INFLUENCE OF PREDATION RISK ON THE ECOLOGY OF MALE EASTERN WILD TURKEYS ( MELEAGRIS GALLOPAVO SILVESTRIS ) by PATRICK H E NRY WIGHTMAN (Under the Direction of Michael J. Chamberlain ) ABSTRACT Restoration of wild turkeys across North America is one of the most successful conservation efforts, after the species was nearly extirpat ed in the early 1900’s However, productivity and abundance of wild turkeys h a ve declined throughout broad areas of the species range, which is occurring simultaneously as populations of natural predators are increasing . Male wild turkeys are not only susceptible to natural predation , but are a highly sought after game bird Therefore, understanding influences of predation risk on behavior of male wild turkey s can provide valuable insight into turkey ecology and management. Male wild turkeys gobble to attract attention from fe males , but gobbling also increases predation risk. Notably, gobbling activity is the primary determinant of hunter satisfaction, so it is of interest to management agencies I develo ped an efficient method known as a convolution neural network to detect gobb les from ambient sound recordings , reducing time and money associated with collecting and processing gobbling data. Using gobbling and environmenta l data coupled with location data collected on turkeys, coyotes, and hunters, I investigated how predation risk influenced gobbling activity, movement s , and resource selection of male wild turkeys. I found that weather conditions influenced gobbling activi ty , likely as a result of factors related to predation risk , and noted that harvest and reproduction were the primary drivers of gobbling. I found that gobbling activity varied across the landscape and declined with the onset and progression of hunting. I found no relationship between coyote predicted use and gobbling activity , but did f ind evidence that gobbling activity increased in areas as sociated with reduced hunter activity I determined that although coyote s and turkeys maintained overlapping ranges , fine scale spatial overlap and known contacts were rare. Furthermore, I found known contacts between coyotes and male turkeys had no influence on male movement s, and that such contacts occurred closer to forest edges. I observed that contacts between hunt ers and male turkeys were associated with males increasing movements, and that such contacts occurred closer to secondary roads . These findings suggest that risk of encountering hunters is more impactful to movement ecology of male wild turkeys than encounters with natural predators. INDEX WORDS: acoustic monitoring, convolutional neural network, coyote, gobbling, hunting, pred ation, resource selection , wild turkey INFLUENCE OF PREDATION RISK ON THE ECOLOGY OF MALE EASTERN WILD TURKEYS ( MELEAGRIS GALLOPAVO SILVESTRIS ) by PATRICK HENRY WIGHTMAN B.S., Paul Smiths College, 2014 M.S., Louisiana State University, 2017 A Dissertation Submitted to the Graduate Faculty of The University of Georgia in Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSPHY ATHENS, GEORGIA 2022 © 2022 Patrick Henry Wightman All Rights Reserved INFLUENCE OF PREDATION RISK ON THE ECOLOGY OF MALE EASTERN WILD TURKEYS ( MELEAGRIS GALLOPAVO SILVESTRIS ) by PATIRCK HENRY WIGHTMAN Major Professor: Michael J. Chamberlain Committee: James A. Martin Michel T. Kohl Bret A . Collier Electronic Version Approved: Ron Walcott Vice Provost for Graduate Education and Dean of the Graduate School The University of Georgia May 2022 DEDICATION This dissertation is dedicated to my wife, Patricia Wightman, I am so grateful for her support, patience, advice, and love throughout this journey. Also, my parents Dean and Kelly. Lastly, my grandfather Floyd Farley, who will not get the opportunity to see me finish this degree like I hoped he would. I hope I make you all proud iv ACKNOWLEDGEMENTS First and foremost, I would like to thank Dr. Michael Chamberlain for his support, mentorship, patience, and guidance during this research. I will forever be thankful that you convinced me to take this opportunity to research and learn here at Warnell, the experiences and knowledge gained have been instrumental for my development as a w ildlife professional. You always had an open door policy that I took advantage of during my time here, whether it was hour long conversations about turkey research, hunting, or just life ; those conversations are memories that I cherish and won’t soon forge t. Your ability to speak and write eloquently about complex topics in a way that everyone can understand is something I will continue to work on and strive to do throughout the remainder of my career. A big thanks to my graduate committee members Dr. James Martin, Dr. Bret Collier, and Dr. Michel Kohl I was fortunate to have Dr. James Martin on my committee and in the office next to mine. This allowed me to ask questions all the time which he graciously answered. He provided me with numerous statistical an d conceptual support during my time here and has taught me skills that I will be able to utilize throughout the remainder of my career. He also brightened office work with his music/singing, jokes, and conversati ons which was much appreciated. Dr. Bret Col lier, was instrumental in preparing me for a Ph.D. during my time at LSU and pushing me to pursue this degree, for that I will always be thankful. Dr s . Collier and Michel Kohl provided valuable insight into development and statistical methods/interpretatio n of this research I will fo rever be grateful for all of your support and mentorship. v I also need to thank the small army of undergraduate students who helped process gobbling data, all field technicians, fellow researchers, and graduate students withou t whose collaboration this dissertation would not have been possible. Specifically, I would like to thank my lab mates and friends that have helped me during my time at Warnell, including Nick Bakner, Calvin Wakefield , Page Goodman, Nick Gulotta, Erin Ulre y, Sara Watkins, Ashley Lohr , Dr. Joey Hinton, Dr. Brad Cohen and Dr. Daniel Sullivan. I also thank the Georgia Department of Natural Resources - Wildlife Resources Division , South Carolina Department of Natural Resources, the Louisiana State University Ag ricultural Center, the United States Department of Energy – Savannah River, the USDA Forest Service, National Wild Turkey Federation, and the Warnell School of Forestry and Natural Resources at the University of Georgia for funding and logistical support. vi TABLE OF CONTENTS Page ACKNOWLEDGEMENTS ................................ ................................ ................................ ........... iv LIST OF TABLES ................................ ................................ ................................ ......................... ix LIST OF FIGURES ................................ ................................ ................................ ...................... xii CHAPTER 1 INRODUCTION AND LITERATURE REVIEW ................................ ........................ 1 LITERATURE CITED ................................ ................................ ............................ 8 2 COMPARISON OF METHODS FOR AUTOMATED IDENTIFICATION OF WILD TURKEY GOBBLES ................................ ................................ ....................... 16 ABSTRACT ................................ ................................ ................................ ........... 17 INTRODUCTION ................................ ................................ ................................ 18 STUDY AREA ................................ ................................ ................................ ...... 20 METHODS ................................ ................................ ................................ ............ 21 RESULTS ................................ ................................ ................................ .............. 24 DISCUSSION ................................ ................................ ................................ ........ 25 ACKNOWLEDGMENTS ................................ ................................ ..................... 28 LITERATURE CIT ED ................................ ................................ .......................... 29 3 INFLUENCE OF WEATHER ON GOBBLING ACTIVITY OF MALE WILD TURKEYS ................................ ................................ ................................ ................... 38 ABSTRACT ................................ ................................ ................................ ........... 39 vii INTRODUCTION ................................ ................................ ................................ 40 STUDY AREA ................................ ................................ ................................ ...... 42 METHODS ................................ ................................ ................................ ............ 44 RESULTS ................................ ................................ ................................ .............. 47 DISCUSSION ................................ ................................ ................................ ........ 48 ACKNOWLEDGMENTS ................................ ................................ ..................... 51 LITERATURE CI TED ................................ ................................ .......................... 53 4 LANDSCAPE CHARACTERISTICS AND PREDATION RISK INFLUENCE GOBBLING ACTIVITY OF MALE WILD TURKEYS ................................ ............ 70 ABSTRACT ................................ ................................ ................................ ........... 71 INTRODUCTION ................................ ................................ ................................ 72 STUDY AREA ................................ ................................ ................................ ...... 75 METHODS ................................ ................................ ................................ ............ 76 RESULTS ................................ ................................ ................................ .............. 84 DISCUSSI ON ................................ ................................ ................................ ........ 89 MANAGEMENT IMPLICATIONS ................................ ................................ ..... 93 ACKNOWLEDGMENTS ................................ ................................ ..................... 93 LITERATURE CITED ................................ ................................ .......................... 94 APPENDICES ................................ ................................ ................................ ..... 121 5 PROACTIVE AND REACTIVE SPATIOTEMPORAL RESPONSES OF MALE WILD TURKEYS TO RISK OF PREDATION ................................ ................. 136 ABSTRACT ................................ ................................ ................................ ......... 137 INTRODUCTION ................................ ................................ ............................... 138 viii STUDY AREA ................................ ................................ ................................ .... 141 MET HODS ................................ ................................ ................................ .......... 143 RESULTS ................................ ................................ ................................ ............ 153 DISCUSSION ................................ ................................ ................................ ...... 157 ACKNOWLEDGMENT S ................................ ................................ ................... 161 LITERATURE CITED ................................ ................................ ........................ 163 APPENDICES ................................ ................................ ................................ ..... 191 6 CONCLUSIONS ................................ ................................ ................................ ........ 214 ix LIST OF TABLES Page Table 2. 1: Detections, gobbles, and precision(gobbles/detections) of wild turkeys calculated from Convolutional Neural Network (CNN) analysis and Raven Pro automated call recognition software collected from autonomous recording units (ARUs) on the Webb Wildlife Management Area Complex in South Carolina (Webb), USA, during 2015 - 2018 and on B.F. Gra nt (B.F. Grant) and Cedar Creek (CC) Wildlife Management Areas in Georgia, USA, during 2017 – 2018. ................................ ................................ ................................ ... 33 Table 3.1 : Detections, gobbles, and gobbles per autonomous recording unit (ARU) for the Webb Wildlife Management Area Complex (Webb), Savannah River Site ( SRS) and Crackerneck Wildlife Management Area (CWMA) in South Carolina and Cedar Creek Wildlife Management Area (CCWMA) and B.F. Grant Wildlife Management Area (BFG) in Georgia from 2014 through 2018. ................................ ................................ ...... 58 Table 3 .2 : Parameters and associated means, standard deviations (sd), and credible intervals from a state - space model evaluating the relationship between daily gobbling activity by male wild turkeys and weather variables for the Webb Wildlife Management Area Complex (Webb), Savannah River Site (SRS) and Cra ckerneck Wildlife Management Area (CWMA) in South Carolina and Cedar Creek Wildlife Management Area (CCWMA) and B.F. Grant Wildlife Management Area (BFG) in Georgia from 2014 through 2018. ....... 59 Table 4 .1 : List of Bayesian regression models and their paramete rs used to estimate the effects on gobbling activity of Eastern wild turkey ( Meleagris gallopavo silvestris ) on Cedar x Creek and B.F. Grant wildlife management areas in Georgia, USA, during March - June 2018 - 2019. Each model includes an interactive term for stage (before, during, and after hunting) and a random intercept term to account for variation across autonomous recording units. Models were compared by marginal and conditional R 2 values and ranked using expected log pointwise predictive density (ELPD) a nd leave - one - out - information - criterion (LOOIC). ................................ ................................ ....................... 105 Table 5.1 : List of Bayesian regression models and their parameter estimates used to estimate the effect of male turkeys interacting with coyotes on Cedar Creek and B.F. Grant wildlife management areas in Georgia, USA during 2018. Shown are parameter estimates (β), standard deviation (SD), 95% credible intervals (CI), and R hat values ........................ 169 Table 5.2: List of Bayesian regression models and their parameter estimates used to estimate the effect of male turkeys interacting with hunters on B.F Grant and Cedar Creek Wildlife Management Areas (WMAs) in Georgia, USA, 2018, Webb WMA in South Carolina, USA, 2014 – 2018, and Tunica Hills WMA in Louisiana, USA, 2012 – 2013 Shown are parameter estimates (β), standard de viation (SD), 95% credible intervals (CI), and R hat values ................................ ................................ ................................ ............................. 170 Table 5.3: Estimated effects of the distance to hardwoods, pine, open, water, shrub, mix, forest edge, primary roads, secondary roads, public access, private property, and percent NDVI on interaction occurrence of male wild turkey and coyotes on B.F. Grant and Cedar Creek Wildlife Management Area in Georgia, USA 2018. Shown are coefficient estimates (β), standard deviation (SD), 95% credible intervals (CI), and R hat values. ........................ 171 Table 5.4: Number of male turkeys and hunters monitored simultaneously, contacts between hunters and male wild turkeys, and the total number male wild turkeys that came in xi contact with a hunter on B.F Grant and Cedar Creek Wildlife Management Areas (WMAs) in Georgia , USA, 2018, Webb WMA in South Carolina, USA, 2014 – 2018, and Tunica Hills WMA in Louisiana, USA, 2012 – 2013 ................................ .............. 171 Table 5.5: Estimated effects of the distance to hardwoods, pine, open, water, shrub, mix, forest edge, primary roads, secondary roads, publi c access, private property, and percent NDVI on contact occurrence of male wild turkey and hunters on B.F Grant and Cedar Creek Wildlife Management Areas (WMAs) in Georgia, USA, 2018, Webb WMA in South Carolina, USA, 2014 – 2018, and Tunica Hills WMA in Lo uisiana, USA, 2012 – 2013 Shown are coefficient estimates (β), standard deviation (SD), 95% credible intervals (CI), and R hat values. ................................ ................................ ................................ .............. 172 xii LIST OF FIGURES Page Figure 2. 1: Location of Webb Wildlife Management Area Complex in South Carolina, USA, and B .F. Grant and Cedar Creek Wildlife Management Areas in Georgia, USA ................... 34 Figure 2. 2: Precision ([True Gobbles]/[True Gobbles + False Gobbles]) for identification of gobbles by wild turkeys calculated from results of Convolutional Neural Network analysis (red) and Raven Pro Automated Call Recognition software (black) of all units on the Webb Wildlife Management Area Complex in South Carolina, USA, during 2015 – 2018. ................................ ................................ ................................ ................................ 35 Figure 3 .1 : Location of Webb Wildlife Management Area Complex, Crackerneck Wildlife Ma nagement Area, the Savannah River Site in South Carolina, USA, and B.F. Grant and Cedar Creek Wildlife Management Areas in Georgia, USA. ................................ ............ 61 Figure 3.2 : Predicted daily gobbling acti vity from state space model (dotted line) with 95% credible intervals (shad ed grey) compared to observed daily gobbling activity (black line) on the Savana River Site in South Carolina, USA, 2014. ................................ .................. 62 Figure 3.3 : The expected number of gobbles ( r expected(t) *20 ) with 95% credible intervals as a function of c hange in barometric pressure (mb ) across all 5 sites in South Carolina and Georgia, USA, 2014 - 2018. "Click here and type figure title.]" ................................ ......... 63 Figure 3. 4 : The expected number of gobbles ( r expected(t) *20 ) with 95% credible intervals as a func tion of temperature ( ° C) across all 5 sites in South Carolina and Georgia, USA, 2014 - 2018. ................................ ................................ ................................ ......................... 64 xiii Figure 3. 5 : The expected number of gobbles ( r expected(t) *20 ) with 95% credible intervals as a function of wind speed ( km/hr ) across all 5 sites in South Carolina and Georgia, USA, 2014 - 2018. ................................ ................................ ................................ ......................... 65 Figure 4.1: Location and la nd cover features of B.F. Grant and Cedar Creek Wildlife Management Area s , and surrounding property in Georgia, USA , 2018 - 2019 ............... 107 Figure 4.2: Parameter estimates and confidence in tervals of full model for second - order resource selection functions for 111 GPS - marked eastern wild turkeys (81 F, 30 M) in Georgia, USA, calculated from daytime locations before (Pre - h unt), during (Hunt), and after (Post - h unt) spring wild turkey hunting season 2018 and 2019 ................................ ................ 108 Figure 4.3: Parameter estimates and confidence intervals of full model for third - order resource selection functions for 111 GPS - marked eastern wild turkeys (81 F, 30 M) in Georgia, USA, calculated from daytime locations before (P re - hunt), during (Hunt), and after (Post - hunt) spring wild turk ey hunting season 2018 and 2019 ................................ ................ 109 Figure 4.4: Parameter estimates of full model for second - and third - order resource selection functions for 36 GPS - collared coyotes in Georgia, USA, calculat ed from daytime locations before (Pre - hunt), during (Hunt), and after (Post - hunt) spring wild turkey hunting season 2018 - 201 9. ................................ ................................ ........................... 110 Figure 4.5: Probability of direction and magnitudes of conditional effects for each predictor variable (parameters) from B ayesian regression models estimating effects of predation risk on gobbling activity of male eastern wild turkeys on Cedar Creek and B.F. Grant wildlife management areas in Georgia, USA, during March - June 2018 - 2019. Each model includes an interactive term f or stage [before (Pre - hunt), during (Hunt), and after (Post - xiv hunt) hunting] and a random intercept term to account for variation ac ross autonomous recording units ................................ ................................ ................................ ................ 111 Figure 4.6: Probability of direction and magnitudes of conditional effects for each predictor variable (parameters) from Bayesian regression models estimating effects of 2 nd and 3 rd order probability of use of male and female eastern wild turkeys on gobbling activity on Cedar Creek and B.F. Grant wildlife management areas in Georgia, USA, during March - June 2018 - 2019. Each model includes an interactive term for stage [before (Pre - hunt), during (Hunt), and after (Post - hunt) hunting] and a random intercept term to account for variation across autonomous recording units ................................ ................................ .. 112 Figure 4.7: Probability of direction and magnitudes of conditional effects for each predictor variable (parameters) from Bayesian regression models estimating effects of land cover on gobbling activity of male eastern wild turkeys on Cedar Creek and B.F. Grant wildlife management areas in Georgia, USA, during March - June 2018 - 2019. Each model includes and interactive term for stage [before (Pre - hunt), during (Hunt), and after (Post - hunt) hunting] and a ra ndom intercept term to account for variation across autonomous recording units. ................................ ................................ ................................ ............... 113 Figure 4.8: The predicted effect with 95% credible intervals of distance to public access on average daily gobbling activity of male eastern wild turkeys during spring hun ting season on B. F. Grant and Cedar Creek Wildlife Management Areas in Georgia, USA, 2018 – 2019 ................................ ................................ ................................ .............................. 114 Figure 4.9: The predicted effect with 95% credible intervals of distance to private property on average daily gobbling activity of male eastern wild tur keys during spring hunting season xv on B. F. Grant and Cedar Creek Wildlife Management Areas in Georgia, USA, 2018 – 2 019 ................................ ................................ ................................ .............................. 115 Figure 4.10: The predicted effect with 95% credible intervals of within home range probability of use for female wild turkeys on a verage daily gobbling activity of male eastern wild turkeys after hunting season on B. F. Grant and Cedar Creek Wildlife Management A reas in Georgia, USA, 2018 – 2019 ................................ ................................ .......................... 116 Figure 5.1: Location of Webb Wildlife Management Area (WMA) Complex in South Carolina, B.F. Grant and Cedar Creek WMAs in Georgia, and Tunica Hills WMA in Louisiana, USA ................................ ................................ ................................ .............................. 173 Figure 5. 2 : Number of male wild turkey and coyote contact s within 100m, in relationship to hour of day (EST) on B.F Grant and Cedar Creek Wildlife Management Areas in Georgia, USA, 2018 ................................ ................................ ................................ .................... 174 Figure 5. 3 : Number of male wild turkey and coyote contact s within 100m, in relationship to day of year on B.F Grant and Cedar Creek Wildlife Management Areas in Georgia, USA, 2018 ................................ ................................ ................................ .............................. 174 Figure 5. 4 : Number of male wild turkey and hunter contacts within 100m in relationship to hour of day (EST) on B.F Grant and Cedar Creek Wildlife Management Areas (WMAs) in Georgia during 2018 , Webb WMA in South Carolina during 2014 – 2018, and Tunica Hills WMA in Louisiana, U SA, during 2012 – 2013 ................................ .................. 175 Figure 5.5: Number of male wild turkey and hunter contacts within 100m in relationship to da te on B.F Grant and Cedar Creek Wildlife Management Areas (WMAs) in Georgia during 2018 , Webb WMA in South Carolina during 2014 – 2018, and Tunica Hills WMA in Louisiana, USA, during 2012 – 2013 ................................ ................................ .......... 175 xvi Figure 5.6: Predicted probability of contact within 100m and 95% credible intervals for coyotes and male wild turkeys relative to distance to forest edge on B. F. Grant and Cedar Cree k Wildlife Management t Areas in Georgia, USA, 2018 – 2019 ................................ ....... 176 Figure 5.7: Predicted probability of contact within 100m and 95% credible intervals for hunters and male wild turkeys relative to distance to secondary roads on B.F Grant and Cedar Creek Wild life Management Areas (WMAs) in Georgia during 2018 , Webb WMA in South Carolina during 2014 – 2018, and Tunica Hills WMA in Louisiana, USA, during 2012 – 2013 ................................ ................................ ................................ .................. 177 Figure 5.8: Parameter estimates of full model for second - and third - order resource selection functions for 36 GPS - collared coyotes in Georgia, USA, calculated from daytime locations before (Pre - hunt), during (Hunt), and after (Post - hunt) spring wild tu rkey hunting season 2018 - 2019 ................................ ................................ ......................... 178 Figure 5.9: Parameter estimates of full model f or second - order resource selection functions for 30 GPS - tagged male wild turkey in Georgia, USA, before (Pre - hunt), during (Hunt), and after (Post - hunt) spring wild turkey hunting season 2018 - 2019 ................................ 179 Figure 5.10: Parameter estimates of full model fo r step selection functions for 30 GPS - tagged male wild turkeys in Georgia, USA, before (Pre - hunt), during (Hunt), and after (Post - hunt) spring wild turkey hunting season 2018 - 2019 ................................ .................. 180 1 CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW Predators influence prey via direct (lethal) or indirect (nonlethal) effects by invoking anti predator behaviors that impact prey survival and reproduction (Sih 1994 , Caro 2005 , Schmitz 2005, Preisser et al. 2007 ) . When n ot directly encountering predators, the perceived risk of predation can influence behavioral strategies in prey species (Laundre et al. 2001, Creel et al. 2008, Laundre at al. 2010, Embar et al 2014 ). Altering foraging, resource selection, movement, and br eeding behaviors to avoid predators can negatively affect prey survival and reproduction ( Lima 1998 , Preisser et al. 200 7 , Zanette 2011, Coleman et al. 2014, Hua 2014). Relative influences of predation risk on resource selection, movement , breeding behavi ors, survival, and reproduction can be more pronounced in species who use ostentatious behaviors for mate acquisition. Mate acquisition strategies vary markedly across mating systems, with strategies typically being more conspicuous in polygamous than mono gamous species (Kirkpatrick 1982). In polygamous mating systems, secondary sexual characteristics and behaviors, such as signaling and displays by males, portray fitness and social dominance, and are used as cues to attract females and secure mating opport unities (Emlen and Oring 1977, Anderson and Simmons 2006). Mate attraction is directly linked to the preference and likelihood of successfully reproducing, but reproductive behaviors also lead to an increased exposure to predation risk ( Hooper and Miller 2 008 , Edward 2015). Mate attraction via signaling manifests in a variety of behaviors such as vocalizations and displays. Risk reward theory predicts that