(PDF) Multiscale habitat suitability index models for priority landbirds in the Central Hardwoods and West Gulf Coastal Plain/Ouachitas Bird Conservation Regions

We developed spatially-explicit bird-habitat models with a variety of site-specific and landscape parameters to predict avian species distributions on southern Appalachian National Forests to aid National Forests with bird conservation planning. These models can be used to assess the effects of different forest manage- ment alternatives on long-term population viability for a variety of avian species. Unlike past planning efforts

United States Department of Agriculture Forest Service Northern Research Station General Technical Report NRS-49 Multiscale Habitat Suitability Index Models for Priority Landbirds in the Central Hardwoods and West Gulf Coastal Plain/Ouachitas Bird Conservation Regions John M. Tirpak D. Todd Jones-Farrand Frank R. Thompson, III Daniel J. Twedt William B. Uihlein, III Abstract Ecoregional conservation planning for priority landbirds requires methods that explicitly link populations to habitat conditions at multiple scales. We developed Habitat Suitability Index (HSI) models to assess habitat quality for 40 priority bird species in the Central Hardwoods and West Gulf Coastal Plain/Ouachitas Bird Conservation Regions. The models incorporated both site and landscape environmental variables derived from one of six nationally consistent datasets: ecological subsections from the National Ecological Unit Hierarchy, National Land Cover Dataset, National Elevation Dataset, National Hydrography Dataset, State Soil Geographic Database, and Forest Inventory and Analysis data. We initially defined potential habitat for each species from unique landform, landcover, and successional age class combinations. Species-specific environmental variables identified from the literature were used to refine initial habitat estimates. We verified models by comparing subsection-level HSI scores and Breeding Bird Survey (BBS) abundance via Spearman rank correlation. To validate models, we developed generalized linear models that predicted BBS abundance as a function of HSI score and Bird Conservation Region. We considered models that included a significant (P ≤ 0.100) positive coefficient on the BBS predictor to be valid and useful for conservation planning. The Authors JOHN M. TIRPAK is science coordinator for the Lower Mississippi Valley Joint Venture, Vicksburg, MS. D. TODD JONES-FARRAND is a postdoctoral fellow with the Department of Fisheries and Wildlife Sciences, University of Missouri, Columbia, MO. FRANK R. THOMPSON, III is a research wildlife biologist and Project Leader with the Northern Research Station, Columbia, MO. DANIEL J. TWEDT is a research wildlife biologist with the U.S. Geological Survey Patuxent Wildlife Research Center, Vicksburg, MS. WILLIAM B. UIHLEIN, III is coordinator for the Lower Mississippi Valley Joint Venture, Vicksburg, MS. Cover Photos Clockwise from top left: Cerulean warbler, U.S. Forest Service; Wood thrush, Steve Maslowski, U.S. Fish & Wildlife Service; Pileated woodpecker, U.S Forest Service; Painted bunting, Deanna K. Dawson, Patuxent Bird Identification InfoCenter, Photo used with permission; Kentucky warbler, U.S. Fish & Wildlife Service; Bewick’s wren, Dave Menke, U.S. Fish & Wildlife Service. Manuscript received for publication 27 August 2008 Published by: U.S. FOREST SERVICE 11 CAMPUS BLVD SUITE 200 NEWTOWN SQUARE PA 19073-3294 July 2009 For additional copies: U.S. Forest Service Publications Distribution 359 Main Road Delaware, OH 43015-8640 Fax: (740)368-0152 Visit our homepage at: http://www.nrs.fs.fed.us/ CONTENTS Introduction.......................................................................................... 1 Study Areas .......................................................................................... 3 Methods ................................................................................................ 4 Priority Bird Species ......................................................................... 4 HSI Model Development .................................................................. 4 Model Testing ................................................................................. 12 Model Accounts ................................................................................. 14 Acadian Flycatcher......................................................................... 14 American Woodcock ...................................................................... 20 Bachman’s Sparrow ....................................................................... 25 Bell’s Vireo ..................................................................................... 29 Bewick’s Wren ................................................................................ 33 Black-and-white Warbler ................................................................ 36 Blue-gray Gnatcatcher ................................................................... 40 Blue-winged Warbler ...................................................................... 44 Brown Thrasher.............................................................................. 47 Brown-headed Nuthatch ................................................................ 51 Carolina Chickadee ........................................................................ 55 Cerulean Warbler ........................................................................... 58 Chimney Swift ................................................................................ 63 Chuck-will’s-widow ......................................................................... 65 Eastern Wood-pewee..................................................................... 68 Field Sparrow ................................................................................. 71 Great Crested Flycatcher ............................................................... 75 Hooded Warbler ............................................................................. 78 Kentucky Warbler ........................................................................... 83 Louisiana Waterthrush ................................................................... 87 Mississippi Kite............................................................................... 92 Northern Bobwhite ......................................................................... 96 Northern Parula ............................................................................ 101 Orchard Oriole.............................................................................. 105 Painted Bunting ............................................................................ 108 Pileated Woodpecker ................................................................... 112 Prairie Warbler ............................................................................. 116 Prothonotary Warbler ................................................................... 120 Red-cockaded Woodpecker ......................................................... 124 Red-headed Woodpecker ............................................................ 129 Swainson’s Warbler...................................................................... 133 Swallow-tailed Kite ....................................................................... 137 Whip-poor-will .............................................................................. 141 White-eyed Vireo.......................................................................... 144 Wood Thrush ................................................................................ 148 Worm-eating Warbler ................................................................... 153 Yellow-billed Cuckoo .................................................................... 157 Yellow-breasted Chat ................................................................... 161 Yellow-throated Vireo ................................................................... 165 Yellow-throated Warbler ............................................................... 169 Current Model Use And Future Directions .................................... 173 Acknowledgments ........................................................................... 173 Literature Cited ................................................................................ 174 INTRODUCTION The primary goal of the North American Landbird Conservation Plan (Rich and others 2004) is to create landscapes that can sustain populations of the 448 native landbird species that breed in the United States and Canada. To attain this goal, the Plan advocates a three-phase approach: 1. Establish population objectives at the continental scale. 2. Allocate these population objectives to speciﬁc Bird Conservation Regions (BCRs). 3. Translate the regional population objectives to habitat goals within each BCR. The ﬁrst two steps of this process have been completed (Panjabi and others 2001, Rosenberg and Blancher 2005), and it is at this third step where the conservation community stands today. Translating target population numbers into concrete habitat goals requires both knowledge of how landbird populations respond to changing habitat conditions and a method for quantifying this relationship. However, there are few data explicitly linking landbird abundance to speciﬁc habitat conditions, nor is there consensus on the optimal methodology to achieve this linkage. The goal of our research is to develop a comprehensive, replicable approach to ecoregional habitat assessment that links habitat conditions to the density of priority bird species. Speciﬁc objectives are to: 1. Assess the ability of landscapes to sustain priority species at prescribed population levels based on the extent and distribution of available habitats. 2. Monitor changes in the ability of landscapes to sustain species. 3. Predict how landscape suitability changes under alternative succession and disturbance patterns, land use, conservation strategies, management practices, and development pressures. To create a replicable and transferable methodology, we selected a Habitat Suitability Index (HSI) modeling approach. HSI models were initially developed by the U.S. Department of the Interior (USDI) Fish and Wildlife Service (FWS) to evaluate habitat quality for a variety of species (Schamberger and others 1982). These models identify and quantify the relationship between key environmental variables and habitat suitability on a scale from 0 to 1. HSI scores are calculated independently for each environmental factor and an appropriate weighting scheme is used to combine individual variables and determine a composite suitability index (SI) score for a particular location. Although the FWS developed HSI models solely with site-speciﬁc habitat variables (e.g., canopy cover) for assessing stand-level habitat suitability, researchers are increasingly developing HSI models that incorporate broad-scale metrics (e.g., percent forest in a 1-km radius) for application to large landscapes (Larson and others 2003). The continued use of the HSI approach by both researchers and managers likely is a result of the intuitive nature of these models as well as their scalability and portability to novel situations. HSI models easily incorporate existing information via a priori hypotheses but also allow generalization of habitat relationships across areas and species where empirical data are limited. Currently, few HSI models include environmental variables at both the site and landscape scale due to the limited site-speciﬁc data across areas that are large enough to exhibit strong 1 diﬀerences in landscape structure or composition. Nevertheless, habitat selection by birds is a multiscale process (Villard and others 1998) and habitat models should reﬂect conditions at multiple scales. This report begins ﬁlling this gap by documenting multiscale HSI models for 40 priority landbird species (Table 1). Table 1.—Partners in Flight regional combined score and USDI Fish and Wildlife Service Bird of Conservation Concern status for 40 priority landbird species in the Central Hardwoods and West Gulf Coastal Plain/ Ouachitas Bird Conservation Regions Central Hardwoods Species Alpha codea Regional combined score Bird of Conservation Concern Regional combined score Bird of Conservation Concern Acadian flycatcher American woodcock Bachman’s sparrow Bell’s vireo Bewick’s wren Black-and-white warbler Blue-gray gnatcatcher Blue-winged warbler Brown thrasher Brown-headed nuthatch Carolina chickadee Cerulean warbler Chimney swift Chuck-will’s-widow Eastern wood-pewee Field sparrow Great crested flycatcher Hooded warbler Kentucky warbler Louisiana waterthrush Mississippi kite Northern bobwhite Northern parula Orchard oriole Painted bunting Pileated woodpecker Prairie warbler Prothonotary warbler Red-cockaded woodpecker Red-headed woodpecker Swainson’s warbler Swallow-tailed kite Whip-poor-will White-eyed vireo Wood thrush Worm-eating warbler Yellow-billed cuckoo Yellow-breasted chat Yellow-throated vireo Yellow-throated warbler ACFL AMWO BACS BEVI BEWR BAWW BGGN BWWA BRTH BHNU CACH CERW CHSW CWWI EAWP FISP GCFL HOWA KEWA LOWA MIKI NOBO NOPA OROR PABU PIWO PRAW PROW RCWO RHWO SWWA STKI WPWI WEVI WOTH WEWA YBCU YBCH YTVI YTWA 16 -20 15 15 13 14 19 15 19 15 19 16 14 15 17 13 13 18 15 14 16 12 17 16 13 18 14 21 16 20 19 17 15 16 18 13 16 16 15 No No Yes Yes Yes No No Yes No No No Yes No No No No No No No Yes No No No No No No Yes No No Yes Yes No Yes No Yes Yes No No No No 17 -20 16 16 16 13 -13 19 16 19 14 16 16 15 13 16 19 18 16 15 13 18 17 16 18 17 21 17 20 18 13 16 15 15 15 13 15 16 Yes No Yes Yes Yes No No No No Yes No Yes No Yes No No No No Yes Yes No No No Yes No No Yes Yes No Yes Yes Yes No No Yes Yes No No No No a 2 West Gulf Coastal Plain/Ouachitas Pyle and DeSante (2003). STUDY AREAS We developed HSI models for landbirds identiﬁed as priorities in the Central Hardwoods (CH) and West Gulf Coastal Plain/Ouachitas (WGCP) BCRs (Fig. 1). The CH, approximately 33 million ha straddling the Mississippi River, is dominated by deciduous hardwood forest. This region is bordered to the north and west by the tallgrass prairie ecosystem, to the east by the Appalachian Mountains, and to the south by the southern pine belt along the Coastal Plain. The vast forests of the CH make it an important breeding area for many area-sensitive species, Figure 1.—Central Hardwoods and West Gulf Coastal Plain/Ouachitas Bird Conservation Regions. 3 including the cerulean warbler, Kentucky warbler, Louisiana waterthrush, and worm-eating warbler (Panjabi and others 2001). The WGCP also is predominantly forested but consists primarily of pine: longleaf pine in the south transitioning to loblolly and shortleaf pine in the north. As a result, this region contains large populations of pine specialists (e.g., red-cockaded woodpecker, brown-headed nuthatch, and pine warbler). The WGCP also contains broad swaths of bottomland hardwood forest, particularly along the Arkansas, Ouachita, and Sabine Rivers, which support substantial populations of the hooded warbler, Kentucky warbler, and Swainson’s warbler (Conner and Dickson 1997). METHODS Priority Bird Species We selected priority bird species for modeling by identifying a subset of the forest-breeding landbirds in the CH or WGCP with a Partners in Flight (PIF) regional combined score of at least 15 (Panjabi and others 2005) or an FWS designation as a Bird of Conservation Concern (USDI Fish and Wildl. Serv. 2002) (Table 1). Forty-nine species initially met these criteria. We eliminated Bachman’s warbler and the ivory-billed woodpecker from consideration due to limited habitat and validation data available within the CH and WGCP for these species. Also, we did not model habitat suitability for the ruﬀed grouse, broad-winged hawk, eastern kingbird, scissor-tailed ﬂycatcher, loggerhead shrike, summer tanager, or eastern towhee. We added American woodcock, blue-gray gnatcatcher, great crested ﬂycatcher, and northern parula to ensure the species modeled were representative of a cross section of habitat associations (e.g., early successional forest, pine savanna, bottomland hardwoods) and conservation priorities (e.g., critical recovery, management attention, planning and responsibility) within these BCRs. HSI Model Development In our adaptation of the HSI approach, we assume that habitat suitability is a function of both composition and structure at the site and landscape scales. To characterize environmental variables at each of these scales, we relied on six nationally consistent datasets: 1. Ecological subsections from the National Ecological Unit Hierarchy. 2. National Landcover Dataset (NLCD) (30-m pixels). 3. National Elevation Dataset (NED) (30-m pixels). 4. National Hydrography Dataset (NHD). 5. State Soil Geographic Database (STATSGO). 6. Forest Inventory and Analysis (FIA) data. The ﬁrst ﬁve datasets are widely available and commonly used to characterize landscape composition and structure. The sixth, FIA, provides information on the composition and structure of vegetation within forest patches (i.e., site scale) from a national ﬁeld survey of forest lands undertaken by the USDA Forest Service. A description of the methodology used to integrate these datasets in a spatially explicit framework is available in Tirpak and others (2009b). 4 Table 2.—Parameters and data sources for inputs in priority forest-breeding landbird Habitat Suitability Index models, Central Hardwoods and West Gulf Coastal Plain/Ouachitas Bird Conservation Regions; numbers correspond to Suitability Index (SI) functions in text Species codea Data source DEM, NLCD, and FIA Landform, landcover, and successional age class NLCD and FIA Early successional patch size (ha) NLCD and NHD Occurrence of water Distance (m) to water NLCD Forest patch size (ha) Landscape composition (percent forest in 1-km radius) Landscape composition (percent forest in 10-km radius) Occurrence of edge Distance (m) to edge Interspersion – 1 landcover class Interspersion – 2 landcover classes Connectivity (km) Grass-open landcover FIA Basal area (m2/ha) Hardwood basal area (m2/ha) Pine basal area (m2/ha) Sawtimber (> 28 cm d.b.h.) tree density (trees/ha) Large (> 50 cm d.b.h) tree density (trees/ha) Large (> 35 cm d.b.h) pine density (trees/ha) Dominant (> 76.2 cm d.b.h.) tree density (trees/ha) Midstory (11–25 cm d.b.h.) density (trees/ha) Snag density (snags/ha) Large (> 30 cm d.b.h.) snag density (snags/ha) Canopy cover (percent) Small stem (< 2.5 cm d.b.h.) density (stems/ha) DEM Slope NHD Distance (m) to stream STATSGO Soil texture Soil moisture ACFL AMWO BACS BEVI BEWR BAWW 1 1 1 1 1 1 2 4 5 2 2 3 3 3 2 2 4 3 3 3 2 4 4 4 5 continued As a ﬁrst step in developing HSI models, we identiﬁed key habitat factors for each species from the literature and compiled all pertinent data from these sources. In the interests of parsimony and processing time, we generally limited our HSI models to ﬁve or fewer suitability indices (Table 2). The ﬁrst SI in all models (with the exception of chimney swift) was a function that assigned SI scores to unique combinations of landform, landcover, and successional age classes. Landform comprised three classes (ﬂoodplain-valley, terrace-mesic, and xeric-ridge) developed from the digital elevation model-derived metrics of aspect, slope, topographic position (the diﬀerence between the elevation value of an individual pixel and the average elevation in a 500and 1,500-m-radius window around it), and relief. Landcover was classiﬁed to seven forest types derived from the NLCD: low-density residential, transitional-shrubland, deciduous, evergreen, 5 Table 2.—continued Species codea Data source DEM, NLCD, and FIA Landform, landcover, and successional age class NLCD and FIA Early successional patch size (ha) NLCD and NHD Occurrence of water Distance (m) to water NLCD Forest patch size (ha) Landscape composition (percent forest in 1-km radius) Landscape composition (percent forest in 10-km radius) Occurrence of edge Distance (m) to edge Interspersion – 1 landcover class Interspersion – 2 landcover classes Connectivity (km) Grass-open landcover FIA Basal area (m2/ha) Hardwood basal area (m2/ha) Pine basal area (m2/ha) Sawtimber (> 28 cm d.b.h.) tree density (trees/ha) Large (> 50 cm d.b.h) tree density (trees/ha) Large (> 35 cm d.b.h) pine density (trees/ha) Dominant (> 76.2 cm d.b.h.) tree density (trees/ha) Midstory (11–25 cm d.b.h.) density (trees/ha) Snag density (snags/ha) Large (> 30 cm d.b.h.) snag density (snags/ha) Canopy cover (percent) Small stem (< 2.5 cm d.b.h.) density (stems/ha) DEM Slope NHD Distance (m) to stream STATSGO Soil texture Soil moisture BGGN BWWA BRTH BHNU CACH CERW 1 1 1 1 1 1 2 2 2 3 3 4 4 2 5 4 4 2 2 3 5 3 3 continued mixed, orchard-vineyard, and woody wetlands. Finally, successional age class was delineated into ﬁve classes based on the average diameter at breast height (d.b.h.) of dominant trees in each stand, ultimately derived from FIA data: grass-forb (trees < 2.5 cm d.b.h.), shrub-seedling (2.5 to 7.5 cm), sapling (7.5 to 12.5 cm), pole (12.5 to 37.5 cm), and sawtimber (> 37.5 cm). We assigned to each of the 105 unique landform, landcover, and successional age class combinations (three landform classes × seven forest type classes × ﬁve successional age classes) an SI value based on the relative habitat suitability rankings reported in the bird habitat matrices in Hamel (1992). These matrices qualitatively assess habitat suitability (marginal, suitable, optimal) for each bird species based on seral stage (4 classes) and forest type (23 classes). To adapt these matrices to our purposes, we crosswalked these forest types to our 6 Table 2.—continued Species codea Data source DEM, NLCD, and FIA Landform, landcover, and successional age class NLCD and FIA Early successional patch size (ha) NLCD and NHD Occurrence of water Distance (m) to water NLCD Forest patch size (ha) Landscape composition (percent forest in 1-km radius) Landscape composition (percent forest in 10-km radius) Occurrence of edge Distance (m) to edge Interspersion – 1 landcover class Interspersion – 2 landcover classes Connectivity (km) Grass-open landcover FIA Basal area (m2/ha) Hardwood basal area (m2/ha) Pine basal area (m2/ha) Sawtimber (> 28 cm d.b.h.) tree density (trees/ha) Large (> 50 cm d.b.h) tree density (trees/ha) Large (> 35 cm d.b.h) pine density (trees/ha) Dominant (> 76.2 cm d.b.h.) tree density (trees/ha) Midstory (11–25 cm d.b.h.) density (trees/ha) Snag density (snags/ha) Large (> 30 cm d.b.h.) snag density (snags/ha) Canopy cover (percent) Small stem (< 2.5 cm d.b.h.) density (stems/ha) DEM Slope NHD Distance (m) to stream STATSGO Soil texture Soil moisture CHSW CWWI EAWP FISP GCFL HOWA 1 1 1 1 1 4 5 2 3 1 2 4 3 2 2 3 3 2 continued landform-landcover classes and adapted the four seral stages to our ﬁve successional age classes (Table 3). First, we identiﬁed which of the 23 forest types occurred in the CH or WGCP (seven types: Sandhills longleaf pine, oak-gum-cypress, elm-ash-cottonwood, loblolly pine-shortleaf pine, mixed pine-hardwood, oak-hickory, and cove hardwoods). We then assigned these forest types to speciﬁc landform and landcover combinations based on the physiography associated with these forest communities. However, not all NLCD landcovers have an analogous forest types in the Hamel classiﬁcation. For example, orchards-vineyards, low-density residential, and transitional-shrubland landcover types provide habitat for many priority species but do not have a speciﬁc forest type association. Therefore, we assigned to orchards-vineyards and low-density residential sites the same SI scores 7 Table 2.—continued Species codea Data source DEM, NLCD, and FIA Landform, landcover, and successional age class NLCD and FIA Early successional patch size (ha) NLCD and NHD Occurrence of water Distance (m) to water NLCD Forest patch size (ha) Landscape composition (percent forest in 1-km radius) Landscape composition (percent forest in 10-km radius) Occurrence of edge Distance (m) to edge Interspersion – 1 landcover class Interspersion – 2 landcover classes Connectivity (km) Grass-open landcover FIA Basal area (m2/ha) Hardwood basal area (m2/ha) Pine basal area (m2/ha) Sawtimber (> 28 cm d.b.h.) tree density (trees/ha) Large (> 50 cm d.b.h) tree density (trees/ha) Large (> 35 cm d.b.h) pine density (trees/ha) Dominant (> 76.2 cm d.b.h.) tree density (trees/ha) Midstory (11–25 cm d.b.h.) density (trees/ha) Snag density (snags/ha) Large (> 30 cm d.b.h.) snag density (snags/ha) Canopy cover (percent) Small stem (< 2.5 cm d.b.h.) density (stems/ha) DEM Slope NHD Distance (m) to stream STATSGO Soil texture Soil moisture KEWA LOWA MIKI NOBO NOPA OROR 1 1 1 1 1 1 3 5 6 2 2 3 2 4 3 5 4 3 2 3 4 2 3 4 4 2 continued as those for deciduous landcovers on the assumption that orchards are composed primarily of deciduous species and low-density residential sites typically are planted with deciduous shade trees. Similarly, we assumed that transitional-shrubland sites are regenerating forests. Where there were transitional-shrubland pixels in ﬂoodplain-valley landforms, we assumed that they were hardwood forest regeneration. Thus, we assigned to them the same SI scores associated with deciduous habitats. On the higher and drier landforms, transitional-shrubland sites likely are dominated by oak and redcedar in the CH and pine in the WGCP, so we assigned to these sites the same SI scores as those for mixed and evergreen forest in each BCR, respectively (Table 3). To assign SI scores to speciﬁc age classes, we used the relative habitat quality values reported in Hamel (1992) for grass-forb, shrub-seedling, and sawtimber seral stages. However, Hamel 8 Table 2.—continued Species codea Data source DEM, NLCD, and FIA Landform, landcover, and successional age class NLCD and FIA Early successional patch size (ha) NLCD and NHD Occurrence of water Distance (m) to water NLCD Forest patch size (ha) Landscape composition (percent forest in 1-km radius) Landscape composition (percent forest in 10-km radius) Occurrence of edge Distance (m) to edge Interspersion – 1 landcover class Interspersion – 2 landcover classes Connectivity (km) Grass-open landcover FIA Basal area (m2/ha) Hardwood basal area (m2/ha) Pine basal area (m2/ha) Sawtimber (> 28 cm d.b.h.) tree density (trees/ha) Large (> 50 cm d.b.h) tree density (trees/ha) Large (> 35 cm d.b.h) pine density (trees/ha) Dominant (> 76.2 cm d.b.h.) tree density (trees/ha) Midstory (11–25 cm d.b.h.) density (trees/ha) Snag density (snags/ha) Large (> 30 cm d.b.h.) snag density (snags/ha) Canopy cover (percent) Small stem (< 2.5 cm d.b.h.) density (stems/ha) DEM Slope NHD Distance (m) to stream STATSGO Soil texture Soil moisture PABU PIWO PRAW 1 1 1 PROW RCWO RHWO 1 1 1 3 2 3 4 3 4 2 2 5 2 3 5 4 3 4 6 5 2 3 2 4 5 4 continued combined sapling- and pole-size trees into a single class, whereas we separated these two successional age classes (a segregation we believed was more appropriate for many of our species). To tease apart the SI scores for sapling and pole age classes, we averaged the value for sapling-pole with shrub-seedling (for sapling) or sawtimber (for pole). This approach assumes that sapling and pole stands have an equal weighting by Hamel in assessing the relative habitat quality for the aggregate age class, and that there is a linear relationship across age classes that allows us to discern the relative inﬂuence of each by simple averaging. After crosswalking Hamel’s forest types and seral stages to our landform-landcover-successional age class matrix, we assigned SI scores to each unique combination based on Hamel’s qualitative assessments. Combinations considered optimal (Hamel 1992) were assigned a value of 1.000; 9 Table 2.—continued Species codea Data source DEM, NLCD, and FIA Landform, landcover, and successional age class NLCD and FIA Early successional patch size (ha) NLCD and NHD Occurrence of water Distance (m) to water NLCD Forest patch size (ha) Landscape composition (percent forest in 1-km radius) Landscape composition (percent forest in 10-km radius) Occurrence of edge Distance (m) to edge Interspersion – 1 landcover class Interspersion – 2 landcover classes Connectivity (km) Grass-open landcover FIA Basal area (m2/ha) Hardwood basal area (m2/ha) Pine basal area (m2/ha) Sawtimber (> 28 cm d.b.h.) tree density (trees/ha) Large (> 50 cm d.b.h) tree density (trees/ha) Large (> 35 cm d.b.h) pine density (trees/ha) Dominant (> 76.2 cm d.b.h.) tree density (trees/ha) Midstory (11–25 cm d.b.h.) density (trees/ha) Snag density (snags/ha) Large (> 30 cm d.b.h.) snag density (snags/ha) Canopy cover (percent) Small stem (< 2.5 cm d.b.h.) density (stems/ha) DEM Slope NHD Distance (m) to stream STATSGO Soil texture Soil moisture SWWA STKI WPWI WEVI WOTH WEWA 1 1 1 1 1 1 2 3 2 2 3 3 4 5 4 5 2 3 2 4 4 3 4 2 continued those considered suitable were assizned a value of 0.667; and those considered marginal had a value of 0333. We assumed that forest types and age classes not assigned a qualitative habitat ranking were not used and assigned to these combinations an SI score of zero. Where a landform-landcover type was represented by more than one of Hamel’s forest types, SI values for the forest types were averaged. For example, deciduous landcover on ﬂoodplain-valley landforms are associated with cove hardwood and elm-ash-cottonwood forest communities. Cove hardwood is suitable (SI = 0.667) for the Acadian ﬂycatcher but elm-ash-cottonwood is optimal (SI = 1.000). Thus, this landform-landcover type combination is assigned a base SI score of 0.834 (i.e., 1.667/2) prior to adjusting for successional age class (Table 4). Finally, we standardized all SI scores in the matrix to ensure that the maximum value was 1.000. 10 Table 2.—continued Species codea Data source DEM, NLCD, and FIA Landform, landcover, and successional age class NLCD and FIA Early successional patch size (ha) NLCD and NHD Occurrence of water Distance (m) to water NLCD Forest patch size (ha) Landscape composition (percent forest in 1-km radius) Landscape composition (percent forest in 10-km radius) Occurrence of edge Distance (m) to edge Interspersion – 1 landcover class Interspersion – 2 landcover classes Connectivity (km) Grass-open landcover FIA Basal area (m2/ha) Hardwood basal area (m2/ha) Pine basal area (m2/ha) Sawtimber (> 28 cm d.b.h.) tree density (trees/ha) Large (> 50 cm d.b.h) tree density (trees/ha) Large (> 35 cm d.b.h) pine density (trees/ha) Dominant (> 76.2 cm d.b.h.) tree density (trees/ha) Midstory (11–25 cm d.b.h.) density (trees/ha) Snag density (snags/ha) Large (> 30 cm d.b.h.) snag density (snags/ha) Canopy cover (percent) Small stem (< 2.5 cm d.b.h.) density (stems/ha) DEM Slope NHD Distance (m) to stream STATSGO Soil texture Soil moisture a YBCU YBCH YTVI YTWA 1 1 1 1 3 3 5 4 2 2 3 4 2 2 3 4 4 Pyle and DeSante 2003; see Table 1. Similarly, we directly assigned SI scores to individual classes for other discrete environmental variables (e.g., occurrence of water). For continuous environmental variables (e.g., canopy cover), we used CurveExpert 1.38 software (Hyams 2001)1 to ﬁt smoothed functions through known data points derived from the literature that quantify the relationship between each speciﬁc environmental factor and HSI scores for particular species. Information sources, assumptions, and functions (type and equation) are detailed in the model accounts. 1 The use of trade, ﬁrm, or corporation names in this publication is for the information and convenience of the reader. Such use does not constitute an oﬃcial endorsement or approval by the U.S. Department of Agriculture or Forest Service of any product or service to the exclusion of others that may be suitable. 11 Table 3.—Crosswalk between landform-landcover class combinations and vegetation types defined in Hamel (1992) Landform Landcover type Hamel vegetation typea Floodplain-valley Low-density residential Transitional-shrubland Deciduous Same as deciduous Same as deciduous Cove hardwoods Elm-ash-cottonwood Loblolly pine-shortleaf pine Mixed pine-hardwood Same as deciduous Oak-gum-cypress Elm-ash-cottonwood Same as deciduous Same as mixed in Central Hardwoods, same as evergreen in West Gulf Coastal Plain/Ouachitas Oak-hickory Cove hardwoods Loblolly pine-shortleaf pine Mixed pine-hardwood Same as deciduous Elm-ash-cottonwood Same as deciduous Same as Mixed in Central Hardwoods, same as evergreen in West Gulf Coastal Plain/Ouachitas Oak-hickory Loblolly pine-shortleaf pine. Also includes Sandhills longleaf pine in West Gulf Coastal Plain/Ouachitas Mixed pine-hardwood Same as deciduous Elm-ash-cottonwood Evergreen Mixed Orchards-vineyards Woody wetlands Terrace-mesic Low-density residential Transitional-shrubland Deciduous Xeric-ridge Evergreen Mixed Orchards-vineyards Woody wetlands Low-density residential Transitional-shrubland Deciduous Evergreen Mixed Orchards-vineyards Woody wetlands a Hamel (1992). To calculate the overall HSI score, we determined the geometric mean of SI scores for sitescale and landscape-scale variables separately and then the geometric mean of these means together. Use of the geometric mean follows recommendations from the published standards for development of HSI models (USDI Fish and Wildl. Serv. 1981). The equal weighting of individual functions within a spatial scale assumes that all variables are required for a habitat to be suitable and that all variables are nonsubstitutable. Further, the equal weighting of functions across scales assumes that site and landscape variables are equally important. The notable exception to use of the geometric mean was for species where both forest patch size and percent forest in the landscape are included as model parameters. In these cases, we used the maximum SI score from these two variables to account for the use of small forest patches by area-sensitive species when small patches are embedded in predominantly forested landscapes (Rosenberg and others 1999). For each species, we solicited at least ﬁve reviewers with an intimate knowledge of the habitat requirements of at least one species. Each reviewer received a standard questionnaire requesting feedback on the appropriateness of the functions included in the model. We revised models based on reviewers’ comments. Model Testing To test the HSI models for reliability, we followed the three-stage framework (calibration, veriﬁcation, and validation) outlined by Brooks (1997). We ﬁrst ensured that the equations 12 Table 4.—Initial assignment of suitability index scores for Acadian flycatcher habitat to landform, landcover type, and successional age classes based on Hamel (1992) Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.834 0.834 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.333 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 1.000 1.000 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.667 0.667 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.333 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.834 0.834 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.333 0.333 0.667 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.333 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.834 0.834 1.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge used to predict SI scores resulted in the full potential range of SI scores given the habitat conditions within each BCR (i.e., calibration). We then used Spearman rank correlation to compare HSI scores to abundance estimates from Breeding Bird Survey (BBS) data summarized by ecological subsection (i.e., veriﬁcation). We ranked subsections by HSI score and BBS abundance for each species and within each BCR independently to compensate for geographical diﬀerences in these regions not explicitly incorporated in the HSI models. We assessed correlations between these variables based on all subsections and based solely on subsections within which each species was detected. The former analysis provides insight into the overall model performance; the latter addresses the potential bias associated with correctly predicting the absence of a rare species in many subsections. Following veriﬁcation, we validated HSI models by developing species-speciﬁc generalized linear models that predicted abundance (as indexed by BBS data) from HSI and BCR predictor variables. We considered HSI models validated if the general linear model was signiﬁcant (P < 0.100) and the coeﬃcient on the HSI predictor variable was both signiﬁcant (P < 0.100) and positive. Detailed results of these analyses are documented in Tirpak and others (2009a). 13 MODEL ACCOUNTS Acadian Flycatcher Status The Acadian ﬂycatcher (Empidonax virescens) is a long-distance migrant found throughout most of the eastern United States. While populations have declined in the northern portion of its range (particularly the Appalachians) over the last 40 years, populations in the South, particularly along the Atlantic and East Gulf Coastal Plains, have increased (Sauer and others 2005). However, the Acadian ﬂycatcher has declined in the WGCP (Table 5), and the FWS classiﬁes this species as John J. Mosesso, images.nbii.gov a Bird of Conservation Concern in the WGCP (Table 1). Similarly, PIF considers the Acadian ﬂycatcher as a planning and responsibility species in the CH (regional combined score of 16). In the WGCP, the ﬂycatcher has a regional combined score of 17, warranting management attention (Table 1). Natural History The Acadian ﬂycatcher is a forest-interior species associated with water throughout most of its range: bottomland hardwood and cypress forests in the Southeast and riparian forests and ravines in the deciduous forests of the Midwest and Northeast (Whitehead and Taylor 2002). This species is found in numerous forest types and uses a variety of tree species for nesting. However, this bird typically is associated with mesic forest stands and avoids upland oak-hickory sites (Klaus and others 2005). Breeding territories are small and average 1 ha (Woolfenden and others 2005). The Acadian ﬂycatcher typically nests in midstory trees and large shrubs in mature forests. Canopy cover typically is dense (> 95 percent; Wilson and Cooper 1998), and the understory usually is sparse (Bell and Whitmore 2000, Wood and others 2004). The Acadian ﬂycatcher is particularly susceptible to forest fragmentation. Aquilani and Brewer (2004) found this species only in forest tracts larger than 55 ha in north-central Mississippi. Blake and Karr (1987) did not observe the Acadian ﬂycatcher in woodlots smaller than 24 ha. In east Texas, the Acadian ﬂycatcher was absent from riparian buﬀer strips less than 70 m wide (Conner and others 2004). Results were similar in Missouri (Peak and others 2004) and Indiana (Ford and others 2001). Even in large forested tracts (> 600 ha), nest predation and parasitism rates may be 10 to 20 percent higher if the surrounding landscape is highly fragmented. Nevertheless, Fauth and Cabe (2005) did not observe signiﬁcant eﬀects of parasitism on a Blue Ridge study site where 75 percent of the landscape was forested, including 45 percent more than 250 m from an edge. Disturbance, whether natural (e.g., tornado or pest outbreak) or anthropogenic (e.g., silvicultural treatments—thinning, selective harvesting, clearcutting, and prescribed burning) reduced the abundance and productivity of the Acadian ﬂycatcher in most landscapes (Artman and others 2001, Duguay and others 2001, Robinson and Robinson 2001, Twedt and others 2001, Prather and Smith 2003, Blake 2005). 14 Table 5.—Trend estimates (percent change per year) for 40 priority landbird species in the Central Hardwoods and West Gulf Coastal Plain/Ouachitas Bird Conservation Regions, 1967 to 2004 (Sauer and others 2005) Central Hardwoods Species Acadian flycatcher American woodcock Bachman’s sparrow Bell’s vireo Bewick’s wren Black-and-white warbler Blue-gray gnatcatcher Blue-winged warbler Brown thrasher Brown-headed nuthatch Carolina chickadee Cerulean warbler Chimney swift Chuck-will’s-widow Eastern wood-pewee Field sparrow Great crested flycatcher Hooded warbler Kentucky warbler Louisiana waterthrush Mississippi kite Northern bobwhite Northern parula Orchard oriole Painted bunting Pileated woodpecker Prairie warbler Prothonotary warbler Red-cockaded woodpecker Red-headed woodpecker Swainson’s warbler Swallow-tailed kite Whip-poor-will White-eyed vireo Wood thrush Worm-eating warbler Yellow-billed cuckoo Yellow-breasted chat Yellow-throated vireo Yellow-throated warbler a b West Gulf Coastal Plain/Ouachitas a Trend P n -0.3 -9.1 --3.2 -6.5 2.3 -1.0 -4.0 -1.4 -0.2 -6.3 -2.6 -0.9 -1.4 -3.2 -0.8 2.7 -0.4 2.6 16.3 -3.1 3.7 -0.9 19.8 1.8 -2.6 0.0 --1.0 ---1.8 -0.4 -0.7 0.4 -1.9 -1.9 0.9 3.8 0.56 0.35 -0.49 0.00 0.21 0.26 0.01 0.00 -0.70 0.00 0.00 0.19 0.00 0.00 0.09 0.08 0.32 0.02 0.16 0.00 0.00 0.01 0.61 0.01 0.00 0.98 -0.09 --0.05 0.20 0.05 0.77 0.00 0.00 0.25 0.00 107 3 -18 61 50 118 62 125 -123 34 124 64 124 125 123 31 108 66 2 125 95 124 5 112 94 52 -115 --71 120 118 44 125 125 99 76 Trend P n -2.0 --b -7.8 -4.7 0.8 -2.9 -0.9 --1.4 -1.4 -2.0 -9.5 -1.1 -1.3 -4.9 -3.7 -1.3 -3.1 -2.2 -1.3 6.4 -4.4 -2.5 -3.0 -0.6 -0.9 -4.4 -5.8 9.0 -3.2 23.5 -6.6 -0.8 -1.4 -2.3 -1.1 1.3 1.1 -0.9 0.05 -0.00 0.03 0.88 0.01 0.36 -0.01 0.18 0.00 0.00 0.15 0.04 0.00 0.01 0.04 0.35 0.00 0.49 0.21 0.00 0.17 0.01 0.48 0.14 0.00 0.00 0.00 0.00 0.23 -0.22 0.19 0.05 0.51 0.00 0.01 0.38 0.65 67 -27 14 11 60 75 -64 52 77 5 76 60 75 45 77 60 73 28 16 75 53 75 63 72 60 53 6 68 26 -11 76 67 28 77 75 62 43 Number of Breeding Bird Survey routes on which trend estimate is based. No trend estimate available. 15 Table 6.—Relationship of landform, landcover type, and successional age class to suitability index scores for Acadian flycatcher habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.050 0.917 1.000 Deciduous 0.000 0.000 0.050 0.917 1.000 Evergreen 0.000 0.000 0.017 0.167 0.333 Mixed 0.000 0.000 0.017 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.050 1.000 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.017 0.333 0.333 Deciduous 0.000 0.000 0.042 0.667 0.834 Evergreen 0.000 0.000 0.017 0.167 0.333 Mixed 0.000 0.000 0.017 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.050 1.000 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.017 0.333 0.333 Deciduous 0.000 0.000 0.033 0.500 0.667 Evergreen 0.000 0.000 0.017 0.167 0.333 Mixed 0.000 0.000 0.017 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.050 1.000 1.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Model Description Our Acadian ﬂycatcher model includes seven variables related to density: landform, landcover type, successional age class, distance to water, canopy cover, forest patch size, and percent forest in a 1-km radius window. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 6). We directly assigned SI scores to these combinations on the basis of habitat suitability data from Hamel (1992) on the relative quality of diﬀerent vegetation types and successional stages for the Acadian ﬂycatcher. However, we reduced SI scores for sapling and evergreen habitats on the basis of data from Hazler (1999). Because the Acadian ﬂycatcher typically is found near water (Whitehead and Taylor 2002), we ﬁt an inverse logistic function to describe the relationship between SI scores for this species and increasing distance to water (SI2; Fig 2). The ﬂycatcher often aligns at least one edge of its 1-ha territory along a stream or wetland (Woolfenden and others 2005). 16 Table 7.—Relationship between distance to water and suitability index (SI) scores for Acadian flycatcher habitat Suitability Index Score 1.0 Distance to water (m)a 0.8 0b 120c 240b 360b 480b 0.6 0.4 0.2 0.0 0 100 200 300 400 500 Distance to Water (m) SI score 1.00 1.00 0.75 0.25 0.00 a Water defined as streams from the National Hydrography Dataset (medium resolution) or classified as water, woody wetlands, or emergent herbaceous wetlands in the National Land Cover Dataset. b Assumed value. c Woolfenden and others (2005). Figure 2.—Relationship between distance to water and suitability index (SI) scores for Acadian flycatcher habitat. Equation: SI score = 1 - (1.049 / (1 + (1664.953 * e -0.021 * distance to water))). Table 8.—Relationship between canopy cover and suitability index (SI) scores for Acadian flycatcher habitat Suitability Index Score 1.0 Canopy cover (percent) 0.8 0a 31b 73b 91b 100a 0.6 0.4 a 0.2 b SI score 0.00 0.00 0.33 1.00 1.00 Assumed value. Prather and Smith (2003). 0.0 0 25 50 75 100 Canopy Cover (%) Figure 3.—Relationship between canopy cover and suitability index (SI) scores for Acadian flycatcher habitat. Equation: SI score = 1.013 / (1.000 + (144082770 * e -0.248 * canopy cover)). Assuming a circular home range, the diameter of the home range (112.8 m) represents the farthest distance from water a bird could be within the home range. On the basis of this assumption, we assigned all locations less than 120 m from water SI scores of 1.000 (Table 7). The Acadian ﬂycatcher also uses sites that are more than 120 m from water but generally are found at lower densities there. Thus, we considered areas 360 m from water (a distance of three home range diameters) as having an SI score that is one-quarter of the optimal value (0.250) and sites at least 480 m from water as nonhabitat (SI score of zero). The habitat suitability model for the Acadian ﬂycatcher also included canopy closure (SI3) as a variable because of the strong aﬃnity of this species for closed-canopy forests (Prather and Smith 2003). For this variable, we used a logistic function (Fig. 3) to extrapolate between known break points in the canopy cover-relative density relationship (Table 8). 17 We also included forest patch size (SI4) as a variable because of the sensitivity of the Acadian ﬂycatcher to fragmentation (Robbins and others 1989) and increasing edge density (Parker and others 2005). We used a logarithmic function (Fig. 4) to describe the relatively quick increase in suitability of a forest patch with increasing area (Robbins and others 1989) (Table 9). We assumed that 312 ha, the minimum forest patch size on which Wallendorf and others (2007) always observed the Acadian ﬂycatcher, was representative of optimal habitat (SI score = 1.000). Nevertheless, the eﬀects of forest patch size on suitability are inﬂuenced by the percentage of forest in the landscape. In predominantly forested landscapes, small forest patches that may not be used in predominantly nonforested landscapes may provide habitat due to their proximity to large forest blocks (Rosenberg and others 1999). To capture this relationship, we ﬁt a logistic function (Fig. 5) to data (Table 10) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoints between these classes (30 and 70 percent forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. We used the maximum value of SI4 or SI5 to assess area sensitivity and to account for small patches in predominantly forested landscapes and large patches in predominantly nonforested landscapes. To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure attributes (SI1 and SI3) and landscape attributes (maximum value of SI4 or SI5 and SI2) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI3)0.500 * (Max(SI4 or SI5) * SI2)0.500)0.500 Veriﬁcation and Validation The Acadian ﬂycatcher was found in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.47) between average HSI score and mean BBS abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the Acadian ﬂycatcher was signiﬁcant (P = 0.095; R2 = 0.054), and the coeﬃcient on the HSI predictor variable was both positive (β = 4.250) and signiﬁcantly diﬀerent from zero (P = 0.043). Therefore, we considered the HSI model for the Acadian ﬂycatcher both veriﬁed and validated (Tirpak and others 2009a). 18 Table 9.—Relationship between forest patch size and suitability index (SI) scores for Acadian flycatcher habitat Suitability Index Score 1.0 Forest patch size (ha) 0.8 0.2a 15a 312b 0.6 a b 0.4 SI score 0.0 0.5 1.0 Robbins and others (1989). Wallendorf and others (2007). 0.2 0.0 0 100 200 300 Forest Patch Size (ha) Suitability Index Score Figure 4.—Relationship between forest patch size and suitability index (SI) scores for Acadian flycatcher habitat. Equation: SI score = 0.174 * ln(forest patch size) + 0.010. 1.0 Table 10.—Relationship between local landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for Acadian flycatcher habitat 0.8 Local landscape composition 0.6 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 5.—Relationship between landscape composition and suitability index (SI) scores for Acadian flycatcher habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * (landscape composition )). 0a 10a 20a 30b 40a 50b 60a 70b 80a 90b 100a a b SI score 0.00 0.00 0.05 0.10 0.25 0.50 0.75 0.90 0.95 1.00 1.00 Assumed that value. Dononvan and others (1997). 19 American Woodcock Status The American woodcock (Scolopax minor) is a popular gamebird found throughout the eastern United States and southeastern Canada. Although this species breeds primarily in the northern portion of its continental range, small numbers breed regularly throughout the wintering range in the Southeast. Singing ground surveys and wing collections from northern latitudes in U.S. Fish & Wildlife Service the Central United States document annual 1.8 percent declines in woodcock since 1968 (Kelley 2003). The status of the relatively small breeding population in the Southeast is unknown. Natural History The American woodcock breeds in early successional habitat throughout its range (Keppie and Whiting 1994). Typically, these young forest stands are on moist, uncompacted soils that allow the woodcock to probe for earthworms, the bird’s preferred food (Steketee 2000). Equally important is an interspersion of the forest with openings that provide sites for both courtship displays and roosting (Sepik and Derleth 1993). Openings used by woodcock in Maine generally were at least 1.2 ha (Dunford and Owen 1973). Given the aﬃnity of the woodcock for openings and early successional habitat, Sprankle and others (2000) recommended even-age forest management in rotational blocks to ensure that both habitat requirements are met. Most of the available quantitative information on breeding habitat for the American woodcock is from the Northeast, particularly Maine and Pennsylvania (Straw and others 1986, McAuley and others 1996). Shrub cover generally is high (75 to 87 percent; Morgenweck 1977), while overstory cover typically is moderate (50 to 64 percent; Dunford and Owen 1973, Gregg and others 2000). Nests are in young forest stands (Morgenweck 1977). McAuley and others (1996) compared nest sites to random sites and found lower basal area and fewer coniferous saplings, but higher densities of deciduous saplings and shrub stems around nests sites. Young broods inhabit young to mid-age forest interspersed with openings; older broods occupy sites with greater basal area but fewer mature trees (Morgenweck 1977). Many habitat variables have been associated with the presence of woodcock (Storm and others 1995; Klute and others 2002). Landcover variables were the best predictors at ﬁne scales whereas indices of landscape heterogeneity were the most important predictors at large spatial scales (Klute and others 2000). Murphy and Thompson (1993) developed a model to predict the density of males on singing grounds in central Missouri that contained small stem density (≤ 2.5 cm d.b.h.), tree density (> 2.5 cm d.b.h.), and ﬁeld size as predictor variables. 20 Table 11.—Relationship of landform, landcover type, and successional age class to suitability index scores for American woodcock habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 1.000 0.667 0.333 Deciduous 0.000 0.000 1.000 0.667 0.333 Evergreen 0.000 0.000 0.500 0.250 0.125 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.667 0.333 0.167 Orchard-vineyard 0.000 0.000 0.667 0.333 0.167 Woody wetlands 0.000 0.000 1.000 0.667 0.333 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.834 0.500 0.250 Deciduous 0.000 0.000 0.834 0.500 0.250 Evergreen 0.000 0.000 0.400 0.200 0.100 Mixed 0.000 0.000 0.500 0.250 0.125 Orchard-vineyard 0.000 0.000 0.500 0.250 0.125 Woody wetlands 0.000 0.000 0.834 0.500 0.250 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.750 0.400 0.167 Deciduous 0.000 0.000 0.750 0.400 0.167 Evergreen 0.000 0.000 0.333 0.167 0.083 Mixed 0.000 0.000 0.400 0.200 0.100 Orchard-vineyard 0.000 0.000 0.500 0.417 0.000 Woody wetlands 0.000 0.000 0.750 0.400 0.167 Model Description The American woodcock HSI model includes seven variables: landform, landcover, successional age class, small stem density (< 2.5 cm d.b.h.), composition of appropriately sized foraging-nesting and courtship-roosting habitat patches in the landscape, soil moisture, and soil texture. The ﬁrst suitability function combines landform, landcover type, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 11). Because the woodcock prefers moist habitats with high deciduous stem densities, we assigned the highest SI scores to sapling-aged transitional, deciduous, and woody wetland cover types in ﬂoodplain-valley landforms. We considered mixed and evergreen forests as well as xericridge landforms as poor habitat for the American woodcock. We included small stem density (SI2) as a model function because the woodcock relies on vertical structure to provide security from predators as it forages, nests, and loafs during the day. McAuley and others (1996) summarized habitat attributes around woodcock nest sites from seven studies in which stem density ranged from 5,051 to 49,250 stems per ha. Due to the relatively small sample size and the lack of geographic representation within the samples 21 Table 12.—Influence of small stem (< 2.5 cm d.b.h.) density (stems*1,000/ha) on suitability index (SI) scores for American woodcock habitat Suitability Index Score 1.0 Small stem density 0.8 0 accc 0.00 3.767b 0.6 0.4 0.25 27.125 c 0.90 49.250 d 1.00 a Assumed value. b Murphy and Thompson (1993). c McAuley and others (1996). d Coon and others (1982). 0.2 0.0 0 10 20 30 40 50 Small Stem Density (stems * 1,000/ha) Figure 6.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems*1000/ha) and suitability index (SI) scores for American woodcock habitat. Equation: SI score = 1.029 * (0.998 – e -0.076 * (small stem density / 1000)). (both New York and Pennsylvania are represented twice), we used the midpoint of this range rather than the average to summarize these data. With three of the studies observing stem densities of at least 44,000 and three observing densities of approximately 14,000 stems per ha (+/- 600 stems/ha), we believed there was adequate evidence to assign to the midpoint of this range (27,125 stems/ha) a higher SI score than average (0.500). Therefore, we assigned 27,125 stems per ha an SI score of 0.900, the maximum stem density (49,250) an SI score of 1.000 and the minimum density (3,767 stems/ha, as reported by Murphy and Thompson [1993]) an SI score of 0.250 (Table 12). We ﬁt a logistic function through these data points to quantify the small stem density-SI score relationship (Fig. 6). The next two variables relate to the minimum size of habitat patches used by the American woodcock. Movement rates within diurnal foraging and nesting habitats often are low, resulting in small diurnal home ranges (≤ 0.3 ha; Hudgins and others 1985). Conversely, the woodcock displays and roosts in relatively large openings at night (≥ 1.6 ha; Keppie and Whiting 1994). We used these data to establish minimum area thresholds for forests and openings, respectively. Nevertheless, the ultimate suitability of either of these habitat types is related to their interspersion with one another, as the woodcock requires both. Ideally, these habitats should be separated by less than 400 m (Hudgins and others 1985) even though the average home range may be at least 74 ha (485-m radius; Keppie and Whiting 1994). Because home ranges may encompass areas of nonhabitat, the American woodcock sometimes is found where the proportion of these habitat types within a typical home range is relatively small (e.g., 0.1; Table 13). We assumed that the woodcock derives greater beneﬁt from increasing proportions of early successional forest habitat than ﬁeld habitat within its home ranges due to greater foraging opportunities and increased protection from predators. Thus, our table deﬁning the relationship between landscape composition (SI3) 22 SI score Table 13.—Suitability index scores for American woodcock habitat based on composition of open and forest habitat within 500-m radius Proportion opena Proportion forestb 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.10 0.05 0.05 0.05 0.05 0.05 0.00 0.00 0.00 0.00 0.00 0.20 0.10 0.10 0.10 0.10 0.10 0.05 0.05 0.05 0.00 0.30 0.20 0.20 0.20 0.20 0.20 0.10 0.10 0.05 0.40 0.40 0.40 0.40 0.40 0.40 0.20 0.10 0.50 0.60 0.60 0.60 0.60 0.60 0.40 0.60 0.80 0.80 0.80 0.80 0.80 0.70 1.00 1.00 1.00 1.00 0.80 1.00 1.00 1.00 0.90 1.00 1.00 1.00 1.00 a Merged grasslands, pasture/hay, fallow, urban/recreational grasses, emergent herbaceous wetlands, grass-forb, and shrubseedling forests ≥1.6 ha. b Sites with a positive SI1 score (Table 11) and ≥ 0.3 ha. and SI scores shows greater increases in suitability with relatively modest increases in diurnal habitat compared to the increases in suitability associated with similar proportional increases in openings. Soil properties also inﬂuence American woodcock habitat suitability. This species feeds nearly exclusively on earthworms, which it probes for preferentially in moist loamy soils (Rabe and others 1983). Because soils with excessive clay or sand contain insuﬃcient, accessible earthworms with which to support a foraging woodcock, we included both soil texture (SI4) and soil drainage (SI5) as variables in the habitat suitability model. We used the STATSGO database to deﬁne soil characteristics. Soil texture classes from STATSGO were crosswalked to soil texture classes from the soil triangle (Table 14) and then assigned SI scores on the basis of texture descriptions in Rabe and others (1983) (Table 15). We also assumed that soil drainage class was associated with soil moisture content and similarly assigned SI scores to these drainage classes (Table 16) based on observations from Rabe and others (1983), who documented higher probing rates in soils with greater moisture contents. To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure (SI1 and SI2) and landscape factors (SI3, SI4 and SI5) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI2)0.500 * (SI3 * SI4 * SI5)0.333)0.500 23 Table 14.—Crosswalk of soil texture classes defined in STATSGO soil database to soil texture triangle classes STATSGO soil texture class Soil texture triangle class Clayey Clay Clayey over loamy Clay Clayey-skeletal Clay Coarse-loamy Sandy loam Coarse-silty Sandy loam Fine Silt Fine-loamy Table 15.—Suitability index (SI) scores for American woodcock habitat based on soil texture triangle classes Soil texture triangle class SI score Clay 0.0a Silty clay 0.0a Silty clay loam 0.2a Silt loam 0.4a Silt 0.0a Loam 1.0b Silt loam Sandy loam 0.8b Fine-loamy over clayey Fine-loamy over sandy or sandyskeletal Silty clay loam Loamy sands 0.0a Sands 0.0b Sandy clay loam 0.4a Fine-silty Silt Fine-silty over clayey Silt Sandy clay 0.0a Loamy Loam Clay loam 0.1b Loamy-skeletal Loam None 0.0a Loamy-skeletal over clayey Loam Not used None Sandy Sand Silt loam Very-fine Silty clay All others None a b Assumed value. Rabe and others (1983). Table 16.—Suitability index (SI) scores for American woodcock habitat based on soil moisture, as defined by drainage class in the STATSGO soil database Soil moisture Very poorly 1.0a Poorly 1.0a Somewhat poorly 0.5a Moderately well 0.1a Well 0.0a Somewhat excessively 0.0a Excessively 0.0a a Rabe and others (1983). Veriﬁcation and Validation The American woodcock was observed only in 50 of the 88 subsections within the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.36) between average HSI score and mean BBS route abundance across all subsections. When the 38 subsections in which the American woodcock was not found were removed from the analysis, the correlation not only remained signiﬁcant (P ≤ 0.001) but also was more strongly positive (rs = 0.68). Thus, the HSI model is predicting habitat for this species in subsections where it was not detected on BBS routes. The generalized linear model predicting BBS abundance from BCR and HSI for the American woodcock was signiﬁcant (P ≤ 0.001; R 2 = 0.218), and the coeﬃcient on the HSI predictor variable was both positive (β = 0.090) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the American woodcock both veriﬁed and validated (Tirpak and others 2009a). 24 SI score Bachman’s Sparrow Status Bachman’s sparrow (Aimophila aestivalis) is a resident bird associated with pine savannas and other open habitats throughout the Southeastern United States. Although its range expanded north to include Illinois, Indiana, and Ohio at the turn of the 20th century (likely in response to widespread land clearing), the range of this species has contracted steadily over the last 100 years. Today, the Bachman’s sparrow is restricted to the extreme Southeast. BBS data from the central United States U.S. Forest Service indicates signiﬁcant annual declines (8.1 percent) over the past 40 years; declines have been particularly steep since 1980 (20.8 percent/year). This species is a Bird of Conservation Concern in both the CH and WGCP (Table 1). Similarly, this bird has a regional combined score of 20 in both regions, and PIF considers this species in need of critical recovery in the CH and immediate management in the WGCP (Table 1). Natural History Bachman’s sparrow occupies two primary habitats in the Southeast: mature (> 80 year old) pine stands that are frequently burned (< 3-year burn interval) and recently cutover areas (< 5 year old; Dunning and Watts 1990). However, productivity is lower in these latter habitats (one vs. three oﬀspring/pair/year; Liu and others 1995, Perkins and others 2003a). On the basis of this lower productivity and the poor colonizing ability of this species—suitable clearcut habitats more than 3 km from a source population generally remained unoccupied in South Carolina (Dunning and others 1995)—Tucker and others (2004) considered Bachman’s sparrow as endemic to mature longleaf pine stands. In all studies of Bachman’s sparrow habitat, two features are identiﬁed repeatedly: a dense grass understory and an open overstory, both of which are maintained through frequent ﬁres (Haggerty 1998, Plentovich and others 1998, Tucker and others 2004, Wood and others 2004). Stands managed for the red-cockaded woodpecker via prescribed burning typically provide excellent habitat for the Bachman’s sparrow as well because the ﬁres are frequent enough to suppress dense woody understories and maintain sparse canopies (Wilson and others 1995, Plentovich and others 1998, Provencher and others 2002, Wood and others 2004). Model Description Our habitat suitability model for the Bachman’s sparrow includes six variables: landform, landcover type, successional age class, forest patch size, canopy cover, and connectivity. The ﬁrst suitability function combines landform, landcover type, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 17). We directly assigned SI scores to these combinations on the basis of data from Hamel (1992) on the relative quality of diﬀerent vegetation types in diﬀerent successional stages for this species. 25 Table 17.—Relationship of landform, landcover type, and successional age class to suitability index scores for Bachman’s sparrow habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 1.000 0.333 0.000 0.000 1.000 Deciduous 1.000 0.333 0.000 0.000 0.000 Evergreen 1.000 0.333 0.000 0.000 1.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 1.000 0.333 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 1.000 0.333 0.000 0.000 1.000 Deciduous 1.000 0.333 0.000 0.000 0.000 Evergreen 1.000 0.333 0.000 0.000 1.000 Mixed 1.000 0.333 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 1.000 0.333 0.000 0.000 1.000 Deciduous 1.000 0.333 0.000 0.000 0.000 Evergreen 1.000 0.333 0.000 0.000 1.000 Mixed 1.000 0.333 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 We also included forest patch size (SI2) as a variable because of the relatively large home range for this species (mean = 2.5 ha; Haggerty 1998). Home ranges varied among regions and habitat types (reviewed in Mitchell 1998). They were slightly larger in evergreen stands (4.8 ha) than in ephemeral, early successional habitats (2.2 ha). We ﬁt a logistic function (Fig. 7) through these data points, assuming that the former represented a stand area that would be occupied reliably and that the latter value was a minimum below which the sparrow would be absent (Table 18). We included canopy cover (SI3) as a third suitability function to satisfy the two-fold requirement for open canopies and dense understories, two habitat components often well correlated (Table 19). Haggerty (1998) observed an average canopy cover of 9.5 percent at sites occupied by the Bachman’s sparrow and 40 percent canopy cover at unoccupied sites. Wood and others (2004) observed 20 times more Bachman’s sparrows in habitats with 25 to 50 percent canopy cover than sites with 50 to 75 percent cover. We ﬁt an inverse logistic function to these data to extrapolate values between these known points (Fig. 8). Because this resident species is restricted to a specialized habitat, occupancy of a site by the Bachman’s sparrow is aﬀected by the ability of dispersers to colonize it. This ability is 26 Table 18.—Relationship between forest patch size and suitability index (SI) scores for Bachman’s sparrow habitat Suitability Index Score 1.0 Forest patch size (ha) 0.8 0.0 2.2b 0.0 3.5 b 0.5 4.8 b 1.0 6.0a 1.0 0.0 0.6 0.4 a 0.2 b 0.0 0.0 2.5 5.0 SI score a Assumed value. Stober (1996), reviewed in Mitchell (1998). 7.5 Forest Patch Size (ha) Figure 7.—Relationship between forest patch size and suitability index (SI) scores for Bachman’s sparrow habitat. Equation: SI score = 1.000 / (1 + (699817.120 * e -3.845 * forest patch size)). Table 19.—Relationship between canopy cover and suitability index (SI) scores for Bachman’s sparrow habitat Suitability Index Score 1.0 Canopy cover (percent) 0.8 0.6 0.4 0.0 a 9.5 b SI score 1.00 1.00 37.5 c 1.00 62.5 c 0.05 100.0a 0.00 a Assumed value. Haggerty (1998). c Wood and others (2004). 0.2 b 0.0 0 25 50 75 100 Canopy Cover (%) Figure 8.—Relationship between canopy cover and suitability index (SI) scores for Bachman’s sparrow habitat. Equation: SI score = 1 - (1.000 / (1 + (126024970 * e -0.3455 * canopy cover))). directly aﬀected by the connectivity (or conversely the isolation) of habitat patches (SI4). Birds are unable to colonize clearcuts more than 3 km distant before succession renders habitat conditions within them unsuitable (Dunning and others 1995). Although isolation also may aﬀect the occupancy of mature evergreen stands, habitat conditions within them are less ephemeral. Thus, the Bachman’s sparrow has a potentially longer time to colonize these stands. To compensate for this diﬀerential temporal window in accessibility, we used a 15-km distance threshold to ﬁt a longer tail to the function relating connectivity of patches to their suitability as Bachman’s sparrow habitat (Table 20, Fig. 9). We also assumed that source populations were restricted to mature evergreen forest stands with a preliminary overall SI score (calculated from SI1, SI2, and SI3) that was greater than 0.8. 27 Suitability Index Score 1.0 Table 20.—Relationship between distance to nearest evergreen sawtimber habitat with initial suitability index (SI) score > 0.8 and SI scores for Bachman’s sparrow habitat 0.8 Habitat connectivity (km) 0 0.6 a 1.00 6b 15 0.4 0.25 b a Dunning and others (1995). b Assumed value. 0.2 0.0 0 5 10 15 Distance to Mature Evergreen Stand (km) Figure 9.—Relationship between distance to nearest evergreen sawtimber habitat with initial suitability index (SI) score >0.8 and SI scores for Bachman’s sparrow habitat. Equation: SI score = 1 / (1.000 + (0.002 * (distance to evergreen sawtimber habitat with initial SI score >0.8)4.066)). To calculate the overall HSI score, we calculated the geometric mean of the two SIs related to forest structure (SI1 and SI3) and landscape attributes (SI2 and SI4) separately and then the geometric mean of these values together. Overall HSI = ((SI1 * SI3)0.500 * (SI2 * SI4)0.500)0.500 Veriﬁcation and Validation Bachman’s sparrow was found only in 29 of the 88 subsections within the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.62) between average HSI score and mean BBS route abundance across all subsections. However, when subsections where the Bachman’s sparrow was not found were removed from the analysis, the relationship was not signiﬁcant (rs = 0.24; P = 0.208). Thus, the HSI model predicts the absence of the Bachman’s sparrow better than its abundance in subsections where it is found. The generalized linear model predicting BBS abundance from BCR and HSI for the Bachman’s sparrow was signiﬁcant (P ≤ 0.001; R 2 = 0.567), and the coeﬃcient on the HSI predictor variable was both positive (β = 0.908) and signiﬁcantly diﬀerent from zero (P = 0.079). Therefore, we considered the HSI model for the Bachman’s sparrow both veriﬁed and validated (Tirpak and others 2009a). 28 SI score 0.00 Bell’s Vireo Status Bell’s vireo (Vireo bellii) is a scrubland specialist that reaches the eastern limit of its range in the CH and WGCP. Throughout both regions this species has declined over the past 40 years, with the most severe declines in the southern portion of the eastern range (-4.7, -6.6, and -10.1 percent annually in Missouri, Oklahoma, and the OzarkOuachita Plateau, respectively; Sauer and others 2005). Steve Maslowski, U.S. Fish & Wildlife Service Bell’s vireo has a regional combined score of 15 in the CH and 16 in the WGCP, and PIF considers the species as requiring management attention in both regions (Table 1). The FWS also recognizes Bell’s vireo as a Bird of Conservation Concern in both BCRs (Table 1). Natural History Bell’s vireo is a small, Neotropical migrant associated with dense, low, shrubby vegetation (Brown 1993). It uses a variety of early successional scrubland habitats that meet these requirements (e.g., riparian woods, brushy ﬁelds, and regenerating forest). Most of the research on this species was conducted in the West, where Bell’s vireo is alternately described as a riparian specialist (particularly the federally endangered subpopulation of least Bell’s vireo in California) or a scrub-shrub generalist. This bird nests in dense shrub or understory vegetation 0.5 to 1.5 m above the ground, making its nests susceptible to both terrestrial and avian predators. Predation and brood parasitism are the primary causes of nest failure (Budnik and others 2000, 2002; Powell and Steidl 2000). Increasing the density of large shrub patches may improve Bell’s vireo habitat in Missouri (Budnik and others 2002). Model Description The model for Bell’s vireo includes six variables: landform, landcover, successional age class, interspersion of forest and open areas, edge, and small stem density. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 21). We directly assigned SI values to these combinations on the basis of data from Hamel (1992) relating vegetation types and successional age class to habitat suitability estimates for Bell’s vireo. Both landcover and age class data were used to identify upland shrublands in grassland landscapes, the preferred habitat for this species in its eastern range (Budnik and others 2000). We used a 10-ha moving window (an average home range; Budnik and others 2000) to assess the interspersion of shrubland and grassland habitats (SI2). We assumed that an area containing 50 percent of each habitat type was ideal (Table 22). To extrapolate from this point we used broad incremental changes in habitat suitability (20 percent) and applied these symmetrically to 10-percent incremental changes in the proportion of scrubland or grassland. Landscapes lacking shrublands or grasslands were unsuitable and assigned an SI score of zero. 29 Table 21.—Relationship of landform, landcover type, and successional age class to suitability index scores for Bell’s vireo habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.500 0.250 0.125 0.000 0.000 Deciduous 0.500 0.250 0.125 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.500 0.250 0.125 0.000 0.000 Woody wetlands 0.500 0.500 0.250 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.500 1.000 0.750 0.000 0.000 Deciduous 0.250 0.500 0.375 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.250 0.500 0.375 0.000 0.000 Woody wetlands 1.000 0.500 0.250 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.500 1.000 0.750 0.000 0.000 Deciduous 0.500 1.000 0.750 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.500 1.000 0.750 0.000 0.000 Woody wetlands 1.000 0.500 0.250 0.000 0.000 Table 22.—Relative composition of scrubland and grassland within 10-ha moving window on suitability index scores for Bell’s vireo habitat Proportion grasslanda Proportion scrubland b 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.1 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.0 0.0 0.1 0.2 0.4 0.4 0.4 0.4 0.4 0.3 0.0 0.1 0.2 0.4 0.6 0.6 0.6 0.6 0.4 0.0 0.2 0.4 0.6 0.8 0.8 0.8 0.5 0.0 0.2 0.4 0.6 0.8 1.0c 0.6 0.0 0.2 0.4 0.6 0.8 0.7 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.9 0.0 0.2 1.0 0.0 a Grasslands/herbaceous, pasture/hay, and grass-forb successional age class. Shrub-seedling and sapling successional age classes. c Budnik and others (2000); all other values assumed. b 30 Table 23.—Influence of edge occurrence on suitability index (SI) scores for Bell’s vireo habitat Suitability Index Score 1.0 3 × 3 pixel window around forest pixel includes fielda 0.8 Yes SI score b 1.0 No 0.6 0.0 a Field defined as any shrub-seedling or grass-forb age class pixel, natural grasslands/herbaceous, or pasture/ hay. Forest defined as any used sapling age class pixel of transitional, shrublands, deciduous, orchard, or woody wetlands. b Grass-forb and seedling-shrub habitats used regardless of edge. 0.4 0.2 0.0 0 10 20 30 40 50 Small Stem Density (stems * 1,000/ha) Figure 10.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for Bell’s vireo habitat. Equation: SI score = 1.001 / (1.000 + (85.005 * e -0.222 * (small stem density / 1000))). Table 24.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1,000/ha) and suitability index (SI) scores for Bell’s vireo habitat Small stem density 0 a SI score 0.00 10a 0.10 25 a 0.75 49 b 1.00 a Assumed value. b Farley (1987). Bell’s vireo uses a variety of young woody habitats (Brown 1993); however, birds also nest along the edges of sapling stands and in hedgerows (Budnik and others 2002). Therefore, we included edge (SI3) as a parameter in the Bell’s vireo HSI model. To identify edges, we examined the eight pixels surrounding each sapling age class pixel to determine whether any were classiﬁed as shrub-seedling or grass-forb age class forest or as a nonforest landcover class. If so, the central pixel in the 3 × 3 pixel window (90 x 90 m) was assigned an SI score of 1.000; if not, it was assigned a zero. We assigned to grass-forb and shrub-seedling pixels an SI score of 1.000 regardless of edge (Table 23). Similarly, we always assigned to pole and sawtimber pixels an SI score of zero regardless of edge. We also included small stem density (SI4) as a component of the overall Bell’s vireo HSI model because of the importance of dense woody shrub cover for this species. Farley (1987) measured an average of 9.8 stems greater than 2 mm per 1-m diameter plot (approximately 392,000 stems/ha) in Bell’s vireo territories. This relatively high stem value included woody and nonwoody stems of all sizes greater than 2 mm; therefore, we assumed that that only one-eighth of these stems (49,000 = ⅛ * 392,000) were woody and less than 2.5 cm d.b.h. and that this value represented optimal habitat (Table 24, Fig. 10). To calculate the overall HSI score for Bell’s vireo, we ﬁrst determined the geometric mean of the suitability indices related to forest structure (SI1 and SI4) and landscape attributes (SI2 and SI3) separately and then determined the geometric mean of these values together. Because SI3 applies only to sapling habitats, HSI scores were calculated diﬀerently for sapling 31 successional age class stands than for grass-forb or shrub-seedling successional age class stands. To determine the overall SI score across the entire BCR, we added suitability scores from individual age classes across the entire landscape. For grass-forb and shrub-seedling habitats: HSIGF and SS = (((SI1 * SI4)0.500 ) * (SI2))0.500 For sapling habitats: HSISap = ((SI1 * SI4)0.500 * (SI2 * SI3)0.500 )0.500 Overall HSI = HSIGF and SS + HSISap Veriﬁcation and Validation Bell’s vireo was found in 54 of the 88 subsections within the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.44) between average HSI score and mean BBS route abundance across all subsections. Removing subsections in which Bell’s vireo was not observed had a minimal eﬀect on these results (rs = 0.46; P ≤ 0.001). The generalized linear model predicting BBS abundance from BCR and HSI for the Bell’s vireo was signiﬁcant (P = 0.042; R2 = 0.072); however, the coeﬃcient on the HSI predictor variable was negative (β = -19.906) and not signiﬁcantly diﬀerent from zero (P = 0.544). Therefore, we considered the HSI model for the Bell’s vireo veriﬁed but not validated (Tirpak and others 2009a). 32 Bewick’s Wren Status Bewick’s wren (Thryomanes bewickii) was once a common resident throughout the Southeast and mid-Atlantic. However, its range has contracted steadily over the last century and today this species is virtually absent east of the Mississippi River (Kennedy and White 1997). BBS data from FWS Region 4 indicates that populations have Dave Menke, U.S. Fish & Wildlife Service declined by 12.8 percent per year over the last 40 years (Sauer and others 2005). The decline of this species coincided with the range expansion of the house wren, which often destroys Bewick’s wren nests in areas where the species’ ranges overlap (Kennedy and White 1996). Bewick’s wren is a Bird of Conservation Concern in both the CH and WGCP (Table 1). PIF identiﬁes the species as requiring both critical recovery in the WGCP (regional combined score = 16) and immediate management attention in the CH (regional combined score = 15). Natural History Bewick’s wren is a small resident passerine that breeds in a variety of vegetation types, including brushy areas, scrub and thickets in open country, and open and riparian woodlands (Kennedy and White 1997). This plasticity has produced conﬂicting reports of habitat associations in the literature (e.g., dry vs. riparian, open woodlands vs. shrub thickets). However, this species likely responds most strongly to the availability of nest sites. Bewick’s wren nests in cavities or opportunistically in crevices up to 10 m high. In the eastern portion of its range, this bird often lives near human habitation, particularly farmland. As mentioned, population declines of this species may be partly the result of competition with the house wren (Kennedy and White 1996). Bewick’s wren is found primarily in grassland scrub while the house wren occurs primarily in secondary growth on abandoned agricultural land and in residential areas. Both species exploit the full range of these habitat types, and populations of both expanded as these latter types increased. However, as scrub habitats declined, Bewick’s wren may have declined because its primary source habitat no longer was abundant. Model Description Our model for Bewick’s wren includes ﬁve variables: landform, landcover, successional age class, interspersion of forest and open habitats, and snag density. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 25). We then directly assigned an SI score to these combinations on the basis of data from Hamel (1992) on the relative quality of Bewick’s wren habitat based on vegetation type and successional age class. We also considered as important for this species the interspersion of forest and grassland habitats (SI2), as Bewick’s wren is most abundant in semi-open areas containing about 40 percent woodland (Pogue and Schnell 1994; Table 26). We relied on data from Pogue and Schnell to deﬁne SI values along the diagonal axis of our interspersion table (where forest and grassland totaled 100 percent) and completed the table from these values. 33 Table 25.—Relationship of landform, landcover type, and successional age class to suitability index scores for Bewick’s wren habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 1.000 1.000 0.500 0.000 0.000 Transitional-shrubland 1.000 1.000 0.500 0.000 0.000 Deciduous 0.500 0.500 0.250 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 1.000 1.000 0.500 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 1.000 1.000 0.500 0.000 0.000 Transitional-shrubland 1.000 1.000 0.500 0.000 0.000 Deciduous 0.500 0.500 0.250 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 1.000 1.000 0.500 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 1.000 1.000 0.500 0.000 0.000 Transitional-shrubland 1.000 1.000 0.500 0.000 0.000 Deciduous 0.500 0.500 0.250 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 1.000 1.000 0.500 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Table 26.—Influence of interspersion between forest and open habitats (as indexed by relative composition within 10-ha moving window) on suitability index scores for Bewick’s wren habitat Proportion opena Proportion forestb 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00c 0.1 0.00 0.00 0.05 0.10 0.10 0.20 0.20 0.20 0.20 0.20c 0.40 0.40 c 0.80 0.80 c 0.80 1.00 c c 0.2 0.3 0.4 0.00 0.00 0.00 0.05 0.05 0.05 0.10 0.20 0.20 0.15 0.25 0.40 0.20 0.60 0.5 0.00 0.05 0.20 0.40 0.80 0.6 0.00 0.10 0.20 0.40 0.80c 0.7 0.00 0.10 0.20 0.40c 0.8 0.00 0.10 0.20c 0.9 0.00 0.10c 1.0 c a 0.00 0.25 1.00 1.00 0.80 0.40 Open = grasslands, herbaceous planted (pasture-hay, fallow, and urban-recreational grasses), emergent herbaceous wetlands. Forest = forested upland, low-density residential, shrubland, transitional, and woody wetlands. c Pogue and Schnell (1994); all other values assumed. b 34 Table 27.—Influence of snag density on suitability index scores for Bewick’s wren habitat Suitability Index Score 1.0 Snag density (snags/ha) 6.2 0.8 0.6 a SI score 0.128 16.4 b 0.500 52.8 a 1.000 a Rumble and Gobeille (2004). b Sedgwick and Knopf (1990). 0.4 0.2 0.0 0 10 20 30 40 50 Snag Density (snags/ha) Figure 11.—Relationship between snag density and suitability index (SI) score for Bewick’s wren habitat. Equation: SI score = 1.0011 / (1 + (21.9129 * e -0.1881 * snag density)). We also included snag density (SI3) in our model of Bewick’s wren habitat because as a secondary cavity nester, this species responds strongly to nest-site availability. We assumed that higher snag densities would decrease competition with other cavity nesters, improving habitat quality. Speciﬁc data relating snag density to Bewick’s wren habitat suitability were not available, so we assumed that the average snag density observed by Sedgwick and Knopf (1990) (16.4 snags/ha) within home ranges of the house wren, a secondary cavity nester of similar size, represented average habitat suitability (SI score = 0.500) for the Bewick’s wren. We coupled this information with data from Rumble and Gobeille (2004) (Table 27) on the relative density of the house wren in habitats with diﬀerent snag densities to build a logistic function quantifying the relationship between habitat suitability and snag density (Fig. 11). To calculate the overall HSI score, we ﬁrst calculated the geometric mean of the two suitability indices related to forest structure attributes (SI1 and SI3), and then the geometric mean of this result and the SI related to interspersion (SI2). Overall HSI = ((SI1 * SI3)0.500 * SI2)0.500 Veriﬁcation and Validation Bewick’s wren was found in 74 of the 88 subsections within the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.40) between average HSI score and mean BBS route abundance across subsections. However, this relationship was weaker (rs = 0.35; P = 0.002) when subsections in which the Bewick’s wren was not detected were removed from the analysis. The generalized linear model predicting BBS abundance from BCR and HSI for the Bewick’s wren was not signiﬁcant (P = 0.517; R2 = 0.015), and the coeﬃcient on the HSI predictor variable was negative (β = -3.193) and not signiﬁcantly diﬀerent from zero (P = 0.857). Therefore, we considered the HSI model for the Bewick’s wren veriﬁed but not validated (Tirpak and others 2009a). 35 Black-and-white Warbler Status The black-and-white warbler (Mniotilta varia) is a neotropical migrant found throughout the eastern United States and southern Canada. This is a forest-interior species and the annual declines of 1.2 percent observed in the United States over the last 40 years likely are the result of increasing forest fragmentation (Sauer and others 2005). This species has a regional combined score of 16 in the WGCP, where it is a species requiring management attention (Table 1). The blackand-white warbler has a regional combined score of only 13 in the CH. The FWS does not recognize the black-and-white warbler as a Bird of Conservation Concern in either BCR (Table 1). Charles H. Warren, images.nbii.gov Natural History As a forest-interior specialist, the black-and-white warbler is found in the mature deciduous hardwood forests of the eastern United States and Canada (Kricher 1995). It is highly sensitive to fragmentation in the landscape (Robbins and others 1989) and typically is absent from small woodlots (< 7.5 ha; Galli and others 1976). Hamel (1992) suggested that 550 ha was the minimum tract size for this species in the Southeast. Few studies have focused exclusively on the habitat ecology of this bird, though Conner and others (1983) found that the black-and-white warbler is associated with mature forest stands with high densities of large (> 32 cm d.b.h.) trees. Although a ground-nesting bird, this species is associated with high densities of hardwood saplings. Conversely, pine saplings negatively aﬀect both the presence and abundance of the black-and-white warbler. This bird occupies upland and bottomland forests but reaches greater densities in the former, with oak-hickory and cove forests considered optimal (Hamel 1992). Nevertheless, successional age may be the most critical habitat factor aﬀecting the black-and-white warbler. Dettmers and others (2002) validated Hamel’s (1992) habitat suitability model for the black-and-white warbler, ﬁnding the model performed well due to the restriction of the black-and-white warbler to older age class forests. However, Thompson and others (1992) and Annand and Thompson (1997) observed the black-and-white warbler in sapling and clearcut stands in Missouri. Model Description Our HSI model for the black-and-white warbler includes six variables: landform, landcover, successional age class, forest patch size, percent forest in a 1-km radius, and canopy cover. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 28). We directly assigned SI scores to these combinations based on vegetation type and age class associations of the black-and-white warbler reported by Hamel (1992). However, we assigned higher values to shrub-seedling stands based on data from Thompson and others (1992) and Annand and Thompson (1997). 36 Table 28.—Relationship between landform, landcover type, age class, and suitability index scores for blackand-white warbler habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.167 0.333 0.333 0.667 Deciduous 0.000 0.167 0.333 0.333 0.667 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.167 0.333 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.167 0.333 0.333 0.333 Low-density residential 0.000 Transitional-shrubland 0.000 0.000 0.167 (0.000) 0.000 0.333 (0.000) 0.000 0.333 (0.000) 0.000 0.333 (0.000) Deciduous 0.000 0.167 0.333 0.333 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.167 0.333 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.167 0.333 0.333 0.333 Low-density residential 0.000 Transitional-shrubland 0.000 0.000 0.167 (0.000) 0.000 0.333 (0.000) 0.000 0.333 (0.000) 0.000 0.333 (0.000) Deciduous 0.000 0.167 0.333 0.333 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.167 0.333 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.167 0.333 0.333 0.333 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Forest patch size (SI2) aﬀects occurrence of this species as it is notably absent from small forest blocks. Therefore, we ﬁt a logarithmic function (Fig. 12) relating forest patch size to SI scores derived from probability of occurrence data from Robbins and others (1989) (Table 29). The relative value of a forest block of a speciﬁc size is inﬂuenced by its landscape context. In predominantly forested landscapes, small forest patches that may not be used in predominantly nonforested landscapes may provide habitat due to their proximity to large forest blocks (Rosenberg and others 1999). To capture this relationship, we ﬁt a logistic function (Fig. 13) to data (Table 30) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. Because of the extreme sensitivity of the black-and-white warbler to fragmented landscapes, we assumed that the midpoint between moderately and lightly fragmented forest deﬁned the speciﬁc cutoﬀ for average (SI score = 0.500) hatitat. We used the maximum value of SI2 or SI3 to account for small patches in predominantly forested landscapes and large patches in predominantly nonforested landscapes. 37 Table 29.—Influence of forest patch size on suitability index (SI) scores for black-and-white warbler habitat Suitability Index Score 1.0 Forest patch size (ha) 0.8 10 a 220 0.6 0.0 b 3,200b a 0.4 b SI score 0.5 1.0 Assumed value. Robbins and others (1989). 0.2 0.0 0 1000 2000 3000 Forest Patch Size (ha) Suitability Index Score Figure 12.—Relationship between forest patch size and suitability index (SI) scores for black-and-white warbler habitat. Equation: SI score = 0.1731 * ln(forest patch size) – 0.4096. 1.0 Table 30.—Relationship between landscape composition (proportion forest in 1-km radius) and suitability index (SI) scores for black-andwhite warbler habitat 0.8 Landscape compositiona 0.6 0.4 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 Landscape Composition (proportion forest in 1-km radius) Figure 13.—Relationship between landscape composition and suitability index (SI) scores for black-and-white warbler habitat. Equation: SI score = 1.047 / (1.000 + (1991.516 * e -10.673 * landscape composition )). 0.00 0.00 0.10 a 0.00 0.20a 0.00 0.30 a 0.00 0.40 a 0.00 0.50 a 0.10 0.60a 0.25 0.70 b 0.50 0.80 a 0.75 0.90 a 0.90 1.00a 1.00 a b Assumed value. Donovan and others (1997). Canopy cover (SI4) also may aﬀect the quality of black-and-white warbler habitat. Thus, we included it as a factor in our HSI model. Prather and Smith (2003) reported higher densities of the black-and-white warbler in forests with relatively open canopies, so we used their data (Table 31) to derive an inverse logistic function (Fig. 14) that quantiﬁed the relationship between canopy cover and SI scores. We calculated the overall HSI score as the geometric mean of the geometric mean of individual SI functions related to forest structure (SI1 and SI4) multiplied by the maximum SI score for forest patch size or percent forest in the 1-km radius landscape. Overall HSI = ((SI1 * SI4)0.500 * Max(SI2 or SI3))0.500 38 SI score a Table 31.—Influence of canopy cover on suitability index (SI) scores for black-and-white warbler habitat. Suitability Index Score 1.0 Canopy cover (percent)a 0.8 0.6 31 1.000 73 0.866 91 0.627 a 0.4 SI score Prather and Smith (2003). 0.2 0.0 0 25 50 75 100 Canopy Cover (%) Figure 14.—Relationship between canopy cover and suitability index (SI) scores for black-and-white warbler habitat. Equation: SI score = 1 - (-4.190 / (1 + (-1890.213 * e -0.055 * canopy cover))). Veriﬁcation and Validation The black-and-white warbler was found in 85 of the 88 subsections within the CH and WGCP. Not surprisingly, Spearman rank correlations based on all subsections and only subsections in which this species was found produced similar results: signiﬁcant (P ≤ 0.001 for both analyses) positive relationships (rs = 0.54 and 0.53, respectively) between average HSI score and mean BBS route abundance. The generalized linear model predicting BBS abundance from BCR and HSI for the black-and-white warbler was signiﬁcant (P ≤ 0.001; R2 = 0.380), and the coeﬃcient on the HSI predictor variable was both positive (β = 3.194) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the black-and-white warbler both veriﬁed and validated (Tirpak and others 2009a). 39 Blue-gray Gnatcatcher Status The blue-gray gnatcatcher (Polioptila caerulea) is a shortdistance migrant found throughout eastern North America and the Southwest. Populations are relatively stable in both the CH and WGCP (Table 5). The FWS does not recognize this species as a Bird of Conservation Concern in either region (Table 1). This bird requires management attention in the CH (regional combined score = 14) but does not have any special designation in the WGCP (regional combined score =13; Table 1). Charles H. Warren, images.nbii.gov Natural History The blue-gray gnatcatcher is a small passerine that inhabits woodland types ranging from shrubland to mature forest (Ellison 1992). It prefers deciduous habitats and is rare or absent in evergreen forests. This species attains its highest numbers in mesic and low-lying areas, but is also found in xeric forests and along ridges. Kershner and others (2001) did not identify speciﬁc microhabitat requirements for this species in Illinois, and considerable variation in nest height (0.8 to 24.4 m) and territory size (0.5 to 8 ha) has been documented across the range. Although often associated with edges, this bird may be area sensitive (Knutson 1995, Kilgo and others 1998). Nest success was greater for nests placed higher and farther from an edge in Illinois (Kershner and others 2001) but did not diﬀer between bottomland hardwood stands and cottonwood plantations in the Mississippi Alluvial Valley (Twedt and others 2001). The abundance of the blue-gray gnatcatcher was higher in bottomland hardwood stands surrounded by ﬁelds than those surrounded by pine forest (Kilgo and others 1998). Model Description The HSI model for the blue-gray gnatcatcher includes seven variables in ﬁve functions: landform, landcover, successional age class, forest patch size, percent forest in a 1-km radius landscape, edge, and basal area. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 32). We directly assigned SI scores to these combinations on the basis of data from Hamel (1992) on the relative quality of vegetation associations and successional age classes for this species. We adjusted Hamel’s values for shrub-seedling and sapling-aged stands to account for the higher densities observed in young forests by Thompson and others (1992) and Annand and Thompson (1997). We included forest patch size (SI2) as a variable to account for the area sensitivity of the blue-gray gnatcatcher. We ﬁt a logarithmic function (Fig. 15) to data from Robbins and others (1989) on the probability of occurrence for this bird in stands of various sizes (Table 33). Nevertheless, the actual use of a forest patch reﬂects both its area and its landscape 40 Table 32.—Relationship of landform, landcover type, and successional age class to suitability index scores for blue-gray gnatcatcher habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Landform Landcover type Grass-forb Shrubseedling Sapling Pole Saw Floodplain-valley Low-density residential 0.000 0.333 0.667 0.667 1.000 Transitional-shrubland 0.000 0.333 0.667 0.667 1.000 Deciduous 0.000 0.333 0.667 0.667 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.083 0.167 0.167 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Terrace-mesic Xeric-ridge Woody wetlands 0.000 0.167 0.333 0.333 1.000 Low-density residential 0.000 0.333 0.667 0.667 1.000 Transitional-shrubland 0.000 Deciduous 0.000 0.333 (0.000) 0.333 0.667 (0.000) 0.667 0.667 (0.000) 0.667 1.000 (0.000) 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.083 0.167 0.167 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.167 0.333 0.333 1.000 Low-density residential 0.000 0.333 0.667 0.667 1.000 Transitional-shrubland 0.000 Deciduous 0.000 0.333 (0.000) 0.333 0.667 (0.000) 0.667 0.667 (0.000) 0.667 1.000 (0.000) 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.083 0.167 0.167 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.167 0.333 0.333 1.000 context (SI3). In predominantly forested landscapes, a small forest patch that otherwise may not be suitable may be occupied due to its proximity to a larger forest block (Rosenberg and others 1999). Because the gnatcatcher also is associated with edges, it may not be as abundant in predominantly forested landscapes that lack signiﬁcant edge habitat. Thus, we assumed that the relationship between habitat suitability of the blue gray gnatcatcher and the amount of forest in the landscape followed a Gaussian function (Fig. 16), with landscapes containing 70 to 80 percent forest as optimal and suitability declining as the proportion of forest in the landscape moved from this ideal (Table 34). We used the maximum suitability score of SI2 or SI3 to simultaneously account for patch area and landscape composition. We also included edge (SI4) in our HSI model because of the association of the blue-gray gnatcatcher with edges within large forest blocks. This species nests along both hard and soft edges (typically within 30 m; Kershner and others 2001). Therefore, we deﬁned edge as the interface among sapling, pole, and sawtimber stands and herbaceous and nonforest landcovers (hard edge) or seedling and grass-forb stands (soft edge). We used a 7 × 7 pixel moving window (210 x 210 m) to identify where these adjacencies occurred but recognized that the blue-gray gnatcatcher is not restricted to edge habitats and applied a residual SI score (0.010) to sites that did not meet this criterion (Table 35). 41 Table 33.—Influence of forest patch size on suitability index (SI) scores for blue-gray gnatcatcher habitat Suitability Index Score 1.0 Forest patch size (ha)a 0.8 6.8 0.6 0.2 0.0 0 1000 2000 3000 Forest Patch Size (ha) 1.0 Suitability Index Score 0.5 3,200 1.0 Robbins and others (1989). Table 34.—Relationship between landscape composition (percent forest in 10-km radius) and suitability index (SI) scores for blue-gray gnatcatcher habitat Landscape composition Figure 15.—Relationship between forest patch size and suitability index (SI) scores for blue-gray gnatcatcher habitat. Equation: SI score = 0.137 * ln(forest patch size) + 0.186. 0.8 0.6 0 a 0.00 0.10 20a 0.20 30 b 0.30 40 a 0.40 50 b 0.50 60a 0.75 70 b 1.00 80 a 1.00 90 b 0.75 10 a b 0.50 Assumed value. Dononvan and others (1997). 0.2 Table 35.—Influence of edge on suitability index (SI) scores for blue-gray gnatcatcher habitat 0.0 0 25 50 75 100 Landscape Composition (% forest in 10-km radius) Figure 16.—Relationship between landscape composition and suitability index (SI) scores for blue-gray gnatcatcher habitat. Equation: SI score = 1.002 * e ((0 – ((landscape composition) – 74.165) ^ 2) / 1064.634) . 7 × 7 pixel window around forest pixel includes fielda SI score Yes 1.00 No 0.01 a Field defined as any shrub-seedling or grass-forb age class forest, or natural grasslands, pasture-hay, fallow, urban-recreational grasses, emergent herbaceous wetlands, open water, high intensity residential, commercial-industrial-transportation, bare rock-sand-clay, quarries-strip mines-gravel pits, row crops, or small grains. Forest defined as any used sapling, pole, or sawtimber age class pixel of low-density residential, transitional, shrublands, deciduous, mixed, evergreen, orchard, or woody wetlands (i.e., SI1 > 0). We ﬁt a quadratic function to data from Annand and Thompson (1997) on the response of the blue-gray gnatcatcher to basal area (SI5; Table 36, Fig. 17), reﬂecting the preference of this species for open forest conditions. 42 SI score a 100a 0.4 0.0 15 a 0.4 SI score Table 36.—Influence of basal area (m2/ha) on suitability index (SI) scores for blue-gray gnatcatcher habitat Suitability Index Score 1.0 Basal areaa 0.8 0.6 0.706 12.33 1.000 22.20 0.412 a 0.4 SI score 3.41 Annand and Thompson (1997). 0.2 0.0 0 10 20 30 Basal Area (m^2/ha) Figure 17.—Relationship between basal area and suitability index (SI) scores for blue-gray gnatcatcher habitat. Equation: SI score = 0.3863 + 0.1105 * (basal area) – 0.0049 * (basal area)2. To calculate the HSI score for sapling, pole, and sawtimber age classes, we determined the geometric mean of SI scores for forest structure (SI1 and SI5) and landscape composition attributes (Max(SI2 or SI3) and SI4) separately and then the geometric mean of these means together. Because edge occurrence (SI4) was not applicable to the shrub-seedling age class, we calculated HSI scores separately for this age class and summed across age classes to determine the overall HSI score for the landscape. Sapling, pole, and sawtimber successional age classes: HSIOld = (((SI1 * SI5)0.500) * ((Max (SI2 or SI3)) * SI4)0.500)0.500 Shrub-seedling successional age classes: HSIShrub = ((SI1 * SI5)0.500 * (Max (SI2 or SI3)))0.500 Overall HSI = HSIOld + HSIShrub Veriﬁcation and Validation The blue-gray gnatcatcher was found in all 88 subsections of the CH and WGCP. Spearman rank correlation analysis on average HSI score and mean BBS route abundance across subsections resulted in a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.58) between these variables. The generalized linear model predicting BBS abundance from BCR and HSI for the blue-gray gnatcatcher was signiﬁcant (P ≤ 0.001; R 2 = 0.210), and the coeﬃcient on the HSI predictor variable was both positive (β = 19.625) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the blue-gray gnatcatcher both veriﬁed and validated (Tirpak and others 2009a). 43 Blue-winged Warbler Status The blue-winged warbler (Vermivora pinus) is a neotropical migrant found from southern New England west to the Lake States and Chandler S. Robbins, Patuxent Bird Identification InfoCenter south through the southern Appalachians and Photo used with permission Ozarks. Across most of its range, this species has been stable and has even increased in some areas (possibly to the detriment of the goldenwinged warbler, with which it sometimes interbreeds; Gill 1980). Once limited to a mostly Midwestern range, this bird expanded into southern New England as forests were cleared and farms were abandoned. However, as the forest has matured in this region, the blue-winged warbler has experienced declines (3.3 and 5.3 percent annually from 1966 to 2004 in the increasingly residential Connecticut and New Jersey, respectively). A similar phenomenon has occurred in the Southeast and BBS data indicate a 3.7 percent decline in FWS Region 4 during this same period (Sauer and others 2005). This species is designated a Bird of Conservation Concern in the CH but not in the WGCP (Table 1), where it rarely breeds. It has a regional combined score of 19 in the CH and requires management attention in that region (Table 1). Natural History The blue-winged warbler is an early successional species (Gill and others 2001) that beneﬁted from European settlement by expanding its range following the initial clearing of forests for agriculture and the subsequent abandonment of farms. Breeding habitat includes early to midsuccessional forest containing dense low growth (shrubs, young trees, thickets). This species makes use of a variety of landform conditions from wetland edges to dry uplands, though mated males have more xeric territories than unmated males. Territories range from 0.2 to 5 ha, with boundaries often aligned along edges. Nests typically are within 30 m of a forest edge in grassy areas with high numbers of small (< 10 cm d.b.h.) trees. Density is inversely related to successional age class, fragmentation, and the abundance of the goldenwinged warbler and brown-headed cowbird. Model Description The blue-winged warbler model includes ﬁve variables: landform, landcover, successional age class, early successional patch size, and canopy cover. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 37). We directly assigned SI scores to these combinations based on habitat associations reported in Hamel (1992) for the blue-winged warbler. We modiﬁed Hamel’s data to maximize SI scores in the transitional-shrubland landcover class in the xeric landform. We also included early successional patch size (SI2) in our model on the basis of data from Rodewald and Vitz (2005) on the relative abundance of the blue-winged warbler in small and large clearcuts (Table 38; Fig. 18). We deﬁned early successional forest by age class and included only grass-forb, shrub-seedling, and sapling age classes in the calculation of patch area. 44 Table 37.—Relationship of landform, landcover type, and successional age class to suitability index scores for blue-winged warbler habitat Successional age class Landcover type Floodplain-valley Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.333 0.167 0.000 0.000 Deciduous 0.000 0.333 0.167 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Xeric-ridge 0.000 0.333 0.167 0.000 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.167 0.083 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.667 0.333 0.000 0.000 Deciduous 0.000 0.667 0.333 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.333 0.167 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.333 0.167 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 1.000 0.500 0.000 0.000 Deciduous 0.000 1.000 0.500 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.333 0.167 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.333 0.167 0.000 0.000 SI score a 0.000 4b 0.786 13b 1.000 b Assumed value. Rodewald and Vitz (2005). Saw 0.000 Early successional patch size (ha) a Pole Orchard-vineyard Table 38.—Influence of early successional patch size on suitability index scores for blue-winged warbler habitat; early successional patches include all adjacent grass-forb, shrub-seedling, and sapling successional age class forest 0 Sapling Mixed 1.0 Suitability Index Score Terrace-mesic Grass-forb Shrubseedling Landform 0.8 0.6 0.4 0.2 0.0 0 5 10 15 Early Successional Patch Size (ha) Figure 18.—Relationship between early successional patch size and suitability index (SI) scores for blue-winged warbler habitat. Equation: SI score = 1.000 / (1 + (14353.617 * e -2.788 * forest patch size)). 45 Table 39.—Influence of canopy cover on suitability index (SI) scores for blue-winged warbler habitat Suitability Index Score 1.0 Canopy cover (percent)a 0.8 0.6 0.4 29.26 1.000 71.86 0.523 93.38 0.034 95.58 0.000 96.59 0.011 a Annand and Thompson (1997). 0.2 0.0 0 25 50 75 100 Canopy Cover (%) Figure 19.—Relationship between canopy cover and suitability index (SI) scores for blue-winged warbler habitat. Equation: SI score = 1 - (1.0381 / (1 + (16277.383 * e -0.1327 * canopy cover))). We used an inverse logistic function (Fig. 19) to quantify the relationship between canopy cover (SI3) and SI scores to reﬂect the lower densities of the blue-winged warbler in forests with increasingly closed canopies. We deﬁned this function by ﬁtting a curve to data from Annand and Thompson (1997) on the relative density of this bird in forest stands with diﬀerent estimates of canopy cover (Table 39). To calculate the overall HSI score for this species , we determined the geometric mean of SI scores for forest structure attributes (SI1 and SI3) and then calculated the geometric mean of this value and early successional patch size (SI2). Overall HSI = ((SI1 * SI3)0.500 * SI2)0.500 Veriﬁcation and Validation The blue-winged warbler was found in 64 of the 88 subsections within the CH and WGCP. We used Spearman rank correlations between average HSI score and mean BBS route abundance at the subsection scale to verify this model. We observed signiﬁcant positive relationships when analyses included all subsections (rs = 0.26; P = 0.014) or only those subsections where this species was detected (rs = 0.28; P = 0.026). The generalized linear model predicting BBS abundance from BCR and HSI for the blue-winged warbler was signiﬁcant (P ≤ 0.001; R2 = 0.232), and the coeﬃcient on the HSI predictor variable was positive (β = 1.717) but not signiﬁcantly diﬀerent from zero (P = 0.334). Therefore, we considered the HSI model for the blue-winged warbler veriﬁed but not validated (Tirpak and others 2009a). 46 SI score Brown Thrasher Status The brown thrasher (Toxostoma rufum) is a shortdistance migrant found throughout eastern North America. Although populations in the CH and WGCP declined by 1.4 percent per year between 1966 and 2004 (Table 5), this species is not considered a Bird Jeffrey A Spendelow, Patuxent Bird Identification InfoCenter of Conservation Concern in either BCR (Table 1). Photo used with permission The brown thrasher has a regional combined score of 13 and 15 in the WGCP and CH, respectively, and is a species warranting management attention in the CH (Table 1). Natural History A ground-foraging passerine, the brown thrasher is associated with edge habitats throughout the eastern United States and Canada (Cavitt and Haas 2000). Breeding habitat includes a variety of vegetation types, but this species reaches its highest densities in shrublands and midsuccessional forests. Grand and Cushman (2003) found that thrashers in Massachusetts were associated predominately with the amount of scrub oak in the landscape. Rumble and Gobeille (2004) found no signiﬁcant diﬀerence in brown thrasher occurrence among seral stages of cottonwood ﬂoodplains in South Dakota, though this bird was detected most often in younger forest classes. Savanna restoration eﬀorts increase thrasher abundance by reducing tree density (Davis and others 2000). Nests are typically low in a tree or shrub but some may be on the ground. Territory size and thrasher density vary according to habitat quality (0.5 to 1.1 ha and 0.1 to 0.4/ha, respectively). The FWS (Cade 1986) developed an HSI model for this species that included three site-speciﬁc variables: density of woody stems, canopy cover, and litter cover. Model Description Our brown thrasher model includes six variables: landform, landcover, successional age class, edge occurrence, small stem density (<2.5 cm d.b.h.), and forest composition in a 10-km radius. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 40). We directly assigned SI scores to these combinations on the basis of habitat associations reported by Hamel (1992) for the brown thrasher in the Southeast. This edge species inhabits thickets and hedgerows in deciduous forests. Because the brown thrasher uses both hard and soft edges, we deﬁned edge (SI2) as the interface between pole age forest and herbaceous or non-forest landcovers (hard edge) and seedling or grass-forb age forest (soft edge). To be suitable, we required pole age forest sites to be adjacent to an edge (Table 41). However, we relaxed this requirement for seedling-shrub and sapling stands, which we considered suitable regardless of edge. 47 Table 40.—Relationship of landform, landcover type, and successional age class to suitability index scores for brown thrasher habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Landform Landcover type Grass-forb Shrubseedling Sapling Pole Saw Floodplain-valley Low-density residential 0.000 0.500 0.333 0.083 0.000 Transitional-shrubland 0.000 0.500 0.333 0.083 0.000 Deciduous 0.000 0.500 0.333 0.083 0.000 Evergreen 0.000 0.667 0.500 0.167 0.000 Mixed 0.000 1.000 0.667 0.167 0.000 Orchard-vineyard 0.000 0.500 0.333 0.083 0.000 Woody wetlands 0.000 0.667 0.417 0.083 0.000 Low-density residential 0.000 0.667 0.417 0.083 0.000 Transitional-shrubland 0.000 0.000 0.000 0.667 (0.500) 0.417 0.167 Deciduous 1.000 (0.667) 0.667 0.083 0.000 Evergreen 0.000 0.667 0.500 0.167 0.000 Mixed 0.000 1.000 0.667 0.167 0.000 Orchard-vineyard 0.000 0.667 0.417 0.083 0.000 Woody wetlands 0.000 0.667 0.500 0.167 0.000 Low-density residential 0.000 1.000 0.667 0.167 0.000 Transitional-shrubland 0.000 0.000 0.667 (0.250) 0.667 0.167 (0.083) 0.167 0.000 Deciduous 1.000 (0.334) 1.000 Evergreen 0.000 Mixed 0.000 0.667 (0.334) 1.000 0.500 (0.250) 0.667 0.167 (0.083) 0.167 0.000 Orchard-vineyard 0.000 1.000 0.667 0.167 0.000 Woody wetlands 0.000 0.667 0.500 0.167 0.000 Terrace-mesic Xeric-ridge Table 41.—Influence of edge on suitability index (SI) scores for brown thrasher habitat 3 × 3 pixel window around forest pixel includes fielda Yes No a b SI score 1.0 0.0 Field defined as any shrub-seedling or grass-forb age class pixel, or natural grasslands, pasture-hay, fallow, urbanrecreational grasses, emergent herbaceous wetlands, open water, high intensity residential, commercial-industrialtransportation, bare rock-sand-clay, quarries-strip minesgravel pits, row crops, or small grains. Forest defined as any used pole age class pixel of low-density residential, transitional, shrublands, deciduous, mixed, evergreen, orchard, or woody wetlands. b Seedling-shrub and sapling habitats used regardless of edge. 48 0.000 0.000 Table 42.—Influence of small stem (< 2.5 cm d.b.h.) density (stems * 1,000/ha) on suitability index (SI) scores for brown thrasher habitat Suitability Index Score 1.0 Small stem densitya 0.8 0.6 0.1 10 1.0 40 0.5 a 0.4 SI score 0 Cade (1986). 0.2 0.0 0 10 20 30 40 50 Small Stem Density (stems * 1,000/ha) Figure 20.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for brown thrasher habitat. Equation: SI score = (0.1 + (0.165 * (small stem density / 1000))) / (1 + (-0.003 * (small stem density / 1000)) + (0.0078 * ((small stem density / 1000))2)). The brown thrasher occupies habitats with numerous small stems (SI3). We ﬁt a smoothed quadratic function (Fig. 20) to HSI cutoﬀ values from the FWS HSI model for this species (Cade 1986; Table 42) to quantify the relationship between small stem density and habitat suitability. Although the brown thrasher is associated with edges, it prefers modestly forested landscapes (Haas 1997). We included forest composition (SI4) in our model, assuming that habitat suitability would be low if there were no woodland (i.e., 0 percent forest, the left side of the function; Fig. 21) or no edges (i.e., 100 percent forest, the right side of the function). Haas (1997) observed higher reproductive success for birds in more isolated shelterbelts and Robbins and others (1989) observed negative relationships between the occurrence of the gray catbird and American robin (species that share similar habitat preferences to those of the brown thrasher) and forest patch size. Further, Perkins and others (2003b) observed an increase in abundance of edge-associated birds as the total amount of woody cover decreased. However, the brown thrasher responded positively to the amount of forest cover in the study area. We interpreted these observations as evidence that this species would exhibit a preference for landscapes with moderate forest landcover. We ﬁt a Gaussian function to landscape proportions reﬂecting this pattern and assumed that landscapes that were 70 percent forested were associated with the maximum SI score (Table 43). 49 Table 43.—Relationship between landscape composition (percent forest in 10-km radius) and suitability index (SI) scores for brown thrasher habitat Suitability Index Score 1.0 Landscape compositiona 0.8 0.6 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 10-km radius) Figure 21.—Relationship between landscape composition and suitability index (SI) scores for brown thrasher habitat. Equation: SI score = 0.998 * e ((0 – ((landscape composition) – 70.304) ^ 2) / 1253.402) . 0 0.00 10 0.05 20 0.10 30 0.25 40 0.50 50 0.75 60 0.90 70 1.00 80 0.90 90 0.75 100 0.50 a Assumed value. We assumed that the brown thrasher used edge as a surrogate to early successional habitat, so we calculated HSI scores separately for young (seedling-shrub and sapling) and old (pole) age class forests. In the former, the geometric mean of forest structure and landscape composition variables deﬁnes the suitability score. For the latter, we included edge occurrence in the calculation. We summed the age class-speciﬁc HSI scores to determine the overall HSI score for all sites. Seedling-shrub and sapling successional age classes: HSIYoung: ((SI1 * SI3)0.500 * SI4)0.500 Pole successional age class: HSIPole: ((SI1 * SI3)0.500 * SI4)0.500 * SI2 Overall SI = HSIYoung + HSIPole Veriﬁcation and Validation The brown thrasher was found in all 88 subsections of the CH and WGCP. Spearman rank correlation did not identify a positive relationship between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the brown thrasher was signiﬁcant (P ≤ 0.001; R2 = 0.719); however, the coeﬃcient on the HSI predictor variable was negative (β = -7.087). Therefore, we considered the HSI model for the brown thrasher neither veriﬁed nor validated (Tirpak and others 2009a). 50 SI score Brown-headed Nuthatch Status The brown-headed nuthatch (Sitta pusilla) is a resident species of mature pine forests along the Piedmont and Coastal Plains of the southeastern United States. Although this species has experienced modest declines throughout most of its range over the last 40 years (1.2 percent per year), only in Florida has the decline been signiﬁcant (4.2 percent annually from 1966 to 2004; Sauer and others 2005). This species is an FWS Bird of Conservation Concern in the WGCP (Table 1), where it has a regional combined score of 19. The brown-headed nuthatch is a rare breeder in the CH (regional combined score = 19), and PIF considers this species one that warrants critical recovery in that region. Fernbank Science Center Photo used with permission Natural History The brown-headed nuthatch is closely associated with pine: it breeds in mature pine forests and forages almost exclusively in pine trees (> 98 percent of observations; Withgott and Smith 1998). Although often associated with the longleaf pine savanna characteristic of the habitat for red-cockaded woodpecker and Bachman’s sparrow, the brown-headed nuthatch has a broader niche than these species (Hamel 1992, Dornak and others 2004). The habitat of this species is deﬁned by two habitat elements: mature pines for foraging and cavities for nesting (Wilson and Watts 1999, Dornak and others 2004). Speciﬁc composition of pine species is not as critical as d.b.h., with an average d.b.h. of 25.6 cm considered optimal (O’Halloran and Conner 1987 cited in Dornak and others 2004). The brownheaded nuthatch nests primarily in large-diameter snags < 3 m tall and may require seven to eight snags per ha to ensure adequate nest and roost sites, particularly in the presence of interspeciﬁc competition for cavities. In urban areas, the brown-headed nuthatch readily adopts nest boxes and may use other manmade cavities, such as streetlights. This species prefers open pine stands with few hardwoods (≤ 17.4 stems/ha and basal area ≤ 5 m2/ha) and an open midstory (Wilson and Watts 1999). Optimal canopy cover is highly variable (15 to 85 percent) but stands with closed canopies are not preferred (O’Halloran and Conner 1987, Wilson and Watts 1999). Undergrowth typically is sparse (roughly 35 percent; Dornak and others 2004). The nuthatch regularly breeds at low densities in suboptimal habitats, including stands with small pines, a large fraction of hardwoods, and dense understories (Withgott and Smith 1998). Area sensitivity apparently is not an issue for this species, which is not an acceptable host for the brown-headed cowbird (Withgott and Smith 1998). Model Description The HSI model for the brown-headed nuthatch includes six variables: landform, landcover, successional age class, snag density, small stem (< 2.5 cm d.b.h.) density, and hardwood basal area. 51 Table 44.—Relationship of landform, landcover type, and successional age class to suitability index scores for brown-headed nuthatch habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.000 Evergreen 0.000 0.000 0.334 0.834 1.000 Mixed 0.000 0.000 0.167 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.000 Evergreen 0.000 0.000 0.334 0.834 1.000 Mixed 0.000 0.000 0.167 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.000 Evergreen 0.000 0.000 0.334 0.834 1.000 Mixed 0.000 0.000 0.167 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 44). We directly assigned SI scores to these combinations on the basis of habitat associations of the brownheaded nuthatch described by Hamel (1992). We included snag density (SI2) in our HSI model because of the importance of cavities to this species. We assumed that the SI score was zero when eight or fewer snags of any size were present (Dornak and others 2004). We ﬁt a logistic function (Fig. 22) to data from Wilson and Watts (1999) (Table 45) to quantify the relationship between snag density and SI scores. We also used small stem density as a function (SI3) in the HSI model to account for the preference of the brown-headed nuthatch for open understories. We ﬁt an inverse logistic function (Fig. 23) to hypothetical data reﬂecting this preference (Table 46). The shape of this function is supported by observations from Wilson and others (1995), who observed a higher abundance of the brown-headed nuthatch in stands immediately following wildlife stand improvements and prescribed burns (when stem density was lowest) with subsequent declines in abundance as stem density increased through time. 52 Table 45.—Influence of snag density on suitability index (SI) scores for brown-headed nuthatch habitat Suitability Index Score 1.0 Snag density (snags/ha) 8 0.8 a 40 0.000 b 66.67 0.6 0.286 b 106.67b 0.4 a b SI score 0.715 1.000 Dornak and others (2004). Wilson and Watts (1999). 0.2 0.0 0 25 50 75 100 125 Snag Density (snags/ha) Figure 22.—Relationship between snag density and suitability index (SI) scores for brown-headed nuthatch habitat. Equation: SI score = 1.000 / (1 + (49.165 * e (-0.073 * snag density))). Table 46.—Influence of small stem (< 2.5 cm d.b.h.) density (stems * 1,000/ha) on suitability index (SI) scores for brown-headed nuthatch habitat Suitability Index Score 1.0 Small stem densitya 0.8 01 0.6 0.4 1.0 10 1 0.9 20 1 0.5 301 0.1 1 0.0 40 a 0.2 SI score Assumed value. 0.0 0 10 20 30 40 50 Small Stem Density (stems * 1,000/ha) Figure 23.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for brown-headed nuthatch habitat. Equation: SI score = 1 (1.010 / (1 + (79.565 * e (-0.217 * (small stem density / 1000))))). Finally, we incorporated hardwood basal area (SI4) as a model variable as birds are less abundant in habitats with a greater hardwood component (Wilson and others 1995, Withgott and Smith 1998, Wilson and Watts 1999). Again, we relied on data from Wilson and Watts (1999) (Table 47) to develop an inverse logistic function to describe the relationship between hardwood basal area and SI score (Fig. 24). To determine the overall HSI score for the brown-headed nuthatch, we calculated the geometric mean of the four individual functions related to forest structure attributes. Overall HSI = (SI1 * SI2 * SI3 * SI4)0.250 53 Table 47.—Influence of hardwood basal area on suitability index (SI) scores for brown-headed nuthatch habitat Suitability Index Score 1.0 Hardwood basal area (m2/ha) 0.8 0.6 0.4 0.0 1.000 4.6 a 0.778 10.5a 0.222 15.0 b 0.000 20.0 b 0.000 a Wilson and Watts (1999). b Assumed value. 0.2 0.0 0 5 10 15 20 Hardwood Basal Area (m^2/ha) Figure 24.—Relationship between hardwood basal area and suitability index (SI) scores for brown-headed nuthatch habitat. Equation: SI score = 1 - (1.018 / (1 + (29.747 * e (-0.441 * hardwood basal area) ))). Veriﬁcation and Validation The brown-headed nuthatch was found in 37 of the 88 subsections within the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.58) between average HSI score and mean BBS route abundance across subsections. This relationship was even stronger (rs = 0.80) when subsections in which the brownheaded nuthatch was not detected were removed from the analysis. The generalized linear model predicting BBS abundance from BCR and HSI for the brown-headed nuthatch was signiﬁcant (P ≤ 0.001; R2 = 0.738), and the coeﬃcient on the HSI predictor variable was both positive (β = 4.712) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the brown-headed nuthatch both veriﬁed and validated (Tirpak and others 2009a). 54 SI score a Carolina Chickadee Status The Carolina chickadee (Parus carolinensis) is a resident species of the southeastern United States. Although populations have been stable in the CH, this species has declined by about 2 percent annually over the last 40 years in the WGCP (Table 5). This bird is a planning and responsibility species in both the CH (regional combined score = 15) and WGCP (regional combined score = 16; Table 1). Charles H. Warren, images.nbii.gov Natural History The Carolina chickadee is a generalist species that breeds in a variety of forest types across a broad spectrum of landforms (Mostrom and others 2002). It nests in cavities of live and dead trees within multilayered forests containing well developed shrub, midstory, and overstory canopies (Hamel 1992). Abundance declines following reduction of hardwoods in pine stands, likely as a result of the loss of midstory trees (Provencher and others 2002). Nest success and adult survival is positively correlated with woodlot area but is lower on edges regardless of patch size (Doherty and Grubb 2002). Nest destruction by the house wren is a major cause of nest failure in areas where the ranges of these species overlap. Territory size ranges from 1.6 to 2.4 ha. Model Description The Carolina chickadee model includes four variables: landform, landcover, successional age class, and snag density. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 48). We directly assigned SI scores to these combinations on the basis of vegetation and successional age class associations of the Carolina chickadee reported in Hamel (1992). We included snag density (SI2) as a variable because of the importance of nest and roost cavities for the chickadee, a secondary cavity nester. Data for the Carolina chickadee were not available but Rumble and Gobeille (2004) and Sedgwick and Knopf (1990) observed the black-capped chickadee in habitats with six snags per hectare (Table 49). Therefore, we assumed that stands with six or more snags per ha were representative of optimal habitat. Because the chickadee can use cavities in live trees, we assumed that stands with no snags were not necessarily nonhabitat and assigned to them a small but non-zero SI score (0.03). We ﬁt a logistic function through these data points to quantify the relationship between snag density and habitat suitability (Fig. 25). We calculated the overall HSI score as the geometric mean of the two individual functions: Overall HSI = (SI1 * SI2)0.500 55 Table 48.—Relationship of landform, landcover type, and successional age class to SI scores for Carolina chickadee habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.167 0.500 0.667 Transitional-shrubland 0.000 0.000 0.167 0.500 0.667 Deciduous 0.000 0.000 0.167 0.500 0.667 Evergreen 0.000 0.000 0.334 0.834 1.000 Mixed 0.000 0.000 0.334 0.834 1.000 Orchard-vineyard 0.000 0.000 0.167 0.500 0.667 Woody wetlands 0.000 0.000 0.167 0.500 0.667 Low-density residential 0.000 0.000 0.167 0.500 0.667 Transitional-shrubland 0.000 0.000 0.334 0.834 1.000 Deciduous 0.000 0.000 0.167 0.500 0.667 Evergreen 0.000 0.000 0.334 0.834 1.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.334 0.834 1.000 Orchard-vineyard 0.000 0.000 0.167 0.500 0.667 Woody wetlands 0.000 0.000 0.167 0.500 0.667 Low-density residential 0.000 0.000 0.167 0.500 0.667 Transitional-shrubland 0.000 0.000 Deciduous 0.000 0.000 0.334 (0.250) 0.167 0.834 (0.667) 0.500 1.000 (0.834) 0.667 Evergreen 0.000 0.000 Mixed 0.000 0.000 0.334 (0.250) 0.334 0.834 (0.667) 0.834 1.000 (0.834) 1.000 Orchard-vineyard 0.000 0.000 0.167 0.500 0.667 Woody wetlands 0.000 0.000 0.167 0.500 0.667 Table 49.—Influence of snag density on suitability index (SI) scores for Carolina chickadee habitat Suitability Index Score 1.0 Snag density (snags/ha) 0.8 0.90 6 a a, c Rumble and Gobeille (2004). b Assumed value. c Sedgwick and Knopf (1990). 0.4 0.2 0.0 0.0 2.5 5.0 7.5 10.0 Snag Density (snags/ha) Figure 25.—Relationship between snag density and suitability index (SI) scores for Carolina chickadee habitat. Equation: SI score = 1.007 / (1.000 + (32.567 * e (-1.403 * snag density))). 56 0.03 4b 0 0.6 SI score a 1.00 Veriﬁcation and Validation The Carolina chickadee was found in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.55) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the Carolina chickadee was signiﬁcant (P ≤ 0.001; R2 = 0.473), and the coeﬃcient on the HSI predictor variable was both positive (β = 5.142) and signiﬁcantly diﬀerent from zero (P = 0.038). Therefore, we considered the HSI model for the Carolina chickadee both veriﬁed and validated (Tirpak and others 2009a). 57 Cerulean Warbler Status The cerulean warbler (Dendroica cerulea) is a long-distance migrant to the eastern United States. Densities are highest in the Ohio River Valley and along the Cumberland Plateau. This species has declined across most of its range, including the CH and WGCP (6.3 and 9.5 percent per year from 1966 to 2004, respectively; Table 5). The cerulean warbler is classiﬁed as a Bird of Conservation Concern requiring U.S. Forest Service critical recovery in the WGCP (regional combined score = 19) and immediate management in the CH (regional combined score = 19) (Table 1). Concern for this species culminated in a petition to the FWS to list the cerulean warbler as threatened. However, this action was deemed unwarranted on the basis of current scientiﬁc information (Federal Register 71:234 [6 December 2006] p. 70717). Natural History A forest interior specialist, the cerulean warbler has experienced some of the most dramatic declines of any songbird over the last 30 years (Hamel 2000). This species has a broad geographic range but is abundant only locally. It may nest semi-colonially, with territories in good habitat highly clumped. The cerulean warbler seems to be highly sensitive to forest fragmentation. Robbins and others (1989) found a 50 percent reduction in observations of this species as forest patch size declined from 3,000 to 700 ha. No birds were detected on forest patches less than 138 ha. Estimates from other researchers suggest that forest tracts as large as 8,000 ha may be required to ensure sustainable populations in the Mississippi Alluvial Valley (summarized in Hamel [2000]). Although it requires large forest tracts, the cerulean warbler establishes territories near interior forest gaps. Weakland and Wood (2005) observed a positive association between this species and forest roads or snags that created small canopy openings. Aside from canopy gaps (a measure of horizontal canopy structure), the cerulean warbler also may respond to the vertical canopy proﬁle. Canopy cover of 6 to 12 m and more than 24 m was preferred in West Virginia (Weakland and Wood 2005). In Ontario, canopy cover of 12 to 18 m and more than 18 m was preferred (Jones and Robertson 2001). The diﬀerence in preferred canopy heights between these studies likely reﬂects diﬀerences in local vegetation structure rather than an absolute diﬀerence in preferred canopy height. The key habitat feature in both is the multilayered character of the overstory canopy. Closed-canopy stands with large trees (both in height and d.b.h.) are commonly associated with the cerulean warbler but likely are a crude proxy for the aforementioned canopy features that provide the true selection criteria for this bird (Hamel 2000). This species is associated with bottomland hardwoods in the Southeast and ridges in West Virginia (Hamel 2000, Weakland and Wood 2005). Again, speciﬁc landforms probably are not directly selected for but are correlated with the location of large tracts of deciduous forest containing large trees and favorable canopy conditions in these landscapes. 58 Table 50.—Relationship of landform, landcover type, and successional age class to suitability index scores for cerulean warbler habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Landform Landcover type Floodplain-valley Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.500 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.400 0.800 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.500 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Terrace-mesic Xeric-ridge Woody wetlands 0.000 0.000 0.000 0.400 0.800 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.400 0.800 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.400 0.800 In “Birds of North America,” Hamel (2000) stated: “Important habitat elements for this species thus appear to be large tracts with big deciduous trees in mature to old-growth forest with horizontal heterogeneity of the canopy. The pattern of vertical distribution of foliage in the canopy is also important.” Model Description The HSI model for the cerulean warbler includes seven variables: landform, landcover, successional age class, forest patch size, percent forest in a 1-km radius, dominant tree density, and canopy cover. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 50). We directly assigned SI scores to these combinations on the basis of habitat associations of the cerulean warbler outlined in Hamel (1992). We derived the suitability function for forest patch size (SI2) by ﬁtting a logistic curve (Fig. 26) to data from Robbins and others (1989) and Rosenberg and others (2000), who 59 Table 51.—Influence of forest patch size on suitability index (SI) scores for cerulean warbler habitat Suitability Index Score 1.0 Forest patch size (ha) 0.8 0.6 400 0.064 700 b 0.500 3,000b 1.000 c 1.000 5,000 0.4 SI score a a Rosenberg and others (2000). Robbins and others (1989). c Assumed value. b 0.2 0.0 0 1000 2000 3000 Forest Patch Size (ha) Figure 26.—Relationship between forest patch size and suitability index (SI) scores for cerulean warbler habitat. Equation: SI score = 1.000 / (1.000 + (524.457 * e -0.0089 * forest patch size)). Table 52.—Relationship between landscape composition and suitability index (SI) scores for cerulean warbler habitat Suitability Index Score 1.0 Landscape composition 0.8 0.6 0.4 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 Landscape Composition (proportion forest in 1-km radius) Figure 27.—Relationship between landscape composition and suitability index (SI) scores for cerulean warbler habitat. Equation: SI score = 1.047 / (1.000 + (1991.516 * e -10.673 * landscape composition)). 0.00 0.00 0.10 a 0.00 0.20a 0.00 0.30 a 0.00 0.40 a 0.00 0.50 a 0.10 0.60a 0.25 0.70 b 0.50 0.80 a 0.75 0.90 a 0.90 1.00a 1.00 a b Assumed value. Donovan and others (1997). observed that about 95 percent of all birds in FWS Region 4 were on tracts of at least 400 ha (Table 51). Recognizing the suitability of a forest patch is aﬀected by its landscape context (Rosenberg and others 1999), we ﬁt a logistic function (Fig. 27) to data (Table 52) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoint between moderately and lightly fragmented forest deﬁned the speciﬁc cutoﬀ for average (SI score = 0.500) habitat. We used the maximum value from SI2 or SI3 to account for the suitability of small patches in predominantly forested landscapes. 60 SI score a Table 53.—Influence of dominant tree density on suitability index (SI) scores for cerulean warbler habitat Suitability Index Score 1.0 Dominant tree density (trees/ha)a 0.8 0 0.6 0.0 1 1.0 14 1.0 a 0.4 SI score Assumed value. 0.2 0.0 0 2 4 Dominant Tree Density (trees/ha) Figure 28.—Relationship between dominant tree density and suitability index (SI) scores for cerulean warbler habitat. Equation: SI score = 1 – e -8.734 * dominant tree density. We used the density of dominant trees (SI4) in the HSI model and assumed that trees with a d.b.h. greater than 76.2 cm would produce the heterogeneous vertical canopy structure preferred by the cerulean warbler. On the basis of qualitative habitat descriptions by Rosenberg and others (2000), we assumed that the cerulean warbler reached its highest density in stands containing at least one dominant tree per ha. Because this bird nests almost exclusively in these trees (Weakland and Wood 2005), we also assumed that it would be absent from stands with a uniform canopy height (i.e., no dominant trees). We ﬁt an exponential function (Fig. 28) to these data points and assumed that stands with at least 14 dominant trees per ha (the maximum number observed in the WGCP during the FIA surveys of the 1990s) were associated with maximum habitat suitability (Table 53). We used data from Rosenberg and others (2000), Jones and others (2001), and Weakland and Wood (2005) to derive an inverse quadratic function (Fig. 29) that predicted habitat suitability for the cerulean warbler from canopy cover (SI5; Table 54). Canopy cover of 50 percent or less is associated with failed reproduction by this species (Jones and others 2001), so we considered these values as nonhabitat (SI score = 0.000). Rosenberg and others (2000) identiﬁed “a tall, but broken, canopy” as one of the few common denominators of cerulean warbler habitat rangewide, and we maximized the SI score at 90 percent canopy closure. However, Weakland and Wood (2005) observed the cerulean warbler selecting internal edges, so we also discounted habitat suitability for closed canopies. Nonetheless, we recognize that a dense upper canopy is needed by this species (Hamel 2000) and assigned to sites with 80 and 100 percent canopy cover an average SI score (0.500). To calculate overall HSI scores for cerulean warbler habitat, we calculated the geometric mean of the three suitability indices related to forest structure (SI1, SI4, and SI5) and the maximum value for the two suitability indices related to landscape composition (SI2 and SI3) separately and then the geometric mean of these values together. Overall SI = ((SI1 * SI4 * SI5)0.333 * Max(SI2 or SI3))0.500 61 Table 54.—Influence of canopy cover on suitability index (SI) scores for cerulean warbler habitat Suitability Index Score 1.0 Canopy cover (percent) 0.8 0.6 50 0.00 70 b 0.25 80 b 0.50 90c 0.4 100 1.00 d a Jones and others (2001). b Hamel (2000). c Rosenberg and others (2000). d Weakland and Wood (2005). 0.2 0.0 0 25 50 75 100 Canopy Cover (%) Figure 29.—Relationship between canopy cover and suitability index (SI) scores for cerulean warbler habitat. Equation: SI score = 1 / (62.548 – (1.369 * canopy cover) + (0.007612 * (canopy cover)2)). Veriﬁcation and Validation The cerulean warbler was found in 60 of the 88 subsections within the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant positive relationship between average HSI score and mean BBS route abundance across all subsections (P ≤ 0.001; rs = 0.44) and those in which this species was detected (P ≤ 0.001; rs = 0.42). The generalized linear model predicting BBS abundance from BCR and HSI for the cerulean warbler was signiﬁcant (P ≤ 0.001; R2 = 0.205), and the coeﬃcient on the HSI predictor variable was both positive (β = 0.627) and signiﬁcantly diﬀerent from zero (P = 0.023). Therefore, we considered the HSI model for the cerulean warbler both veriﬁed and validated (Tirpak and others 2009a). 62 SI score a 0.50 Chimney Swift Status The chimney swift (Chaetura pelagica) is a familiar bird found across most of North America east of the Rocky Mountains. Populations have declined in both the CH and WGCP over the last 40 years (2.6 and 1.1 percent per year). However, the high annual variability in abundance for this species prevents the identiﬁcation of signiﬁcant trends (Sauer and others 2005; Table 5). This Ron Austing, used with permission bird has a regional combined score of 16 and requires management attention in the CH. However, in the WGCP, the chimney swift is only a planning and responsibility species with a regional combined score of 14 (Table 1). Natural History The range of the chimney swift, a small, long-distance migrant, expanded dramatically with European settlement and the increase in artiﬁcial nest structures (e.g., chimneys) that followed (Cink and Collins 2002). Prior to European settlement, this species probably was distributed thinly and relied on tree cavities for nesting. Nesting in trees is now rare (Graves 2004) and most nests and roosts are concentrated in urban areas (Cink and Collins 2002). This species is weakly territorial (typically one nest per cavity), and population declines may be due to the loss of nest sites as large, open chimneys become scarce. Home ranges are largely unknown. Model Description For a bird that occurs in such close association with humans, few data are available on the habitat preferences of the chimney swift. We assumed that habitat suitability for this species was primarily a function of the availability of nest and roost sites within the proper landscape context (i.e., open chimneys near foraging areas). To identify these locations, we estimated the proportion of foraging habitats in a 1-km buﬀer around each pixel of developed landcover. We assumed that this bird could travel 1 km from nesting-roosting areas to foraging habitats (deﬁned as water, grassland, pasture-hay, recreational grasses, or forest landcover classes) and that these habitats had to be more than 1 ha to accommodate the aerial foraging maneuvers of this species. Because the chimney swift is semi-colonial, we also assumed that that as foraging habitat increased in the 1-km buﬀer, developed pixels were increasingly isolated and would be of lower suitability (Table 55). We used a quadratic curve (Fig. 30) to quantify the relationship between landscape composition and habitat suitability for this species. Veriﬁcation and Validation The chimney swift occurred in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.50) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the chimney swift was signiﬁcant (P ≤ 0.001; R2 = 0.208), and the coeﬃcient on the HSI predictor variable was positive (β = 5.043) but not signiﬁcantly diﬀerent from zero (P = 0.524). Therefore, we considered the HSI model for the chimney swift veriﬁed but not validated (Tirpak and others 2009a). 63 Table 55.—Influence of proportion of foraging habitata within 1-km buffer around potential nesting-roosting sitesb on suitability index (SI) scores for chimney swift habitat Suitability Index Score 1.0 Proportionc of foraging habitat around potential nestingroosting sites 0.8 0.6 0.4 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 Foraging Habitat in 1-km Radius Around Nest Habitat Figure 30.—Relationship between proportion of foraging habitat within 1-km buffer around potential nesting/roosting sites on suitability index (SI) scores for chimney swift habitat. Equation: SI score = (-0.0769 + (4.0734 * proportion foraging cover) - (3.8462 * (proportion foraging cover2))). SI score 0.0 0.00 0.1 0.25 0.2 0.50 0.3 0.75 0.4 1.00 0.5 1.00 0.6 1.00 0.7 1.00 0.8 0.75 0.9 0.25 1.0 0.25 a Foraging habitat = water, grassland, pasture-hay, recreational grasses, forest > 1 ha. b Nesting-roosting site = any developed landcover. Assumed value. c 64 Chuck-will’s-widow Status The chuck-will’s-widow (Caprimulgus carolinensis) is a neotropical migrant that breeds in the southeastern United States. It has experienced small yet signiﬁcant declines in the WGCP over the last 40 years (1.3 percent per year; Sauer and others 2005). Populations in the CH have remained relatively stable during the Chandler S. Robbins, Patuxent Bird Identification InfoCenter same period (Table 5). Chuck-will’s-widow is as Photo used with permission Bird of Conservation Concern and a PIF species in need of management attention in the WGCP (regional combined score = 16). This species has no special conservation status in the CH (regional combined score = 14; Table 1). Natural History The chuck-will’s-widow, like all nightjars, is nocturnal and most active on moonlit nights. Because of this behavior and its cryptic coloration, this species is diﬃcult to study and few systematic investigations of its habitat, demography, or population status have been conducted. Most of the information on chuck-will’s-widow is anecdotal and coincident to studies of other species (Straight and Cooper 2000). The chuck-will’s-widow occupies woodland habitats interspersed with large openings in which the bird forages at night. Calling males are equally abundant among suburban, pasture, and forested landscapes (Cooper 1981). Urban habitats are unsuitable (Straight and Cooper 2000). The chuck-will’s-widow prefers more open habitats than the whip-poor-will (Cooper 1981) and is unaﬀected by forest fragmentation (it may even beneﬁt from it). Drier sites also are preferred. Model Description The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 56). We directly assigned SI scores to these combinations on the basis of data from Hamel (1992) on the habitat associations of the chuck-will’s-widow in the Southeast. The realized suitability of the sites identiﬁed in SI1 depends largely on landscape context. Cooper (1981) found that the abundance of chuck-will’s-widow was highest in areas with equal amounts of forest and agriculture. Therefore, we used the proportion of these two habitats in a 500-m radius window (SI2) in the HSI model. We assigned the maximum SI score to landscapes characterized by 50 percent forest and 50 percent agriculture. We reduced these scores as landscapes varied from this optimal conﬁguration towards a more open or a more forested composition with a stronger reduction in suitability for increasingly forested landscapes (Table 57). The overall HSI score for chuck-will’s-widow is based solely on SI2, which incorporates the results from SI1. Overall HSI = SI2 65 Table 56.—Relationship of landform, landcover type, and successional age class to suitability index scores for chuck-will’s-widow habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.334 0.834 1.000 Deciduous 0.000 0.000 0.083 0.167 0.167 Evergreen 0.000 0.000 0.334 0.834 1.000 Mixed 0.000 0.000 0.334 0.834 1.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.334 0.834 1.000 Deciduous 0.000 0.000 0.083 0.167 0.167 Evergreen 0.000 0.000 0.334 0.834 1.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.334 0.834 1.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 Deciduous 0.000 0.000 0.334 (0.250) 0.167 0.834 (0.583) 0.333 1.000 (0.667) 0.333 Evergreen 0.000 0.000 Mixed 0.000 0.000 0.334 (0.250) 0.334 0.834 (0.583) 0.834 1.000 (0.667) 1.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Veriﬁcation and Validation The chuck-will’s-widow was found in 86 of the 88 subsections within the CH and WGCP. Spearman rank correlations yielded similar results when analysis included all subsections and only those subsections in which this species was detected: signiﬁcant (P ≤ 0.001 and 0.003, respectively) positive associations (rs = 0.34 and 0.32, respectively) between average HSI score and mean BBS route abundance. The generalized linear model predicting BBS abundance from BCR and HSI for the chuck-will’s-widow was signiﬁcant (P ≤ 0.001; R2 = 0.312), and the coeﬃcient on the HSI predictor variable was positive (β = 0.569) but not signiﬁcantly diﬀerent from zero (P = 0.415). Therefore, we considered the HSI model for the chuck-will’s-widow veriﬁed but not validated (Tirpak and others 2009a). 66 Table 57.—Suitability index scores for chuck-will’s-widow habitat based on proportion of nesting-roosting and foraging habitat within 500-m radius landscape Proportion foraginga Proportion nest and roostb 0.0 0.0 0.0 0.1 0.0 0.2 0.0 0.3 0.0 0.4 0.0 0.5 0.0 0.6 0.0 0.7 0.0 0.8 0.0 0.9 0.0 0.1 0.0 0.0 0.0 0.1 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.0 0.0 0.1 0.2 0.4 0.6 0.6 0.6 0.5 0.3 0.0 0.1 0.2 0.4 0.6 0.6 0.8 0.8 0.4 0.0 0.2 0.4 0.6 0.8 0.8 1.0 0.5 0.0 0.2 0.4 0.6 0.8 1.0c 0.6 0.0 0.2 0.4 0.6 0.8 0.7 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.9 0.0 0.2 1.0 0.0 1.0 0.0 a Foraging = pasture-hay, recreational grasses, grasslands, and emergent herbaceous wetland landcovers or grass-forb and shrub-seedling successional age classes. b Nest and roost = habitats identified in SI1 (Table 56). c Cooper (1981). 67 Eastern Wood-pewee Status The eastern wood-pewee (Contopus virens) is a longdistance neotropical migrant that breeds throughout the temperate regions of eastern North America (McCarty 1996). This species reaches its highest densities in the Ozark Mountain region of the CH, where it has a regional combined score of 15 (Table 1). In the WGCP, the eastern wood-pewee has a regional combined score of 16. This bird is one requiring management attention in both BCRs, with declining populations in both regions (Sauer and others 2005) (Table 5). Jeffrey A Spendelow, Patuxent Bird Identification InfoCenter Photo used with permission Natural History The eastern wood-pewee is a common species in woodlands of all types (deciduous, mixed, and evergreen). However, this species consistently selects open park-like conditions on xeric sites with limited canopy cover and low shrub densities (Robbins and others 1989; McCarty 1996). The eastern wood-pewee is positively associated with increasing density of sawtimber trees, reaching a threshold at 100 trees per ha where a negative relationship develops (Best and Stauﬀer 1986, Robbins and others 1989). The eastern wood-pewee, common in both forest interiors and edges, generally is areainsensitive, and may occupy fragments as small as 0.3 ha (Blake and Karr 1987, Robbins and others 1989). Its cryptic nests high in the canopy may limit predation and parasitism, allowing the pewee to occupy small fragments without the adverse eﬀects on reproduction common to other open-cup nesters (McCarty 1996, Knutson and others 2004, Underwood and others 2004). This species is not found in riparian corridors with less than 24 percent forest cover in the landscape (Perkins and others 2003b). Model Description The HSI model for the eastern wood-pewee includes ﬁve variables: landform, landcover, successional age class, percent forest in a 1-km radius, and density of sawtimber trees (> 28 cm d.b.h.). The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 58). We directly assigned SI scores to these combinations on the basis of habitat associations of the eastern wood-pewee reported by Hamel (1992). This species can occupy small forest fragments but may require a minimum amount of forest in the landscape. Therefore, our model did not include a forest patch size function but relied solely on landscape composition (SI2). We used a logistic function (Fig. 31) to predict SI scores from the percentage of forest in the landscape (Table 59). 68 Table 58.—Relationship of landform, landcover type, and successional age class to suitability index scores for eastern wood-pewee habitat. Values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.167 0.250 0.500 0.667 Transitional-shrubland 0.000 0.167 0.250 0.500 0.667 Deciduous 0.000 0.167 0.250 0.500 0.667 Evergreen 0.000 0.250 0.333 0.667 1.000 Mixed 0.000 0.000 0.167 0.667 1.000 Orchard-vineyard 0.000 0.167 0.250 0.500 0.667 Woody wetlands 0.000 0.250 0.333 0.417 0.500 Low-density residential 0.000 0.000 0.167 0.583 0.834 Transitional-shrubland 0.000 1.000 0.000 0.167 (0.333) 0.167 0.667 Deciduous 0.000 (0.333) 0.000 0.583 0.834 Evergreen 0.000 0.250 0.333 0.667 1.000 Mixed 0.000 0.000 0.167 0.667 1.000 Orchard-vineyard 0.000 0.000 0.167 0.583 0.834 Woody wetlands 0.000 0.250 0.333 0.500 0.667 Low-density residential 0.000 0.000 0.167 0.667 1.000 Transitional-shrubland 0.000 1.000 0.000 0.167 (0.250) 0.167 0.667 Deciduous 0.000 (0.167) 0.000 0.667 1.000 Evergreen 0.000 1.000 0.000 0.333 (0.250) 0.167 0.667 Mixed 0.250 (0.167) 0.000 0.667 1.000 Orchard-vineyard 0.000 0.000 0.167 0.667 1.000 Woody wetlands 0.000 0.250 0.333 0.500 0.667 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge We included density of sawtimber trees in the HSI model and used the threshold of 100 trees per ha observed by Best and Stauﬀer (1986) as the optimal value in a quadratic function (Fig. 32) that links density of sawtimber trees (SI3) to habitat suitability. Because Best and Stauﬀer (1986) observed a reduction in wood-pewee abundance at sawtimber tree densities less than 100 trees per ha and Robbins and others (1989) observed a negative relationship between occurrence and tree density, we assumed a symmetrical decline in habitat quality as sawtimber tree density increased or decreased above or below the optimum (Table 60). To calculate the overall HSI score, we determined the geometric mean of individual SI functions relating to forest structure (SI1 and SI3) and then calculated the geometric mean of this value and landscape composition (SI2). Overall HSI = ((SI1 * SI3)0.500 * SI2)0.500 69 Suitability Index Score 1.0 Table 59.—Relationship between landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for eastern woodpewee habitat 0.8 Landscape composition 0a 0.6 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 31.—Relationship between landscape composition and suitability index (SI) scores for eastern wood-pewee habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * (landscape composition) )). Suitability Index Score 0.00 10 a 0.00 20 a 0.05 30b 0.10 40 a 0.25 50 b 0.50 60 a 0.75 70b 0.90 80 a 0.95 90 b 100 1.00 a 1.00 a Assumed value. b Dononvan and others (1997). Table 60.—Influence of sawtimber tree (≥ 28 cm d.b.h.) density (trees/ha) on suitability index (SI) scores for eastern wood-pewee habitat 1.0 Sawtimber tree density 0.8 0 0.6 0.4 a 0.0 100 1.0 200 a 0.0 a Assumed value. b Best and Stauffer (1986). 0.2 0 50 100 150 SI score b 0.0 200 Sawtimber Tree Density (trees/ha) Figure 32.—Relationship between sawtimber tree (≥ 28 cm d.b.h.) density and suitability index (SI) scores for eastern wood-pewee habitat. Equation: SI score = (0.0200 * sawtimber tree density) - (0.0001 * (sawtimber tree density2)). Veriﬁcation and Validation The eastern wood-pewee was found in all 88 subsections of the CH and WGCP. Spearman rank correlation on average HSI score and mean BBS route abundance identiﬁed a signiﬁcant (P ≤ 0.001) positive association (rs = 0.46) between these two variables at the subsection scale. The generalized linear model predicting BBS abundance from BCR and HSI for the eastern wood-pewee was signiﬁcant (P ≤ 0.001; R 2 = 0.472), and the coeﬃcient on the HSI predictor variable was both positive (β = 5.183) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the eastern wood-pewee both veriﬁed and validated (Tirpak and others 2009a). 70 SI score Field Sparrow Status The ﬁeld sparrow (Spizella pusilla) is a shortdistance migrant found throughout North America east of the Rocky Mountains. Associated with early successional habitats, Deanna K. Dawson, Patuxent Bird Identification InfoCenter this species has experienced the sharp declines Photo used with permission typical of many scrub-shrub and grassland species in the East. BBS data indicate declines in populations of the ﬁeld sparrow in both the CH and WGCP (Sauer and others 2005; Table 5). The ﬁeld sparrow has a regional combined score of 17 and 15 in the CH and WGCP, respectively, but is not a Bird of Conservation Concern in either BCR (Table 1). About 20 percent of the continental population occurs in the CH (Panjabi and others 2001). Natural History The ﬁeld sparrow breeds in a variety of vegetation types, including brushy pastures, secondgrowth scrub, forest openings and edges, Christmas tree farms, orchards, nurseries, and roadsides and railroads near open ﬁelds (Carey and others 1994). Abundance increases in forested landscapes managed for early successional habitat (Yahner 2003), and this bird commonly occupies reclaimed mines (DeVault and others 2002) and savanna restoration sites (Davis and others 2000). Abundance is positively related to the size of old ﬁelds in Arkansas (Bay 1994). The ﬁeld sparrow nests on or near the ground in early spring but may nest in saplings or shrubs later in the year. Brood parasitism rates vary geographically but the ﬁeld sparrow generally is a poor cowbird host. Parasitism rates are higher in thinned forest stands than in regenerating plantations (Barber and others 2001). This species also uses grasslands, though at lower densities than in shrub-scrub habitats (Horn and others 2002). Grass type aﬀects habitat suitability, with warm-season grasses supporting higher abundance (Giuliano and Daves 2002, Walk and Warner 2000), nest density (Farrand 2005), and productivity than cool-season grasses (Giuliano and Daves 2002). Conservation Reserve Program ﬁelds serve as source habitat for the ﬁeld sparrow in Missouri (McCoy and others 1999). Model Description The model predicting habitat suitability for the ﬁeld sparrow includes six variables: landform, land cover, successional age class, canopy cover, density of small stems (< 2.5 cm d.b.h.), and the presence of grassy landcover. The ﬁrst suitability function of the ﬁeld sparrow HSI model combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 61). We used habitat associations of the ﬁeld sparrow reported by Hamel (1992) to assign SI scores to these combinations. 71 Table 61.—Relationship of landform, landcover type, and successional age class to suitability index scores for field sparrow habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.333 0.000 0.000 0.000 Deciduous 0.000 0.333 0.000 0.000 0.000 Evergreen 0.667 1.000 0.000 0.000 0.000 Mixed 0.667 1.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.333 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Woody wetlands 0.000 0.167 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.667 1.000 0.000 0.000 0.000 Deciduous 0.000 0.667 0.000 0.000 0.000 Evergreen 0.667 1.000 0.000 0.000 0.000 Mixed 0.667 1.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.667 0.000 0.000 0.000 Woody wetlands 0.000 0.333 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.667 1.000 0.000 0.000 0.000 Deciduous 0.000 1.000 0.000 0.000 0.000 Evergreen 0.667 1.000 0.000 0.000 0.000 Mixed 0.667 1.000 0.000 0.000 0.000 Orchard-vineyard 0.000 1.000 0.000 0.000 0.000 Woody wetlands 0.000 0.333 0.000 0.000 0.000 We included canopy cover (SI2) and small stem density (SI3) as SIs in our model to account for the absence of the ﬁeld sparrow from closed-canopy forests or forested sites with an open understory. We used data from Annand and Thompson (1997) (Tables 62 and 63) to ﬁt a quadratic function to canopy cover and a Gaussian function to small stem density for predicting SI scores (Fig. 33 and 34). The negative relationship between the ﬁeld sparrow and stem density is supported by Carey and others (1994), who observed a reduction in habitat suitability as “thickets of trees spread in the habitat.” Sousa (1983) constructed an HSI model that contained a negative relationship between habitat suitability and percent shrub cover. Suitability of habitat for the ﬁeld sparrow declined from optimal at 50 percent shrub cover (deﬁned as the percentage of ground shaded by a vertical projection of the canopies of woody vegetation less than 5 m) to unsuitable at 75 percent shrub cover. We did not have a quantitative estimate of the relationship between small stem density and shrub cover, so we assumed that 40,000 stems per ha would shade 75 percent of the ground. We were conservative with this estimate; lacking quantitative data, we did not want to exclude stands that might provide habitat for this species. 72 Table 62.—Influence of canopy cover on suitability index (SI) scores for field sparrow habitat Suitability Index Score 1.0 Canopy cover (percent) 0.00 0.8 0.6 a 1.000 29.26b 1.000 71.86 b 0.555 93.38 b 100.00 0.4 1 2 0.2 SI score 0.000 a 0.000 Assumed value. Annand and Thompson (1997). 0.0 0 25 50 75 100 Canopy Cover (%) Figure 33.—Relationship between canopy cover and suitability index (SI) scores for field sparrow habitat. Equation: SI score = 1.0038 + 0.0040 * (canopy cover) – 0.0001475 * (canopy cover)2. Table 63.—Influence of small stem (< 2.5 cm d.b.h.) density (stems * 1,000/ha) on suitability index (SI) scores for field sparrow habitat Small stem density SI score 0a 3.812 0.5 8.148 b 1.0 40.000a 1.0 a Suitability Index Score 0.1 b b 0.8 0.0 Sousa (1983). Annand and Thompson (1997). 0.6 0.4 Table 64.—Relationship between grass landcover and suitability index (SI) scores for field sparrow habitat 0.2 Landcover 0.0 0 10 20 30 40 Small Stem Density (stems * 1,000/ha) Grassland-herbaceous Pasture-haya a Figure 34.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for field sparrow habitat. Equation: SI score = 1.003 * e (-((small stem density / 1000) – 8.461)^2 )/ 31.0472 . SI score a 1.0 0.5 Must occur ≤ 170 meters from forested landcover. The ﬁeld sparrow often is associated with grasslands with suﬃcient perches (Carey and others 1994, Kahl and others 1985). Therefore, we included an SI function related to grasslands (SI4) in the model. Many useable grassland sites may have insuﬃcient woody cover to be classiﬁed as shrublands in the NLCD, so we required all grassland types (natural as well as pasture and hayﬁelds) to be within 170 m of a wooded edge—a distance approximating a large ﬁeld sparrow territory (Best 1974)—to be considered useable. Natural grasslands also are more likely to contain dense grass nesting sites than pastures and hayﬁelds (Giuliano and Daves 2002, Farrand 2005), so we assigned to useable natural grasslands an SI score of 1.000 and to useable pasture-hayﬁelds a score of 0.500 (Table 64). 73 To calculate the HSI score for ﬁeld sparrow habitat in forested landcovers, we calculated the geometric mean of the SI scores relating to forest structure (SI1, SI2, and SI3). We added the SI score for grasslands (SI4) to this value to determine the overall HSI score. Overall HSI = ((SI1 * SI2 * SI3)0.333 + SI4) Veriﬁcation and Validation The ﬁeld sparrow was found in 87 of the 88 subsections within the CH and WGCP. Spearman rank correlation on average HSI score and mean BBS route abundance identiﬁed a signiﬁcant (P ≤ 0.001) positive association (rs = 0.55) between these two variables within subsections where this species was detected. The generalized linear model predicting BBS abundance from BCR and HSI for the ﬁeld sparrow was signiﬁcant (P ≤ 0.001; R2 = 0.690), and the coeﬃcient on the HSI predictor variable was both positive (β = 37.060) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the ﬁeld sparrow both veriﬁed and validated (Tirpak and others 2009a). 74 Great Crested Flycatcher Status The great crested ﬂycatcher (Myiarchus crinitus), a neotropical migrant, is found throughout the forests of eastern North America and the riparian habitats of the Mississippi River watershed. Populations have remained relatively stable across most of its range, though in the WGCP they have declined by 1.3 percent per year since 1966 (Sauer and others 2005) (Table 5). This species has a regional combined score of 13 in both the CH and WGCP (Table 1). Natural History Deanna K. Dawson, Patuxent Bird Identification InfoCenter Photo used with permission The great crested ﬂycatcher is an obligate cavity nester in deciduous forest habitats of the eastern United States; it generally is absent in pure evergreen stands (Lanyon 1997). This species is not area sensitive but does require a minimum amount of forested habitat in the landscape. It may nest in patches as small as 0.2 ha and abundance may decline in forest interiors (Robbins and others 1989). The great crested ﬂycatcher does not occupy riparian corridors surrounded by less than 14.7 percent forest (Perkins and others 2003b), and detection probabilities steadily increase with increasing corridor width (Groom and Grubb 2002). The great crested ﬂycatcher forages by sallying from exposed perches (Lanyon 1997), so open forest stands are preferred. Holmes and others (2004) found that abundance was highest in heavily cut stands where one-third or more of the basal area was removed. Similarly, Moorman and Guynn (2001) found that the great crested ﬂycatcher was associated with large (0.5 ha) canopy gaps in bottomland hardwood forest in South Carolina. Snags not only provide exposed perches for foraging but also cavities for nesting, and the great crested ﬂycatcher is negatively aﬀected by the removal of snags associated with certain forestry practices (Lohr and others 2002). Where snags are lacking, this species will use nest boxes and other artiﬁcial cavities; this enables it to occupy cemeteries, suburban parks, and wooded pastures. Wakeley and Roberts (1996) found that this bird is associated with mesic sites, but this may reﬂect a preference for bottomland hardwoods over evergreen uplands in the Southeast. Model Description The HSI model for great crested ﬂycatcher includes ﬁve variables: landform, landcover, successional age class, snag density, and distance to edge. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 65). We directly assigned SI scores to these combinations on the basis of relative habitat quality associations reported by Hamel (1992) for the great crested ﬂycatcher. 75 Table 65.—Relationship of landform, landcover type, and successional age class to suitability index scores for great crested flycatcher habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.167 0.500 0.667 Transitional-shrubland 0.000 0.000 0.167 0.500 0.667 Deciduous 0.000 0.333 0.333 0.500 0.667 Evergreen 0.000 0.333 0.333 0.500 0.667 Mixed 0.000 0.333 0.333 0.667 1.000 Orchard-vineyard 0.000 0.000 0.167 0.500 0.667 Woody wetlands 0.000 0.333 0.333 0.667 1.000 Low-density residential 0.000 0.333 0.333 0.583 0.834 Transitional-shrubland 0.000 0.333 0.333 Deciduous 0.000 0.333 0.333 0.667 (0.500) 0.583 1.000 (0.667) 0.834 Evergreen 0.000 0.333 0.333 0.500 0.667 Mixed 0.000 0.333 0.333 0.667 1.000 Orchard-vineyard 0.000 0.000 0.167 0.583 0.834 Woody wetlands 0.000 0.333 0.333 0.667 1.000 Low-density residential 0.000 0.000 0.167 0.667 1.000 Transitional-shrubland 0.000 Deciduous 0.000 0.333 (0.250) 0.333 0.333 (0.250) 0.333 0.667 (0.500) 0.667 1.000 (0.667) 1.000 Evergreen 0.000 0.667 0.000 0.333 (0.250) 0.333 0.500 Mixed 0.333 (0.250) 0.333 0.667 1.000 Orchard-vineyard 0.000 0.333 0.333 0.667 1.000 Woody wetlands 0.000 0.333 0.333 0.667 1.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge The great crested ﬂycatcher relies on snags (SI2) for nesting and foraging. We ﬁt a logistic function (Fig. 35) through average snag values (8.5/ha) observed by Lohr and others (2002), assuming that this value represented average habitat suitability (SI score = 0.500) and that a higher abundance of snags would not be detrimental but increase the likelihood that this bird will use a site (Table 66). This species is associated with edges (Lanyon 1997), and its abundance declines with increasing distance from an edge (SI3). Small and Hunter (1989) found that more than 60 percent of all ﬂycatchers were less than 60 m from an edge. We assumed maximum habitat suitability at the edge and modeled the relationship between distance to edge and SI score as an inverse logistic function through these data points (Fig. 36, Table 67). To calculate the overall HSI, we determined the geometric mean of SI scores for forest structure (SI1 and SI2) and then calculated the geometric mean of this value with the edge function (SI3). Overall HSI = ((SI1 * SI2)0.500 * SI3)0.500 76 Table 66.—Influence of snag density on suitability index (SI) scores for great crested flycatcher habitat Suitability Index Score 1.0 Snag density (snags/ha) 0.8 0.6 0.4 0.0 0.000 1.9 a 0.133 8.5 a 0.500 20.0b 1.000 b 1.000 25.0 a b 0.2 SI score a Lohr and others (2002). Assumed value. 0.0 0 10 20 30 Snag Density (snags/ha) Figure 35.—Relationship between snag density and suitability index (SI) scores for great crested flycatcher habitat. Equation: SI score = 1.001 / (1 + (18.704 * e (-0.346 * snag density))). Table 67.—Influence of distance (m) to edgea on suitability index (SI) scores for great crested flycatcher habitat Suitability Index Score 1.0 Distance to edge 0.8 0 b 60 0.6 1.0 c 0.6 120b 0.1 b 0.0 150 0.4 SI score a Edge defined by nonhabitat pixels adjacent to habitat pixels (defined by SI1). b Assumed value. c Small and Hunter (1989). 0.2 0.0 0 50 100 150 200 Distance to Edge (m) Figure 36.—Relationship between distance to edge and suitability index (SI) scores for great crested flycatcher habitat. Equation: SI score = 1 - (1.000 / (1 + (28.950 * e -0.049 * distance to edge))). Veriﬁcation and Validation The great crested ﬂycatcher was found in all 88 subsections within the CH and WGCP. Spearman rank correlation on average HSI score and mean BBS route abundance failed to identify a signiﬁcant (P ≤ 0.001) association (rs = 0.55) between these two variables. The generalized linear model predicting BBS abundance from BCR and HSI for the great crested ﬂycatcher was not signiﬁcant (P = 0.152; R2 = 0.043), and the coeﬃcient on the HSI predictor variable was negative (β = -2.740) and not signiﬁcantly diﬀerent from zero (P = 0.151). Therefore, we considered the HSI model for the great crested ﬂycatcher neither veriﬁed nor validated (Tirpak and others 2009a). 77 Hooded Warbler Status The hooded warbler (Wilsonia citrina) is a long-distance migrant found throughout the deciduous forests of eastern North America. Because of area sensitivity, it is restricted to forested landscapes and disappears from the forest-prairie ecotone at the western edge of its range faster than other silvicolous species (e.g., eastern wood-pewee). Populations in the WGCP declined prior to 1990 but have since remained stable. Conversely, populations in the CH have increased (Sauer and U.S. Fish & Wildlife Service others 2005) (Table 5). This species is not a Bird of Conservation Concern in either BCR (Table 1) but it is a planning and responsibility species in the WGCP (regional combined score = 16; Table 1). Nearly 30 percent of the continental population of the hooded warbler breeds in the WGCP (Panjabi and others 2001). Natural History The hooded warbler breeds in a variety of habitats, from mixed-hardwood forests in the northern portion of its range to cypress-gum swamps in the South. Regardless of forest type, it prefers mesic sites in large forest tracts (> 15 ha; Evans-Ogden and Stutchbury 1994). Although nest success in small forest patches is not signiﬁcantly lower than in large patches (Buehler and others 2002), females may avoid small fragments and males use edge less than its availability (Norris and Stutchbury 2002, Norris and others 2000). Occupancy of a site by a nesting pair increases with shrub height and the percentage of vegetation between 1 and 2 m. This species nests in shrubs within small forest clearings or in the dense understories of closed-canopied forests. As a result, territories often include a mix of open and closed canopies. Gaps created by tree fall or selective logging are particularly attractive (≤ 0.5 ha; Annand and Thompson 1997, Moorman and others 2002, Whittam and others 2002), and the hooded warbler colonizes these sites within 1 to 5 years. Nest sites in Canada had denser ground vegetation, fewer tree stems, lower basal area of small trees, and greater basal area of large trees than control sites (Whittam and others 2002). Bisson and Stutchbury (2000) concluded that canopy gaps and density of understory vegetation were the most important factors aﬀecting site selection. Repeated burning, which removed understory vegetation, reduced hooded warbler abundance in Ohio (Artman and others 2001). This species is a common cowbird host, which may explain its sensitivity to fragmentation (Donovan and Flather 2002). Model Description The HSI model for the hooded warbler includes seven variables: landform, land cover, successional age class, small stem (< 2.5 cm d.b.h.) density, canopy cover, forest patch size, and percent forest in a 1-km landscape. 78 Table 68.—Relationship of landform, landcover type, and successional age class to suitability index scores for hooded warbler habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.167 0.667 1.000 Deciduous 0.000 0.000 0.167 0.667 1.000 Evergreen 0.000 0.000 0.000 0.334 0.667 Mixed 0.000 0.000 0.167 0.500 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.667 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.500 (0.334) 0.667 0.667 Deciduous 0.167 (0.000) 0.167 1.000 Evergreen 0.000 0.000 0.000 0.334 0.667 Mixed 0.000 0.000 0.167 0.500 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.667 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 Deciduous 0.000 0.000 0.167 (0.000) 0.167 0.500 (0.167) 0.667 0.667 (0.334) 1.000 Evergreen 0.000 0.000 0.000 Mixed 0.000 0.000 0.167 0.334 (0.167) 0.500 0.667 (0.334) 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.667 1.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 68). We directly assigned SI scores to these combinations on the basis of relative habitat quality rankings from Hamel (1992) for the hooded warbler in the Southeast. This species occupies dense understories in mature forested habitats, so we included both small stem density (SI2) and canopy cover (SI3) in our model. We ﬁt a logistic function (Fig. 37) that links small stem density to SI scores on the basis of data from Annand and Thompson (1992) and Moorman and others (2002) (Table 69). We assumed that the average stem density measured at nest sites by Moorman and others (2002) (4,700 stems/ha) was representative of ideal habitat conditions for the hooded warbler and that there was no upper threshold above which habitat suitability declined. We also ﬁt a logistic function (Fig. 38) to data from Annand and Thompson (1997) (Table 70) to link canopy cover values to SI scores. We included forest patch size (SI4) as a model predictor because of the negative eﬀect of fragmentation on this species. We used an exponential curve (Fig. 39) to predict habitat 79 Table 69.—Influence of small stem (< 2.5 cm d.b.h.) density (stems * 1,000/ha) on suitability index (SI) scores for hooded warbler habitat Suitability Index Score 1.0 Small stem density 0.8 0.6 0.000 0.000 2.077 b 0.039 4.700c 1.000 4.717 0.4 SI score a b 10.000 1.000 a 1.000 a Assumed value. b Annand and Thompson (1992). c Moorman and others (2002). 0.2 0.0 0.0 2.5 5.0 7.5 10.0 Small Stem Density (stems * 1,000/ha) Figure 37.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) on suitability index (SI) scores for hooded warbler habitat. Equation: SI score = 1.000 / (1.000 + (102634.340 * e -4.017 * (small stem density / 1000))). Table 70.—Influence of canopy cover on suitability index (SI) scores for hooded warbler habitat Canopy cover (percent) Suitability Index Score 0.00a 0.0 29.26 b 0.0 71.86 b 0.6 93.38b 1.0 95.58 b 1.0 96.59 b 1.0 a b Assumed value. Annand and Thompson (1997). 0.0 0 25 50 75 100 Canopy Cover (%) Figure 38.—Relationship between canopy cover on suitability index (SI) scores for hooded warbler habitat. Equation: SI score = 1.024 / (1.000 + (3823.776 * e -0.120 * canopy cover)). suitability from forest patch size on the basis of data from Evans-Ogden and Stutchbury (1994) and Kilgo and others (1998). To convert riparian widths reported by Kilgo and others (1998) to forest patch sizes, we assumed that all riparian strips were 10 km long (Table 71). The suitability of a speciﬁc forest patch is inﬂuenced by the percentage of forest in the landscape (SI5). Small patches that otherwise would be unsuitable may be occupied when in close proximity to a large forest block or in a predominantly forested landscape (Rosenberg and others 1999). To capture this relationship, we ﬁt a logistic function (Fig. 40) to data (Table 72) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoints between these classes (30 and 70 percent 80 SI score Table 71.—Influence of forest patch size on suitability index (SI) scores for hooded warbler habitat Suitability Index Score 1.0 Forest patch size (ha) 0.8 0.6 0.4 0.2 15 0.00 25 b 0.65 50 b 0.74 100b 0.86 200 b 0.97 500 b 0 1000 1.00 2,500b 1.00 a 2000 b Forest Patch Size (ha) 1.00 b 1,000 0.0 SI score a Evans-Ogden and Stutchbury (1994). Kilgo and others (1998). Figure 39.—Relationship between forest patch size and suitability index (SI) scores for hooded warbler habitat. Equation: SI score = 0.994 * (1 - e -0.024 * forest patch size). Table 72.—Relationship between landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for hooded warbler habitat Suitability Index Score 1.0 Landscape composition 0.8 SI score 0a 0.6 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 40.—Relationship between landscape composition and suitability index (SI) scores for hooded warbler habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * (landscape composition) )). 0.00 10 a 0.00 20 a 0.05 30 b 0.10 40a 0.25 50 b 0.50 60 a 0.75 70 b 0.90 80a 0.95 90 b 100 a b 1.00 a 1.00 Assumed value. Dononvan and others (1997). forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. We used the maximum SI score from SI4 or SI5 to account for the higher suitability of small forest patches in a heavily forested landscape. The overall HSI score was calculated as the geometric mean of the geometric mean of the SI values from the landform, landcover, and successional age class matrix, small stem density, and canopy cover functions (SI1, SI2, and SI3) multiplied by the maximum value of either the forest patch size or percent forest in the 1-km radius landscape functions (SI4 and SI5). Overall HSI = ((SI1 * SI2 * SI3)0.333 * Max(SI4 or SI5))0.500 81 Veriﬁcation and Validation The hooded warbler was found in 84 of the 88 subsections within the CH and WGCP. Spearman rank correlations identiﬁed signiﬁcant positive associations between average HSI score and mean BBS route abundance across all subsections (P ≤ 0.001; rs = 0.49) and subsections within which this species was detected (P ≤ 0.001; rs = 0.42). The generalized linear model predicting BBS abundance from BCR and HSI for the hooded warbler was signiﬁcant (P ≤ 0.001; R2 = 0.551), and the coeﬃcient on the HSI predictor variable was both positive (β = 8.190) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the hooded warbler both veriﬁed and validated (Tirpak and others 2009a). 82 Kentucky Warbler Status The Kentucky warbler (Oporornis formosus) breeds throughout the southeastern United States; densities are highest west of the Appalachian front. Populations have been stable in the CH over the last 40 years, but have declined in the WGCP by 2.2 percent per year during this period (Table 5). This species requires management attention in both regions (regional combined score = 18 and 19 in the CH and WGCP, respectively). A high percentage of the continental population breeds in both U.S. Fish & Wildlife Service BCRs (28 and 22 percent, respectively; Panjabi and others 2001). The species is an FWS Bird of Conservation Concern in the WGCP (Table 1). Natural History The Kentucky warbler, a long-distance migrant, breeds in mature moist deciduous forests of the Southeast. It is a forest-interior specialist, primarily because of low productivity and survival in edge and early successional habitats (Morse and Robinson 1999; Robinson and Robinson 2001). The Kentucky warbler occupies fragments as small as 2.4 ha (Blake and Karr 1987) but tracts larger than 500 ha are considered the minimum size necessary to support sustainable populations (McDonald 1998). A dense understory is a common feature of nesting sites. Ground cover averaged 46 percent in Kentucky warbler territories in Missouri (Wenny and others 1993), and vegetation of less than 1.5 m was denser around nests than random sites in South Carolina (Kilgo and others 1996). Dense vegetation (0.3 to 1 m) was also associated with higher numbers of the Kentucky warbler in Maryland (Robbins and others 1989). Mesic sites are universally selected (McShea and others 1995, McDonald 1998, Gram and others 2003). Model Description The habitat suitability model for the Kentucky warbler includes six variables: landform, landcover, successional age class, small stem (< 2.5 cm d.b.h.) density, forest patch size, and percent forest in the landscape. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 73). We relied on relative habitat quality associations reported by Hamel (1992) to assign SI scores to these combinations. However, we increased SI scores for shrub-seedling stands on the basis of data from Thompson and others (1992). The Kentucky warbler nests at the base of shrubs and occupies habitats containing high densities of small stems (SI2). We used data on the relative abundance of this species from Wenny and others (1993), Kilgo and others (1996), and Annand and Thompson (1997) to derive a logistic function (Fig. 41) that predicts habitat suitability from small stem density (Table 74). 83 Table 73.—Relationship of landform, landcover type, and successional age class to suitability index scores for Kentucky warbler habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.667 0.417 0.667 0.667 Deciduous 0.000 0.667 0.417 0.667 0.667 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.333 0.167 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 1.000 0.667 1.000 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 Deciduous 0.000 0.333 (0.000) 0.667 0.167 (0.000) 0.334 0.333 (0.000) 0.667 0.333 (0.000) 0.667 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.333 0.167 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 1.000 0.667 1.000 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 Deciduous 0.000 0.333 (0.000) 0.500 0.167 (0.000) 0.250 0.333 (0.000) 0.500 0.333 (0.000) 0.500 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.333 0.167 0.333 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 1.000 0.667 1.000 1.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge We used a logarithmic function (Fig. 42) to quantify the relationship between forest patch size (SI3) and habitat suitability on the basis of minimum patch size observations by Hayden and others (1985) and occupancy rates in diﬀerent patch sizes reported by Robbins and others (1989) (Table 75). However, the suitability of a speciﬁc forest patch is inﬂuenced by its landscape context (SI4). Because the Kentucky warbler is particularly sensitive to fragmentation (Lynch and Whigham 1984), we used a 10-km window to characterize the landscape. We ﬁt a logistic function (Fig. 43) to data (Table 76) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoints between these classes (30 and 70 percent forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. 84 Table 74.—Influence of small stem (< 2.5 cm d.b.h.) density (stems/ha) on suitability index (SI) scores for Kentucky warbler habitat Suitability Index Score 1.0 Small stem density 0.8 0.6 0.000 2.077b 0.000 0.316 3.000c 0.500 b 1.000 3.812 0.4 8.148b 47.600 0.2 a 0.0 0.0 b 2.5 5.0 7.5 10.0 Small Stem Density (stems * 1,000/ha) SI score a 1.000 d 1.000 Assumed value. Annand and Thompson (1997). c Wenny and others (1993). d Kilgo and others (1996). Figure 41.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for Kentucky warbler habitat. Equation: SI score = 1.026 / (1.000 + (111.558 * e -1.707 * (small stem density / 1000))). Table 75.—Influence of forest patch size on suitability index (SI) scores for Kentucky warbler habitat Suitability Index Score 1.0 Forest patch size (ha) 8 0.8 17 0.0 b 300 0.6 SI score a 0.5 b 1.0 a Hayden and others (1985). b Robbins and others (1989). 0.4 0.2 0.0 0 100 200 300 Forest Patch Size (ha) Figure 42.—Relationship between forest patch size and suitability (SI) scores for Kentucky warbler habitat. Equation: SI score = 0.248 * ln(forest patch size) – 0.377. To calculate the overall HSI score, we determined the geometric mean of SI scores for functions relating to forest structure (SI1 and SI2) and landscape composition (SI3 and SI4) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI2)0.500 * (SI3 * SI4)0.500)0.500 85 Suitability Index Score 1.0 Table 76.—Relationship between landscape composition (percent forest in 10-km radius) and suitability index (SI) scores for Kentucky warbler habitat 0.8 Landscape composition 0a 0.6 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 10-km radius) Figure 43.—Relationship between landscape composition and suitability index (SI) scores for Kentucky warbler habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * (landscape composition) )). 0.00 10 a 0.00 20 a 0.05 30b 0.10 40 a 0.25 50 b 0.50 60 a 0.75 70b 0.90 80 a 0.95 90 b 100 a b 1.00 a Assumed value. Dononvan and others (1997). Veriﬁcation and Validation The Kentucky warbler was found in all 88 subsections of the CH and WGCP. Spearman rank correlations identiﬁed a signiﬁcant positive association between average HSI score and mean BBS route abundance across all subsections (P ≤ 0.001; rs = 0.71). The generalized linear model predicting BBS abundance from BCR and HSI for the Kentucky warbler was signiﬁcant (P ≤ 0.001; R2 = 0.346), and the coeﬃcient on the HSI predictor variable was both positive (β = 6.351) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the Kentucky warbler both veriﬁed and validated (Tirpak and others 2009a). 86 SI score 1.00 Louisiana Waterthrush Status The Louisiana waterthrush (Seirus motacilla) is a longdistance neotropical migrant found throughout the deciduous forests of the eastern and central United States. The small population in the WGCP has remained relatively stable since 1966 while the larger population in the CH has increased by 2.6 percent annually (Sauer and others 2005) (Table 5). This species is a Bird of Conservation Concern in both regions (Table 1). Charles H. Warren, images.nbii.gov However, PIF diﬀerentiates the priority for this species in the CH (planning and responsibility, regional combined score = 15) and WGCP (management attention, regional combined score = 18; Table 1). Natural History As its name implies, the Louisiana waterthrush is associated with water throughout its range (Robinson 1995). Densities are highest along gravel-bottomed, ﬁrst- and second-order streams ﬂowing through large (> 350 ha) tracts of mature deciduous forest (Robbins and others 1989, Robinson 1995). Birds also breed at lower densities along mud-bottomed streams in cypress swamps and bottomland hardwood forests (Hamel 1992, Robinson 1995). Prosser and Brooks (1998) developed and validated an HSI model for the Louisiana waterthrush in central Pennsylvania that included eight variables: canopy cover (> 80 percent considered ideal), shrub cover (< 25 percent), ratio of deciduous to conifer cover (30 to 69 percent, mostly reﬂecting hemlock dominance along streams in the Northeast), herbaceous cover (< 25 percent), stream order (ﬁrst- or second-order with well developed pools and riﬄes), water clarity and substrate (clear and rocky or sandy), nesting cover (presence of uprooted trees or creviced, steep banks), and forest area (> 350 ha). Model Description Our HSI model for the Louisiana waterthrush included eight variables: landform, landcover, successional age class, distance to stream, canopy cover, small stem (< 2.5 cm d.b.h.) density, forest patch size, and percent forest in a 1-km radius. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 77). We directly assigned SI scores to these combinations on the basis of vegetation and successional age class associations outlined in Hamel (1992). We included distance to stream (SI2) as a variable because the waterthrush uses streams and creeks for foraging and nesting. The Louisiana waterthrush restricts its foraging to the streambed and bank, so we assumed a sharp decline in suitability with increasing distance to a stream (Table 78). We used an inverse logistic function to characterize this relationship (Fig. 44). 87 Table 77.—Relationship of landform, landcover type, and successional age class to suitability index (SI) scores for Louisiana waterthrush habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.500 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.167 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Woody wetlands 0.000 0.000 0.000 0.500 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.500 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.167 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.500 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.250 0.500 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.167 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.334 0.667 Table 78.—Relationship between distance to stream and suitability index (SI) scores for Louisiana waterthrush habitat. Suitability Index Score 1.0 Distance to stream (m)a 0.8 0.6 0.4 0.0 0 50 100 150 Distance to Stream (m) Figure 44.—Relationship between distance to stream and suitability index (SI) scores for Louisiana waterthrush habitat. Equation: SI score = 1 - (1.0015 / (1 + (104411.5 * e -0.1926 * distance to stream ))). 88 0 1.0 30 1.0 60 0.5 90 0.0 120 0.0 a 0.2 SI score Assumed value. Table 79.—Relationship between canopy cover and suitability index (SI) scores for Louisiana waterthrush habitat Suitability Index Score 1.0 Canopy cover (percent)a 0.8 0.6 0.4 0.2 0.0 0 25 50 75 100 Canopy Cover (%) Figure 45.—Relationship between canopy cover and suitability index (SI) scores for Louisiana waterthrush habitat. Equation: SI score = (1.0313 / (1 + (175.8083 * e -0.0864 * canopy cover))). 0.0 10 0.0 20 0.0 30 0.0 40 0.2 50 0.2 60 0.7 70 0.7 80 0.7 90 1.0 a Prosser and Brooks (1998). Table 80.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1,000/ha) and suitability index (SI) scores for Louisiana waterthrush habitat 1.0 Suitability Index Score SI score 0 Small stem density 0.8 0 a 1.000 5.803a 0.767 9.086 0.4 1.000 a 2.519 0.6 a 25.000 a 0.2 SI score b 0.349 b 0.000 Prosser and Brooks (1998). Assumed value. 0.0 0 10 20 30 Small Stem Density (stems * 1,000/ha) Figure 46.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for Louisiana waterthrush habitat. Equation: SI score = 1 - (1.000 / (1 + (113.261 * e -0.592 * (small stem density / 1000)))). We also included canopy cover (SI3) and small stem density (SI4) as variables based on the preference of this species for mature forested sites with closed canopies and open understories. We ﬁt logistic (Fig. 45) and inverse logistic (Fig. 46) functions to data adapted from the HSI model of Prosser and Brooks (1998) for canopy cover (Table 79) and small stem density (Table 80), respectively. Forest patch size (SI5) aﬀects the occupancy of habitats by the Louisiana waterthrush. To predict habitat suitability from forest patch size, we ﬁt a logarithmic function (Fig. 47) to data from Hayden and others (1985) and Robbins and others (1989) (Table 81) 89 Table 81.—Relationship between forest patch size and suitability index (SI) scores for Louisiana waterthrush habitat Forest patch size (ha) SI score Suitability Index Score 1.0 0.8 42.2a 350 0.6 0.0 b 3,200 0.5 b 1.0 a Hayden and others (1985). b Robbins and others (1989). 0.4 0.2 0.0 0 1000 2000 3000 Forest Patch Size (ha) Figure 47.—Relationship between forest patch size and suitability index (SI) scores for Louisiana waterthrush habitat. Equation: SI score = 1.000 – (1.010 * e -0.0003 * (forest patch size ^ 1.321)). Table 82.—Relationship between landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for Louisiana waterthrush habitat Suitability Index Score 1.0 Landscape composition 0.8 0 0.6 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 48.—Relationship between landscape composition and suitability index (SI) scores for Louisiana waterthrush habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * (landscape composition) )). a 0.00 10a 0.00 20 a 0.05 30 b 0.10 40 a 0.25 50b 0.50 60 a 0.75 70 b 0.90 80 a 0.95 90b 100 a b 1.00 a Assumed value. Donovan and others (1997). on the detection probabilities of the Louisiana waterthrush in patches of varying size. However, forest patch size alone may not be an appropriate measure of a site’s suitability. In predominantly forested landscapes, small patches otherwise not suitable may be occupied due to their proximity to large forest blocks (Rosenberg and others 1999). To capture this relationship, we ﬁt a logistic function (Fig. 48) to data (Table 82) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed the midpoints between these classes (30 and 70 percent forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. We used the maximum SI score from 90 SI score 1.00 SI5 or SI6 to ensure that small forest blocks in predominantly forested landscapes were assigned an appropriate suitability score. To calculate the overall HSI, we determined the geometric mean of SI scores for forest structure (SI1, SI3, and SI4) and landscape composition (Max (SI5 or SI6) and SI2) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI3 * SI4)0.333 * (Max (SI5 or SI6) * SI2)0.500)0.500 Veriﬁcation and Validation The Louisiana waterthrush was found in all 88 subsections of the CH and WGCP. Spearman rank correlation on average HSI score and mean BBS route abundance per subsection identiﬁed a signiﬁcant (P ≤ 0.001) positive association (rs = 0.56) between these two variables. The generalized linear model predicting BBS abundance from BCR and HSI for the Louisiana waterthrush was signiﬁcant (P ≤ 0.001; R2 = 0.263), and the coeﬃcient on the HSI predictor variable was both positive (β = 3.664) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the Louisiana waterthrush both veriﬁed and validated (Tirpak and others 2009a). 91 Mississippi Kite Status The Mississippi kite (Ictinia mississippiensis), a neotropical migrant raptor, is restricted to the Coastal Plains as well as the lower Mississippi and Red River Valleys. Like many birds of prey, this species has exhibited dramatic recoveries over the last 25 years from historical lows in the 1970s. However, its general scarcity prevents BBS from detecting statistically signiﬁcant trends (Sauer and others 2005; Table 5). The Mississippi kite is not a Bird of Conservation Concern in the CH or WGCP (Table 1). It has a regional combined score of 14 in the CH and 16 in the WGCP. Peter S. Weber, www.wildbirdphotos.com Photo used with permission Natural history The Mississippi kite exhibits two breeding strategies within its range. In the southern Great Plains, it is a colonial nester that often inhabits urban areas. In the Mississippi Valley and farther east, this bird is less colonial and nests singly in large trees in bottomland forest and riparian woodlands. Nests from birds within the eastern population generally are located in large (> 22 ha) unfragmented forest near open habitats where birds forage aerially (Parker 1999). Model Description The HSI model for the Mississippi kite includes six variables: landform, land cover, successional age class, forest patch size, interspersion of forest and open habitats, and density of dominant trees. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 83). We directly assigned SI scores to these combinations on the basis of relative habitat quality ranks reported by Hamel (1992) for this species. However, we restricted the Mississippi kite to sawtimber stands based on its preference for mature forest stands (Parker 1999). We also included forest patch size (SI2) in the model and used the range and mean of patch sizes reported by Barber and others (1998) to deﬁne the minimum, maximum, and average patch sizes associated with nonhabitat, optimal, and average habitat suitability for this function, respectively (Table 84; Fig. 49). The Mississippi kite requires large patches of forest and grassland in a speciﬁc landscape context (Parker 1999, Coppedge and others 2001). We used the relative amount of these habitats within a 1-km radius as an index to their interspersion at the landscape scale (SI3). We assumed that habitat suitability was optimal in open habitats with few trees (70 to 90 percent agriculture or grassland) or landscapes containing moderate forest cover interspersed with open habitats (60 to 70 percent forest; Table 85). 92 Table 83.—Relationship of landform, landcover type, and successional age class to suitability index scores for Mississippi kite habitat. Values in parentheses apply to West Gulf Coastal Plain/Ouachitas. Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.500 Transitional-shrubland 0.000 0.000 0.000 0.000 0.500 Deciduous 0.000 0.000 0.000 0.000 0.500 Evergreen 0.000 0.000 0.000 0.000 0.333 Mixed 0.000 0.000 0.000 0.000 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.500 Woody wetlands 0.000 0.000 0.000 0.000 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.500 Transitional-shrubland 0.000 0.000 0.000 0.000 0.333 Deciduous 0.000 0.000 0.000 0.000 0.500 Evergreen 0.000 0.000 0.000 0.000 0.333 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.000 0.000 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.500 Woody wetlands 0.000 0.000 0.000 0.000 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.500 Transitional-shrubland 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.333 (0.167) 0.500 Evergreen 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.333 (0.167) 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.500 Woody wetlands 0.000 0.000 0.000 0.000 1.000 The Mississippi kite nests in dominant trees (SI4) that extend above the canopy. Parker (1999) identiﬁed old-growth stands and isolated trees as preferred nesting substrates for this species, and Barber and others (1998) observed the Mississippi kite using nest trees that were higher and larger in d.b.h. than those in the surrounding overstory. We assumed that a tree with a d.b.h. greater than 76.2 cm in a sawtimber stand would extend above the canopy and provide an adequate nest substrate for this species. We further assumed that one dominant tree per ha would satisfy this requirement and that the Mississippi kite would be absent from stands with a uniform canopy (zero dominant trees/ha). We ﬁt an exponential function (Fig. 50) to the values between these data points. Stands with 14 dominant trees per ha (the maximum observed in the WGCP during the FIA surveys of the 1990s) were associated with maximum habitat suitability (Table 86). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure (SI1 and SI4) and landscape composition (SI2 and SI3) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI4)0.500 * (SI2 * SI3)0.500)0.500 93 Table 84.—Influence of forest patch size on suitability index (SI) scores for Mississippi kite habitat Suitability Index Score 1.0 Forest patch size (ha)a 0.8 0.6 22 0.0 683 0.5 3,000 1.0 a 0.4 SI score Barber and others (1998). 0.2 0.0 0 1000 2000 3000 Forest Patch Size (ha) Figure 49.—Relationship between forest patch size and suitability index (SI) scores for Mississippi kite habitat. Equation: SI score = 1.002 – (1.000 * e -0.0002 * (forest patch size ^ 1.278)). Table 85.—Suitability index scores for Mississippi kite habitat based on proportion of cells providing roosting and nesting habitat within 1-km radius Proportion foresta Proportion agriculturegrasslandb 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.50 0.50 0.50 0.50 0.1 0.00 0.00 0.00 0.00 0.20 0.20 0.40 0.60 0.60 0.60 0.2 0.00 0.00 0.00 0.00 0.40 0.40 0.60 0.80 0.80 0.3 0.00 0.00 0.00 0.00 0.60 0.60 0.80 1.00 0.4 0.35 0.40 0.40 0.60 0.80 0.80 1.00 0.5 0.50 0.50 0.55 0.70 0.70 0.60 0.6 0.60 0.70 0.75 0.90 0.80 0.7 0.70 0.75 1.00 1.00 0.8 0.80 0.90 1.00 0.9 0.80 1.00 1.0 0.80 a b Woody wetlands, deciduous forest, low-density residential. Open water, open fields (natural or cultivated), emergent herbaceous wetland. 94 Table 86.—Influence of dominant tree (d.b.h. > 76.2 cm) density (trees/ha) on suitability index (SI) scores for Mississippi kite habitat Suitability Index Score 1.0 Dominant tree densitya 0.8 0.6 0 0.0 1 1.0 14 1.0 a 0.4 SI score Assumed value. 0.2 0.0 0 2 4 Dominant Tree Density (trees/ha) Figure 50.—Relationship between dominant tree (> 76.2 cm d.b.h.) density and suitability index (SI) scores for Mississippi kite habitat. Equation: SI score = 1 – e -8.734 * dominant tree density. Veriﬁcation and Validation The Mississippi kite was found in 49 of the 88 subsections within the CH and WGCP. Spearman rank correlations based on all subsections yielded a signiﬁcant (P = 0.003) positive association (rs = 0.31) between average HSI score and mean BBS route abundance. However, this association was not evident when the correlation considered only subsections in which this species was found. The generalized linear model predicting BBS abundance from BCR and HSI for the Mississippi kite was signiﬁcant (P ≤ 0.001; R2 = 0.287); however, the coeﬃcient on the HSI predictor variable was negative (β = -0.176). Therefore, we considered the HSI model for the Mississippi kite veriﬁed but not validated (Tirpak and others 2009a). 95 Northern Bobwhite Status The northern bobwhite (Colinus virginianus) is a resident gamebird found throughout the eastern United States and Great Plains. Populations have declined by 3 percent per year since 1966 (Sauer and others 2005). Declines in the CH and WGCP have been equally dramatic (3.1 and 4.4 percent per year, respectively) during this period (Table 5). As a resident gamebird, this species is not aﬀorded special status by the FWS (protection is relegated to state wildlife agencies). Nevertheless, PIF has designated this U.S. Forest Service bird as one requiring management attention in both the CH and WGCP (regional combined scores = 16 and 15, respectively) (Table 1). To address rangewide declines in populations, the Northern Bobwhite Conservation Initiative was established in 2002. Natural History The northern bobwhite is an economically important gamebird in the southern and central United States (Brennan 1999). It is associated with early successional vegetation, making use of agricultural ﬁelds, grasslands, grass-shrub rangelands, park-like pine forests and mixed pine-hardwood forests. At the county scale in Texas, the area in cultivated land and livestock density show curvilinear relationships to bobwhite population indices (Lusk and others 2002a). In Oklahoma, bobwhite indices decrease with the proportion of the landscape in mature woodland, but increase with the proportion of brushy prairie or early successional habitat (Guthery and others 2001). Guthery and others (2001) found that populations were highest in areas lacking cropland agriculture. However, Williams and others (2000) found that the bobwhite selected cropland when it accounted for a small proportion of the landscape. Patterns of use and survival diﬀer between crop-dominated and rangelanddominated areas during the hunting season in Kansas (Williams and others 2000). Bobwhite densities vary across the range depending on habitat quality but are highest in areas with small (0.5 to 5.0 ha) interspersed patches of habitat. Frequency and intensity of disturbance are important for this species, especially in southern pine forests where prescribed burning is a useful management tool. Cram and others (2002) reported higher bobwhite abundance in pine-grassland restoration areas in Arkansas as conifer and hardwood basal area decreased and woody structure less than 2 m tall increased. The bobwhite also occupies cottonwood reforestation plots less than 4 years old in Mississippi and Louisiana (Twedt and others 2002). Most management for this species has been at the local scale, but Guthery (1999) showed that optimal conﬁguration of patch types and sizes has variability (slack), and Williams and others (2004) promoted a regional management strategy that focused on useable space (i.e., more patches of native prairies, savanna, and other favored vegetation types). Weather aﬀects bobwhite populations, including positive eﬀects of summer temperature and fall precipitation (Lusk and others 2002a) and negative eﬀects of spring ﬂooding and 96 low winter temperatures (Applegate and others 2002). Bridges and others (2001) found a negative correlation between drought indices in dry regions and bobwhite abundance, but this pattern did not hold in wetter regions of Texas. Lusk and others (2002b) also found that climatic variables were more important than landscape variables for predicting bobwhite abundance in Oklahoma. Nests are constructed of litter (grass or pine needles) in areas of high structural complexity (Townsend and others 2001); brood cover is found in open areas with dense forbs that still permit mobility at ground level. Nevertheless, Taylor and others (1999) did not ﬁnd any habitat attributes associated with higher probabilities of adult survival or nest success. White and others (2005) examined multiple landscape buﬀers (radii of 250 to 1,000 m) around nest sites and random points to examine landscape eﬀects on nest site selection. Bobwhite responded to both composition and conﬁguration of landscapes, including proportions of open-canopy planted pine and fallow ﬁelds, interspersion-juxtaposition index, and patch density. A model containing all four of these variables applied at the largest landscape had the best predictive ability, but was closely followed by a model containing only proportion of open-canopy planted pine applied at the smallest landscape size. Several other types of habitat models have been developed for the bobwhite: HSI (Schroeder 1985), PATREC (Roseberry and Sudkamp 1998), and logistic regression (Burger and others 2004). Tests of these models showed that they perform poorly (Roseberry and Sudkamp 1998, Burger and others 2004, Jones-Farrand and Millspaugh 2006). Model Description Habitat quality for bobwhite is aﬀected by many parameters that are not measured easily at any scale: the proportion of forbs or open areas in grasslands, herbaceous vegetation height, grasslands and crop-ﬁeld management, and intra- and inter-annual climatic variations. Therefore, we restricted our habitat suitability model to aspects of landscape composition and forest structure that were quantiﬁable from available datasets. Our ﬁnal model includes seven variables: landform, landcover, successional age class, hardwood basal area, evergreen basal area, grass landcover, and interspersion of open and forest habitats. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 87). We directly assigned SI scores to these combinations on the basis of habitat associations for the northern bobwhite outlined in Hamel (1992). Forested sites used by the northern bobwhite typically are woodlands with low hardwood and pine basal area (SI2 and SI3, respectively). We used data from Cram and others (2002) and Palmer and Wellendorf (2006) to inform inverse logistic functions that predict SI scores for the bobwhite at various basal area levels (Tables 88-89; Figs 51-52). We directly assigned SI scores to grass landcover (SI4) classes based on their potential to provide feeding, nesting, and brood-rearing habitat (Guthery 1997) (Table 90). We assumed that natural grassland-herbaceous landcovers had the greatest potential to provide these habitats, though it is likely that a given patch can satisfy only two of the three requisites 97 Table 87.—Relationship of landform, landcover type, and successional age class to suitability index (SI) scores for northern bobwhite habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.167 0.167 0.083 0.000 0.000 Deciduous 0.167 0.167 0.083 0.000 0.000 Evergreen 1.000 1.000 0.667 0.500 0.667 Mixed 0.667 1.000 0.667 0.333 0.333 Orchard-vineyard 0.167 0.167 0.083 0.000 0.000 Woody wetlands 0.334 0.334 0.250 0.250 0.334 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 1.000 0.667 Deciduous 0.667 (1.000) 0.333 0.667 0.333 0.333 (0.500) 0.000 0.333 (0.667) 0.000 Evergreen 1.000 1.000 0.667 0.500 0.667 Mixed 0.667 1.000 0.667 0.333 0.333 Orchard-vineyard 0.333 0.667 0.333 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.667 (0.834) 0.333 1.000 (0.834) 1.000 0.667 0.333 (0.667) 0.000 0.333 (0.667) 0.000 1.000 (0.834) 0.667 1.000 (0.834) 1.000 0.667 0.667 0.500 (0.667) 0.333 0.333 Orchard-vineyard 0.333 1.000 0.500 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Deciduous Evergreen Mixed 0.500 0.667 at any point in time (Stoddard 1931). We assumed that areas in small grain production provided foraging opportunities but had little residual value for nesting or brood rearing. Similarly, fallow ﬁelds provide marginal nest and brood habitat but little forage. Finally, pasture-hay and row crops may provide foraging, nesting, and brood-rearing habitat but their value likely is limited due to management practices that produce unsuitable vegetative structure during most of the breeding season. The bobwhite relies on landscapes comprised of interspersed vegetation types (White and others 2005, Guthery 2000). We used the composition of open and forest landcovers within a 1-km landscape (SI5) to index the interspersion of these cover types. Guthery (1999, 2000) and others before him (see Schroeder 1985 and references therein) have noted that this species can tolerate a broad range of landscape conﬁgurations. On the basis of suggestions from Fred Guthery (2006, Oklahoma State University, pers. commun.), we assumed that high quality habitat was characterized by 10 to 40 percent forest land and 60 to 90 percent open habitat (Table 91). 98 Table 88.—Influence of hardwood basal area on suitability index (SI) scores for northern bobwhite habitat Suitability Index Score 1.0 Hardwood basal area (m2/ha) 0.8 0.6 0.4 0.0 1.000 2.6 b 1.000 3.3b 0.439 5.0 a 0.100 6.5 b 10.0 0.2 SI score a 0.000 a 0.000 a Assumed value. b Cram and others (2002). 0.0 0 5 10 15 Hardwood Basal Area (m^2/ha) Figure 51.—Relationship between hardwood basal area and suitability index (SI) scores for northern bobwhite habitat. Equation: SI score = 1/ (1.000 + (0.053 * (hardwood basal area)5.068)). Table 89.—Influence of pine basal area on suitability index (SI) scores for northern bobwhite habitat Suitability Index Score 1.0 Pine basal area (m2/ha) 0.8 0.6 0.4 0.2 SI score 0.00 a 1.000 9.20 b 1.000 12.30a 1.000 13.78 b 0.500 15.40 c 0.228 17.20 c 0.000 18.37b 0.000 a 0.0 Assumed value. Palmer and Wellendorf (2006). c Cram and others (2002). b 0 5 10 15 20 Pine Basal Area (m^2/ha) Figure 52.—Relationship between pine basal area and suitability index (SI) scores for northern bobwhite habitat. Equation: SI score = 1 - (0.984 / (1 + (83605490 * e -1.305 * pine basal area ))). We calculated the overall HSI score by ﬁrst determining the geometric mean of SI scores for forest structure attributes (SI1, SI2, and SI3). Open habitats lacking forest structure were assigned SI score independently (SI4). The landscape context of these forest and open habitats were incorporated into the HSI calculation by determining the geometric mean of these site-level and landscape-level variables (SI5) together. Overall HSI = (((SI1 * SI2 * SI3)0.333 + SI4) * SI5)0.500 99 Table 90.—Relationship between open and grassy landcover and suitability index (SI) scores for northern bobwhite habitat Landcover typea SI score Grassland-herbaceous 1.0 Pasture-hay 0.1 Row crops 0.1 Small grains 0.4 Fallow 0.2 a Assumed value. Table 91.—Suitability index scores for northern bobwhite habitat based on the proportion of cells providing: 1) good nesting, feeding, and brood-rearing habitat (open landcovers); 2) escape and thermal cover (forest landcovers) within 1-km radius Proportion foresta Proportion openb 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.1 0.00 0.00 0.10 0.15 0.25 0.25 0.25 0.20 0.15 0.10 0.2 0.00 0.10 0.15 0.25 0.35 0.35 0.30 0.25 0.20 0.3 0.00 0.30 0.35 0.45 0.45 0.45 0.40 0.30 0.4 0.00 0.50 0.50 0.50 0.50 0.50 0.50 0.5 0.00 0.70 0.70 0.70 0.70 0.70 0.6 0.00 0.90 0.90 0.90 0.90 0.7 0.00 0.90 1.00 1.00 0.8 0.00 0.90 1.00 0.9 0.00 0.90 1.0 0.00 a b Forest = landcovers with positive SI1 score (Table 87). Open = landcovers identified in SI4 (Table 90). Veriﬁcation and Validation The northern bobwhite was found in all 88 subsections of the CH and WGCP. Spearman rank correlation support a signiﬁcant (P = 0.006) positive association (rs = 0.29) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the northern bobwhite was signiﬁcant (P ≤ 0.001; R2 = 0.440); however, the coeﬃcient on the HSI predictor variable was negative (β = -37.119). Therefore, we considered the HSI model for the northern bobwhite veriﬁed but not validated (Tirpak and others 2009a). 100 Northern Parula Status The northern parula (Parula americana), a longdistance neotropical migrant, breeds in two disjunct zones of eastern North America: New Englandsouthern Canada and the southeastern United States. This species is notably absent from the southern Great Lakes. It depends on epiphytes—Spanish moss in the south and old man’s beard in the north—as a nesting Chandler S. Robbins, Patuxent Bird Identification InfoCenter substrate. Parula populations have been stable in most Photo used with permission regions during the last 40 years and have increased in some areas including the CH (Table 5). This species is not considered a Bird of Conservation Concern in the CH or WGCP (regional combined score = 12 and 13, respectively; Table 1). Natural History The northern parula is common in the bottomland hardwood and riverine forests of the Southeastern United States (Moldenhauer and Regelski 1996). It also occupies mixed pinehardwoods, though at lower densities (Moldenhauer and Regelski 1996). The northern parula has two competing habitat requirements: a preference for canopy gaps and large forest blocks. Moorman and Guynn (2001) found that this species is more abundant near canopy gaps than forest-interior sites with an unbroken canopy in bottomland hardwoods, and Annand and Thompson (1997) observed the highest northern parula densities in forests with canopy gaps resulting from single-tree selection. However, the probability of detecting the northern parula increases with riparian buﬀer width (Kilgo and others 1998) and forest patch size (Robbins and others 1989). The northern parula forages in the mid- to upper canopy layers (Moldenhauer and Regelski 1996), so it is not surprising that it prefers microsites with high basal area (Robbins and others 1989), high canopy cover, and tall canopies (James 1971), and avoids areas with dense understories (often associated with open canopies) (Torres and Leberg 1996). In the Southeast, this species nests almost exclusively in Spanish moss (Moldenhauer and Regelski 1996). However, no studies have identiﬁed Spanish moss as limiting. Model Description The HSI model for the northern parula includes six variables: landform, landcover, successional age class, forest patch size, percent forest in a 1-km radius, and canopy cover. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 92). We directly assigned SI scores to these combinations on the basis of habitat associations of northern parulas reported by Hamel (1992) for the Southeast. We derived a logarithmic function (Fig. 53) from data on the occupancy rate of northern parulas in forest blocks of varying size (SI2; Hayden and others 1985, Robbins and others 1989) (Table 93) to predict habitat suitability from patch area. However, small forest 101 Table 92.—Relationship of landform, landcover type, and successional age class to suitability index scores for northern parula habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.083 0.500 0.834 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.167 0.500 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Woody wetlands 0.000 0.000 0.250 0.750 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.250 0.500 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.167 0.500 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.667 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.167 0.333 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.167 0.500 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.667 1.000 patches in predominantly forested landscapes may provide habitat due to their proximity to large forest blocks (Rosenberg and others 1999). To capture this relationship, we ﬁt a logistic function (Fig. 54) to data (Table 94) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoints between these classes (30 and 70 percent forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. We used the maximum SI score from SI2 or SI3 to account for small patches in predominantly forested landscapes. We included canopy cover (SI4) in our model to capture the preference of the northern parula for interior edges. James (1971), Collins and others (1982), and Morgan and Freedman (1986) found that the northern parula is associated with increased canopy cover. Nonetheless, there seems to be a threshold above which suitability declines. Robbins and others (1989) observed an inverse relationship between canopy cover and northern parula abundance, and Annand and Thompson (1997) observed a threefold increase of parulas in single-tree selection stands characterized by a heterogeneous canopy than in mature forest habitats with closed canopies. On the basis of these studies, we assumed that 102 Table 93.—Influence of forest patch size on suitability index (SI) scores for northern parula habitat Suitability Index Score 1.0 Forest patch size (ha) 23.6 0.8 520 0.0 b 3,200 0.6 SI score a 0.5 b 1.0 a Hayden and others (1985). b Robbins and others (1989). 0.4 0.2 0.0 0 1000 2000 3000 Forest Patch Size (ha) Suitability Index Score Figure 53.—Relationship between forest patch size and suitability index (SI) scores for northern parula habitat. Equation: SI score = 0.199 * ln(forest patch size) – 0.661. 1.0 Table 94.—Relationship between local landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for northern parula habitat 0.8 Landscape composition 0 0.6 a 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 54.—Relationship between local landscape composition and suitability index (SI) scores for northern parula habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * (local landscape composition) )). 0.00 a 0.00 20a 0.05 30 b 0.10 40 a 0.25 50 b 0.50 60a 0.75 70 b 0.90 80 a 0.95 90 b 1.00 10 0.4 SI score 100a 1.00 a Assumed value. Donovan and others (1997). b habitat suitability was optimal at 90 percent canopy cover and decreased as the canopy became increasingly open or closed. We ﬁt an inverse quadratic function (Fig. 55) to data demonstrating this relationship (Table 95). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure attributes (SI1 and SI4) and then calculated the geometric mean of this value and landscape composition (Max of SI2 or SI3). Overall HSI = ((SI1 * SI4)0.500 * Max(SI2 or SI3))0.500 103 Table 95.—Influence of canopy cover on suitability index (SI) scores for northern parula habitat Suitability Index Score 1.0 Canopy cover (percent)a 0.8 0.6 0.4 0.2 60 0.2 70 0.4 80 0.8 85 0.9 90 1.0 95 0.9 100 0.8 a 0.0 0 25 50 75 Assumed value. 100 Canopy Cover (%) Figure 55.—Relationship between canopy cover and suitability index (SI) scores for northern parula habitat. Equation: SI score = 1 / (37.3645 – (0.8127 * canopy cover) + (0.00454 * (canopy cover2))). Veriﬁcation and Validation The northern parula was found in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.51) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the northern parula was signiﬁcant (P ≤ 0.001; R2 = 0.276), and the coeﬃcient on the HSI predictor variable was both positive (β = 5.250) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the northern parula both veriﬁed and validated (Tirpak and others 2009a). 104 SI score Orchard Oriole Status The orchard oriole (Icterus spurius), a neotropical migrant, is found throughout most of the United States east of the Rocky Mountains except for New England and the northern Great Lakes. Although this species has experienced increases along the edges of its distribution, populations have declined in the core of its range where Deanna K. Dawson, Patuxent Bird Identification InfoCenter densities are highest. In the WGCP, populations have Photo used with permission declined by 3 percent per year since 1967 (Table 5). Populations in the adjacent Mississippi Alluvial Valley have declined 4 percent. The orchard oriole is a Bird of Conservation Concern in the WGCP and has been identiﬁed as a species requiring management attention in both the CH and WGCP (regional combined score = 17 and 18, respectively; Table 1). Natural History The orchard oriole breeds in wooded riparian zones, ﬂoodplains, marshes, and shorelines (Scharf and Kren 1996) but also in open shrublands and low-density human-dominated areas (e.g., farms and parklands). It is semi-colonial in optimal habitat but relatively solitary in marginal areas. This species is a common host of the brown-headed cowbird. Model Description The HSI model for the orchard oriole includes ﬁve variables: landform, landcover, successional age class, forest within a 1-km radius, and basal area. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 96). We directly assigned SI scores to these combinations based on vegetation and successional age class associations in Hamel (1992). However, we adjusted Hamel’s values to account for the preference of the orchard oriole for mesic habitats (e.g., riparian zones, ﬂoodplains, and marshes; Scharf and Kren 1996). The orchard oriole is not area sensitive but generally is restricted to forested landscapes. Therefore, we included only local forest composition (SI2) in our model to discount forest patches that were isolated within a matrix of nonforest landcover. Conversely, this is an edge species whose abundance declines in heavily forested regions (Scharf and Kren 1996). Therefore, we assumed that landscapes with 70 to 80 percent forest provided optimal habitat suitability and reduced suitability symmetrically as landscape composition shifted from these optima (Table 97, Fig. 56). This species is most abundant in areas with scattered trees. Heltzel and Leberg (2006) observed signiﬁcantly fewer orioles in stands with an average basal area of 25 m2 per ha than in recently harvested stands with an average basal area of 18 m2 per ha. We assumed that habitat suitability was optimal for the orchard oriole at lower basal areas and modeled 105 Table 96.—Relationship of landform, landcover type, and successional age class to suitability index scores for orchard oriole habitat Landform Landcover type Floodplain-valley Low-density residential 0.000 0.000 0.500 1.000 1.000 Transitional-shrubland 0.000 0.000 0.500 1.000 1.000 Deciduous 0.000 0.000 0.500 1.000 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.500 1.000 1.000 Terrace-mesic Xeric-ridge Grass-forb Successional age class Shrubseedling Sapling Pole Woody wetlands 0.000 0.000 0.500 1.000 1.000 Low-density residential 0.000 0.000 0.333 0.667 0.667 Transitional-shrubland 0.000 0.000 0.250 0.500 0.500 Deciduous 0.000 0.000 0.250 0.500 0.500 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.333 0.667 0.667 Woody wetlands 0.000 0.000 0.500 1.000 1.000 Low-density residential 0.000 0.000 0.333 0.667 0.667 Transitional-shrubland 0.000 0.000 0.250 0.500 0.500 Deciduous 0.000 0.000 0.250 0.500 0.500 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.333 0.667 0.667 Woody wetlands 0.000 0.000 0.500 1.000 1.000 the basal area (SI3)-habitat suitability relationship as a quadratic function (Fig. 57) that maximized SI scores at intermediate basal area values (12.5 m2/ha; Table 98). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure indices (SI1 and SI3) and then determined the geometric mean of this value and landscape composition (SI2). Overall HSI = ((SI1 * SI3)0.500 * SI2)0.500 Veriﬁcation and Validation The orchard oriole was found in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.34) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the orchard oriole was signiﬁcant (P = 0.088; R2 = 0.056), and the coeﬃcient on the HSI predictor variable was positive (β = 2.442) but not signiﬁcantly diﬀerent from zero (P = 0.221). Therefore, we considered the HSI model for the orchard oriole veriﬁed but not validated (Tirpak and others 2009a). 106 Saw Table 97.—Relationship between landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for orchard oriole habitat Suitability Index Score 1.0 Landscape compositiona 0.8 0.6 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 56.—Relationship between landscape composition and suitability index (SI) scores for orchard oriole habitat. Equation: SI score = 1.011 * e ((0 – ((landscape composition * 100) – 74.945) ^ 2) / 863.949). SI score 0 0.00 10 0.00 20 0.05 30 0.10 40 0.25 50 0.50 60 0.75 70 1.00 80 1.00 90 0.75 100 0.50 a Assumed value. Table 98.—Influence of basal area (m2/ha) on suitability index (SI) scores for orchard oriole habitat Suitability Index Score 1.0 Basal area (m2/ha) 0.0 0.8 a 1.0 25.0b 0.0 a b 0.4 0.0 a 12.5 0.6 SI score Assumed value. Heltzel and Leberg (2006). 0.2 0.0 0 10 20 30 Basal Area (m^2/ha) Figure 57.—Relationship between basal area and suitability index (SI) scores for orchard oriole habitat. Equation: SI score = (0.16 * basal area) - (0.00639 * (basal area2)). 107 Painted Bunting Status The painted bunting (Passerina cyanea) occurs as two allopatric populations that may represent separate species (Lowther and others 1999). The western population inhabits the southern Great Plains and the western edges of the CH and WGCP, while the eastern population inhabits the Atlantic Coastal Plain from North Carolina to Florida. Populations have been relatively stable across Deanna K. Dawson, the WGCP as a whole (Table 5), but populations have Patuxent Bird Identification InfoCenter Photo used with permission declined in Arkansas (5.8 percent per year from 1967 to 2004), Louisiana (3.5 percent), and Texas (2.4 percent) but increased in Oklahoma (1.3 percent; Sauer and others 2005). The painted bunting is not an FWS Bird of Conservation Concern but is a PIF management attention priority in both the CH and WGCP (regional combined score = 16 and 17, respectively; Table 1). Natural History The habitat requirements of the painted bunting are poorly understood. This species generally occupies areas of scattered woody vegetation. Kopachena and Crist (2000a) characterized painted bunting habitat in northeast Texas as “wooded areas in otherwise open habitat” as opposed to the indigo bunting, which occurs in “open areas in otherwise wooded habitat.” The painted bunting use smaller, more heterogeneous groups of trees than the indigo bunting, but microhabitats diﬀer little between these species (Kopachena and Crist 2000b). The painted bunting occupies narrow riparian strips in eastern Texas and its abundance decreases quickly as widths exceed 70 m (Conner and others 2004). The painted bunting nests in low, woody vegetation (Lowther and others 1999) and its territory size varies with its population density. In Missouri, territories ranged from 0.64 to 6.66 ha and included 80 percent pasture and 20 percent woodland. This species is a common host of both the brown-headed and bronzed cowbird. Model Description The HSI model for the painted bunting includes six variables: landform, landcover, successional age class, distance to edge, interspersion of open and forested lands, and small stem (< 2.5 cm d.b.h.) density. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 99). We directly assigned SI scores to these combinations on the basis of relative habitat rankings for vegetation and successional age class associations of painted buntings reported by Hamel (1992). We assigned higher values to the shrub-seedling age class than Hamel (1992) on the basis of qualitative descriptions in Lowther and others (1999). An early-successional species, the painted bunting is associated with edges. We used data on territory density from Lanyon and Thompson (1986; Table 100) to deﬁne an inverse logistic function linking SI scores to distance from an edge (SI2; Fig. 58). 108 Table 99.—Relationship of landform, landcover type, and successional age class to suitability index scores for painted bunting habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.500 0.500 0.250 0.000 Deciduous 0.000 0.500 0.500 0.250 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.500 0.500 0.250 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Woody wetlands 0.000 1.000 0.750 0.500 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.500 0.500 0.250 0.000 Deciduous 0.000 0.500 0.500 0.250 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.500 0.500 0.250 0.000 Woody wetlands 0.000 1.000 0.750 0.500 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.500 0.500 0.250 0.000 Deciduous 0.000 0.500 0.500 0.250 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.500 0.500 0.250 0.000 Woody wetlands 0.000 1.000 0.750 0.500 0.000 Table 100.—Influence of distance to edge on suitability index (SI) scores for painted bunting habitat Suitability Index Score 1.0 Distance to edge (m) 0.8 0 a 90 0.6 0.4 1.0 a 0.7 150a 0.3 210 a 0.0 270 b 0.0 a 0.2 SI score b Lanyon and Thompson (1986). Assumed value. 0.0 0 10 20 30 Distance to Edge (m/10) Figure 58.—Relationship between distance to edge and suitability index (SI) scores for painted bunting habitat. Equation: SI score = 1 - (1.034 / (1 + (39.685 * e -0.301 * (distance to edge / 10 m)))). 109 Table 101.—Suitability index scores for painted bunting habitat based on the proportion of open and forest landcovers within 5-ha area Proportion opena Proportion forestb 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.2 0.0 0.0 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.3 0.0 0.0 0.5 0.7 0.7 0.7 0.7 0.7 0.4 0.0 0.0 0.5 0.7 0.9 0.9 0.9 0.5 0.0 0.0 0.5 0.7 0.9 1.0c 0.6 0.0 0.0 0.5 0.7 0.9 0.7 0.0 0.0 0.5 0.7 0.8 0.0 0.0 0.5 0.9 0.0 0.0 1.0 0.0 a Open = herbaceous natural, cultivated, and emergent herbaceous wetland. b Forest = upland forested, transitional, woody wetland, and orchard/vineyard. Unpublished data. c The presence of both forest and open landcovers in the landscape (SI3) is perhaps the most important component of painted bunting habitat. We maximized SI scores for this species in landscapes containing 50 percent forest and 50 percent open habitats based on unpublished data (Jeﬀrey Kopachena, 2006, Texas A&M University—Commerce, pers. commun.). Norris and Elder (1982, cited in Lowther and others 1999) observed the painted bunting in landscapes with forest cover of 20 to 80 percent forest. We used these values as cutoﬀs for forest cover in our interspersion function for the painted bunting (Table 101). As an early successional species, the painted bunting occupies habitats containing high densities of small stems (SI4). We assumed that the mean stem density values (6,400 stems/ha) reported by Kopachena and Crist (2000b) were characteristic of average habitat suitability (SI score = 0.500). However, because of the high standard error (6,300 stems/ha) associated with this estimate, we assumed that a stem density that was twice the mean was necessary to ensure optimal habitat (Table 102). We ﬁt a smoothed logistic function through these data points (Fig. 59) to quantify the relationship between small stem density and SI scores for painted bunting habitat. To calculate the HSI score for sapling and pole successional age class stands, we determined the geometric mean of SI scores for forest structure (SI1 and SI4) and landscape composition (SI2 and SI3) separately and then the geometric mean of these means together. HSISap-pole = ((SI1 * SI4)0.500 * (SI2 * SI3)0.500)0.500 110 Table 102.—Influence of small stem density (stems * 1,000/ha) on suitability index (SI) scores for painted bunting habitat Suitability Index Score 1.0 Small stem density 0.8 SI score 0.0a 6.4 0.6 0.4 0.5 12.8 a 1.0 25.0 a 1.0 a b 0.2 0.0 b Assumed value. Kopachena and Crist (2000b). 0.0 0 10 20 30 Small Stem Density (stems * 1,000/ha) Figure 59.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for painted bunting habitat. Equation: SI score = (1.000 / (1 + (1178.674 * e -1.105 * (small stem density / 1000)))). We assumed that shrub-seedling successional age class stands were suitable regardless of edge or landscape composition. Thus, we calculated the HSI score as the geometric mean of forest structure attributes alone (SI1 and SI4). HSIShrub = (SI1 * SI4)0.500 The overall HSI score is the sum of the two age class speciﬁc SIs: Overall HSI = SISap-pole + SIShrub Veriﬁcation and Validation The painted bunting was found in only 38 of the 88 subsections within the CH and WGCP. Nevertheless, Spearman rank correlations based on either all subsections or only subsections in which the painted bunting occurred produced similar results: signiﬁcant (P ≤ 0.001 in both analyses) positive associations (rs = 0.56 and 0.58, respectively) between average HSI score and mean BBS route abundance at the subsection scale. The generalized linear model predicting BBS abundance from BCR and HSI for the painted bunting was signiﬁcant (P ≤ 0.001; R2 = 0.480), and the coeﬃcient on the HSI predictor variable was both positive (β = 70.737) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the painted bunting both veriﬁed and validated (Tirpak and others 2009a). 111 Pileated Woodpecker Status The pileated woodpecker (Dryocopus pileatus) breeds throughout eastern North America, southern Canada, and the montane forests of the West. Populations have been stable across most of its range, including the WGCP, over the last 40 years and have increased along the northern limit of this bird’s distribution. In the CH, populations have increased by 1.8 percent per year since 1967 (Sauer and others 2005) (Table 5). This species is a management attention priority in the WGCP (regional combined score = 16) but has no special conservation status in the CH (regional combined score = 13; Table 1). U.S Forest Service Natural History The pileated woodpecker uses a variety of forest types across its range but typically is associated with older successional age classes (Bull and Jackson 1995, Annand and Thompson 1997). The key component to pileated woodpecker habitat is an abundance of large snags—the more the better. Diﬀerent researchers deﬁne “large” diﬀerently (Renken and Wiggers 1989, Savignac and others 2000, Showalter and Whitmore 2002) but the pileated woodpecker is invariably associated with the largest available size class. In Missouri, this species is associated with bottomland hardwood forest (Renken and Wiggers 1993); in east Texas, the pileated woodpecker is equally abundant in bottomland hardwoods, longleaf pine savanna, and mixed pine-hardwood stands, so long as suitable snags are available (Shackelford and Conner 1997). Closed canopies (canopy cover of 75 to 96 percent) are the norm (Renken and Wiggers 1989). Because it has a large home range (53 to 160 ha), it is not surprising that the pileated woodpecker is sensitive to forest area. Robbins and others (1989) did not detect this species in woodlots less than 42 ha and larger areas likely are required for breeding pairs. Schroeder (1982) considered 130 ha as the minimum forest patch size for this species. Model Description The pileated woodpecker model includes six variables: landform, land cover, successional age class, large snag (> 30 cm d.b.h.) density, forest patch size, and percentage of forest in a 1-km radius. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 103). We used the habitat associations of the pileated woodpecker outlined in Hamel (1992) to assign SI scores to these combinations. Large snags (SI2) are used for roosting, nesting, and foraging and are an important component of pileated woodpecker habitat. We ﬁt a logistic function (Fig. 60) to data from Renken and Wiggers (1989) on the relative density of this species on sites with varying large snag densities to predict SI scores based on this habitat feature (Table 104). 112 Table 103.—Relationship of landform, landcover type, and successional age class to suitability index scores for pileated woodpecker habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.042 0.083 0.167 Transitional-shrubland 0.000 0.000 0.083 0.583 1.000 Deciduous 0.000 0.000 0.083 0.583 1.000 Evergreen 0.000 0.000 0.167 0.333 0.333 Mixed 0.000 0.000 0.167 0.500 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.667 1.000 Low-density residential 0.000 0.000 0.042 0.083 0.167 Transitional-shrubland 0.000 0.000 0.167 Deciduous 0.000 0.000 0.000 0.500 (0.333) 0.500 0.667 (0.333) 1.000 Evergreen 0.000 0.000 0.167 0.333 0.333 Mixed 0.000 0.000 0.167 0.500 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.667 1.000 Low-density residential 0.000 0.000 0.042 0.083 0.167 Transitional-shrubland 0.000 0.000 Deciduous 0.000 0.000 0.167 (0.083) 0.000 0.500 (0.167) 0.500 0.667 (0.167) 1.000 Evergreen 0.000 0.000 Mixed 0.000 0.000 0.167 (0.083) 0.167 0.333 (0.167) 0.500 0.333 (0.167) 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.667 1.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Table 104.—Influence of large snag (> 30 cm d.b.h.) density (snags/ha) on suitability index (SI) scores for pileated woodpecker habitat Suitability Index Score 1.0 Large snag density 0.8 0. 0.6 0.4 0.2 0.0 0 10 Large Snag Density (snags/ha) 20 a SI score 0.0 2.5 a 0.0 6.1 b 0.1 7.6 b 0.5 10.0b 1.0 15.0 a 1.0 12.5 a 1.0 a Assumed value. b Renken and Wiggers (1989). Figure 60.—Relationship between large snag (> 30 cm d.b.h.) density and suitability index (SI) scores for pileated woodpecker habitat. Equation: SI score = (1.0054 / (1 + (747.0936 * e -0.8801 * large snag density))). 113 Table 105.—Influence of forest patch size on suitability index (SI) scores for pileated woodpecker habitat Suitability Index Score 1.0 Forest patch size (ha)a 0.8 0.6 42.2 0.0 165 0.5 3,200 1.0 a 0.4 SI score Robbins and others (1989). 0.2 0.0 0 1000 2000 3000 Forest Patch Size (ha) Figure 61.—Relationship between forest patch size and suitability index (SI) scores for pileated woodpecker habitat. Equation: SI score = 0.230 * ln(forest patch size) – 0.877. Table 106.—Relationship between landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for pileated woodpecker habitat Suitability Index Score 1.0 0.8 Landscape composition 0a 0.6 10a 0.00 a 0.05 30b 0.10 40 a 0.25 50 b 0.50 60 a 0.75 70b 0.90 80 a 0.95 90 b 20 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 62.—Relationship between landscape composition and suitability index (SI) scores for pileated woodpecker habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * (local landscape composition) )). 100 a b 1.00 a Assumed value. Donovan and others (1997). We incorporated forest patch size (SI3) and percent forest in the local landscape (SI4) as predictors of habitat suitability. Large home ranges for the pileated woodpecker necessitate large forest patches. We ﬁt a logarithmic function (Fig. 61) to data from Robbins and others (1989) on the eﬀect of forest patch size on occupancy rates (Table 105). We also included percent forest in the landscape because small forest patches that may not be used in predominantly nonforested landscapes may provide habitat in predominantly forested landscapes due to their proximity to large forest blocks (Rosenberg and others 1999). To capture this relationship, we ﬁt a logistic function (Fig. 62) to data (Table 106) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite 114 SI score 0.00 1.00 communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoints between these classes (30 and 70 percent forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. We used the maximum SI score from SI3 or SI4 to account for the higher suitability of small forest patches in predominantly forested landscapes. To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure attributes (SI1 and SI2) and multiplied that by the maximum value of forest patch size (SI3) or percent forest in the 1-km radius landscape (SI4) and calculated the geometric mean of that product. Overall HSI = ((SI1 * SI2)0.500 * Max(SI3 or SI4))0.500 Veriﬁcation and Validation The pileated woodpecker was observed in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.002) positive association (rs = 0.33) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the pileated woodpecker was signiﬁcant (P ≤ 0.001; R2 = 0.313), and the coeﬃcient on the HSI predictor variable was both positive (β = 8.852) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the pileated woodpecker both veriﬁed and validated (Tirpak and others 2009a). 115 Prairie Warbler Status The prairie warbler (Dendroica discolor), a neotropical migrant, occupies early successional habitats throughout the eastern United States. Like many early successional species, populations of this bird have declined throughout the eastern and central United States since 1967, including a drop of 2.6 percent per year in the CH and 4.4 percent per year in the WGCP (Table 5). The prairie warbler is an FWS Bird of Conservation Concern and a management attention priority in both BCRs (regional combined score = 18 in the CH and WGCP; Table 1). Deanna K. Dawson, Patuxent Bird Identification InfoCenter Photo used with permission Natural History The prairie warbler breeds in shrubby vegetation under an open canopy (Nolan and others 1999). Typical associations in the CH and WGCP include shrubby southern pine forest, pine barrens, scrub oak barrens, abandoned ﬁelds and pastures, regenerating forest, abandoned orchards, grassland-forest edge, Christmas tree farms, and reclaimed strip mine spoils. The prairie warbler uses a variety of landforms from xeric uplands in Arkansas to palustrine swamps in Virginia. In comparison to other early successional warblers, this bird occupies sites with fewer dense shrubs than the blue-winged warbler, more dense vegetation and drier areas than the yellow warbler, and less dense vegetation and higher vegetation strata than the common yellowthroat or yellow-breasted chat (Nolan and others 1999). The prairie warbler nests in shrubs and small trees that are more than 20 m from a ﬁeldforest edge (Nolan and others 1999, Woodward and others 2001). However, in eastern Texas this species typically occurs in narrow riparian zones, with abundance decreasing quickly as widths increase (Conner and others 2004). Mean territory size varies inversely with population density, ranging from 0.2 to 3.5 ha in Indiana (Nolan and others 1999). Territory size also varies with shape of forest patch; it is larger in more linear patches. Although males do not limit movements to their defended territory, a female’s home range usually is contained within a male’s defended territory. This species is a cowbird host. Although parasitism has little eﬀect on hatching success, it can signiﬁcantly reduce ﬂedging rates. Model Description Our HSI model for the prairie warbler includes seven variables: landform, landcover, successional age class, early-successional patch size, small stem (< 2.5 cm d.b.h.) density, edge occurrence, and canopy cover. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 107). We directly assigned SI scores to these combinations on the basis of habitat associations for the prairie warbler documented in Hamel (1992). 116 Table 107.—Relationship of landform, landcover type, and successional age class to suitability index scores for prairie warbler habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.333 0.167 0.000 0.000 Deciduous 0.000 0.333 0.167 0.000 0.000 Evergreen 0.000 0.667 0.334 0.000 0.000 Mixed 0.000 1.000 0.500 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.333 0.167 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.500 (0.334) 0.333 0.000 Deciduous 1.000 (0.667) 0.667 0.000 0.000 Evergreen 0.000 0.667 0.334 0.000 0.000 Mixed 0.000 1.000 0.500 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.333 0.167 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.500 (0.250) 0.500 0.000 Deciduous 1.000 (0.500) 1.000 0.000 0.000 Evergreen 0.000 0.000 0.000 0.334 (0.250) 0.500 0.000 Mixed 0.667 (0.500) 1.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.333 0.167 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Both Woodward and others (2001) and Rodewald and Vitz (2005) observed edge avoidance by this species. Thus, we used a 3 × 3 pixel (90 x 90 m) window to identify early successional habitats (i.e., grass-forb, shrub-seedling, or sapling successional age class forest) adjacent to mature forest stands (i.e., pole or sawtimber successional age class) and reduced the suitability of locations adjacent to edges by half (SI2; Table 108). We also included early successional patch size (SI3) as an explanatory variable because the prairie warbler is absent from small clearings and edge habitats. We used data from Larson and others (2003) (Table 109) to ﬁt a logistic function (Fig. 63) that characterized the relationship between habitat suitability and early successional patch size. We also included small stem density (SI4) as a variable because the prairie warbler is associated with dense understory vegetation. We used point count and habitat data reported by Annand and Thompson (1997) (Table 110) to derive a logistic function (Fig. 64) that predicted habitat suitability for the prairie warbler from small stem density. 117 Table 108.—Influence of edge on suitability index (SI) scores for prairie warbler habitat Suitability Index Score 1.0 3 × 3 pixel window around early successional habitat includes mature foresta 0.8 0.6 SI score Yes 0.5 No 1.0 a Early successional = grass-forb, shrub-seedling, and sapling successional age classes; mature forest = pole or sawtimber successional age classes. 0.4 0.2 0.0 0 2 4 6 Early Successional Patch Size (ha) Figure 63.—Relationship between early successional patch size and suitability index (SI) scores for prairie warbler habitat. Equation: SI score = (1.002 / (1 + (1207.332 * e -3.757 * forest patch size))). Table 109.—Influence of early successional patch size on suitability index (SI) scores for prairie warbler habitat; early successional patches only include grass-forb, shrub-seedling, and sapling successional age classes Early successional patch size (ha)a Suitability Index Score 1.0 0.8 0.18 0.0 0.36 0.0 1.89 0.5 3.42 1.0 5.00 1.0 a 0.6 Larson and others (2003). 0.4 Table 110.—Influence of small stem (< 2.5 cm d.b.h.) density (stems * 1,000/ha) on suitability index (SI) scores for prairie warbler habitat 0.2 0.0 0.0 2.5 5.0 7.5 10.0 Small Stem Density (stems * 1,000/ha) Figure 64.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for prairie warbler habitat. Equation: SI score= (1.000 / (1 + (99.749 * e -1.001 * (small stem density / 1000)))). Small stem density SI score 0.0a 0.00 3.8 b 0.31 8.1 b 1.00 a b Assumed value. Annand and Thompson (1997). Finally, we used data from Sheﬃeld (1981) to inform an inverse logistic function (Fig. 65) that discounted SI scores at increasingly high canopy closures (SI5; Table 111). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure attributes (SI1, SI4, and SI5) and landscape composition (SI2 and SI3) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI4 * SI5)0.333 * (SI2 * SI3)0.500)0.500 118 SI score Table 111.—Influence of canopy cover on suitability index (SI) scores for prairie warbler habitat 1.0 Suitability Index Score Canopy cover (percent)a 0.8 0.6 0.4 0 1.0 25 1.0 50 0.5 75 0.0 100 0.0 a 0.2 SI score Sheffield (1981). 0.0 0 25 50 75 100 Canopy Cover (%) Figure 65.—Relationship between canopy cover and suitability index (SI) scores for prairie warbler habitat. Equation: SI score = 1 - (1.003 / (1 + (26950.420 * e -0.204 * canopy cover))). Veriﬁcation and Validation The prairie warbler was found in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.41) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the prairie warbler was signiﬁcant (P = 0.005; R2 = 0.117), and the coeﬃcient on the HSI predictor variable was both positive (β = 15.317) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the prairie warbler both veriﬁed and validated (Tirpak and others 2009a). 119 Prothonotary Warbler Status The prothonotary warbler (Protonotaria citrea) is a longdistance neotropical migrant associated with bottomland hardwood and ﬂoodplain forests of the Southeast. Densities are highest in the Mississippi Alluvial Valley; this species is notably absent from the central and southern Appalachians. Populations in the CH have remained relatively stable while those in the WGCP, where the prothonotary warbler is a Bird of Conservation Concern (Table 1), have declined by 5.8 percent per John and Karen Hollingsworth, year since 1967 (Table 5). This bird is a planning and U.S. Fish & Wildlife Service responsibility species in the CH (regional combined score = 14) and a management attention species in the WGCP (regional combined score = 17). Natural History Because it nests in cavities and readily accepts nest boxes, the prothonotary warbler has been well-studied. Petit (1999) provided an excellent, detailed description of this bird’s habitat requirements: Key (and nearly universal) features are presence of water near wooded area with suitable cavity nest sites. Nest usually placed over or near large bodies of standing or slow-moving water, including seasonally ﬂooded bottomland hardwood forest, baldcypress swamps, and large rivers or lakes (Walkinshaw 1953, Blem and Blem 1991). Many other forms of water also chosen, such as creeks, streams, backyard ponds, and even swimming pools. Nests located away from permanent water are usually in low-lying, temporarily ﬂooded spots (Walkinshaw 1953). Other important habitat correlates include low elevation, ﬂat terrain, shaded forest habitats with sparse understory, and in some places, presence of baldcypress (Kahl and others 1985, Robbins and others 1989). Common overstory trees in nesting habitat include willows, maples, sweet gum, willow oak, ashes, elms, river birch, black gum, tupelo, cypress, and other species associated with wetlands. Buttonbush is the most common subcanopy species. Canopy height 12-40 m (usually 16-20), canopy cover usually 50-75 percent; ground vegetation usually very sparse and of low stature (< 0.5 m; Kahl and others 1985). Exhibits area sensitivity, avoiding forests <100 ha in area and avoiding waterways with wooded borders <30 m wide (Kahl and others 1985). Model Description The HSI model for prothonotary warbler includes seven variables: landform, landcover, successional age class, water, forest patch size, percentage of forest in the local (1-km radius) landscape, and snag density. 120 Table 112.—Relationship of landform, landcover type, and successional age class to suitability index scores for prothonotary warbler habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.100 0.300 0.400 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.100 0.300 0.400 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Woody wetlands 0.000 0.000 0.300 0.800 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.200 0.600 0.800 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.200 0.600 0.800 The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 112). We directly assigned SI scores to these combinations on the basis of relative rankings of habitat associations reported by Hamel (1992) for the prothonotary warbler in the Southeast. This species is rarely found more than 200 m from water during the breeding season, so we used a 9 × 9 pixel window (270 x 270 m) to examine whether water was close enough to each site to make it suitable (SI2). If water was present in any of the 81 pixels comprising the window, we assigned the center pixel a value of 1.000. If water was absent, we assigned the center pixel a value of zero (Table 113). We also included forest patch size (SI3) as a variable in the HSI model because prothonotary warbler abundance is lower in small isolated fragments and thin riparian buﬀer strips (Table 114; Fig. 66). However, this species occupies small forest fragments within heavily forested landscapes so we included the percentage of forest in the local landscape as a variable (SI4). To capture this relationship, we ﬁt a logistic function (Fig. 67) to data (Table 115) derived 121 Table 113.—Influence of occurrence of water on suitability index (SI) scores for prothonotary warbler habitat Suitability Index Score 1.0 9 × 9 pixel window contains water 0.8 0.6 No 0.0 Table 114.—Influence of forest patch size on suitability index (SI) scores for prothonotary warbler habitat 0.2 Forest patch area (ha)a 0 100 200 300 Forest Patch Size (ha) Figure 66.—Relationship between forest patch size and suitability index (SI) scores for prothonotary warbler habitat. Equation: SI score = 1.002 – 1.001 * e -0.031 * (forest patch size ^ 0.968). SI score 0 0.00 50 0.75 200 1.00 500 1.00 a Assumed value. Table 115.—Relationship between local landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for prothonotary warbler habitat 1.0 Suitability Index Score 1.0 0.4 0.0 Landscape composition 0.8 0 a 0.4 0.2 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 67.—Relationship between landscape composition and suitability index (SI) scores for prothonotary warbler habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * (landscape composition) )). SI score 0.00 a 0.00 20a 0.05 30 b 0.10 40 a 0.25 50 b 0.50 60a 0.75 70 b 0.90 80 a 0.95 90 b 1.00 10 0.6 0.0 100a a b Assumed value. Donovan and others (1997). from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoints between these classes (30 and 70 percent forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. We applied the maximum value of SI3 or SI4 to all sites to compensate for the higher suitability of small forest blocks in predominantly forested landscapes. 122 SI score Yes 1.00 Table 116.—Influence of snag density on suitability index (SI) scores for prothonotary warbler habitat Suitability Index Score 1.0 Snag density (snags/ha) 0.8 0 a 5 b 20 0.6 SI score 0.25 1.00 a 1.00 a Assumed value. b McComb and others (1986). 0.4 0.2 0.0 0.0 2.5 5.0 7.5 10.0 Snag Density (snags/ha) Figure 68.—Relationship between snag density and suitability index (SI) scores for prothonotary warbler habitat. Equation: SI score = 1.000 / (1 + (3.113 * e -3.689 * snag density)). The prothonotary warbler is a cavity nester and uses snags (SI5) for nesting. McComb and others (1986) recommended 212 snags per 40 ha to satisfy the requirements of the primary cavity-nesting bird guild. We assumed that ﬁve snags per ha (Table 116) was suﬃcient for this bird (a secondary cavity-nesting species), but we recognized that this species also uses both cavities in live trees and crevices as nest sites. Therefore, we assigned a residual SI score (0.25) to sites lacking snags. We ﬁt a logistic function through these points to quantify the snag density-habitat suitability relationship (Fig. 68). To calculate the overall HSI, we calculated the geometric mean of the two SIs related to forest structure (SI1 and SI5) and the product of the maximum of the two SIs related to landscape composition (SI3 or SI4) and SI2 separately and then the geometric mean of these values together. Overall HSI = ((SI1 * SI5)0.500 * (Max(SI3 or SI4) * SI2))0.500 Veriﬁcation and Validation The prothonotary warbler was found in 83 of the 88 subsections within the CH and WGCP. Spearman rank correlations identiﬁed signiﬁcant positive associations between average HSI score and mean BBS route abundance across all subsections (P ≤ 0.001; rs = 0.39) and subsections within which the prothonotary warbler were detected (P ≤ 0.001; rs = 0.41). The generalized linear model predicting BBS abundance from BCR and HSI for the prothonotary warbler was signiﬁcant (P ≤ 0.001; R2 = 0.249), and the coeﬃcient on the HSI predictor variable was both positive (β = 2.271) and signiﬁcantly diﬀerent from zero (P = 0.002). Therefore, we considered the HSI model for the prothonotary warbler both veriﬁed and validated (Tirpak and others 2009a). 123 Red-cockaded Woodpecker Status The red-cockaded woodpecker (Picoides borealis) is a federally endangered, nonmigratory resident of old-growth pine forest (particularly longleaf pine) throughout the Southeast (Jackson 1994). Due to the low detection rate for this species (0.05 bird/ route in the WGCP), BBS data poorly estimates population trends (Table 5). The red-cockaded woodpecker is designated as a species warranting critical recovery in both the WGCP and CH (regional combined score = 21), though it is extirpated from the latter region. Natural History John and Karen Hollingsworth, U.S. Fish & Wildlife Service Due to the limited availability of suitable habitat, the red-cockaded woodpecker lives in loose family groups and engages in cooperative breeding (Jackson 1994). Home ranges are large (average = 76.1 ha) but highly variable (17.2 to 159.5 ha; reviewed in Doster and James 1998). Suitable habitat is deﬁned by two primary habitat components. The ﬁrst is the presence of large pines. Pines at least 35 cm d.b.h. generally are required for a stand to be occupied by the red-cockaded woodpecker (Davenport and others 2000, James and others 2001, Walters and others 2002). However, once large pine density exceeds 80 per ha, family group size (a demographic parameter related to productivity; Heppell and others 1994) declines (Walters and others 2002). Similarly, as the average d.b.h. of overstory pines increases above 35 cm, habitat quality declines (Davenport and others 2000), though these declines likely are linked to the maturation of the forests rather than to the negative eﬀects of large trees directly. Similar patterns have been observed for overstory pine basal area and small pine tree density in occupied stands, where values for these habitat attributes are lower than local maxima (James and others 2001, Rudolph and others 2002, Walters and others 2002). Open midstory is the second notable feature of high-quality habitat for the red-cockaded woodpecker. Hardwood midstory trees should be less than 3.26 m tall and ideally less than 1.8 m (Davenport and others 2002, Walters and others 2002). The open midstory typically is maintained through periodic ﬁre (burn interval of 1 to 3 years), which also facilitates a wiregrass understory (James and others 2001). Because this species is nonmigratory and suitable habitat is disjunct, connectivity of patches is critical for the long-term persistence of this species across the landscape. Model Description The HSI model for the red-cockaded woodpecker includes eight variables: landform, landcover, successional age class, forest patch size, pine basal area, hardwood basal area, connectivity, and large pine (> 35 cm d.b.h.) density. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 117). 124 Table 117.—Relationship between landform, landcover type, age class, and suitability scores for redcockaded woodpecker habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.000 Evergreen 0.000 0.000 0.200 0.600 0.800 Mixed 0.000 0.000 0.200 0.400 0.400 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.000 Evergreen 0.000 0.000 0.200 0.600 0.800 Mixed 0.000 0.000 0.200 0.400 0.400 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous Evergreen 0.000 0.000 0.000 0.000 0.000 0.200 0.000 0.600 (0.700) 0.000 0.800 (1.000) Mixed 0.000 0.000 0.200 0.400 0.400 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.000 We directly assigned SI scores to these combinations on the basis of relative rankings of vegetation types and successional age classes for red-cockaded woodpeckers reported by Hamel (1992). We included forest patch size (SI2) as a variable because of the large home ranges of the red-cockaded woodpecker. We assumed that the minimum and maximum home range sizes reported by Doster and James (1998) represented patch size thresholds for nonsuitable and optimal habitat, respectively. To inform the shape of the curve between these points, we assumed that the minimum area requirement of habitat identiﬁed in the red-cockaded woodpecker recovery plan (USDI Fish and Wildl. Serv. 2003) deﬁned average (SI score = 0.500) habitat suitability. We used these data (Table 118) to deﬁne a logarithmic function to predict SI scores from forest patch size (Fig. 69). Pine basal area (SI3) is a key component of red-cockaded woodpecker habitat, and sites with pine basal areas that are too low or too high are of poor quality. We ﬁt a quadratic function (Fig. 70) to data from Conner and others (1995) and Walters and others (2002; Table 119) on the relative abundance of this species in habitats with varying levels of pine basal area. 125 Table 118.—Relationship between forest patch size and suitability index (SI) scores for red-cockaded woodpecker habitat Suitability Index Score 1.0 Forest patch size (ha) 0.8 0.6 SI score 17 a 0.0 49 b 0.5 170a 0.4 a Doster and James (1998). b USDI Fish and Wildl. Serv. (2003). 1.0 0.2 0.0 0 50 100 150 200 Forest Patch Size (ha) Figure 69.—Relationship between forest patch size and suitability index (SI) scores for red-cockaded woodpecker habitat. Equation: SI score = 0.4334 * ln(forest patch size) – 1.2133. Table 119.—Relationship between basal area of pines and suitability index (SI) scores for redcockaded woodpecker habitat Suitability Index Score 1.0 Pine basal area (m2/ha) 0.8 0.0 SI score a 0.00 2.3b 0.6 0.4 0.2 0.50 12.7 c 1.00 14.2 c 1.00 20.0 a 0.25 a Assumed value. b Walters and others (2000). c 0.0 0 10 20 Conner and others (1995). Pine Basal Area (m^2/ha) Figure 70.—Relationship between pine basal area and suitability index (SI) scores for red-cockaded woodpecker habitat. Equation: SI score = 0.0367 + 0.2006 * (pine basal area) – 0.009507 * (pine basal area)2. Mid- and overstory hardwoods reduce habitat suitability for red-cockaded woodpeckers. We ﬁt an inverse logistic function (Fig. 71) to data from Kelly and others (1993) and Wilson and others (1995) (Table 120) on the amount of hardwood basal area (SI4) around woodpecker nest cavities to predict habitat suitability based on this habitat feature. As a resident species occupying disjunct habitat patches, the red-cockaded woodpecker exists in metapopulations. Therefore, dispersal between suitable forest patches is critical for the persistence of this species on the landscape. Isolated patches lacking a breeding female have no productivity, so we used the median dispersal distance for females (3.2 km; Jackson 126 Table 120.—Relationship between basal area of hardwoods (m2/ha) and suitability index (SI) scores for red-cockaded woodpecker habitat Suitability Index Score 1.0 Hardwood basal area (m2/ha) 0.8 0.6 SI score 0.0 a 1.0 3.9 b 1.0 8.6c 0.4 0.5 14.6 c 0.0 20.0 a 0.0 a Assumed value. b Wison and others (1995). c Kelly and others (1993). 0.2 0.0 0 10 20 Hardwood Basal Area (m^2/ha) Figure 71.—Relationship between hardwood basal area and suitability index (SI) scores for red-cockaded woodpecker habitat. Equation: SI score = 1 - (1.001 / (1 + (5745.304 * e -1.006 * hardwood basal area ))). Table 121.—Relationship between distance to nearest habitat patch and suitability index (SI) scores for red-cockaded woodpecker habitat Suitability Index Score 1.0 Distance to nearest habitat patch (m) 0.8 0a 0.6 3,200 1.00 b 20,000 0.4 a b SI score 0.50 a 0.01 Assumed value. Jackson (1994). 0.2 0.0 0 10000 20000 30000 Distance to Nearest Patch (m) Figure 72.—Relationship between habitat connectivity and suitability index (SI) scores for red-cockaded woodpecker habitat. Equations: SI score = e -0.0002 * distance to nearest habitat patch. 1994) to deﬁne average SI score (0.500). However, long-distance dispersal does occur (Larry Hedrick, 2006, U.S. Forest Service, pers. commun.), so we assigned to patches isolated more than 20 km from any other suitable site at least some residual suitability (0.010). We ﬁt an exponential relationship (Fig. 72) through these data points (Table 121) to describe how the connectivity of patches inﬂuences habitat suitability. Large pines (SI6) are a necessary component of red-cockaded woodpecker habitat because this bird disproportionately forages and nests in large pines. However, there is a threshold above which habitat suitability declines and increasingly large trees reduce the preferred open 127 Table 122.—Relationship between large pine (> 35 cm d.b.h.) density (trees/ha) and suitability index (SI) scores for red-cockaded woodpecker habitat Suitability Index Score 1.0 Large pine density 0.8 0 a 0.4 0.2 0.000 b 0.647 30b 0.765 45 b 0.882 60 b 1.000 75 b 1.000 15 0.6 90b 0.0 0 50 100 150 Large Pine Density (trees/ha) 105 a b 1.000 b Assumed value. Walters and others (2002). Figure 73.—Relationship between large pine tree (> 35 cm d.b.h.) density and suitability index (SI) scores for redcockaded woodpecker habitat. Equation: SI score = 0.0269 * (pine tree density) – 0.000193 * (pine tree density)2 + 0.1127. character of the forest. We ﬁt a quadratic function (Fig. 73) to data from Walters and others (2002), who identiﬁed this threshold at 60 to 90 large pines per ha (Table 122). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure (SI1, SI3, SI4, and SI6) and landscape composition (SI2 and SI5) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI3 * SI4 * SI6)0.250 * (SI2 * SI5)0.500)0.500 Veriﬁcation and Validation The red-cockaded woodpecker was found in only 10 of the 88 subsections within the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.49) between average HSI score and mean BBS route abundance across all subsections. However, when subsections where the red-cockaded woodpecker was not found were removed from the analysis, the relationship was not signiﬁcant (P = 0.645; rs = 0.17). Thus, the HSI model predicts the absence of the red-cockaded woodpecker better than its abundance in subsections where it is found. The generalized linear model predicting BBS abundance from BCR and HSI for the red-cockaded woodpecker was signiﬁcant (P ≤ 0.001; R2 = 0.203), and the coeﬃcient on the HSI predictor variable was both positive (β = 0.094) and signiﬁcantly diﬀerent from zero (P = 0.042). Therefore, we considered the HSI model for the red-cockaded woodpecker both veriﬁed and validated (Tirpak and others 2009a). 128 SI score 0.824 Red-headed Woodpecker Status The red-headed woodpecker (Melanerpes erythrocephalus) is found throughout North America east of the Rocky Mountains; however, it is absent from New England and the higher elevations of the central and southern Appalachians. Since 1967, populations have declined by 3.2 percent per year in the WGCP and by 1 percent in the CH (Sauer and others 2005) (Table 5). This species is a Bird of Conservation Concern and a management attention priority in both the CH and WGCP (regional combined score = 16 and 17, respectively; Table 1). Dave Menke, U.S. Fish & Wildlife Service Natural History The red-headed woodpecker is one of the most recognizable birds of the eastern United States and southern Canada, but few in-depth studies of this species have been conducted (Smith and others 2000). Nesting habitat consists of deciduous woodlands, including upland and bottomland hardwoods, riparian strips, open woods, open wooded swamps, groves of dead and dying trees, orchards, shelterbelts, parks, open agricultural lands, savannas, forest edges, roadsides, and utility poles (Smith and others 2000). It prefers xeric sites with large, tall trees, high basal area, and a sparse understory. The red-headed woodpecker exhibits seasonal shifts in habitat use. Population dynamics are linked to annual ﬂuctuations in oak acorn crops, and migration occurs in northern and western populations when hard mast is limited (Rodewald 2003). More locally, winter territories are established around small food caches within forest interiors; breeding territories are larger (3.1 to 8.5 ha in Florida) and concentrated along edges (Smith and others 2000). Occurrence of the red-headed woodpecker varies with mean patch dimension, edge density of agricultural land, and the area of urban landcover (Lukomski 2003). It is a primary cavity excavator and snag availability may drive habitat selection (Giese and Cuthbert 2003). This species often is associated with high snag densities (Conner and others 1994) in mature stands near openings (Conner and Adkisson 1977, Brawn and others 1984). Snag density and basal area of dead elm distinguish nest sites from random sites in Minnesota (Giese and Cuthbert 2003). Similarly, loblolly pine stands with both standing and down dead woody debris removed contain fewer birds (Lohr and others 2002). Snags retained as groups provide multiple snags for roosting and foraging. Hardwood snags are used predominantly for foraging, whereas pine snags are more commonly used for nesting (Smith and others 2000). Thinnings and prescribed ﬁres that open the understory and create snags are beneﬁcial. Model Description The HSI model for the red-headed woodpecker includes seven variables: landform, landcover, successional age class, snag density, large snag (> 20 cm d.b.h.) density, sawtimber tree (> 28 cm d.b.h.) density, and the occurrence of edge. 129 Table 123.—Relationship of landform, landcover type, and successional age class to suitability index scores for red-headed woodpecker habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.125 0.250 0.250 Transitional-shrubland 0.000 0.000 0.125 0.250 0.250 Deciduous 0.000 0.000 0.125 0.250 0.250 Evergreen 0.000 0.000 0.250 0.500 0.500 Mixed 0.000 0.000 0.250 0.500 0.500 Orchard-vineyard 0.000 0.000 0.125 0.250 0.250 Woody wetlands 0.000 0.000 0.250 0.625 0.750 Low-density residential 0.000 0.000 0.125 0.375 0.500 Transitional-shrubland 0.000 0.000 0.250 0.500 0.500 Deciduous 0.000 0.000 0.125 0.375 0.500 Evergreen 0.000 0.000 0.250 0.500 0.500 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.250 0.500 0.500 Orchard-vineyard 0.000 0.000 0.125 0.375 0.500 Woody wetlands 0.000 0.000 0.250 0.500 0.500 Low-density residential 0.000 0.000 0.250 0.750 1.000 Transitional-shrubland 0.000 0.000 0.250 Deciduous 0.000 0.000 0.250 0.500 (0.750) 0.750 0.500 (1.000) 1.000 Evergreen 0.000 0.000 0.250 Mixed 0.000 0.000 0.250 0.500 (0.750) 0.500 0.500 (1.000) 0.500 Orchard-vineyard 0.000 0.000 0.250 0.750 1.000 Woody wetlands 0.000 0.000 0.250 0.750 1.000 The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 123). We directly assigned SI scores to these combinations on the basis of data from Hamel (1992) on the relative value of various vegetation types and successional age classes as red-headed woodpecker habitat in the Southeast. This species relies heavily on snags for nesting, foraging, and roosting. King and others (2007) observed 31.8 snags per ha in savanna habitat used by the red-headed woodpecker, though basal area was only 0.9 m2 per ha in that study. Therefore, we adjusted snag densities to reﬂect the intermediate basal area values (12 to 15 m2/ha; Heltzel and Leberg 2006) characteristic of stands used by the red-headed woodpecker in the WGCP and CH BCRs. We assumed that 500 snags per ha represented an upper threshold above which maximal suitability was achieved and that 200 snags per ha represented a threshold below which sites were unsuitable (Table 124). We ﬁt a logistic function (Fig. 74) through these data to predict how habitat suitability varied with snag density (SI2). Because the snag density in SI2 includes all dead trees greater than 2.5 cm d.b.h., we also included large snag (> 20 cm d.b.h.) density (SI3) as a variable. This additional requirement ensured the presence of snags suitable for nesting 130 Table 124.—Influence of snag density on suitability index (SI) scores for red-headed woodpecker habitat Suitability Index Score 1.0 Snag density (snags/ha)a 0.8 0.6 0.4 SI score 0 0.00 200 0.00 400 0.75 500 1.00 a Assumed value. 0.2 0.0 0 100 200 300 400 500 Snag Density (snags/ha) Figure 74.—Relationship between snag density (snags * 100/ha) and suitability index (SI) scores for red-headed woodpecker habitat. Equation: SI score = 1.006 / (1 + (249051.2 * e (-0.0338 * snag density))). Table 125.—Influence of large snag (> 20 cm d.b.h.) density (snags/ha) on suitability index (SI) scores for red-headed woodpecker habitat Suitability Index Score 1.0 Large snag density 0.8 0.6 0.0 a 8.5 b 12.0 SI score 0.0 0.1 a 1.0 a Assumed value. b Lohr and others (2002). 0.4 0.2 0.0 0 10 20 Large Snag Density (snags/ha) Figure 75.—Relationship between large snag (> 20 cm d.b.h.) density and suitability index (SI) scores for redheaded woodpecker habitat. Equation: SI score = 1.006 / (1 + (90614077 * e (-1.899 * large snag density))). in high-quality habitats. We relied on data from Lohr and others (2002) to inform an inverse logistic function (Fig. 75) that linked habitat suitability to large snag density (Table 125). The red-headed woodpecker breeds in relatively open habitats with widely spaced large trees near openings (King and others 2007). Therefore, we included sawtimber tree density (SI4) and edge occurrence (SI5) as variables. We assumed that habitat suitability was highest when sawtimber tree density was 20 or fewer trees per ha and lowest when sawtimber tree density exceeded 50 trees per ha (Table 126). We ﬁt a logistic function (Fig. 76) through these data points to quantify the relationship between sawtimber tree density and SI scores. To identify edges, we used a 7 × 7 pixel moving window (210 x 210 m) to locate the transitions between 131 Table 126.—Influence of sawtimber tree (> 28 cm d.b.h.) density (trees/ha) on suitability index (SI) scores for red-headed woodpecker habitat Suitability Index Score 1.0 Sawtimber tree densitya 0.8 0.6 0.4 0.2 SI score 0 1.00 20 1.00 30 0.75 35 0.25 50 0.00 70 0.00 a Assumed value. 0.0 0 25 50 75 Sawtimber Tree Density (trees/ha) Figure 76.—Relationship between sawtimber tree (≥ 28 cm d.b.h.) density (trees * 10/ha) and suitability index (SI) scores for red-headed woodpecker habitat. Equation: SI score =1 – (1.000 / (1 + (1615169 * e (-0.4398 * sawtimber tree density)))). Table 127.—Influence of edge on suitability index (SI) scores for red-headed woodpecker habitat 7 × 7 window around forest pixel includes fielda SI score Yes 1.0 No 0.1 a Field defined as any shrub-seedling or grass-forb age class pixel, or natural grasslands, pasture-hay, fallow, urban-recreational grasses, emergent herbaceous wetlands, open water, high intensity residential, commercial-industrial-transportation, bare rock-sand-clay, quarries-strip mines-gravel pits, row crops, or small grains. Forest defined as any used sapling, pole, or sawtimber age class pixel of low-density residential, transitional, shrublands, deciduous, mixed, evergreen, orchard, or woody wetlands. forest and non-forest landcovers or sapling-pole-sawtimber and grass-forb-shrub-seedling successional age class stands. We assigned to edge habitats the maximal SI score and discounted areas with no edge (Table 127). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure attributes (SI1, SI2, SI3, and SI4) and multiplied this product by the SI score for edge occurrence (SI5). Overall HSI = ((SI1 * SI2 * SI3 * SI4)0.250) * SI5 Veriﬁcation and Validation The red-headed woodpecker was found in all 88 subsections of the CH and WGCP. Spearman rank correlation failed to identify a positive association between average HSI score and mean BBS abundance. The generalized linear model predicting BBS abundance from BCR and HSI for the red-headed woodpecker was signiﬁcant (P ≤ 0.001; R2 = 0.225); however, the coeﬃcient on the HSI predictor variable was negative (β = -3.359). Therefore, we considered the HSI model for the red-headed woodpecker neither veriﬁed nor validated (Tirpak and others 2009a). 132 Swainson’s Warbler Status The Swainson’s warbler (Limnothlypis swainsonii) is a neotropical migrant that breeds in dense thickets across the Southeast. Due to its overall low density and occurrence in habitats not well sampled by BBS, estimates of population trends based on this dataset are not reliable (Sauer and others 2005) (Table 5). Nonetheless, this species is a Bird of Conservation Concern and has a regional combined score of 20 in both the CH and WGCP (Table 1). An estimated 46 percent of the continental population of the Swainson’s warbler breeds in the WGCP (Panjabi and others 2001). Natural History Chandler S. Robbins, Patuxent Bird Identification InfoCenter Photo used with permission The Swainson’s warbler is distributed locally across the Southeast (Brown and Dickson 1994). Once believed to be restricted to canebrakes in bottomland hardwood and swamp forests of the Atlantic and Gulf Coastal Plains, it now has been documented breeding at low densities in regenerating clearcuts in Texas and rhododendron-mountain laurel thickets in the southern Appalachians (Graves 2002). Territory size is large for a wood warbler (3.2 ha) (Brown and Dickson 1994), and this species demonstrates area sensitivity. In Illinois, the Swainson’s warbler is not observed on tracts smaller than 350 ha (Eddleman and others 1980). This species does not use canopy height, basal area, successional age class, or species composition as habitat cues (Eddleman and others 1980, Graves 2002), but selects habitat based on understory characteristics. Dense thickets are required, and stem densities of about 35,000 stems per ha are optimal (Graves 2002). Canopy gaps are important for encouraging this dense growth, and canopy cover typically is high (70 to 80 percent) but rarely closed (> 90 percent) (Eddleman and others 1980, Graves 2001, Somershoe and others 2003). Understory vegetation is primarily woody; herbaceous cover is typically sparse (< 25 percent) (Eddleman and others 1980, Brown and Dickson 1994). Leaf litter is abundant and provides an important foraging substrate (Graves 2001, Somershoe and others 2003). Hydrology is a critical factor inﬂuencing the habitat suitability for this warbler. In bottomland and ﬂoodplain habitats, birds select areas that typically are drier than surrounding sites (Graves 2001, Somershoe and others 2003). Inundation of otherwise suitable habitat from March - September negatively aﬀects the quality of an otherwise suitable site (Graves 2002). This species occasionally breeds in xeric uplands with appropriate understory characteristics (Carrie 1996). Model Description The HSI model for the Swainson’s warbler includes six variables: landform, landcover, successional age class, forest patch size, proportion of forest in a 1-km radius, and small stem (< 2.5 cm d.b.h.) density. 133 Table 128.—Relationship of landform, landcover type, and successional age class to suitability index scores for Swainson’s warbler habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.400 0.000 0.000 Deciduous 0.000 0.000 0.400 0.900 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.400 0.900 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.200 0.000 0.000 Deciduous 0.000 0.000 0.200 0.500 0.600 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.400 0.800 0.800 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.200 0.000 0.000 Deciduous 0.000 0.000 0.200 0.500 0.600 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.400 0.800 0.800 The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 128). We adjusted the relative habitat quality rankings of Hamel (1992) for Swainson’s warbler vegetation and successional age class associations to maximize habitat suitability in woody wetland habitats along ﬂoodplains, and to ensure that transitional sapling stands that may be used in the WGCP were assigned SI scores (Carrie 1996). We included forest patch size (SI2) in the model because of the preference of the Swainson’s warbler for interior sites within large forest tracts. We assumed that the minimum patch size in which Eddleman and others (1980) observed this species (350 ha) represented optimal habitat. Because this study was at the northern limit of the range of the Swainson’s warbler, we assumed that birds would occupy signiﬁcantly smaller tracts (Table 129). We based a logistic function on these assumptions to predict the impact of forest patch size on habitat suitability (Fig. 77). Nevertheless, the suitability of a speciﬁc patch size also is inﬂuenced by its landscape context (SI3). In predominantly forested landscapes, small forest patches that otherwise may not be suitable may be occupied due to their proximity to large forest blocks (Rosenberg and others 1999). To capture this relationship, we ﬁt a logistic function (Fig. 78) to data (Table 130) derived from Donovan and others (1997), who observed diﬀerences 134 Table 129.—Influence of forest patch size on suitability index (SI) score for Swainson’s warbler habitat Suitability Index Score 1.0 Forest patch size (ha) 0.8 0a 35 0.6 0.4 0.00 a 0.01 250 a 0.50 350 b 1.00 500a 1.00 a 0.2 SI score b Assumed value. Eddleman and others (1980). 0.0 0 100 200 300 400 500 Forest Patch Size (ha) Figure 77.— Relationship between forest patch size and suitability index (SI) scores for Swainson’s warbler habitat. Equation: SI score = (1.001 / (1 + (31096.960 * e -0.041 * (forest patch size)))). Table 130.—Relationship between landscape composition (proportion forest in 1-km radius) and suitability index (SI) scores for Swainson’s warbler habitat Suitability Index Score 1.0 Landscape composition 0.8 0.6 0.4 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 Landscape Composition (proportion forest in 1-km radius) Figure 78.—Relationship between landscape composition and suitability index (SI) scores for Swainson’s warbler habitat. Equation: SI score = 1.047 / (1.000 + (1991.516 * e -10.673 * landscape composition )). SI score 0.00a 0.00 0.10 a 0.00 0.20 a 0.00 0.30 a 0.00 0.40a 0.00 0.50 a 0.10 0.60 a 0.25 0.70 b 0.50 0.80a 0.75 0.90 a 0.90 1.00 a 1.00 a Assumed value. b Donovan and others (1997). in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoint between moderately and lightly fragmented forest deﬁned the speciﬁc cutoﬀ for average (SI score = 0.500) habitat. We used the maximum score from SI2 or SI3 to account for the higher suitability of small patches in predominantly forested landscapes relative to their size alone. The Swainson’s warbler breeds in dense thickets and stem densities of approximately 35,000 stems per ha are optimal (SI score = 1.000) (Graves 2002). Stem densities can be even higher in early-successional bottomland hardwoods (> 200,000/ha), but we assumed habitat 135 Table 131.—Influence of small stem (< 2.5 cm d.b.h.) density (stems * 1,000/ha) on suitability index (SI) scores for Swainson’s warbler habitat Suitability Index Score 1.0 Small stem density 0.8 0.000a 7.550 0.6 0.4 0.0 b 0.1 17.365 b 0.5 34.773 b 1.0 72.999b 1.0 a 0.2 b Assumed value. Graves (2002). 0.0 0 10 20 30 40 50 Small Stem Density (stems * 1,000/ha) Figure 79.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for Swainson’s warbler habitat. Equation: SI score = 1.008 / (1.000 + (59.233 * e -0.235 * (small stem density / 1000))). suitability was not negatively aﬀected by stem density. Therefore, we ﬁt a logistic function (Fig. 79) to data from Graves (2002) that captured the eﬀect of varying stem density on habitat suitability (Table 131). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure (SI1 and SI4) and multiplied that by the maximum SI score for forest patch size (SI2) or percent forest in the 1-km landscape (SI3) and ﬁnally calculated the geometric mean of that product. Overall HSI = ((SI1 * SI4)0.500 * Max(SI2 or SI3))0.500 Veriﬁcation and Validation The Swainson’s warbler was found only in 31 of the 88 subsections within the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.010) positive relationship (rs = 0.31) between average HSI score and mean BBS route abundance across all subsections. However, when subsections where this species was not found were removed from the analysis, the relationship was not signiﬁcant (P = 0.893; rs = -0.03). Thus, the HSI model better predicts the absence of the Swainson’s warbler than its abundance in subsections where this species is found. The generalized linear model predicting BBS abundance from BCR and HSI for the Swainson’s warbler was signiﬁcant (P ≤ 0.001; R2 = 0.260); however, the coeﬃcient on the HSI predictor variable was negative (β = -0.298). Therefore, we considered the HSI model for the Swainson’s warbler veriﬁed but not validated (Tirpak and others 2009a). 136 SI score Swallow-tailed Kite Status The swallow-tailed kite (Elanoides forﬁcatus) is a neotropical raptor that reaches the northern limit of its distribution in the Unites States. Once ranging throughout the Mississippi River drainage as far north as Minnesota, this species now is restricted to seven states in the Southeast. There are too few swallowtailed kites detected on BBS routes in the WGCP to estimate a population trend; however, this species is a Bird of Conservation Concern and immediate D.A. Rintoul, Patuxent Bird Identification InfoCenter management attention priority in this BCR (regional Photo used with permission combined score = 18; Table 1). The swallow-tailed kite no longer breeds in the CH and this species warrants critical recovery eﬀorts in this region (regional combined score = 19). Natural History The swallow-tailed kite is a rare breeder in the continental United States. The current restriction of this species to seven southern states (with limited distributions in all but Florida) represents a signiﬁcant contraction of its former range. Most of the information on this bird in the United States is from Florida (Meyer 1995). The swallow-tailed kite has a large home range (500 to 1800 ha) that increases substantially (> 20,000 ha) when the long but regular foraging forays characteristic of this species are included. With such a large home range, the important role of landscape structure on habitat suitability is not surprising. Critical habitat elements are large, tall trees for nesting and open habitats containing prey (Meyer 1995, Sykes and others 1999). Any interspersion of these features is useable (e.g., trees adjacent to prairie, wetlands, or marsh). Landscapes containing bottomland hardwood forest interspersed with scattered openings are particularly attractive. The edges of pine forests along swamps and riparian zones also are commonly used along the Coastal Plains. The Mississippi kite typically occupies habitats that are drier and contain more contiguous forest than the habitats of the swallow-tailed kite. Model Description The HSI model for the swallow-tailed kite includes six variables: landform, landcover, successional age class, forest patch size, landscape composition, and dominant tree density. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 132). We then directly assigned SI scores to these combinations on the basis of relative habitat quality rankings from Hamel (1992) for the swallow-tailed kite. However, we assumed that only stands in the sawtimber successional age class provided suitable habitat for this species We also included forest patch size (SI2) as a variable because of this bird’s large home range and association with large blocks of forested wetlands. We ﬁt a logarithmic function (Fig. 80) 137 Table 132.—Relationship of landform, landcover type, and successional age class to SI scores for swallowtailed kite habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.500 Evergreen 0.000 0.000 0.000 0.000 0.500 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.000 0.000 0.500 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 1.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.800 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.000 0.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.000 0.800 to data (Table 133) from Zimmerman (2004) on the mean value of forest in 5-km buﬀers around swallow-tailed kite nest sites and the maximum home range size reported by Cely and Sorrow (1990) to assess the impact of forest patch size on habitat suitability scores for the swallow-tailed kite. Like the Mississippi kite, the swallow-tailed kite forages aerially in open habitats, so it requires both forested sites for nesting and open areas for foraging (SI3). We based the ideal composition of vegetation types in the landscape on data from Sykes and others (1999), who observed 20 percent open habitat within 200-ha core areas in Florida. We maximized habitat suitability at this threshold and reduced SI scores in landscapes containing greater or lower proportions of open habitat (Table 134, Fig. 81). The swallow-tailed kite nests in dominant trees (SI4) that extend above the canopy. We assumed that trees with a d.b.h. greater than 76.2 cm would extend above the canopy in the sawtimber stands that provide the exclusive habitat for this species. We assumed that one dominant tree per ha would satisfy this requirement and that the swallow-tailed kite would be absent from stands with a uniform canopy (zero dominant trees/ha). We ﬁt an exponential function (Fig. 82) to the values between these data points and assumed that 138 Table 133.—Influence of forest patch size on suitability index (SI) scores for swallow-tailed kite habitat Suitability Index Score 1.0 Forest patch size (ha) 4,300a 0.8 40,000b 0.6 a b SI score 0.5 1.0 Zimmerman (2004). Cely and Sorrow (1990). 0.4 0.2 0.0 0 10000 20000 30000 40000 Forest Patch Size (ha) Figure 80.—Relationship between forest patch size and suitability index (SI) scores for swallow-tailed kite habitat. Equation: SI score = 0.224 * ln(forest patch size) – 1.376. Table 134.—Suitability index scores for swallowtailed kite habitat based on landscape composition (percent of open habitat) within 1,200-ha landscape Suitability Index Score 1.0 Landscape compositiona 0.8 6 1.0 25b 1.0 b 0.1 75 0.4 0.1 c 20 0.6 SI score b a Water, grasslands, cultivated lands, and emergent wetlands. Assumed value. c Sykes and others (1999). b 0.2 0.0 0 25 50 75 100 Landscape Composition (% open habitat in 1,200-ha area) Figure 81.—Relationship between landscape composition and suitability index (SI) scores for swallow-tailed kite habitat. Equation: SI score = (0.001 * 0.885(percent open habitat)) * (percent open habitat)3.065. stands with 14 dominant trees per ha (the maximum value from the WGCP during the FIA surveys of the 1990s) were associated with maximum habitat suitability (Table 135). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure attributes (SI1 and SI4) and landscape composition (SI2 and SI3) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI4)0.500 * (SI2 * SI3)0.500)0.500 139 Table 135.—Influence of dominant tree (> 76.2 cm d.b.h.) density (trees/ha) on suitability index (SI) scores for swallow-tailed kite habitat Suitability Index Score 1.0 Dominant tree densitya 0.8 0 0.6 1.0 14 1.0 Assumed value. 0.2 0.0 0 2 4 Dominant Tree Density (trees/ha) Figure 82.—Relationship between dominant tree density and (SI) scores for swallow-tailed kite habitat. Equation: SI score = 1 – e -8.734 * dominant tree density. Veriﬁcation and Validation The swallow-tailed kite was found in 8 of the 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.73) between average HSI score and mean BBS route abundance across all subsections. However, when subsections where this species was not found were removed from the analysis, the relationship was not signiﬁcant (P = 0.432; rs = 0.33). Thus, the HSI model better predicts the absence of the swallow-tailed kite than its abundance in subsections where this species is found. The generalized linear model predicting BBS abundance from BCR and HSI for the swallow-tailed kite was signiﬁcant (P ≤ 0.001; R2 = 0.522), and the coeﬃcient on the HSI predictor variable was both positive (β = 0.725) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the swallow-tailed kite both veriﬁed and validated (Tirpak and others 2009a). 140 0.0 1 a 0.4 SI score Whip-poor-will Status The whip-poor-will (Caprimulgus vociferus) is a neotropical migrant with a more northerly range than the chuck-will’s-widow, though the ranges of the two are not exclusive and overlap broadly across the CH. The whip-poor-will has declined by 1.8 percent per year since 1967 in the CH (Sauer and others 2005) (Table 5), where this species is a Bird of Conservation Chandler S. Robbins, Patuxent Bird Identification InfoCenter Concern and has a regional combined score of 17 Photo used with permission (Table 1). A large proportion of the continental population (35.5 percent) breeds in the CH (Panjabi and others 2001). This species is a rare breeder in the WGCP (regional combined score = 13). Natural History Owing to its cryptic coloration and crepuscular activity pattern, the whip-poor-will is one of the least studied birds in North America (Cink 2002). Breeding habitat in the CH and WGCP consists of xeric deciduous and mixed forests with a sparse understory. This species also is associated with open areas, such as rural farmland, powerline and roadway rights-ofway, clearcuts and selectively logged forest, old ﬁelds, and reclaimed surface mines. Shaded forest stands with limited ground cover adjacent to open areas for foraging provide ideal whip-poor-will habitat. This species usually is absent from extensive areas of closed canopy forest, but there are no data on minimum or maximum thresholds for forest patch size. Small, isolated woodlots in a Maryland agricultural landscape are not used (Reese 1996, cited in Cink 2002). In Massachusetts, Grand and Cushman (2003) found that the whip-poorwill is strongly associated with complex patch shapes and high contrast edges. This species nests on the forest ﬂoor and hatching is synchronized with the full moon to optimize the foraging time of adults. Whip-poor-wills are not strongly territorial; home range varies from 2.8 to 11.1 ha. Model Description The HSI model for whip-poor-will includes four variables: landform, landcover, successional age class, and the relative composition of forest and open habitats in the landscape. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 136). We directly assigned SI scores to these combinations on the basis of relative habitat rankings for vegetation and successional age class associations of the whip-poor-will reported by Hamel (1992). The whip-poor-will nests in forest and forages in openings. As a result, it requires landscapes with an interspersion (SI2) of these landcover types. We assumed that a landscape with 70 percent forest and 30 percent open habitat was optimal (Michael Wilson, 2006, College of William & Mary, pers. commun.) and that landscapes with a greater proportion of forest 141 Table 136.—Relationship of landform, landcover type, and successional age class to suitability index scores for whip-poor-will habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.334 0.667 0.667 Deciduous 0.000 0.000 0.334 0.667 0.667 Evergreen 0.000 0.000 0.334 0.667 0.667 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.334 0.834 1.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.333 0.333 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.334 0.834 1.000 Deciduous 0.000 0.000 0.334 0.667 0.667 Evergreen 0.000 0.000 0.334 0.667 0.667 Mixed 0.000 0.000 0.334 0.834 1.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.333 0.333 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.334 0.834 1.000 Deciduous 0.000 0.000 0.334 0.667 0.667 Evergreen 0.000 0.000 0.334 0.667 0.667 Mixed 0.000 0.000 0.334 0.834 1.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.167 0.333 0.333 were more suitable than those with less forest cover so long as some openings were present (Table 137; sensu Cooper 1981). We calculated the overall HSI score as the geometric mean of the two component variables. Overall HSI = (SI1 * SI2)0.500 Veriﬁcation and Validation The whip-poor-will was found in 76 of the 88 subsections within the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P = 0.005) positive relationship (rs = 0.30) between average HSI score and mean BBS route abundance across subsections. This relationship was even stronger (rs = 0.47) when subsections in which the whip-poor-will was not detected were removed from the analysis. The generalized linear model predicting BBS abundance from BCR and HSI for the whip-poor-will was signiﬁcant (P = 0.002; R2 = 0.139), and the coeﬃcient on the HSI predictor variable was positive (β = 1.270) but not signiﬁcantly diﬀerent from zero (P = 0.229). Therefore, we considered the HSI model for the whip-poor-will veriﬁed but not validated (Tirpak and others 2009a). 142 Table 137.—Suitability index scores for whip-poor-will habitat based on the relative proportion of cells providing open and forest landcover within 500-m radius Proportion opena Proportion forestb 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.0 0.1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.3 0.00 0.10 0.10 0.10 0.10 0.10 0.10 0.10 0.4 0.00 0.25 0.25 0.25 0.25 0.25 0.25 0.5 0.00 0.50 0.50 0.50 0.50 0.50 0.6 0.00 0.70 0.90 0.90 0.90 0.7 0.00 0.80 0.90 1.00 0.8 0.00 0.80 0.90 0.9 0.00 0.80 1.0 0.00 a Open = pasture/hay, recreational grasses, grasslands/herbaceous, and emergent herbaceous wetland landcovers or grass-forb and shrub-seedling successional age class stands. b Forest = any habitats with positive SI1 values (Table 136). 143 White-eyed Vireo Status The white-eyed vireo (Vireo griseus) is a neotropical migrant that breeds throughout the southeastern United States. Populations have been stable in both the CH and WGCP over the last 40 years, but have been increasing in the WGCP by 1.6 percent annually since 1980 (Sauer and others 2005; Table 5). This species requires management attention in both the CH and WGCP (regional combined score = 15 and 16, respectively) but is not a Bird of Conservation Concern in either BCR (Table 1). David Arbour, U.S. Forest Service Natural History A small secretive songbird, the white-eyed vireo is associated with dense vegetation in secondary deciduous scrub-shrub, wood margins, overgrown pastures, abandoned farmlands, streamside thickets, and even mid- to late successional forests (Hopp and others 1995). This species shares habitats with the blue-gray gnatcatcher, Carolina wren, gray catbird, and brown thrasher, but prefers later successional forest than the yellow-breasted chat, prairie warbler, and Bell’s vireo. In Texas, the white-eyed vireo breeds in areas of shrubby vegetation (0 to 1 m) with dense foliage (Conner and Dickson 1997). Similarly, in Virginia, it prefers habitats with an extensive undergrowth of shrubs, brambles, and saplings interspersed with taller trees (10 to 20 percent of area). Vireo densities are higher in glade and regenerating forest habitat than edges in Missouri (Fink and others 2006). Densities also are inversely related to vegetation height, foliage density at 12 to 15 m, density of pole trees, and percent canopy closure (Conner and others 1983). Prather and Smith (2003) found that this species was more abundant in tornado-damaged forest in Arkansas than in undamaged areas. In South Carolina, abundance was positively related to gap size in bottomland forest that had been harvested by group-selection (Moorman and Guynn 2001). Territory size (0.1 to 1.8 ha) and population density vary with habitat quality. Brood parasitism aﬀects nearly half of all nests and may signiﬁcantly reduce productivity. The white-eyed vireo is more abundant in wide riparian strips of bottomland hardwood forest than in narrow strips (Kilgo and others 1998). Model Description The HSI model for the white-eyed vireo includes six variables: landform, landcover, successional age class, edge occurrence, canopy cover, and small stem (< 2.5 cm d.b.h.) density. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 138). We directly assigned SI scores to these combinations on the basis of data from Hamel (1992) on the habitat associations of the white-eyed vireo in the Southeast. 144 Table 138.—Relationship of landform, landcover type, and successional age class to suitability index (SI) scores for white-eyed vireo habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 1.000 0.834 0.500 0.333 Deciduous 0.000 1.000 0.834 0.500 0.333 Evergreen 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.000 0.000 0.000 Orchard-vineyard 0.000 1.000 0.834 0.500 0.333 Woody wetlands 0.000 1.000 0.834 0.500 0.333 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.667 0.500 0.333 0.167 Deciduous 0.000 0.667 0.500 0.333 0.167 Evergreen 0.000 0.667 0.500 0.333 0.167 Mixed 0.000 0.667 0.500 0.333 0.167 Orchard-vineyard 0.000 0.667 0.500 0.333 0.167 Woody wetlands 0.000 1.000 0.834 0.500 0.333 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.667 0.500 0.333 0.167 Deciduous 0.000 0.667 0.500 0.333 0.167 Evergreen 0.000 0.667 0.500 0.333 0.167 Mixed 0.000 0.667 0.500 0.333 0.167 Orchard-vineyard 0.000 0.667 0.500 0.333 0.167 Woody wetlands 0.000 1.000 0.834 0.500 0.333 Table 139.—Influence of edge on suitability index (SI) scores for white-eyed vireo habitat 3 × 3 pixel window around forest pixel includes field? a SI score b 1.00 Yes In older forest stands, the white-eyed vireo 0.01 concentrates on edges (SI2) and other areas with No a Field defi ned as any sapling, shrub-seedling, or grass-forb dense vegetation (Conner and Dickson 1997). age class pixel, or natural grasslands, pasture-hay, fallow, We used a 3 × 3 pixel window (90 x 90 m) urban-recreational grasses, emergent herbaceous wetlands, open water, high-intensity residential, commercial-industrialto identify the interfaces between pole and transportation, bare rock-sand-clay, quarries-strip mines-gravel sawtimber successional age class forest and pits, row crops, or small grains. Forest defined as any pole or sawtimber age class pixel of low-density residential, transitional, herbaceous and nonforest landcovers (hard shrublands, deciduous, mixed, evergreen, orchard, or woody edge) or shrub-seedling, grass-forb, and sapling wetlands. b Seedling-shrub and sapling habitats used regardless of edge. successional age class forest (soft edge). We assumed that pole and sawtimber stands adjacent to these edges would have the highest SI score but applied a residual suitability value (0.01) to areas not identiﬁed as edge habitats to compensate for small forest gaps and openings that may be used. Shrub-seedling and sapling stands were suitable habitat regardless of edge (Table 139). 145 Table 140.—Influence of canopy cover on suitability index (SI) scores for white-eyed vireo habitat Suitability Index Score 1.0 Canopy cover (percent) 0.8 0.6 0.4 0.2 0.0 0 25 50 75 100 29.26 1.000 31.00 b 1.000 71.86 a 0.482 73.00b 0.493 91.00 b 0.000 93.38 a 0.024 95.58 a 0.036 96.59b 0.012 a b Canopy Cover (%) SI score a Annand and Thompson (1997). Prather and Smith (2003). Figure 83.—Relationship between canopy cover and suitability index (SI) scores for white-eyed vireo habitat. Equation: SI score = 1 - (1.0101 / (1 + (127952.58 * e -0.1629 * canopy cover))). Table 141.—Influence of small stem (< 2.5 cm d.b.h.) density (stems * 1,000/ha) on suitability index (SI) scores for white-eyed vireo habitat Suitability Index Score 1.0 Small stem densitya 0.8 0.6 2 0.01 4 0.50 8 1.00 a 0.4 SI score Annand and Thompson (1997). 0.2 0.0 0.0 2.5 5.0 7.5 10.0 Small Stem Density (stems * 1,000/ha) Figure 84.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for white-eyed vireo habitat. Equation: SI score = (1.000 / (1 + (14512.121 * e -2.396 * (small stem density / 1000)))). To reﬁne the association of the white-eyed vireo with canopy gaps, we modeled the eﬀect of canopy cover (SI3) on SI scores as an inverse logistic function (Fig. 83) that captured the absence of this species in closed-canopy forests (Table 140). Finally, we ﬁt a logistic function (Fig. 84) to data from Annand and Thompson (1997) (Table 141) on the inﬂuence of small stem (< 2.5 cm d.b.h.) density (SI4) on the relative density of the white-eyed vireo to quantify the relationship between SI scores and this habitat feature. Assuming that this species uses edge as a surrogate to its preferred shrub-seedling and sapling habitats, we calculated HSI scores separately for shrub-seedling-sapling and pole-sawtimber 146 forest stands. In the former, the geometric mean of forest structure variables alone deﬁnes the suitability score. For the latter, landscape composition (edge occurrence) also was a factor in the calculation. Shrub-seedling and sapling (young) successional age classes: HSIYoung: (SI1 * SI3 * SI4)0.333 Pole and sawtimber (old) successional age classes: HSIOld: ((SI1 * SI3 * SI4)0.333 * SI2)0.500 To determine the overall HSI score, we summed the age class speciﬁc HSIs: Overall HSI = HSIYoung + HSIOld Veriﬁcation and Validation The white-eyed vireo was found in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P = 0.002) positive association (rs = 0.33) between average HSI score and mean BBS route abundance across all subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the white-eyed vireo was signiﬁcant (P ≤ 0.001; R2 = 0.529); however, the coeﬃcient on the HSI predictor variable was negative (β = -9.070). Therefore, we considered the HSI model for the white-eyed vireo veriﬁed but not validated (Tirpak and others 2009a). 147 Wood Thrush Status The wood thrush (Hylocichla mustelina) is a familiar woodland migrant to the forests of the eastern and central United States. Population declines for this species in the Midwest are linked to higher predation and parasitism rates in fragmented landscapes (Robinson and others 1995, Sauer and others 2005) (Table 5). The wood thrush is both a Bird of Conservation Concern and a management attention priority in the CH and WGCP (regional combined score = 16 and 15, respectively; Table 1). Steve Maslowski, U.S. Fish & Wildlife Service Natural History The wood thrush is a long-distance neotropical migrant that exempliﬁes the decline in songbirds due to forest fragmentation. Due to its general abundance, ease of nest location and monitoring, and area sensitivity, the wood thrush is easy to study and there is a large body of knowledge on this bird (Roth and others 1996). This species is common in deciduous and mixed forests but rare in pure evergreen stands (Roth and others 1996). Mesic, upland forests with a moderate density of midcanopy trees and shrubs for nesting and an open understory with abundant leaf litter for foraging are optimal (Roth and others 1996). Closed overstory canopies are commonly used (Roth and others 1996, Bell and Whitmore 2000). The wood thrush displays area sensitivity in productivity but not in its occupancy of habitats. It nests in forest fragments as small as 0.3 ha, albeit at low densities (Tilghman 1987, Weinberg and Roth 1998), and in narrow (< 150 m wide) riparian strips (Sargent and others 2003). However, nest predation and parasitism rates are extremely high in fragments of less than 80 ha and in riparian buﬀers less than 530 m wide (Donovan and others 1995, Hoover and others 1995, Peak and others 2004). Landscapes with greater amounts of forest cover (particularly unfragmented forest) mitigate some of these eﬀects in small woodlots (Donovan and others 1997, Driscoll and Donovan 2004, Driscoll and others 2005). Nest success is predicted better by the amount of forest in the landscape than by the structural characteristics of microhabitat around nests (Hoover and Brittingham 1998, Driscoll and others 2005). Model Description The HSI model for the wood thrush includes seven variables: landform, landcover, successional age class, forest patch size, percent forest in the local (1-km radius) landscape, small stem (< 2.5 cm d.b.h.) density, and canopy cover. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 142). We directly assigned SI scores to these combinations on the basis of habitat associations reported by Hamel (1992) but made minor adjustments to increase SI scores for sapling stands on the basis of data from Thompson and others (1992). 148 Table 142.—Relationship of landform, landcover type, and successional age class to suitability index scores for wood thrush habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Landform Landcover type Floodplain-valley Low-density residential 0.000 0.250 0.750 0.750 1.000 Transitional-shrubland 0.000 0.250 0.750 0.750 1.000 Deciduous 0.000 0.250 0.750 0.750 1.000 Evergreen 0.000 0.167 0.000 0.000 0.000 Mixed 0.000 0.167 0.333 0.333 0.667 Orchard-vineyard 0.000 0.250 0.333 0.333 0.667 Woody wetlands 0.000 0.250 0.500 0.500 1.000 Low-density residential 0.000 0.250 0.500 0.500 0.834 Transitional-shrubland 0.000 Deciduous 0.000 0.167 (0.000) 0.250 0.333 (0.000) 0.500 0.333 (0.000) 0.500 0.667 (0.000) 0.834 Evergreen 0.000 0.167 0.000 0.000 0.000 Mixed 0.000 0.167 0.333 0.333 0.667 Orchard-vineyard 0.000 0.250 0.333 0.333 0.667 Woody wetlands 0.000 0.334 0.667 0.667 1.000 Low-density residential 0.000 0.334 0.667 0.667 1.000 Transitional-shrubland 0.000 Deciduous 0.000 0.167 (0.000) 0.334 0.333 (0.000) 0.667 0.333 (0.000) 0.500 0.667 (0.000) 0.667 Evergreen 0.000 0.000 0.000 0.000 0.000 Terrace-mesic Xeric-ridge Grass-forb Successional age class Shrubseedling Sapling Pole Saw Mixed 0.000 0.167 0.333 0.333 0.667 Orchard-vineyard 0.000 0.334 0.333 0.333 0.667 Woody wetlands 0.000 0.334 0.667 0.667 1.000 Although the wood thrush will occupy small forest fragments, its density may be lower within them. Therefore, we included forest patch size (SI2) in the HSI model. We ﬁt an exponential function (Fig. 85) to data from Robbins and others (1989) and Kilgo and others (1998) (riparian strips in this study were assumed to be 10 km long) that documented changes in relative occurrence with changes in patch size (Table 143). Nevertheless, the suitability of a forest patch is inﬂuenced not only by its size but also by its landscape context (SI3). To capture this relationship, we ﬁt a logistic function (Fig. 86) to data (Table 144) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoints between these classes (30 and 70 percent forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. We used the maximum SI score from SI2 or SI3 to increase the suitability of small patches in heavily forested landscapes. 149 Table 143.—Influence of forest patch size on suitability index (SI) scores for wood thrush habitat Suitability Index Score 1.0 Forest patch size (ha) 0.8 0 a 1 a 25 0.6 0.0 0.5 b 500a a 0.4 b SI score 1.0 1.0 Robbins and others (1989). Kilgo and others (1998). 0.2 0.0 0 10 20 30 Forest Patch Size (ha) Figure 85.—Relationship between forest patch size and suitability index (SI) scores for wood thrush habitat. Equation: SI score = 1.000 – (1.017 * e -0.710 * (forest patch size ^ 0.797)). Table 144.—Relationship between landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for wood thrush habitat Suitability Index Score 1.0 Landscape composition 0.8 0 a 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 86.—Relationship between landscape composition and suitability index (SI) scores for wood thrush habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * landscape composition)). 0.00 a 0.00 20a 0.05 30 b 0.10 40 a 0.25 50 b 0.50 60a 0.75 70 b 0.90 80 a 0.95 90 b 1.00 10 0.6 100a a b Assumed value. Donovan and others (1997). The wood thrush forages in leaf litter on the forest ﬂoor and is most common in stands with an open understory. We included small stem density (SI4) in the model as a proxy to understory cover. Although some researchers suggest that the wood thrush selects habitats with higher stem densities than generally are available, the controls in these studies typically are in mature forest and the wood thrush may simply be selecting habitats with locally high stem densities (Artman and Downhower 2003). We assumed that the average stem density (1,988 stems/ha) observed by Hoover and Brittingham (1998) around wood thrush nests was representative of optimal habitat. We discounted habitat suitability as small stem density increased due to presumed reductions in leaf litter, the preferred foraging substrate (Roth and others 1996). Nonetheless, Hoover and Brittingham (1998) observed wood thrush 150 SI score 1.00 Table 145.—Influence of small stem (< 2.5 cm d.b.h.) density (stems * 100/ha) on suitability index (SI) scores for wood thrush habitat Suitability Index Score 1.0 Small stem densitya 0.8 0.6 0.4 0 1.0 20 1.0 40 0.7 80 0.1 100 0.0 a 0.2 SI score Assumed value. 0.0 0 25 50 75 100 Small Stem Density (stems * 100/ha) Figure 87.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 100/ha) and suitability index (SI) scores for wood thrush habitat. Equation: SI score = 1 - (0.963 / (1 + (243.780 * e -0.116 * (small stem density / 100))). Table 146.—Influence of canopy cover (percent) on suitability index (SI) scores for wood thrush habitat Suitability Index Score 1.0 Canopy cover (percent) 0.00 70b 0.25 25 0.8 90 0.6 SI score a b 100 0.90 b 1.00 a 0.4 Hoover and Brittingham (1998). b Annand and Thompson (1997). 0.2 0.0 0 25 50 75 100 Canopy Cover (%) Figure 88.—Relationship between canopy cover and suitability index (SI) scores for wood thrush habitat. Equation: SI score = 1.032 / (1 + (141241.64 * e -0.153 * canopy cover)). utilizing sites with extraordinarily high small stem densities (58,500 stems/ha, no doubt localized). Therefore, we assigned residual SI scores to sites with these characteristics. We ﬁt an inverse logistic function (Fig. 87) to small stem density numbers that reﬂected this relationship (Table 145). The wood thrush also is associated with closed-canopied forests, so we included canopy cover (SI5) as a variable and ﬁt a logistic function (Fig. 88) to data from Annand and Thompson (1997) and Hoover and Brittingham (1998) to predict SI scores from canopy cover values (Table 146). 151 To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure attributes (SI1, SI4, and SI5) and then calculated the geometric mean of this value and the maximum of SI scores from forest patch size or percent forest in the landscape (Max(SI2 or SI3)). Overall HSI = ((SI1 * SI4 * SI5)0.333 * Max(SI2 or SI3))0.500 Veriﬁcation and Validation The wood thrush was found in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.52) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the wood thrush was signiﬁcant (P ≤ 0.001; R2 = 0.311), and the coeﬃcient on the HSI predictor variable was both positive (β = 9.992) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the wood thrush both veriﬁed and validated (Tirpak and others 2009a). 152 Worm-eating Warbler Status The worm-eating warbler (Helmitheros vermivorus) breeds on forested slopes of the eastern deciduous forest. It is notably absent from the Mississippi ﬂoodplain and the relatively ﬂat forest-prairie ecotone immediately east of the Great Plains. Its preference for rugged terrain and its high-pitched, insect-like song result in underestimations of its density from roadside surveys. As a result, there are no credible trends from BBS data Charles H. Warren, images.nbii.gov for this species (Table 5). Nevertheless, this species is a Bird of Conservation Concern in both BCRs. However, PIF designates the worm-eating warbler as a management attention priority in the CH (regional combined score = 18) and a planning and responsibility species in the WGCP (regional combined score = 15; Table 1). Natural History The worm-eating warbler is a neotropical migrant that breeds in forest interiors of the Eastern United States (Hanners and Patton 1998). Minimum area requirements range from 21 ha in the mid-Atlantic (Robbins and others 1989) to more than 800 ha in Missouri (Wenny and others 1993). This species nests on the ground along moderate to steep slopes (≥ 20 percent) with dense (≥ 48 percent) shrub understories in mature deciduous and mixed deciduous-coniferous forests (Gale and others 1997). Both Artman and others (2001) and Blake (2005) found that the worm-eating warbler was less abundant in recently burned stands due to the loss of leaf litter, a preferred nesting and foraging substrate. Canopy closure exceeded 95 percent in both Missouri (Wenny and others 1993) and Connecticut (Gale and others 1997). Model Description The HSI model for the worm-eating warbler includes seven variables: landform, landcover, successional age class, slope, forest patch size, percent forest in the landscape, and small stem (< 2.5 cm d.b.h.) density. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 147). We directly assigned SI scores to these combinations on the basis of habitat associations reported by Hamel (1992). We included slope (SI2) in our model because of the prevalence of steep slopes in the territories of the worm-eating warbler. We deﬁned slope classes on the basis of data from Gale and others (1997) who identiﬁed the relative preference of various slopes for this species (Table 148). We also included forest patch size (SI3) as a variable to account for the preference of the worm-eating warbler for forest interiors. We ﬁt a modiﬁed exponential function (Fig. 89) to data from Robbins and others (1989) to quantify the relationship between patch size 153 Table 147.—Relationship of landform, landcover type, and successional age class to suitability index scores for worm-eating warbler habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.300 0.700 0.800 Evergreen 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.200 0.400 0.400 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.200 0.500 0.600 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.300 0.800 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.200 0.400 0.400 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.200 0.400 0.400 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.200 0.600 0.800 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.200 0.400 0.400 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.200 0.400 0.400 and habitat suitability (Table 149). The suitability of a forest patch is inﬂuenced by its size and landscape context (SI4). To capture this relationship, we ﬁt a logistic function (Fig. 90) to data (Table 150) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoints between these classes (30 and 70 percent forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. We assigned the maximum SI score of SI3 or SI4 to each site to account for the higher suitability of small forest patches in heavily forested landscapes. We relied on data from Wenny and others (1993) and Annand and Thompson (1997) (Table 151) to quantify the relationship between SI scores and small stem density (SI5; Fig. 91). We assumed that the worm-eating warbler occupied forests with low stem densities, but these sites had lower suitability scores than sites with well developed understories characterized by dense stems. 154 Table 148.—Influence of slope on suitability index (SI) scores for worm-eating warbler habitat Suitability Index Score 1.0 Slope (percent) a 0.8 0.6 SI score <5 0.0 5-20 0.5 21 1.0 a Gale and others (1997). 0.4 0.2 0.0 0 1000 2000 3000 Forest Patch Size (ha) Figure 89.—Relationship between forest patch size and suitability index (SI) scores for worm-eating warbler habitat. Equation: SI score = 1.035 * e -109.238 / (forest patch size). Table 149.—Influence of forest patch size on suitability index (SI) scores for worm-eating warbler habitat Forest patch size (ha) 21 a 0.0 120b 3,200 a b SI score 0.5 a 1.0 Robbins and others (1989). Assumed value. Suitability Index Score 1.0 0.8 Table 150.—Relationship between landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for worm-eating warbler habitat 0.6 0.4 Landscape composition 0a 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 90.—Relationship between landscape composition and suitability index (SI) scores for worm-eating warbler habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * (landscape composition) )). SI score 0.00 10 a 0.00 20 a 0.05 30 b 0.10 40a 0.25 50 b 0.50 60 a 0.75 70 b 0.90 80a 0.95 90 b 100 a b 1.00 a 1.00 Assumed value. Donovan and others (1997). 155 Table 151.—Influence of small stem (< 2.5 cm d.b.h.) density (stems/ha) on suitability index (SI) scores for worm-eating warbler habitat Suitability Index Score 1.0 Small stem density 0.8 0 a 0.773 4,200c 1.000 b 1.000 4,717 0.4 0.500 b 2,077 0.6 a Assumed value. Annand and Thompson (1997). c Wenny and others (1993). b 0.2 0.0 0 1000 2000 3000 4000 5000 Small Stem Density (stems/ha) Figure 91.—Relationship between small stem (< 2.5 cm d.b.h.) density and suitability index (SI) scores for worm-eating warbler habitat. Equation: SI score = 1.000 / (1 + e 18.707 – 0.006 * (small stem density)) ^ 1 / 26.989 Equation takes the general form: y = a/(1 + eb-cx)1/d. To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure (SI1 and SI5) and landscape composition (Max(SI3 or SI4) and SI2) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI5)0.500 * (Max(SI3 or SI4) * SI2)0.500)0.500 Veriﬁcation and Validation The worm-eating warbler was found in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.66) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the worm-eating warbler was signiﬁcant (P ≤ 0.001; R2 = 0.408), and the coeﬃcient on the HSI predictor variable was both positive (β = 1.798) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the worm-eating warbler both veriﬁed and validated (Tirpak and others 2009a). 156 SI score Yellow-billed Cuckoo Status The yellow-billed cuckoo (Coccyzus americanus) is a neotropical migrant that breeds throughout North America east of the Rocky Mountains. The yellow-billed cuckoo is abundant in the CH and WGCP (10.43 and 12.93 birds/route, respectively), but populations in these BCRs have declinded slightly (Table 5). Although the yellow-billed cuckoo is not a Bird of Conservation Concern in either BCR, it is a management attention priority in both due to the importance of these regions (the core of this bird’s range) for the sustainability of the continental population (Table 1). Natural History A long-distance migrant, the yellow-billed cuckoo breeds in low, U.S. Forest Service dense scrub near streams, marshes, and wetlands within otherwise open woodlands (Hughes 1999). It is among the most common birds in ﬂoodplain habitats along the Mississippi River and occupies both young cottonwood-willow stands and mature silver maple forests (Knutson and others 2005). This species exhibits some area sensitivity. Conner and others (2004) found that the yellow-billed cuckoo was most abundant in riparian strips more than 70 m wide, and Aquilani and Brewer (2004) recorded highest abundances in forest tracts larger than 55 ha. Breeding success is correlated with insect outbreaks, particularly those of hairy caterpillars, and population densities vary greatly with food supply. Nests are located in dense, broadleaved, deciduous shrubs or trees within 10 m of the ground. Twedt and others (2001) reported no diﬀerence in nest success between bottomland hardwoods and cottonwood plantations, nor did Wilson (1999) report a diﬀerence in nest success among stands subject to alternative thinning rates in Arkansas. On the basis of anticipated harvest scenarios, Klaus and others (2005) predicted that populations of the yellow-billed cuckoo would decline by approximately 37 percent on the Cherokee National Forest over the next 60 years. Model Description The HSI model for the yellow-billed cuckoo includes seven variables: landform, landcover, successional age class, edge occurrence, midstory tree (11 to 25 cm d.b.h.) density, percent forest in the landscape (10-km radius), and forest patch size. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 152). We directly assigned SI scores to these combinations on the basis of habitat associations of the yellow-billed cuckoo reported by Hamel (1992). We increased SI scores within ﬂoodplainvalley and terrace-mesic landforms to account for the higher abundance of the yellow-billed cuckoo on these sites in the CH and WGCP. 157 Table 152.—Relationship of landform, landcover type, and successional age class to suitability index scores for yellow-billed cuckoo habitat. Values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.500 0.667 1.000 Transitional-shrubland 0.000 0.000 0.500 0.667 1.000 Deciduous 0.000 0.000 0.500 0.667 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.167 0.333 0.333 Orchard-vineyard 0.000 0.000 0.500 0.667 1.000 Woody wetlands 0.000 0.000 0.333 0.667 0.667 Low-density residential 0.000 0.000 0.500 0.667 1.000 Transitional-shrubland 0.000 0.000 Deciduous 0.000 0.000 0.500 (0.000) 0.500 0.667 (0.000) 0.667 1.000 (0.000) 1.000 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.167 0.333 0.333 Orchard-vineyard 0.000 0.000 0.500 0.667 1.000 Woody wetlands 0.000 0.000 0.333 0.667 0.667 Low-density residential 0.000 0.000 0.250 0.333 0.500 Transitional-shrubland 0.000 0.000 Deciduous 0.000 0.000 0.250 (0.000) 0.250 0.333 (0.000) 0.333 0.500 (0.000) 0.500 Evergreen 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.000 0.000 0.083 0.167 0.167 Orchard-vineyard 0.000 0.000 0.250 0.333 0.500 Woody wetlands 0.000 0.000 0.167 0.333 0.333 This species is more abundant within edge (SI2) habitats than within forest interiors (Kroodsma 1984). We used a 9 × 9 pixel moving window (270 x 270 m) to identify habitat edges and assumed that these locations represented optimal habitat. Nevertheless, nonedge habitats also are used by the yellow-billed cuckoo so we assigned to these sites only a slightly lower SI score (0.667; Table 153). The yellow-billed cuckoo breeds in forest stands with welldeveloped midstories (SI3). We ﬁt a quadratic function (Fig. 92) to data from Annand and Thompson (1997) on the relative densities of this species in stands with diﬀerent midstory tree densities (Table 154) to predict how SI scores responded to changes in this habitat variable. 158 Table 153.—Influence of edge on suitability index (SI) scores for yellow-billed cuckoo habitat 9 × 9 pixel window around forest pixel includes fielda SI score Yes 1.000 No 0.667 a Field defined as any shrub-seedling or grass-forb age class pixel, or natural grasslands, pasturehay, fallow, urban-recreational grasses, emergent herbaceous wetlands, open water, high-intensity residential, commercial-industrial-transportation, bare rock-sand-clay, quarries-strip mines-gravel pits, row crops, or small grains. Forest defined as any used sapling, pole, or sawtimber age class pixel of low-density residential, transitional, shrublands, deciduous, mixed, evergreen, orchard, or woody wetlands. Table 154.—Influence of midstory tree (11–25 cm d.b.h.) density (trees/ha) on suitability index (SI) scores for yellow-billed cuckoo habitat Suitability Index Score 1.0 Midstory tree densitya 0.8 0.6 0.4 SI score 70 0.439 320 1.000 361 0.902 506 0.244 a Annand and Thompson (1997). 0.2 0.0 0 100 200 300 400 500 Midstory Tree Density (trees/ha) Suitability Index Score Figure 92.—Relationship between midstory tree (11–25 cm d.b.h.) density and suitability index (SI) scores for yellow-billed cuckoo habitat. Equation: SI score = 0.0078 * (midstory tree density) – 0.00001 * (midstory tree density)2 – 0.0355. 1.0 Table 155.—Relationship between landscape composition (percent forest in 10-km radius) and suitability index (SI) scores for yellow-billed cuckoo habitat 0.8 Landscape compositiona 0.6 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 10-km radius) Figure 93.—Relationship between landscape composition and suitability index (SI) scores for yellow-billed cuckoo habitat. Equation: SI score = 1.002 * e((0 – ((landscape forest composition * 100) – 74.165) ^ 2) / 1064.634) . SI score 0 0.00 10 0.10 20 0.20 30 0.30 40 0.40 50 0.50 60 0.75 70 1.00 80 1.00 90 0.75 100 0.50 a Assumed value. Although a forest-breeding species, the yellow-billed cuckoo is associated with fragmented landscapes (Robbins and others 1989, Hughes 1999). We assumed that 70 to 80 percent forest in a 10-km landscape (SI4) was characteristic of ideal habitat (Table 155) and ﬁt a function that reduced SI scores symmetrically as forest compositions departed from these ideal proportions (Fig. 93). Nevertheless, the cuckoo exhibits area sensitivity and may be absent or at low densities in small fragments (Robbins and others 1989, Bancroft and others 1995, Hughes 1999). Therefore, we used data from these sources to derive a logistic function (Fig. 94) that quantiﬁed the relationship between habitat suitability and forest patch size (SI5; Table 156). 159 Table 156.—Influence of forest patch size on suitability index (SI) scores for yellow-billed cuckoo habitat Suitability Index Score 1.0 Forest patch size (ha) 0.8 0 a 0.00 7.5 0.6 b 0.00 22c 0.25 d 1.00 50 0.4 SI score a Assumed value. Bancroft and others (1995). c Hughes (1999). d Robbins and others (1989). b 0.2 0.0 0 10 20 30 40 50 Forest Patch Size (ha) Figure 94.—Relationship between forest patch size and suitability index (SI) scores for yellow-billed cuckoo habitat. Equation: SI score = 1.000 / (1.000 + (20350.850 * e -0.401 * forest patch size )). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure (SI1 and SI3) and landscape composition (SI2, SI4, and SI5) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI3)0.500 * (SI2 * SI4 * SI5)0.333)0.500 Veriﬁcation and Validation The yellow-billed cuckoo was found in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P = 0.024) positive relationship (rs = 0.24) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the yellow-billed cuckoo was signiﬁcant (P ≤ 0.001; R2 = 0.190), and the coeﬃcient on the HSI predictor variable was positive (β = 5.265) but not signiﬁcantly diﬀerent from zero (P = 0.302). Therefore, we considered the HSI model for the yellow-billed cuckoo veriﬁed but not validated (Tirpak and others 2009a). 160 Yellow-breasted Chat Status The yellow-breasted chat (Icteria virens) is a neotropical migrant that breeds in early successional habitats across the eastern United States. The distribution of this species in the West is patchy. Populations have responded to the loss of early successional habitat and have Chandler S. Robbins, Patuxent Bird Identification InfoCenter declined sharply across the northern edge of Photo used with permission this bird’s distribution (Sauer and others 2005). Within the CH, where this species has a regional combined score of 16 and is a management attention priority, populations have declined by approximately 2 percent per year during the last 40 years (Table 5). Conversely, at the southern limit of their range, populations have increased (1.3 percent annual increases in the WGCP from 1966 to 2005; Table 5). Natural History The yellow-breasted chat breeds in low, dense, deciduous and evergreen vegetation within forests lacking a closed canopy (Eckerle and Thompson 2001). Habitat associations include forest edges and openings, regenerating forest, powerline rights-of-way, fencerows, upland thickets, abandoned farms, and shrubby areas along streams, swamps, and ponds. Chats are most abundant in 6- to 9-year-old cottonwood plantations in the Mississippi Alluvial Valley (Twedt and others 1999). However, Annand and Thompson (1997) observed similar abundance across stands subject to alternative forest management prescriptions. In east Texas, density is positively correlated with foliage density at 0 to 3 m, the percentage of saplings that are pine, and the number of shrub species. Densities are negatively aﬀected by increasing vegetation height, percent canopy cover, foliage density at 12 to 15 m, and density of pole trees (Conner and others 1983). In Missouri, the yellow-breasted chat nests more than 20 m from the edge of large early successional patches characterized by high densities of small stems (Burhans and Thompson 1999). Nest success increases with patch size; territories range from 0.5 to 1.6 ha. Model Description The HSI model for the yellow-breasted chat includes six variables: landform, landcover, successional age class, edge, early successional patch size, and small stem (< 2.5 cm d.b.h.) density. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 157). We directly assigned SI scores to these combinations based on data from Hamel (1992). However, we assumed that shrub-seedling habitats were optimal and that pole stands were nonhabitat. We ignored landform eﬀects in assessing habitat suitability for this species. Chats prefer to nest more than 20 m from the edge of mature forest (SI2) (Woodward and others 2001). Thus, we used a 3 × 3 pixel window (90 x 90 m) to identify suitable early 161 Table 157.—Relationship of landform, landcover type, and successional age class to suitability index scores for yellow-breasted chat habitat Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.000 1.000 0.500 0.000 0.000 Deciduous 0.000 1.000 0.500 0.000 0.000 Evergreen 0.333 0.667 0.500 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Mixed 0.333 0.667 0.334 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.667 0.334 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.333 1.000 0.500 0.000 0.000 Deciduous 0.167 1.000 0.500 0.000 0.000 Evergreen 0.333 0.667 0.500 0.000 0.000 Mixed 0.333 0.667 0.334 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.667 0.334 0.000 0.000 Low-density residential 0.000 0.000 0.000 0.000 0.000 Transitional-shrubland 0.333 1.000 0.500 0.000 0.000 Deciduous 0.333 1.000 0.500 0.000 0.000 Evergreen 0.333 0.667 0.500 0.000 0.000 Mixed 0.333 0.667 0.334 0.000 0.000 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.667 0.334 0.000 0.000 successional forest sites immediately adjacent to pole or sawtimber successional age class forest. We reduced the suitability of these sites by half (SI score = 0.500; Table 158). Table 158.—Influence of edge on suitability index (SI) scores for yellow-breasted chat habitat 3 × 3 pixel window around early successional pixel includes mature foresta SI score Yes 0.5 The yellow-breasted chat is associated with large No 1.0 a Early successional = grass-forb, shrub-seedling, and patches of early successional forest (SI3). We sapling successional age classes; mature forest = pole or aggregated all grass-forb, shrub-seedling, and sapling sawtimber successional age classes. successional age class sites to calculate patch sizes for this species. We ﬁt a logarithmic function (Fig. 95) to data from Rodewald and Vitz (2005) on the relative abundance of the yellow-breasted chat in early successional patches of various sizes to quantify the relationship between patch size and habitat suitability (Table 159). This species occupies sites with high small stem densities (SI4). Therefore, we ﬁt a logistic function (Fig. 96) to data from Annand and Thompson (1997) relating the relative density of the yellow-breasted chat to small stem densities (Table 160) to predict the eﬀect of this habitat characteristic on habitat suitability. 162 Table 159.—Influence of early successional patch size on suitability index (SI) scores for yellowbreasted chat habitat; early successional patches only include grass-forb, shrub-seedling, and sapling successional age classes Suitability Index Score 1.0 0.8 Early successional patch size (ha)a 0.6 0.4 SI score 6 0.6 14.5 1.0 a Rodewald and Vitz (2005). 0.2 0.0 0 5 10 15 Early Successional Patch Size (ha) Figure 95.—Relationship between early successional patch size and suitability index (SI) scores for yellow-breasted chat habitat. Equation: SI score = -0.212 + 0.453 * ln(forest patch size). Table 160.—Influence of small stem (< 2.5 cm d.b.h.) density (stems * 1,000/ha) on suitability index (SI) scores for yellow-breasted chat habitat Small stem densitya SI score Suitability Index Score 1.0 0.8 0.6 0.0 0.000 3.8 0.516 8.1 1.000 a 0.4 Annand and Thompson (1997). 0.2 0.0 0.0 2.5 5.0 7.5 10.0 Small Stem Density (stems * 1,000/ha) Figure 96.—Relationship between small stem (< 2.5 cm d.b.h.) density (stems * 1000/ha) and suitability index (SI) scores for yellow-breasted chat habitat. Equation: SI score = (1.000 / (1 + (1148216.200 * e -3.689 * (small stem density / 1000)))). To calculate the overall HSI score for the yellow-breasted chat, we determined the geometric mean of the SI scores for forest structure attributes (SI1 and SI4) and the SI score for landscape composition (SI2 and SI3) separately and then the geometric mean of these values together. Overall HSI = ((SI1 * SI4)0.500 * (SI2 * SI3)0.500)0.500 163 Veriﬁcation and Validation The yellow-breasted chat was found in all 88 subsections of the CH and WGCP. Spearman rank correlation identiﬁed a signiﬁcant (P ≤ 0.001) positive relationship (rs = 0.40) between average HSI score and mean BBS route abundance across subsections. The generalized linear model predicting BBS abundance from BCR and HSI for the yellow-breasted chat was signiﬁcant (P ≤ 0.001; R2 = 0.379), and the coeﬃcient on the HSI predictor variable was both positive (β = 93.367) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the yellow-breasted chat both veriﬁed and validated (Tirpak and others 2009a). 164 Yellow-throated Vireo Status The yellow-throated vireo (Vireo ﬂavifrons) is a neotropical migrant found throughout North America east of the Great Plains. Populations in both the CH and WGCP are stable (Sauer and others 2005) (Table 5). This species is not a Bird of Conservation Concern in either region (Table 1) but is a planning and responsibility species in both the CH (regional combined score = 16) and WGCP (regional combined score = 15). Approximately 20 percent of the continental population breeds in these two BCRs (Panjabi and others 2001). Chandler S. Robbins, Patuxent Bird Identification InfoCenter Photo used with permission Natural History The yellow-throated vireo breeds along the edges of mature forest stands; its abundance may even decline within forest interiors (Rodewald and James 1996). Appropriate edges include streams, rivers, swamps, and roads. Parks, orchards, and suburban habitats also may be used (Rodewald and James 1996). This species uses both bottomland and upland sites but is restricted to deciduous and mixed-forest habitats. As a forest edge species, it is not area sensitive and may beneﬁt from canopy gaps. However, Robbins and others (1989) observed a positive relationship between the abundance of the yellow-throated vireo and forest cover within a 2-km buﬀer. Similarly, this bird did not use riparian forests strips that were less than 70 m wide in east Texas (Conner and others 2004). Thus, the yellow-throated vireo prefers canopy gaps within forested landscapes. The key component of its habitat is canopy structure, and this species selects taller trees (> 20 m) than other vireos (James 1976). Robbins and others (1989) also noted a positive relationship between abundance and canopy height. Speciﬁc tree species do not aﬀect selection (Gabbe and others 2002). Model Description Our HSI model for the yellow-throated vireo includes six variables: landform, landcover, successional age class, forest patch size, percent forest in the landscape (1-km radius), and canopy cover. The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 161). We directly assigned SI scores to these combinations on the basis of relative rankings of habitat associations for the yellow-throated vireo described in Hamel (1992). Although a forest edge species, the yellow-throated vireo is aﬀected by forest area (SI2) and the percentage of forest in the landscape (SI3). We ﬁt a logarithmic function (Fig. 97) to data from Blake and Karr (1987) and Kilgo and others (1998) to describe the relationship between forest patch size and habitat suitability (Table 162). Similarly, we used a logistic function to predict habitat suitability from percent forest cover in a 1-km radius landscape (Fig. 98) based on data (Table 163) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 165 Table 161.—Relationship of landform, landcover type, and successional age class to SI scores for yellowthroated vireo habitat Landform Landcover type Floodplain-valley Low-density residential 0.000 0.000 0.000 0.333 0.667 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.333 0.667 Evergreen 0.000 0.000 0.000 0.000 0.000 Terrace-mesic Xeric-ridge Grass-forb Successional age class Shrubseedling Sapling Pole Mixed 0.000 0.000 0.000 0.167 0.333 Orchard-vineyard 0.000 0.000 0.000 0.333 0.667 Woody wetlands 0.000 0.000 0.000 0.417 0.834 Low-density residential 0.000 0.000 0.000 0.250 0.500 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.250 0.500 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.167 0.333 Orchard-vineyard 0.000 0.000 0.000 0.250 0.500 Woody wetlands 0.000 0.000 0.000 0.500 1.000 Low-density residential 0.000 0.000 0.000 0.334 0.667 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.334 0.667 Evergreen 0.000 0.000 0.000 0.000 0.000 Mixed 0.000 0.000 0.000 0.167 0.333 Orchard-vineyard 0.000 0.000 0.000 0.334 0.667 Woody wetlands 0.000 0.000 0.000 0.500 1.000 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoints between these classes (30 and 70 percent forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. The aﬃnity of the yellow-throated vireo for canopy gaps led us to incorporate canopy cover in the HSI model for this species (SI4). We ﬁt a smoothed quadratic function (Fig. 99) to data from Kahl and others (1985) (Table 164) on the relative density of this species at varying canopy closures, and assumed that Kahl’s optimal designation of canopy cover (80 to 90 percent) was associated with maximum SI scores. Further, we assumed that habitat suitability declined symmetrically as canopy cover departed from this optimum. 166 Saw Table 162.—Influence of forest patch size on suitability index (SI) scores for yellow-throated vireo habitat Suitability Index Score 1.0 Forest patch size (ha) 0.8 6.5 25 0.6 a b 0.2 0.381 100 b 0.429 200 b 0.524 500 b 0.794 1000b 0.0 0 250 500 750 1000 0.000 0.365 50b 0.4 SI score a b Forest Patch Size (ha) 1.000 Blake and Karr (1987). Kilgo and others (1998). Suitability Index Score Figure 97.—Relationship between forest patch size and suitability index (SI) scores for yellow-throated vireo habitat. Equation: SI score = 0.180 * ln(forest patch size) – 0.323. 1.0 Table 163.—Relationship between landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for yellow-throated vireo habitat 0.8 Landscape composition 0 0.6 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 98.—Relationship between landscape composition and suitability index (SI) scores for yellow-throated vireo habitat. Equation: SI score = 1.005 / (1.000 + (221.816 * e -0.108 * (landscape composition) )). a SI score 0.00 10a 0.00 20 a 0.05 30 b 0.10 40 a 0.25 50b 0.50 60 a 0.75 70 b 0.90 80 a 0.95 90b 100 a b 1.00 a 1.00 Assumed value. Donovan and others (1997). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure (SI1 and SI4) and landscape composition attributes (SI2 and SI3) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI4)0.500 * (SI2 * SI3)0.500)0.500 167 Table 164.—Influence of canopy cover (percent) on suitability index (SI) scores for yellow-throated vireo habitat Suitability Index Score 1.0 Canopy cover (percent) 0.8 SI score 0a 0.6 0.4 70 0.75 80 b 1.00 90 a 0.90 a b 0.2 0.00 b Assumed value. Kahl and others (1985). 0.0 0 25 50 75 100 Canopy Cover (%) Figure 99.—Relationship between canopy cover and suitability index (SI) scores for yellow-throated vireo habitat. Equation: SI score = 1.011 * e (0-((canopy cover – 82.319)^2 / 508.869)). Veriﬁcation and Validation The yellow-throated vireo was found in all 88 subsections of the CH and WGCP. Spearman rank correlation on average HSI score and mean BBS route abundance per subsection identiﬁed a signiﬁcant (P ≤ 0.001) positive association (rs = 0.51) between these two variables. The generalized linear model predicting BBS abundance from BCR and HSI for the yellow-throated vireo was signiﬁcant (P = 0.002; R2 = 0.133), and the coeﬃcient on the HSI predictor variable was both positive (β = 2.811) and signiﬁcantly diﬀerent from zero (P ≤ 0.001). Therefore, we considered the HSI model for the yellow-throated vireo both veriﬁed and validated (Tirpak and others 2009a). 168 Yellow-throated Warbler Status The yellow-throated warbler (Dendroica dominica) is a neotropical migrant that breeds in the southeastern United States and reaches its highest densities in the Ohio River Valley. Deanna K. Dawson, Patuxent Bird Identification InfoCenter This species has remained relatively stable Photo used with permission in the WGCP over the past 40 years but has increased considerably in the CH (3.8 percent per year since 1967; Table 5). The yellowthroated warbler is not a Bird of Conservation Concern in either BCR but is a planning and responsibility species in the CH (regional combined score = 15; Table 1). Natural History The yellow-throated warbler breeds in two distinct habitat types: mature bottomland hardwood forest and dry, upland oak-pine forest (Hall 1996). It is more common in the former. This species shows a strong aﬃnity for cypress along the Coastal Plains, but prefers sycamore along inland rivers (Hall 1996, Gabbe and others 2002). Where Spanish moss is found, it is used for both foraging and nesting (Hall 1996). Elsewhere, the warbler forages by creeping along limbs and probing leaf clusters and pinecones. This bird is both an interior and edge species and may occupy woodlots as small as 6 ha (Blake and Karr 1987). Robbins and others (1989) associated this species with large tree (> 38 cm d.b.h.) density, forest in a 2-km buﬀer, and coniferous canopy cover. Model Description Our HSI model for the yellow-throated warbler includes six variables: landform, landcover, successional age class, large tree (> 50 cm d.b.h.) density, distance to water, and percent forest in the landscape (1-km radius). The ﬁrst suitability function combines landform, landcover, and successional age class into a single matrix (SI1) that deﬁnes unique combinations of these classes (Table 165). We directly assigned SI scores to these combinations on the basis of habitat associations outlined by Hamel (1992) for the yellow-throated warbler in the Southeast. We also incorporated large tree density (SI2) into the HSI model for the yellow-throated warbler because of its aﬃnity for nesting and foraging in large trees (Hamel 1992, Robbins and others 1989). Lacking data points from the literature to ﬁt a curve, we assumed that SI scores were logistically related to large tree density up to 50 trees per ha and remained optimal above this threshold (Fig. 100, Table 166). The yellow-throated warbler typically nests near water (Hall 1996, Hamel 1992). Thus, we included distance to water (SI3) in the HSI model. We assumed that sites closer to water had a higher suitability. Lacking quantitative data on the potential eﬀect of water on habitat suitability, we assumed that the size of the yellow-throated warbler’s territory is similar to that of the Acadian ﬂycatcher but that the warbler is not as dependent on water as the 169 Table 165.—Relationship of landform, landcover type, and successional age class to suitability index (SI) scores for yellow-throated warbler habitat; values in parentheses apply to West Gulf Coastal Plain/Ouachitas Successional age class Grass-forb Shrubseedling Sapling Pole Saw Low-density residential 0.000 0.000 0.000 0.250 0.500 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.250 0.500 Evergreen 0.000 0.000 0.000 0.333 0.667 Mixed 0.000 0.000 0.000 0.333 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.834 1.000 Low-density residential 0.000 0.000 0.000 0.167 0.167 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.167 0.167 Evergreen 0.000 0.000 0.000 0.333 0.667 Mixed 0.000 0.000 0.000 0.333 0.667 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Landform Landcover type Floodplain-valley Terrace-mesic Xeric-ridge Woody wetlands 0.000 0.000 0.000 0.500 1.000 Low-density residential 0.000 0.000 0.000 0.167 0.333 Transitional-shrubland 0.000 0.000 0.000 0.000 0.000 Deciduous 0.000 0.000 0.000 0.167 0.333 Evergreen 0.000 0.000 0.000 0.667 (0.334) 0.667 Mixed 0.000 0.000 0.000 0.333 (0.167) 0.333 Orchard-vineyard 0.000 0.000 0.000 0.000 0.000 Woody wetlands 0.000 0.000 0.000 0.500 1.000 ﬂycatcher. Therefore, we assumed that all sites less than 100 m from water were optimal but reduced SI more slowly for the yellow-throated warbler than the Acadian ﬂycatcher as distance to water increased (Fig. 101; Table 167). The yellow-throated warbler responds to the percentage of forest in the landscape (SI4). To capture this relationship, we ﬁt a logistic function (Fig. 102) to data (Table 168) derived from Donovan and others (1997), who observed diﬀerences in predator and brood parasite communities among highly fragmented (< 15 percent), moderately fragmented (45 to 50 percent), and lightly fragmented (> 90 percent forest) landscapes. We assumed that the midpoints between these classes (30 and 70 percent forest) deﬁned the speciﬁc cutoﬀs for poor (SI score ≤ 0.10) and excellent (SI score ≥ 0.90) habitat, respectively. 170 Table 166.—Influence of large tree (> 50 cm d.b.h.) density (trees/ha) on suitability index (SI) scores for yellow-throated warbler habitat Suitability Index Score 1.0 Large tree densitya 0.8 0.6 0.4 0 0.00 20 0.25 40 0.75 50 1.00 75 1.00 a 0.2 SI score Assumed value. 0.0 0 25 50 75 100 Large Tree Density (trees/ha) Figure 100.—Relationship between large tree (> 50 cm d.b.h.) density and suitability index (SI) scores for yellow-throated warbler habitat. Equation: SI score = 1.000 / (1.0000 + (38.185 * e -0.123 * large tree density)). Table 167.—Relationship between distance to water and suitability index (SI) scores for yellow-throated warbler habitat Suitability Index Score 1.0 Distance to water (m) a 0.8 0.6 0.4 SI score 100 b 1.00 300 b 0.75 400 b 0.25 500b 0.00 a Water defined as NHD streams or NLCD water, woody wetlands, and emergent herbaceous wetlands classes. b Assumed value. 0.2 0.0 0 100 200 300 400 500 Distance to Water (m) Figure 101.—Relationship between distance to water and suitability index (SI) scores for yellow-throated warbler habitat. Equation: SI score = 1 - (1.050 / (1 + (1661.322 * e -0.021 * distance to water ))). To calculate the overall HSI score, we determined the geometric mean of SI scores for forest structure (SI1 and SI2) and landscape composition attributes (SI3 and SI4) separately and then the geometric mean of these means together. Overall HSI = ((SI1 * SI2)0.500 * (SI3 * SI4)0.500)0.500 171 Table 168.—Relationship between landscape composition (percent forest in 1-km radius) and suitability index (SI) scores for yellow-throated warbler habitat Suitability Index Score 1.0 Landscape composition 0.8 SI score 0a 0.6 0.4 0.2 0.0 0 25 50 75 100 Landscape Composition (% forest in 1-km radius) Figure 102.—Relationship between landscape composition and suitability index (SI) scores for yellow-throated warbler habitat. Equation: SI score= 1.005 / (1.000 + (221.816 * e -0.108 * (landscape composition) )). 0.00 10 a 0.00 20 a 0.05 30b 0.10 40 a 0.25 50 b 0.50 60 a 0.75 70b 0.90 80 a 0.95 90 b 100 a b 1.00 a Assumed value. Donovan and others (1997). Veriﬁcation and Validation The yellow-throated warbler was found in 87 of the 88 subsections within the CH and WGCP. Spearman rank correlation on average HSI score and mean BBS route abundance identiﬁed a signiﬁcant (P ≤ 0.001) positive association (rs = 0.48) between these two variables within subsections where this species was detected. The generalized linear model predicting BBS abundance from BCR and HSI for the yellow-throated warbler was signiﬁcant (P = 0.003; R 2 = 0.125), and the coeﬃcient on the HSI predictor variable was both positive (β = 2.870) and signiﬁcantly diﬀerent from zero (P = 0.020). Therefore, we considered the HSI model for the yellow-throated warbler both veriﬁed and validated (Tirpak and others 2009a). 172 1.00 CURRENT MODEL USE AND FUTURE DIRECTIONS For species with veriﬁed and validated models, we developed geospatial datasets that summarize the habitat suitability and estimated population size of these species within each subsection for two periods (1992 and 2001). These datasets are being used to assess changes in habitats through time and identify which model variables are associated with these changes. We also are using these datasets as conservation design tools to identify the speciﬁc location and type of management practice that may most eﬀectively increase the habitat quality and population size of target species. Population estimates explicitly tied to habitat suitability are allowing the reﬁnement of landbird population objectives and spatial depiction of these objectives at the ecological subsection scale. We are developing a decisionsupport tool based on these model outputs that will estimate the magnitude of management that may be required to achieve population objectives for a particular species and will assess the simultaneous impacts of diﬀerent management options on populations of multiple species. With conservation informed by these models in both the CH and WGCP, these models are informing the status at the continental scale of species with a signiﬁcant portion of their populations in these BCRs (e.g., Kentucky warbler; Panjabi and others 2005). Adoption and application of these models in other BCRs (the East Gulf Coastal Plain Joint Venture references the use of these models in its Implementation Plan [East Gulf Coastal Plain Joint Venture 2008]) may provide a framework for assessing the status of additional species at the continental scale. However, the use of these models outside the CH and WGCP will require careful scrutiny and additional testing to ensure that the habitat associations remain valid as diﬀerences in forest types among regions (particularly outside the Southeast) likely will aﬀect the SI scores in the landform, forest type, and successional age class matrix derived from Hamel (1992). ACKNOWLEDGMENTS We thank the following reviewers who provided invaluable feedback for improving our HSI models: Eric Baka, Laurel Moore Barnhill, Charles Baxter, Michelle Beck, Jim Bednarz, Jeﬀ Buler, Wes Burger, Dirk Burhans, Bob Cooper, Dean Demarest, Randy Dettmers, Jim Dickson, Therese Donovan, Rob Doster, John Dunning, Jane Fitzgerald, Jim Giocomo, Bill Giuliano, Gypsy Gooding, Fred Guthery, Carola Haas, Tom Haggerty, Paul Hamel, Kirsten Hazler, Larry Hedrick, Mark Howery, Pamela Hunt, Chuck Hunter, James Ingold, Brad Jacobs, Christopher Kellner, Eric Kershner, John Kilgo, Melinda Knutson, Jeﬀ Kopachena, David Krementz, Scott Lanyon, Chester Martin, Dan McAuley, Joe Meyers, Warren Montague, Christopher Moorman, Rua Mordecai, Allan Mueller, Wayne Norling, Brainard Palmer-Ball, David Pashley, Bruce Reid, Rochelle Renken, Amanda Rodewald, Scott Rush, Kristin Schaumburg, Brian Smith, Phil Stauﬀer, Jeﬀrey Stratford, Wayne Thogmartin, Bill Vermillion, Shawchyi Vorisek, R. Montague Whiting, Jr., Mike Wilson, Randy Wilson, and Doug Zollner. 173 LITERATURE CITED Annand, E.M.; Thompson, F.R., III. 1997. Forest bird response to regeneration practices in central hardwood forests. Journal of Wildlife Management. 61(1): 159-171. Applegate, R.D.; Williams, C.K.; Lutz, R.S. 2002. The eﬀect of ﬂooding on northern bobwhites. Western North American Naturalist. 62(2): 227-229. Aquilani, S.M.; Brewer, J.S. 2004. Area and edge eﬀects on forest songbirds in a nonagricultural landscape in Northern Mississippi, USA. Natural Areas Journal. 24(4): 326-335. Artman, V.L.; Downhower, J.F. 2003. Wood thrush (Hylocichla mustelina) nesting ecology in relation to prescribed burning of mixed-oak forest in Ohio. Auk. 120(3): 874-882. Artman, V.L.; Sutherland, E.K.; Downhower, J.F. 2001. Prescribed burning to restore mixed-oak communities in Southern Ohio: eﬀects on breeding-bird populations. Conservation Biology. 15(5): 1423-1434. Bancroft, G.T.; Strong, A.M.; Carrington, M. 1995. Deforestation and its eﬀects on forest-nesting birds in the Florida Keys. Conservation Biology. 9(4): 835-844. Barber, D.R.; Martin, T.E.; Melchiors, M.A.; Thill, R.E.; Wigley, T.B. 2001. Nesting success of birds in diﬀerent silvicultural treatments in southeastern U.S. pine forests. Conservation Biology. 15(1): 196-207. Barber, J.D.; Wiggers, E.P.; Renken, R.B. 1998. Nest-site characteristics and reproductive success of Mississippi kites in the Mississippi River ﬂoodplains. Journal of Wildlife Management. 62(4): 1373-1378. Bay, M.D. 1994. An ecological analysis of breeding bird communities in early successional old ﬁelds: eﬀects of habitat area and vegetation. Fayetteville, AR: University of Arkansas. 114 p. Ph.D. dissertation. Bell, J.L.; Whitmore, R.C. 2000. Bird nesting ecology in a forest defoliated by gypsy moths. Wilson Bulletin. 112(4): 524-531. Best L.B. 1974. Breeding ecology of the ﬁeld sparrow (Spizella pusilla). Urbana, IL: University of Illinois at Urbana-Champaign. 163 p. Ph.D. dissertation. Best, L.B.; Stauﬀer, D.F. 1986. Factors confounding evaluation of bird-habitat relationships. In: Verner, J.; Morrison, M.L.; Ralph, C.J., eds. Wildlife 2000: modeling habitat relationships of terrestrial vertebrates. Madison, WI: University of Wisconsin Press: 209-216. 174 Bisson, I.A.; Stutchbury, B.J.M. 2000. Nesting success and nest-site selection by a Neotropical migrant in a fragmented landscape. Canadian Journal of Zoology. 78(5): 858-863. Blake, J.G. 2005. Eﬀects of prescribed burning on distribution and abundance of birds in a closed-canopy oak-dominated forest, Missouri, USA. Biological Conservation. 121(4): 519-531. Blake, J. G.; Karr, J. R. 1987. Breeding birds of isolated woodlots: area and habitat relationships. Ecology. 68: 1724-1734. Blem, C.R.; Blem, L.B. 1991. Nest-box selection by Prothonotary warblers. Journal of Field Ornithology. 62(3): 299-307. Brawn, J.D.; Tannenbaum, B.; Evans, K.E. 1984. Nest site characteristics of cavity nesting birds in central Missouri. Res. Note NC-314. St. Paul, MN: U.S. Department of Agriculture, Forest Service, North Central Forest Experiment Station. 6 p. Brennan, L.A. 1999. Northern bobwhite (Colinus virginianus). Number 397. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Bridges, A.S.; Peterson, M.J.; Silvy, N.J.; Smeins, F.E.; Wu, X.B. 2001. Diﬀerential inﬂuence of weather on regional quail abundance in Texas. Journal of Wildlife Management. 65(1): 10-18. Brooks, R.P. 1997. Improving habitat suitability index models. Wildlife Society Bulletin. 25(1): 163-167. Brown, B.T. 1993. Bell’s vireo (Vireo bellii). Number 35. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Brown, R.E.; Dickson, J.G. 1994. Swainson’s warbler (Limnothlypis swainsonii). Number 126. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Budnik, J.M.; Ryan, M.R.; Thompson, F.R., III. 2000. Demography of Bell’s vireos in Missouri grassland-shrub habitats. Auk. 117(4): 925-935. Budnik J.M; Thompson, F.R., III; Ryan, M.R. 2002. Eﬀect of habitat characteristics on the probability of parasitism and predation of Bell’s vireo nests. Journal of Wildlife Management. 66(1): 232-239. 175 Buehler, D.M.; Norris, D.R.; Stutchbury, B.J.M.; Kopysh, N.C. 2002. Food supply and parental feeding rates of hooded warblers in forest fragments. Wilson Bulletin. 114(1): 122-127. Bull, E.L.; Jackson, J.E. 1995. Pileated woodpecker (Dryocopus pileatus). Number 148. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Burger, L.W., Jr.; Linder, E.T.; Hamrick, R. 2004. A regional northern bobwhite habitat model for planning and prioritization of habitat restoration in the Central Hardwoods Bird Conservation Region. The Plains, VA: American Bird Conservancy. 39 p. Burhans, D.E.; Thompson, F.R., III. 1999. Habitat patch size and nesting success of yellow-breasted chats. Wilson Bulletin. 111(2): 210-215. Burnham, K.P.; Anderson, D.R. 1998. Model selection and inference: a practical information-theoretic approach. New York, NY: Springer-Verlag. 353 p. Cade, B.S. 1986. Habitat suitability index models: brown thrasher. Biol. Rep. 82 (10.118). Washington, DC: U.S. Department of the Interior, Fish and Wildlife Service. 26 p. Carey, M.; Burhans, D.E.; Nelson, D.A. 1994. Field sparrow (Spizella pusilla). Number 103. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Carrie, N.R. 1996. Swainson’s warblers nesting in early seral pine forests in East Texas. Wilson Bulletin. 108(4): 802-804. Cavitt, J.F.; Haas, C.A. 2000. Brown thrasher (Toxostoma rufum). Number 557. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Cely, J.E.; Sorrow, J.A. 1990. The American swallow-tailed kite in South Carolina. Nongame and Heritage Trust Sect. Pub. No. 1. Columbia, SC: South Carolina Wildlife and Marine Resources Department. 160 p. Cink, C.L. 2002. Whip-poor-will (Caprimulgus vociferus). Number 620. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Cink, C.L.; Collins, C.T. 2002. Chimney swift (Chaetura pelagica). Number 646. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. 176 Collins, S.L.; James, F.C.; Risser, P.G. 1982. Habitat relationships of wood warblers (Parulidae) in northern central Minnesota USA. Oikos. 39(1): 50-58. Conner, R.N.; Adkisson, C.S. 1977. Principal component analysis of woodpecker nesting habitat. Wilson Bulletin. 89(1): 122-129. Conner, R.N.; Dickson, J.G. 1997. Relationships between bird communities and forest age, structure, species composition and fragmentation in the West Gulf Coastal Plain. Texas Journal of Science. 49(3): 123-138. Conner, R.N.; Dickson, J.G.; Locke, B.A.; Segelquist, C.A. 1983. Vegetation characteristics important to common songbirds in east Texas. Wilson Bulletin. 95(3): 349-361. Conner, R.N.; Dickson, J.G.; Williamson, J.H.; Ortega, B. 2004. Width of forest streamside zones and breeding bird abundance in eastern Texas. Southeastern Naturalist. 3(4): 669-682. Conner, R.N.; Jones, S.D.; Jones, G.D. 1994. Snag condition and woodpecker foraging ecology in a bottomland hardwood forest. Wilson Bulletin. 106(2): 242-257. Conner, R.N.; Saenz, D.; Rudolph, D.C. 1995. Fauna using nest boxes in four timber types in eastern Texas. Bulletin of the Texas Ornithological Society. 28(1): 2-6. Coon, R.A.; Williams, B.K.; Lindzey, J.S.; George, J.L. 1982. Examination of woodcock nest sites in central Pennsylvania. In: Dwyer, T.J.; Storm, G.L., eds. Woodcock ecology and management. Wildl. Res. Rep. 14. Washington, DC: U.S. Department of the Interior, Fish and Wildlife Service: 55-62. Cooper, R.J. 1981. Relative abundance of Georgia caprimulgids based on call-counts. Wilson Bulletin. 93(3): 363-371. Coppedge, B.R.; Engle, D.M.; Masters, R.E.; Gregory, M.S. 2001. Avian response to landscape change in fragmented southern Great Plains grasslands. Ecological Applications. 11(1): 47-59. Cram, D.S.; Masters, R.E.; Guthery, F.S.; Engle, D.M.; Montague, W.G. 2002. Northern bobwhite population and habitat response to pine-grassland restoration. Journal of Wildlife Management. 66(4): 1031-1039. Davenport, D.E.; Lancia, R.A.; Walters, J.R.; Doerr, P.D. 2000. Red-cockaded woodpeckers: a relationship between reproductive ﬁtness and habitat in the North Carolina Sandhills. Wildlife Society Bulletin. 28(2): 426-434. 177 Davis, M.A.; Peterson, D.W.; Reich, P.B.; Crozier, M.; Query, T.; Mitchell, E.; Huntington, J.; Bazakas, P. 2000. Restoring savanna using ﬁre: impact on the breeding bird community. Restoration Ecology. 8(1): 30-40. Dettmers, R.; Buehler, D.A.; Franzreb, K.A. 2002. Testing habitat-relationship models for forest birds of the southeastern United States. Journal of Wildlife Management. 66(2): 471-424. DeVault, T.L.; Scott, P.E.; Bajema, R.A.; Lima, S.L. 2002. Breeding bird communities of reclaimed coal mine grasslands in the American Midwest. Journal of Field Ornithology. 73(3): 268-275. Doherty, P.F., Jr.; Grubb, T.C., Jr. 2002. Nest usurpation is an “edge eﬀect” for Carolina chickadees Poecile carolinensis. Journal of Avian Biology. 33(1): 77-82. Donovan, T.M.; Flather, C.H. 2002. Population trends in North American songbirds: interactions between habitat fragmentation and landscape occupancy patterns. Ecological Applications. 12(2): 364-374. Donovan, T.M.; Jones, P.W.; Annand, E.M.; Thompson, F.R, III. 1997. Variation in localscale edge eﬀects: mechanisms and landscape context. Ecology. 78(7): 2064-2075. Donovan, T.M.; Thompson, F.R., III; Faaborg, J.; Probst, F.R. 1995. Reproductive success of migratory birds in habitat sources and sinks. Conservation Biology. 9(6): 13801395. Dornak, L.L.; Burt, D.B.; Coble, D.W.; Conner, R.N. 2004. Relationships between habitat and snag characteristics and the reproductive success of the brown-headed nuthatch (Sitta pusilla) in Eastern Texas. Southeastern Naturalist. 3(4): 683-694. Doster, R.H.; James, D.A. 1998. Home range size and foraging habitat of red-cockaded woodpeckers in the Ouachita Mountains of Arkansas. Wilson Bulletin. 110(1): 110117. Driscoll, M.J.L.; Donovan, T.M. 2004. Landscape context moderates edge eﬀects: nesting success of wood thrushes in central New York. Conservation Biology. 18(5): 13301338. Driscoll, M.J.L.; Donovan, T.; Mickey, R.; Howard, A; Fleming, K.K. 2005. Determinants of wood thrush nest success: a multi-scale, model selection approach. Journal of Wildlife Management. 69(2): 699-709. Duguay, J.P.; Wood, P.B.; Nichols, J.V. 2001. Songbird abundance and avian nest survival rates in forests fragmented by diﬀerent silvicultural treatments. Conservation Biology. 15(5): 1405-1415. 178 Dunford, R.D.; Owen, R.B. 1973. Summer behavior of immature radio-equipped woodcock in central Maine. Journal of Wildlife Management. 37(4): 462-469. Dunning, J.B., Jr.; Borgella, R., Jr.; Clements, K., Meﬀe, G.K. 1995. Patch isolation, corridor eﬀects, and colonization by a resident sparrow in a managed pine woodland. Conservation Biology. 9(3): 542-550. Dunning, J.B., Jr.; Watts, B.D. 1990. Regional diﬀerences in habitat occupancy by Bachman’s sparrow. Auk. 107(3): 463-472. East Gulf Coastal Plain Joint Venture. 2008. Implementation plan, version 1. Auburn, AL: East Gulf Coastal Plain Joint Venture. 32 p. Eckerle, K.P.; Thompson, C.F. 2001. Yellow-breasted chat (Icteria virens). Number 575. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Eddleman, W.R.; Evans, K.E.; Elder, W.H. 1980. Habitat characteristics and management of Swainson’s warbler in southern Illinois. Wildlife Society Bulletin. 8(3): 228-233. Ellison, W.G. 1992. Blue-gray gnatcatcher (Polioptila caerulea). Number 23. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Evans-Ogden, L.J.; Stutchbury, B.J. 1994. Hooded warbler (Wilsonia citrina). Number 110. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Farley, G.H. 1987. Comparative breeding strategies of two coexisting passerines: Bell’s vireo (Vireo bellii) and Bewick’s wren (Thryomanes bewickii). Manhattan, KS: Kansas State University. 52 p. M.S. thesis. Farrand, D.T. 2005. Wildlife response to farm conservation practices in northeast Missouri. Columbia, MO: University of Missouri-Columbia. 237 p. Ph.D. dissertation. Fauth, P.B.; Cabe, P.R. 2005. Reproductive success of Acadian ﬂycatchers in the Blue Ridge Mountains of Virginia. Journal of Field Ornithology. 76(2): 150-157. Fink, A.D.; Thompson, III, F.R.; Tudor, A.A. 2006. Songbird use of regenerating forest, glade, and edge habitat types. Journal of Wildlife Management. 70(1): 180-188. Ford, T.B.; Winslow, D.E.; Whitehead, D.R.; Koukol, M.A. 2001. Reproductive success of forest-dependent songbirds near an agricultural corridor in south-central Indiana. Auk. 118(4): 864-873. 179 Gabbe, A.P.; Robinson, S.K.; Brawn, J.D. 2002. Tree-species preferences of foraging insectivorous birds: implications for ﬂoodplain forest restoration. Conservation Biology. 16(2): 462-470. Gale, G.A.; Hanners, L.A.; Patton, S.R. 1997. Reproductive success of worm-eating warblers in a forested landscape. Conservation Biology. 11(1): 246-250. Galli, A.E.; Leck, C.F.; Forman, R.T.T. 1976. Avian distribution patterns in forest islands of diﬀerent sizes in central New Jersey. Auk. 93(2): 356-364. Giese, C.L.A.; Cuthbert, F.J. 2003. Inﬂuence of surrounding vegetation on woodpecker nest tree selection in oak forest of the Upper Midwest, USA. Forest Ecology and Management. 179(3): 523-534. Gill, F.B. 1980. Historical aspects of hybridization between blue-winged and goldenwinged warblers. Auk. 97(1): 1-18. Gill, F.B.; Canterbury, R.A.; Confer, J.L. 2001. Blue-winged warbler (Vermivora pinus). Number 584. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Giuliano, W.M; Daves, S.E. 2002. Avian response to warm-season grass use in pasture and hayﬁeld management. Biological Conservation. 106(1): 1-9. Gram, W.K.; Porneluzi, P.A.; Clawson, R.L.; Faaborg, J.; Richter, S.C. 2003. Eﬀects of experimental forest management on density and nesting success of bird species in Missouri Ozark forests. Conservation Biology. 17(5): 1324-1337. Grand, J.; Cushman, S.A. 2003. A multi-scale analysis of species-environment relationships: breeding birds in a pitch pine-scrub oak (Pinus rigida-Quercus ilicifolia) community. Biological Conservation. 112(3): 307-317. Graves, G.R. 2001. Factors governing the distribution of Swainson’s warbler along a hydrological gradient in Great Dismal Swamp. Auk. 118(3): 650-664. Graves, G.R. 2002. Habitat characteristics in the core breeding range of the Swainson’s warbler. Wilson Bulletin. 114(2): 210-220. Graves, G.R. 2004. Avian commensals in colonial America: when did Chaetura pelagica become the chimney swift? Archives of Natural History. 31(2): 300-307. Gregg, I.D.; Wood, P.B.; Samuel, D.E. 2000. American woodcock use of reclaimed surface mines in West Virginia. In: McAuley, D.G.; Bruggink, J.G.; Sepik, G., eds. Proceedings of the ninth American woodcock symposium. Info. and Technol. Rep. USGS/BRD/ITR2000-0009. Laurel, MD: U.S. Geological Survey, Biological Resources Division, Patuxent Wildlife Research Center: 9-22. 180 Groom, J.D.; Grubb, T.C., Jr. 2002. Bird species associated with riparian woodland in fragmented, temperate-deciduous forest. Conservation Biology. 16(3): 832-836. Guthery, F.S. 1997. A philosophy of habitat management for northern bobwhites. Journal of Wildlife Management. 61(2): 291-301. Guthery, F.S. 1999. Slack in the conﬁguration of habitat patched for northern bobwhites. Journal of Wildlife Management. 63(1): 245-250. Guthery, F.S. 2000. On bobwhites. College Station, TX: Texas A&M University Press. 213 p. Guthery, F.S.; Green, M.C.; Masters, R.E.; Demaso, S.J.; Wilson, H.M.; Steubing, F.B. 2001. Land cover and bobwhite abundance on Oklahoma farms and ranches. Journal of Wildlife Management. 65(4): 838-849. Haas, C.A. 1997. What characteristics of shelterbelts are important to breeding success and return rate of birds? American Midland Naturalist. 137(2): 225-238. Haggerty, T. M. 1998. Vegetation structure of Bachman’s sparrow breeding habitat and its relationship to hone range. Journal of Field Ornithology. 69(1): 45-50. Hall, G.A. 1996. Yellow-throated warbler (Dendroica dominica). Number 223. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Hamel, P.B. 1992. Land manager’s guide to the birds of the South. Chapel Hill, NC: Nature Conservancy, Southeastern Region. 437 p. Hamel, P.B. 2000. Cerulean warbler (Dendroica cerulea). Number 511. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Hanners, L.A.; Patton, S.R. 1998. Worm-eating warbler (Helmitheros vermivorus). Number 367. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Hayden, T.J.; Faaborg, J.; Clawson, R.L. 1985. Estimates of minimum area requirements for Missouri forest birds. Transactions, Missouri Academy of Science. 19: 11-22. Hazler, K.R. 1999. An assessment of pine plantations as breeding habitat for Acadian ﬂycatchers and hooded warblers. Raleigh, NC: North Carolina State University. 91 p. M.S. thesis. 181 Heltzel, J.M.; Leberg, P.L. 2006. Eﬀects of selective logging on breeding bird communities in bottomland hardwood forests in Louisiana. Journal of Wildlife Management. 70(5): 1416-1424. Heppell, S.S.; Walters, J.R.; Crowder, L.B. 1994. Evaluating management alternatives for red-cockaded woodpeckers: a modeling approach. Journal of Wildlife Management. 58(3): 479-488. Holmes, S.B.; Burke, D.M.; Elliott, K.A.; Cadman, M.D.; Friesen, L. 2004. Partial cutting of woodlots in an agriculture-dominated landscape: eﬀects on forest bird communities. Canadian Journal of Forest Research. 34(12): 2467-2476. Hopp, S.L.; Kirby, A.; Boone, C.A. 1995. White-eyed vireo (Vireo griseus). Number 168. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Horn, D.J.; Koford, R.R.; Braland, M.L. 2002. Eﬀects of ﬁeld size and landscape composition on grassland birds in south-central Iowa. Journal of the Iowa Academy of Science. 109(1): 1-7. Hoover, J.P.; Brittingham, M.C. 1998. Nest-site selection and nesting success of wood thrushes. Wilson Bulletin. 110(3): 375-383. Hoover, J.P.; Brittingham, M.C.; Goodrich, L.J. 1995. Eﬀects of forest patch size on nesting success of wood thrushes. Auk. 112(1): 146-155. Hudgins, J.E.; Storm, G.L.; Wakeley, J.S. 1985. Local movements and diurnal habitat selection by male American woodcock in Pennsylvania. Journal of Wildlife Management. 49(3): 614-619. Hughes, J.M. 1999. Yellow-billed cuckoo (Coccyzus americanus). Number 418. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Hyams, D.G. 2001. CurveExpert 1.3. [online]. Available: http://curveexpert.webhop.biz/. Jackson, J.A. 1994. Red-cockaded woodpecker (Picoides borealis). Number 85. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. James, F.C. 1971. Ordinations of habitat relationships among breeding birds. Wilson Bulletin. 83(3): 215-236. James, F.C.; Hess, C.A.; Kicklighter, B.C.; Thum, R.A. 2001. Ecosystem management and the niche gestalt of the red-cockaded woodpecker in longleaf pine forests. Ecological Applications 11(3): 854-870. 182 James, R.D. 1976. Foraging behavior and habitat selection of three species of vireos in southern Ontario. Wilson Bulletin. 88(1): 62-75. Jones, J.; DeBruyn, R.D.; Barg, J.J.; Robertson, R.J. 2001. Assessing the eﬀects of natural disturbance on a Neotropical migrant songbird. Ecology. 82(9): 2628-2635. Jones, J.; Robertson, R.J. 2001. Territory and nest-site selection of cerulean warblers in eastern Ontario. Auk. 118(3): 727-735. Jones-Farrand, D.T.; Millspaugh J. J. 2006. Eﬀects of scale and land cover map resolution on estimating bobwhite habitat quality in northeast Missouri. Transactions, Missouri Academy of Science. 40: 53-66. Kahl, R.B.; Baskett, T.S.; Ellis, J.A.; Burroughs, J.N. 1985. Characteristics of summer habitats of selected nongame birds in Missouri. Res. Bull. 1056. Columbia, MO: University of Missouri-Columbia. 155 p. Kelley, J.R., Jr. 2003. American woodcock population status, 2003. Laurel, MD: U.S. Department of the Interior, Fish and Wildlife Service. 20 p. Kelly, J.F.; Pletschet, S.M.; Leslie, D.M., Jr. 1993. Habitat associations of red-cockaded woodpecker cavity trees in an old-growth forest of Oklahoma. Journal of Wildlife Management. 57(1): 122-128. Kennedy, E.D.; White, D.W. 1996. Interference competition from house wrens as a factor in the decline of Bewick’s wrens. Conservation Biology. 10(1): 281-284. Kennedy, E.D.; White, D.W. 1997. Bewick’s wren (Thryomanes bewickii). Number 315. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Keppie, D.M.; Whiting, R.M., Jr. 1994. American woodcock (Scolopax minor). Number 100. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Kershner, E.L.; Bollinger, E.K.; Helton, M.N. 2001. Nest-site selection and renesting in the blue-gray gnatcatcher (Polioptila caerulea). American Midland Naturalist. 146(2): 404-413. Kilgo, J.C.; Sargent, R.A.; Chapman, B.R.; Miller, K.V. 1998. Eﬀect of stand width and adjacent habitat on breeding bird communities in bottomland hardwoods. Journal of Wildlife Management. 62(1): 72-83. Kilgo, J.C.; Sargent, R.A.; Miller, K.V.; Chapman, B.R. 1996. Nest sites of Kentucky warblers in bottomland hardwoods of South Carolina. Journal of Field Ornithology. 67(2): 300-306. 183 King, R.S.; Brashear, K.E.; Reiman, M. 2007. Red-headed woodpecker nest-habitat thresholds in restored savannas. Journal of Wildlife Management. 71(1): 30-35. Klaus, N.A.; Buehler, D.A.; Saxton, A.M. 2005. Forest management alternatives and songbird breeding habitat on the Cherokee National Forest, Tennessee. Journal of Wildlife Management. 69(1): 222-234. Klute, D.S.; Lovallo, M.J.; Tzilkowski, W.M.; Storm, G.L. 2000. Determining multiscale habitat and landscape associations for American woodcock in Pennsylvania. In: McAuley, D.G.; Bruggink, J.G.; Sepik, G., eds. Proceedings of the ninth American woodcock symposium. Info. and Technol. Rep. USGS/BRD/ITR-2000-0009. Laurel, MD: U.S. Geological Survey, Biological Resources Division, Patuxent Wildlife Research Center: 42-49. Klute, D.S.; Lovallo, M.J.; Tzilkowski, W.M. 2002. Autologistic regression modeling of American woodcock habitat use with spatially dependent data. In: Scott, J.M.; Heglund, P.J.; Morrison, M.L., eds. Predicting species occurrences: issues of accuracy and scale. Washington, DC: Island Press: 335-343. Knutson, M.G. 1995. Birds of large ﬂoodplain forests: local and regional habitat associations on the Upper Mississippi River. Ames, IA: Iowa State University. 121 p. Ph.D. dissertation. Knutson, M.G.; McColl, L.E.; Suarez, S.A. 2005. Breeding bird assemblages associated with stages of forest succession in large river ﬂoodplains. Natural Areas Journal. 25(1): 55-70. Knutson, M.G.; Niemi, G.J.; Newton, W.E.; Friberg, M.A. 2004. Avian nest success in Midwestern forests fragmented by agriculture. Condor. 106(1): 116-130. Kopachena, J.G.; Crist, C.J. 2000a. Macro-habitat features associated with painted and indigo buntings in northeast Texas. Wilson Bulletin. 112(1): 108-114. Kopachena, J.G.; Crist, C.J. 2000b. Microhabitat features associated with the song perches of painted and indigo buntings (Passeriformes:Cardinalidae) in northeast Texas. Texas Journal of Science. 52(2): 133-144. Kricher, J.C. 1995. Black-and-white warbler (Mniotilta varia). Number 158. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Kroodsma, R.L. 1984. Eﬀect of edge on breeding forest bird species. Wilson Bulletin. 96(3): 426-436. Lanyon, S.M.; Thompson, C.F. 1986. Site ﬁdelity and habitat quality as determinants of settlement pattern in male painted buntings. Condor. 88(2): 206-210. 184 Lanyon, W.E. 1997. Great crested ﬂycatcher (Myiarchus crinitus). Number 300. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Larson M.A.; Dijak, W.D.; Thompson, F.R., III, Millspaugh, J.J. 2003. Landscape-level habitat suitability models for twelve species in southern Missouri. Gen. Tech. Rep. NC-233. St. Paul, MN: U.S. Department of Agriculture, Forest Service, North Central Research Station. 51 p. Liu, J.; Dunning, J.B., Jr.; Pulliam, H.R. 1995. Potential eﬀects of a forest management plan on Bachman’s sparrows (Aimophila aestivalis): linking a spatially explicit model with GIS. Conservation Biology. 9(1): 62-75. Lohr, S.M.; Gauthreaux, S.A.; Kilgo, J.C. 2002. Importance of coarse woody debris to avian communities in loblolly pine forests. Conservation Biology. 16(3): 767-777. Lowther, P.E.; Lanyon, S.M., Thompson, C.W. 1999. Painted Bunting (Passerina ciris). Number 398. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Lukomski, L. 2003. Eﬀect of land-cover change on red-headed woodpecker populations at high vs. low abundance range locations. East Lansing, MI: Michigan State University. 100 p. M.S. thesis. Lusk, J.J.; Guthery, F.S.; DeMaso, S.J. 2002b. A neural network model for predicting northern bobwhite abundance in the Rolling Red Plains of Oklahoma. In: Scott, J.M.; Heglund, P.J.; Morrison, M.L., eds. Predicting species occurrences: issues of accuracy and scale. Washington, DC: Island Press: 345-356. Lusk, J.J.; Guthery, F.S.; George, R.R.; Peterson, M.J.; DeMaso, S.J. 2002a. Bobwhite abundance in relation to weather and land use. Journal of Wildlife Management. 66(4): 1040-1051. Lynch, J.F.; Whigham, D.F. 1984. Eﬀects of forest fragmentation on breeding bird communities in Maryland, U.S.A. Biological Conservation. 38(3): 287-324. McAuley, D.G.; Longcore, J.R.; Sepik, G.F.; Pendleton, G.W. 1996. Habitat characteristics of American woodcock nest sites on a managed area in Maine. Journal of Wildlife Management. 60(1): 138-148. McCarty, J.P. 1996. Eastern wood-pewee (Contopus virens). Number 245. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. 185 McComb, W.C.; Bonney, S.A.; Sheﬃeld, R.M.; Cost, N.D. 1986. Snag resources in Florida: are they suﬃcient for average populations of primary cavity-nesters? Wildlife Society Bulletin. 14(1): 40-48. McCoy, T.D.; Ryan, M.R.; Kurzejeski, E.W.; Burger, L.W., Jr. 1999. Conservation reserve program: source or sink habitat for grassland birds in Missouri? Journal of Wildlife Management. 63(2): 530-538. McDonald, M.V. 1998. Kentucky warbler (Oporornis formosus). Number 324. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. McShea, W.J.; McDonald, M.V.; Morton, E.S.; Meier, R.; Rappole, J.H. 1995. Long-term trends in habitat selection by Kentucky warblers. Auk. 112(2): 375-381. Meyer, K.D. 1995. Swallow-tailed kite (Elanoides forﬁcatus). Number 138. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Mitchell, W.A. 1998. Species proﬁle: Bachman’s sparrow (Aimophila aestivalis) on military installations in the southeastern United States. Tech. Rep. SERDP-98-11. Vicksburg, MS: U.S. Army Corps of Engineers. 30 p. Moldenhauer, R.R.; Regelski, D.J. 1996. Northern parula (Parula americana). Number 215. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Moorman, C.E.; Guynn, D.C., Jr. 2001. Eﬀects of group-selection opening size on breeding bird habitat use in a bottomland forest. Ecological Applications. 11(6): 16801691. Moorman, C.E.; Guynn, D.C., Jr.; Kilgo, J.C. 2002. Hooded warbler nesting success adjacent to group-selection and clearcut edges in a southeastern bottomland forest. Condor. 104(2): 366-377. Morgan, K.; Freedman, B. 1986. Breeding bird communities in a hardwood forest succession in Nova Scotia. Canadian Field-Naturalist. 100(4): 506-519. Morgenweck, R.O. 1977. Diurnal high use areas of hatching-year female American woodcock. In: Keppie, D.M.; Owen, R.B., Jr., eds. Proceedings of the sixth woodcock symposium. Fredericton, NB: New Brunswick Department of Natural Resources: 155160. Morse, S.F.; Robinson, S.K. 1999. Nesting success of a neotropical migrant in a multipleuse, forested landscape. Conservation Biology. 13(2): 327-337. 186 Mostrom, A.M.; Curry, R.L.; Lohr, B. 2002. Carolina chickadee (Poecile carolinensis). Number 636. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Murphy, D.W.; Thompson, F.R., III. 1993. Breeding chronology and habitat of the American woodcock in Missouri. In: Longcore, J.R.; Sepik, G.F., eds. Proceedings of the eighth woodcock symposium. Biol. Rep. 16. Washington, DC: U.S. Department of the Interior, Fish and Wildlife Service:12-18. Nolan, V., Jr.; Ketterson, E.D.; Buerkle, C.A. 1999. Prairie warbler (Dendroica discolor). Number 455. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Norris, D.J.; Elder, W.H. 1982. Distribution and habitat characteristics of the painted bunting in Missouri. Transactions, Missouri Academy of Science. 16:77-83. Norris, D.R.; Pitcher, T.E.; Stutchbury, B.J.M. 2000. The spatial response of male Hooded Warblers to edges in isolated fragments. Condor. 102(3): 595-600. Norris, D.R.; Stutchbury, B.J.M. 2002. Sexual diﬀerences in gap crossing ability of a forest songbird revealed through radiotracking. Auk. 119(2): 528-532. O’Halloran, K.A.; Conner, R.N. 1987. Habitat use by brown-headed nuthatches. Bulletin of the Texas Ornithological Society. 20 :7-13. Palmer, B.; Wellendorf, S. 2006. Timber and quail density. Tall Timbers Research Station & Land Conservancy Research Notes. 5(5): 1. Panjabi, A.; Beardmore, C.; Blancher, P; Butcher, G.; Carter, M.; Demarest, D.; Dunn, E.; Hunter, C.; Pashley, D.; Rosenberg, K.; Rich, T.; Will, T. 2001. The Partners in Flight handbook on species assessment and prioritization. Version 1.1. Brighton, CO: Rocky Mountain Bird Observatory. 25 p. Panjabi, A.O.; Dunn, E.H.; Blancher, P.J.; Hunter, W.C.; Altman, B.; Bart, J.; Beardmore, C.J.; Berlanga, H.; Butcher, G.S.; Davis, S.K.; Demarest, D.W.; Dettmers, R.; Easton, W.; Gomez de Silva Garza, H.; Iñigo-Elias, E.E.; Pashley, D.N.; Ralph, C.J.; Rich, T.D.; Rosenberg, K.V.; Rustay, C.M.; Ruth, J.M.; Wendt, J.S.; Will, T.C. 2005. The Partners in Flight handbook on species assessment. Version 2005. Partners in Flight Tech. Ser. No. 3. Brighton, CO: Rocky Mountain Bird Observatory. 29 p. Parker, J.W. 1999. Mississippi kite (Ictinia mississippiensis). Number 402. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Parker, T.H.; Stansberry, B.M.; Becker, C.D.; Gipson, P.S. 2005. Edge and area eﬀects on the occurrence of migrant forest songbirds. Conservation Biology. 19(4): 1157-1167. 187 Peak, R.G.; Thompson, F.R., III; Shaﬀer, T. 2004. Factors aﬀecting songbird nest survival in riparian forests in a Midwestern agricultural landscape. Auk. 121(3): 726-737. Perkins, M.W.; Johnson, R.J.; Blankenship, E.E. 2003b. Response of riparian avifauna to percentage and pattern of woody cover in an agricultural landscape. Wildlife Society Bulletin. 31(3): 642-660. Perkins, D.W.; Vickery, P.D.; Shriver., W.G. 2003a. Spatial dynamics of source-sink habitats: eﬀects on rare grassland birds. Journal of Wildlife Management. 67(3): 588599. Petit, L.J. 1999. Prothonotary warbler (Protonotaria citrea). Number 408. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Plentovich, S.; Tucker, J.W., Jr.; Holler, N.R.; Hill, G.E. 1998. Enhancing Bachman’s sparrow habitat via management of red-cockaded woodpeckers. Journal of Wildlife Management. 62(1): 347-354. Pogue, D.W.; Schnell, G.D. 1994. Wilson Bullet Habitat characterization of secondary cavity-nesting birds in Oklahoma. in. 106(2): 203-226. Powell, B.F.; Steidl, R.J. 2000. Nesting habitat and reproductive success of southwestern riparian birds. Condor. 102(4): 823-831 Prather, J.W.; Smith, K.G. 2003. Eﬀects of tornado damage on forest bird populations in the Arkansas Ozarks. Southwestern Naturalist. 48(2): 292-297. Prosser, D.J.; Brooks, R.P. 1998. A veriﬁed habitat suitability index for the Louisiana waterthrush. Journal of Field Ornithology. 69(2): 288-298. Provencher, L.; Gobris, N.M.; Brennan, L.A.; Gordon, D.R.; Hardesty, J.L. 2002. Breeding bird response to midstory hardwood reduction in Florida sandhill longleaf pine forests. Journal of Wildlife Management. 66(3): 641-661. Pyle, P.; DeSante, D.F. 2003. Four-letter and six-letter alpha codes for birds recorded from the American Ornithologists’ Union check-list area. North American BirdBander 28: 64-79. Rabe, D.L.; Prince, H.H.; Beaver, D.L. 1983. Feeding-site selection and foraging strategies of American woodcock. Auk. 100(3): 711-716. Reese, J.G. 1996. Whip-poor-will. In: Robbins, C.S., ed. Atlas of the breeding birds of Maryland and the District of Columbia. Pittsburgh, PA: University of Pittsburgh Press: 194-195 188 Renken, R.B.; Wiggers, E.P. 1989. Forest characteristics related to pileated woodpecker territory size in Missouri. Condor. 91(3): 642-652. Renken, R.B.; Wiggers, E.P. 1993. Habitat characteristics related to pileated woodpecker densities in Missouri. Wilson Bulletin. 105(1): 77-83. Rich, T.D.; Beardmore, C.J.; Berlanga, H.; Blancher, P.J.; Bradstreet, M.S.W; Butcher, G.S.; Demarest, D.W.; Dunn, E.H.; Hunter, W.C.; Inigo-Elias, E.E.; Kennedy, J.A.; Martell, A.M.; Panjabi, A.O.; Pashley, D.N.; Rosenberg, K.V.; Rustay, C.M.; Wendt, J.S.; Will, T.C. 2004. Partners in Flight North American landbird conservation plan. Ithaca, NY: Cornell Lab of Ornithology. 84 p. Robbins, C.S.; Dawson, D.K.; Dowell, B.A. 1989. Habitat area requirements of breeding forest birds of the Middle Atlantic states. Wildlife Monographs 103. 34 p. Robinson, S.K.; Thompson, F.R., III; Donovan, T.M.; Whitehead, D.R.; Faaborg, J. 1995. Regional forest fragmentation and the nesting success of migratory birds. Science. 267(5206): 1987-1990. Robinson, S.K.; Robinson, W.D. 2001. Avian nesting success in a selectively harvested north temperate deciduous forest. Conservation Biology. 15(6): 1763-1771. Robinson, W.D. 1995. Louisiana waterthrush (Seiurus motacilla). Number 151. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Rodewald, A.D. 2003. The decline of oak forests and implications for forest wildlife conservation. Natural Areas Journal. 23(4): 368-371. Rodewald, A.D.; Vitz, A.C. 2005. Edge- and area-sensitivity of shrubland birds. Journal of Wildlife Management. 69(2): 681-688. Rodewald, P.G.; James, R.D. 1996. Yellow-throated vireo (Vireo ﬂavifrons). Number 247. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Roseberry, J.L.; Sudkamp, S.C. 1998. Assessing the suitability of landscapes for northern bobwhite. Journal of Wildlife Management. 62(3): 895-902. Rosenberg, K.V.; Barker, S.E.; Rohrbaugh, R.W. 2000. An atlas of cerulean warbler populations: ﬁnal report to U.S. Fish and Wildlife Service: 1997-2000 breeding seasons. Ithaca, NY: Cornell Lab of Ornithology. 56 p. Rosenberg, K.V.; Blancher, P.J. 2005. Setting numerical population objectives for priority landbird species. In: Ralph, C.J.; Rich, T.D., eds. Bird conservation implementation and integration in the Americas: proceedings of the third international Partners in Flight conference, Volume 1; 2002 March 20-24; Asilomar, CA. Gen. Tech. 189 Rep. PSW-GTR-191. Albany, CA: U.S. Department of Agriculture, Forest Service, Paciﬁc Southwest Research Station: 57-67. Rosenberg, K.V.; Rohrbaugh, R.W.; Barker, S.E.; Lowe, J.D.; Hames, R.S.; Dhondt, A.A. 1999. A land manager’s guide to improving habitat for scarlet tanagers and other forest-interior birds. Ithaca, NY: Cornell Lab of Ornithology. 23 p. Roth, R.R.; Johnson, M.S.; Underwood, T.J. 1996. Wood thrush (Hylocichla mustelina). Number 246. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Rudolph, D.C.; Conner, R.N.; Schaeﬀer, R.R. 2002. Red-cockaded woodpecker foraging behavior in relation to midstory vegetation. Wilson Bulletin. 114(2): 235-242. Rumble, M.A.; Gobeille, J.E. 2004. Avian use of successional cottonwood (Populus deltoides) woodlands along the middle Missouri River. American Midland Naturalist. 152(1): 165-177. Sargent, R.A.; Kilgo, J.C.; Chapman, B.R.; Miller, K.V. 2003. Nesting ecology of wood thrush (Turdidae: Passeriformes) in hardwood forests of South Carolina. Southeastern Naturalist. 2(2): 217-222. Sauer, J.R.; Hines, J.E.; Fallon, J. 2005. The North American Breeding Bird Survey, results and analysis 1966-2005. Version 6.2.2006. [online]. Available: http://www. mbr-pwrc.usgs.gov/bbs/bbs.html. Savignac, C.; Desrochers, A.; Huot, J. 2000. Habitat use by pileated woodpeckers at two spatial scales in eastern Canada. Canadian Journal of Zoology. 78(2): 219-225. Schamberger, M.A.; Farmer, A.H.; Terrel, J.W. 1982. Habitat suitability index models: introduction. FWSI-OBS-82/10. Fort Collins, CO: U.S. Department of the Interior, Fish and Wildlife Service. 2 p. Scharf, W.C.; Kren, J. 1996. Orchard oriole (Icterus spurius). Number 255. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Schroeder, R.L. 1982. Habitat suitability index models: pileated woodpecker. FWS/OBS82/10.39. Washington, DC: U.S. Department of Interior, Fish and Wildlife Service. 15 p. Schroeder, R.L. 1985. Habitat suitability index models: northern bobwhite. Biol. Rep. 82 (10.104). Washington, DC: U.S. Department of the Interior, Fish and Wildlife Service. 32 p. 190 Sedgwick, J.A.; Knopf, F.L. 1990. Cavity-nesting birds in cottonwood ﬂoodplains: habitat relationships and nest site characteristics. Journal of Wildlife Management. 54(1): 112-124. Sepik, G.F.; Derleth, E.L. 1993. Habitat use, home range size, and patterns of moves of the American woodcock in Maine. In: Longcore, J.R.; Sepik, G.F., eds. Proceedings of the eighth woodcock symposium. Biol. Rep. 16. Washington, DC: U.S. Department of the Interior, Fish and Wildlife Service: 41-49. Shackelford, C.E.; Conner, R.N. 1997. Woodpecker abundance and habitat use in three forest types in eastern Texas. Wilson Bulletin. 109(4): 614-629. Sheﬃeld, R.M. 1981. Multiresource inventories: techniques for evaluating nongame bird habitat. Res. Pap. SE-218. Asheville, NC: U.S. Department of Agriculture, Forest Service, Southeastern Forest Experiment Station. 28 p. Small, M.F.; Hunter, M.L., Jr. 1989. Responses of passerines to abrupt forest-river and forest-powerline edges in Maine. Wilson Bulletin. 101(1): 77-83. Showalter, C.R.; Whitmore, R.C. 2002. The eﬀect of gypsy moth defoliation on cavitynesting bird communities. Forest Science. 48(2): 273-281. Smith, K.G.; Withgott, J.H.; Rodewald, P.G. 2000. Red-headed woodpecker (Melanerpes erythrocephalus). Number 518. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Somershoe, S.G.; Hudman, S.P.; Chandler, C.R. 2003. Habitat use by Swainson’s warblers in a managed bottomland forest. Wilson Bulletin. 115(2): 148-154. Sousa, P.J. 1983. Habitat suitability index models: ﬁeld sparrow. FWS/OBS-82/10. Washington, DC: U.S. Department of the Interior, Fish and Wildlife Service. 14 p. Sprankle, K.E.; Sepik, G.F.; McAuley, D.G.; Longcore, J.R. 2000. Assessing habitat selection in spring by male American woodcock in Maine with a geographic information system. In: McAuley, D.G.; Bruggink, J.G.; Sepik, G., eds. Proceedings of the ninth American woodcock symposium. Info. and Technol. Rep. USGS/BRD/ITR2000-0009. Laurel, MD: U.S. Geological Survey, Biological Resources Division, Patuxent Wildlife Research Center: 117. Steketee, A.K. 2000. Predicting habitat suitability for American woodcock and landscape-level assessment of habitat in West Virginia. Morgantown, WV: West Virginia University. 123 p. Ph.D. dissertation. 191 Stober, J. 1996. Territory dynamics and basic biology of the Bachman’s sparrow (Aimophila aestivalis) at the Savannah River Site, South Carolina. Athens, GA: University of Georgia. 108 p. M.S. thesis. Stoddard, H.L. 1931. The bobwhite quail: its life history and management. New York: Charles Scribner’s Sons. 559 p. Storm, G.L.; Lovallo, M.J.; Klute, D.S.; Tzilkowski, W.M.; Delong, T. 1995. Predicting American woodcock presence in Pennsylvania from local and landscape-scale habitat variables. Northeast Wildlife. 52:39-48. Straight, C.A.; Cooper, R.A. 2000. Chuck-will’s-widow (Caprimulgus carolinensis). Number 499. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Straw, J.A, Jr.; Wakeley, J.S.; Hudgins, J.E. 1986. A model for management of diurnal habitat for American woodcock in Pennsylvania. Journal of Wildlife Management. 50(3): 378-383. Sykes, P.W., Jr.; Kepler, C.B.; Litzenberger, K.L.; Sansing, H.R.; Lewis, E.T.R.; Hatﬁeld, J.S. 1999. Density and habitat of breeding swallow-tailed kites in the lower Suwanee ecosystem, Florida. Journal of Field Ornithology. 70(3): 321-336. Taylor, J.S.; Church, K.E.; Rusch, D.H.; Cary, J.R. 1999. Macrohabitat eﬀects on summer survival, movements, and clutch success of northern bobwhite in Kansas. Journal of Wildlife Management. 63(2): 675-685. Thompson, F.R., III; Dijak, W.D.; Kulowiec, T.G.; Hamilton, D.A. 1992. Breeding bird populations in Missouri Ozark forests with and without clearcutting. Journal of Wildlife Management. 56(1): 23-30. Tilghman, N.G. 1987. Characteristics of urban woodlands aﬀecting breeding bird diversity and abundance. Landscape and Urban Planning. 14: 481-495. Tirpak, J.M.; Jones-Farrand, D.T.; Thompson, F.R., III; Twedt, D.J.; Baxter, C.K.; Fitzgerald, J.A.; Uihlein, W.B., III. 2009a. Assessing ecoregional-scale Habitat Suitability Index models for priority landbirds. Journal of Wildlife Management. 73. Tirpak, J.M.; Jones-Farrand, D.T.; Thompson, F.R., III; Twedt, D.J.; Nelson, M.D.; Uihlein, W.B., III. 2009b. Predicting bird habitat quality from a geospatial analysis of FIA data. In: McRoberts, R.E.; Reams, G.A.; Van Deusen, P.C.; McWilliams, W.H., eds. Proceedings, 8th annual Forest Inventory and Analysis symposium; 2006 October 16-19; Monterey, CA. Gen. Tech. Rep. WO-79. Washington, DC: U.S. Department of Agriculture, Forest Service. 192 Torres, A.R.; Leberg, P.L. 1996. Initial changes in habitat and abundance of cavitynesting birds and the Northern parula following Hurricane Andrew. Condor. 98(3): 483-490. Townsend, D.E.; Masters, R.E.; Lochmiller, R.L.; Leslie, D.M., Jr.; DeMaso, S.J.; Peoples, A.D. 2001. Characteristics of nest sites of northern bobwhites in western Oklahoma. Journal of Range Management. 54(3): 260-264. Tucker, J.W., Jr.; Robinson, W.D.; James, J.B. 2004. Inﬂuence of ﬁre on Bachman’s sparrow, an endemic North American songbird. Journal of Wildlife Management. 68(4): 1114-1123. Twedt, D.J.; Wilson, R.R.; Henne-Kerr, J.L.; Grosshuesch, D.A. 2002. Avian response to bottomland hardwood reforestation: the ﬁrst 10 years. Restoration Ecology. 10(4): 645-655. Twedt, D.J.; Wilson, R.R.; Henne-Kerr, J.L.; Hamilton, R.B. 1999. Impact of forest type and management strategy on avian densities in the Mississippi alluvial valley. Forest Ecology. and Management. 123(2-3): 261-274. Twedt, D.J.; Wilson, R.R.; Henne-Kerr, J.L.; Hamilton, R.B. 2001. Nest survival of forest birds in the Mississippi Alluvial Valley. Journal of Wildlife Management. 65(3): 450460. Underwood, T.J.; Sealy, S.G.; McLaren, C.M. 2004. Eastern wood-pewees as brownheaded cowbird hosts: accepters but infrequently parasitized. Journal of Field Ornithology. 75(2): 165-171. U.S. Department of the Interior, Fish and Wildlife Service. 1981. Standards for the development of habitat suitability index models. Ecological Services Manual 103. Washington, DC: U.S. Department of the Interior,, Fish and Wildlife Service, Division of Ecological Services. 170 p. U.S. Department of the Interior, Fish and Wildlife Service. 2002. Birds of conservation concern 2002. Arlington, VA: U.S. Department of the Interior,, Fish and Wildlife Service, Division of Migratory Bird Management. 99 p. U.S. Department of the Interior, Fish and Wildlife Service. 2003. Recovery plan for the red-cockaded woodpecker (Picoides borealis): second revision. Atlanta, GA: U.S. Department of the Interior, Fish and Wildlife Service. 296 p. Villard, M-A; Schmidt, E.V.; Maurer, B.A. 1998. Contribution of spatial modeling to avian conservation. In: Marzluﬀ, J.M.; Sallabanks, R., eds. Avian conservation: research and management. Washington, DC: Island Press. 563 p. 193 Wakeley, J.S.; Roberts, T.H. 1996. Bird distributions and forest zonation in a bottomland hardwood wetland. Wetlands. 16(3): 296-308. Walk, J.W.; Warner, R.E. 2000. Grassland management for the conservation of songbirds in the Midwestern USA. Biological Conservation. 94(2): 165-172. Walkinshaw, L.H. 1953. Life-history of the prothonotary warbler. Wilson Bulletin. 65(3): 152-168. Wallendorf, M.J.; Poneluzi, P.A.; Gram, W.K.; Clawson, R.L.; Faaborg, J. 2007. Bird response to clear cutting in Missouri Ozark forests. Journal of Wildlife Management. 71(6): 1899-1905. Walters, J.R.; Daniels, S.J.; Carter, J.H.; Doerr, P.D. 2002. Deﬁning quality of redcockaded woodpecker foraging habitat based on habitat use and ﬁtness. Journal of Wildlife Management. 66(4): 1064-1082. Weakland, C.A.; Wood, P.B. 2005. Cerulean warbler (Dendroica cerulea) microhabitat and landscape-level habitat characteristics in southern West Virginia. Auk. 122(2): 497-508. Weinberg, H.J.; Roth, R.R. 1998. Forest area and habitat quality for nesting wood thrushes. Auk. 115(4): 879-889. Wenny, D.G.; Clawson, R.L.; Faaborg, J.; Sheriﬀ, S.L. 1993. Population density, habitat selection and minimum area requirements of three forest-interior warblers in central Missouri. Condor. 95(4): 968-979. White, C.G.; Schweitzer, S.H.; Moore, C.T.; Parnell, I.B.; Lewis-Weis, L.A. 2005. Evaluation of the landscape surrounding northern bobwhite nest sites: a multiscale analysis. Journal of Wildlife Management. 69(4): 1528-1537. Whitehead, D.R.; Taylor, T. 2002. Acadian ﬂycatcher (Empidonax virescens). Number 614. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Whittam, R.M.; McCracken, J.D.; Francis, C.F.; Gartshore, M.E. 2002. The eﬀects of selective logging on nest-site selection and productivity of hooded warblers (Wilsonia citrina) in Canada. Canadian Journal of Zoology. 80(4): 644-654. Williams, C.K.; Lutz, R.S.; Applegate, R.D. 2004. Winter survival and additive harvest in Kansas northern bobwhite coveys. Journal of Wildlife Management. 68(1): 94-100. Williams, C.K.; Lutz, R.S.; Applegate, R.D.; Rusch, D.H. 2000. Habitat use and survival of northern bobwhite in Kansas cropland and rangeland ecosystems during the harvest season. Canadian Journal of Zoology 78(9): 1562-1566. 194 Wilson, C.W.; Masters, R. E.; Bukenhofer, G.A. 1995. Breeding bird response to pinegrassland community restoration for red-cockaded woodpeckers. Journal of Wildlife Management. 59(1): 56-67. Wilson, J.K. 1999. Breeding biology of the yellow-billed cuckoo relative to timber harvest in a bottomland hardwood forest. Athens, GA: University of Georgia. 107 p. Ph.D. dissertation. Wilson, M.D.; Watts, B.D. 1999. Response of brown-headed nuthatches to thinning of pine plantations. Wilson Bulletin. 111(1): 56-60. Wilson, R.R., Cooper, R.J. 1998. Acadian ﬂycatcher nest placement: does placement inﬂuence reproductive success? Condor. 100(4): 673-679. Withgott, J.H.; Smith, K.G. 1998. Brown-headed nuthatch (Sitta pusilla). Number 349. In: Poole, A.; Gill, F., eds. The birds of North America. Philadelphia, PA: Academy of Natural Sciences; Washington, DC: American Ornithologists’ Union. Wood, D.R.; Burger, L.W., Jr.; Bowman, J.L.; Hardy, C.L. 2004. Avian community response to pine-grassland restoration. Wildlife Society Bulletin. 32(3): 819-829. Woodward, A.A.; Fink, A.D.; Thompson, F.R., III. 2001. Edge eﬀects and ecological traps: eﬀects on shrubland birds in Missouri. Journal of Wildlife Management. 65(4): 668-675. Woolfenden, B.E.; Stutchbury, B.J.M.; Morton, E.S. 2005. Male Acadian ﬂycatchers, Empidonax virescens, obtain extrapair fertilizations with distant females. Animal Behaviour. 69(4): 921-929. Yahner, R.H. 2003. Responses of bird communities to early successional habitat in a managed landscape. Wilson Bulletin. 115(3): 292-298. Zimmerman, G.M. 2004. Studies of the annual cycle of the swallow-tailed kite (Elanoides forﬁcatus): migration, habitat use, and parasites. Statesboro, GA: Georgia Southern University. 131 p. M.S. thesis. 195 Tirpak, John M.; Jones-Farrand, D. Todd; Thompson, Frank R., III; Twedt, Daniel J.; Uihlein, William B., III. 2009. Multiscale habitat suitability index models for priority landbirds in the Central Hardwoods and West Gulf Coastal Plain/ Ouachitas Bird Conservation Regions. Gen. Tech. Rep. NRS-49. Newtown Square, PA: U.S. Department of Agriculture, Forest Service Northern Research Station. 195 p. Habitat Suitability Index (HSI) models were developed to assess habitat quality for 40 priority bird species in the Central Hardwoods and West Gulf Coastal Plain/ Ouachitas Bird Conservation Regions. The models incorporated both site and landscape environmental variables from one of six nationally consistent datasets. Potential habitat was first defined from unique landform, landcover, and successional age class combinations. Species-specific environmental variables identified from the literature were used to refine initial habitat estimates. Models were verified by comparing subsection-level HSI scores and Breeding Bird Survey (BBS) abundance via Spearman rank correlation. Generalized linear models that predicted BBS abundance as a function of HSI were used to validate models. KEY WORDS: Conservation planning, ecoregion, forest, Forest Inventory and Analysis, National Landcover Dataset, validation. The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs and activities on the basis of race, color, national origin, age, disability, and where applicable, sex, marital status, familial status, parental status, religion, sexual orientation, genetic information, political beliefs, reprisal, or because all or part of an individual’s income is derived from any public assistance program. (Not all prohibited bases apply to all programs.) Persons with disabilities who require alternate means for communication of program information (Braille, large print, audiotape, etc.) should contact USDA’s TARGET Center at (202)720-2600 (voice and TDD). To file a complaint of discrimination, write to USDA, Director, Office of Civil Rights, 1400 Independence Avenue, S.W., Washington, DC 20250-9410, or call (800)795-3272 (voice) or (202)720-6382 (TDD). USDA is an equal opportunity provider and employer. Printed on Recycled Paper Capitalizing on the strengths of existing science capacity in the Northeast and Midwest to attain a more integrated, cohesive, landscape-scale research program www.nrs.fs.fed.us View publication stats

RELATED PAPERS

RELATED TOPICS

Log In

Multiscale habitat suitability index models for priority landbirds in the Central Hardwoods and West Gulf Coastal Plain/Ouachitas Bird Conservation Regions

Multiscale habitat suitability index models for priority landbirds in the Central Hardwoods and West Gulf Coastal Plain/Ouachitas Bird Conservation Regions

Related Papers

RELATED PAPERS

RELATED TOPICS