Thesis PDF - Faculty Sites
Transcription
Thesis PDF - Faculty Sites
ABSTRACT WINIARSKI, JASON MATTHEW. Breeding Ecology and Space-use of Bachman’s Sparrow (Peucaea aestivalis) at the Northern Periphery of its Range. (Under the direction of Christopher E. Moorman). Longleaf pine (Pinus palustris) forests provide habitat for unique assemblages of flora and fauna, and contribute significantly to biodiversity in the southeastern United States. Historically, the longleaf pine ecosystem covered ~30 million ha from Virginia south through Texas, but less than 5% remains as scattered and degraded remnants due to extensive landuse change and fire suppression. Strategies for restoring and conserving this ecosystem include a variety of habitat management approaches that occur at the site-level, but less consideration has been given to approaches at the landscape-scale. Knowledge of certain species’ natural history and demographics is essential for directing such multi-scale strategies. Bachman’s sparrow (Peucaea aestivalis) is a declining ground-nesting bird that is strongly associated with fire-maintained longleaf pine forests, and has been considered to be a surrogate species for this system because of its reliance on frequent fire regimes, open canopy, and diverse groundcover characterizing healthy longleaf pine communities. Despite the species’ importance, relatively little is known about the demographic mechanisms driving Bachman’s sparrow declines, or its breeding season ecology at the species’ northern range limits where some of the greatest declines are occurring. To address this knowledge gap and inform management in this region, we investigated nest-site selection, space-use, and reproductive success of Bachman’s sparrows across a fragmentation gradient in southeastern North Carolina. Despite clear differences in nest-site selection by physiographic region (Coastal Plain and Sandhills), we documented no significant predictors of nest survival in the Sandhills, but nest survival declined later in the nesting season in the Coastal Plain. Daily nest survival rates (Coastal Plain = 0.953, Sandhills = 0.938) were similar between physiographic regions, and were consistent with published studies from the species’ core range. Within home ranges, Bachman’s Sparrows used locations with a greater density of woody vegetation and forbs, and intermediate levels of grass density. Home ranges varied from 2 ha to 18 ha, but we documented no significant predictors of variation in home range size. Bachman’s sparrow home ranges at our study sites were larger than previously reported estimates for the southern portion of the species’ range, likely because of methodological differences among studies. Finally, we examined whether local- or landscape-scale factors influenced two components of reproduction: pairing success and fledging offspring. Percent habitat within 3 km of a male’s territory positively influenced pairing success, with males in highly fragmented landscapes remaining unpaired throughout the breeding season. Conversely, habitat amount did not affect the probability of fledging offspring for paired males, indicating that reduced connectivity may be hindering dispersal. Overall, this study suggests that creating and maintaining regionally-specific habitat characteristics via prescribed fire should increase the amount of nesting cover and microhabitat features, but may not influence nest survival rates or home range sizes. More importantly, increasing landscape-level habitat amount is the most effective means of promoting breeding opportunities for Bachman’s sparrow and potentially other dispersal-limited species in the longleaf pine ecosystem. © Copyright 2016 Jason M. Winiarski All Rights Reserved Breeding Ecology and Space-use of Bachman’ Sparrow (Peucaea aestivalis) at the Northern Periphery of its Range by Jason Matthew Winiarski A thesis submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the degree of Master of Science Fisheries, Wildlife, and Conservation Biology Raleigh, North Carolina 2016 APPROVED BY: _______________________________ Dr. Christopher Moorman Committee Chair _______________________________ Dr. Christopher DePerno _______________________________ Dr. George Hess DEDICATION This work is dedicated to my parents for their love and support. ii BIOGRAPHY Jay Winiarski grew up in North Kingstown, Rhode Island. As a child he developed a passion for wildlife and the outdoors while feeding and identifying backyard birds, and during many days spent fishing with his father and brother in Narragansett Bay. He pursued a BS in Wildlife & Conservation Biology at the University of Rhode Island, where he graduated in 2008. Upon graduation, he worked for several years as a seasonal research technician on ecological field projects in Rhode Island, New York, Jamaica, Costa Rica, and Oregon. In January 2014, he entered the Fisheries, Wildlife, and Conservation Biology Program at North Carolina State University. iii ACKNOWLEDGMENTS I thank my advisor, Dr. Chris Moorman, for this opportunity, and for all of his support, sound advice, and patience through all aspects of my graduate research and education. Also, I appreciate the feedback, insight, and support provided by my committee members, Drs. Chris DePerno and George Hess. Several Bachman’s sparrow researchers including John Carpenter, Alex Fish, Paul Taillie, Jeff Marcus, and Scott Anderson provided helpful knowledge, feedback, and guidance on this project. Foremost, I am extremely grateful to John Carpenter, who offered invaluable advice and field support during this project. Alex Fish also provided much expertise from his concurrent research, helpful conversation, and collected much of the data and answered many questions that made the first chapter of this thesis possible. Many thanks to the North Carolina Wildlife Resources Commission and North Carolina State University for providing research funding for this project. The Department of Defense and the Endangered Species Branch at Ft. Bragg provided funding for some of the nest data collected and used in the first chapter of my thesis. I am grateful to all my research assistants that made this work possible: April Bartelt, David Bernasconi, Amber Bledsoe, Taylor Root, and Mike Wallgren. Susie Masecar, Mark McAlister, and Mikayla Seamster also volunteered for several long days to help complete vegetation surveys. The staff of Holly Shelter Game Land provided use of their facilities, which were invaluable for planning field logistics and entering data, and I thank them for their flexibility and cooperation. Several land owners and land managers made this research possible by graciously allowing access to their properties and providing fire history data: Angie Carl (The Nature Conservancy), Bobby Harrelson (Compass Pointe), Kiley Hinson (Molpus Timberlands), iv Ray Hudson (Sleepy Creek Farms), Hervey McIver (The Nature Conservancy), and Leslie Starke (NC Plant Conservation Program). Jeff Humphries and Henry Wood also let us borrow some of the radio-telemetry equipment used to track sparrows. I thank Cindy Burke who worked hard to help with purchases, housing, and paying the bills. Also, I thank Sarah Slover for guiding me through all of the graduate student procedures, and quickly responding to and resolving the many questions I had for her. Last but not least, thank you to my family (Jayce, Lenny, Kris, Sarah, and Fletcher) for their love and support on the path toward completing this degree and during my previous endeavors. I am deeply indebted to my partner, April Bartelt, for embarking on this journey with me to North Carolina, assisting with many aspects of this project, and her love and encouragement. v TABLE OF CONTENTS LIST OF TABLES ............................................................................................................... viii LIST OF FIGURES ................................................................................................................ x CHAPTER 1: NEST-SITE SELECTION AND NEST SURVIVAL OF BACHMAN’S SPARROWS IN CONTRASTING LONGLEAF PINE ECOSYSTEMS ......................... 1 ABSTRACT ....................................................................................................................... 1 INTRODUCTION ............................................................................................................ 3 METHODS ........................................................................................................................ 5 Study area ..................................................................................................................... 5 Nest-searching and monitoring..................................................................................... 6 Nest-site vegetation measurements .............................................................................. 7 Landscape .................................................................................................................... 8 Statistical analyses ....................................................................................................... 8 RESULTS ........................................................................................................................ 11 Nest-site selection ...................................................................................................... 11 Nest survival ............................................................................................................... 11 DISCUSSION .................................................................................................................. 12 MANAGEMENT AND CONSERVATION IMPLICATIONS .................................. 16 LITERATURE CITED .................................................................................................. 18 CHAPTER 2: SPACE-USE AND HABITAT SELECTION BY BACHMAN’S SPARROW AT THE NORTHERN RANGE PERIPHERY ............................................ 32 ABSTRACT ..................................................................................................................... 32 INTRODUCTION .......................................................................................................... 33 METHODS ...................................................................................................................... 35 Study area ................................................................................................................... 35 Field data collection ................................................................................................... 36 Statistical analyses ..................................................................................................... 38 RESULTS ........................................................................................................................ 41 Home range characteristics ........................................................................................ 41 Microhabitat selection ................................................................................................ 41 Home range variation ................................................................................................ 42 DISCUSSION .................................................................................................................. 42 LITERATURE CITED .................................................................................................. 48 CHAPTER 3: REPRODUCTIVE CONSEQUENCES OF HABITAT FRAGMENTATION IN A DECLINING RESIDENT LONGLEAF PINE BIRD ........ 60 ABSTRACT ..................................................................................................................... 60 INTRODUCTION .......................................................................................................... 62 METHODS ...................................................................................................................... 64 Study area ................................................................................................................... 64 vi Field data collection ................................................................................................... 65 Statistical analyses ..................................................................................................... 69 RESULTS ........................................................................................................................ 73 Territory and landscape characteristics ..................................................................... 73 Reproductive success .................................................................................................. 73 DISCUSSION .................................................................................................................. 74 LITERATURE CITED .................................................................................................. 79 APPENDICES ....................................................................................................................... 91 APPENDIX A .................................................................................................................. 92 APPENDIX B .................................................................................................................. 93 APPENDIX C .................................................................................................................. 94 vii LIST OF TABLES Chapter 1: Nest-site Selection and Nest Survival of Bachman’s sparrows in Contrasting Longleaf Pine Ecosystems Table 1. Description of variables considered for examining nest-site selection and nest survival by Bachman’s sparrows in North Carolina, USA, 2014–2015. ................................ 23 Table 2. Summary statistics (mean ± SD) for nest and reference plot vegetation variables, and Mann-Whitney U statistics comparing nest-site vegetation between the Coastal Plain and Sandhills physiographic regions in North Carolina, USA, 2014-2015. ................................. 24 Table 3. Mixed-effects logistic regression models of nest-site selection by Bachman’s sparrows in the Coastal Plain and Sandhills physiographic regions of North Carolina, USA, 2014-2015. Paired plot ID was used as a random effect in all models. Models were compared using Akaike’s Information Criterion corrected for small sample size (AICc). ..................... 25 Table 4. Parameter estimates for the top nest-site selection models for the Coastal Plain and Sandhills physiographic regions of North Carolina, USA, 2014-2015. ................................. 26 Table 5. Bachman’s sparrow nest fates by physiographic region and study site (n = total nests found, A = abandoned, S = successful, F = failed, U = unknown fate, P = parasitized, B = burned by prescribed fire or wildfire, and T = total nests used for data summaries and analyses). ................................................................................................................................ 27 Table 6. Bachman’s sparrow nest survival models for the Coastal Plain and Sandhills physiographic regions of North Carolina, USA, 2014-2015. Models were compared using Akaike’s Information Criterion corrected for small sample size (AICc). ............................... 28 Chapter 2: Space-use and Habitat Selection by Bachman’s Sparrow at the Northern Range Periphery Table 1. Summary statistics (mean ± SD) for used and reference plot vegetation variables for 37 radio-tracked Bachman’s Sparrows in the Coastal Plain physiographic region in North Carolina, USA, 2014-2015. .................................................................................................... 53 Table 2. Top 10 mixed-effects logistic regression models of microhabitat selection by Bachman’s sparrows in the Coastal Plain region of North Carolina, USA, 2014-2015. BirdID was used as a random effect in all models. Models were compared using Akaike’s Information Criterion corrected for small sample size (AICc). .............................................. 54 Table 3. Model-averaged parameter estimates for the top microhabitat selection models (∆AICc ≤ 2) for 37 radio-tracked male Bachman’s Sparrows in the Coastal Plain region of North Carolina, USA, 2014-2015. Shaded cells indicate parameter estimates that do not overlap zero. ........................................................................................................................... 55 viii Table 4. Linear models examining home range variation for Bachman’s sparrows in the Coastal Plain region of North Carolina, USA, 2014-2015. Model sets tested habitat and nonhabitat predictors. Models were compared using Akaike’s Information Criterion corrected for small sample size (AICc). ....................................................................................................... 56 Table 5. Summary of published studies of Bachman’s Sparrow home-range size in comparison to findings presented here. .................................................................................. 57 Chapter 3: Reproductive Consequences of Habitat Fragmentation in a Declining Resident Longleaf Pine Bird Table 1. Variables used in candidate models testing the effects of local- and landscape-scale factors on reproductive success of male Bachman’s sparrows in southeastern North Carolina, 2014-2015. .............................................................................................................................. 84 Table 2. Summary of Bachman’s sparrow territories monitored in the Coastal Plain of North Carolina, USA, 2014-2015. Percent of males that fledged young included paired males only. Note that 8 banded males were monitored in both years. ....................................................... 85 Table 3. Top (∆AICc ≤ 2) mixed-effects logistic regression models of pairing and fledging success for Bachman’s sparrows in the Coastal Plain region of North Carolina, USA, 20142015. Site was used as a random effect in all models. Models were compared using Akaike’s Information Criterion corrected for small sample size (AICc). Model parameters: Habitat (percent habitat within 3 km of a territory centroid), TSF (time since fire), PC2 (habitat principal component 2), and Wing (wing length). .................................................................. 86 Table 4. Parameter estimates and 95% confidence intervals for model-averaged variables (∆AICc ≤ 2) used to model pairing and fledging success for 96 male Bachman’s Sparrows in the Coastal Plain region of North Carolina, USA, 2014-2015. Model parameters: Hab (percent habitat within 3 km of a territory centroid), TSF (time since fire), PC1 (habitat principal component 1), PC2 (habitat principal component 2), and Wing (wing length). ..... 87 ix LIST OF FIGURES Chapter 1: Nest-site Selection and Nest Survival of Bachman’s sparrows in Contrasting Longleaf Pine Ecosystems Figure 1. Locations of study sites and physiographic regions where nest-site selection and nest survival data were collected during 2014-2015. Physiographic region designation follows level III classifications defined by Bailey (1995). The Sandhills is a distinct level IV subregion of the Southeastern Plains. Study sites are indicated with four-letter abbreviation; from west to east, Fort Bragg Military Installation (FTBR; n = 85 nests), Green Swamp Savanna Preserve (GSPR; n = 3 nests), Compass Pointe (COPO; n = 1 nest), Holly Shelter Game Land (HSGL; n = 42 nests), and Shaken Creek Savanna Preserve (SCPR; n = 1 nest). ................................................................................................................................................. 29 Figure 2. Predicted probability of selection of a nesting site for Bachman’s sparrows by physiographic region in North Carolina, USA, in 2014–2015. ‘Hits’ refers to the total number of vegetation contacts on the entire pole averaged across a plot. Shaded regions indicate 95% confidence intervals. .............................................................................................................. 30 Figure 3. Predicted probability of Bachman’s sparrow daily nest survival rate in relation to Julian date for the Coastal Plain of North Carolina, USA, in 2014–2015. Shaded region indicates 95% confidence interval. ........................................................................................ 31 Chapter 2: Space-use and Habitat Selection by Bachman’s Sparrows at the Northern Range Periphery Figure 1. Locations of study sites and home ranges of 37 Bachman’s Sparrows that were radio-tracked in North Carolina, USA, 2014-2015. ............................................................. 58 Figure 2. Predicted probability of microhabitat use by Bachman’s sparrows in the Coastal Plain physiographic region, North Carolina, USA, in 2014–2015, for woody vertical density, grass vertical density, and forb groundcover density. Shaded regions indicate 95% confidence intervals. ............................................................................................................ 59 Chapter 3: Reproductive Consequences of Habitat Loss and Fragmentation in a Declining Resident Songbird Figure 1. Study sites in southeastern North Carolina where reproductive data were collected for 96 Bachman’s sparrow territories, 2014-2015. Study sites are indicated with four-letter abbreviation: Boiling Spring Lakes Preserve (BSLP), Compass Pointe (COPO), Green Swamp Preserve (GSPR), Holly Shelter Game Land (HSGL), Molpus Timberlands (MOLP), Pinch Gut Game Land (PGGL), Shaken Creek Preserve (SCPR), and Sleepy Creek Farms (SCFA). .................................................................................................................................. 88 x Figure 2. Mean (± SD) habitat amount (%) within 3 km of male territory centroids for 8 sites in the Coastal Plain region of North Carolina, USA, 2014-2015. Sample size of males at each site is listed below the site code. ............................................................................................ 89 Figure 3. Predicted probability of pairing in relation to percent habitat within 3 km. Shaded regions indicate 95% confidence intervals. ........................................................................... 90 xi CHAPTER 1 NEST-SITE SELECTION AND NEST SURVIVAL OF BACHMAN’S SPARROWS IN CONTRASTING LONGLEAF PINE ECOSYSTEMS ABSTRACT Longleaf pine (Pinus palustris) communities of the southeastern United States have experienced high rates of habitat loss and fragmentation, coinciding with dramatic population declines for a variety of taxa that inhabit the system. Bachman’s sparrow (Peucaea aestivalis) is closely associated with the longleaf pine ecosystem and is listed as a species of conservation concern across its entire range. Understanding aspects of Bachman’s sparrow breeding biology is crucial to mitigating population declines and directing restoration and management of remnant longleaf pine forest, but factors such as nest-site selection and nest survival have received little attention due to difficulty in locating the species’ nests. To address this knowledge gap, we located 132 Bachman’s sparrow nests in the Coastal Plain and Sandhills physiographic regions of North Carolina during 2014-2015, and modeled nestsite selection and nest survival as a function of vegetation characteristics, temporal factors, and landscape-level habitat amount. There were distinct differences in nest-site selection between regions, with Bachman’s sparrows in the Coastal Plain region selecting greater woody vegetation density and lower grass density at nest sites than at non-nest locations. Conversely, sparrows selected nest sites with intermediate grass density and relatively high tree basal area in the Sandhills. Despite clear differences in habitat selection, we detected no predictors of nest survival in the Sandhills, but nest survival declined later in the nesting season in the Coastal Plain. Daily survival rates were similar between regions, and were consistent with published studies from the species’ core range. Creating and maintaining 1 regionally-specific vegetation characteristics should increase the amount of nesting cover, but may not increase nest survival rates for Bachman’s sparrows. 2 INTRODUCTION Although knowledge of habitat selection is important for avian conservation, perhaps more critical is understanding how habitat factors drive demographic processes such as nest survival. Nest survival is an important component of avian fitness (Clark and Martin 2007), and predation is the primary cause of nest failure in birds (Martin 1993). Nest predation often results in complete reproductive failure (Ricklefs 1969, Martin 1993), and nest success directly impacts a bird’s fitness. Therefore, it generally is assumed that nest-site selection is adaptive, such that selected habitat features should minimize predation risk and maximize reproductive output and fitness (Martin 1998). Theoretically, birds that nest in locations that increase nest survival should pass their genotypes on to subsequent generations, and their offspring would be adapted to select more productive habitat features (Jaenike and Holt 1991). Despite the prevailing theory that habitat selection should be adaptive, empirical studies often demonstrate a lack of congruence between nest-site selection and nest survival (Chalfoun and Schmidt 2012). Although nest-site selection can lead to improved nest survival through concealment (Martin 1998, Moorman et al. 2002), it may ultimately be overwhelmed by factors occurring at multiple spatial and temporal scales. For example, landscape context (Thompson et al. 2002), edge effects (Chalfoun et al. 2002), nest initiation date or year (Grant et al. 2005, Bulluck and Buehler 2008), nest age (Dinsmore et al. 2002, Grant et al. 2005), and weather patterns (Rotenberry and Wiens 1989, Conway and Martin 2000, Morrison and Bolger 2002) can be more important drivers of nest survival than site selection. Also, anthropogenic habitat change may lead to mismatches between previously adaptive nest-site selection and nest survival (i.e., ecological traps; Misenhelter and 3 Rotenberry 2000, Robertson and Hutto 2007). Additionally, predator communities could be so diverse that birds may not be able to select for nest sites that conceal their nests from varied predator search strategies (Filliater et al. 1994, Davis 2005). Although nest-site selection and nest survival rarely are straightforward, an improved understanding of the factors that influence nest survival is critical for species of conservation concern. Bachman's sparrow (Peucaea aestivalis) is a species of conservation concern primarily inhabiting longleaf pine (Pinus palustris) forests of the southeastern US. Bachman’s sparrows nest on the ground in areas with a diverse understory of low shrubs, grasses, and forbs maintained by frequent fires (Dunning and Watts 1990, Plentovich et al. 1998). Populations of Bachman’s sparrows are declining throughout the species’ range due to fire suppression and habitat loss in an ecosystem with ~3% of its original extent remaining (Noss et al. 1995, Dunning 2006). Breeding Bird Survey data indicate that populations have declined by ~ 3% per year survey-wide since 1966, with greater declines reported at the species’ contracting range margins (Sauer et al. 2014). These population declines are leading to extirpation of the species from areas that have supported Bachman’s sparrows during the past ~125 years (Dunning 2006). Consequently, Bachman’s sparrow is a state-listed species of conservation concern in every state it inhabits (Cox and Widener 2008). Understanding aspects of Bachman’s sparrow reproductive success is crucial to mitigating population declines. Yet, published data on the breeding biology of Bachman’s sparrows is limited due to the difficulty in finding the species’ cryptic nests (Dunning 2006). Previous studies have estimated nest survival (Haggerty 1988, Stober and Krementz 2000, Perkins et al. 2003, Tucker et al. 2006, Jones et al. 2013) and evaluated nest-site selection (Haggerty 1988, Jones et al. 2013). Of these studies, only one has sought to quantitatively 4 link Bachman’s sparrow nest-site selection to nest survival (Jones et al. 2013) and none have included the effects of space or time in nest survival models. Additionally, there is a high degree of geographic variation in longleaf vegetation depending on local climate, soils, and topography (Peet 2006). Given the lack of nesting information and variability of longleaf pine communities, research regarding Bachman’s sparrow responses to local environmental conditions is needed to guide conservation efforts. We modeled Bachman’s sparrow nest-site selection and survival within contrasting longleaf pine ecosystems of the Middle Atlantic Coastal Plain and Sandhills physiographic regions of North Carolina, USA. Specifically, the objectives of our research were to: 1) identify important vegetation features selected for nest-sites, and 2) determine whether there was a relationship between nest survival and temporal, local-, or landscape-scale covariates within each region. Using a unique dataset with the largest sample size of Bachman’s sparrow nests published to date, we provide one of the first comprehensive reproductive assessments for the species. METHODS Study Area In 2014 and 2015, we located and monitored nests at five sites within the Sandhills and the Middle Atlantic Coastal Plain (hereafter, Coastal Plain) physiographic regions in North Carolina, USA (Figure 1). The Sandhills physiographic region is located in central North Carolina and is characterized by rolling, hilly terrain dissected by abundant blackwater streams (Sorrie et al. 2006). We focused our work in the Sandhills at a single study site, Fort Bragg Military Installation (hereafter, Ft. Bragg), which occupies 73,469 ha in portions of Cumberland, Harnett, Hoke, and Moore counties. Ft. Bragg and other adjacent areas in the 5 Sandhills comprise the largest contiguous block of longleaf pine-wiregrass (Aristida stricta) ecosystem remaining in North Carolina. Wetlands (i.e., streamhead pocosins and sandhill seeps) are interspersed within the xeric longleaf pine uplands, and are dominated by switchcane (Arundinaria gigantea), ferns, or dense shrubs, depending on fire intensity (Sorrie et al. 2006). Ft. Bragg is divided into >1000 burn units, which are managed with prescribed fire on a 3-year return interval, with burns usually occurring between February and August. In contrast, the Coastal Plain consists of flat terrain with little topographic relief. Four study sites within the region were primarily composed of mesic longleaf pine forests bordered by pocosin wetlands. Wet pine savannas in this region were characterized by seasonally-saturated soils, and the understory was more or less dominated by grasses, forbs, low shrubs, and insectivorous plants depending on soil moisture and fire frequency. Most prescribed fires in this region were implemented with a less frequent return interval (~3-5 years) than fire prescriptions at Ft. Bragg. Study sites in the Coastal Plain were imbedded in a matrix of urban and rural residential development and agricultural lands, and were managed by The Nature Conservancy, the North Carolina Wildlife Resources Commission, and one private landowner. Nest Searching and Monitoring Nest searching and monitoring efforts were conducted from 1 April - 10 August, 2014 - 2015. Nests were located by: 1) observing adults carrying food and nest material, or exhibiting behaviors elicited when predators were near nests (i.e. flushing from the nest and performing distraction displays) during weekly reproductive index surveys and territorymapping of color-banded male Bachman’s sparrows (Vickery et al. 1992); and 2) radiotagging and tracking male (Coastal Plain) and female (Sandhills) sparrows to nests for 6 concurrent studies. All Bachman’s sparrows (both sexes) that we captured were banded with a USGS band and a unique combination of colored leg bands. For a subset of birds, we attached a 0.64-g (~3.5% of body mass) radio-transmitter (Blackburn Transmitters, Nacogdoches, Texas, USA) using a figure-eight leg loop harness (Rappole and Tipton 1991, Streby et al. 2015). We used hand-held antennas and radio receivers to track males and females to nests. Following discovery by either nest-searching method, nests were flagged from 5-15 m away and monitored every 1– 4 days using standard nest-monitoring methods that minimize disturbance to nests (Martin and Geupel 1993). During each nest visit, we recorded the status of nest contents (i.e., number of eggs, number of nestlings, and estimated age of nestlings) until nests fledged or failed. We considered a nest successful if at least one Bachman’s sparrow young fledged, and confirmed fledging by observing parental behavior (i.e., alarm calls and feeding) around the nest during checks and reproductive surveys. Because of differences in nest-searching methods, nests in the Coastal Plain typically were located by observing adults attending nests, which resulted in finding nests mostly during the nestling stage (73% of nests). Conversely, nests in the Sandhills were located mainly by following radio-tagged females, and 59% of nests found were in the nest-building or incubation stage. Nest-Site Vegetation Measurements We measured vegetation structure at every meter along two perpendicular 10-m transects centered at the nest location. Vegetation categories included grass, woody vineshrub (hereafter woody vegetation), and forb-fern following Taillie et al. (2015). We quantified vegetation structure using indices of density by recording the number of ‘hits’ (i.e., contacts) by each vegetation category on 0.1-m sections of a 1.5-m vertical pole. 7 Vertical density of the vegetation was measured the number of hits on the entire pole, while groundcover density was calculated using the number of hits on the first 0.1-m section of the pole (Wiens and Rotenberry 1981). Maximum height of understory vegetation was recorded as the tallest 0.1-m section of the pole with a vegetation contact (Wiens and Rotenberry 1981, Moorman and Guynn 2001, Taillie et al. 2015). Canopy closure was estimated by recording the presence or absence of live canopy using an ocular sighting tube at every meter. Lastly, we used a 10-factor prism from the plot center to determine the total basal area of pines surrounding the nest. We repeated these same measurements at a reference plot (i.e., a non-nest location) 50 m away from the nest in a random direction. We chose this distance to eliminate overlap between nest and reference points, yet constrain reference locations within a bird’s home range (7.92 ± 4.05 ha, mean ± SD, J. Winiarski, unpublished data). Vegetation sampling was completed within 14.6 ± 11.4 days (range = 0-69 days) of a nest fledging or failing to account for short-term changes in vegetation structure. Landscape A binary Bachman’s sparrow species distribution model (Pickens et al., in prep.) was used to quantify landscape-level habitat amount. We used the spatialEco package (Evans 2015) in R to derive the amount of habitat within a 1-km buffer around each nest. We chose this metric because it offered a simple yet comprehensive interpretation of different landscape attributes (Fahrig 2013) and has been shown to be an important predictor of Bachman’s sparrow occupancy (Taillie et al. 2015). Statistical analyses 8 Based on previous studies as well as our own field experience, we developed a list of variables that might influence selection at the nest-site scale (Table 1). We used nonparametric Mann-Whitney U tests to compare nest-site vegetation variables between the Sandhills and the Coastal Plain. For all descriptive statistics of vegetation variables we present means ± SD, and we considered probability tests significant at P ≤ 0.05. To assess the importance of individual vegetation variables on nest-site selection, we used an information-theoretic approach (Burnham and Anderson 2002) in R (R Core Development Team 2015). We constructed mixed-effects logistic regression models in the package lme4 (Bates et al. 2015) to determine variables that differentiated nest sites from reference locations. Because of strong regional variation in vegetation structure, we developed distinct model sets for each physiographic region. Prior to analysis, we tested for correlations (r > |0.60|) between all pairwise combinations of variables. To determine the best model for each region, we used a manual forward-selection approach (Burnham and Anderson 2002). We began by comparing the null model with univariate models, and retained variables if they had a lower AICc value compared with the null model. We continued to build more complex models by adding covariates that decreased the AICc value. In addition to a single linear effect of grass, we included a quadratic effect of vertical grass density because dense grass can restrict movement on the ground in the absence of fire (Brooks and Stouffer 2010, Taillie et al. 2015). For all models, we included ‘plot pair ID’ as a random effect to account for non-independence between each nest and paired reference plot. Nest observation data were formatted for logistic exposure analysis (Shaffer 2004), where a nest’s exposure was calculated as the number of days between nest visits, and its 9 survival during the observation period was coded as a binary variable (0 = failed, 1 = survived). The logistic exposure model is a generalized linear model that specifies a binomial error distribution and a logit-link function that accommodates variable exposure length for each nest, thus eliminating potential bias due to different exposure periods (Shaffer 2004). We applied the same manual forward-selection approach to model daily survival rate in relation to 13 variables (Table 1) for each region. We included a year effect because annual variation in survival may occur due to weather events, cyclical increases in predator or prey abundance, and resource availability (Rotenberry and Wiens 1989). Linear and quadratic models of Julian date on nest survival were considered to assess temporal effects potentially correlated with factors such as weather, food availability, and predators (Grant et al. 2005). In addition, we modeled daily survival rate in relation to nest stage (incubation or nestling) because predation risk may vary between stages. We included vegetation characteristics selected by Bachman’s sparrows at nest sites because we expected that they were adaptive and increased nest survival. We tested the effect of time since fire because it can alter predator communities (Jones et al. 2004) or food and cover availability, and has been shown to limit reproductive performance of Bachman’s sparrows when time since fire exceeds 3 years (Tucker et al. 2006). Also, we included an effect of habitat amount to determine the influence of landscape on nest survival. Daily nest survival rates were used to derive estimates of nesting success for each region by exponentiating the respective daily survival rate to the 22-day nesting period (incubation and nestling stages) reported by Haggerty (1988) and Tucker et al. (2006). We determined the top models of nest-site selection and nest survival for each region by selecting models that were < 2 ΔAICc using the MuMIn package (Barton 2016). For each 10 of these model sets, a parameter was considered as strongly affecting the response variable if the 95% CI did not include 0, and was considered to have no relationship if the 95% CI overlapped 0. RESULTS Nest-site selection We located 132 Bachman’s sparrow nests between 2014 and 2015. We used 130 in the nest-site selection analysis because two nests were destroyed by prescribed fire and could not be measured for vegetation characteristics. Several nest-site selection variables differed between physiographic regions (Table 2). Maximum height of forbs (U = 2506.0, P ≤ 0.005) was greater in the Coastal Plain, and woody vegetation measurements including vertical density (U = 3429.0, P ≤ 0.005), groundcover density (U = 3802.0, P ≤ 0.005), and max height (U = 842.0, P ≤ 0.005) were greater at nest-sites in the Coastal Plain than in the Sandhills. Top mixed-effects logistic regression models differed for each region (Table 3). Nest sites had lower grass vertical density and greater vertical density of woody vegetation than reference locations in the Coastal Plain (Table 4; Figure 2). In the Sandhills, sparrows selected nest-sites with intermediate vertical grass densities and greater tree basal area compared with reference locations (Table 4; Figure 2). Nest survival We included 114 active nests in the nest survival analysis. We excluded 2 nests destroyed by prescribed fire, 4 nests discovered the same day the nest fledged, 3 nests with unknown fates, and 7 nests considered abandoned by adults (Table 5). The best-fit model for nest survival in the Coastal Plain included a negative correlation between increasing Julian date and daily survival rate (DSR; β = -0.02; 95% CI: -0.048, -0.004; Figure 3). The next11 best model was competitive (ΔAICc ≤ 2; Table 6), and differed from the top model only by the addition of a quadratic term for Julian date. However, 95% confidence intervals around the parameter estimates for this model overlapped zero. For the Sandhills region, there were no predictors significantly affecting daily survival rate (the 95% CI of the parameter estimates of all the predictor variables included 0), and the null model ranked among competing models (ΔAICc = 1.41, Table 6). Daily survival rate was 0.953 (95% CI: 0.919 0.973) from the best-fit Coastal Plain model and 0.938 (95% CI: 0.916 - 0.955) from the null Sandhills model. Nest survival derived from these models was 35% (95% CI: 16 – 55%) for the Coastal Plain, and 24% (95% CI: 15 – 36%) for the Sandhills. Nest survival did not significantly differ between study regions. DISCUSSION Our results emphasize the importance of examining nest-site selection and survival in different habitat types and geographic regions across the species’ range. Bachman’s sparrows selected distinct nest sites with specific vegetation characteristics, and these relationships differed among study regions. Our results for the Coastal Plain were consistent with recent work in the Red Hills region of Georgia, showing that Bachman’s sparrows used nest sites with greater woody vegetation cover and lower grass cover and standing crop than random locations (Jones et al. 2013). Greater woody vegetation structure has been hypothesized to improve access to nest sites and create favorable microclimate conditions (Haggerty 1988). Moreover, Haggerty (1988) and Jones et al. (2013) noted the presence of bare ground at the entrances of nest sites, which may facilitate nest attendance or fledgling movement upon leaving the nest. We documented similar features (e.g., bare ground and sphagnum moss) for nests in this study, and likewise noted that adults carrying food frequently landed within 12 close proximity of nest-sites and continued on the ground (Haggerty 1988, Jones et al. 2013). Recent studies examining factors that influence Bachman’s sparrow distribution have documented decreasing occupancy beyond a threshold of grass (Brooks and Stouffer 2010, Taillie et al. 2015), and dense grass may restrict movement and foraging activity near the nest (Haggerty 1998). Also, dense grass has been hypothesized to negatively impact nest microclimate because it retains water from dew and rain events (Jones et al. 2013). Woody vegetation may provide cover from solar radiation and protect nests from heat stress (With and Webb 1993), but quickly outcompetes herbaceous vegetation in the absence of fire (Jones et al. 2013). Most differences in Bachman’s sparrow nest-site selection between physiographic regions likely can be attributed to the available understory plant composition. Like the Red Hills region, mesic soils in the Coastal Plain are relatively nutrient-rich compared to most soil types associated with longleaf pine forests (Jones et al. 2013), and support a diverse plant community (Peet 2006). In contrast, soil productivity of longleaf pine uplands at Ft. Bragg and other locations in the Sandhills is relatively low (Perry and Amecher 2007, Lashley et al. 2015). Grasses tend to dominate the groundcover layer in these upland areas, and the occurrence of woody stems often is limited (Lashley et al. 2014, Just et al. 2015), perhaps due to the homogeneous application of prescribed fire (Lashley et al. 2014). Compared to the Coastal Plan, vertical woody density at Ft. Bragg was approximately 2.5 times lower and groundcover density 7 times lower, suggesting the availability of woody vegetation for nestsite selection was much lower at Fort Bragg. Also, woody vegetation height also was greater at Ft. Bragg, and tall shrubs are less likely to be selected for nest sites (Jones et al. 2013). Lastly, woody vegetation at Ft. Bragg was composed primarily of re-sprouting oaks (Quercus 13 spp.), which are not characterized by the same growth structure as gallberry (Ilex glabra), huckleberry (Gaylussacia frondosa), blueberry (Vaccinium spp.), and swamp redbay (Persea palustris) that were dominant at Coastal Plain sites. Despite clear selection of habitat features, vegetation structure at Bachman’s sparrow nests did not influence nest survival. This is not uncommon in many nest survival studies (e.g., Davis 2005, Bulluck and Buehler 2008), and is consistent with previous research examining the breeding biology of Bachman’s sparrows (Haggerty 1995, Jones et al. 2013). Similar results have been explained by random patterns of nest predation (Filliater et al. 1994) or the absence of ‘safe’ nest sites due to myriad search behaviors by diverse predator communities (Davis 2005). Hence, birds may not be able to select nest-sites that deter predation from all possible sources. Rather, important vegetation variables from this study may be related to nest access and microclimate hypotheses proposed previously. Regardless of the mechanism or degree to which nest-site selection is adaptive, Bachman’s sparrows appear to select nest-sites with specific habitat attributes, and ensuring the presence of these nest-site characteristics in the landscape certainly is important for the conservation of this species. We detected few significant effects for temporal covariates on nest survival. We hypothesized that older Bachman’s sparrow nests containing nestlings were more likely to be depredated than nests in the incubation stage, but found no support for this relationship in either study region. However, nest survival in the Coastal Plain declined significantly later in the breeding season, likely because of changes in the predator community. Other studies have demonstrated a similar trend of nest survival decreasing over time within a single breeding season (Grant et al. 2005, Davis et al. 2005), which may correspond to an increase in 14 predator abundance or activity later in the season as temperatures increase (Grant et al. 2015). Snakes were abundant at our study sites (A. Fish and J. Winiarski, personal observation), and are the most important nest predators in the southeastern US (Thompson and Ribic 2012, DeGregorio et al. 2014). Minimal disturbance at most of the depredated nests suggests that snakes might be major nest predators in our study, and snakes were a significant cause of predation during the post-fledging period in a fledgling survival study at Ft. Bragg (A. Fish, personal communication). Snake activity peaks seasonally in response to temperature changes or mating behavior, and can drive seasonal variation in nest survival (Sperry et al. 2008). Time since fire did not influence nest survival, which was unexpected given the importance of frequent fire for increasing the abundance and reproductive success of Bachman’s sparrows (Tucker et al. 2004, 2006). Specifically, reproductive success has been shown to quickly decline when time since fire exceeds 3 years (Tucker et al. 2006); however, it is important to note that 89% of nests in this study were located in stands last burned ≤ 3 years, effectively excluding longer fire intervals from our analyses. We expected nest survival to increase with greater habitat amount in the surrounding landscape. Most of the nests located in both physiographic regions were in found in relatively continuous landscapes, and lack of a landscape effect may have been due to the limited sample size of nests from more isolated sites. In the Coastal Plain, few male Bachman’s sparrows in isolated locations showed indications of pairing, suggesting that pairing success may be of greater concern to reproductive success in more fragmented landscapes (Winiarski 2016). 15 We estimated daily survival rates that were within the range of those previously reported. Estimates of daily survival were 0.919 – 0.965 in Arkansas pine plantations (Haggerty 1988), 0.952 in the Coastal Plain of South Carolina (Stober and Krementz 2000), and 0.899 – 0.960 in Florida dry prairie (Perkins et al. 2003). More recent work by Jones et al. (2013) showed a daily survival rate of 0.960 for nests in mature longleaf pine forest in Georgia (Jones et al. 2013). Bachman’s sparrow populations are declining throughout their range, with the greatest declines occurring at the species’ shrinking range margins (Sauer et al. 2014). Yet, our results and those from core Bachman’s sparrow populations suggest that reduced nest survival may not pose the most significant threat to this declining species at the current northern extent. Recent work by Taillie et al. (2015) suggests that Bachman’s sparrows are negatively affected by habitat fragmentation, but the demographic factors driving these responses are not well understood. Hence, further research is needed to determine the influence of habitat, landscape, and temporal factors on other components of reproductive success (i.e., pairing success and fledgling survival) and adult survival rates. MANAGEMENT AND CONSERVATION IMPLICATIONS Based on our results, the frequent application of prescribed fire should be promoted as an essential management tool to improve nest-site conditions for Bachman’s sparrows. Frequent prescribed fire (i.e., 2-3 year cycle) can be used to maintain both low grass density (Coastal Plain and Sandhills) and low-statured woody vegetation (Coastal Plain; Glitzenstein et al. 2003). While our study indicated that Bachman’s sparrows exhibit strong patterns of nest-site selection, vegetation structure did not improve nest survival as we expected. Bachman’s sparrow conservation depends partly on a better understanding of factors related to other components of reproductive success (i.e., pairing success and fledgling survival). 16 Therefore, until the mechanisms influencing reproductive success of Bachman’s sparrows are fully understood, habitat-based management approaches alone may be ineffective at increasing breeding productivity. 17 LITERATURE CITED Bailey, R. G. (1995). Descriptions of the ecoregions of the United States. Report #1391. U.S. Department of Agriculture, Forest Service. Barton, K. (2016). MuMIn: Multi-model inference R package 1.9.13. http:// CRANRprojectorg/package.MuMIn Bates, D., M. Maechler, B. Bolker, and S. Walker (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67:1–48. doi:10.18637/jss.v067.i01 Brooks, M. E., and P. C. Stouffer. (2010). Effects of Hurricane Katrina and salvage logging on Bachman's Sparrow. The Condor 112:744-753. Bulluck, L. P., and D. A. Buehler. (2008). Factors influencing Golden-winged Warbler (Vermivora chrysoptera) nest-site selection and nest survival in the Cumberland Mountains of Tennessee. Auk 125:551-559. Burnham, K. P., and D. R. Anderson. (2002). Model selection and multimodel inference: a practical information-theoretic approach, 2nd ed. Springer-Verlag, New York. Chalfoun, A. D., and K. A. Schmidt. (2012). Adaptive breeding-habitat selection: is it for the birds? Auk 129:589-599. Chalfoun, A. D., M. J. Ratnaswamy, and F. R. Thompson, III. (2002). Songbird nest predators in forest-pasture edge and forest interior in a fragmented landscape. Ecological Applications 12:858-867. Clark, M. E., and T. E. Martin. (2007). Modeling tradeoffs in avian life history traits and consequences for population growth. Ecological Modelling 209:110-120. Conway, C. J., and T. E. Martin. (2000). Evolution of passerine incubation behavior: influence of food, temperature, and nest predation. Evolution 54:670-685. Cox, J. A., and B. Widener. (2008). Lightning-season burning: friend or foe of breeding birds. Tall Timbers Research Station Miscellaneous Publication 17. Davis, S. K. (2005). Nest-site selection patterns and the influence of vegetation on nest survival of mixed-grass prairie passerines. The Condor 107:605-616. DeGregorio, B. A., S. J. Chiavacci, P. J. Weatherhead, J. D. Willson, T. J. Benson, and J. H. Sperry. (2014). Snake predation on North American bird nests: culprits, patterns and future directions. Journal of Avian Biology 45: 325-333. 18 Dinsmore, S. J., G. C. White, and F. L. Knopf. (2002). Advanced techniques for modeling avian nest survival. Ecology 83:3476-3488. Dunning Jr, J. B., and B. D. Watts. (1990). Regional differences in habitat occupancy by Bachman's Sparrow. Auk 107:463-472. Dunning, J. B. (2006). Bachman’s sparrow (Peucaea aestivalis). The Birds of North America. Number 038. Fahrig, L. (2013). Rethinking patch size and isolation effects: The habitat amount hypothesis. Journal of Biogeography 40:1649–1663. Filliater, T. S., R. Breitwisch, and P. M. Nealen. (1994). Predation on northern cardinal nests: does choice of nest site matter? Condor 96:761–768. Grant, T. A., T. L. Shaffer, E. M. Madden, and P. J. Pietz. (2005). Time-specific variation in passerine nest survival: new insights into old questions. The Auk 122:661-672. Gotmark, F., D. Blomqvist, O. C. Johansson, and J. Bergkvist. (1995). Nest site selection: a trade-off between concealment and view of the surroundings? Journal of Avian Biology 26:305-312. Haggerty, T. M. (1986). Reproductive ecology of Bachman’s Sparrow (Aimophila aestivalis) in central Arkansas. Dissertation. University of Arkansas, Fayetteville, USA. Haggerty, T. M. (1988). Aspects of the breeding biology and productivity of Bachman’s Sparrow in central Arkansas. Wilson Bulletin 100:247–255. Haggerty, T. M. (1995). Nest-site selection, nest design and nest-entrance orientation in Bachman’s sparrow. The Southeastern Naturalist 40:62–67. Jaenike, J., and R. D. Holt. (1991). Genetic variation for habitat preference: evidence and explanations. The American Naturalist 137:67-90. Jones, D. D., M. L. Conner, T. H. Storey, and R. J. Warren. (2004). Prescribed fire and raccoon use of longleaf pine forests: implications for managing nest predation? Wildlife Society Bulletin 32:1255-1259. Jones, C. D., J. A. Cox, E. Toriani-moura, J. Robert, and R. J. Cooper. (2013). Nest-site characteristics of Bachman’s sparrows and their relationship to plant succession following prescribed burns. Wilson Journal of Ornithology 125:293–300. Just, M. G., M. G. Hohmann, and W. A. Hoffmann. (2015). Where fire stops: vegetation structure and microclimate influence fire spread along an ecotonal gradient. Plant Ecology 114. 19 Lashley, M. A., M.C. Chitwood, A. A. Prince, M. B. Elfelt, E. L. Kilburg, C. S. Deperno, and C. E. Moorman. (2014). Subtle effects of a managed fire regime: a case study in the longleaf pine ecosystem. Ecological Indicators 38:212-217. Lashley, M. A., M. C. Chitwood, C. A. Harper, C. E. Moorman, and C. S. DePerno. (2015). Poor soils and density-mediated body weight in deer: forage quality or quantity? Wildlife Biology 21: 213-219. Martin, T. E. (1993). Nest predation and nest sites: new perspectives on old patterns. BioScience 43:523–532. Martin, T. E., and G. R. Geupel. (1993). Nest-monitoring plots: methods for locating nests and monitoring nest success. Journal of Field Ornithology 64:507–519. Martin, T. E. (1998). Are microhabitat preferences of coexisting species under selection and adaptive? Ecology 79:656-670. Misenhelter, M. D., and J. T. Rotenberry. (2000). Choices and consequences of habitat occupancy and nest site selection in sage sparrows. Ecology 81:2892–2901. Moorman, C. E., and D. C. Guynn Jr. (2001). Effects of group-selection opening size on breeding bird habitat use in a bottomland forest. Ecological Applications 11:1680–1691. Moorman, C. E., D. C. Guynn Jr., and J. C. Kilgo. (2002). Hooded Warbler nesting success adjacent to group-selection and clearcut edges in a southeastern bottomland forest. Condor 104: 366-377. Morrison, S. A., and D. T. Bolger. (2002). Variation in a sparrow’s reproductive success with rainfall: food and predator-mediated processes. Oecologia 133:315-324. Noss, R. F., E. T. LaRoe, III, and J. M. Scott. (1995). Endangered ecosystems of the United States: a preliminary assessment of loss and degradation. U.S. Department of Interior National Biological Service, Biological Report 28, Washington, D.C., USA. Peet, R. K. (2006). Ecological classification of longleaf pine woodlands. Pages 51-93 in J. Shibu, E. J. Jokela, D. L. Miller, editors. The Longleaf Pine Ecosystem. Springer New York, Perkins, D. W., P. D. Vickery, and W. G. Shriver. (2003). Spatial dynamics of source-sink habitats: effects on rare grassland birds. Journal of Wildlife Management 67:588-599. Perry, C. H. and Amacher, M. C. (2007). Forest resources of the United States. – < www.fs.fed.us/rm/pubs_other/wo_gtr078_042_044.pdf > pp. 42– 44. 20 Plentovich, S., J. W. Tucker Jr., N. R. Holler, and G. E. Hill. (1998). Enhancing Bachman's Sparrow habitat via management of Red-cockaded Woodpeckers. Journal of Wildlife Management 62:347-354. R Development Core Team. (2015). R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Rappole, J. H., and A. R. Tipton (1991). New harness design for attachment of radio transmitters to small passerines. Journal of Field Ornithology 62:335-337. Ricklefs, R. E. (1969). An analysis of nesting mortality in birds. Smithsonian Contributions to Zoology 9:1–48. Robertson, B. A., and R. L. Hutto. (2007). Is selectively harvested forest an ecological trap for Olive-sided Flycatchers? The Condor 109:109-121. Rotenberry, J. T., and J. A. Wiens. (1989). Reproductive biology of shrubsteppe passerine birds: geographical and temporal variation in clutch size, brood size, and fledging success. Condor 91: 1-14. Sauer, J. R., J. E. Hines, J. E. Fallon, K. L. Pardieck, D. J. Ziolkowski, Jr., and W. A. Link. (2014). The North American Breeding Bird Survey, Results and Analysis 1966 - 2013. Version 01.30.2015. Shaffer, T. L. (2004). A unified approach to analyzing nest success. Auk 121:526-540. Sorrie, B. A., J. B. Gray, and P. J. Crutchfield. (2006). The vascular flora of the longleaf pine ecosystem of Fort Bragg and Weymouth Woods, North Carolina. Castanea 71:129–161. Sperry, J. H., R. G. Peak, D. A. Cimprich, and P. J. Weatherhead. (2008). Snake activity affects seasonal variation in nest predation risk for birds. Journal of Avian Biology 39: 379383. Stober, J. M. and D. G. Krementz. (2000). Survival and reproductive biology of the Bachman’s Sparrow. Proceedings of the Annual Conference of Southeastern Association of Fish and Wildlife Agencies 54:383–390. Streby, H. M., T. L. McAllister, S. M. Peterson, G. R. Kramer, J. A. Lehman, and D. E. Andersen (2015). Minimizing marker mass and handling time when attaching radiotransmitters and geolocators to small songbirds. The Condor 117:249-255. Taillie, P. J., M. N. Peterson, and C. E. Moorman. (2015). The relative importance of multiscale factors in the distribution of Bachman's Sparrow and the implications for ecosystem conservation. Condor 117: 137-146. 21 Thompson III, F. R., and C. A. Ribic. (2012). Conservation implications when nest predators are known. Studies in Avian Biology 43: 23–33. Thompson, F. R. III, T. M. Donovan, R. M. DeGraaf, J. Faaborg, and S. K. Robinson. (2002). A multi-scale perspective of the effects of forest fragmentation on birds in Eastern forests. Studies in Avian Biology 25:8-19. Tucker Jr., J. W., W. D. Robinson, and J. B. Grand (2004). Influence of fire on Bachman’s Sparrow, an endemic North American songbird. Journal of Wildlife Management 68:1114– 1123. Tucker, J. W., W. D. Robinson, and J. B. Grand. (2006). Breeding productivity of Bachman’s sparrows in fire-managed longleaf pine forests. Wilson Journal of Ornithology 118:131–137. Vickery, P. D., M. L. Hunter Jr., and J. V. Wells. (1992). Use of a new reproductive index to evaluate relationship between habitat quality and breeding success. Auk 109:697-705. Wiebe, K. L. (2001). Microclimate of tree cavity nests: is it important for reproductive success in Northern Flickers? The Auk 118:412-421. Wiens, J. A., and J. T. Rotenberry. (1981). Habitat associations and community structure of birds in shrubsteppe environments. Ecological Monographs 51:21-41. With, K. A., and D. R. Webb. (1993). Microclimate of ground nests: the relative importance of radiative cover and wind breaks for three grassland species. Condor 95:401-413. 22 Table 1. Description of variables considered for examining nest-site selection and nest survival by Bachman’s sparrows in North Carolina, USA, 2014–2015. Variable Description Nest-site selection Nest survival 2 -1 PineBA Pine basal area (m /ha ) x x Canopy Canopy cover (%) xa,b xa,b VerGR Grass vertical density (hits) x x VerWD Woody vertical density (hits) x x VerFB Forb vertical density (hits) x x GrdGR Grass groundcover density (hits) xa,b xa,b a GrdWD Woody groundcover density (hits) x xa GrdFB Forb groundcover density (hits) x x MaxGR Max grass height (m) x x MaxWD Max woody height (m) x x MaxFB Max forb height (m) xa xa Habitat Habitat amount within 1 km (%) x Fire Time since fire (years) x Date Julian date (linear and quadratic) x Year Study year (2014 or 2015) x Stage Nest stage (incubation or nestling) x a Variables that were omitted because of correlations with other variables (Coastal Plain). b Variables that were omitted because of correlations with other variables (Sandhills). 23 Table 2. Summary statistics (mean ± SD) for nest and reference plot vegetation variables, and Mann-Whitney U statistics comparing nest-site vegetation between the Coastal Plain and Sandhills physiographic regions in North Carolina, USA, 2014-2015. Variable PineBA VerGR VerWD VerFB GrdGR GrdWD GrdFB MaxGR MaxWD MaxFB *** P ≤ 0.05 Coastal Plain Nest-site Reference 11.78 ± 6.90 11.36 ± 5.76 5.61 ± 3.47 7.90 ± 4.57 2.90 ± 1.34 2.11 ± 1.12 0.65 ± 0.86 0.35 ± 0.53 1.40 ± 0.90 1.94 ± 1.04 1.33 ± 0.64 1.00 ± 0.65 0.12 ± 0.12 0.07 ± 0.12 0.37 ± 0.11 0.42 ± 0.13 0.24 ± 0.09 0.23 ± 0.10 0.27 ± 0.11 0.25 ± 0.12 Sandhills Nest-site Reference 13.39 ± 5.12 11.43 ± 5.57 5.29 ± 2.45 4.08 ± 2.88 1.11 ± 0.62 1.12 ± 1.00 0.52 ± 0.77 0.63 ± 0.83 1.48 ± 0.75 1.18 ± 0.85 0.18 ± 0.17 0.18 ± 0.20 0.12 ± 0.20 0.18 ± 0.35 0.39 ± 0.11 0.39 ± 0.17 0.37 ± 0.15 0.43 ± 0.26 0.18 ± 0.15 0.19 ± 0.15 U 1539.5 1922.0 3429.0*** 2167.0 1775.0 3802.0*** 2106.0 1724.0 842.0*** 2056.0*** 24 Table 3. Mixed-effects logistic regression models of nest-site selection by Bachman’s sparrows in the Coastal Plain and Sandhills physiographic regions of North Carolina, USA, 2014-2015. Paired plot ID was used as a random effect in all models. Models were compared using Akaike’s Information Criterion corrected for small sample size (AICc). Model Coastal Plain VerWD + VerGR + GrdFB VerWD + VerGR VerWD Null Sandhills MaxWD + PineBA + VerGR + VerGR2 PineBA + VerGR + VerGR2 MaxWD + VerGR + VerGR2 MaxWD + PineBA + VerGR VerGR + VerGR2 PineBA + VerGR MaxWD + VerGr VerGR Null K Loglik AICc ∆AICc wi 5 4 3 2 -54.22 -56.04 -59.34 -63.77 119.14 120.54 124.95 131.67 0.00 1.40 5.81 12.54 0.64 0.32 0.04 0.00 6 5 5 5 4 4 4 3 2 -101.06 -102.88 -105.86 -106.38 -107.81 -108.25 -110.14 -112.19 -116.45 214.63 216.13 222.08 223.13 223.86 224.75 228.53 230.52 236.97 0.00 1.50 7.45 8.50 9.23 10.11 13.89 15.89 22.34 0.65 0.31 0.02 0.01 0.01 0.00 0.00 0.00 0.00 25 Table 4. Parameter estimates for the top nest-site selection models for the Coastal Plain and Sandhills physiographic regions of North Carolina, USA, 2014-2015. Parameter Coastal Plain VerWD VerGR GrdFB Sandhills MaxWD PineBA VerGR VerGR2 β SE 95% CI P 0.55 -0.14 3.68 0.20 0.06 1.20 (0.17, 0.96) (-0.27, -0.04) (-0.10, 7.95) 0.006 0.012 0.066 -1.68 0.10 0.91 -0.06 0.90 0.03 0.24 0.02 (-3.52, 0.04) (0.04, 0.17) (0.47, 1.41) (-0.10, -0.02) 0.063 0.003 <0.001 0.002 26 Table 5. Bachman’s sparrow nest fates by physiographic region and study site (n = total nests found, A = abandoned, S = successful, F = failed, U = unknown fate, P = parasitized, B = burned by prescribed fire or wildfire, and T = total nests used for data summaries and analyses). Region Site n A S F U P B T Coastal Plain COPO 1 0 1 0 0 0 0 1 Coastal Plain GSPR 3 0 3 0 0 0 0 3 Coastal Plain HSGL 42 2 25 14 0 5 1 39 Coastal Plain SCPR 1 0 1 0 0 0 0 1 Sandhills FTBR 85 7 32 38 3 3 1 70a a 4 nests were found on their fledge date (exposure = 0), and were excluded from the nest survival analysis 27 Table 6. Bachman’s sparrow nest survival models for the Coastal Plain and Sandhills physiographic regions of North Carolina, USA, 2014-2015. Models were compared using Akaike’s Information Criterion corrected for small sample size (AICc). Model Coastal Plain Date Date + Date2 Habitat Null Sandhills Stage Null Year Habitat MaxWD VerWD VerFB HorWD VerGR + VerGR2 Date VerGR Fire MaxGR HorFB MaxFB Date + Date2 K Loglik AICc ∆AICc wi 2 3 2 1 -42.78 -42.69 -43.85 -45.38 89.67 91.60 91.81 92.80 0.00 1.93 2.14 3.13 0.52 0.20 0.18 0.11 2 1 2 2 2 2 2 2 3 2 2 2 2 2 2 3 -92.87 -94.59 -93.76 -94.10 -94.18 -94.22 -94.28 -94.42 -93.40 -94.43 -94.44 -94.50 -94.54 -94.54 -94.57 -93.84 189.79 191.20 191.59 192.26 192.43 192.50 192.63 192.91 192.92 192.93 192.94 193.06 193.14 193.14 193.20 193.81 0.00 1.41 1.80 2.47 2.64 2.71 2.83 3.12 3.13 3.13 3.14 3.26 3.35 3.35 3.41 4.02 0.21 0.11 0.09 0.06 0.06 0.06 0.05 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.03 28 Figure 1. Locations of study sites and physiographic regions where nest-site selection and nest survival data were collected during 2014-2015. Physiographic region designation follows level III classifications defined by Bailey (1995). The Sandhills is a distinct level IV subregion of the Southeastern Plains. Study sites are indicated with four-letter abbreviation; from west to east, Fort Bragg Military Installation (FTBR; n = 85 nests), Green Swamp Savanna Preserve (GSPR; n = 3 nests), Compass Pointe (COPO; n = 1 nest), Holly Shelter Game Land (HSGL; n = 42 nests), and Shaken Creek Savanna Preserve (SCPR; n = 1 nest). 29 Figure 2. Predicted probability of selection of a nesting site for Bachman’s sparrows by physiographic region in North Carolina, USA, in 2014–2015. ‘Hits’ refers to the total number of vegetation contacts on the entire pole averaged across a plot. Shaded regions indicate 95% confidence intervals. 30 Figure 3. Predicted probability of Bachman’s sparrow daily nest survival rate in relation to Julian date for the Coastal Plain of North Carolina, USA, in 2014–2015. Shaded region indicates 95% confidence interval. 31 CHAPTER 2 SPACE-USE AND HABITAT SELECTION BY BACHMAN’S SPARROWS AT THE NORTHERN RANGE PERIPHERY ABSTRACT Bachman’s sparrow (Peucaea aestivalis) populations have declined range-wide during the past 40 years. Peripheral populations show some of the worst declines, and the species is now rare or absent over the northern extent of its range. Although home range size and habitat selection have been quantified in the sparrow’s core range, northern populations have been largely unstudied. To better describe space-use and habitat relationships in this region of sparrow decline, we quantified microhabitat selection and the influence of body condition and reproductive status on Bachman’s sparrow home range size. We captured and radio-tracked 37 male Bachman’s sparrows between April and July 2014-2015 in the Coastal Plain of North Carolina, USA. Within home ranges, Bachman’s sparrows used locations with greater density of woody vegetation and forbs, and intermediate levels of grass density. Home range size varied from 2.6 ha to 18 ha, but we documented no significant predictors of variation in home range size. Because home range size was variable and not linked with reproductive status, it should not be used as an indication of habitat quality without more detailed studies on the factors that affect space-use in this species. 32 INTRODUCTION Knowledge of habitat selection and its influence on home range size are important for understanding a species’ ecology and natural history, and for implementing better conservation and management actions (Anich et al. 2012). Landscapes with more abundant resources are expected to contain individuals with smaller, high quality home ranges and support larger populations (Anich et al. 2010, Tingley et al. 2014). Moreover, habitat selection may vary across a species’ range (e.g., Zhu et al. 2012, Tack and Fedy 2015), and space-use may change in accordingly. In particular, populations at the edge of a species’ range experience strong limiting factors in the form of atypical habitat types or reduced habitat quality (Sexton et al. 2009). Yet, peripheral populations are important for species’ survival and evolution in the context of increasing environmental change, and are often of high conservation value (Lesica and Allendorf 1995). Thus, understanding geographicallyspecific estimates of habitat use and home range size could identify resources limiting species of management concern and guide more successful management and restoration activities (van Beest et al. 2011, Tingley et al. 2014). One such species of concern is the Bachman’s sparrow (Peucaea aestivalis), a ground-nesting bird that is strongly associated with fire-maintained longleaf pine (Pinus palustris) ecosystems in the southeastern United States. Extensive study has shown that Bachman’s sparrows select specific vegetation conditions that reflect a frequent fire regime (≤ 3 years), in particular the presence of open canopy cover, and dense ground cover of grasses and forbs (Dunning 2006, Tucker et al. 2006). Due to extensive habitat loss and firesuppression, the longleaf pine ecosystem has been reduced to ~3% of its historic range (Frost 2006), and ranks among the most endangered ecosystems in North America (Noss et al. 33 1995). Consequently, Bachman’s sparrow populations have steadily declined in recent decades, particularly along the periphery of the species’ range (Dunning and Watts 1990, Sauer et al. 2014). Given their status as an indicator species for functional longleaf pine forests (Taillie et al. 2015), Bachman’s sparrows have become the focus of numerous studies in recent years. Extensive research has documented Bachman’s sparrow-habitat relationships (Plentovich et al. 1998, Tucker et al. 2004, Cox and Jones 2007, Brooks and Stouffer 2010); yet, only a few studies have been conducted outside of the species’ southern range extent (e.g., Dunning and Watts 1990, Haggerty 1998, Taillie et al. 2015). Although prior research has been valuable for describing second-order habitat selection, only a handful of studies have examined habitat features at the home range-level (Haggerty 1998, Jones 2008, Cox and Jones 2007, Brown 2012). For Bachman’s sparrow, no studies have examined habitat selection at finer spatialscales (i.e., used locations within the home range) to quantify microhabitat characteristics, which likely are more informative than coarser home-range-level selection (Anich et al. 2012). Habitat selection is likely to influence Bachman’s sparrow home range size, with smaller home ranges in areas with better habitat quality (Stober and Krementz 2006). In the species’ core range, home ranges average 3-5 ha (Stober and Krementz 2006, Cox and Jones 2007, Brown 2012). However, home range size appears to vary considerably depending on pairing status (Jones 2008), vegetation characteristics (Haggerty 1998), habitat type (Stober and Krementz 2006), weather conditions (Brown 2012), and study methodology. Most studies of Bachman’s sparrow home range size have relied on spot-mapping singing perches of unmarked (McKitrick 1979, Meanley 1990) or color-banded males (Haggerty 1998, Cox 34 and Jones 2007), and these estimates can be substantially lower than home range estimates from radio-tagged birds (Anich et al. 2009, Streby et al. 2012). Further, Bachman’s sparrow home-range size has never been empirically estimated in North Carolina. Given recent extirpations (e.g., Virginia) and the increasing rarity of Bachman’s sparrow populations at the species’ shrinking range limits (Dunning 2006), we addressed information needs for Bachman’s sparrow conservation and longleaf pine forest restoration and management at the northern periphery of the species’ range in North Carolina. Our primary objectives included 1) quantifying microhabitat selection within Bachman’s sparrow home ranges, 2) estimating home-range sizes, 3) determining which factors influence variation in home-range size, and 4) comparing our results to those of previous studies conducted in the core of the species’ range. METHODS Study area We studied Bachman’s sparrows at four sites within the Middle Atlantic Coastal Plain (hereafter, Coastal Plain) physiographic region of southeastern North Carolina, USA (Figure 1) from 2014-2015. Bachman’s sparrows are uncommon across most of the region, so we relied on Bachman’s sparrow presence and accessibility to locate study sites. We selected Boiling Spring Lakes Preserve, Green Swamp Preserve, Holly Shelter Game Land, and Sleepy Creek Farms as study sites (Figure 1). Study sites in the Coastal Plain were imbedded in a matrix of urban and rural residential development and agricultural lands, and were managed by a variety of state and private landowners. The Coastal Plain recently has been recognized as a global biodiversity hotspot, and longleaf pine communities within this region have among the highest levels of plant richness 35 per square meter of any ecosystem in North America (Noss et al. 2015). Study sites were primarily composed of mesic longleaf pine forests bordered by pocosin wetlands. Wet pine savannas in this region were characterized by an open tree canopy, seasonally saturated soils, and a herbaceous layer dominated by graminoids, forbs, and low shrubs. In addition to longleaf pine, tree species dominating Bachman’s sparrow habitat included loblolly (P. taeda) and pond pine (P. serotina). Dominant groundcover species included wiregrass (Aristida stricta), switchcane (Arundinarea gigantea), bracken fern (Pteridium aquilinum) and inkberry (Ilex glabra). Field data collection Capture and radio-telemetry We documented Bachman’s sparrow home range and microhabitat selection by fitting VHF radio-transmitters to adult males. We initially targeted Bachman’s sparrows for capture by playing audio recordings of territorial calls at potential study sites. Once located, we captured sparrows with audio playback of conspecific vocalizations and mist nets. Upon capture, we measured wing length and mass, and attached a small transmitter (~0.64 g, Blackburn Transmitters, Nacogdoches, Texas, USA) using an elastic leg loop harness (Rappole and Tipton 1991, Streby et al. 2015). Bachman’s sparrows captured during the course of this study weighed 18.78 ± 0.636 g (mean ± SD), and transmitters weighed ~3.5% of an average adult bird’s mass. Additionally, we fit all birds with a unique combination of colored plastic leg bands and a uniquely numbered U.S. Geological Survey aluminum leg band. We collected home range data on Bachman’s sparrows between April and July, 20142015. We located sparrows 1 - 2 times daily for 5 - 6 days per week to ensure ≥ 30 telemetry 36 locations necessary to estimate home range size before transmitters failed (Seaman et al. 1999). All telemetry locations were gathered ≥ 2 h apart, which ensured independence between successive locations because sparrows could move across the largest home range during this time interval. We located sparrows via homing and typically confirmed their locations visually or aurally. When birds did not sing or were visually obscured by dense groundcover, the strength of the telemetry signal allowed us to determine the bird’s approximate location (i.e., within 5 m). When sparrows were located, we recorded spatial coordinates (Universal Transverse Mercator Zone 17N or 18N) using a hand-held GPS unit and marked the location with a labeled flag for future vegetation sampling. Reproductive status We were interested in whether reproductive status influenced home range size. To determine reproductive status of each radio-tagged sparrow, we collected behavioral information at one-week intervals using a protocol modified from Vickery et al. (1992) and Tucker et al. (2006). During each visit, one observer spent a minimum of 60 minutes on a birds’ territory and determined the reproductive status of the male by noting behaviors such as associating with a female, feeding nestlings, or feeding fledglings. Bachman’s sparrows can produce 3 broods (Dunning 2006), and we recorded the number of nesting attempts or successful broods. Also, we located nests during visits and tracking activities, and determined nest fates using established protocols (Martin and Geupel 1993). After the completion of radio-tracking, males were assigned a reproductive score based on weekly observations and the fates of nests (Appendix A). Vegetation measurements 37 We measured vegetation structure within a 5-m radius plot centered on a subset of 10 randomly-selected telemetry locations prior to the conclusion of each season (24 ± 16 days after completion of radio-tracking each male). Vegetation categories included grass, woody vine-shrub (hereafter woody vegetation), switchcane, and forb-fern (Taillie et al. 2015). We quantified vegetation structure using indices of density by recording the presence of each vegetation category (i.e., ‘hits’ or vegetation contacts) on 0.1-m sections of a 1.5-m vertical pole. Vertical density included vegetation hits along the entire length of the pole, while ground density was calculated as the number of hits on the first 0.1-m section of the pole (Wiens and Rotenberry 1981). Maximum vegetation height was obtained by recording the tallest hit on the pole and rounding to the nearest 0.1-m section (Wiens and Rotenberry 1981, Moorman and Guynn 2001, Taillie et al. 2015). Measurements were taken at every meter along four 5-m transects running in each cardinal direction. At the same 1-m increments along each transect, we estimated canopy closure by recording the presence or absence of live canopy using an ocular sighting tube. We used a 10-factor prism from the plot center to determine the basal area (m2/ha) of pines surrounding each used location. We repeated these same measurements at a reference plot (i.e., an available non-use location) 50 m away from each telemetry location in a random direction. We chose this distance to eliminate overlap between used and available points, and to constrain reference locations within a bird’s home range. Statistical analyses Estimating home range size All statistical analyses were performed in R 3.2.2 (R Development Core Team 2015). We estimated home-range size using kernel density techniques (Worton 1989) and the ‘ks’ 38 package (Duong 2007), which assumes a bivariate normal density fixed kernel. We used the ‘plug-in’ method for calculating the bandwidth parameter (Millspaugh et al. 2006). The plugin method generally outperforms other bandwidth estimators (e.g., reference and least-square cross validation; Gitzen et al. 2006), and has been increasingly used in avian telemetry studies (e.g., Rota et al. 2014, Stanton et al. 2014, Lorenz et al. 2015). Home range size estimates were based on 95% utilization distribution contours. Modeling microhabitat selection Microhabitat selection was analyzed with the lme4 package (Bates et al. 2015) and generalized linear mixed effects models specifying a logit link function and binomial response (where 1 = used location, 0 = reference location). Prior to analysis, we tested for correlations (r > |0.60|) between all pairwise combinations of variables. Among 17 potential vegetation variables, maximum height measurements were correlated with vertical density measurements, and we omitted all maximum height variables in final models. Grass vertical density was highly correlated with grass groundcover density, so we omitted the latter variable. Also, we included a quadratic effect of vertical grass density because dense grass may restrict movement on the ground in the absence of fire (Brooks and Stouffer 2010, Taillie et al. 2015). For all models, we included ‘BirdID’ as a random effect to account for repeated measures of each individual and potential variation in error associated with tracking different sparrows. We examined models with BirdID nested within a random ‘Site’ effect, but determined that models with BirdID performed better. To explore the most important vegetation variables, we looked at all possible linear combinations of variables. Home range variation 39 Variation in home range size was investigated using linear models. Home-range size was log-transformed prior to analyses to meet assumptions for statistical analysis. For the home range size analysis, we selected variables based on published literature and our observations of Bachman’s sparrow space-use, along with habitat, temporal, and individual characteristics important in space-use for other bird species (e.g., Mazerolle and Hobson 2004, Anich et al. 2010, Lorenz et al. 2015). Following Anich et al. (2010), we used 2 model sets examining the effects of 1) habitat and 2) non-habitat predictors on home range size. Due to limited sample size, we tested univariate models in each model set. For the habitat model set, we tested the same vegetation variables used in the microhabitat selection analysis, and time since fire. The non-habitat model set included 5 predictors: body condition (wing length), reproductive status (i.e., reproductive index score), mean tracking date, site, and year. For vegetation variables, Bachman’s sparrow locations and the associated reference locations were combined into a single mean value for each individual to characterize vegetation within a home range. Model selection Habitat selection and home range variation models were selected using Akaike’s Information Criterion (AICc; Burnham and Anderson 2002) using the ‘MuMIn’ package (Barton 2016). Models were ranked according to their AICc values, and model averaging was performed across the top candidate models (∆AICc ≤ 2) when appropriate. A parameter was considered as strongly affecting microhabitat selection if the 95% CI did not include 0, and was considered to have no relationship if the 95% CI overlapped 0. 40 RESULTS Home range characteristics We collected 1199 locations on 37 male Bachman’s sparrows from 2014-2015. We obtained ≥ 30 telemetry locations for 27 sparrows for an average of 36 ± 5 points (range: 30 – 46). For the 27 birds, there was no statistically significant correlation between home-range size and the number of location points (r = 0.36, P = 0.06). Mean 95% kernel home range size was 7.92 ± 4.05 ha, and varied from 2.59 – 18.18 ha. Microhabitat selection We collected vegetation data for 37 sparrows between 2014 and 2015 (Table 1). The top models showed strong support for an influence of woody vegetation, grass, and forbs on microhabitat selection. Specifically, the top models (∆AICc ≤ 2) included the effect of woody vertical density, a quadratic effect of grass vertical density, and groundcover forb density (Table 2). Also, the confidence limits of the model-averaged parameter estimates for these terms did not overlap zero (Table 3). Four other terms appeared in our top models, but confidence limits of the model-averaged parameter estimates overlapped zero (pine basal area, groundcover density of woody vegetation, and both density measurements of dead vegetation; Table 3). Of the three significant predictors, vertical woody density had the strongest relationship with microhabitat selection. The relationship between woody plants and forbs and microhabitat selection was positive, with the probability of use increasing with an increase in the density of woody vegetation and forbs (Figure 2). Also, selection probability increased with increasing grass density, but there was a threshold beyond which relative probability of use began to decrease (Figure 2). 41 Home range variation Home range size predictors were poorly supported in model sets, with all models ranking lower than the null model (Table 4). In the habitat model set, models containing vertical density of dead vegetation, time since fire, and vertical and groundcover density of forbs were competitive with the null model (Table 4). However, confidence intervals for these parameters overlapped zero and were not considered significant. Similarly, reproductive index score and mean tracking date were among the top predictors in the nonhabitat model set, but these variables had weak effects on home range variation (Table 4). DISCUSSION Bachman’s sparrow showed consistent selection for several vegetation characteristics within home ranges. Although the importance of high grass density often has been emphasized (Dunning and Watts 1990, Plentovich et al. 1998) when describing Bachman’s sparrow habitat, our results are consistent with more recent studies documenting threshold responses to grass density (Brooks and Stouffer 2010, Taillie et al. 2015). This threshold for grass density suggests that Bachman’s sparrows select patchy grass cover, possibly because too much herbaceous cover can hamper foraging efficiency (Haggerty 1998). In the absence of frequent fire, the formation of dense grass may restrict movement of Bachman’s sparrows, rendering locations within home ranges unsuitable. Sparrows selected microhabitat with more forb groundcover. Bachman’s sparrows primarily glean prey from the ground (Dunning 2006), and forbs may represent a valuable foraging substrate and could reflect the availability of arthropods as a prey resource. However, the forb species most important as hosts for arthropods consumed by Bachman’s 42 sparrows are not known. Additionally, forbs may further facilitate open ground for movement. Contrary to previous studies, greater woody vegetation density was an important variable for microhabitat selection. Woody vegetation, particularly tall shrubs, has been considered to be detrimental to Bachman’s sparrow occupancy (Brooks and Stouffer 2010, Taillie et al. 2015). Indeed, as more time elapses without fire, fire intolerant woody species become taller and often shade herbaceous vegetation (Engstrom et al. 1984, Glitzenstein et al. 2003). However, in frequently burned longleaf pine forest, woody vegetation at the homerange level likely provides several important benefits. Numerous studies have documented the importance of song-perch structures for Bachman’s sparrows (Meanley 1959, Dunning and Watts 1990, Jones et al. 2013), and we frequently observed males singing from horizontal and terminal branches of woody shrubs. Also, woody shrubs may provide escape cover from predators (Pulliam and Mills 1977, Dunning 2006), and constitute important features for Bachman’s sparrow nest-site selection (Jones et al. 2013, Winiarski 2016). Interestingly, we observed male Bachman’s sparrows frequently using the edges of densely vegetated swales that had wetter soils than surrounding longleaf pine forest, as well as ecotones representing the gradient between longleaf pine woodlands and pocosins. In these areas, forbs and woody vegetation dominated, and grass density was low. Thus, the microhabitat relationships we observed in this study may reflect the high use of these areas within home ranges. Although this habitat association has not been reported elsewhere in the species’ range, it appears to be consistent with habitat use exhibited by Bachman’s sparrows in the Sandhills physiographic region of central North Carolina (Alex Fish, personal 43 observation), suggesting that habitat use may differ for the species at the northern limits of its range. Home range sizes in this study were larger than most previously reported estimates for Bachman’s sparrows. Home-range estimates across the species’ range have varied substantially, perhaps due to differences in habitat quality, home range estimation techniques, and methods of gathering location points (Table 5). Several early studies relied on spotmapping male locations and singing perches (McKitrick 1979, Haggerty 1998, Cox and Jones 2007). However, home ranges are generally larger than song territories, and standard observational methods can underestimate home range size compared to telemetry-based methods (Anich et al. 2009, Streby et al. 2012). For example, Golden-winged Warbler (Vermivora chrysoptera) home ranges estimated using radio-telemetry were 3 times larger than those estimated from spot-mapping (Streby et a. 2012). Our estimated home range sizes for some individuals were much larger than previously had been documented for Bachman’s sparrows, which suggests that spot-mapping may have underestimated home range size (but see Stober and Krementz 2006). Elsewhere, the 95 or 100% MCP was used in 4 of 6 published studies, despite considerable biases associated with this estimator (Börger et al. 2006). In addition to the work presented here, only one other study has implemented both radio-telemetry and fixed kernel estimators to determine home range size for Bachman’s sparrows (Brown 2012). Mean home range size (7.92 ± 4.05 ha) presented here was ~2.1 times larger than an estimate provided by Brown (2012), but ~1.2 times smaller compared with home range size during a severe drought year in the same study (Table 5). Further research, particularly radio-tracking studies using kernel density estimators, are needed to 44 obtain more robust estimates of Bachman’s sparrow home range size and to compare spaceuse across studies. Despite clear selection for vegetation features within home ranges, home range size was variable (2.59 – 18.18 ha) and not associated with any measured habitat or individual characteristics. Little is known about factors influencing variation in Bachman’s sparrow home-range size, but Stober and Krementz (2006) reported that home ranges were smaller in young pine stands (≤ 4 years old) than mature stands. In pine plantations in Arkansas, Bachman’s sparrow home range size was negatively related to percent forb cover, ground cover, forb height, percent grass cover, and density of vegetation between 0 - 90 cm (Haggerty 1998). Paired males were shown to have larger home ranges than unpaired males in Georgia, perhaps due to long-distance movements to locate and deliver food to nestlings and young (Jones 2008). Jones (2008) similarly showed that increases in forb cover and time since burn were positively related to the home-range size of male Bachman’s sparrows, although parameter estimates indicated weak effects of these relationships. Severe drought may cause significant increases in daily movements and home range size due to decreased habitat quality (Brown 2012), but drought conditions did not occur over the course of this study. Home range size was not correlated with pairing success or fledging young (i.e., reproductive index scores), which are important indicators of habitat quality. However, the correlation between home range size and years since fire was positive, indicating that habitat quality may decline as time since fire elapses, but parameter estimates were not significant. Most (90%) males in this study occurred in stands which were burned ≤ 3 years prior, which is considered to be an optimal fire-return interval to promote Bachman’s sparrow abundance 45 and productivity (Tucker et al. 2004, 2006). Consistent with male home ranges in clearcut pine stands (Haggerty 1998), home range size decreased with greater forb cover, though this effect was weak. In contrast, forb cover was positively associated with Bachman’s sparrow home range size and breeding success in Georgia (Jones 2008). Successful males in that study had large home ranges, and often occurred in upland areas where forb cover was lower. The effect of forbs on Bachman’s sparrow home range size remains unclear, but likely warrants further study. Other factors, such as relationships to neighbors or mates, may influence space-use (Anich et al. 2010), but we lacked sufficient data to evaluate the influence of neighboring individuals. Body size and age can influence territory sizes (i.e., the ideal dominance model of habitat selection; Adams 2001), with larger, older individuals typically holding smaller, high quality home ranges. While we accounted for body size (i.e., wing length) when modeling variability in home range size, we were not able to determine precise estimates of age because Bachman’s sparrow adults can only be categorized as afterhatch-year (Pyle 1997). Bachman’s sparrows have been declining in abundance, with the greatest declines at the periphery of their range (Sauer et al. 2014). In North Carolina, we measured several structural vegetation characteristics associated with both sparrow use and prescribed fire at the home range level. In particular, the microhabitat characteristics we documented support the value of variable-intensity prescribed fire that maintains small-statured shrubs, high forb cover, and patchy grass cover (Glitzenstein et al. 2003, Thaxton and Platt 2006). Further study is needed to determine whether these habitat associations have a biologically significant link to Bachman’s sparrow vital rates. Also, we provide the first estimate of home range size for Bachman’s sparrow at the northern limits of its range, but factors that influence 46 home range size in this species are not yet clear. The idea that home range size varies with habitat quality may not be an appropriate assumption for this species, and we believe that further studies are needed to understand factors that influence space-use in Bachman’s sparrows. 47 LITERATURE CITED Adams, E.S. (2001). Approaches to the study of territory size and shape. Annual Review of Ecology, Evolution, and Systematics 32: 277–303. Anich, N. M., T. J. Benson, and J. C. Bednarz (2009). Estimating territory and home-range sizes: do singing locations alone provide an accurate estimate of space use? The Auk 126:626-634. Anich, N. M., T. J. Benson, and J. C. Bednarz (2010). Factors influencing home-range size of Swainson’s warblers in eastern Arkansas. The Condor 112:149–158. Anich, N. M., T. J. Benson, and J. C. Bednarz (2012). What factors explain differential use within Swainson's Warbler (Limnothlypis swainsonii) home ranges? The Auk 129:409-418. Barton, K. (2016). MuMIn: Multi-model inference R package 1.9.13. http:// CRANRprojectorg/package.MuMIn Bates, D., M. Maechler, B. Bolker, and S. Walker (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67:1–48. doi:10.18637/jss.v067.i01 Börger, L., N. Franconi, G. De Michele, A. Gantz, F. Meschi, A. Manica, S. Lovari, and T. Coulson (2006). Effects of sampling regime on the mean and variance of home range size estimates. Journal of Animal Ecology 75:1393–1405. Brooks, M. E., and P. C. Stouffer. (2010). Effects of Hurricane Katrina and salvage logging on Bachman's Sparrow. The Condor 112:744-753. Brown, S. K. (2012). Movements, home range and habitat selection of Bachman’s Sparrows (Peucaea aestivalis) on a longleaf sandhill forest: Implications for fire management. M.S. thesis, University of Georgia, Athens, GA, USA. Burnham, K. P., and D. R. Anderson (2002). Model selection and multimodel inference: a practical information-theoretic approach, 2nd ed. Springer-Verlag, New York. Cox, J. A., and C. D. Jones (2007). Home range and survival characteristics of male Bachman’s Sparrows in an old-growth forest managed with breeding season burns. Journal of Field Ornithology 78:263-269. Dunning Jr, J. B., and B. D. Watts (1990). Regional differences in habitat occupancy by Bachman's Sparrow. The Auk 107:463-472. Dunning, J. B. 2006. Bachman’s sparrow (Peucaea aestivalis). The Birds of North America. Number 038. 48 Duong, T. (2007). ks: kernel density estimation and kernel discriminant analysis for multivariate data in R. Journal of Statistical Software 21:1–16. Engstrom, R. T., R. L. Crawford, and W. W. Baker (1984). Breeding bird populations in relation to changing forest structure following fire exclusion: a 15-year study. Wilson Bulletin 96:437-450. Frost, C. (2006). History and future of the longleaf pine ecosystem. In The Longleaf Pine Ecosystem: Ecology, Silviculture, and Restoration (S. Jose, E. J. Jokela, and D. L. Miller, Editors). Springer, New York, NY, USA. pp. 9–48. Gitzen, R. A., J. J. Millspaugh, and B. J. Kernohan (2006). Bandwidth selection for fixedkernel analysis of animal utilization distributions. Journal of Wildlife Management 70:13341344. Glitzenstein, J. S., D. R. Streng, and D. D. Wade (2003). Fire frequency effects on longleaf pine (Pinus palustris P. Miller) vegetation in South Carolina and Northeast Florida, USA. Natural Areas Journal 23:22–37. Haggerty, T. M. (1998). Vegetation structure of Bachman’s Sparrow breeding habitat and its relationship to home range. Journal of Field Ornithology 69:45–50. Jones, C. D. (2008). Response of Bachman’s Sparrow (Aimophila aestivalis) to growingseason prescribed fires in longleaf pine savannas of southern Georgia. Master’s thesis, University of Georgia, Athens, GA, USA. Jones, C. D., J. A. Cox, E. Toriani-moura, J. Robert, and R. J. Cooper (2013). Nest-site characteristics of Bachman’s sparrows and their relationship to plant succession following prescribed burns. Wilson Journal of Ornithology 125:293–300. Lesica, P., and F. W. Allendorf (1995). When are peripheral populations valuable for conservation? Conservation Biology 9:753–760. Lorenz, T. J., K. T. Vierling, J. M. Kozma, J. E. Millard, and M. G. Raphael (2015). Space use by white-headed woodpeckers and selection for recent forest disturbances. Journal of Wildlife Management. 79:1286-1297. Martin, T. E., and G. R. Geupel (1993). Nest-monitoring plots: methods for locating nests and monitoring nest success. Journal of Field Ornithology 64:507–519. Mazerolle, D. F., and K. A. Hobson (2004). Territory size and overlap in male Ovenbirds: contrasting a fragmented and contiguous boreal forest. Canadian Journal of Zoology 82:1774-1781. 49 McKitrick, M. C. (1979). Territory size and density of Bachman’s Sparrow in south central Florida. Florida Field Naturalist 7:33–34. Meanley, B. (1959). Notes on Bachman’s Sparrow in central Louisiana. Auk 76:232–234. Meanley, B. (1990). Some observations of the singing behavior of Bachman’s Sparrow. Chat 54:63. Millspaugh, J. J., R. M. Nielson, L. McDonald, J. M. Marzluff, R. A. Gitzen, C. D. Rittenhouse, M. W. Hubbard, and S. L. Sheriff (2006). Analysis of resource selection using utilization distributions. Journal of Wildlife Management 70:384–395. Moorman, C. E., and D. C. Guynn Jr. (2001). Effects of group-selection opening size on breeding bird habitat use in a bottomland forest. Ecological Applications 11:1680–1691. Noss, R. F., E. T. LaRoe, III, and J. M. Scott (1995). Endangered ecosystems of the United States: a preliminary assessment of loss and degradation. U.S. Department of Interior National Biological Service, Biological Report 28, Washington, D.C., USA. Noss, R. F., W. J. Platt, B. A. Sorrie, A. S. Weakley, D. B. Means, J. Costanza, J., and R. K. Peet (2015). How global biodiversity hotspots may go unrecognized: lessons from the North American Coastal Plain. Diversity and Distributions 21:236-244. Plentovich, S., J. W. Tucker Jr., N. R. Holler, and G. E. Hill (1998). Enhancing Bachman's Sparrow habitat via management of Red-cockaded Woodpeckers. Journal of Wildlife Management 62:347-354. Pulliam, H. R., and G. S. Mills (1977). The use of space by wintering sparrows. Ecology 58:1393-1399. Pyle, P. (1997). Identification Guide to North American Birds, Part I: Columbidae to Ploceidae. Slate Creek Press, Bolinas, CA, USA. R Core Team (2013). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.r-project.org/. Rappole, J. H., and A. R. Tipton (1991). New harness design for attachment of radio transmitters to small passerines. Journal of Field Ornithology 62:335-337. Rota, C. T., M. A. Rumble, J. J. Millspaugh, C. P. Lehman, and D. C. Kesler. (2014). Spaceuse and habitat associations of Black-backed Woodpeckers (Picoides arcticus) occupying recently disturbed forests in the Black Hills, South Dakota. Forest Ecology and Management 313:161-168. 50 Sauer, J. R., J. E. Hines, J. E. Fallon, K. L. Pardieck, D. J. Ziolkowski, Jr., and W. A. Link (2014). The North American Breeding Bird Survey, Results and Analysis 1966 - 2013. Version 01.30.2015. USGS Patuxent Wildlife Research Center, Laurel, MD, USA. http://www.mbr-pwrc.usgs.gov/bbs/bbs2013.html Seaman, D. E., 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 63:739-747. Sexton, J. P., P. J. McIntyre, A. L. Angert, and K. J. Rice (2009). Evolution and ecology of species range limits. Annual Review of Ecology, Evolution, and Systematics 40:415–436. Stanton Jr., R. A., D. C. Kesler, and F. R. Thompson III (2014). Resource configuration and abundance affect space use of a cooperatively breeding resident bird. The Auk 131:407-420. Stober, J. M., and D. G. Krementz (2006). Variation in Bachman’s sparrow home-range size at the Savannah River site, South Carolina. The Wilson Journal of Ornithology 118:138-144. Streby, H.M., J. P. Loegering, and D. E. Andersen (2012). Spot-mapping underestimates song-territory size and use of mature forest by breeding golden-winged warblers in Minnesota, USA. Wildlife Society Bulletin 36:40-46. Streby, H. M., T. L. McAllister, S. M. Peterson, G. R. Kramer, J. A. Lehman, and D. E. Andersen (2015). Minimizing marker mass and handling time when attaching radiotransmitters and geolocators to small songbirds. The Condor 117:249-255. Tack, J. D., and B. C. Fedy (2015). Landscapes for energy and wildlife: conservation prioritization for golden eagles across large spatial scales. PloS One 10:e0134781. Taillie, P. J., M. N. Peterson, and C. E. Moorman (2015). The relative importance of multiscale factors in the distribution of Bachman's Sparrow and the implications for ecosystem conservation. The Condor 117:137-146. Thaxton, J. M., and W. J. Platt (2006). Small-scale fuel variation alters fire intensity and shrub abundance in a pine savanna. Ecology 87:1331-1337. Tingley, M. W., R. L. Wilkerson, M. L. Bond, C. A. Howell, and R. B. Siegel (2014). Variation in home-range size of Black-backed Woodpeckers. The Condor: Ornithological Applications 116:325–340. Tucker Jr., J. W., W. D. Robinson, and J. B. Grand (2004). Influence of fire on Bachman’s Sparrow, an endemic North American songbird. Journal of Wildlife Management 68:1114– 1123. 51 Tucker, J. W., W. D. Robinson, and J. B. Grand (2006). Breeding productivity of Bachman’s sparrows in fire-managed longleaf pine forests. The Wilson Journal of Ornithology 118:131– 137. van Beest, F. M., I. M. Rivrud, L. E. Loe, J. M. Milner, and A. Mysterud (2011). What determines variation in home range size across spatiotemporal scales in a large browsing herbivore? Journal of Animal Ecology 80:771–785. Vickery, P. D., M. L. Hunter Jr., and J. V. Wells (1992). Use of a new reproductive index to evaluate relationship between habitat quality and breeding success. The Auk 109:697-705. Wiens, J. A., and J. T. Rotenberry (1981). Habitat associations and community structure of birds in shrubsteppe environments. Ecological Monographs 51:21-41. Worton, B. J. (1989). Kernel methods for estimating the utilization distribution in homerange studies. Ecology 70:164-168. Zhu, X., D. S. Srivastava, J. N. M. Smith, and K. Martin (2012). Habitat selection and reproductive success of Lewis’s Woodpecker (Melanerpes lewis) at its northern limit. PLoS One 7:e44346. 52 Table 1. Summary statistics (mean ± SD) for used and reference plot vegetation variables for 37 radio-tracked Bachman’s sparrows in the Coastal Plain physiographic region in North Carolina, USA, 2014-2015. Variable Used Reference Canopy 35.76 ± 25.4 35.85 ± 25.49 PineBA 11.14 ± 6.62 10.85 ± 7.00 VerWD 1.40 ± 0.75 1.08 ± 0.65 VerGR 2.09 ± 1.17 2.39 ± 1.29 VerFB 0.32 ± 0.35 0.28 ± 0.36 VerSC 0.02 ± 0.09 0.02 ± 0.09 VerDD 0.11 ± 0.17 0.08 ± 0.11 GrdWD 0.49 ± 0.23 0.44 ± 0.24 GrdGR 0.51 ± 0.26 0.58 ± 0.27 GrdFB 0.07 ± 0.11 0.07 ± 0.10 GrdSC 0.00 ± 0.01 0.00 ± 0.01 GrdDD 0.01 ± 0.03 0.01 ± 0.03 MaxWD 0.21 ± 0.14 0.16 ± 0.11 MaxGR 0.28 ± 0.15 0.31 ± 0.16 MaxFB 0.07 ± 0.08 0.06 ± 0.08 MaxSC 0.01 ± 0.03 0.01 ± 0.03 MaxDD 0.04 ± 0.06 0.02 ± 0.04 Canopy = Canopy cover (%); PineBA = Pine basal area (m2/ha-1); VerWD = woody vertical density (hits); VerGR = grass vertical density (hits); VerFB = forb vertical density (hits); VerSC = switchcane vertical density (hits); VerDD = dead vertical density (hits); GrdWD = woody groundcover density (hits) GrdGR = grass groundcover density (hits); GrdFB = forb groundcover density (hits); GrdSC = switchcane groundcover density (hits); GrdDD = dead groundcover density (hits); MaxWD = woody max. height (m); MaxGR = grass max. height (m); MaxFB = forb groundcover density (hits); MaxSC = switchcane max. height (m); MaxDD = dead max. height (m). 53 Table 2. Top 10 mixed-effects logistic regression models of microhabitat selection by Bachman’s sparrows in the Coastal Plain region of North Carolina, USA, 2014-2015. BirdID was used as a random effect in all models. Models were compared using Akaike’s Information Criterion corrected for small sample size (AICc). Model GrdFB+VerDD+VerGR+VerGR2+VerWD GrdFB+VerGR+VerGR2+VerWD GrdFB+PineBA+VerDD+VerGR+VerGR2+VerWD GrdFB+VerDD+VerFB+VerGR+VerGR2+VerWD GrdDD+GrdFB+VerDD+VerGR+VerGR2+VerWD GrdFB+GrdWD+VerDD+VerGR+VerGR2+VerWD GrdFB+VerDD+VerGR+VerGR2+VerWD GrdFB+GrdSC+VerDD+VerGR+VerGR2+VerWD GrdDD+GrdFB+VerGR+VerGR2+VerWD GrdFB+GrdWD+VerGR+VerGR2+VerWD K 7 6 8 8 8 8 8 8 7 7 Loglik -481.09 -482.96 -480.92 -481.04 -481.06 -481.06 -481.06 -481.09 -482.55 -482.84 ∆AICc 0.00 1.69 1.70 1.94 1.98 1.98 1.98 2.04 2.91 3.50 wi 0.24 0.10 0.10 0.09 0.09 0.09 0.09 0.09 0.06 0.04 54 Table 3. Model-averaged parameter estimates for the top microhabitat selection models (∆AICc ≤ 2) for 37 radio-tracked male Bachman’s sparrows in the Coastal Plain region of North Carolina, USA, 2014-2015. Shaded cells indicate parameter estimates that do not overlap zero. Confidence limits Variable Parameter estimate a 2.5% 97.5% (Intercept) 0.186 -0.01 0.38 GrdFB 0.253 0.07 0.43 VerDD 0.146 -0.01 0.34 VerGR -0.139 -0.31 0.04 VerGR2 -0.186 -0.32 -0.06 VerWD 0.451 0.27 0.63 PineBA 0.006 -0.11 0.20 VerFB -0.003 -0.22 0.16 GrdDD 0.002 -0.14 0.18 GrdWD -0.003 -0.23 0.18 VerSC -0.002 -0.18 0.14 a Parameter estimates and 95% CIs were derived from standardized variables. 55 Table 4. Linear models examining home range variation for Bachman’s sparrows in the Coastal Plain region of North Carolina, USA, 2014-2015. Model sets tested habitat and nonhabitat predictors. Models were compared using Akaike’s Information Criterion corrected for small sample size (AICc). Model Habitat models Null Dead vertical density Time since fire Forb groundcover density Forb vertical density Woody groundcover density Switchcane vertical density Switchcane groundcover density Woody vertical density Grass vertical density Pine basal area Dead groundcover density Grass vertical density (quadratic) Non-habitat models Null Reproductive index score Mean tracking date Year Wing length Site K Loglik AICc ∆AICc wi 2 3 3 3 3 3 3 3 3 3 3 3 4 -20.59 -19.46 -20.05 -20.07 -20.10 -20.35 -20.44 -20.46 -20.49 -20.52 -20.55 -20.57 -20.06 45.68 45.96 47.14 47.19 47.24 47.74 47.92 47.96 48.02 48.08 48.14 48.19 49.95 0.00 0.28 1.46 1.51 1.56 2.07 2.24 2.28 2.35 2.40 2.46 2.51 4.27 0.18 0.16 0.09 0.08 0.08 0.06 0.06 0.06 0.06 0.05 0.05 0.05 0.02 2 3 3 3 3 4 -20.59 -20.20 -20.22 -20.35 -20.48 -20.58 45.68 47.44 47.48 47.75 48.00 50.98 0.00 1.76 1.80 2.07 2.32 5.30 0.39 0.16 0.16 0.14 0.12 0.03 56 Table 5. Summary of published studies of Bachman’s Sparrow home-range size in comparison to findings presented here. Study Habitat State Haggerty 1998 Plantation AR McKitrick 1979 Unknown FL Dean & Vickery 2003 a Dry prairie FL b e Stober & Krementz 2006 Mixed SC Cox & Jones 2007 Old-growth GA Cox & Jones 2007 Old-growth GA Brown 2012 c Sandhills FL c Brown 2012 Sandhills FL Brown 2012 c Sandhills FL d Brown 2012 Sandhills FL Brown 2012 d Sandhills FL Brown 2012 d Sandhills FL Present study Flatwoods NC Present study Flatwoods NC a Conducted during the non-breeding season b Sample includes male and female home ranges c First year of study (2010) d Second year of study (2011) e Mature longleaf pine and young pine plantations N 25 6 8 4 46 46 8 8 7 23 23 22 27 27 Method Mapping Mapping Telemetry Telemetry Mapping Mapping Telemetry Telemetry Telemetry Telemetry Telemetry Telemetry Telemetry Telemetry Estimator MCP MCP MCP MCP MCP Fixed kernel MCP Fixed kernel Convex hull MCP Fixed kernel CH MCP Fixed kernel Percentile 100th 100th 95th 95th 100th 95th 100th 95th 95th 100th 95th 95th 95th 95th Mean 2.50 5.10 0.65 2.95 1.80 3.10 4.10 3.80 2.90 33.00 9.60 5.20 3.47 7.92 SE 0.22 0.49 0.10 0.57 1.40 3.50 0.76 0.41 0.43 10.20 0.82 0.68 0.36 0.78 57 Figure 1. Locations of study sites and home ranges of 37 Bachman’s sparrows that were radio-tracked in North Carolina, USA, 2014-2015. 58 Figure 2. Predicted probability of microhabitat use by Bachman’s sparrows in the Coastal Plain physiographic region, North Carolina, USA, in 2014–2015, for woody vertical density, grass vertical density, and forb groundcover density. Shaded regions indicate 95% confidence intervals. 59 CHAPTER 3 REPRODUCTIVE CONSEQUENCES OF HABITAT FRAGMENTATION IN A DECLINING RESIDENT LONGLEAF PINE BIRD ABSTRACT The longleaf pine (Pinus palustris) ecosystem is among the most imperiled in North America, primarily as a result of land conversion and fire suppression. Of its original extent, less than 5% remains as scattered or degraded patches. Consequently, longleaf conservation efforts recently have expanded, but a better understanding of local- and landscape-scale effects on demographic responses for wildlife species associated with longleaf pine is needed to guide effective conservation strategies. To address this knowledge gap, we studied the breeding ecology of Bachman’s sparrows (Peucaea aestivalis), a species of conservation concern that is closely associated with longleaf pine forests in the southeastern United States. We investigated the effects of local- and landscape-scale factors on different components of reproductive success (i.e., pairing success and probability of fledging ≥ 1 offspring) for 96 male sparrows at 8 sites in the Coastal Plain of southeastern North Carolina. We used a ‘species-centered’ approach and species distribution modeling to quantify Bachman’s sparrow habitat amount across our study area. We then constructed mixed effects logistic regression models for each reproductive component as a function of male body size, territory-level vegetation structure and time since fire, and habitat amount in the surrounding landscape. Pairing success was lower in highly fragmented longleaf pine landscapes compared to more continuous landscapes. In contrast, habitat amount did not affect the probability of fledging offspring for paired males. Overall, our work indicates that low pairing success is inhibiting successful reproduction in highly fragmented longleaf pine 60 landscapes, perhaps due to disrupted dispersal. Low pairing success may be partly responsible for Bachman’s sparrow declines, but further research is needed to quantify the effects of habitat fragmentation on other vital rates (i.e., adult and juvenile survival) that may be more influential in population declines. Nevertheless, to promote breeding opportunities for Bachman’s sparrows and other longleaf-pine associated wildlife, managers should prioritize resources to patches near large, pre-existing longleaf pine forest. 61 INTRODUCTION Anthropogenic change in the amount and configuration of native vegetation communities is hypothesized to be a major driver of bird population declines across landscapes (Lampila et al. 2005). Habitat destruction and fragmentation transform landscapes by reducing habitat area, increasing distance and isolation between fragments, and increasing the amount of edge (Haddad et al. 2015). In addition to diminishing available habitat, these processes can negatively affect a suite of demographic rates (Lampila et al. 2005) that drive bird density, distribution, and richness (McGarigal and McComb 1995, Villard et al. 1999, Donovan and Flather 2002). While it is well known that landscape changes can have widespread negative effects on bird distribution and abundance, the specific demographic parameters involved in these declines often remain unknown for many species of conservation concern (Lampila et al. 2005). Several hypotheses have been proposed to explain bird population declines in response to habitat loss and fragmentation. First, changes in landscape composition (e.g., patch size reduction and greater exposure to edges) can initiate changes in predator and brood parasite abundance (Chalfoun et al. 2002), leading to elevated nest predation or brood parasitism (Robinson et al. 1995), and lower adult survival (Lampila et al. 2005). Second, as fragments become more isolated and surrounded by a matrix of human-transformed land cover types, connectivity is reduced. Hence, mobility becomes inhibited in highly fragmented landscapes, potentially impacting pairing success and survival of dispersing females (Dale 2001, Cooper and Walters 2002, Lampila et al. 2005, Robles et al. 2008). Third, smaller patches may be of lower quality and contain less food resources and nest sites (Burke and Nol 1998, Zanette 2001, Lampila et al. 2005), which can negatively affect demographic 62 parameters such as nest success, clutch size, and fledgling condition. Lastly, habitat loss and fragmentation often lead to a reduction in abundance of males in the remaining habitat patches (Dale 2001). Because females may use territory density as a habitat selection cue, this change may lead to a reduction in male pairing success (Villard et al. 1993, Ward and Schlossberg 2004). Given the myriad impacts of landscape change on bird populations, a better understanding of the specific demographic mechanisms responsible for declines is crucial to address adverse fragmentation effects (Lampila et al. 2005). We focused on the breeding response of the Bachman’s sparrow (Peucaea aestivalis) to habitat loss and fragmentation as a possible explanation for population declines. Bachman’s sparrow is a ground-nesting bird that is strongly associated with fire-maintained longleaf pine forests in the southeastern United States (Dunning 2006). Due to extensive habitat loss and fire suppression, this ecosystem has been reduced to 3-5% of its historic range (Frost 2006), and much of the remaining longleaf pine ecosystem exists as scattered and degraded remnant patches (Van Lear et al. 2005). Consequently, Bachman’s sparrow is listed as a species of conservation concern throughout its range (Cox and Widener 2008), and habitat loss and fragmentation likely are serious threats to its population persistence (Dunning and Watts 1990) and other longleaf pine biota (Van Lear et al. 2005). Additionally, the species has been considered to be a surrogate species for longleaf and open-pine ecosystems because of its reliance on frequent fire regimes (typically ≤ 3 years; Tucker et al. 2004, Tucker et al. 2006), open canopy, and diverse groundcover characterizing these forests (Shelton 2014). Therefore, understanding the effects of habitat fragmentation on Bachman’s sparrow demographics may have important implications for the conservation of other longleaf pine-associated species. 63 Suitable patches of longleaf pine forest often are unoccupied by Bachman’s sparrows (Dunning and Watts 1990), and previous research suggests that habitat loss and fragmentation play an important role in determining the distribution of this species. Indeed, isolated patches of habitat are less likely to be colonized than connected patches, possibly due to limited dispersal ability (Dunning et al. 1995, Dunning and Kilgo 2000). Taillie et al. (2015) found that landscape-level habitat amount was the most influential predictor of Bachman’s sparrow occupancy, suggesting a strong effect of habitat loss on this species. Despite the emergent patterns of Bachman’s sparrow occupancy that have been demonstrated by these studies, the demographic mechanisms underlying these landscape-scale relationships remain largely unknown. In this paper, we focused on three of the four hypotheses discussed above: 1) lower productivity in response to increased nest predation and brood parasitism, and reduced pairing success due to 2) disrupted landscape connectivity or 3) low habitat and male quality in isolated locations. Specifically, we examined the influence of habitat loss and fragmentation on these key components of reproductive success by comparing Bachman’s sparrows residing in highly fragmented and relatively continuous longleaf pine landscapes in North Carolina. METHODS Study area We studied Bachman’s sparrows at 8 sites within the Middle Atlantic Coastal Plain physiographic region (hereafter, ‘Coastal Plain’) of southeastern North Carolina (Figure 1). This region is part of the North American Coastal Plain, which has recently been recognized as a global biodiversity hotspot because of considerable plant endemism and habitat loss 64 (Noss et al. 2015). Longleaf pine ecosystems represented < 20% of the study area, while development, agricultural lands, and pine plantations were the most abundant land cover types (Southeast GAP Analysis Program 2008). The majority of longleaf pine woodlands occurred on public landholdings, which comprised approximately 17% of the study area and were managed by the US Department of Defense, North Carolina Forest Service, North Carolina Wildlife Resources Commission (NCWRC), and the North Carolina Plant Conservation Program (NCPCP). The Nature Conservancy (TNC) also managed 2% of landholdings. Because of the species’ rarity in the study area, we relied on previously published information on Bachman’s sparrow distribution in North Carolina (NCWRC, unpublished data), reconnaissance surveys, and accessibility to select study sites. Bachman’s sparrow habitat primarily was composed of mesic longleaf pine flatwoods bordered by pocosin wetlands. Wet pine savannas in this region are characterized by an open canopy, seasonally saturated soils, and a herbaceous layer dominated by graminoids, forbs, and low shrubs. In addition to longleaf pine, tree species dominating sparrow habitat included loblolly (P. taeda) and pond pine (P. serotina). Dominant groundcover species in these vegetation types included wiregrass (Aristida stricta), switchcane (Arundinarea gigantea), bracken fern (Pteridium aquilinum) and inkberry (Ilex glabra). Prescribed fire typically was applied on a 3-5 year cycle, but sometimes exceeded 5 years on privately owned sites where longleaf pine management was not an objective. Field data collection Reproductive status and territory mapping We captured adult male Bachman’s sparrows at the beginning of each breeding season from 2014-2015. Female Bachman’s sparrows are not easily observed or captured, so 65 we focused our study on the reproductive behaviors of male sparrows. We lured males into mist nests using playback of Bachman’s sparrow vocalizations following methods described by Jones and Cox (2007), and by placing nets in frequently used flight paths. Once captured, we measured unflattened wing length and fitted individuals with a unique combination of three colored leg bands and a US Geological Survey aluminum band to allow for visual identification of individual sparrows. Also, we studied a small subset of males that we were unable to capture, and their identity was confirmed through weekly territory mapping and the presence of neighboring color-banded males (described below). After banding, male territories were visited for 60 min between 5:30 and 13:00 EST once per week between mid-April and the end of July to determine reproductive success for each male (hereafter, ‘reproductive visit’). Territories were visited at different times of the day and by different observers throughout a season. Males were considered territorial if they remained in a given territory for ≥ 4 weeks (Vickery et al. 1992); we used this time span to ensure that non-territorial ‘floaters’ were not included in the analyses. We considered males to be paired if we observed a female sparrow in close proximity to the male, and if we recorded mating behaviors (e.g., mate-guarding, copulation, or nest-building) during a reproductive visit. For territories containing a mated pair, we located nests by observing nesting behaviors of males and females (e.g., distraction displays and carrying nest material, food, or fecal sacs) or opportunistically. When discovered, we checked nests using standard nest monitoring protocols until they fledged or failed (Martin and Geupel 1993). Nests were considered unsuccessful if they were depredated, abandoned, destroyed by weather or fire, or otherwise produced no fledglings. Bachman’s Sparrows are to able raise multiple broods per season (Dunning 2006), so we recorded the number and fate of nesting attempts per territory. 66 Based on weekly observations, we determined three components of reproductive success for males at the end of each field season: 1) pairing success (i.e. paired vs. unpaired), 2) if paired, was a male successful at fledging ≥ 1 offspring, and 3) the number of successful broods produced. Because nests and fledglings are difficult to locate and observe for this species (Dunning 2006), we were not able to precisely estimate productivity (i.e., number of nestlings fledged) for each male. During reproductive visits, we mapped territory boundaries by marking male locations with handheld GPS units (± 5 m) and flagging. We defined a territory as the entire area used during the breeding season, and recorded locations where males were observed foraging, singing, following mates, and feeding fledglings. During visits to a territory, we followed males from a distance to avoid influencing routine behaviors, and recorded 1-5 GPS points of male locations until we lost track of the bird. GPS locations were taken only when a male had moved a substantial distance (>15 m) from its previous location to avoid taking points repeatedly within the same area. To delineate territory boundaries for each male, we used the adehabitatHR package (Calenge 2006) to create 95% minimum convex polygons (MCP) from territory mapping locations (20 ± 12 SD GPS points). Vegetation measurements Territory-level vegetation structure was quantified for each male. We measured vegetation structure within a 5-m radius plot centered on five randomly selected male locations recorded during weekly reproductive visits. Vegetation categories included grass, woody vine-shrub (hereafter ‘woody vegetation’), and forb-fern following Taillie et al. (2015). We quantified vegetation structure using indices of density by recording the presence of each vegetation category (i.e., ‘hits’ or vegetation contacts) on 0.1-m sections of a 1.5-m 67 vertical pole. Vertical density included the number of vegetation hits along the entire length of the pole, while groundcover density was calculated as the presence of vegetation on the first 0.1-m section of the pole (Wiens and Rotenberry 1981). Also, maximum vegetation height was obtained by recording the tallest hit on the pole and rounding to the nearest 0.1-m section (Wiens and Rotenberry 1981, Moorman and Guynn 2001, Taillie et al. 2015). Measurements were taken at every meter along four 5-m transects running in each cardinal direction. At the same 1-m increments along each transect, we estimated canopy closure by recording the presence or absence of live canopy using an ocular sighting tube. We used a 10-factor prism from the center point to determine the total basal area of pines surrounding the plot center. Measurements were then averaged across plots for each male. We obtained time since fire for each male’s territory from GIS databases (NCWRC, NCPCP, and TNC, unpublished data) or by speaking with private land managers. Data substitution was necessary for 3% of territories with missing data. For a single territory monitored in 2014 and 2015, time since fire was unknown, but to our knowledge had not occurred in ≥ 10 years. Therefore, we coded time since fire as 10 years to prevent the inclusion of outliers in our models. For another 2 territories, the most recent fire occurred 7 – 10 years prior (K. Hinson, personal communication), so we selected the median date for those males. For territories that spanned multiple burn units, we used the average time since fire across burn units within the territory. Averaging time since fire was necessary for 29% (n = 28) of males included in the analyses. 68 Statistical analyses Territory characteristics All statistical analyses were carried out using program R, version 3.2.2 (R Core Team 2015) and we report mean ± SD unless noted otherwise. To reduce the number of variables used to measure territory-level vegetation characteristics, we used principal component analysis (PCA). PCA reduces a large number of interdependent variables to 2 or 3 dimensions that extract the underlying environmental gradients (McGarigal et al. 2000). We described the ecological characteristics of principal components based on significant factor loadings (≤ -0.3 and ≥ 0.3; McGarigal et al. 2000). We retained two principal components describing vegetation structure within male Bachman’s sparrow territories, which explained > 85% of the original variance. Principal component 1 (PC1) represented increasing vertical density and max height of grass, and principal component 2 (PC2) represented increasing vertical density and max height of woody vegetation (Appendix B). Landscape-level habitat amount We used landscape-scale habitat amount (hereafter ‘habitat amount’) surrounding male territories as a metric to assess the effects of habitat loss and fragmentation on Bachman’s sparrow reproductive success. We chose this metric because it offers a simple interpretation of both patch size and isolation (Fahrig 2013), and habitat amount is an important predictor of Bachman’s sparrow occupancy (Taillie et al. 2015). To quantify habitat amount surrounding each male Bachman’s sparrow territory, we adopted a ‘speciescentered’ approach (Shirley et al. 2013, Betts et al. 2014). Most landscape studies quantify habitat using human-defined cover types, which may be inaccurate or have little relation to how species respond to a landscape (Betts et al. 2014). For example, longleaf pine habitat 69 type classified for the Southeast GAP analysis dataset is sometimes incorrect (J. Winiarski, personal observation), and Bachman’s sparrow can use other habitat types such as open-pine woodlands, clearcuts, and powerline right-of-ways (Dunning 2006). Alternatively, it is possible to accurately quantify landscape-level habitat amount using species’ distribution data and unclassified Landsat reflectance bands (Shirley et al. 2013) that do not suffer from classification biases. This species-centered approach is a promising method that has been used to reveal the effects of landscape-scale habitat loss and fragmentation on species’ occupancy and vital rates (reviewed in Betts et al. 2014). To quantify a species-centered view of habitat amount within the study area, we constructed a Bachman’s sparrow species distribution model (SDM) using maximum entropy modeling (Maxent; Phillips et al. 2006). Maxent generally performs well and is considered to be one of the most effective approaches to creating SDMs with presence-only data (Elith et al. 2011). Maxent uses environmental data from known occurrence locations to predict the expected distribution of a species, and produces a raster map where each grid cell represents the predicted habitat suitability (Elith et al. 2011). We used Bachman’s sparrow occurrence data obtained from 119 point counts conducted across the study area during April-July, 20072014 (NCWRC and NCSU, unpublished data), and Landsat reflectance bands and normalized difference vegetation index (NDVI) calculated from an April 2014 Landsat image of the study region (Shirley et al. 2013, Betts et al. 2014). We used 2014 because it closely represented the landscape conditions that existed during our study period. To assess model performance, we performed 10-fold cross-validation and calculated the area under the curve of the receiver-operator plot (Phillips et al. 2006; Appendix C). The resulting SDM 70 was converted to a binary habitat model by classifying values ≥ 0.5 to a value of 1 (habitat) and values < 0.5 to a value of 0 (non-habitat; Betts et al. 2014). To obtain habitat amount from the binary habitat model, we first used the rgeos package (Bivand and Rundel 2015) and the ‘gCentroid’ function to extract coordinates of each male’s territory centroid from its 95% MCP. Finally, we used the ‘land.metrics’ function in the spatialEco package (Evans 2016) and the binary habitat model to derive percent habitat amount within a 3-km buffer around each male’s territory centroid. This 3-km scale was important for predicting the effects of habitat amount on Bachman’s sparrow occupancy in North Carolina (Taillie et al. 2015), and corresponds with the estimated dispersal distance for this species based on home range size (Cox and Jones 2007, Brown 2012, Winiarski 2016) and the proportional relationship between territory size and median dispersal described by Bowman (2003). Reproductive success We conducted analyses of reproductive success using generalized linear mixed effects models (GLMM) with the lme4 package (Bates et al. 2015). We included ‘Site’ as a random effect in all models to account for the spatial interdependence of male territories within study sites. For each component of reproductive success, we built separate models as a function of vegetation structure variables from the principal component analysis (PC1 and PC2), time since fire, wing length (i.e., body size), and habitat amount (Table 1). We used the median value for wing length for 10 birds missing this measurement (~10% of our sample). In a concurrent study investigating Bachman’s sparrow habitat selection and space-use (Winiarski 2016), we affixed radio transmitters to a subset of males (n = 37), which can negatively affect productivity (Barron et al. 2010). Therefore, we first tested whether radio-tagged males had 71 reduced pairing (χ2 = 3.17, df = 1, P = 0.07) or reproductive success (χ2 = 0.03, df = 1, P = 0.86) compared to untagged males. Because there were no differences between groups, we discarded this variable in subsequent analyses. Due to our study design (more isolated sites were sampled in 2015), we could not include year as a predictor variable in the models. While we acknowledge that reproductive success may vary annually in response to fluctuating weather and predator abundance, we did not detect a difference in the number of males that paired (χ2 = 1.14, df = 1, P = 0.29) or fledged young (χ2 = 0.00, df = 1, P = 1.00) at sites that we were able to sample in both years of the study (n = 64 territories). Pairing success (0 = unpaired, 1 = paired) and fledging success (0 = no offspring fledged, 1 = at least 1 offspring fledged) were fitted to binomial distributions. We conducted additional analyses examining 1) number of successful broods fledged as the response variable using the same variables as above (fitted to a Poisson distribution), and 2) daily nest survival in response to temporal, local-, and landscape-scale covariates in a separate study (Winiarski 2016); however, we had similar results as the binomial fledging success analysis and do not include these analyses here. For both model sets, only territorial males (males present ≥ 4 weeks) were included in the analyses, and only paired males were included in the fledging success models. To explore the most important variables for each model set, we tested all possible linear combinations of variables. The top pairing and fledging success models were selected using Akaike’s Information Criterion corrected for small sample sizes (AICc; Burnham and Anderson 2002) using the ‘MuMIn’ package (Barton 2016). Models were ranked according to their AICc values, and model averaging was performed across the top candidate models (∆AICc ≤ 2). A parameter was considered significant if the 95% CI did 72 not include 0, and was considered to have no relationship with the response variable if the 95% CI overlapped 0. RESULTS Territory and landscape characteristics We monitored 112 Bachman’s sparrow territories, and were able to record reproductive success for 96 Bachman’s sparrow territories where males were present for ≥ 4 weeks: 43 males at 5 sites in 2014, and 53 males at 5 sites in 2015 (Table 2). Of the 96 males, 89 (93%) were color banded, and 8 of these banded males were monitored in both years of the study (Table 2). Local- and landscape-scale factors were variable across territories, with time since fire ranging from 0.2 – 10 years (2.0 ± 1.95 years) and habitat amount ranging from 3 – 25% (15.9 ± 7.45%). Reproductive success Pairing success Approximately 69% of males were paired across all years and sites combined (Table 2). The GLMM analysis for pairing success included three top candidate models with ∆AICc ≤ 2 (Table 3), and habitat amount appeared in all of the top models. Following model averaging, habitat amount did not overlap zero (Table 4), reflecting the consistently lower pairing rate in locations with low habitat amount within 3 km (Figure 3). Two other terms featured in the two top models, time since fire and PC2 (Table 3), but the confidence limits of the model averaged parameter estimates overlapped zero. Neither PC1 or wing length appeared in models with ∆AICc ≤ 2 (Table 3). Fledging success 73 We documented no support for any of the variables to predict the probability of fledging ≥ 1 offspring. Sixty-six paired males were included in the fledging success GLMM analysis. Of these 66 males, approximately 23% produced zero offspring, 64% fledged a single brood, and 14% raised two broods. No males were observed to successfully fledge more than two broods. There were eight top candidate models in the fledging success model set, with the null model ranking the highest (Table 3). PC1, PC2, wing length, time since fire, and habitat amount appeared in the top models, but none of these variables were significant following model averaging (Table 4). DISCUSSION As expected, we saw lower pairing success in areas with less habit within 3 km of male territories, but the probability of fledging offspring was unaffected. Thus, differential pairing success as a function of habitat amount may be partly responsible for the lower probability of occurrence of this species in landscapes with low levels of habitat (Taillie et al. 2015). Although local vegetation conditions (i.e., herbaceous groundcover) and frequent fire are often emphasized as the most critical components of Bachman’s sparrow breeding habitat (Dunning and Watts 1990, Haggerty 1998, Tucker et al. 2004, Tucker et al. 2006), we conclude that pairing success was more strongly influenced by habitat amount at the landscape-scale. Our results indicated no relationship between habitat amount and fledging success. This result is supported by previous work, which found no indication of reduced nest survival for Bachman’s sparrows in more fragmented areas compared to relatively continuous landscapes in the Coastal Plain of North Carolina (Winiarski 2016). Rather, reproductive success of males in this study that were able to pair was high, regardless of habitat amount in 74 the surrounding landscape. Thus, it appears the mechanisms responsible for negative effects of habitat fragmentation (i.e., elevated nest predation and brood parasitism) for many species (Robinson et al. 1995, Rodewald and Yahner 2001) differ for Bachman’s sparrow. For example, Winiarski (2016) found that parasitism by brown-headed cowbirds (Molothrus ater) occurred for only 11% of Bachman’s sparrow nests, and low levels of nest parasitism have been reported elsewhere (Haggerty 1988, Stober and Krementz 2000, Tucker et al. 2006). Little is known of the predator community responsible for Bachman’s sparrow nest predation, or whether the predator community is differentially affected by habitat loss and fragmentation. Future research on Bachman’s sparrow nest predators will help clarify the relationships observed in this study. Reduced pairing success for males in isolated habitat patches has been observed in migratory (Gibbs and Faaborg 1990, Burke and Nol 1998, Bayne and Hobson 2001) and resident species (Cooper and Walters 2002, Robles et al. 2008) in a variety of fragmented ecosystems. There are several explanations for limited breeding opportunities for males in highly fragmented landscapes. One hypothesis is that males in highly isolated patches may fail to pair because these patches are rarely colonized by potential mates (the ‘disrupted dispersal’ hypothesis; Dale 2001, Cooper and Walters 2002, Lampila et al. 2005, Robles et al. 2008). Female-biased natal dispersal is common in birds (Greenwood 1980), and it has been proposed that females in fragmented landscapes have 1) a limited ability to find mates, and 2) may effectively be ‘lost’ from the breeding population if they disperse into areas devoid of conspecifics (Dale 2001). This disruption of female dispersal would result in maleskewed populations, which is often the case for small and isolated bird populations in decline (Dale 2001, Donald 2007). In agreement with this hypothesis, we observed zero females at 75 two sites and only one female at four sites, all of which were small and isolated. Disrupted dispersal seems a plausible explanation for low pairing success of Bachman’s sparrows in our study, and these movements have been negatively impacted by habitat fragmentation in a more mobile longleaf pine specialist, the red-cockaded woodpecker (Picoides borealis; Kesler et al. 2012). However, removal or translocation experiments are needed to test this hypothesis (e.g., Cooper and Walters 2002) and the resistance of different matrix conditions to Bachman’s sparrow movement (e.g., Jones 2013). Alternatively, females might not select isolated patches due to the potentially lower quality of these areas and the males residing within them (Cooper and Walters 2002). However, we accounted for measures of habitat quality (i.e., vegetation structure principal components and time since fire) and male quality (i.e., body size) in the pairing success analysis, and showed no confounding effects of these factors. Earlier work showed that breeding productivity was greatest for Bachman’s sparrows in areas burned 1-3 years previously, and declined significantly after 3 years (Tucker et al. 2006). Although time since fire exceeded 3 years for ~30% of unpaired males and was longer for unpaired males (2.71 ± 2.60 years) than paired males (1.6 ± 1.49 years), differences were not significant. Male age and breeding experience can be an important determinant of pairing success (e.g., Bayne and Hobson 2001, Dale 2011), but we were unable to evaluate the importance of male age because adult Bachman’s sparrows can only be classified initially as after-hatch-year (Pyle 1997). However, several males of known age or at least after-second-year did not indicate that experience was related to reduced pairing success in this study; 2 males banded as nestlings in 2014 bred successfully as adults the following breeding season, and approximately 19% of unpaired males (n = 6) banded in 2014 and monitored in 2015 were at 76 least 3 years old. Active habitat-selection by females may be a possible explanation for the low pairing success of sparrows in isolated patches, but more research is needed to understand the effects of food availability, male age, and conspecific attraction on pairing success in Bachman’s sparrow. Although reduced pairing success in highly fragmented landscapes may be partly responsible for Bachman’s sparrow declines, it remains unclear which demographic parameters are contributing the most to population growth rates, and how they are affected by landscape continuity. Adult survival is generally the most important vital rate affecting population growth for a wide diversity of birds (Sæther and Bakke 2000), and previous work suggests this relationship may apply to Bachman’s sparrow populations. Pulliam et al. (1992) concluded that adult and juvenile survival were the primary parameters influencing growth rates in Bachman’s sparrows using a spatially-explicit simulation approach. Likewise, adult survival had a larger influence on population growth rates than annual productivity for Bachman’s sparrows in an old-growth longleaf forest in Georgia (Cox and Jones 2010). However, both studies selected reproductive rates from the literature and assumed that juvenile survival produced a stable population growth rate (λ = 1.0). Therefore, obtaining accurate estimates of reproductive success and survival is necessary to gain a better understanding of Bachman’s sparrow population dynamics and implement effective conservation strategies. In the rapidly urbanizing southeastern US, wildlife species associated with open pine woodlands, such as Bachman’s sparrow, are likely to experience the highest rates of habitat loss under future land-use scenarios (Martinuzzi et al. 2015). Urban sprawl in this region is expected to double or triple within the next 50 years, putting already vulnerable species at 77 further risk by decreasing habitat amount and hindering management actions (i.e., prescribed fire) that are necessary to maintain open pine ecosystems (Terando et al. 2014). Therefore, understanding the effects of landscape transformation on Bachman’s sparrow demographics will become increasingly critical. Promoting breeding opportunities, and perhaps long-term persistence of Bachman’s sparrow, depends on protection and management of all remaining patches of longleaf pine (especially via prescribed burning; Tucker et al. 2004, Tucker et al. 2006), and – even more importantly – increasing landscape-level habitat amount. Managers can ensure a probability of pairing close to 100% by coordinating restoration and management activities in landscapes comprised of ≥20% habitat (~560 ha) within 3 km. Otherwise, restoration efforts may fail to accommodate Bachman’s sparrow and other longleaf pine inhabitants, despite suitable conditions at the local-level. 78 LITERATURE CITED Barron, D. G., J. D. Brawn, and P. J. Weatherhead. (2010). Meta-analysis of transmitter effects on avian behavior and ecology. Methods in Ecology and Evolution 1:180-187. Barton, K. (2016). MuMIn: Multi-model inference R package 1.15.6. http:// CRANRprojectorg/package=MuMIn Bates, D., M. Maechler, B. Bolker, and S. Walker (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67:1–48. doi:10.18637/jss.v067.i01 Bayne, E. M., and K. A. Hobson. (2001). Effects of habitat fragmentation on pairing success of ovenbirds: importance of male age and floater behavior. Auk 118:380-388. Betts, M.G., L. Fahrig, A. S. Hadley, K. E. Halstead, J. Bowman, W. D. Robinson, J. A. Wiens, and D. B. Lindenmayer. (2014). A species‐centered approach for uncovering generalities in organism responses to habitat loss and fragmentation. Ecography 37:517-527. Bivand, R, and C. Rundel. (2015). Interface to geometry engine - open source (GEOS) R package 0.3-12. http:// CRANR-projectorg/package=rgeos Bowman, J. (2003). Is dispersal distance of birds proportional to territory size? Canadian Journal of Zoology 81:195-202. Brown, S. K. (2012). Movements, home range and habitat selection of Bachman’s Sparrows (Peucaea aestivalis) on a longleaf sandhill forest: Implications for fire management. M.S. thesis, University of Georgia, Athens, GA, USA. Burke, D. M., and E. Nol. (1998). Influence of food abundance, nest-site habitat, and forest fragmentation on breeding Ovenbirds. Auk 115:96–104. Burnham, K. P., and D. R. Anderson. (2002). Model selection and multimodel inference: a practical information-theoretic approach, 2nd ed. Springer-Verlag, New York. Calenge, C. (2006). The package “adehabitat” for the R software: a tool for the analysis of space and habitat use by animals. Ecological Modelling 197:516–519 Cerame, B., J. A. Cox, R. T. Brumfield, J. W. Tucker, and S. S. Taylor (2014). Adaptation to ephemeral habitat may overcome natural barriers and severe habitat fragmentation in a firedependent species, the Bachman’s Sparrow (Peucaea aestivalis). PLOS One 9:e105782. Chalfoun, A. D., M. J. Ratnaswamy, and F. R. Thompson III. (2002). Songbird nest predators in forest-pasture edge and forest interior in a fragmented landscape. Ecological Applications 12:858-867. 79 Cooper C. B., and J. R. Walters. (2002). Experimental evidence of disrupted dispersal causing decline of an Australian passerine in fragmented habitat. Conservation Biology 16:471– 478. Cox, J. A., and C. D. Jones. (2010). Estimating probabilities of survival of male Bachman's Sparrows from plot-based, mark-resighting, off-plot surveys and multi-strata models. Condor 112:663-669. Cox, J. A., and B. Widener. (2008). Lightning-season burning: friend or foe of breeding birds. Tall Timbers Research Station Miscellaneous Publication 17. Dale S. (2001). Female-biased dispersal, low female recruitment, unpaired males, and the extinction of small and isolated bird populations. Oikos 92:344–356. Dale, S. (2011). Lifetime patterns of pairing success in male Ortolan Buntings Emberiza hortulana. Ibis 153:573-580. Donald, P. F. (2007). Adult sex ratios in wild bird populations. Ibis 149:671-692. Donovan, T. M., and C. H. Flather. (2002). Relationships among North American songbird trends, habitat fragmentation, and landscape occupancy. Ecological Applications 12:364-374. Dunning Jr., J. B. (2006). Bachman’s sparrow (Peucaea aestivalis). The Birds of North America. Number 038. Dunning Jr., J. B., and B. D. Watts. (1990). Regional differences in habitat occupancy by Bachman's Sparrow. Auk 107:463-472. Dunning, J. B., Jr., and J. C. Kilgo (Editors) (2000). Avian research at the Savannah River Site: A model for integrating basic research and long-term management. Studies in Avian Biology 21. Dunning, J. B., Jr., R. Borgella, Jr., K. Clements, and G. K. Meffe (1995). Patch isolation, corridor effects, and colonization by a resident sparrow in a managed pine woodland. Conservation Biology 9:542–550. Elith, J., S. J. Phillips, T. Hastie, M. Dudík, Y. E. Chee, and C. J. Yates. (2011). A statistical explanation of MaxEnt for ecologists. Diversity and Distributions 17:43-57. Evans, J. S. (2016). spatialEco: spatial analysis and modelling R package 0.1-4. http:// CRANR-projectorg/package=spatialEco Fahrig, L. (2013). Rethinking patch size and isolation effects: The habitat amount hypothesis. Journal of Biogeography 40:1649–1663. 80 Frost, C. (2006). History and future of the longleaf pine ecosystem. In The Longleaf Pine Ecosystem: Ecology, Silviculture, and Restoration (S. Jose, E. J. Jokela, and D. L. Miller, Editors). Springer, New York, NY, USA. pp. 9–48. Gibbs J. P., and J. Faaborg. (1990). Estimating the viability of Ovenbird and Kentucky Warbler populations in forest fragments. Conservation Biology 4:193–196. Greenwood, P. J. (1980). Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour, 28:1140-1162. Haddad, N.M., L. A. Brudvig, J. Clobert, K. F. Davies, A. Gonzalez, R. D. Holt, T. E. Lovejoy, J. O. Sexton, M. P. Austin, C. D. Collins, and W. M. Cook. (2015). Habitat fragmentation and its lasting impact on Earth’s ecosystems. Science Advances 1:p.e1500052. Haggerty, T. M. (1988). Aspects of the breeding biology and productivity of Bachman’s Sparrow in central Arkansas. Wilson Bulletin 100:247–255. Haggerty, T. M. (1998). Vegetation structure of Bachman’s Sparrow breeding habitat and its relationship to home range. Journal of Field Ornithology 69:45–50. Jones, C. D. (2013). Landscape-level investigations of avian species at Fort Benning, Georgia. Ph.D. dissertation, University of Georgia, Athens, GA, USA. Jones, C. D., and J. A. Cox. (2007). Field procedures for netting Bachman’s Sparrows. North American Bird Bander 32:114-117. Kesler, D. C., and J. R. Walters (2012). Social composition of destination territories and matrix habitat affect Red-cockaded Woodpecker dispersal. Journal of Wildlife Management 76:1028–1035. Lampila, P., M. Mönkkönen, and A. Desrochers (2005). Demographic responses by birds to forest fragmentation. Conservation Biology 19:1537–1546. Martin, T. E., and G. R. Geupel. (1993). Nest-monitoring plots: methods for locating nests and monitoring nest success. Journal of Field Ornithology 64:507–519. Martinuzzi, S., J. C. Withey, A. M. Pidgeon, A. J. Plantinga, A. J. McKerrow, S. G. Williams, D. P. Helmers, and V. C. Radeloff. (2015). Future land‐use scenarios and the loss of wildlife habitats in the southeastern United States. Ecological Applications 25:160-171. McGarigal, K., and W. C. McComb (1995). Relationships between landscape structure and breeding birds in the Oregon Coast Range. Ecological Monographs 65:235–260. McGarigal, K., S. Cushman, and S. Stafford. (2000). Multivariate Statistics for Wildlife and Ecology Research. Springer-Verlag, New York. 81 Moorman, C. E., and D. C. Guynn Jr. (2001). Effects of group-selection opening size on breeding bird habitat use in a bottomland forest. Ecological Applications 11:1680–1691. Noss, R. F., W. J. Platt, B. A. Sorrie, A. S. Weakley, D. B. Means, J. Costanza, J., and R. K. Peet (2015). How global biodiversity hotspots may go unrecognized: lessons from the North American Coastal Plain. Diversity and Distributions 21:236-244. Phillips, S. J., R. P. Anderson, and R. E. Schapire. (2006). Maximum entropy modeling of species geographic distributions. Ecological Modelling 190:231–259. Pulliam, H. R., J. B. Dunning Jr, and J. Liu. (1992). Population dynamics in complex landscapes: a case study. Ecological Applications 2:165-177. Pyle, P. (1997). Identification Guide to North American Birds, Part I: Columbidae to Ploceidae. Slate Creek Press, Bolinas, CA, USA. R Development Core Team. (2015). R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Robinson, S. K., F. R. Thompson III, T. M. Donovan, D. R. Whitehead, and J. Faaborg. (1995). Regional forest fragmentation and the nesting success of migratory birds. Science 267:1987-1990. Robles, H., C. Ciudad, R. Vera, P. P. Olea, and E. Matthysen. (2008). Demographic responses of middle spotted woodpeckers (Dendrocopos medius) to habitat fragmentation. Auk 125:131-139. Rodewald, A. D., and R. H. Yahner. (2001). Avian nesting success in forested landscapes: influence of landscape composition, stand and nest-patch microhabitat, and biotic interactions. Auk 118:1018–1028. Sæther, B. E., and Ø. Bakke. (2000). Avian life history variation and contribution of demographic traits to the population growth rate. Ecology 81:642-653. Shelton, S. (2014) Birds of a feather, linked to an open pine forest. U.S. Fish and Wildlife Service. http://www.fws.gov/ landscape-conservation/stories/birds-of-a-feather-linked-toanopen-pine-forest.html Shirley, S. M., Z. Yang, R. A. Hutchinson, J. D. Alexander, K. McGarigal, and M. G. Betts. (2013). Species distribution modelling for the people: unclassified landsat TM imagery predicts bird occurrence at fine resolutions. Diversity and Distributions 19:855-866. Stober, J. M. and D. G. Krementz. (2000). Survival and reproductive biology of the Bachman’s Sparrow. Proceedings of the Annual Conference of Southeastern Association of Fish and Wildlife Agencies 54:383–390. 82 Southeast GAP Analysis Project. (2008). Southeast GAP Regional Land Cover (NC Subsection <http://www.basic.ncsu.edu/segap/index.html> Taillie, P. J., M. N. Peterson, and C. E. Moorman. (2015). The relative importance of multiscale factors in the distribution of Bachman's Sparrow and the implications for ecosystem conservation. Condor 117:137-146. Terando, A. J., J. Costanza, C. Belyea, R. R. Dunn, A. McKerrow, and J. A Collazo. (2014). The southern megalopolis: using the past to predict the future of urban sprawl in the Southeast US. PloS one 9:e102261. Tucker Jr., J. W., W. D. Robinson, and J. B. Grand (2004). Influence of fire on Bachman’s Sparrow, an endemic North American songbird. Journal of Wildlife Management 68:1114– 1123. Tucker Jr., J. W., W. D. Robinson, and J. B. Grand. (2006). Breeding productivity of Bachman’s sparrows in fire-managed longleaf pine forests. Wilson Journal of Ornithology 118:131–137. Van Lear, D. H., W. D. Carroll, P. R. Kapeluck, and R. Johnson (2005). History and restoration of the longleaf pine–grassland ecosystem: Implications for species at risk. Forest Ecology and Management 211:150–165. Vickery, P. D., M. L. Hunter Jr., and J. V. Wells. (1992). Use of a new reproductive index to evaluate relationship between habitat quality and breeding success. Auk 109:697-705. Villard, M. A., P. R. Martin, and C. G. Drummond. (1993). Habitat fragmentation and pairing success in the Ovenbird (Seiurus aurocapillus). Auk 759-768. Villard, M. A., M. K. Trzcinski, and G. Merriam. (1999). Fragmentation effects on forest birds: relative influence of woodland cover and configuration on landscape occupancy. Conservation Biology 13:774-783. Ward, M. P., and S. Schlossberg. (2004). Conspecific attraction and the conservation of territorial songbirds. Conservation Biology 18:519–525. Wiens, J. A., and J. T. Rotenberry. (1981). Habitat associations and community structure of birds in shrubsteppe environments. Ecological Monographs 51:21-41. Zanette, L. (2001). Indicators of habitat quality and the reproductive output of a forest songbird in small and large fragments. Journal of Avian Biology 32:38-46. 83 Table 1. Variables used in candidate models to test the effects of local- and landscape-scale factors on reproductive success of male Bachman’s sparrows in southeastern North Carolina, 2014-2015. Variable Wing length Vegetation PC1 & PC2 Time since fire Habitat amount (%) Hypothesis Larger, higher quality males will have a higher probability of pairing and fledging offspring. Vegetation structure variables may indicate habitat quality. Males in recently burned stands will have high quality territories with more resources, and will have better reproductive success. Males in more contiguous landscapes will have a better chance of pairing and producing young due to increased connectivity and decreased nest predation and parasitism. 84 Table 2. Summary of Bachman’s sparrow territories monitored in the Coastal Plain of North Carolina, USA, 2014-2015. Percent of males that fledged young included paired males only. Note that 8 banded males were monitored in both years. N focal males Study site County Year Banded Unbanded Paired (%) Holly Shelter Pender 2014 31 3 32/34 (94) Holly Shelter Pender 2015 28a 0 24/28 (86) Compass Pointe Brunswick 2014 1 0 1/1 (100) Compass Pointe Brunswick 2015 1b 0 0/1 (0) Sleepy Creek Bladen 2014 4 0 1/4 (25) Shaken Creek Pender 2014 1 1 1/2 (50) Molpus Brunswick 2014 2 0 1/2 (50) Green Swamp Brunswick 2015 13 3 6/16 (38) Pinch Gut Brunswick 2015 2 0 0/2 (0) Boiling Spring Brunswick 2015 6 0 0/6 (0) a includes 7 color banded males monitored in both 2014 and 2015 b same color banded male monitored in both 2014 and 2015 Fledged (%) 24/32 (75) 19/24 (79) 1/1 (100) 0/0 (0) 0/1 (0) 1/1 (100) 0/1 (0) 6/6 (100) 0/2 (0) 0/6 (0) 85 Table 3. Top (∆AICc ≤ 2) mixed-effects logistic regression models of pairing and fledging success for Bachman’s sparrows in the Coastal Plain region of North Carolina, USA, 20142015. Site was used as a random effect in all models. Models were compared using Akaike’s Information Criterion corrected for small sample size (AICc). Model parameters: Hab (percent habitat within 3 km of a territory centroid), TSF (time since fire), PC1 (habitat principal component 1), PC2 (habitat principal component 2), and Wing (wing length). Model K Loglik AICc ∆AICc wi Pairing success Hab 3 -41.68 89.63 0.00 0.21 Hab + TSF 4 -41.00 90.44 0.81 0.14 Hab + PC2 4 -41.41 91.26 1.63 0.10 Fledging success Null 2 -35.37 74.94 0.00 0.11 TSF 3 -34.33 75.05 0.11 0.11 Hab + TSF 4 -33.47 75.60 0.66 0.08 PC2 3 -34.84 76.07 1.13 0.06 Wing 3 -34.88 76.15 1.21 0.06 TSF + Wing 4 -33.92 76.49 1.55 0.05 Hab 3 -35.08 76.54 1.60 0.05 PC1 3 -35.26 76.91 1.97 0.04 Notes: K = degrees of freedom, Loglik = log liklihood, ∆AICc = change in AICc in relation to the highest-ranked model, and wi = AICc weight. 86 Table 4. Parameter estimates and 95% confidence intervals for model-averaged variables (∆AICc ≤ 2) used to model pairing and fledging success for 96 male Bachman’s Sparrows in the Coastal Plain region of North Carolina, USA, 2014-2015. Model parameters: Hab (percent habitat within 3 km of a territory centroid), TSF (time since fire), PC1 (habitat principal component 1), PC2 (habitat principal component 2), and Wing (wing length). Variable Parameter estimate a 2.5% 97.5% Pairing success Hab 1.46 0.88 2.03 TSF -0.09 -0.79 0.21 PC2 -0.04 -0.76 0.34 Fledging success TSF -0.23 -1.32 0.21 Hab -0.13 -1.70 0.59 PC2 -0.03 -0.80 0.25 Wing -0.05 -0.81 0.29 PC1 0.01 -0.46 0.74 a Parameter estimates and 95% CIs were derived from standardized variables. 87 Figure 1. Study sites in southeastern North Carolina where reproductive data were collected for 96 Bachman’s sparrow territories, 2014-2015. Study sites are indicated with four-letter abbreviation; Boiling Spring Lakes Preserve (BSLP), Compass Pointe (COPO), Green Swamp Preserve (GSPR), Holly Shelter Game Land (HSGL), Molpus Timberlands (MOLP), Pinch Gut Game Land (PGGL), Shaken Creek Preserve (SCPR), and Sleepy Creek Farms (SCFA). 88 Figure 2. Mean (± SD) habitat amount (%) within 3 km of male territory centroids for 8 sites in the Coastal Plain region of North Carolina, USA, 2014-2015. Sample size of males at each site is listed below the site code. 89 Figure 3. Model-averaged predicted probability of pairing for male Bachman’s sparrows in relation to percent habitat within 3 km of a territory centroid. Shaded regions indicate 95% confidence intervals. 90 APPENDICES 91 Appendix A Appendix A. Reproductive index scores modified from Vickery et al. (1992) and Tucker et al. (2006) used as an index of Bachman’s Sparrow reproductive success. Score 1 2 3 4 5 6 7 8 9 Description Male was on territory for ≥ 4 weeks but showed no indication of pairing. Male seen consorting with a female, but no evidence of nesting. Female seen carrying nesting material or nest found. Adults seen carrying food or found feeding nestlings. Direct observation of fledglings or nest known to have successfully fledged. Renesting after successful fledging of a first brood. Direct observation of fledglings for a second brood. Active nest after successful fledging of a second brood. Direct observation of fledglings for a third brood. 92 Appendix B Appendix B. Results of principal component analysis of 15 vegetation variables measured at Bachman’s sparrow territory locations. The first two components explained over half of the variance and were included as covariates in the pairing and fledging success models. Values indicate factor loadings for each component. Principal components were interpreted using significant (≤ -0.3 and ≥ 0.3) factor loading values (bold). Habitat characteristic Eigenvalue Percent of variance explained Vertical woody density Vertical grass density Vertical forb density Vertical switchcane density Vertical dead density Groundcover woody density Groundcover grass density Groundcover forb density Groundcover switchcane density Groundcover dead density Maximum woody height Maximum grass height Maximum forb height Maximum switchcane height Maximum dead height PC1 1.41 62.49 -0.148 0.614 -0.007 0.005 -0.021 -0.023 0.123 0.014 0.000 -0.002 -0.291 0.699 -0.064 0.018 -0.082 PC2 0.86 23.08 0.446 0.244 0.050 0.026 0.046 0.042 0.023 0.003 0.000 0.000 0.779 0.237 0.163 0.099 0.184 93 Appendix C Appendix C. The Bachman’s sparrow Maxent model performed better than random, and had high predictive success based on area under the curve values (AUC; training AUC = 0.934, test AUC = 0.942, SD = 0.010). Variables with the highest predictive contributions were reflectance band 5 and NDVI (41.8% and 26.9%, respectively; Figure 1). Figure 1. Percent contribution of 8 predictors used in Bachman’s sparrow Maxent model for the Coastal Plain of North Carolina, USA. 94