REPRODUCTIVE ECOLOGY OF EASTERN WILD TURKEY HENS IN SUSSEX COUNTY DELAWARE. Eric L. Ludwig

REPRODUCTIVE ECOLOGY OF EASTERN WILD TURKEY HENS IN SUSSEX COUNTY DELAWARE by Eric L. Ludwig A thesis submitted to the Faculty of the University of Delaware in partial fulfillment of the requirements for the degree of Master of Science in Wildlife Ecology Spring 2012 2012 Eric Ludwig All rights Reserved

REPRODUCTIVE ECOLOGY OF EASTERN WILD TURKEY HENS IN SUSSEX COUNTY DELAWARE by Eric L. Ludwig Approved: Jacob L. Bowman, Ph.D. Professor in charge of thesis on behalf of the Advisory Committee Approved: Douglas W. Tallamy, Ph.D. Chair of the Department of Entomology and Wildlife Ecology Approved: Robin W. Morgan, Ph.D. Dean of the College of Agriculture and Natural Resources Approved: Charles G. Riordan Ph.D Vice Provost for Graduate and Professional Education

ACKNOWLEDGEMENTS I would like to thank my funding sources Delaware Division of Fish and Wildlife, McIntire Stennis research funds, the University of Delaware, and the National Wild Turkey Federation for their support. I thank my committee, Bob Eriksen, Matt Dibona, Greg Shriver, and Jacob Bowman for all their time and willingness to assist on technical questions throughout the duration of the project. I would also like to thank John Mackenzie and Ben Mearns for their help with GIS problems that arose during the project. A special thanks goes to the staff of Redden State Forest and to all of the cooperating landowners. I greatly thank everyone who worked on the project including J. Ashling, J. Baird, C. Corddry, S. Dougherty, A. Dunbar, K. Duren, N. Hengst, D. Kalb, H. Kline, M. Miller, J. Rogerson, M. Springer, E. Tymkiw. I would especially like to thank C. Rhoads for his willingness to trap with me on New Year s Day. iii

TABLE OF CONTENTS LIST OF TABLES.vi LIST OF FIGURES...vii ABSTRACT.viii Chapter 1 INTRODUCTION...1 2 HEN SURVIVAL...3 Abstract...3 Introduction...3 Study Area...6 Methods...7 Results...10 Discussion...10 Management Implications...12 3 NESTING ECOLOGY OF EASTERN WILD TURKEYS IN AN AGRICULTURAL LANDSCAPE...16 Abstract...16 Introduction...17 Study Area...18 Methods...20 Results...24 Discussion...24 Management Implications...27 4 ROOST SITE SELECTION BASED ON LANDSCAPE CHARACTERISTICS IN AN AGRICULTURAL LANDSCAPE...36 Abstract...36 Introduction...36 Study Area...37 Methods...39 Results...41 Discussion...41 Management Implications...42 5 POULT SURVIVAL IN SUSSEX COUNTY DELAWARE...45 Abstract...45 iv

Introduction...45 Study Area...47 Methods...48 Results...50 Discussion...50 Management Implications...51 6 ANNUAL AND SEASONAL HOME RANGE SIZE OF WILD TURKEY HENS IN DELAWARE...52 Abstract...52 Introduction...52 Study Area...53 Methods...54 Results...56 Discussion...56 Management Implications...57 7 MANAGEMENT IMPLICATIONS...59 8 LITERATURE CITED...61 v

LIST OF TABLES Table 1. Table 2. Table 3. Table 4. Table 5. Table 6. Table 7. Table 8. Mortality causes (percent) of eastern wild turkey hens in Sussex County Delaware, March 15, 2010 March 15, 2012...14 Landscape variables affecting nest failure of eastern wild turkey hens Sussex County, Delaware from 2010-2011...28 Microhabitat variables affecting nest site selection of eastern wild turkey hens Sussex County, Delaware from 2010-2011...29 Landscape variables affecting nest site selection of eastern wild turkey hens Sussex County, Delaware from 2010-2011...30 Summary of nest fates for eastern wild turkey hens in Sussex County, Delaware, 2010 and 2011...31 Summary microhabitat and landscape variables used to investigate nest site selection of eastern wild turkey hens Sussex County, Delaware from 2010-2011...32 Landscape variables affecting roost site selection of eastern wild turkey hens Sussex County, Delaware from 2010-2012...43 Summary landscape variables used to investigate roost site selection of eastern wild turkey hens Sussex County, Delaware from 2010-2012...44 Table 9. Hen home range sizes (ha) for wild turkey hens March 2010 March 2012 in Sussex County, Delaware...58 vi

LIST OF FIGURES Figure 1. Figure 2. Figure 3. Figure 4. The study area for the reproductive ecology of eastern wild turkey hens in Sussex County, Delaware, 2010 2012...13 The probability of nest failure for eastern wild turkeys as a function of distance to the nearest edge in Sussex County, Delaware, 2010-2011...33 The probability of nest failure for eastern wild turkeys as a function of distance to the nearest road in Sussex County, Delaware, 2010-2011...34 The probability of nest failure for eastern wild turkeys as a function of distance to the nearest stream in Sussex County, Delaware, 2010-2011...35 vii

ABSTRACT With the increase popularity in Turkey hunting in the State of Delaware and the increase in turkey numbers since 1984, the state of Delaware and the university of Delaware started a project to investigate the reproduction success of Eastern wild turkey hens (Melagris gallopavo silvestris). I captured 106 turkeys using rocket nets and placed backpack transmitters on 76 hens in December 2009-March 2010 and December 2010- March 2011. I classified each bird as adult or juvenile, marked them with unique bands and placed transmitters on hens weighing >3.4 kg (>7.5lbs). To estimate survival, I took locations from fixed ground telemetry stations 3-7 times a week. The estimated survival rate of adult hens in 2010 was 0.47(SE=0.09), which was 0.21 less than the 2011 survival rate of 0.68 (SE=0.07; X 2 1 = 2.82, P = 0.093). Predation accounted for 87.1% (n = 27) of the moralities with foxes being attributed to most predation events (85.2%, n =23). Most mortalities (61.3%, n=19) occurred during the nesting season (15 April-15 June). The variation in annual survival justifies the need for continued monitoring of this population. Nest success has been found to be an important factor in overall population success. Nesting habitat variables and landscape variables have been studied before with a wide range of what hens select for a nest and what habitat variables effect nest success. I placed 76 transmitters on hens, and used telemetry to determine nesting times and locations from fixed ground stations. I walked in on nest locations to find an exact position, and used telemetry location to estimate other nest sites, creating 2 nest data sets. viii

I sampled 68 nests (2010, n = 27; 2011, n = 41) from 61 hens during the 2010 and 2011. Nest initiation date ranged from 23 April 28 June with most (80%) occurring the first week of May. Most (89%) hatching occurred during the first week of June with a range of 30 May 18 June. The average number of eggs per nests was 8.2 (SE = 0.635). Most nests (75.4%) failed, and I documented hen mortality (30.9%), unknown fate (39.1%), and predation (4%) as causes of nest failure. The estimated probability of daily nest failure was 0.046 (SE = 0.006) and estimated probability of nest failure after 28 days was 0.731 (SE = 0.049). The probability of nest failure increased with increasing distance from the nearest edge, road, and stream. Ground cover was the most important microhabitat variable for nest site selection, which was 40% greater than random plots. Hens selected nests that were closer to roads but farther from edges and streams than random points based on landscape variables. Hen nesting success needs further monitoring to fully understand nesting success on the population. Roost site habitat is an important factor allowing hens a place to avoid predation and thermoregulation during the night. I estimated roost site location by taking at least 2 roost site locations after night fall using telemetry equipment from fixed ground stations. I used ArcGIS to analyze distances from buildings, edge, roads, and streams. I paired these locations with a randomly selected point in the same habitat type. I had 678 roost site locations. Distance to the nearest had the great influence on roost site selection (Table 7). Roost site were 20m farther from roads than random points, whereas distance to nearest edge and stream were similar for roost sites and random points. Protection of large tracts will ensure the quality roost habitat. ix

Poult survival is another important factor affecting population growth. Poult survival shows managers the recruitment into the population. I used telemetry to find nest and then track successful hens. I used flush counts and lost poult calls inorder to see the hens and associated poults. I investigated poult survival from 76 collared hens. I used the 16 (2010, n = 8; 2011, n = 8) successful nests from these hens to estimate poult survival. I estimated poult survival as 0.343. and 0.525 for 2010 and 2011, respectively. My annual average poult survival was 0.434. Poult survival was excellent in Delaware and attributes to a healthy turkey population Home ranges encompass the area that is used by a turkey. This home range estimate can be used to ascertain if there is quality habitat in the area that birds use. The larger the home range could potentially mean the inferior habitat. I collected telemetry location 13 times a month from fixed ground stations. Home ranges did not differ for adult hens among seasons (50%, F3,121=0.68, P=0.565; 95%, F3,121=0.95, P=0.417) or between years (50%, F1,121=1.43, P=0.234; 95%, F1,121=1.27, P=0.263). The adult hen home range size ranged 346-971 ha and 2039-7595 ha for 50% and 95% distributions, respectively. I had too few juvenile birds to test difference in seasonal and annual home range sizes. Juveniles had larger home ranges than hens in all seasons expect the fall. Home ranges in Delaware are similar in size and suggest that there were adequate food sources to support the turkey population. Overall analyzing all aspects of reproduction in hens in Delaware, the population is stable. Nesting success clutch size and survival were all low compared to other studies; however poult survival was much greater than in other studies. The poult survival was compensating for the other deficiencies in reproduction among hens. x

Chapter 1 INTRODUCTION The eastern wild turkey (Meleagris gallopavo silvestris) was extirpated from many states during the late 1800 s by overharvest, market hunting, and habitat loss. A few remnant populations remained in few remote areas on the eastern United States. Early restoration efforts in the 1930 s and 40s used farmed raised birds, which had very little success. Starting in the 1950 s with the invention of rocket fired capture techniques, eastern states started moving wild birds from the remnant populations (Earl 1997, Dickson 1992) As a result, populations increased from 500,000 in 1959 to near 3.5 million birds in 1990 (Dickson 1992). Some states such as Delaware, Pennsylvania, and Arkansas have experienced declines in turkey numbers since the late 90s (Thogmartin 1999). Recently with the concern of stable or declining population in other states and the increased popularity of turkey hunting in the state of Delaware biologists wanted to determine the population dynamics of the turkey population. In Delaware, turkey populations have increased from 1,500 in the early 1990 s to 4,500 by 2010 (Taeply et al. 2011) but Delaware biologists are concerned with the reproduction potential of wild turkeys to maintain a popular hunting season (K. Reynolds, Delaware Department of Natural Resources and Environmental Control Division of Fish and Wildlife, personal communication). Due to the lack of research pertaining to eastern 1

wild turkeys in Delaware, research is needed to determine what factors might impact population demographics. Wild turkey populations are regulated by factors that affect reproductive success like nest success, hen survival, and poult survival (Roberts et al. 1995, Miller et al 1998, Thogmartin 1999). Nest success affects the number of poults produced. Adult hen survival is important because like most populations, female survival rates often have the greatest influence on population dynamics. Understanding causes of mortality can permit managers to developed management strategies to increase hen survival. Poult survival is the third key component into understanding reproductive success. Poults need to be recruited into the population at least as fast as the mortality to adult birds to keep the population stable. These three aspects make up the factors influencing population dynamics. Furthermore, roost site locations cannot be overlooked as a valuable habitat where hens spend most of their time, that allows them to thermoregulation and avoid predators. Estimating home range size and composition allows estimation of the required area needed to support wild turkeys. Reproductive potential is the most important limiting factor for wild turkey populations (Roberts et al. 1995). Currently, no data exist for wild turkey reproductive ecology or spatial ecology in Delaware. My objectives were to estimate survival of adult hens, estimate poult survival, estimate factors affecting roost use, estimate home range size, and determine what factors effect nest success of wild turkeys in Delaware. 2

Chapter 2 HEN SURVIVAL Abstract: Eastern wild turkey populations in Delaware and other Mid-Atlantic states have been stable or decreasing. Hen survival is one of the most important parameters in determining population stability and overall population success. I placed backpack transmitters on 76 hens. I monitored hens daily during the nesting season and at least 3 times a week during the remaining year. I investigated mortality causes by visual inspection. I documented 31 mortalities and estimated survival for 2010 as 0.47 (SE=0.09) and 2011 as 0.68 (SE=0.07; P = 0.093). Fox predation caused 80% of the mortality and most of the mortality (67%) was caused during the nesting season (15 April - 15 June). Natural predation was the largest cause of mortality limiting hen survival. With proper habitat management, hen survival should increase and continue the success story of turkey reintroductions. KEYWORDS: Eastern wild turkey, Melagris gallopavo, adult hen, survival, Kaplan-meier, mortality Introduction The eastern wild turkey (Meleagris gallopavo silvestris) was extirpated from many states during the late 1800 s by overharvest and market hunting. A few remnant populations remained in the rugged areas on the eastern United States, which served as 3

source populations for reintroductions. Populations have increased from 500,000 in 1959 to approximately 3.5 million birds in 1990 (Dickson 1992). Some states such as Delaware, Pennsylvania, and Arkansas have experienced declines in turkey numbers in the late 1990s (Thogmartin 1999). Although populations have recovered at the initial introduction sites, populations remain low is other areas (K. Reynolds, Delaware Department of Natural Resources and Environmental Control Division of Fish and Wildlife, personal communication). More information is needed about wild turkeys populations to better understand what factors might be limiting populations in some areas. Survival rates of hens are important to understanding the population dynamics of wild turkeys (Dickson 1992), but these data are lacking for Delaware; therefore, an estimate of these number will provide wildlife biologists with additional information to recover wild turkey populations in Delaware Adult hen survival is important because like most populations, female survival rate often has the greatest influence on population dynamics (Dickson 1992). Annual survival rates for wild turkeys vary among states and years within states (Vander Hagen et al 1988, Palmer et al. 1993, Wright et al 1996, Miller et al. 1998). In Massachusetts, Missouri, and Wisconsin, annual survival rates for hens ranged from 43% to 69% (Vander Hagen et al 1988, Vangilder et al. 1992, Wright et. al. 1996), whereas in Mississippi, Palmer et al. (1993) had survival rates as high as 81%. Annual survival rates also vary by years within states. For example, adult hen survival ranged from 22% to 77% over an 11 year period in Mississippi (Miller et al. 1998). Climatic factors are one factor that could explain variation in survival rates. Miller et al. (1998) speculated that the lowest year of survival was due to drought. Snowfall events have also been 4

demonstrated to be detrimental to turkey populations through direct winter mortality and reduced breeding potential (Roberts et al 1995, Wright et al 1996). Predation and harvest are the greatest cause of mortalities for adult hens. Miller et al. (1998) attributed mortality of adult hens to 46% predation [9 great horned owls (Bubo virgianus), 6 bobcat (Lynx rufus), 5 coyote (Canis latrans), 4 raccoon (Procyon lotor), 5 unknown mammals, 22 unknown predator], 10% illegal harvest, and 3% vehicle collisions. In Iowa, red foxes (Vulpes vulpes) have been reported as the greatest mammalian predator for hens and their poults (Hubbard et al. 1999). Predation risk increases in high fragmented areas and during the nesting and brood rearing seasons (Wilcove 1985, Miller et al. 1998). Roberts et al. (1995) documented that 46% of hen mortality occurred in the spring with 73% of deaths caused by predation. Illegal harvest may significantly affect turkey populations (Pack et al. 1999). Additionally, Pack et al. (1999) found that in hunted populations a decrease in fall hen survival has a significant effect on populations. Understanding the impact of predation and harvest on turkeys will allow a better understanding of how mortality affects the turkey population in Delaware. Many studies have investigated survival and cause specific-mortality in wild turkey hens (Vander Haegen 1988, Palmer et al. 1993, Miller et al 1998, Thogmartin and Schaffer 2000). Many of those studies have not been able to monitor hens sufficiently to identify all of the causes of mortalities, citing unknown reason for many of the events (e.g., Miller et al. 1998). Better understanding the causes of mortality and the amount caused by each factor is the first step to development management strategies to improve adult hen survival. Additionally, many studies in forested habitats have investigated survival and cause-specific mortality (Vander Haegen 1988, Palmer et al. 1993, 5

Thogmartin and Schaffer 2000), but few studies have been conducted in agricultural landscapes on the coastal plain (Morgan et. al 1996). I conducted my research on the coastal plain in a highly fragmented area and monitored birds 3-7 times per week to allow a clearer understanding of the causes of mortality. My objectives were to estimate annual survival rates of adult hens and documents causes of mortality. Study Area I conducted my research in central Sussex County, Delaware. My study was focused on Redden State Forest (hereafter Redden) and surrounding private land. Redden (4,642 ha) was located in Georgetown, Delaware (Figure 1). Redden consisted primarily of loblolly pine plantations (Pinus taeda), which were undergoing extensive thinning. Common species in the understory were greenbrier (Similax spp.), American holly (Ilex opaca), loblolly pine, and sweet gum regeneration in thinned stands. Redden had 16 discontinuous tracts varying in size (18-870ha) with roads, rural housing, and agricultural fields dividing the property. The private land surrounding Redden consisted of large farms 60-121 ha (150-300 ac) and a more diverse composition of tree species. The common canopy species of surrounding forests were oak (Quercus spp), sweet gum (Liquidambar stryaciflua), and black cherry (Prunus serotina), Virginia pine (Pinus virginiana), and loblolly pines. The common understory species were greenbrier, American holly, highbush blueberry (Vaccinium corymbosum), and early lowbush blueberry (Vaccinium pallidum). The 30-year average (1971-2000) for daily temperatures in Sussex County was -3.9 6.4 C in January and 18.2 30.5 C in July (Georgetown; National Oceanic and Atmospheric Administration 2010). Annual precipitation in Sussex County averaged 115cm (1971-2000, Georgetown; National Oceanic and 6

Atmospheric Administration 2010). The temperatures during my study were -0.3 C in January and 26.2 C in July, and these values fell within the range for the long-term average. Annual precipitation during my study was 86.8cm and was similar to the longterm average; however, we did experience several extreme snowfall events in 2010 (Redden 2012). The 30-term average (1971-2000) was 16.3 cm. In 2010, 88.9 cm of snow was deposited on the study area during 3 snow fall events in February and March, whereas in 2011, 49.5 cm of snow was deposited on the study area, which was similar to the long-term average (K. Brinson, Delaware Environmental Observing System, personal communication). Methods I captured 106 turkeys between December and March of 2010 and 2011. I captured turkeys using rocket nets baited corn and black oil sunflower seed (Schemnitz 2005). After capture, I placed birds in transport boxes until they could be processed. When I removed birds from the boxes, I placed loose fitting bags over their heads to decrease environmental stimuli to calm the birds. Of the 106 captured birds, I placed a VHF backpack transmitter (100g; Advanced Telemetry Systems, Isanti, MN) with an eight hour mortality sensor on 76 hens (>3.4 kg [>7.5lbs]). I secured backpack transmitters to the bird using elastic 3/16 inch shock cord (Norman et al 1997). Each captured turkey received a metal leg band on each leg secured by pop rivets (Schemnitz 2005). The bands had a unique bird identification number and a phone number to report the band if found. I weighed each bird using a manual hand held spring scale (Pesola Macro Line Kapuskasing, ON) and estimated age of the birds based on banding on the 9 th and 10 th primaries (Schroeder and Rob 2005), fan characteristics, and leg color (Dickson 7

1992). Capture myopathy has been shown to affect birds within the first 14 days following capture (Nicholson et al 2000), so we monitored hens daily after capture to document any deaths due to capture myopathy. The University of Delaware Institutional Animal Care and Use Committee approved my capture and handling methods (#1197). I collected telemetry locations on each turkey 1-7 times a week beginning 2 weeks after capture. I located turkeys using a hand-held receiver and yagi antenna (Advanced Telemetry Systems, Isanti, MN) from permanent stations, which were georeferenced using a handheld GPS (GeoExplorer 3, Trimble, Sunnyvale, California, USA). Of the collected bearings, we used the two bearings closest to 90 degrees (excluding any angles >120 or <60) and no more than 15 minutes apart to estimate the location of an individual. I collected telemetry locations between December 2009 and March 15 2011 I used radio telemetry to document the time and cause of mortality. When a mortality signal was emitted from a transmitter, I located the transmitter and categorized the mortality as mammalian predation, avian predation, harvest, vehicle collisions, illegal harvest, or capture related. I categorized mammalian predation as dog (Canis lupus), fox (Vulpes vulpes), or raccoon (Procyon lotor), because bobcats (Lynx rufus) and coyotes(canis latrans) did not inhabit the study area. I categorized a morality event as dog if the carcass was chewed on but not consumed, and/or dog scat or tracks were located within 5m around the kill site, or the transmitter was found in a residential area with dogs (Campa et al 1987, Thogmartin and Schaffer 2000, M. Cassalena personal communication). I categorized a mortality event as fox predation if the transmitter had teeth marks, fox scat and/or tracks were located within 5 m of the kill site, and/or the transmitter was located at a den site (Campa et al 1987, Thogmartin and Schaffer 2000, 8

M. Cassalena personal communication). I categorized a mortality event as raccoon if the transmitter had teeth marks, and/or raccoon scat or tracks were present within 5m around the kill site (Campa et al 1987, Thogmartin and Schaffer 2000, M. Cassalena personal communication). I categorized avian predation into hawks or owls. I categorized a mortality event as owl if I located transmitters at the base of a tree with an owl nest and/or the carcass was decapitated with owl feathers at the kill site (Campa et al 1987, Thogmartin and Schaffer 2000, M. Cassalena personal communication). I categorized a mortality event as hawks if I located the transmitter at the base of a hawk nest and/or hawk feathers were found at the kill site (Campa et al 1987, Thogmartin and Schaffer 2000, M. Cassalena personal communication). I separated harvest into legal and illegal categories. Delaware does not permit fall turkey hunting but bearded hens are legal during the spring turkey season, so I categorized a mortality event as legal harvest if a hunter reported harvesting a bird. I categorized a mortality event as illegal harvest if I located the transmitter in human residence, garbage cans, near obvious boot prints, and/or buried with shovel marks present and the carcass was gone. I categorized a mortality event as vehicle collision if I located the carcass or transmitter along a road, and/or the carcass was flattened (Campa et al 1987, Thogmartin and Schaffer 2000, M. Cassalena personal communication). I estimated annual survival rates using a Kaplan-Meier procedure (Pollock et al. 1989, Allison 2010). I removed capture related mortalities from the analysis, and I center censored any birds that went missing during the study (Lindsey and Ryan 1998, Murray 2006). If the date of mortality was unknown for a turkey, I used the midpoint between the date of the last telemetry location and the date found dead (Lindsey and Ryan 1998, 9

Murray 2006). I compared survival rates between years using a log-rank test (Allison 2010). I conducted all analyses using PROC LIFETEST in SAS (version 9.2, Cary, NC) with an alpha level of 0.10. Results The estimated survival rate of adult hens in 2010 was 0.47(SE = 0.09), which was 0.21 lower than the 2011 survival rate of 0.68 (SE = 0.07; X 2 1 = 2.82, P = 0.093). I documented 31 mortalities with mortalities attributed to foxes, owls, illegal harvest, vehicle, or unknown (Table 1). Predation accounted for 87.1% (n = 27) of the moralities with foxes being attributed to most predation events (85.2%, n = 23). I did not detect coyote, dog, hawk, or raccoon mortalities. Most mortalities (61.3%, n = 19) occurred during the nesting season (15 April-15 June). Discussion My estimates of survival fell within the range reported in the literature, but my survival estimates differed between years (Vangilder 1992, Wright et al 1996, Miller et al 1998). Although my survival estimate for the first year was lower than some other studies (Vander Haegen et. al 1988, Palmer et al 1993), my survival estimate for the second year was more comparable to recent turkey literature (Vangilder 1992, Wright et al 1996, Miller et al 1998). My results were similar to other studies that reported most mortality occurred during the nesting season (Vander Haegen et. al 1988, Palmer et al. 1993, Miller et al. 1998). Several studies have documented the impact of climatic factors on wild turkey survival (Roberts et al 1995, Wright et al 1996, Miller et al. 1998). The climatic factors during my study were normal during my study except extremely heavy snowfall during 10

late winter in 2010. Although we did not observe any direct morality like that reported by Roberts et al. (1995) and Wright et al. (1996), I think the heavy snowfall increased a hen s probability of predation during the nesting season. The extreme weather may have reduced body conditions going into nesting season and therefore increasing the likelihood of predation. I attributed most mortality to mammalian predation, which is similar to previous research on wild turkeys (Vander Haegen et al. 1988, Palmer et al. 1993, Miller et al. 1998). The suite of predators was less diverse in Delaware compared to previous studies (Miller et al 1998, Thogmartin and Schaeffer 1999), because coyotes and bobcats were not found on my study area. Additionally, I did not observe any raccoon predation even though raccoons were present on the study area. Most of the predation during the nesting season was caused by red fox. Delaware is somewhat unique in the fact one predator has such a strong effect on cause-specific mortality. Foxes are very abundant in Delaware, because until recently (put the year here) a hunting or trapping season did not exist in Delaware. Reducing the fox population could increase the survival of turkeys in the area (Frey and Conover 2007); however other factors affecting turkey survival could compensate for fox predation if the fox population was reduced. The red fox in Delaware are the apex predators but coyotes we absent from our study area. Coyotes regulate the number of foxes in an area by direct and indirect pressure on the foxes (Crooks and Soule 1999). The coyote has been recorded in areas around the study area and their population is expect to increase, which could cause the red fox population to decline. Illegal harvest was only documented during the spring season and only occurred twice during my study. In comparison to previous studies, illegal harvest had little impact on survival (Godfrey 11

and Norman 2000, Pack et al. 1999). My results were similar to past research that documented mammalian predation as being the primary cause of mortality but differed because red fox was the primary predator. Stable or increasing turkey populations have been associated with hen survival estimates of 72% (Roberts et al. 1995, Vangilder and Kurzejeski 1995, Vangilder 1996, Wright et al 1996 Thogmartin and Johnson 1999), whereas decreasing turkeys population have been associated with hen survival estimates of <50% (Paisley et al. 1998, Thogmartin and Johnson 1999 ). My survival estimate the first year suggests a decline population, whereas my estimate for the second year suggests a stable or increasing population. I suspect the survival estimate for the first year is uncharacteristically low because of the abnormally high snowfall. Management Implications Predation was the largest factor affecting adult hen survival, with foxes being the primary predator. Illegal harvest was not a factor during the study and has relatively little to no impact on the population. Stochastic weather events can be the cause for the increase of mortality during the nesting season. The low survival during the first year can be overcome as long as the following years have survival rates similar to year 2. My data for hen data survival suggests that the population is stable in Delaware. 12

Figure 1. The study area for the reproductive ecology of eastern wild turkey hens in Sussex County, Delaware, 2010-2012. 13

Table 1. Mortality causes (percent) of eastern wild turkey hens in Sussex County Delaware, March 15, 2010 March 15, 2012 Mortality Cause 2010 (n) 2011 (n) Average (n) Fox 70.6 (12) 78.6 (11) 74.2 (23) Owl 17.6 (3) 7.1 (1) 12.9 (4) Illegal Harvest 5.9 (1) 7.1 (1) 6.5 (2) Vehicle 0.0 (0) 7.1 (1) 3.2 (1) Unknown 5.9 (1) 0.0 (0) 3.2 (1) 14

Chapter 3 NESTING ECOLOGY OF EASTERN WILD TURKEYS IN AN AGRICULTURAL LANDSCAPE Abstract Turkey populations have decreased in size in certain areas of the eastern United States. Nest success and nest site selection are important variables for reproductive success. I placed back pack transmitters on 76 hens, and used telemetry to monitor nests throughout the nesting season (15 April -15 June). I monitored nests daily during the nesting. I sampled 68 nests (2010, n = 27; 2011, n = 41) from 61 hens during the 2010 and 2011. Nest incubation date ranged from 23 April 28 June with most (80%) occurring the first week of May. Most (89%) hatching occurred during the first week of June with a range of 30 May 18 June. The average number of eggs per nests was 8.2 (SE = 0.635). Most nests (75.4%) failed, and I documented hen mortality (30.9%), unknown fate (39.1%), predated (4%) as causes of nest failure. The estimated probability of daily nest failure was 0.046 (SE = 0.006) and estimated probability of nest failure after 28 days was 0.731 (SE = 0.049). The probability of nest failure increased with increasing distance from the nearest edge, road, and stream. Ground cover was the most important microhabitat variable for nest site selection, which was 40% higher than random plots. Hens selected nests that were closer to roads but farther from edges and 16

streams than random points based on landscape variables. Hen nesting success needs further monitoring to fully understand nesting success on the population. KEYWORDS: Melagris gallapavo, nesting success, wild turkey, habitat location, nests. Introduction Adequate nesting and brood habitat is crucial for hens to have successful broods, and the nesting and brood rearing seasons are when most poult and adult hen mortality occurs (Palmer et al. 1993, Roberts et. al 1995, Miller et. al. 1998). Nest success and poult survival are the two factors that affect recruitment (Paisley et al. 1998, Miller et al 1998,Thogmartin and Johnson 1999). Habitat is often a driving factor in determining nest success and poult survival. Understanding the relationship of habitat and nest success is critical to development habitat management guidelines for managing wild turkeys. Nest success is often linked to habitat. In Mississippi, turkeys that had successful nests used upland habitat, whereas unsuccessful nesters used bottomlands (Miller et al. 1999). Turkeys nested most often in pine regeneration which may provide dense vegetation a critical variable for nest success (Miller et al. 2000). Nest predation decreased as vegetation density increased because vegetative density may allow for turkeys to reduce the effectiveness of olfactory senses of potential predators (Bowman and Harris 1980). Large patches are most typically associated with nesting success, while highly fragmented areas may decrease nest success (Thogmartin 1999). Habitat type, interspersion, and patch size are often the most important factors for successful nesting (Heske 1995, Thogmartin 1999, Thogmartin 2001). Nest site selection and 17

success are often associated with quality brood habitat, which was high forb ground cover (Campo et. al. 1989, Miller et. al. 2000). Nesting success has varied among states from 20-38% (Roberts et al. 1995, Miller et al. 1998, Paisley et al. 1998, Thogmartin and Johnson 1999). In declining population, nest success was 20-21% (Paisley et al. 1998, Thogmartin and Johnson 1999), whereas nest success was 24%-38% that were stable or increasing population (Roberts et al. 1995, Miller et al. 1998). Nesting success may be the most important factor in determining the success of a turkey population, so understanding the factors that affect nesting success is critical to managing habitats for wild turkeys. Studies investigating factors influencing nesting success have found it is hard to define what criteria hens are using to select nest sites and what habitat factors contribute to successful nesting (Badyaev 1995, Thogmartin 1999, Thogmartin 2001, Miller and Conner 2005). Additionally, few studies have investigated nest success in a highly fragmented landscape with small forested patches and whether predator influence nest success in these environments (Heske 1995, Thogmartin 1999, Thogmartin 2001). My objectives were to determine what habitat variable influence nesting success, estimate nest success, and determine what habitat variables influence nest site selection of hens in Delaware. Study Area I conducted my research in central Sussex County, Delaware. My study was focused on Redden State Forest (hereafter Redden) and surrounding private land. Redden (4642 ha) was located in Georgetown, Delaware (Figure 1). Redden consisted primarily of loblolly pine plantations (Pinus taeda), which were undergoing extensive 18

thinning. Common species in the understory were greenbrier (Similax spp.), American holly (Ilex opaca), loblolly pine, and sweet gum regeneration in thinned stands. Redden had 16 discontinuous tracts varying in size (18-870ha) with roads, rural housing, and agricultural fields dividing the property. The private land surrounding Redden consisted of large farms 60-121 ha (150-300 ac) and a more diverse composition of tree species. The common canopy species of surrounding forests were oak (Quercus spp), sweet gum (Liquidambar stryaciflua), and black cherry (Prunus serotina), Virginia pine (Pinus virginiana), and loblolly pines. The common understory species were greenbrier, American holly, highbush blueberry (Vaccinium corymbosum), and early lowbush blueberry (Vaccinium pallidum). The 30-year average (1971-2000) for daily temperatures in Sussex County was -3.9 6.4 C in January and 18.2 30.5 C in July (Georgetown; National Oceanic and Atmospheric Administration 2010). Annual precipitation in Sussex County averaged 115cm (1971-2000, Georgetown; National Oceanic and Atmospheric Administration 2010). The temperatures during my study were -0.3 C in January and 26.2 C in July, and these values fell within the range for the long-term average. Annual precipitation during my study was 86.8cm and was similar to the longterm average; however, we did experience several extreme snowfall events in 2010 (Redden 2012). The 30-term average (1971-2000) was 16.3cm. In 2010, 88.9 cm of snow was deposited on the study area during 3 snow fall events in February and March, whereas in 2011, 49.5cm of snow was deposited on the study area, which was similar to the long-term average (K. Brinson Delaware Environmental Observing System, personal communication). 19

Methods I captured 106 turkeys between December and March of 2010 and 2011. I captured turkeys using rocket nets baited corn and black oil sunflower seed (Schemnitz 2005). After capture, I placed birds in transport boxes until they could be processed. When I removed birds from the boxes, I placed loose fitting bags over their heads to decrease environmental stimuli to calm the birds. Of the 106 captured birds, I placed a VHF backpack transmitter (100g; Advanced Telemetry Systems, Isanti, MN) with an eight hour mortality sensor on 76 hens (>3.4kg [>7.5lbs]). I secured backpack transmitters to the bird using elastic 3/16 inch shock cord (Norman et al 1997). Each captured turkey received a metal leg band on each leg secured by pop rivets (Schemnitz 2005). The bands had a unique bird identification number and a phone number to report the band if found. I weighed each bird using manual hand held spring scale (Pesola Macro Line Kapuskasing, ON) and estimated age of the birds based on banding on the 9 th and 10 th primaries (Schroeder and Rob 2005), fan characteristics, and leg color (Dickson 1992). Capture myopathy has been shown to affect birds within the first 14 days following capture (Nicholson et al 2000), so we monitored hens daily after capture to document any deaths due to capture myopathy. The University of Delaware Institutional Animal Care and Use Committee approved my capture and handling methods (#1197). I collected telemetry locations between December 2009 and March 15 2011. I collected telemetry locations on each turkey 1-7 times a week beginning 2 weeks after capture. I located turkeys using a hand-held receiver and yagi antenna (Advanced Telemetry Systems, Isanti, MN) from permanent stations, which were geo-referenced using a handheld GPS (GeoExplorer 3, Trimble, Sunnyvale, California, USA). Of the 20

collected bearings, we used the two bearings closest to 90 degrees (excluding any angles >120 or <60) and no more than 15 minutes apart to estimate the location of an individual. I estimated telemetry locations using Location of a Signal (LOAS, Ecological Software Solutions, Sacramento, CA). I conducted a telemetry test to estimate my telemetry error. A colleague placed transmitters in areas frequented by turkeys with transmitters and I did not know the locations of the transmitters. They placed transmitters on bottles filled with a saline solution 1 meter off the ground. My average telemetry error polygon was 0.95 ha. I conducted daily telemetry during the nesting season (15 April 15 June; Bob Eriksen, National Wild Turkey Federation Certified Biologist, personal communication), and assumed hens were nesting when 2 consecutive locations had <10 degree azimuth differences (Weinstein et. al. 1999). I allowed 15 days for egg laying and then I used telemetry locations to estimate the nest location from the roadway. I used the telemetry equipment to encircle the nest to estimate the location. I marked the nest site with 2 flags approximately 50 meters from the nest (Godfrey and Norman 2000, Thogmartin 1999, Weinstein et al. 1999). I stayed approximately 50 meters from the nest to prevent hens from abandoning their nest due to disturbance (Weinstein et al. 1999). Approximately 30 days after incubation was detected and when bearings were >15 degrees from the previous day s reading, I assumed the hen s nest had hatched (Thogmartin 1999, Weinstein et al. 1999). I examined the nest and classified them as successful or unsuccessful. A successful nest was defined as one that produced at least one hatched egg and an unsuccessful nest produced no hatched eggs (Thogmartin 1999, Roberts et al 1995). Because I waited until 15 days after the start of incubation to locate the nests, I 21

could not find nests that were abandoned prior to the 15 days; therefore, I had 2 sets of nest locations: those verified by finding the actual nest locations and those with estimated nest locations based on radio telemetry, which included nests that were abandoned prior to the 15 days. After the hen left the nest, I located the actual nest location. I conducted vegetation sampling with the nest as the center point of the vegetation plot, and a random plot for each nest site in the same habitat type. I picked random sites that were in the same habitat type. I used excel to randomly generate a list distances and bearings that I used to find random location in the same habitat type. I measured vegetation density using a Nudds board. I took a measurement in each of the 4 cardinal directions 15 meters from plot center and averaged the values for an estimate at each point (Nudds 1977). I used a factor 5 and 10 prism to estimate basal area. In the 4 cardinal directions, I estimated canopy cover using a spherical densitometer and averaged the 4 readings for the estimate at each plot. I visually estimated the percentage ground cover in a 1-meter square centered on the nest and in four cardinal directions touching the first square. I averaged the four squares touching the center plot to estimate ground cover for the plot. I used ArcGIS (9.3, Redlands, California) to determine distance to the nearest building edge, road, and water source from each nest. I used GIS to randomly select a point to pair with each nest location on the landscape within 500 meter buffer around each nest. I limited random point for each nest to the same habitat as the nest location. I used a correlation matrix to reduce redundancy in the microhabitat and landscape variables (Sokal and Rohlf 1995). If two variables were correlated (r > 0.5), I selected the most biologically appropriate variable (e.g., distance to edge over distance to 22

building). This variable reduction procedure reduced the microhabitat variables to factor 10 basal area, canopy cover, ground cover, and vegetation density and landscape variables to edge, streams, and roads. I used MCestimate to estimate daily and overall probabilities of nest-failure due to specific causes (Etterson et al. 2007, Etterson 2011). I used all nest locations for this analysis (i.e., those with verified nest locations and those with nest locations estimated using radio telemetry). I first modeled what factors affected nest failure using the covariates edge, streams, and roads. I had 6 a priori models (Table 2). I did not include microhabitat variables because these were not available for nests that failed prior to 15 days. To estimate support for each model, I used Akaike s Information Criterion (AIC) and Akaike model weights (w i ) for each model calculated by MCestimate (Burnham and Anderson 2002). Because no clear single model had sufficient support, I used delta AIC and model weights to determine which variables had the greatest influence on nest site selection (Burnham and Anderson 2002, Arnold 2010). To investigate the influence of vegetation variables on nest site selection, I used a case-control logistic regression in SAS (version 9.2, Cary, NC; Allison 1999, Stoke et al. 2000, Manly et al. 2002, Thomas and Taylor 2006). I compared nest sites to random sites at the microhabitat level and the landscape level. I used only nests with verified locations for these analyses. I had 5 a priori models at the microhabitat level (Table 3) and 4 a priori models at the landscape level (Table 4). To estimate support for each model, I calculated Akaike s Information Criterion (AIC) and Akaike model weights (w i ) for each model (Burnham and Anderson 2002). Because no clear single model had sufficient 23

support, I used delta AIC and model weights to determine which variables had the greatest influence on nest site selection (Burnham and Anderson 2002, Arnold 2010). Results I sampled 68 nests (2010, n = 27; 2011, n = 41) from 61 hens during the 2010 and 2011. Nest incubation dates ranged from 23 April 28 June with most (80%) occurring the first week of May. Most (89%) hatching occurred during the first week of June with a range of 30 May 18 June. The average number of eggs per nests was 8.2 (SE = 0.635; 2010 = 8.4, SE = 0.87; 2011 = 8.1, SE = 0.97). Most nests (75.4%) failed, and I documented hen mortality (30.9%), unknown fate (39.1%), predated (4%) as causes of nest failure (Table 5). The estimated probability of daily nest failure was 0.046 (SE = 0.006) and estimated probability of nest failure after 28 days was 0.731 (SE = 0.049). The distance to the nearest edge, road, and stream influenced the probability of nest failure (Table 2). The probability of nest failure increased with increasing distance from the nearest edge, road, and stream (Figure 2-4). Most microhabitat variable had little influence on nest site selection (Table 3). Ground cover was the most important microhabitat variable for nest site selection (Table 3). I documented the nest sites have 40% more ground cover than random points (Table 6). Landscape variable had similar influence on nest site selection (Table 4). Hen selected nests that were closer to roads but farther from edges and streams than random points (Table 6). Discussion Nest success was the most important factor influencing recruitment and overall population size (Miller and Leopold 1992, Palmer et al. 1993). My nest success 24

estimates were lower than most studies (Speake 1980, Palmer et al. 1993, Nicholson et al. 1995, Roberts et al. 1995, Vangilder and Kurzejeski 1995, Badyaev et al. 1996, Miller et al. 1995). Only a few papers reported nest success lower than 27 percent, which was my estimate of nest survival (Paisley et al. 1998, Thogmartin 1999). Most researchers did not include nest failures prior to 15 days after the start of incubation, which biased their estimates of nest success high. My estimate and those of Paisley et al. (1998) and Thogmartin (1999) included these nest failures and produced similar nest success estimates. The nest failures were due mainly to hen predation especially during the first year. A possible explanation for increased the hen predation, was the major snowfall events that occurred in 2010. Melting snow produced large areas standing water, which created island areas for nesting habitat. These islands reduced search times for foxes and may have played a role in increased hen mortality during 2010. The unknown causes were nest failures that were suspected to be predated nests or abandonment. Habitat quality has been shown to limit nest success (Bowman and Harris 1980, Miller and Conner 2005). Miller and Conner (2005) found that hens selected upland pine plantation regeneration areas over bottomland hardwood forest. In my study, I did not document hens using regeneration areas but instead areas with a mixed understory. This difference was likely from bottomland hardwood forests in Mississippi having a greater chance of being flooded than mixed hardwood stands on my study area. Nest success in my study was influenced by distance to the nearest edge, roads, and stream but the relationship was negative for each variable. The habitat on edge is more dense and diverse. These results are similar to previous research that found increased nest success in area of greater vegetation density (Bowman and Harris 1980, Badyaev 1995, Miller 25

and Conner 2005). Additionally, Thogmartin (1999) found hens to nest closer to edge though to the detriment of nest success. Like edge the habitat near roads in more dense and diverse, which would increase nest success (Bowman and Harris 1980, Badyaev 1995, Miller and Conner 2005). Another benefit of being closer to roads could be reduced fox predation because these areas are avoided by foxes (Heske et al 1995). On my study, many of the streams were in open areas, which were frequently mowed. Along with poor nest success, the clutch size was surprisingly small compared to other studies. Our clutch size was lower than the 9-14 eggs reported in other turkey studies (Vangilder 1992, Badyaev et al. 1996, Thogmartin 1999), whereas out estimates were similar to Miller et al (1995) that reported 8.5 eggs per clutch. A study on the behavior of turkey nest predators in Delaware documented that predators did not destroy the entire nest during some nest predation events (Bowman, unpublished data). In this study, foxes and sometime raccoon returned to a nest on multiple times removing one egg each visit instead of destroying the entire nest. Additionally, these predators often removed only part of the clutch and then never returned. The reduced clutch sizes that I observed could have resulted from mesocarnivores removing part of my clutches. Habitat influences nest site selection in wild turkeys (Badyaev 1995, Thogmartin 1999, Miller et al. 2000). In my study, microhabitat had little influence on nest site selection except for ground cover (Badyaev 1995 Bowman and Harris 1980). My results were similar to other studies that documented hens selecting nest sites with denser vegetation (Badyaev et al 1996, Miller et al. 2000, Spears et al 2005). In addition to ground cover, I documented that distance to edge, streams, and roads influenced nest site selection. In my study, hens tended to nest further from streams in constant to previous 26