alverson 1997). The deer herd will have a propensity to act as a confounding factor for the MDMVA in achieving their INRMP goals. Traditional deer herd management practices involving hunter harvest are difficult to implement at FCTC while maintaining base security. Hunting is limited with respect to time when compared to other areas in Michigan with only 5-6 days to hunt annually. In the 5 to 6 days the public is allotted to hunt FCTC, military operations come to a halt. FCTC‟s purpose is to provide training to the Michigan Army National Guard (MIANG) and other components of the armed forces; FCTC plays a critical role in national defense, both abroad and at home. Countless numbers of troops have performed training exercises before being deployed to both Iraq and Afghanistan. After hurricane Katrina, evacuees from 10 the south called FCTC home temporarily. Figure 1.3. chronologically summarizes the events that molded FCTC into it‟s present state. 11 WWII and FCTC increased in size FCTC area settled 1830 11818 30 1910 1920 1930 1940 Deer hunting ceased at FCTC 1950 WWI and establishment of FCTC 1960 1970 FCTC reduced to present size Figure 1.3. A chronological summary of events that shaped FCTC. 12 Public hunt established in 1986 1980 During a 1985 wetland survey carcasses of starved deer observed 1990 2000 INRMP implemented To assist the MDMVA in meeting the requirements of their INRMP and for the MIANG to achieve their military objectives, methodology needs to be developed to analyze the deer herd population and its impact on the landscape of Ft. Custer. The historical biodata from FCTC provides evidence that the deer herd‟s physical condition is poor when compared to the rest of the surrounding areas, and areas with similar forest communities. For the MDMVA to make informed and effective management decisions they will need a better understanding of the effect of the deer herd on forest dynamics, the population parameters of the deer herd, and the impact of the hunter harvest on the population parameters of the deer herd. 13 Objectives This study was designed to assist the MDMVA in achieving their INRMP goals in conjunction with FCTC‟s ability to fulfill their primary objective of military preparedness, with respect to the deer herd‟s impact on the landscape, overall health, and the ability to conduct military training operations. The objectives will provide the MDMVA personnel the knowledge and means to assess the deer herd‟s overall health, size, and effects on forest regeneration and composition. The objectives are: 1. Develop survey methodologies that will provide an accurate estimation of deer population parameters (e.g. population size, sex ratio, fawn-to-doe ratio, survival, and sources of mortality). 2. Examine deer movement patterns to determine the annual pattern of area/habitat use on FCTC and the season specific level of movement between FCTC and the surrounding area, with particular emphasis on fall movements into areas that cannot be hunted. 3. Quantify the effects of deer herbivory on forest regeneration, with respect to species composition and density. 4. Provide training to MDMVA/FCTC personnel so that they can continue the monitoring/research efforts beyond the completion date of this project 5. Provide management recommendations to MDMVA/FCTC personnel that are in compliance with the INRMP. 14 CHAPTER 2: METHODOLOGY Spring Capture Preliminary results suggested there was a need to study fawn survival, movement, and sources of mortality to account for a low sighting rate of fawns during the spotlight survey; so, the original protocol of the project was amended to include a spring capture period of neonatal fawns. I captured neonatal fawns during the spring of 2006 with assistance of MDMVA personnel and qualified volunteers. In 2007, three field assistants aided in the capture of neonatal fawns. My field assistants and I performed capture of neonatal fawns in accordance with Michigan State University‟s (MSU) Institutional Animal Care and Use Committee guidelines (Application No. 01/04-006-00) and protocols detailed in the scientific collectors permit number SC 816 issued by the MDNR. All captured neonatal fawns were weighed, ear tagged, aged, and fitted with expandable VHF M-2110 radio-collars manufactured by Advance Telemetry Systems, (Ishanti, Minnesota, USA). The spring capture period began in the second week of May in 2006 and 2007. I anticipated a peak period around June 1 based on other research that entailed the capture of neonatal fawns in southern Michigan (Pusateri et al. 2006, Hiller 2007). Fawns were located and captured based on several techniques: behavioral clues from does, grid searching, and searching established home ranges of radio-collared does combined with grid searching (Downing and McGinnes 1969, Lund 1975, Bartush and Lewis 1978, Huegel et al. 1985, Ballard 15 et al. 1999). A fishing net was used to assist in the capture of fawns if they were flushed. Captured fawns were aged based on the condition of the umbilical cord (Haugen and Speake 1958), the amount of new hoof growth (Haugen and Speake 1958), and behavior (Haugen and Speake 1958, Downing and McGinnes 1969, White et al. 1972). Captured fawns were manually restrained, and blindfolded to subdue the animal and lessen capture related stress. If the cover was disturbed where the fawn was captured, it was moved to an area no more than several meters away where it remained concealed from predators. All positions of captured fawns were recorded using a hand held global positioning system (GPS) unit (Model GPS V; GARMIN International, Incorporated, Olathe, Kansas, USA). At each capture location, I recorded the vegetation characteristics, handling duration of the fawn, time of day, and weather conditions. Winter Capture I captured, radio-collared, and ear tagged deer from 2005 to 2008 from December to late March each year except the first capture season that began in February. I conducted trapping and baiting in accordance with protocols detailed in the scientific collectors permit number SC 816 issued by the MDNR and with Michigan State University‟s (MSU) animal care and use forms (Application No. 01/04-006-00). Deer were captured using clover traps (Clover 1954, Clover 1956, McCullough 1975) when the weather conditions (e.g. cold temperatures 16 and snow cover) increased the likelihood of capture. In April of 2005, drop-nets (Ramsey 1968) were used in addition to clover traps because only three deer were caught with clover traps during that trapping season. Drop-nets were not the ideal trapping method for two reasons: at least 8-10 people are needed because several deer can be captured at once, and increased capture related stress and trauma placed on the captured deer increases trap related mortality rates (Beringer et al. 1996, Haulton et al. 2001). In addition, FCTC lacks multiple drop-netting sites, since it is mostly closed canopy forest. Deer captured would have come from the same area, so, that would have been a poor sample of the deer herd. Clover traps were distributed throughout FCTC, with half placed in the interior and the other half along the exterior (Figure 2.1). Traps were not set during inclement weather if the predicted ambient air temperature was below 12.2 C°, or if the predicted wind chill factor was below -17.8 C°. Under those conditions traps were tied open with the area around the trap baited. I discovered through trial and error that traps would not work correctly if it rained and the nighttime temperatures fell below freezing. The netting on the trap would freeze and prevent the door closing properly, and on several occasions, I observed deer sign inside the trap, but a hung door where the deer escaped. To avoid a negative trap response from a deer, I tied the trap doors open when the netting of the traps were saturated with moisture and nighttime temperatures fell below freezing. 17 Traps were checked once or twice each day. Each deer caught was manually restrained, blindfolded, sexed, and aged as an adult or juvenile based on tooth wear and replacement (Severinghaus 1949). To reduce the amount of stress on a deer, processing time, noise and movement were kept to a minimum. Figure 2.1. Clover trap distribution during winter capture seasons. 18 All abrasions and lacerations were treated with Furall™ (4% furazolidone) an aerosol topical antibiotic. All captured deer were fitted with color-coded area-specific ear tags with unique values used to identify individuals. Ear tags were placed in the right ear for males, and the left for females. All captured deer were fitted with hermetically sealed VHF MOD-500 radio-collars manufactured by Telonics, Inc. (Mesa, AZ, USA) with a minimum battery life of 3 years and a motion sensitive mortality sensor. Juvenile males were the exception to the rule; they were collared on a case-by-case basis since male deer have such a large variation in neck size from 6 months old to the latter adult stages. I would determine if it was possible to fit a juvenile male with a collar that has a foam insert that will not be overly constrictive when the deer attains full body size and when the neck swells during the rut. The foam insert in radio-collars wears out over time, which aids in fitting juvenile deer with radio-collars. Juvenile females were typically fitted with a foam insert collar, but do not have a similar amount of variation in neck size throughout their life. I used a different style collar on the juvenile deer during the last year of winter capture (2008), since I had M-2110 expandable fawn collars manufactured by Advanced Telemetry Systems (Ishanti, Minnesota, USA) leftover from the spring capture period. Any deer less than 1.5 years old that was captured and fitted with a radio-collar in 2007-2008 had a fawn collar to lessen the chance of a slipped collar or a collar that would not fit properly as the deer grew. All deer were ear tagged in the same manner as previous years. 19 Monitoring Radio-Collared Deer I monitored radio-collared deer for alive or dead status daily for 30 days after capture as recommended by Haulton et al. (2001) because of the possibility of capture related myopathy. Deer were monitored at least once per week throughout the duration of the project for alive or dead status to determine survival rates. All mortality signals were promptly investigated, and all mortalities transported to the MDNR wildlife disease lab to have a necropsy performed to determine the cause of death. I classified mortalities into 7 categories: hunter harvest, trauma, deer vehicle collisions, coyote (Canis latrans) predation, dehydration, abandoned and malnutrition, and unknown. The location of mortality sites were recorded with a global positioning system (GPS) unit (Model GPS IV; GARMIN International, Incorporated, Olathe, Kansas, USA) and thoroughly investigated to find any clues to help determine the cause of death. I provided any pertinent information and details relevant to the mortality to the pathologist to assist in determining the cause of the mortality. During periods associated with high activity, deer were monitored for alive or dead status, and located 2 or 3 times a week. Periods of high activity include the fawning period in the spring and the fall rut and hunting season (Nelson and Mech 1992). It is not uncommon for deer to disperse from their natal ranges, or migrate from their seasonal ranges in spring or fall (Nelson 1998). Typically, yearling males are the most likely to exhibit such behavior (Nelson 1993, Purdue et al. 2000). During the spring the doe-fawn interaction is the driver for dispersal events, while in the fall male-male interactions are the predominant force in 20 causing changes in white-tailed deer movement patterns (Holzenbien and Marchinton 1992, Ozoga and Verme 1985, Rossenburry et al. 2001). Radio-collared deer were located at a minimum of once a week during other periods of the year. I used hand held yagi triple beam antennas with R1000 telemetry receivers manufactured by Communication Specialists, Inc. (Orange, California, USA) to locate deer. I triangulated the location of radiocollared deer by inputting the data into the computer program Locate III (Pacer, Truro, Nova Scotia, Canada) which calculates point locations and 95% confidence area ellipses based on maximum likelihood estimate (Lenth 1981) as recommended by White and Garrot (1990) and Nams and Boutin (1991). Distances to a radio-collared deer were kept minimal because bearings were measured at multiple telemetry stations at different times; it is possible that the radio-collared deer moved between bearing readings. Reducing the distance to the transmitter shortened the amount of time to travel from one telemetry station to the next, decreasing the distance the radio-collared deer may move. Greater distance to a transmitter also increases the weight of bearing error (Hupp and Ratti 1983, Saltz and Alkon 1985, Nams 1989). My field assistants and I also recorded all visual locations with GPS, along with the deer‟s behavior, physical condition, what type of cover it was in, antler characteristics if applicable, and if one or more fawns accompanied a radio-collared doe. I estimated home ranges with the GIS software packages ArcView V3.2 (Environmental System Research Institute, Redland California, USA) with the Spatial Analyst and Animal Movement Extensions using fixed kernel estimates 21 with least cross squares validation (LCSV). Fixed kernel methods provide more accurate home range estimates than do minimum convex polygons and harmonic mean methods (Worton 1995, Seaman and Powell 1996, Swihart and Slade 1997). Kernel methods produce a utilitarian distribution (UD) which is a three dimensional estimate of home range (Worton 1989). The third dimension is associated with time, and the likelihood of the animal being in that region of its home range. Seaman et al. (1999) reported that sample size is critical when estimating a home range, and recommended that a minimum of 30 observations be required to calculate a home range. When examining home range habitat use and preference, analyses were restricted to radio-collared deer that had a minimum of 30 observations to avoid overestimation of the home range size (Seaman et al. 1999). There were occasions when a deer slipped a radio-collar, died, or suffered some other fate before the minimum number of observations was met for an accurate estimate of home range. If the criteria for the minimum number of observations were not meet then results were treated on a case by case basis, and the findings were reported, but not treated with the same weight. Habitat Use and Resource Selection I performed compositional analysis as outlined by Aebischer et al. (1993) to assess habitat use of radio-collared deer. MDMVA personnel provided spatial vegetation data layers from a 2001 vegetation sampling of FCTC conducted by Michigan Natural Features Inventory (MNFI). I used ArcView v3.2 to reclassify 22 the vegetation cover types into seven categories: oak hardwoods (quercus spp.), mixed hardwoods, fields, pines (pinus spp.), brush, locust (robinia spp.), and wetlands. The core home ranges of radio-collared deer were used in the compositional analysis with the software Resource Selection for Windows (Fred Leban, University of Idaho). Compositional analysis compares the composition of the seven categories of vegetation cover that is available within FCTC to the composition of the vegetation cover types within each radio-collared deer‟s core home range. The Resource Selection for Windows (RSW) software program performs a log ratio test between the habitat used and the habitat available by radio-collared deer. Vegetation cover types were ranked by strongest preference to the weakest. I performed compositional analysis by gender and age (neonatal fawns, mature fawns, yearlings, and adults). I considered fawns less than 6 months old as neonatal, while fawn less than a year old, but older than 6 months were classified as mature fawns. The multiple age classification of fawns was due to capture during two distinctly different time periods, spring and winter, where new individuals would be added to a pool at approximately 6 months old. It is reasonable to assume the home range fidelity of a mature fawn may not be the same as that of a neonatal fawn, since home range size typically increases as a fawn matures, in addition to seasonal differences in the home range of the whitetailed deer. 23 Estimation of Population Parameters I pooled deer by gender and age for survival analysis. There were three general pools: adults, yearlings, and fawns. The fawn pool also had two time subcategories, for a couple of reasons. Fawns were captured at two distinctly different periods, spring and winter capture periods. Neonatal fawn survival during the first several months may vary over the first several months, so I examined survival of neonatal fawns over their first 30 and 60 days. Fawns captured during winter were in a distinct category from neonatal fawns captured during spring. Radio-collared neonatal fawns that survived up too, and past the winter capture period were then subsequently categorized in the same category as winter captured fawns. The data structure dictated grouping fawns into two classes for survival analysis, since capture could have occurred in either spring or winter capture periods. In the survival analysis of fawns, I will refer to spring capture fawns as neonatal fawns, while referring to mature fawns for any fawn captured during the winter capture season. Any neonatal fawn surviving past 6 months of age became classified as mature fawn for survival analysis purposes. I grouped survival data together over the entire duration of the study because of the relatively small sample size. I used the Mayfield Method to estimate the survival of radio-collared deer (Mayfield 1961, 1975). The variances were calculated using the MICROMORT software. Survival analysis included staggered entry since not all deer were radio-collared at the same time (Pollack et al. 1989). I used a seven day acclimation period to the radio-collars. Captured deer may experience capture related stress which could result in 24 mortality; data indicated that capture stress was no longer attributable at seven days, so the acclimation period was defined to be over. The Mayfield Method was used to determine survival. In recent years a preponderance of the journal articles published have used the Kaplan-Meier method, one reason being that it does not assume constant survival. But, the Kaplan-Meier method typically requires a minimum of 50 animals per treatment period for accurate survival estimates and the Mayfield Method should be used with sample sizes smaller than 50 animals (Winterstein et al. 2001). Since survival is assumed constant for the duration of the study with the Mayfield Method, survival was assessed over different time intervals that have constant survival and overall survival was calculated as the product of the estimates from the different time intervals (Johnson 1979, Bart and Robson 1982, Heisey and Fuller 1985). To test the assumption that constant survival holds for the shorter time intervals the hazard function was examined as suggested by Winterstein et al. (2001). Population Size Estimate I utilized the sex-age-kill (SAK) model as an estimator of the population size at FCTC. I calculated the variance estimates of the SAK model as described by Skalski and Millspaugh (2002) and Millspaugh et al. (2009). The generic equation I used is as follows: 25 ˆ H ˆ ˆ ˆ ˆ NT  [1  RF / M  RF / M * RJ / F ] ˆ ˆ M B equation (2.1) T where ˆ NT = estimate of total abundance; ˆ H = estimated adult male harvest in year i; ˆ M T = total annual mortality rate of adult males; ˆ B = proportion of total male mortality due to harvest (male recovery rate); ˆ RF / M = estimated ratio of adult females to adult males in the population; ˆ RJ / F = estimated ratio of juveniles to adult females in the population. ˆ ˆ I used survival data of radio-collared deer to estimate M T and B . To estimate the adult female, adult male and juvenile segments of the population I used the SAK model outlined by Skalski and Millspaugh (2002). Forward Looking Infrared On two occasions, Vision Air Research Incorporated surveyed FCTC, FCRA, and the Hart‟s Lake area using forward looking infrared (FLIR) to estimate the size of the white-tailed deer herd. One FLIR survey took place on March 24, 2007, while the other FLIR survey occurred on November 13 and 14, 2007. Both FLIR surveys began after sunset for improved detectability of white-tailed deer. Fog interrupted the November survey on the 13th, which resumed the morning of the November 14. 26 Vision Air Research used a PolyTech Kelvin 350 II (Sweden) mounted on o the left wing of a Cessna 206 “Stationair”. The sensor gimbal allows 330 of o azimuth and 90 of elevation allowing us to look in all directions except directly behind the airplane. The infrared sensor installed in the gimbal is the high resolution Agema Thermovision 1000, which is a long wave system (8-12 micron). It has 800 by 400 pixels providing good resolution with the ability to determine animals by their morphology or body shape. The thermal delta is less o o than 1 C, which means it can detect objects with less than 1 C different than o o the background. There are 2 fields of view (FOV): wide (20 ) and narrow (5 ). At 305 m. above ground level looking straight down using the wide FOV the footprint or area covered by the sensor is 110 m. x 71 m. while the narrow FOV provides a footprint 27 m. x 18 m. Transects were spaced 213 m. apart and running north – south. The o sensor look angle was approximately 45 elevation or down look angle. The pilot used a Garmin 496 which provided the transect locations and flight track covered. Flight altitude was 305 m. above ground level of the highest point along the transect flown and the adjacent transect for flight safety. The portion of the flight within the study area was recorded on videotape. The pilot and sensor operator communicated to verify the location of the boundaries to turn the tape off and on. The sensor operator turned the tape off at the transect end and commenced recording at the start of the transect. The tapes were reviewed by playing the tape backward and forward and in slow motion and frame by frame as needed to identify deer groups, count individuals 27 within the group, and map group location. Deer were located by observing their level of emitted infrared energy versus background levels. Vision Air has the ability to switch fields of view to zoom in and confirm subject as needed. Duplicates or repeat groups were identified. For each hour of tape time it took 3 – 4 hours to review the tape and map the group location and run a tally of groups and total count. Groups were mapped at their observed position not the position of the airplane. Vision Air performed an additional check of the data through sampling the videotape for detection verification, and checking for duplicate groups. White-Tailed Deer Exclosures As part of the white-tailed deer ecology study conducted by Michigan State University, eight 20m x 20m deer proof exclosures were constructed to quantify the effects of deer herbivory on forest regeneration, composition and structure. This portion of the study is intended to continue after MSU personnel ceased conducting research, with the procedures for vegetation sampling outlined for Ft. Custer Training Center (FCTC) personnel. Continued monitoring of the exclosures is needed to ascertain if deer herbivory at FCTC is significantly affecting forest regeneration and composition in such a manner that will make it difficult to meet the objectives outlined in the Integrated Natural Resources Management Plan (INRMP). Site selection included 4 areas with mixed hardwoods and 4 areas with oak hardwoods with canopy cover greater than 85%. No sites were selected in 28 training area 9 because of the potential for UXO (unexploded ordinance) existing in that area, and no suitable sites were located in the training areas 1 and 2. Figure 2.2 displays the distribution of exclosures and paired control sites. Since FCTC incorporates prescribed burns into their land management practices, 2 mixed hardwoods sites and 2 oak hardwoods sites underwent prescribed burns. There appeared to be similar initial conditions between the exclosure and paired controlled site in terms of ground cover, midstory, and canopy species composition and the amount of cover. Data were collected upon completion of the exclosures to determine any significant statistical difference between the exclosure and paired control site before deer herbivory could be a factor. Areas selected did not have large variation in terms of elevation to avoid heterogeneity in soil characteristics (e.g. moisture content and soil type). Similarity in initial abiotic and biotic conditions within each site was based on visual observation. A large degree of variation in soil characteristics could preclude a species from occurring throughout the site area. In the scope of this design soil characteristics should not be a limiting factor to make within site comparisons between the exclosure and the paired controlled site. This could also lead to false positives or false negatives when making such comparisons between the paired control site and exclosure. Any differences in species composition and density will be assumed to be a result of deer herbivory, and not because of soil characteristics. The paired controlled site and exclosure each measures 20 x 20 m on the perimeter while their nearest point lies approximately 20 m from each other. The 29 distance between the exclosure and paired control site acts as a buffer between the effects caused by deer encountering the exclosure impeding their travel. Over time such behavior will produce edge effects around the exclosure as deer wander around the edge of the exclosure. The height of each exclosure is 2.4m. Each exclosure was constructed with galvanized steel farm fence that has a 9 gauge top and bottom wire, and 11 gauge filler wire with 6 inch vertical stays. The fencing material is mounted on 16 posts spaced 5 meters apart. The posts consist of a variety of materials ranging from treated yellow pine 4x4‟s 12 feet long, black locusts posts 12 feet long with diameters from 4 to 6 inches, and galvanized steel street sign posts salvaged from the Michigan Department of Transportation (MDOT). Each of the 8 herbivory study areas has an exclosure and a paired control site (Figure 2.2). Each control site is an unfenced 20m x 20m area in close proximity (~20m) to the exclosure. Paired control and exclosure site were selected to have similar vegetation characteristics, along with similar soil types, drainage and moisture characteristics so that any differences that may occur as time progresses are from deer herbivory, and not some other exogenous factor. In this study I considered the exclosure, the fenced in area, as the experimental unit, and the area vulnerable to browse the control. The location and forest type for each exclosure and paired control site are listed in Appendix B. The coordinates provided correspond to the northwest corner of the control site and the exclosure, which also is the location of the access point for every exclosure. The access point has had the wire fencing cut, and patched to function as a gate. 30 I used 3 vegetation metrics (frequency of occurrence, density, and vertical cover) to quantify and monitor changes in the plant community. Monitoring and analyzing changes in these vegetation characteristics will allow FCTC personnel to evaluate the effectiveness of their management policies. Frequency of occurrence is the proportion of sample units in which a species occurs (Bonham 1989), and describes the distribution of plants within a community (MuellerDombois and Ellenberg 1974). Absolute frequency is the number of times a species occurs in each plot. Vertical cover, the vertical projection of a stem or leaf onto the surface of the ground, determines the dominance of a species within a community (Higgins et al 2005). Density provides an added description of a plant community when coupled with the frequency of occurrence of a species (Higgins et al 2005). Vegetation sampling began in June and was completed by mid to late August. In this study the understory is from 0.0m > 0.5m high, the midstory ranges from 0.5m ≥ 2.0m, the overstory is > 2.0m. To avoid including any changes in the plant community associated with the disturbance from the construction of the exclosure sampling took place > 1 meter inside the perimeter of each of the exclosure and paired control sites. General Overview of Vegetation Sampling The line intercept method (Canfield 1941) estimates the amount of cover in the understory, midstory, and overstory. The line intercept method consists of placing a tape measure between two points (an 18 meter long transect), and having an observer measure the amount of vegetation touching the tape or any 31 vertical projections. Each site was sampled with 5 transects, each 18m long. The distribution of transects is determined by placing the initial transect parallel to the perimeter‟s edge at a random interval 1 to 4 m from the fence. The second transect was placed randomly 3 to 4 m away and parallel to the initial transect; the next three transects were spaced 3 m apart. The amount of cover was measured in centimeters and expressed as a percentage. Appendix B contains a data sheet for recording data while measuring vertical cover and stem density. A database will also be passed on to FCTC personnel of the data collected to date. Three 1x2m nested plots were randomly placed off each of the line intercept transects to sample species density and frequency of occurrence in the understory. Sampling included both woody and herbaceous species. The plots were marked with flagging to allow me to return to the same plot if needed throughout the sampling period to correctly identify all species. A prolonged sampling period may have occurred when some species were indistinguishable from other species since they were not in bloom at that time; so, I returned to a plot at a latter period when species were in bloom. The species composition sampling methods were the same in the midstory, but have three 1x5m nested plots instead of 1x2m nested plots. The plots were marked with flagging, and sampled as needed throughout the spring and summer to identify all the vegetation that occurs in the nested plots of the midstory. Sampling of species composition in the overstory was accomplished by counting and identifying all the species in each paired control site and exclosure. There are two categories for 32 overstory species: diameter at breast height (DBH) ≥ 10.16 cm or a DBH < 10.16 cm, based on measurements with a Biltmore stick. A theoretical layout for sampling a exclosure or paired control site is in Figure (2.3). Figure 2.2. A map of the exclosure site locations at FCTC. 33 T1 T2 T3 T4 T5 B C A Figure 2.3. A schematic layout of either an exclosure or a paired control site for vegetation sampling. There are five transect lines marked as T1, T2 … T5 to use for the vegetation density in the understory, midstory, and overstory. The rectangles marked A, B, and C are the nested plots that estimate species composition and frequency in the understory and midstory. The midstory nested plot is 1m x 5m, while the understory plot is a subplot of the midstory and measures 1m x 2m. 34 CHAPTER 3: RESULTS Winter Capture I captured 62 deer using clover traps from 2004-2008, and four more deer were captured with drop nets in 2004 (Table 3.1). Thirty-eight of the deer captured were females, all of which were radio-collared. Twenty-four of the twenty-eight males captured were fitted with a radio-collar. I captured 24 female adult deer, 14 juvenile female deer, 13 adult male deer, and 15 juvenile male deer (Table 3.1). I had a success rate, the total number of deer captured per the total number of trap days, for the winter capture that varied from 4.6% in 2005, and 0.51% in 2006 (Table 3.2.). The success rate was 2.38% in 2004, 1.41% in 2007, and 3.19% in 2008 (Table 3.2.). A mild winter attributed to the extremely low success rate for trapping during the 2006 season. During the capture season from December 28, 2005 to March 5, 2006 FCTC lacked snow cover (personal observation) and unseasonably warm temperatures coupled with an unusual amount of other forage decreased the effectiveness of baited traps. In the month of January 2006 the daily mean temperature was 1.1 °C, while the daily high averaged 3.9 °C, and the low averaged -1.6 °C (Midwestern Regional Climate Center, Champaign, Illinois, USA). Annually the daily high averages -1.1 °C and the average daily low is -8.8 °C, which yields a daily mean of -4.9 °C (Midwestern Regional Climate Center, Champaign, Illinois, USA). In January 2006 it snowed 13 cm, compared to an average of 38 cm (Midwestern Regional Climate Center, Champaign, Illinois, USA). In January 2006 there were only 4 35 days with snow cover on the ground, with an average depth of 2.5 cm during that time frame (Midwestern Regional Climate Center, Champaign, Illinois, USA). The month of February in 2006 had 16 cm of snow compared to the average 23 cm (Midwestern Regional Climate Center, Champaign, Illinois, USA). Competition for bait amongst by-catch species lowered success rates when these species ate the available bait and tripped the traps. Additionally, several deer exhibited a trap-happy response, where once captured they were recaptured multiple times. I captured a male fawn on 14 different occasions at 2 different trap sites, which were a kilometer away from each other during winter of 2007. This caused me to remove the traps from both sites since the likelihood of capturing another deer was low and I wanted to avoid unnecessary stress on this particular deer. Table 3.1. The number of deer captured at FCTC during the winters of 2004 to 2008. Female Male Adults Yearlings Adults Yearlings Total Year 2004 6 3 0 0 9 2005 7 10 3 11 31 2006 2 0 1 0 3 2007 5 0 4 1 10 2008 4 1 5 3 13 Total 24 14 13 15 66 36 Table 3.2. Capture success rate of winter captured deer at FCTC based on number of trap days and total number of captures. Total Recapture Total Number of Captures events Trap Days Success Rate Year 2004 9 2 378 2.38% 2005 31 22 674 4.60% 2006 3 0 593 0.51% 2007 10 19 709 1.41% 2008 13 7 408 3.19% Total 66 50 2762 2.39% Spring Capture During the two spring capture seasons, I captured 14 neonatal fawns, all of which were radio-collared, weighed, ear-tagged, and aged. I captured two males and two females in 2006. The following year 10 neonatal fawns were captured, six of which were females and four were males (Table 3.3). Table 3.3. The average age and weights of spring captured neonatal fawns at FCTC in 2006 and 2007. Year 2006 2007 Total Average Age (days) 2.20 7.93 6.29 SE 0.41 2.09 1.64 Average Weight (Kg) 2.97 5.67 4.90 SE 0.11 0.69 0.59 Table 3.4. Capture success based on neonates observed and number of personnel hours per capture during the spring capture at FCTC. Year 2006 2007 Total Observed 9 17 26 Captured 4 10 14 Percentage of Observed Captured 44.4% 58.8% 53.8% 37 Hours averaged per capture 58.5 34.2 46.4 A neonatal fawn carcass was discovered in addition to the 10 fawns captured in 2007. A necropsy performed by the MDNR at the Wildlife Disease Lab indicated that the male fawn was a stillborn. Since the deer was never alive, it has not been included in any analysis, or counted as captured. On average, fawns weighed 4.90 ± 0.59 Kg for both spring capture seasons. However, in 2006 fawns on average weighed 2.97 ± 0.11, while fawns weighed 5.67 ±0.69 kg in 2007; but this shift in weight is explained by the change in the average age at capture. Fawns typically were 2.20 ±0.41 days old and 7.93 ±2.09 days old respectively in 2006 and 2007 when captured. The overall average age of capture was 6.29 ±1.64 days old. The age at capture differed between the two seasons due to my field assistants‟ and my own experience catching and observing neonatal fawns. After flushing several neonatal fawns we noticed that if the fawn had its‟ ears upright versus laid back on its head, it would flush. Thereafter, when we noticed that behavior, my field assistants and I would slowly back away from the fawn. At that point, we would enlist assistance from several personnel of the MDMVA environmental staff to capture the fawn. Anticipating the flush of the fawn personnel would surround the deer, and slowly encircle it in a concerted team effort. Once captured, it took on average 4.6 ±0.46 minutes to process the animal (i.e. age, weigh, ear-tag, radio-collar, and determine its‟ gender). 38 Survival Spring Captured Neonatal Fawns The data structure dictated grouping fawns into two classes for survival analysis, since capture could have occurred in either spring or winter capture periods. In the survival analysis of fawns, I will refer to spring capture fawns as neonatal fawns, while referring to mature fawns for any fawn captured during the winter capture season. Any neonatal fawn surviving past 6 months of age became classified as a mature fawn for survival analysis purposes. I grouped survival data together over the entire duration of the study because of the relatively small sample size. Neonatal fawns (n=14) experienced a period survival rate of 0.493 for their first six months; males and females included in the same group for fawns because there are minimal differences at a young age (Ozoga and Verme 1986). Neonatal fawns had 5 different sources of mortalities: probable coyote (Canis latrans) predation (n=3), hunter harvest (n=2), deer vehicle collision (n=1), abandoned/malnourishment (n=1), and dehydration (n=1) (Table 3.5). Collars remained on the deer for an average of 244 days (SE=68.2) for deer that were censored due to collar deterioration or being alive at the end of data collection. On one occasion a collar prematurely expanded, and came off the fawn after only 16 days. The next shortest duration a collar remained on a fawn was for 79 days, while another fawn retained a collar for 442 days. 39 Table 3.5. Fate of neonatal captured fawns at FCTC from 2006 to 2007. Fate Female Male Total Percent Possible coyote predation 1 2 3 21.43% Hunter Harvest 2 0 2 14.29% Abandoned/malnutrition 0 1 1 7.14% Dehydration 1 0 1 7.14% Deer vehicle collision 0 1 1 7.14% Slipped collar 1 1 2 14.29% Censored 3 1 4 28.57% Winter Captured Deer I categorized deer by gender and age for survival analysis since differences are likely based on those categories. Survival data were pooled together across years because each year‟s sample size was relatively small. I also included all neonatal deer that survived to 1 December, and reclassified them as mature fawns from that point forward. Deer that survived past June 1 were grouped into the next age group (i.e. mature fawns became yearlings, and yearlings became adults) for survival estimates. The daily survival rate for mature fawns (n=22) was 0.997. Yielding a period (6 months) survival rate of 0.581. The annual survival rate for all fawns equaled 0.289 when combining the neonatal rate with the mature fawn rate. Adult does (n=26) experienced the highest daily and annual survival rates, 0.999 and 0.756 respectively. A yearling doe (n=10) had a 0.998 daily survival rate, or a 0.443 annual survival rate. The daily and annual rates for adult bucks (n=14) were 0.998 and 0.493, respectively. Yearling bucks (n=9) survived at a 0.997 daily rate and a 0.379 annual rate. The survival rates of deer are listed with their variances in Table 3.6. 40 Other than the hunter harvested radio-collared deer, the cause specific mortality was determined by field observations, the history of the animal, and a necropsy of the carcass performed by a wildlife pathologist at the MDNR Wildlife Disease Laboratory in East Lansing, MI. Winter captured deer died from the following sources of mortality: hunter harvest (n=21), possible canid predation by coyotes (Canis latrans) (n=6), stress and trauma (n=3), accident and starvation (n=1), unknown mortality source (n=1), and deer vehicle collisions (n=1). A complete listing by age and gender of the fate of winter captured radio-collared deer is contained in Tables (3.7 and 3.8). In addition to the mortalities suffered by the radio-collared deer, 19 were alive at the conclusion of the study, and 11 more deer slipped their radio-collars. The relatively high number of slipped collars was attributed to age class of the deer radio-collared, and I tried to avoid collaring the animals too tightly. A total of four fawns, four yearlings, and three does slipped their respective radio-collars (Table 3.7 and 3.8). The deer, an adult doe, which died from an accident and starvation, was located underground in the tunnel system at FCTC that is associated with the live firing ranges. The doe fell into the tunnel, and apparently could not escape, and then died from starvation. In the cases involving predation, I found carcasses mutilated by coyotes (Canis latrans) and scavenged with the radio-collar forcibly removed from the deer with clear indications of bite marks from canines. In the instance of the unknown mortality source, I did not discover the carcass in time to have a necropsy performed with a conclusive cause of death. I found the carcass of this ten-month-old doe heavily scavenged by turkey vultures 41 (Cathartes aura) and other scavengers, and I suspected that the mortality sensor did not activate upon its‟ death due to the scavenging. 42 Table 3.6. Annual and daily survival rate estimates of radio-collared deer at FCTC based on the Mayfield Method. Period Lower 95% CI Upper 95% Sample Daily Survival Period of period CI of period Gender Age Size Survival Variance (One Year) Variance survival survival Female and Male Neonatal Fawns1 14 0.996 2.11E-06 0.493 0.017 0.230 0.755 1 Female and Male Mature Fawns 22 0.997 1.46E-06 0.581 0.017 0.324 0.838 Female Yearling 9 0.998 1.24E-06 0.443 0.032 0.083 0.803 Female Adult 26 0.999 4.88E-08 0.756 0.004 0.634 0.878 Male Yearling 10 0.997 1.75E-06 0.379 0.034 0.012 0.746 Male Adult 14 0.998 5.32E-07 0.493 0.017 0.230 0.757 2 Female and Male Fawn 14,22 n/a n/a 0.289 0.010 0.147 0.568 1. The period survival is for 6 months. 2. The fawn age class survival rate is a combination of the neonatal and mature fawn age class, and no daily survival rate can be calculated. The period survival rate estimate for fawn is a multiple of the neonatal and mature fawn 43 Table 3.7. Fate of female winter captured radio-collared deer at FCTC. Fate Fawns Yearlings Adults Total Percent Hunter harvest 0 2 8 10 20.00% Possible coyote predation 2 0 3 5 10.00% Stress/trauma 1 1 0 2 4.00% Deer vehicle collision 0 1 0 1 2.00% Accident/starvation 0 0 1 1 2.00% Slipped collar 3 1 3 7 14.00% Unknown mortality source 1 0 0 1 2.00% 1 Alive 0 2 9 11 22.00% 2 Collar malfunction 0 0 1 1 2.00% 3 Censored 6 5 0 11 22.00% 1. The radio-collared deer survived to the end of the study June 1, 2008. 2. The radio-collared deer was sighted and identified by the unique ear tag associated with the individual, but no radio signal was detected. Detection of the radio signal of the collared deer had been sporadic after four months of capture. 3. Censored radio-collared deer survived long enough to be reclassified into the next older age category (i.e. fawns became yearlings, and yearlings became adults on June 1 each year). 44 Table 3.8. Fate of male winter captured radio-collared deer at FCTC. Fate Fawns Yearlings Adults Total Percent Hunter harvest 1 4 6 11 34.38% Possible coyote predation 1 0 0 1 3.13% Stress/trauma 0 0 1 1 3.13% Slipped collar 1 3 0 4 12.50% 1 Alive 0 2 6 8 25.00% 2 Censored 6 1 0 7 21.88% 1. The radio-collared deer survived to the end of the study June 1, 2008. 2. Censored radio-collared deer survived long enough to be reclassified into the next older age category (i.e. fawns became yearlings, and yearlings became adults on June 1each year). 45 Home Range Estimates I compiled location data from visual locations recorded with gps and telemetry location data to estimate home ranges of deer. I calculated home ranges for deer for which I had a minimum of 30 locations (Seaman et al. 1999, Kernohan et al. 2001). I made an exception for one male fawn (n=28). The smaller sample size of this individual did not seem to greatly affect the home range estimate, for several reasons: there is a degree of arbitrariness for a minimum sample size n=30 and this individual had the fifth smallest home range area. Other than the exception to the sample size rule, my sample sizes for number of locations ranged from n=31 to n=259. Forty-four radio-collared deer fulfilled the criteria for a fixed kernel home range estimate and were used to estimate home range size. Fawns exhibited the smallest home range size amongst all age classes, fawns, yearlings, and adult (Table 3.9 and Table 3.10). Male yearlings (n=2) had the largest home ranges, averaging 112.98 hectares (ha) (SE=107.75) for a core home range while the 95% probability overall home range was X =847.59 ha (SE=812.89). All radio-collared deer home range estimates and locations are depicted in Appendix C. 46 Table 3.9. The mean core home range size of radio-collared deer in hectares by age and gender. Gender Age n Mean Std Error LCL UCL Class female adult 19 18.31 3.03 11.95 24.67 female fawn 4 7.63 2.56 0.00 15.78 female yearling 6 58.01 26.37 0.00 125.78 male adult 9 43.36 6.21 29.03 57.68 male fawn 3 11.58 2.83 0.00 23.75 male yearling 2 112.98 107.75 0.00 1482.12 Table 3.10. The mean overall home range size of radio-collared deer in hectares by age and gender. The overall home range represents the area where radio-collared deer have 95% chance of ranging. Gender Age Class female adult female fawn female yearling male adult male fawn male yearling n 19 4 6 9 3 2 Mean 132.22 73.22 266.41 324.82 52.99 847.59 Std Error 16.05 23.27 114.24 37.69 6.02 812.89 LCL 98.49 0.00 0.00 237.90 27.07 0.00 UCL 165.95 147.29 560.08 411.74 78.91 11176.38 Habitat Use Forty-four deer were used to calculate core home-ranges to perform compositional analysis as outlined by Aebischer et al. (1993) to test for random habitat use and rank the habitats according to resource utilization. I categorized deer into data subsets based on gender and age to test for differences in those categories with respect to resource utilization and random habitat use. The groups are: adults n=26, bucks n=9, does n=19, yearlings n=8, and fawns n=7. The yearling group contains six females and two males, while there were three male fawns and four female fawns. 47 Results of compositional analysis indicate each group of deer had nonrandom habitat use (Table 3.11). The oak hardwoods (quercus spp.) were the preferred vegetation type for all groups, except for fawns that were most commonly in the open fields of FCTC. Fawns selected the oak hardwood (quercus spp.) habitat as their second highest preference, while maple hardwoods (acer spp.) were selected second for habitat among yearlings, both male and female, bucks and does. 48 Table 3.11. Rankings of the preferred vegetation types used by radio-collared deer at FCTC. Vegetation Classification Oak Mixed Deer classification Hardwoods Hardwoods Fields Conifers Brush Wetlands All radio-collared deer 1 3 2 4 6 5 Radio-collared bucks 1 2 3 4 6 5 Radio-collared does 1 2 3 4 7 5 Radio-collared fawns 2 3 1 4 7 6 Radio-collared adults 1 2 3 4 6 5 49 Locust 7 7 6 7 7 Population Estimates Spotlight Survey From 8/30/2004 to 9/10/2004 a spotlight survey was conducted at dusk eight times to estimate adult sex ratio, and fawn-to-doe ratio. The survey took approximately an hour and 45 minutes each night. In the design of the spot-light route areas 3 and 4 were omitted from the survey because there was no feasible way to include them. Areas 3 and 4 do not contain enough suitable sites to conduct a survey in a timeframe of around an hour and 45 minutes and sample an adequate number of sites. A total of 593 deer were observed, 218 does, 154 bucks, 148 fawns, and 73 unknowns, yielding, about 68 fawns for every 100 does and 1.42 does for every buck. The survey results led researchers to believe that the survey should be conducted one to two weeks earlier because of a lower than expected number of fawn observations; fawns may have been mistaken for does. The 2005 spotlight mark-resight survey yielded a population estimate of 397 adult deer with the Lincoln-Petersen estimator: ^ nn N 1 2 m2 The number of deer sighted during a given night is n1, the total number of radiocollared deer on FCTC is denoted by n2, and m2 is the number of radio-collared deer sighted on a given night. Since no fawns were marked during the spotlight survey they were not included in the population estimate. In 2005 there were 3.41 does for every buck and 0.344 fawns per doe. 50 Sex-Age-Kill Model The Sex-Age-Kill model (SAK) indicated that the deer population has decreased at FCTC over the last several years (Figure 3.1.). I used two versions of the model, which varied with the parameter estimating male mortality (Mt) and the ratio of 1.5-year-old male deer as a percentage of the male population (Pym) (Skalski and Millspaugh 2002, Millspaugh et. al. 2009). Survival analysis based on the Mayfield Method of radio-collared deer yielded an estimate of (Mt) = 0.51. The hunt data yielded an estimate of Pym = 0.703 with a variance of 0.000878 (Table 3.12). The two models produced similar results, but the Pym model population estimates were lower than the Mt model estimates (Table 3.12). The Mt model only produced estimates for the entire population while the P ym model estimated the number of adult does, adult bucks, and juvenile deer. The deer population has declined steadily from 2004 to 2009 in both of the SAK models; the Mt model has estimated the population at 712 deer in 2004 and 496, 488, 352, 328, and 320 deer in the subsequent years. The deer population estimate went from 516 deer in 2004 to 360 deer in 2005, 354 deer in 2006, 255 deer in 2007, 238 in 2008, and 232 deer in 2009 in the SAK Pym model (Table 3.12). In terms of density, the deer population for 2009 is estimated at 19.62 deer per mi 2 by the SAK Pym model, or 27.04 deer per mi2 by the SAK Mt model (Table 3.14). 51 Population Trend of Deer at FCTC 600 Number of Deer 500 400 300 200 100 0 2004 2005 2006 2007 2008 Year Total Population Bucks Figure 3.1. the population trend of deer from the SAK Pym model. 52 Does Juveniles 2009 Table 3.12. SAK model population estimates. Model 1 2 Pym Mt Year 2004 516 712 2005 360 496 2006 354 488 2007 255 352 2008 238 328 2009 232 320 1. Pym is the proportion of 1.5 year old males harvested versus the entire male population harvested. 2. Mt is estimated from the annual survival rate of radio-collared buck using the Mayfield Method. Table 3.13. SAK Pym model annual populations estimates of age classes by gender with variance. Gender Age Class Year Number of deer Variance Bucks Adult 2004 148 39 Bucks Adult 2005 103 19 Bucks Adult 2006 101 18 Bucks Adult 2007 73 9 Bucks Adult 2008 68 8 Bucks Adult 2009 66 8 Does Adult 2004 256 1455970 Does Adult 2005 179 706571 Does Adult 2006 176 683962 Does Adult 2007 127 355859 Does Adult 2008 127 309261 Does Adult 2009 118 294359 Bucks and Does Juvenile 2004 113 n/a Bucks and Does Juvenile 2005 79 n/a Bucks and Does Juvenile 2006 77 n/a Bucks and Does Juvenile 2007 56 n/a Bucks and Does Juvenile 2008 51 n/a Bucks and Does Juvenile 2009 50 n/a 53 Table 3.14. The number of deer per square mile and kilometer with the SAK model. Pym Mt Year 2004 2005 2006 2007 2008 2009 Deer/mi2 43.65 30.41 29.92 21.58 20.11 19.62 Deer/km2 16.86 11.74 11.55 8.33 7.77 7.58 Deer/mi2 60.17 41.91 41.24 29.75 27.72 27.04 Deer/km2 23.24 16.19 15.93 11.49 10.71 10.44 Forward Looking Infrared During the night of March 27, 2007, Vision Air Research, Inc. conducted a forward-looking infrared (FLIR) flight over FCTC, FCRA, and the Heart Lake area. 285 deer in 92 different groups were observed with detection rate ranging from 70-100% (Table 3.15). The largest group size observed was 16, while several groups of one were observed. Vision Air Research, Inc. claimed an 82 – 87% detection rate in deciduous forest types, 100% in open meadows and agricultural fields, and 30-60% in conifer cover types. The deer that were sighted during the FLIR were concentrated in several areas (Figure 3.2). On 13 November 2007, Vision Air Research, Inc. flew another FLIR flight, but midway through the flight dense fog caused safety concerns preempting completion of the survey. The following morning of 14 November 2007 the fog no longer posed a safety issue to the flight that allowed the completion of the FLIR survey. The 13 November flight had 198 deer in 102 groups; during the following morning flight, 113 deer were observed in 66 groups (Table 3.16). Vision Air Research Inc. again claimed the same detection rates: 70-100% overall, 82-87% in deciduous forest cover types, 100% in open meadows and 54 agricultural fields, and 30-60% in conifer cover types. The Vision Air detection rates were estimates from their past studies in similar vegetation types. Their report indicated there were a number of leaves lingering on trees, but the leaves were relatively dry and not undergoing photosynthesis, which did not change detection rates. Deer sighted during the November FLIR were distributed throughout the survey area (Figure 3.3). The deer density from the March FLIR survey yielded estimates of 6.04 deer per km2 at FCTC and 6.64 deer per km2 over the entire survey area that included FCRA and Hearts Lake (Table 3.17). The November FLIR survey estimated 6.30 deer per km2 and 7.25 deer per km2 respectively over the same areas (Table 3.17). To account for the detection rate under 100% in the various cover types, I examined each data point from the FLIR surveys for the cover classification and group size for those particular points. I then adjusted the number of deer in the group based on the detection rates for the associated cover types. The detection rates had a range for each cover type, so I used both the high and low end of the range for these calculations, that led to a high and low adjustment of deer sighted for the two FLIR surveys. The detection rate adjusted values of deer density during the March flight ranged from 7.35 to 7.72 deer per km 2, the deer density ranged from 8.22 to 8.67 deer per km2 in the November survey (Table 3.18). 55 Table 3.15. The number of deer sighted during the March 2007 FLIR survey. Number of Mean Deer Max Deer Min Deer Deer Group Size Group Size Group Size Total Deer FCTC 185 62 2.98 10 1 IR Survey Area 285 92 3.09 16 1 Table 3.16. The number of deer sighted from the November 2007 FLIR survey. Min Number of Mean Max Group Group Deer Groups Group Size Size Size Total Deer FCTC¹ 193 120 1.6 5 1 IR Survey Area¹ 311 181 1.69 5 1 1. The IR survey was delayed for approximately eight hours due to weather and the survey area was split in half, which led to the possibility of deer being either omitted or counted twice during the survey. Further analysis of the IR data indicates that 8 deer could have been counted twice reducing the number to 185 deer sighted on FCTC and 303 deer for the entire survey area. Table 3.17. The deer density at FCTC and FCRA based on both FLIR surveys. FCTC FCTC and FCRA 2 2 2 2 Deer/mi Deer/km Deer/mi Deer/km Survey Period March 2007 15.64 6.04 17.21 6.64 November 2007 16.31 6.30 18.78 7.25 Table 3.18. The deer density estimate from the two forward looking infrared surveys conducted in 2007. There are low and high adjustments in the deer density to account for a slightly diminished detection rate associated with FLIR in deciduous cover types. 2 Deer/mi Low High Adjustment Adjustment March 2007 FLIR 19.06 20.01 November 2007 FLIR 21.29 22.46 56 Deer/km 2 Low High Adjustment Adjustment 7.35 7.72 8.22 8.67 Figure 3.2 The locations of deer from the March 2007 forward looking infrared flight. 57 Figure 3.3 Locations of deer groups from the forward looking infrared survey on the evening of 13 November and morning of 14 November 2007. An asterisk indicates the locations from the 13th, while the locations from the 14 th are indicated by a dot. Fog interrupted the flight after completion of approximately half of the survey. 58 Hunt Biodata The beam diameter of year old male antlers did not change from 2004 to 2009 significantly according to an ANOVA (F5, 199 = 0.83, P=0.53). The average number of antler points did show significant differences during the same time span (F5, 162 = 3.61, P=0.0041). The number of antler points for year old males fluctuated somewhat from 2004 to 2008. In 2004, year old males averaged 4.74 (SE ± 0.31) pts, rising to 5.6 (SE ± 0.33) pts in 2006, then decreasing the following year to 4.11 (SE ± 0.33) pts , 2008 averaged 4.74 (SE ± 0.38) pts, while 2009 averaged 3.45 (SE ± 0.41) pts (Figure 3.3). Beam diameters of year old males increased incrementally each year from 19.4 (SE ± 0.65) mm in 2004 to 21.36 (SE ± 0.75) mm in 2009 (Figure 3.4). The sample sizes of year old males from 2004 to 2009 were (n=55, n=47, n=48, n=30, n=27, and n=11) respectively. Two year old males average beam diameter did not show significant changes from 2004 to 2009 (F5, 35 = 1.06, P=0.40), and neither did the average number of antler points (F5, 31 = 0.37, P=0.86). The average beam diameter, and the number of antler points fluctuated from 2004 to 2009 (Figures 3.3 and 3.4). The sample sizes of two year old males from 2004 to 2009 were (n=16, n=6, n=3, n=5, n=6, and n=7) respectively. Beam diameter size of two year olds ranged from an average of 20.8 (SE ± 1.70) mm in 2005 to 27.0 (SE ± 1.00) mm in 2006. The number of antler points ranged from 6.40 (SE ± 1.16) pts in 2005 to 8.00 (SE ± 1.00) pts in 2006. The beam diameter size did not show any statistically significant changes from 2004 to 2009 for three year olds (F5, 30 = 1.42, P=0.24). The number of 59 antler points was relatively consistent (F5, 29 = 0.76, P=0.58). From 2004 to 2007 (no four year old deer were harvested after 2007) beam diameter size in four year olds changed significantly (F3, 12 = 5.46, P=0.013), while the number of antler point have not show signs of significant statistical changes (F3, 10 = 1.45, P=0.29) FCTC Harvested Deer Antler Point Averages by Age 11 10 9 Points 8 7 6 5 4 3 2 2004 2005 2006 2007 2008 2009 Year 1 Year Olds 3 Year Olds 2 Year Olds 4 Year Olds Figure 3.4. The annual trend in hunt biodata of the mean antler point counts by age. 60 FCTC Harvested Deer Average Beam Diameters by Age 36 34 Beam Diameter (mm) 32 30 28 26 24 22 20 18 2004 2005 2006 2007 2008 2009 Year 1 Year Olds 2 Year Olds 3 Year Olds 4 Year Olds Figure 3.5. The annual trend of means with respect to beam diameter by age. No four-year-old deer were harvested in 2008 or 2009. Vegetation Sampling Typically, the amount of vertical cover in the understory did not differ between the exclosure and the paired control site. Site 1, consisting of mixed hardwoods, was the only site that had a significant difference in vertical cover measurements. The differences occurred in the understory in 2004 and 2007 (P=0.025, df=8) and (P=0.014, df=8), respectively. Site 1‟s exclosure understory averaged 180 (SE ± 28.11) and 339 (SE ± 77.35) cm per 1800 cm transect in 61 2004 and 2007 respectively. The paired control sites averaged 94 cm (SE ± 14.28) and 87 cm (SE ± 22.40) cm per 1800 cm transect in 2004 and 2007 (Tables 3.19-3.42). The vegetation in the midstory in terms of vertical cover did not show major differences between the exclosures and paired control sites, except in one instance. In 2006 the midstory at the oak hardwood site 6, the exclosure had more vertical cover than the paired control, averaging 63 (SE ± 14.63) and 6 (SE ± 6.00) cm per 1800 cm transect respectively. Results from a t-test indicated a significant difference between the means (P=0.0069, df=8) (Tables 3.19-3.42). The overstory did not have any instances of significant differences between the exclosure and paired control site. However, the 2005 overstory vertical cover at site 6, an oak hardwood site, yielded a t-test on the fringe of having a significant difference. The exclosure averaged 1690 (SE ± 29.32) cm per 1800 cm transect, while the paired control mean equaled 1558 (SE ± 50.34) cm per 1800 cm transect (P=0.053, df=8) (Tables 3.19-3.42). On one occasion from 2004 to 2008, the stem density in the understory of the exclosure and the paired control were statistically significantly different. The exclosure at site 2 in 2006 averaged more stems per plot than the paired control (P=0.02, df=4) in 2006. At site 2, in 2006, on average 21.33 (SE ± 1.45) stems per 2 m2 plot were sampled in the exclosure, and the paired control averaged 7.00 (SE ± 3.21) stems per 2 m2 plot. The following year 9.00 (SE ± 4.16) stems per 2 m2 plot were sampled in the paired control, while the exclosure had 29.67 (SE ± 7.42) stems per 2 m2 plot. 62 Stem density did not differ significantly between the exclosure and paired control sites in the midstory. The mixed hardwood, site 3, in 2007 was the closest to having a significant difference (P=0.09, df=4). In that instance, 26.33 (SE ± 6.67) stems per 5 m2 plot were sampled in the paired control compared to 10.67 (SE ± 2.19) stems per 5 m2 plot in the exclosure. The stem density of the understory tended to increase at each site from 2004 to 2007, while in 2008 the stem density tended to decrease from the 2007 means. This pattern occurred in both the exclosure and the paired control. The pattern was most pronounced at sites 4 and 5. Site 4 was composed of oak hardwoods, while site 5 consisted of mixed hardwoods. The understory of site 4 in the exclosure averaged 14.67 (SE ± 2.91), 32.67 (SE ± 7.26), 48.67 (SE ± 7.17), 100.67 (SE ± 23.95), and 51.00(SE ± 10.21) stems per 2 m 2 plot from 2004 to 2008, while the paired control averaged 20.33 (SE ± 1.67), 48.67 (SE ± 8.69), 83.00 (SE ± 13.23), 179.00 (SE ± 44.02), and 78.67(SE ± 9.21) stems per 2 m2 plot (Tables 3.49-3.59). The understory stem density at site 5 followed a similar trend, where the exclosure averaged 6.33 (SE ± 0.67), 22.67 (SE ± 2.85), 85.00 (SE ± 17.62), 247.33 (SE ± 33.95), and 87.33(SE ± 11.29) stems per 2 m 2 plot from 2004 to 2008. This trend was not unique to the exclosure at site 5, the paired control mimicked the exclosure‟s pattern with an average of 8.00 (SE ± 1.15), 28.67 (SE ± 6.33), 98.33 (SE ± 7.17), 259.33 (SE ± 12.78), and 97.67 (SE ± 3.38) stem per 2 m2 plot during the same time span. The pattern was not strongly pronounced, or did not occur at sites 1 and 2. The stem density in the paired control at site 2 was relatively flat, it averaged 10.00 (SE ± 2.08), 6.33 (SE 63 ± 1.45), 7.00 (SE ± 3.21), 9.00 (SE ± 4.16) stem per 2 m 2 plot, and 14.33 (SE ± 7.88) from 2004 to 2008. The exclosure at site 2 averaged 11.67 (SE ± 2.19), 13.00 (SE ± 3.21), 21.33(SE ± 1.45), 29.67(SE ± 7.42), and 18.67 (SE ± 2.40) stems per 2 m2 plot. The complete results of the stem density sampling of the units are found in Tables (3.43-3.59). 64 Table 3.19. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means. Site 1 Mixed Hardwoods Understory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 180 28.11 94 14.28 8 0.025 Yes 2005 259 113.12 73 9.95 8 0.14 No 2006 222 45.90 165 40.81 8 0.38 No 2007 339 77.35 87 22.40 8 0.014 Yes 2008 256 58.16 163 43.88 8 0.23 No 1. The mean is estimated from a transect 1800 cm long. Table 3.20. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means. Site 1 Mixed Hardwoods Midstory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 181 41.11 261 26.2 8 0.17 No 2005 108 48.34 91 17.9 8 0.75 No 2006 270 47.03 126 48.0 8 0.068 No 2007 264 95.47 67 28.8 8 0.084 No 2008 136 35.37 136 21.7 8 0.13 No 1. The mean is estimated from a transect 1800 cm long. Table 3.21. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means. Site 1 Mixed Hardwoods Overstory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 1784 16.00 1717 45.43 8 0.20 No 2005 1732 22.23 1780 20.00 8 0.15 No 2006 1762 16.25 1757 13.93 8 0.82 No 2007 1691 16.85 1751 11.14 8 0.061 No 2008 1754 5.31 1759 8.42 8 0.63 No 1. The mean is estimated from a transect 1800 cm long. 65 Table 3.22. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means. Site 2 Mixed Hardwoods Understory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 240 47.86 253 32.35 8 0.83 No 2005 293 24.37 231 32.76 8 0.17 No 2006 263 43.29 224 38.79 8 0.53 No 2007 268 34.59 160 40.20 8 0.075 No 2008 266 24.54 217 10.54 8 0.10 No 1. The mean is estimated from a transect 1800 cm long. Table 3.23. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means. Site 2 Mixed Hardwoods Midstory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 52 12.80 38 14.16 8 0.48 No 2005 20 12.40 14 40.36 8 0.66 No 2006 34 18.53 49 21.00 8 0.61 No 2007 110 21.37 101 21.64 8 0.76 No 2008 54 4.73 50 10.96 8 0.77 No 1. The mean is estimated from a transect 1800 cm long. Table 3.24. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two Site 2 Mixed Hardwoods Overstory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 1784 16.00 1790 10.00 8 0.76 No 2005 1786 8.72 1782 13.56 8 0.81 No 2006 1782 11.14 1787 8.31 8 0.73 No 2007 1792 8.00 1784 9.80 8 0.54 No 2008 1786 7.07 1778 2.61 8 0.95 No 1. The mean is estimated from a transect 1800 cm long. 66 Table 3.25. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means. Site 3 Mixed Hardwoods Understory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 929 57.59 1000 37.02 8 0.32 No 2005 958 111.99 827 58.02 8 0.33 No 2006 906 81.82 852 128.96 8 0.73 No 2007 787 52.41 830 31.02 8 0.50 No 2008 718 63.77 817 21.78 8 0.80 No 1. The mean is estimated from a transect 1800 cm long. Table 3.26. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means. Site 3 Mixed Hardwoods Midstory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 498 36.53 389 41.95 8 0.087 No 2005 287 50.34 348 70.21 8 0.49 No 2006 359 49.05 443 54.55 8 0.29 No 2007 393 76.21 416 24.22 8 0.82 No 2008 406 47.48 396 27.14 8 0.86 No 1. The mean is estimated from a transect 1800 cm long. Table 3.27. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means. Site 3 Mixed Hardwoods Overstory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 1788 12.00 1782 11.14 8 0.72 No 2005 1772 19.60 1764 22.27 8 0.79 No 2006 1758 13.56 1757 13.92 8 0.96 No 2007 1774 16.61 1791 5.56 8 0.36 No 2008 1770 5.40 1783 9.34 8 0.25 No 1. The mean is estimated from a transect 1800 cm long. 67 Table 3.28. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means. Site 4 Oak Hardwoods Understory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 732 70.52 849 67.68 8 0.27 No 2005 721 51.17 837 183.40 8 0.56 No 2006 701 90.88 731 129.08 8 0.85 No 2007 699 71.58 794 64.87 8 0.36 No 2008 703 34.20 799 104.33 8 0.40 No 1. The mean is estimated from a transect 1800 cm long. Table 3.29. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means. Site 4 Oak Hardwoods Midstory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 418 61.02 347 46.16 8 0.38 No 2005 164 37.20 176 31.08 8 0.81 No 2006 148 20.04 183 44.99 8 0.50 No 2007 174 22.90 131 17.63 8 0.16 No 2008 230 27.53 205 20.74 8 0.49 No 1. The mean is estimated from a transect 1800 cm long. Table 3.30. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means. Site 4 Oak Hardwoods Overstory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 1735 21.79 1686 20.15 8 0.14 No 2005 1658 9.69 1670 34.29 8 0.66 No 2006 1685 18.03 1660 15.25 8 0.32 No 2007 1663 18.95 1648 28.88 8 0.68 No 2008 1694 8.62 1678 12.15 8 0.30 No 1. The mean is estimated from a transect 1800 cm long. 68 Table 3.31. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means. Site 5 Mixed Hardwoods Understory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 1449 51.34 1533 50.09 8 0.27 No 2005 1425 65.13 1564 42.70 8 0.11 No 2006 1476 87.62 1375 87.62 8 0.37 No 2007 1262 183.81 1377 60.35 8 0.57 No 2008 1384 57.24 1465 35.57 8 0.26 No 1. The mean is estimated from a transect 1800 cm long. Table 3.32. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means. Site 5 Mixed Hardwoods Midstory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 49 19.39 34 10.41 8 0.51 No 2005 77 27.21 50 28.42 8 0.50 No 2006 40 18.91 51 18.80 8 0.69 No 2007 79 12.08 73 27.18 8 0.85 No 2008 53 17.72 62 23.05 8 0.78 No 1. The mean is estimated from a transect 1800 cm long. Table 3.33. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means. Site 5 Mixed Hardwoods Overstory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 1766 18.33 1766 16.00 8 1.00 No 2005 1740 31.47 1726 28.74 8 0.75 No 2006 1765 15.65 1759 13.45 8 0.78 No 2007 1764 16.39 1752 15.29 8 0.61 No 2008 1778 13.89 1750 17.02 8 0.22 No 1. The mean is estimated from a transect 1800 cm long. 69 Table 3.34. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means. Site 6 Oak Hardwoods Understory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 343 48.57 344.8 45.85 8 0.98 No 2005 698 94.85 786 111.50 8 0.56 No 2006 299 52.50 249 52.09 8 0.52 No 2007 382 123.37 442 44.77 8 0.66 No 2008 428 57.67 456 40.37 8 0.70 No 1. The mean is estimated from a transect 1800 cm long. Table 3.35. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means. Site 6 Oak Hardwoods Midstory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 42 29.39 19 9.27 8 0.47 No 2005 63 14.63 6 6.00 8 0.0069 Yes 2006 44 17.42 33 10.56 8 0.60 No 2007 37 8.75 17 7.35 8 0.12 No 2008 55 14.99 25 7.21 8 0.11 No 1. The mean is estimated from a transect 1800 cm long. Table 3.36. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means. Site 6 Oak Hardwoods Overstory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 1550 53.01 1584 41.55 8 0.63 No 2005 1690 29.32 1558 50.34 8 0.053 No 2006 1658 24.58 1665 38.18 8 0.88 No 2007 1661 25.42 1666 23.15 8 0.89 No 2008 1646 16.21 1616 23.72 8 0.32 No 1. The mean is estimated from a transect 1800 cm long. 70 Table 3.37. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means. Site 7 Oak Hardwoods Understory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 175 15.19 211 29.55 8 0.31 No 2005 434 91.68 434 36.67 8 0.27 No 2006 485 93.90 518 125.24 8 0.84 No 2007 413 38.42 579 118.68 8 0.22 No 2008 375 39.83 414 36.19 8 0.48 No 1. The mean is estimated from a transect 1800 cm long. Table 3.38. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means. Site 7 Oak Hardwoods Midstory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 32 11.20 75 24.25 8 0.14 No 2005 61 3.67 47 9.95 8 0.22 No 2006 51 21.93 56 17.73 8 0.85 No 2007 76 13.55 117 25.28 8 0.19 No 2008 63 3.76 73 16.14 8 0.55 No 1. The mean is estimated from a transect 1800 cm long. Table 3.39. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means. Site 7 Oak Hardwoods Overstory Exclosure Paired control significant 1 Mean Mean1 Year SE SE df p value difference 2004 1675 30.25 1734 20.88 8 0.15 No 2005 1666 24.21 1632 39.17 8 0.48 No 2006 1629 32.11 1619 36.99 8 0.84 No 2007 1626 20.88 1663 26.27 8 0.31 No 2008 1674 11.98 1663 13.16 8 0.57 No 1. The mean is estimated from a transect 1800 cm long. 71 Table 3.40. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the understory. A standard t-test was used to test for differences between the two means. Site 8 Oak Hardwoods Understory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 713 60.86 815 110.49 8 0.44 No 2005 1226 110.11 881 67.55 8 0.028 Yes 2006 1044 91.58 1029 133.56 8 0.93 No 2007 1030 85.49 1027 68.22 8 0.98 No 2008 1006 42.67 939 47.58 8 0.32 No 1. The mean is estimated from a transect 1800 cm long. Table 3.41. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the midstory. A standard t-test was used to test for differences between the two means. Site 8 Oak Hardwoods Midstory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 290 34.56 378 30.03 8 0.093 No 2005 199 59.72 268 48.18 8 0.39 No 2006 292 59.38 323 48.95 8 0.70 No 2007 303 49.76 263 39.33 8 0.54 No 2008 273 33.87 309 18.53 8 0.38 No 1. The mean is estimated from a transect 1800 cm long. Table 3.42. Comparison of means between the paired control unit and the exclosure from the line intercept method measuring vertical cover in the overstory. A standard t-test was used to test for differences between the two means. Site 8 Oak Hardwoods Overstory Exclosure Paired control significant 1 1 Mean Mean Year SE SE df p value difference 2004 1725 27.39 1756 19.66 8 0.38 No 2005 1772 19.60 1756 19.13 8 0.57 No 2006 1768 13.19 1748 15.94 8 0.36 No 2007 1716 29.76 1720 10.49 8 0.90 No 2008 1758 5.80 1747 14.42 8 0.49 No 1. The mean is estimated from a transect 1800 cm long. 72 Table 3.43. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 1 Mixed Hardwoods Understory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 13.00 3.61 12.33 1.76 4 0.88 No 2005 13.33 6.06 18.00 7.00 4 0.64 No 2006 18.00 7.51 12.33 2.96 4 0.52 No 2007 36.33 14.68 8.67 1.20 4 0.13 No 2008 19.67 6.33 16.00 2.31 4 0.62 No 2 1. The mean is estimated from the number of stems sampled in a 2 m plot 1 1 Table 3.44. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 2 Mixed Hardwood Understory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 11.67 2.19 10.00 2.08 4 0.61 No 2005 13.00 3.21 6.33 1.45 4 0.13 No 2006 21.33 1.45 7.00 3.21 4 0.02 Yes 2007 29.67 7.42 9.00 4.16 4 0.07 No 2008 18.67 2.40 14.33 7.88 4 0.63 No 2 1. The mean is estimated from the number of stems sampled in a 2 m plot 1 1 Table 3.45. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 3 Mixed Hardwoods Understory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 13.00 1.15 14.33 5.24 4 0.82 No 2005 39.67 23.21 32.33 6.01 4 0.78 No 2006 43.00 16.92 48.33 1.76 4 0.77 No 2007 82.00 28.36 96.00 17.24 4 0.69 No 2008 41.33 10.97 46.67 3.38 4 0.67 No 2 1. The mean is estimated from the number of stems sampled in a 2 m plot 1 1 73 Table 3.46. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 4 Mixed Hardwoods Understory Exclosures Paired Control Signficant Mean1 Mean1 Year SE SE df P-Value difference 2004 14.67 2.91 20.33 1.67 4 0.17 No 2005 32.67 7.26 48.67 8.69 4 0.23 No 2006 48.67 7.17 83.00 13.23 4 0.08 No 2007 100.67 23.95 179.00 44.02 4 0.19 No 2008 51.00 10.21 78.67 9.21 4 0.11 No 2 1. The mean is estimated from the number of stems sampled in a 2 m plot Table 3.47. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 5 Mixed Hardwoods Understory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 6.33 0.67 8.00 1.15 4 0.28 No 2005 22.67 2.85 28.67 6.33 4 0.44 No 2006 85.00 17.62 98.33 7.17 4 0.52 No 2007 247.33 37.95 259.33 12.78 4 0.78 No 2008 87.33 11.29 97.67 3.38 4 0.43 No 2 1. The mean is estimated from the number of stems sampled in a 2 m plot 1 1 Table 3.48. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 6 Mixed Hardwoods Understory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 6.33 3.18 18.00 5.20 4 0.13 No 2005 24.33 6.84 17.67 7.67 4 0.55 No 2006 23.00 6.43 27.67 2.40 4 0.53 No 2007 44.00 16.04 50.33 13.30 4 0.78 No 2008 23.67 7.80 28.00 2.52 4 0.62 No 2 1. The mean is estimated from the number of stems sampled in a 2 m plot 1 1 74 Table 3.49. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 7 Mixed Hardwoods Understory Exclosures Paired Control Signficant 1 1 Mean Mean Year SE SE df P-Value difference 2004 7.33 2.33 8.00 1.15 4 0.81 No 2005 9.33 2.03 8.67 3.28 4 0.87 No 2006 23.33 2.91 39.33 13.38 4 0.31 No 2007 53.00 10.69 98.00 33.18 4 0.27 No 2008 23.00 2.31 43.67 14.19 4 0.22 No 2 1. The mean is estimated from the number of stems sampled in a 2 m plot Table 3.50. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 8 Mixed Hardwoods Understory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 12.00 5.86 13.67 2.03 4 0.80 No 2005 52.00 7.21 44.00 2.65 4 0.36 No 2006 57.00 4.62 68.33 19.10 4 0.60 No 2007 108.67 20.09 143.33 52.72 4 0.57 No 2008 55.67 5.84 65.33 16.34 4 0.61 No 2 1. The mean is estimated from the number of stems sampled in a 2 m plot 1 1 75 Table 3.51. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 1 Mixed Hardwoods Midstory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 2.33 1.20 2.33 0.33 4 1.00 No 2005 2.67 0.88 0.67 0.67 4 0.14 No 2006 5.33 1.33 3.67 1.20 4 0.41 No 2007 2.00 1.53 1.33 0.33 4 0.69 No 2008 3.00 1.00 1.33 0.88 4 0.28 No 2 1. The mean is estimated from the number of stems sampled in a 5 m plot 1 1 Table 3.52. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 2 Mixed Hardwoods Midstory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 0.67 0.67 1.33 0.88 4 0.58 No 2005 1.00 0.58 0.33 0.33 4 0.37 No 2006 2.33 0.88 2.33 0.33 4 1.00 No 2007 1.33 1.33 1.33 0.88 4 1.00 No 2008 1.33 0.88 1.00 0.58 4 0.77 No 2 1. The mean is estimated from the number of stems sampled in a 5 m plot 1 1 Table 3.53. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 3 Mixed Hardwoods Midstory Exclosures Paired Control Signficant Mean1 Mean1 Year SE SE df P-Value difference 2004 5.33 2.33 2.00 0.58 4 0.24 No 2005 4.00 1.53 4.33 1.33 4 0.88 No 2006 9.33 2.03 11.00 1.53 4 0.55 No 2007 10.67 2.19 26.33 6.67 4 0.09 No 2008 6.33 1.45 8.33 1.86 4 0.44 No 2 1. The mean is estimated from the number of stems sampled in a 5 m plot 76 Table 3.54. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 4 Oak Hardwoods Midstory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 3.67 1.67 5.67 2.91 4 0.58 No 2005 5.67 1.76 3.67 1.20 4 0.40 No 2006 13.67 2.03 16.00 3.79 4 0.62 No 2007 16.67 1.86 24.00 4.58 4 0.21 No 2008 8.67 1.76 11.33 1.76 4 0.35 No 2 1. The mean is estimated from the number of stems sampled in a 5 m plot 1 1 Table 3.56. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 5 Mixed Hardwoods Midstory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 1.67 0.88 1.67 1.67 4 1.00 No 2005 0.67 0.67 0.33 0.33 4 0.68 No 2006 5.33 1.86 4.33 2.85 4 0.78 No 2007 9.00 5.57 8.00 4.04 4 0.89 No 2008 3.67 2.19 3.67 2.19 4 1.00 No 2 1. The mean is estimated from the number of stems sampled in a 5 m plot 1 1 Table 3.57. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 6 Oak Hardwoods Midstory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 1.00 1.00 1.00 1.00 4 1.00 No 2005 1.00 1.00 0.33 0.33 4 0.56 No 2006 2.67 2.19 1.67 0.88 4 0.69 No 2007 2.00 1.15 1.67 0.88 4 0.83 No 2008 2.00 1.15 1.00 0.58 4 0.48 No 2 1. The mean is estimated from the number of stems sampled in a 5 m plot 1 1 77 Table 3.58. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 7 Oak Hardwoods Midstory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 3.00 0.58 3.33 0.33 4 0.64 No 2005 0.58 1.00 2.33 1.20 4 0.64 No 2006 8.67 1.67 10.00 1.00 4 0.53 No 2007 10.33 3.71 13.00 5.29 4 0.70 No 2008 5.67 1.20 6.67 0.88 4 0.54 No 2 1. The mean is estimated from the number of stems sampled in a 5 m plot 1 1 Table 3.59. The annual mean of stem density, with a standard t-test to compare differences between the exclosure and its paired control site. Site 8 Oak Hardwoods Midstory Exclosures Paired Control Signficant Mean Mean Year SE SE df P-Value difference 2004 2.33 1.33 5.00 1.00 4 0.18 No 2005 8.67 1.33 7.33 2.60 4 0.67 No 2006 17.33 3.53 19.00 4.16 4 0.78 No 2007 20.00 4.51 22.00 3.61 4 0.75 No 2008 11.00 3.06 13.33 2.85 4 0.61 No 2 1. The mean is estimated from the number of stems sampled in a 5 m plot 1 1 78 CHAPTER 4: DISCUSSION It is common for natural resource managers to face challenges in their quest to conserve resources; however, the specific challenges that the MDMVA faces at FCTC are unique. The primary mission of FCTC is to function as a military base; coincidently the base encompasses 7,560 acres of diverse flora and fauna. The INRMP dictates a secondary objective at FCTC of ecosystem restoration and rehabilitation. It is implicit in the INRMP that management of the deer herd at FCTC plays an integral role in meeting the criteria promulgated within the INRMP. To manage the deer herd the MDMVA needs to mitigate deer effects on the ecosystem at FCTC and serve as an intermediary between stakeholders wanting different outcomes with the deer herd (e.g. population size). To provide assistance managing the deer herd at FCTC, I examined and developed a management policy for the MDMVA addressing the deer herd based on the outcome of the research conducted from 2004-2008. I have accounted for the nuances at FCTC that differ from other deer management areas and policies. Winter and Spring Capture Success of the winter capture period depended on the severity of winter weather. I attribute the lack of overall success during the winter capture period in 2005-2006, where only 3 deer were captured, to an extremely mild winter. Ironically, that year clover traps were placed in the field earlier than in any other year, but FCTC lacked snow cover for nearly the entire trapping period. Snow 79 cover limits deer access to forage, while cold temperatures increase the caloric and nutritional requirements for the survival of deer. Therefore, a mild winter reduced the allure of bait piles in and around the trap site. Other factors that decreased capture success included capturing other species (e.g. turkeys [Meleagris gallopavo], fox squirrels [Sqiurus niger], cottontail rabbits [Sylvilagus floridanus], opossums [Didelphis virginiana], raccons [Procyon lotor], and various passerines) and having the trap tripped inadvertently for numerous reasons. Successful capture of deer relied on snow cover, cold temperatures, and knowledge of movement patterns of deer at FCTC. The distribution of traps was relatively homogenous throughout FCTC and FCRA, and every trap site had a considerable amount of deer sign. Areas 5 and 7 proved problematic, since I did not capture any deer in these areas during any of the winter capture seasons (a map of FCTC is contained in Figure 1.2 on page 5). In part, this difficulty arose from a probable lower density of deer and bad luck (traps with false trips, high amount of raccoon and turkey activity). Capturing neonatal fawns in the spring proved more time intensive than anticipated. Other studies that captured neonatal fawns did not require the personnel hours that this study required capturing individuals. Pusateri (2003) averaged 7.5 and 10 personnel hours per fawn capture, respectively, for two spring capture periods of neonatal fawns. Over a three-year period, Hiller (2007) averaged 25 hours per neonatal fawn capture, while it improved each succeeding year with 39.2, 30.8, and 12.3 hours per neonatal fawn capture respectively. I 80 averaged 46.4 hours per neonatal fawn capture, in which the first year took 58.5 hours per neonatal fawn and 34.2 hours per neonatal fawn in 2007. The large decrease year-to-year required to capture a neonatal is indicative of the subtleties necessary to determine an ideal location to search for neonatal fawns, which search techniques to use, and observers refining their search image of fawns. My field team‟s ability to capture older neonatal fawns changed during the second season, and that led to an increase in the number captured (see spring capture in Methods for more detail). A combination of techniques in all likelihood led to the greatest chance for capture of neonatal fawns. Locating solitary does that are in habitat conducive to fawning and exhibiting behavior indicative of having a fawn nearby led to the highest capture success. Closely monitoring radio-collared does during the fawning season contributed to several captures of neonatal fawns. I theorized that when does with radio-collars displayed an extremely high degree of a fidelity to a certain site; they most likely had recently given birth. The area was then grid searched, and on several occasions, neonatal fawns were either captured or observed. Survival and Fate Assessment The greatest source of mortality was that of hunter harvest, where hunter harvest accounted for 64% of all mortalities followed by predation at 18% and stress and trauma related mortalities at 9% in winter captured deer. Hunter harvest accounted for 85% of the mortalities for adult female deer in Hiller‟s 81 (2007) south central Michigan deer study. Hiller (2007) did not radio-collar any adult males, so no comparisons could be made with respect to adult male deer. Hunter harvest was the greatest source of mortality (65%) with deer vehicle collisions second (27%) in a study in the southwestern lower peninsula of Michigan in adult female deer (Pusateri 2003). The preponderance of hunter harvest as the leading cause of mortality clearly illustrates the integral role hunting plays in managing the size of the southern Michigan deer herd. Predation of adult and yearling radio-collared deer occurred during the winter months when snow depths and conditions disadvantage deer to coyotes (Canis latrans) the majority of the time (Patterson and Messier 2000). The conditions during my study were similar to that of the Patterson and Messier (2000) study in terms of daily mean temperature in January where both sites averaged approximately -5 °C and snow depths were not different, they both did not exceed 30 cm (Midwestern Regional Climate Center, Champaign, Illinois, USA). During the one notable relatively mild winter in 2006, no predation of radio-collared deer occurred. The rate of predation may vary with the severity of winter and density of the deer herd. Spring captured neonatal fawns did not have a statistically significant single source of mortality. Possible coyote predation occurred 3 times, and accounted for 37.5% of mortalities; while hunter harvest accounted for 2 additional mortalities at 25%. On two occasions where coyote predation was the likely source of mortality, the head of the fawn was discovered buried underground and no body was recovered. The necropsy results of the 82 examination of both fawn heads indicated canine bites that commonly occur with coyote predation. The survival of fawns over their first 180 days was 0.49 (95% CL = 0.23 – 0.75), and the annual survival was estimated at 0.29 (95% CL = 0.15-0.57) which differed significantly from the 0.75 (95% CL = 0.59–0.91; Kaplan-Meier method) estimate for southwestern Lower Michigan (Pusateri-Burroughs et al. 2006), and Hiller‟s (2007) estimate of 0.51 (95% CL = 0.37–0.66). A smaller sample size, and thus an increased sensitivity to stochastic effects, may account for some of the lower survival rates observed in my study (n=14); compared to Pusateri (2003) (n=75) and Hiller (2007) (n=34). Differences in human density and land use among the study sites may also affect the mortality sources. It is possible that FCTC has a higher coyote density than that of the other study sites where people may harvest coyotes thus reducing the density in those areas. SAK Model Many state agencies use the SAK model to estimate the abundance of deer. The application of the SAK model typically occurs at a relatively broad geographic scale, and model fitness for precise estimates occurs with populations over 10,000 animals (Millspaugh et al. 2009). Model precision, as evaluated by Millspaugh et al. (2009) is dependent on the accuracy of harvest data that forms the model. The SAK model typically used by agencies utilizes harvest data to estimate the proportion of mortality in male deer caused by hunting. The SAK Mt model I derived depended on data from radio-collared deer 83 to estimate the amount of male deer mortality from hunter harvest. In addition to the parameter estimates derived from radio-collared deer data to increase precision, the hunter harvest data at FCTC is for every individual deer and this avoids the possibility of bias from hunters‟ selectively deciding what deer to check or not to check. The harvest data is not biased on other factors that are problematic with voluntary check stations, such as distance to the check station, gender, size, and other intangible factors that may or may not make the hunter feel inclined to check the deer. The SAK models show a desirable population trend from a management standpoint with the population decreasing. A significant reduction in the size of the deer herd has been the overarching goal since the establishment of the public hunt in 1986. Forward Looking Infrared (FLIR) The FLIR conducted in March of 2007 was flown under better conditions (i.e. snow cover, colder temperatures, and lack of residual leaf cover on trees obscuring thermal images) than was the November 2007 survey. The March results were in line with expectations based on SAK model estimates and my personal observations of the deer herd at FCTC. Results from the November survey were slightly lower than the anticipated size of the deer herd. The November survey took place when trees still had some residual leaf cover and no snow cover, which could lower detection rates. Additionally, the March survey had a greater proportion of deer in open fields where detection rates are at or 84 near 100%, where the November survey had fewer deer and a smaller proportion of deer in open fields. The majority of deer in the November survey were detected in deciduous forests where there is a detection rate ranging from 82%87%. The cover deer were typically located in during the November survey, weather, and leaf cover may have affected the results slightly. The spatial distributions of deer from the FLIR surveys detected were different. The March flight depicted a heterogeneous distribution, while the distribution of deer in November was more homogeneous. The larger group sizes and distribution of deer from the March flight is attributed to seasonal variability in resource selection and behavioral differences. I attribute the small group sizes in mid November to behavior associated with the rut and breeding cycle of deer. The March flight results are representative of deer selecting cover and forage at FCTC in addition to being in their typical social units. I would recommend that the MDMVA have a FLIR performed every 5 years to verify the abundance of the deer herd relative to the other indices (the SAK model, vegetation sampling, and hunt biodata). The SAK model and hunt biodata coupled with the vegetation sampling will indicate if the size of the deer herd is increasing significantly. A significant increase in the abundance of deer at FCTC will affect the MDMVA long term management goals. The ideal time to conduct a FLIR is during the winter with an absence of leaf cover obscuring thermal images. In addition, the likelihood of misidentifying thermal signatures as a deer is minimal, while deer are more distinguishable. 85 Spotlight Survey In 2004 and 2005 the spotlight survey was performed and its‟ effectiveness was greatly limited by lack of suitable terrain. A spotlight survey requires terrain that has numerous areas where deer can be observed from a distance. In the case of FCTC, areas offering observers a reasonable chance to spot deer is limited. Training areas 1, 2, and the northern edge of 9 offer the greatest likelihood. Originally, it was believed that training areas 3, 4, and six might also offer areas with relatively good chance at observing deer. However, the few open areas in training area 3 and 4 had a variety of grass and shrub species that are tall enough to obscure deer from an observer‟s view. FCTC‟s size and landscape composition limited the effectiveness of the spotlight survey. Hunt Biodata Since the inception of the public hunt at FCTC, the biodata gathered has indicated an improvement in the quality of forage available, implying a decrease in deer density. Biometric indices measuring the physical characteristics of deer have steadily improved. The mean deer weight for fawns harvested has increased significantly from 1986 to 2009. In 1986, female fawns averaged 42.2 (S.E. ± 1.1) lbs versus a mean weight of 63.7 (S.E. ± 3.9) lbs in 2008 (only one female fawn was harvested in 2009 and it weighed 59 lbs). Male fawns averaged 44.5 (S.E. ± 1.0) lbs, and 69.5 (S.E. ± 3.9) in 1986 and 2008 respectively. The mean beam diameter has increased across the board in year old bucks with statistical significance. In 1986 bucks averaged 18.8 (S.E. ± 0.51) 86 mm, while in 2009 the mean beam diameter was 21.1 (S.E. ± 0.76) mm. The average number of antler points in 1986 was 4.17 (S.E. ± 0.23) points while in 2009 the mean was 4.7 (S.E. ± 0.38) points. A decreasing deer density from 1986 to 2009 has allowed the quality of forage to improve, which accounts for an increase in mean beam diameter, number of antler points, and the weight of fawns in the hunt biodata. Beam diameter of yearling males is indicative of the quality of the forage during summer months (Severinghaus and Moen 1983, Schmidt et al. 2001). The high deer densities that cause the quality of forage to degrade does not happen immediately; so, changes in the physical characteristics of deer correlated with the quality of forage subsequently lags behind an increase in deer density (Patterson and Power 2002). Such a lag between deer density and range quality has the potential to encourage pronounced fluctuations in deer populations if managers base decisions solely on biodata (May 1981; Messier et al. 1988; Fryxell et al. 1991). Despite being a useful tool, biodata alone is inadequate in detecting changes in range conditions and approximating deer densities. The time lag inherent with biodata being correlated with changes associated with range conditions can prevent a managerial change before the deer density increases to a point where the herd may negatively impact their range. To assess changes in the abundance of the deer herd the biodata should be used in conjunction with continued collection of data from the exclosures and SAK model estimates. 87 Resource Utilization and Habitat use The results from the compositional analysis indicate that deer at FCTC have a strong selection preference for the two vegetation types that the MDMVA is trying to preserve and restore (oak hardwoods and prairies „fields‟). Adult deer heavily relied on both types of hardwoods and fields. While fawns were also reliant on the oak and mixed hardwood, their most highly selected vegetation type was fields. The hardwoods preference of adults most likely stems from the fact they provide good mast crops that have high nutritional value. Fawns selected habitat that provided cover over habitat that provided forage because their mother fulfilled a significant portion of their nutritional needs. Until their first autumn fawns will continue to nurse, but by 10 weeks the have become functionally weaned (Marchinton and Hirth1984). This selection of habitat use by deer complicates the MDMVA‟s ability to achieve their management goals because deer act as keystone herbivores and can influence the composition of a landscape. The influence that deer have on a landscape‟s composition is density dependent. Alverson et al. (1988) suggested that deer densities should not exceed 4 deer/km2 (10.4 deer/mi2). Since the deer herd is selecting the oak hardwoods and prairie resources, they are also more likely to have a detrimental impact on the very landscape the MDMVA is trying to preserve. The oak hardwoods and prairies provide deer with mast crops and cover for fawns, so it is not unexpected to observe these trends of resource utilization at FCTC. If the 88 MDMVA is going to achieve their natural resources management objectives, it‟s essential to manage the size of the deer herd. In 2009, when the MDMVA had assumed responsibility to monitor the exclosures, Liparis liliifolia, a species listed as special concern, was found inside the exclosure at Site 3. Two individuals of L. liliifolia were observed during the annual vegetation survey of the exclosures. The occurrence of this orchid is rare throughout its range, and the prevalence of L. liliifolia leads to a relatively low probability of detection. There is not sufficient data at the current time to conclude that deer herbivory has limited the abundance of L. liliifolia. L. liliifolia is found in a variety of habitats; it is usually found in areas that have been disturbed, and prescribed burns are beneficial for the species (Michigan Natural Species Inventory, Lansing, Michigan, USA). The contiguous closed canopy forest at FCTC differed from the landscape composition of two other deer studies in the southern Lower Peninsula of Michigan conducted by Pusateri Burroughs et al. (2006), and Hiller (2007) where 34% and 20% of their study area was composed of agricultural fields respectively. Differences in the study areas‟ landscape composition did not seem to influence home range size. The sizes of home ranges of deer amongst the three studies in southern Michigan were similar. The mean home range size for fawns in southwestern Michigan was 75.36 ha (SE ± 4.47) with values ranging from 38.38 ha to 118.5 ha (Pusateri 2003). Comparatively, my annual home range size for fawns was 73.22 (SE ± 23.27) ha. My average home range size for adult females was 89 132.22 (SE ± 16.05) ha, which is inline with home range estimates from south central Michigan at 140.4 ha during the growing season, and 157.7 ha for deer > 6 months and non-dispersers in southwestern Michigan (Hiller 2007, Pusateri 2003, respectively). I did not use deer with less than 30 locations for comparisons because a small sample size may result in an underestimate of the size of fixed kernel home range estimates leading to poor bandwidth selection and any outliers have greater weight (Seaman et al. 1999). I did not observe dispersal in any adult deer, male or female. Adult deer demonstrated a high degree of fidelity to their capture areas. Yearling males were the only deer that dispersed in my study. I adopted Kenward et al. (2001, 2002) definition of dispersal; a white-tailed deer permanently emigrated from a natal range to a distinct adult range, such that predispersal locations did not overlap postdispersal locations. Despite being a fenced in area, the deer herd at FCTC can move on and off the base. The fence is rather porous in terms of constraining a deer‟s movement. Male deer dispersal from their natal range is a common behavior; Ozoga and Verme (1985) suggested that yearling males disperse from their natal ranges due to domination by older female relatives and a basic drive for fraternal membership to establish their dominance. Holzenbien and Marchinton (1992) demonstrated that maternal domination did serve as a mechanism to promote yearling male dispersal, and they theorized that their dispersal reduced inbreeding and increased the inclusive fitness of their mothers. 90 Vegetation Sampling In 2004, when I selected the 8 exclosure and paired control sites I attempted to find similar type sites in a broad spectrum of two basic cover types, oak hardwoods and mixed hardwoods. I selected sites that were homogenous and relatively identical in regards to the exclosure and paired control site. Upon completion of the vegetation sampling in 2008 of the exclosures and paired control sites, there is no evidence that deer are significantly affecting their range at FCTC. Data from exclosures and the paired control sites indicate the current deer densities produce minimal impacts regarding changes in the structure and composition of forest types. The results from the vegetation analysis indicate that the deer herd has not fundamentally changed the species composition and vegetation structure at FCTC. The tolerance to deer herbivory by the plant community varies from one geographic region to another with varying biotic and abiotic components (Horsley et al. 2003, Côté et al. 2004). The effect on forest resources of a given deer density is dependent on the surrounding landscape and available forage (deCalesta and Stout 1997). DeCalesta and Stout (1997) assessed studies by Tilghman (1989), deCalesta (1992), and Palmer et al. (1997) concluding that those studies indicated that the impact of deer on a landscape needed to be scaled to the specific area where they occur. Kittredge and Ashton (1995) claimed diverse forest regeneration of New England forests was incompatible with deer densities in excess of 23 deer per 91 square mile. The precise density of deer that could roam FCTC without causing significant changes to the forest structure and composition is difficult to estimate. At the same time, managing the size of the deer herd to a level that precise is an impossible task. The density levels would fluctuate year to year with different weather patterns and food availability, especially in the case of FCTC where oak mast may be highly variable one year to the next. However, the range of the deer densities over the duration of this study, in which few changes occurred in structure and composition of forest types, would be a reasonable range in the deer density. Relative Deer Density Studies concerning the sustainability of oak and hickory forests in the Lower Peninsula of Michigan with respect to a specific deer density are nonexistent at the current time. However, the concept of relative deer density (RDD) is applicable to FCTC. RDD, a concept proposed by deCalsta and Stout (1997) incorporates theories by McCullough (1984) and Marquis et al (1992) and is consistent with the theory of managing the effects of deer relative to their occurrence. McCullough (1984) defined carrying capacity (K) as the residual population productivity declining to zero. Carrying capacity is further defined as the maximum number of animals an environment will support on a sustained basis. DeCalsta and Stout (1997) combined McCullough‟s concept of K carrying capacity and his population recruitment curve (McCullough 1979) as backdrop for defining RDD relating to sustaining ecosystem components and deer harvests 92 where RDD is the proportion of deer density at K, carrying capacity (Figure 4.1). It allows stakeholders to prioritize their management decisions based on specific needs at local or regional levels. It should be noted that deCalasta and Stout (1997) predicted that a RDD that would sustain biodiversity is < 1/3 the maximum sustained yield of deer for hunter harvest. The RDD that would allow the maximum number of resources maintained in an ecosystem, is considered sustaining biodiversity by deCalsta and Stout (1997), and is annotated as RDD s (Figure 4.1) Figure 4.1. Relative deer density (RDD) displayed on McCullough's (1984) graph. DeCalasta and Stout (1997) approximated the location of RDDS, the level associated with sustaining biodiversity, RDDT, the level at which timber productivity is sustained, and RDDI, the level associated with maximum sustained yield of deer numbers for harvest. 93 Marquis et al. (1992) work on a deer impact index is integrated with McCullough‟s concepts to form the idea of a RDD. Marquis et al. (1992) postulated at any given deer density, the amount of forage available in a landscape determines the impact of deer herbivory on forest regeneration. This suggests that deer at even low densities can significantly affect a landscape. The protocol and management recommendations I have outlined herein factor in the theory of RDD. The continued monitoring and analysis of the data from the deer exclosures will indicate if the deer herd has reached a size where they are significantly altering their range. Those significant changes in vegetation are indicative of a higher density deer herd. The biodata gathered during the public hunt is sensitive to changes in deer density, as the density increases, quality forage decreases, and as quality forage decreases physical characteristics of deer change (e.g. beam diameters, point counts, and weight all decrease). The change in these physical characteristics is most notable in oneyear-old males. The FLIR and SAK model estimates provide population estimates with a level of precision that will allow managers at FCTC to make sound decisions. The population estimates used in conjunction with the deer density indices constitutes a RDD for FCTC. In addition to the aforementioned methodology contributing to a RDD at FCTC, it would behoove the MDMVA to routinely walk and inspect FCTC for deer browse lines, overwinter deaths, and other deer sign every spring to further validate the RDD. The concept of RDD is applicable to the type of management done at FCTC. Proper management of the deer herd and estimation of the deer density 94 rely on several types of qualitative and quantitative methods estimating the approximate density of deer. The MDMVA needs to incorporate the host of indices estimating deer density and browse effects on the forest composition and structure. In addition to incorporating the various methods, managers should not overreact to a single set of data that may change in one-year‟s time, yet managers should be aware that those changes might be indicative of a change in the RDD. Significant changes in biodata may occur over a single year due to the variability of weather and other stochastic effects. If the data from one index or population estimate indicates a significant change, the MDMVA should assess the other indices, and consider the possibility of revising the current management protocol if appropriate. The managerial scenarios and recommendations are discussed later on in this document. Hunter Participation Hunter participation is a critical aspect in managing the deer herd at FCTC. Over the last few years a trend has developed where the number of hunters participating the public hunt has declined significantly (Figure 4.2). An insufficient amount of hunter participation may decrease the number of deer harvested to a point where the deer population will begin to increase. Despite the significant decrease in the number of hunters and hunter days, the hunting efficiency in terms of number of hunter days per deer harvested has remained rather static (Figure 4.3). 95 Annual Trend of Hunter Participation at FCTC and Number of Deer Harvested 1800 1600 1656 1412 1400 1289 1200 1000 800 739 600 457 400 200 173 156 77 156 73 0 2004 2005 2006 2007 2008 59 2009 Year Hunters Deer harvested Figure 4.2. The annual trend of total hunter days in the public hunt at FCTC and number of deer harvested (from data provided by the MDMVA). 96 Hunter Efficiency at FCTC 0.14 Harvest Rate per Hunter Day 0.13 0.13 0.12 0.12 0.11 0.11 0.10 0.10 0.10 0.09 0.08 2004 2005 2006 2007 2008 2009 Year Figure 4.3. The hunter efficiency at FCTC based on the total number of hunter days and deer harvested. No hunter efficiency estimate is available for 2007 because the total number of hunter days is unknown. Management Implications The public hunt has played an integral part in managing the deer herd, reducing the population below maximum carrying capacity and within levels such that the deer herd does not significantly affect their range. With the current size of the deer herd at FCTC, deer should not negatively influence the MDMVA ecosystem restoration and rehabilitation goals. It has taken approximately 30 years to achieve the primary goal in significantly reducing the size of the deer herd at FCTC. Continuation of the public hunt is critical in maintaining the deer 97 herds‟ current abundance. However, the public hunt does not guarantee that the deer herd will not increase in size. The reduction in the size of the deer herd consequently reduced the number of individuals hunting annually at FCTC. Hunter participation may continue to decline, thus reducing the number of deer harvested at FCTC that will in turn lead to growth in the size of the deer herd. Lacking a significant harvest of deer over several years the deer population may then grow exponentially. The reproductive potential of deer coupled with less hunter participation poses a great risk to the MDMVA long-term management goals. For these reasons, it is essential that the MDMVA use indices (SAK models, FLIR, monitor exclosures and paired control sites, maintain biodata collection from the public hunt) to monitor changes in the population. The cost of a FLIR will vary to a degree. Vision Air charges by flight time, a per diem expense, and the rates may change over time. The costs of the two flights were $7,900 and $9,170, with the November 2007 flight costing more because of the weather delay and rate change from $700 per hour to $725 per hour. There are alternatives to holding an annual public hunt to manage the deer herd. Though undesirable because of expense and a likely tarnished reputation to FCTC, an annual cull could act as a substitute or in addition to a public hunt to harvest deer. Additionally, an annual cull does not fulfill the obligations outlined by the Sikes Act, whereas the public hunt is within the spirit of the legislation. The public hunt should be considered as a success in attaining 98 the goal of reducing the size of the deer herd, and to drastically change the role of the public hunt is not justifiable under the present set of circumstances. Drastic changes could permanently alienate the hunter stakeholder group and hunter participation could drop below a threshold that harvests a sufficient number of deer. To maintain interest in the public hunt the MDMVA needs to focus on the public relations with outreach programs. The MDMVA must continue the annual After Action Report (AAR) that summarizes the past years‟ deer hunt at FCTC and has multiple stakeholders in attendance ranging from the public, the MDNRE, and base personnel. This annual meeting allows communication amongst various stakeholders and raises involvement in hunters while fostering ideals integral for the continued success of the public hunt. It is important to remember that the primary stakeholder in this case is the MIANG and the base‟s function is to serve as a training facility for military operations. The deer hunter group plays a critical role by assisting the MDMVA in management of the deer herd. Other stakeholders at FCTC include the MDNRE, MNFI, Kalamazoo Nature Center, and Native Connections where their interests are related to deer density dependent effects on the ecology of FCTC. The MDMVA should consider engaging communication among all stakeholders to raise the awareness levels of the various interests and outcomes desired by all parties involved. This objective could be achieved with a meeting or a newsletter with articles representative of the various positions. 99 For the stakeholders to accept the management policy at FCTC they will need not only to be in agreement with the recommended outcomes, but also believe that the prescribed course of action will achieve those goals (Enck et al. 2006). Therefore, the management policy at FCTC needs to engage the various stakeholder groups in the process of managing the deer herd (Lancia et al. 1996, Riley et al. 2002). Lischka et al. (2008) stated that the traditional management practices solely focused on the size of wildlife populations might be insufficient in achieving the desired outcomes. The acceptance capacity of the deer is influenced by perceptions of the effects of interactions (Lischka et al. 2008). At varying levels of abundance, in terms of both social carrying and biological carrying capacity, deer have differing effects on a social and ecological level (Figure 4.4). Hunter perception of deer abundance may confound the MDMVA‟s ability to manage the population. 100 Low Limited negative deer browse effects Hunter harvest rates high, but less "trophy bucks" Deer pose problems to day to day operations at FCTC Deer affect daily operations at FCTC Deer starvation during winter Larger deer Limited deer impact on landscape Deer Hunter Satisfaction Significant negative deer impacts on landscape Deer seen with regularity High Lower deer harvest rates, but "trophy bucks harvested Deer vehicle collisions occur, but not considered problematic Small deer are harvested Low number of deer harvest annually No "trophy bucks" Significant deer browse effects on Decreased satisfaction among deer range hunters Deer vehicle collisions occur, but not considered problematic Deer seldom seen Deer starvation during winter Low High Deer Density Figure 4.4. The ecological and perceived impacts of deer on stakeholders at high and low abundance levels with respect to stakeholder satisfaction and deer density. The stakeholder satisfaction is relative to the perception that individuals or groups of individuals might have about the deer herd (e.g. too few deer, too many deer because they are causing damage to the ecosystem). Whereas the relative deer density is in terms of biological levels being low or high with respect to resource availability acting as a constraint on population growth. Other stakeholders may have different views than those expressed by the deer hunters. Van Deelen and Etter (2003) stressed the importance of integrating the understanding of the functional response of deer hunters to declining deer densities into management practices. Data indicates that as deer densities 101 decline the relationship to deer hunter effort required to harvest a deer is a curvilinear function, and an asymptotic relationship exists with the effort to harvest or see a deer in Midwestern habitat when deer density decreases below 15 deer per km2 (~39 deer per mi2) (Figures 4.5 and 4.6) (Van Deelen and Etter 2003). Van Deelen and Etter (2003) used five data sets to create Figures 4.5 and 4.6, with each data set resulting from an independent study. The study sites in the Midwest ranged from central Wisconsin to northern Michigan to Ontario (Van Deelen and Etter 2003). The results of the metadata produce identical relationships, though the rates varied site to site, which yielded 5 independent functions (Figure 4.5 and 4.6). It is reasonable to assume that FCTC hunters are experiencing a similar relationship to these other studies. Given the current density of deer at FCTC hunters may experience significant increases in the amount of time needed to either harvest or see a deer. 102 Figure 4.5. Deer seen as a function of deer density in five Midwestern data sets; the figure is a reproduction from Van Deelen and Etter (2003). 103 Figure 4.6. Deer killed as a function of deer density in five Midwestern data sets; the figure is a reproduction from Van Deelen and Etter (2003). 104 The significance of the relationship of deer density to that of hunter effort being curvilinear is twofold. One, hunters may become discouraged when deer densities require an exponential increase in effort to harvest or see a deer, and secondly the perceived density of deer by hunters at levels less than approximately 15 deer km2 may be significantly less than the true density yielding a perception gap between the true density and perceived density (Figure 4.7). Figure 4.7. Hypothetical relationship between hunter effort, perceptions, and deer density. If hunter perception of deer density is based on an assumed linear relationship but is in reality hyperbolic, then for any given level of effort hunters will perceive fewer deer than there actually are. Moreover, this bias will accelerate as deer density decreases. The figure is a reproduction from Van Deelen and Etter (2003). 105 The danger of this perception gap of deer density may lead to a further decline in hunter participation at FCTC. A decrease in hunter participation potentially could result in a hunter harvest that is insufficient to maintain the current size of the deer herd at FCTC. As a result, the deer population at FCTC will increase in all likelihood. One scenario is that over a relatively short time of 2 to 3 years the deer population grows exponentially and the deer densities are again at or near the biological carrying capacity. Another possibility is, as the deer population increases the number of hunters increase, which should yield higher harvests. The later scenario may create a situation where the population would increase, then decrease as hunter participation would increase, and then subsequently decrease as the deer population changed periodically. A hunter survey conducted by the MDMVA could be insightful into the perception gap associated with hunter effort and deer density. The hunter survey serves multiple purposes ranging from being a part of public relations, it weighs public opinion, and it assesses support for or against an action. If a hunter survey were performed every 3-5 years, or before changes were made to the current hunt policy, it could potentially predict an increase or decrease in hunter participation. To maintain steady interest in the public hunt, the MDMVA is in a unique situation to draw hunters with special hunt periods. The MDMVA already hosts a special hunt in September to Purple Heart recipients. The MDMVA could consider several different special hunts to bolster hunter participation and raise 106 the awareness of hunters in Kalamazoo and Calhoun counties with a crossbow hunt, a youth hunt, or a second opener. In doing so, the MDMVA should consider engaging their core-hunting base of group leaders, and or the numerous military base personnel that hunt. Alienation of these two groups would be detrimental in the long-term continued success at FCTC of deer herd management. An additional option to arouse public interest in the deer hunt at FCTC is publishing articles in local newspapers touting FCTC‟s relative uniqueness in the immediate area of large contagious forested tracts of land offering a quality deer hunt. 107 Management Recommendations  Continue to collect and analyze public hunt biodata. o If biodata indicates a significant decrease in physical characteristics of the deer herd, a change in management of the herd may be necessary.  Maintain interest in the public hunt by continuing and developing an outreach program.  It is imperative that the MDMVA assess the FCTC area each spring for deer browse lines, and dead deer because these qualitative assessments provide insight into the state of the ecological well being of the deer herd and ecosystem at FCTC.  Continue to monitor and maintain exclosures and paired control sites, and test for significant statistical differences every two years.  Use SAK model to estimate size of deer herd.  Contract a FLIR survey every 5 years in January, or when multiple population indices indicate a significant change in the deer density to validate changes in the indices.  Conduct a hunter survey every 3 to 5 years  Contact an outreach specialist in 5 to 8 years to learn what resonates with the public at the current time with respect to deer hunting management. 108 These management recommendations are based on the current state of FCTC; however, management of a deer herd is a dynamic endeavor. The set of circumstances might change in the near future, or the long-term future. I have tried to anticipate various scenarios and incorporate the management techniques applicable for those situations. The MDMVA needs to continually assess, and decide the appropriate course of action under a given set of circumstances with the methodology prescribed in estimating the size, and quantifying effects of the deer on their range. 109 APPENDICES 110 APPENDIX A Historical Biodata of the deer herd at FCTC 111 ANOVA and a Tukey‟s studentized range test were conducted to examine if FCTC was significantly different from Allegan, Barry, Calhoun, and Kalamazoo counties with respect to beam diameters and antler points. Multiple-comparisons were restricted to 2001-present because FCTC biodata is a subset of both Calhoun and Kalamazoo counties and there is no logistically feasible way to separate the biodata until 2001. There has been unique biodata kept for FCTC, but it is also incorporated into the biodatabase of Calhoun and Kalamazoo counties maintained by the MDNR. The results from ANOVA with respect to total points show a statistically significant difference among the five areas from 2001-2004 (F4, 3497=9.33; P<0.0001). The Tukey studentized range test indicated that the mean number of total points for FCTC was significantly less than for Allegan, Barry, Calhoun, and Kalamazoo Counties (Table A.1), while other areas had no significant difference. Looking at the trend of consecutive years from 2001-2004 ANOVA indicated a statistically significant difference in beam diameters with (F4, 3681=8.64; P<0.0001) among the five areas. The Tukey studentized range test showed that the mean beam diameter for FCTC was significantly less than that for Allegan, Calhoun, Barry, and Kalamazoo Counties, while there was no significant difference among the other areas (Table A.2). Figures A.1 and A.2 display a graphical representation of the trends for the five areas from 1987 to 2004, they compare the annual means of number of points and beam diameter over time. The results indicate that even with reintroduction of hunting to FCTC addressing issues of 112 overabundance there is evidence that the deer herd‟s condition is still below standards of the surrounding areas, and areas with similar habitat. Table A.1. The results from a Tukey studentized range test comparison of yearling bucks total point counts between FCTC and surrounding counties from 2001-2004 showing the difference of the means between areas with a 95% confidence intervals. FCTC was significantly less than the four other areas. Simultaneous 95% Difference between Confidence Limits Area of Comparison Means Lower Upper FCTC and Allegan -0.75199 -1.14 -0.36 FCTC and Barry -0.43028 -0.83 -0.03 FCTC and Calhoun -0.71068 -1.19 -0.23 FCTC and Kalamazoo -0.54442 -0.97 -0.11 Table A.2. The results from a Tukey studentized range test comparison of yearling bucks mean beam diameters between FCTC and surrounding counties from 2001-2004 showing the difference of the means between areas with a 95% confidence intervals. FCTC was significantly less than the four other areas. Simultaneous 95% Difference between Confidence Limits Area of Comparison Means Lower Upper FCTC and Allegan -1.2683 -1.89 -0.65 FCTC and Barry -1.2604 -1.89 -0.63 FCTC and Calhoun -1.4274 -1.86 -0.47 FCTC and Kalamazoo -1.1668 -2.03 -0.65 113 FCTC Calhoun County Kalamazoo County 20 04 20 02 20 00 Year 19 98 19 96 19 94 19 92 19 90 19 88 19 86 Number of Points Average Number of Antler Points 6.5 6 5.5 5 4.5 4 3.5 3 Allegan County Barry County Figure A.1. The temporal trend among the five areas of interest in regards to average number of points for bucks age = 1.5 years. Ft. Custer Barry County Kalamazoo County Year 20 04 20 02 20 00 19 98 19 96 19 94 19 92 19 90 19 88 24 23 22 21 20 19 18 17 19 86 Beam Diameter (mm) Average Beameter of 1.5 Year Old Males Allegan County Calhoun County Figure A.2. The temporal trend of average beam diameter among the five areas of interest for bucks age=1.5 years. 114 APPENDIX B Coordinates of the exclosure sites and paired control units and vegetation sampling data sheets 115 Table B.1. Locations and forest type of each exclosure and paired control site. The locations listed are based on the NAD 83 datum and projected in UTM 16T. Each site's forest type is listed as a mixed hardwood or as oak hardwood. 1 X coordinate Exclosure 1 mixed hardwoods 633991.39 4682991.95 Paired control 1 mixed hardwoods 634033.12 4682984.41 Exclosure 2 mixed hardwoods 634586.05 4684144.44 Paired control 2 mixed hardwoods 634615.08 4684106.86 Exclosure 3 mixed hardwoods 635548.73 4683292.98 Paired control 3 mixed hardwoods 635595.60 4683294.48 Exclosure 4 oak hardwoods 636316.63 4684553.25 Paired control 4 oak hardwoods 636362.63 4684530.91 Exclosure 5 mixed hardwoods 638245.82 4684257.31 Paired control 5 mixed hardwoods 638244.22 4684226.29 Exclosure 6 oak hardwoods 638512.32 4684852.50 Paired control 6 oak hardwoods 638545.65 4684889.51 Exclosure 7 oak hardwoods 639662.48 4683182.36 Paired control 7 oak hardwoods 639700.91 4683163.46 Exclosure 8 oak hardwoods 640625.81 4682668.86 Paired control 8 oak hardwoods 640610.66 4682696.57 1 The coordinates for the sites are for the northwest corners. 116 Y coordinate 1 Forest type Site Vertical Cover Percentage Forest type:____________________ Observers:________________________ Burn treatment: yes / no Site:__________________ Date:__________ Line intercept length: 18m Exclosure Height Strata* Strata Height Transect 1 Transect 2 Transect 3 Transect 4 Transect 5 mean (cm) mean % covered 0.0m-0.5m __________ __________ __________ __________ __________ __________ __________ 0.5m-2.0m __________ __________ __________ __________ __________ __________ __________ >2m ______ ______ ______ ______ ______ ______ ______ Paired Control Height Strata* Strata Height Transect 1 Transect 2 Transect 3 Transect 4 Transect 5 mean (cm) mean % covered 0.0m-0.5m __________ __________ __________ __________ __________ __________ __________ 0.5m-2.0m __________ __________ __________ __________ __________ __________ __________ >2m ______ ______ ______ ______ ______ ______ ______ *measured in centimeters Figure B.1. The vertical cover data sheet used during vegetation sampling. 117 Exclosure Species Composition Forest type:______________________ Observers:_______________________ Site type:________________________ Burn treatment: yes / no Date:______________ Site #:_____ Understory (0.0m-0.5m) plot size:1x2m Species: Plot A Plot B ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ ____________________________________________ _______ _______ Total number of species _______ _______ Figure B. 2. The understory stem density data sheet. 118 Plot C _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ _______ Exclosure Species Composition Forest type:______________________ Observers:_______________________ Site type:________________________ Burn treatment: yes / no Date:______________ Site #:_____ Midstory (0.5m-2m) plot size: 1x5m Species: Plot A Plot B Plot C ____________________________________________ _______ _______ _______ ____________________________________________ _______ _______ _______ ____________________________________________ _______ _______ _______ ____________________________________________ _______ _______ _______ ____________________________________________ _______ _______ _______ ____________________________________________ _______ _______ _______ ____________________________________________ _______ _______ _______ ____________________________________________ _______ _______ _______ ____________________________________________ _______ _______ _______ ____________________________________________ _______ _______ _______ ____________________________________________ _______ _______ _______ Total number of species _______ _______ _______ Overstory (>2m and DBH <4in) Species: Plot A plot size:20x20m ____________________________________________ _______ ____________________________________________ _______ ____________________________________________ _______ ____________________________________________ _______ ____________________________________________ _______ ____________________________________________ _______ ____________________________________________ _______ ____________________________________________ _______ ____________________________________________ _______ ____________________________________________ _______ Total number of species _______ Figure B.3. The stem density data sheets for the midstory and overstory. 119 APPENDIX C Home Range Estimates and Locations of Radio-Collared Deer 120 Figure C.1. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.004. Figure C.2. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.010. 121 Figure C.3. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.020. Figure C.4. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.029. 122 Figure C.5. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.039. Figure C.6. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.049. 123 Figure C.7. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.060. Figure C.8. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.069. 124 Figure C.9. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.090. Figure C.10. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.173. 125 Figure C.11. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.273. Figure C.12. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.280. 126 Figure C.13. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.290-1. Figure C.14. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.300. 127 Figure C.15. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.310. Figure C.16. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.315. 128 Figure C.17. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.330. Figure C.18. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.334. 129 Figure C.19. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.340. Figure C.20. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.353. 130 Figure C.21. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.370. Figure C.22. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.380. 131 Figure C.23. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.390. Figure C.24. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.400. 132 Figure C.25. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.420-2. Figure C.26. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.430. 133 Figure C.27. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.440. Figure C.28. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.470. 134 Figure C.29. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.480-1. Figure C.30. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.480-2. 135 Figure C.31. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.490. Figure C.32. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.519. 136 Figure C.33. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.530. Figure C.34. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.540. 137 Figure C.35. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.560. Figure C.36. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.580-1. 138 Figure C.37. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.580-2. Figure C.38. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.590-2. 139 Figure C.39. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.630. Figure C.40. The fixed kernel home range estimate with the 95% and 50% probability of occupancy for radio frequency 150.710. 140 LITERATURE CITED 141 LITERATURE CITED Aebischer, N. J., P. A. Robertson, and R. E. Kenward. 1993. Compositional analysis of habitat use from animal radio-tracking data. Ecology 74:13131325. Alverson, W. S., D. M. Waller, and S. I. Solheim. 1988. Forests too deer: edge effects in northern Wisconsin. Conservation Biology 2:348-358. Ballard, W. B., H. A. Whitlaw, S. J. Young, R. A. Jenkins, and G. J. Forbes. 1999. Predation and survival of white-tailed deer fawns in northcentral New Brunswick. Journal of Wildlife Management 63:574–579. Bart, J., and D. S. Robson. 1982. Estimating survivorship when subjects are visited periodically. Ecology 63:1078-1090. Bartush, W. S., and J. C. Lewis. 1978. Behavior of white-tailed does and fawns during the parturition period. Proc. Ann. Conf. S.E. Assoc. Fish and Wildl. Agencies 32:246-255. Beringer, J. L., L. P. Hansen, W. Wilding, J. Fischer, and S. Sheriff. 1996. Factors affecting capture myopathy in white-tailed deer. Journal of Wildlife Management 60:373-380. Bonham, C. D. 1989. Measurements for terrestrial vegetation. John Wiley and Sons, New York. USA. Brewer, L.R., H.A. Raup, and T.W. Holder. 1984. Presettlement vegetation of southwest Michigan (map). Western Michigan University, Department of Geology, Kalamazoo, Michigan. Canfield, R. H. 1941. Application of the line interception method in sampling range vegetation. Journal of Forestry 39:388-394. Chapman, K.A. 1984. An ecological investigation of native grassland in southern Lower Michigan. M.S. Thesis, Western Michigan University, Kalamazoo, Michigan, USA. Clover, M.R. 1954. A portable deer trap and catch-net. California Fish and Game 40:373-376. _____. 1956. Single-gate deer trap. California Fish and Game 42:199-201. Comer, P. J. 1995. Michigan‟s Native Landscape, as interpreted from the 142 general land office surveys 1816-1856. Report to the U.S. Environmental Protection Agency, Water Division, and the Michigan Department of Natural Resources, Wildlife Division. Michigan Natural Features Inventory, Lansing, Michigan, USA. 76pp. Cote, S.D., Rooney, T.P., Tremblay, J.P., Dussault, C., Waller, D.M., 2004. Ecological impacts of deer overabundance. Annual Review of Ecology, Evolution, and Systematics 35:113–147. deCalesta, D . S. 1992. Impact of deer on species diversity of Allegheny hardwood stands. Proc. Northeast. Weed Sci. Soc. 46:135. deCalesta, D. S., and S. L. Stout. 1997. Relative deer density and sustainability: a conceptual framework for integrating deer management with ecosystem management. Wildlife Society Bulletin 25:252-258. Downing, R.L., and B.S. McGinnes. 1969. Capturing and marking white-tailed deer fawns. Journal of Wildlife Management 33(3):711-714. Enck, J. W., D. J. Decker, S. J. Riley, J. F. Organ, L. H. Carpenter, and W. F. Siemer. 2006. Integrating ecological and human dimensions in adaptive management of wildlife-related impacts. Wildlife Society Bulletin 34:698705. Fryxell J. M., Hussel D.J., Lambert A. B. 1991. Time lags and population fluctuations in white-tailed deer. Journal of Wildlife Management 55:377–385. Harder, J. D., and Kirkpatrick, R.L. 1993. Physiological methods in wildlife research. Pages 275-306 in T.A. Bookhout, ed. Research and management techniques for wildlife and habitats. The Wildlife Society, Bethesda, M.D. Haugen, A.O., and D.W. Speake. 1958. Determining age of young fawn whitetailed deer. Journal of Wildlife Management 22(3):319-321. Haulton, S. M., W. F. Porter, and B. A. Rudolph. 2001. Evaluating 4 methods to capture white-tailed deer. Wildlife Society Bulletin 29:255-264. Healy, W. M. 1997. Influence of deer on the development of forest structure and composition in central Massachusetts. Pages 249-266 in W. J. McShea, H. B. Underwood, andJ. H. Rappole, eds. The science of overabundance: deer ecology and population management. Smithsonian Inst. Press, Washington, D.C. Heisey, D. M., and T. K. Fuller. 1985. Evaluation of survival and cause-specific 143 mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. Herman, K. D., L. A. Masters, M. R. Penskar, A. A. Reznick, G. S. Wilhelm, and W. W. Brodowicz. 1996. Floristic quality assessment with wetland categories and computer application programs for the state of Michigan. Michigan Department of Natural Resources, Wildlife Division, Natural heritage program. Lansing, Michigan, USA. 21pp. Higgins, K. F., K. J. Jenkins, G. K. Clambey, D. W. Uresk. D. E. Naugle, J. E. Norland, and W. T. Barker. 2005. Vegetation sampling and measurement. Pages 524-553 in C. E. Braun ed. Techniques for Wildlife Investigators and Management. The Wildlife Society. Bethasda, Maryland. USA. Hiller, T. L. 2007. Land-use patterns and population characteristics of whitetailed deer in an agro-forest ecosystem in South Central Michigan. Dissertation, Michigan State University, East Lansing, Michigan, USA. Holzenbein, S., and R. L. Marchinton. 1992. Emmigration and mortality in orphaned male white-tailed deer. Journal of Wildlife Management 56:147-153. Horsley, S.B., Stout, S.L., deCalesta, D. S., 2003. White-tailed deer impact on the vegetation dynamics of a northern hardwood forest. Ecological Applications 13:98–118. Huegel, C. N., R.B. Dahlgreen, and L.H. Gladfelter. 1985. Use of doe behavior to capture white-tailed deer fawns. Wildlife Society Bulletin 13:287-289. Hupp, J. W., and J. T. Ratti. 1983. A test of radio telemetry triangulation accuracy in heterogeneous environments. Pages 31-46 in D. G. Pincock, ed. Proceedings Fourth International Conference Wildlife Biotelemetry, Applied Microelectronics Institute and Technical University of Nova Scotia, Halifax. Johnson, D. H. 1979. Estimating nest success: the Mayfield method and an alternative. Auk 96:651-661. Kenoyer, L. 1930. Ecological notes on Kalamazoo County, Michigan, based on original land survey. Papers, Michigan Acad. Sc. Arts. Letters 11:211-217. Kernohan, J. K., R. A. Gitzen, J. J. Millspaugh. 2001. Analysis of animal space use and movements. Pages 125-166 in J.J. Millspaugh, and J.M. Marzluff, eds. Radio tracking and animal populations. Academic Press. San Diego, California, USA. 144 Kenward R. E., S. S. Walls, K. H. Hodder. 2001. Life path analysis: scaling indicates priming effects of social and habitat factors on dispersal distances. Journal of Animal Ecoloogy 70:1–13. Kenward, R. E., S.P. Rushton, C. M. Perrins, D. W. Macdonald, and A. B. South. 2002. From marking to modelling: dispersal study techniques for land vertebrates. Pages 50-71 in Bullock J.M., R. E. Kenward, R. S. Hails, eds. Dispersal ecology. Blackwell Publishing. Malden, MA, USA. Kittredge, D. B., and P. M. Ashton. 1995. Impact of deer browsing on regeneration in mixed stands in southern New England. Northern Journal of Applied Forestry 12:115-120. Lancia, R. A., C. E. Braun, M. W. Collopy, R. D. Dueser, J. G. Kie, C. J. Martinka, J. D. Nichols, T. D. Nudds, W. R. Porath, and N. G. Tilghman. 1996. ARM! for the future: adaptive resource management in the wildlife profession. Wildlife Society Bulletin 24:436–442. Legge, J. T., P. J. Higman, P. J. Comer, M. R. Penskar, and M. L. Rabe. 1995. A floristic and natural features inventory of Fort Custer Training Center, Augusta, Michigan. Michigan Natural Features Inventory, Lansing, Michigan, USA. 151pp. Lenth. R. V. 1981. On finding the source of a signal. Technometrics 23:149-154. Lischka, S. A., S. J. Riley, and B. A. Rudolph. 2008. Effects of impact perception on acceptance capacity for white-tailed deer. Journal of Wildlife Management 72:502-509. Lund, R. C. 1975. The capture and marking of wild, newborn white-tailed deer fawns. Transactions of the Northeast Fish and Wildlife Conference 32:25-33. Marchinton, R.L. and D.H. Hirth. 1984. Behavior. Pages 129-168 in L.K. Halls, ed. White-tailed deer ecology and management. Wildlife Manage. Inst. stackpole Books. Harrisburg, Pennsylvania. Marquis, D. A., R. L. Ernst, and S. L. Stout. 1992. Prescribing silvicultural treatments in hardwood stands of the Alleghenies (revised). U. S. Dep. Agric. For. Serv. Northeast. Exp. Stn. Gen. Tech. Rep. NE-96 101pp. May R. M. 1981. Models for single populations. Pages 5-30 in May R. M. ed. Theoretical ecology: principles and applications. Sinauer, Sunderland, Mass. USA. 145 Mayfield, H. 1961. Nesting success calculated from exposure. Wilson Bulletin 73:255– 261. Mayfield, H. 1975. Suggestions for calculating nest success. Wilson Bulletin 87:456–466. McCullough, D.R. 1975. Modification of the clover deer trap. California Fish and Game 61(4):242-244. McCullough, D. R. 1979. The George Reserve deer herd: population ecology of a K-selected species. University of Michigan Press, Ann Arbor. 271pp. McCullough, D. R. 1982. Antler characteristics of George Reserve white-tailed deer. Journal Wildlife Management. 46:821-826. McCullough, D. R. 1984. Lessons from the George Reserve, Michigan. Pages 211-242 in L. K. Halls, ed. White-tailed deer ecology and management. Stackpole Books, Harrisburg, Pa. Messier F., J. Huot, and D. L. Henaff. 1988. Demography of the George River caribou herd: evidence of population regulation by forage exploitation and range expansion. Arctic 41:279–287. Millspaugh, J. J., J. R. Skalski, R.L. Towsend, D.R. Diefenbach, M.S. Boyce, L. P. Hansen, and K. Kammermeyer. 2009. An evaluation of sex-age-kill (SAK) model performance. Journal of Wildlife Management 73:442-451. Moen, A. N., and C. W. Severinghaus. 1981. The annual weight cycle and survival of white-tailed deer in New York. New York Fish and Game Journal 28(2):162-177. Mueller-Dombois, D., and H. Ellenberg. 1974. Aims and methods of vegetation ecology. John Wiley and Sons, New York. USA. Nams, V. O. 1989. Effects of radio-telemetry error on sample size and bias when testing for habitat selection. Canadian Journal of Zoology 67:1631–1636. Nams, V. O., and S. Boutin. 1991. What is wrong with error polygons? Journal of Wildlife Management 55:172–176. Nelson, M. E., and L. D. Mech. 1992. Dispersal in female white-tailed deer. Journal of Mammalogy 73:891-894. Nelson, M. E. 1993. Natal dispersal and gene flow in white-tailed deer in northeastern Minnesota. Journal of Mammalogy 74:316-322. 146 Nelson, M. E. 1998. Development of migratory behavior in northern whitetailed deer. Canadian Journal of Zoology 76:426-432. Ozoga, J. J., and L. J. Verme. 1985. Comparitive breeding behavior and performance of yearling vs. prime-age white-tailed bucks. Journal of Wildlife Management 49:364-372. Ozoga, J. J., and L. J. Verme. 1986. Relation of maternal age to fawnrearing success in white-tailed deer. Journal of Wildlife Management 50:480-486. Palmer, W. L., G. L. Storm, W. M. Tzilkowski, and M. J. Lovallo. 1997. Demographic profiles of deer under different management strategies in Pennsylvania. Pages 151-163 in W. J. McShea, H. B. Underwood, and J. H. Rappole, eds. The science of overabundance: deer ecology and population management. Smithsonian Inst. Press, Washington, D.C. Patterson, B. R., and Messier, F. 2000. Factors influencing killing rates of whitetailed deer by coyotes in Eastern Canada. Journal of Wildlife Management 64:721-732. Patterson, B. R., and V. A. Power. 2002. Contributions of forage competition, harvest, and climate fluctuation to changes in population growth of northern white-tailed deer. Oecologia 130:62-71. Pollack, K.H., S.R. Winterstein, C.M. Bunck, and P.D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal Wildlife Management 53(1):7-15. Potvin, F., P. Beaupre, and G. Laprise. 2003. The eradication of balsam fir stands by white-tailed deer on Anticosti Island, Quebec: a 150-year process. Ecoscience 10:487-495. Purdue, J. R., M. H. Smith, and J. C. Patton. 2000. Female philopatry extreme spatial genetic heterogeneity in white-tailed deer. Journal of Mammalogy 81:179-185. Pusateri, J. S. 2003. White-tailed deer population characteristics and landscape use patterns in southwestern Lower Michigan. Thesis, Michigan State University, East Lansing, Michigan. USA Pusateri Burroughs, J., H. Campa III, S. R. Winterstein, B. A. Rudolph and W. E. Moritz. 2006. Cause-specific mortality and survival of white-tailed deer fawns in southwestern Lower Michigan. Journal of Wildlife Management 70:743-751. 147 Putnam, R. J., P. J. Edwards, J. C. E. Mann, R. C. How, and S. D. Hill. 1989. Vegetational and faunal changes in an area of heavily grazed woodland following relief of grazing. Biology and Conservation 47:13-32. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:18- 190. Rasmussen, G.P. 1985. Antler measurements as an index to physical condition and range quality with respect to white-tailed deer. New York Fish Game Journal. 32:97-113. Riley, S. J., D. J. Decker, L. H. Carpenter, J. F. Organ, W. F. Siemer, G. F. Mattfeld, and G. Parsons. 2002. The essence of wildlife management. Wildlife Society Bulletin 30:585-593. Riney, T. 1955. Evaluating condition of free-ranging red deer (Cervus elaphus), with special reference to New Zealand. N.Z. J. Sci. Technol. Sect. B 36:429-463. Rossenburry, C. S., M. C. Conner, and R. A. Lancia. 2001. Behavior and dispersal of white-tailed deer during the breeding season. Canadian Journal of Zoology 79:171-174. Saltz, D., and P. U. Alkon. 1985. A simple computer-aided method for estimating radio-location error. Journal of Wildlife Management 58:181-184. Schmidt, K. T., A. Stien, S. D. Albon, and F. E. Guinness. 2001. Antler length of yearling red deer is determined by population density, weather and early life history. Oecologia 127:191-197. Seaman, D.E., and R.A. Powell. 1996. An evaluation of the accuracy of kernel density estimators for home range analysis. Ecology 77:2075-2085. _____, J.J. Millspaugh, B.J. Kernohan, G.C. Brundige, K.J. Raedeke, and R.A. Gitzen. 1999. Effects of sample size on kernel home range estimates. Journal of Wildlife Management 63(2):739-747. Severinghaus, C. W. 1949. Tooth development and wear as criteria of age in white-tailed deer. Journal of Wildlife Management 13:195-216. _____, H.F. Maguire, R.A. Cookingham, and J.E. Tanick. 1950. Variations by age class in the antler beam diameters of white-tailed deer related range conditions. Trans. North Am. Wildl. Conf. 15:551-570. _____, and A.N. Moen. 1983. Prediction of weight and reproductive rates of a 148 white-tailed deer (Odocoileus-virginianus) population from records of antler beam diameter among yearling males. New York Fish and Game Journal 30:30-35, Skalski, J. R., and J. J. Millspaugh. 2002. Generic variance expressions, precision, and sampling optimization for the sex-age-kill model of population reconstruction. Journal of Wildlife Management 66:1308-1316. Stoymayer, K. A. K., and Warren, R. J. 1997. Are overabundant deer herds in the eastern United States creating alternate stable states in forest plant communities?. Wildlife Society Bulletin 25(2):227-234. Stoynoff, N.A. 1983. Whitman Lake Wetland: A floristic and phytogeographic analysis. Thesis, Michigan State University, East Lansing, Michigan, USA. Strole, T.A. 1988. Influences of white-tailed deer (Odocoileus virginianus) browsing on successional patterns and woody species composition in central Illinois upland forests. Thesis, Illinois State University, Normal, Illinois, USA. _____, and R. C. Anderson. 1992. White-tailed deer browsing: species preferences and implications for central Illinois forests. Natural Areas Journal 12(3):139-144. Swihart, R. K., and N. A. Slade. 1997. On testing for independence of animal movements. Journal of Wildlife Management 39:118-123. Tilghman, N. G. 1989. Impacts of white-tailed deer on forest regeneration in northwestern Pennsylvania. Journal of Wildlife Management 53(3):524532. Van Deelen, T. R., and D. R. Etter. 2003. Effort and the functional response of deer hunters. Human Dimensions of Wildlife 8: 97-108. Waller, D. M., and W. S. Alverson. 1997. The white-tailed deer: a keystone herbivore. Wildlife Society Bulletin 25(2):217-226. White, G. C., and R. A. Garrot. 1990. Analysis of wildlife radio-tracking data. Academic Press, San Diego, California, USA. White, H. A., F. F. Knowlton, and W. C. Glazener. 1972. Effects of damnewborn fawn behavior on capture and mortality. Journal of Wildlife Management 36(6):897-906. Winterstein, S.R., K.H. Pollock, and C.M. Bunck. 2001. Analysis of survival data 149 from radiotelemetry studies. Pages 351-380 in J.J. Millspaugh, and J.M. Marzluff, eds. Radio tracking and animal populations. Academic Press. San Diego, California, USA. Worton, B. J. 1989. Kernel methods for estimating the utilization distribution in home-range studies. Ecology 70:164–168. Worton, B. J., 1995. Using Monte Carlo simulations to evaluate kernel-based home range estimators. Journal of Wildlife Management 59:794-800. 150