FLORIDA WILD TURKEY NEST SITE SELECTION AND NEST SUCCESS ACROSS MULTIPLE SCALES

FLORIDA WILD TURKEY NEST SITE SELECTION AND NEST SUCCESS ACROSS MULTIPLE SCALES By JOHN M. OLSON A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2011

2011 John M. Olson 2

To my family and Rosemary, LLC 3

ACKNOWLEDGMENTS I would like to extend the sincerest of thanks to my parents, family, and fiancé, who have always supported me and pushed me to do more. I would also like to thank Dr. William Giuliano, Dr. Holly Ober, Dr. Emma Willcox, and John Denton for their guidance and support; Mitchell Blake and the technicians who assisted in data collection for their parts in this project; and the University of Florida and the Florida Fish and Wildlife Conservation Commission for providing the financial and technical support necessary to the project. 4

TABLE OF CONTENTS page ACKNOWLEDGMENTS... 4 LIST OF TABLES... 6 LIST OF FIGURES... 7 ABSTRACT... 8 CHAPTER 1 INTRODUCTION... 10 Study Objectives... 12 Study Sites... 12 2 METHODS... 14 Data Collection... 14 Analysis... 19 3 RESULTS... 28 Selection... 28 Success... 30 4 DISCUSSION... 57 Selection... 57 Success... 61 LIST OF REFERENCES... 66 BIOGRAPHICAL SKETCH... 71 5

LIST OF TABLES Table page 2-1 Variable names, abbreviations, and definitions used in a priori models to predict nest habitat selection and success at microhabitat and patch levels... 22 2-2 Variable categories, names, abbreviations, and definitions used in a priori models to predict nest habitat selection and success at the landscape level... 23 3-1 Ranked models used to predict nest habitat selection at the microhabitat level... 33 3-2 Parameter estimates, 95% confidence intervals (CI), and odds ratios (OR) for most supported models for selection at the microhabitat level... 34 3-3 Ranked models used to predict nest habitat selection at the microhabitat level... 35 3-4 Parameter estimates, 95% confidence intervals (CI), and odds ratios (OR) for most supported models for selection at the patch level... 36 3-5 Ranked models used to predict nest habitat selection at the landscape level... 37 3-6 Parameter estimates, 95% confidence intervals (CI), and odds ratios (OR) for most supported models for selection at the landscape level... 43 3-7 Most supported a priori model(s) from each variable category predicting nest habitat selection at the landscape level... 44 3-8 Ranked models used to predict nest success at the microhabitat level... 45 3-9 Parameter estimates, 95% confidence intervals (CI), and odds ratios (OR) for most supported models for success at the microhabitat level... 46 3-10 Ranked models used to predict nest success at the patch level... 47 3-11 Parameter estimates, 95% confidence intervals (CI), and odds ratios (OR) for most supported models for success at the patch level... 48 3-12 Ranked models used to predict nest success at the landscape level... 49 3-13 Parameter estimates, 95% confidence intervals (CI), and odds ratios (OR) for most supported models for success at the landscape level... 55 3-14 Most supported a priori model(s) from each variable category predicting nest success at the landscape level... 56 6

LIST OF FIGURES Figure page 2-1 Schematic of vegetation sampling plot for microhabitat level... 26 2-2 Schematic of vegetation sampling plot for patch level... 27 7

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science FLORIDA WILD TURKEY NEST SITE SELECTION AND NEST SUCCESS ACROSS MULTIPLE SCALES Chair: William M. Giuliano Major: Wildlife Ecology and Conservation By John M. Olson August 2011 Landscapes and land-use practices in Florida continue to change and possibly degrade the quality of habitat available to Florida wild turkey hens (Meleagris gallopavo osceola) with respect to their nest site selection and subsequent success. This study attempted to understand wild turkey hen nest site selection and habitat effects on success at the microhabitat, patch, and landscape levels using logistic regression and AIC model selection at two sites in southern Florida, 2008-2010. Hens selected nest sites in dense vegetation (e.g., saw palmetto; Serenoa repens) that provided lateral and vertical cover for concealment at the microhabitat level (i.e., area within 7 m of the nest bowl), while selecting for a more open habitat at the patch level (i.e., 0.25 ha area surrounding the nest). This presumably allowed hens to survey the area for predators prior to ingress or egress, while also providing concealment. At the landscape level, hens continued this trend, selecting for areas characterized by patchy, dense, hardy vegetation, increasing possible nest locations, while allowing access to forage locations and brood rearing habitat. Areas in which vegetation was managed (i.e., areas burned or roller-chopped) were avoided. Successful hens (i.e., hatching of 1 egg) selected for lower basal area and dense saw palmetto cover at the microhabitat level and more open 8

habitat at the patch level. At the landscape level, nest success was associated with a greater distance from habitat edges and areas burned 0.5-2 years prior, which may have decreased the probability of predation by locating nests in the center of habitat patches, away from edge corridors. Overall, it appears that a combination of treatments, both prescribed burning and roller-chopping, may best benefit Florida wild turkey hens by creating a mosaic habitat characterized by patches of dense vegetation within an open landscape. 9

CHAPTER 1 INTRODUCTION In Florida, changing land-use practices may degrade or destroy native habitats, through urban development, road construction, fragmentation, conversion to agriculture, fire exclusion, invasive species, and changes in natural disturbance regimes. Of particular importance is change in disturbance regimes such as the frequency and timing of fire. In Florida s native plant communities, fire suppression, reduced fire frequency, and a switch to dormant season burning has led to the proliferation and ultimate dominance of woody shrubs, which, in high densities, degrades habitat quality for many species dependent upon early successional habitats with more open understories. In much of the native rangeland and forest remaining in Florida, saw palmetto has become the dominant understory component due to changes in fire regimes (Tanner and Marion 1990). This has resulted in a reduction of native grasses, herbaceous plants, and shrubs beneficial to wildlife as food and cover (Tanner et al. 1986). These conditions may also make some habitat unusable or inaccessible by forming barriers to wildlife movement. In recent years, many managers have sought to mitigate the proliferation and abundance of problematic woody shrub species. This type of management typically involves habitat restoration through treatments such as prescribed fire and roller-chopping. These two treatments aid in opening the understory, allowing herbaceous plants and other vegetation of value to Florida s wildlife to proliferate (Willcox and Giuliano 2010). Research has shown that these practices can improve habitat quality for many species, including several species of critical concern for the state of Florida such as the gopher tortoise (Gopherus polyphemus) and red-cockaded woodpecker (Picoides borealis), and popular game species such as northern bobwhite 10

(Colinus virginianus). On many public and private lands, managers have implemented habitat restoration specifically designed to benefit northern bobwhite. However, whether this type of management benefits Florida wild turkey (Meleagris gallopavo osceola) hens, their nest site selection, or their nesting success has yet to be determined. Little is known about Florida wild turkey nest habitat selection and its effects upon nest success. Williams and Austin (1988), Williams (1991), and Dickson (1992) characterized Florida wild turkey nests. They reported Florida wild turkey hens select areas in transition zones between palmetto prairie and oak scrub, where they could conceal themselves. Williams and Austin (1988) reported that hens favored saw palmetto, specifically palmetto ecotones, which concurs with Dickson s (1992) accounts that hens nesting in dense vegetation were less likely to flush, reducing detection probability. However, much of this is only anecdotal evidence. Additionally, although it is presently unknown whether prescribed burning and roller-chopping benefit nesting Florida wild turkeys, these treatments may provide complex vegetation structure preferred by nesting hens (Badyaev 1995). Several projects have examined habitat selection of other wild turkey subspecies found in the United States, particularly the eastern subspecies (Meleagris gallopavo silvestris), but are equivocal. Research indicated that eastern wild turkey hens in the Southeast selected for denser understories and more open midstories for nesting, though higher levels of visual obstruction due to lateral cover at the nest was the most important factor in selection (Godfrey and Norman 2001). Others have found that hens selected against bottomland hardwoods in favor of pine (Pinus spp.) stands, and that nests located in areas with less lateral cover, closer to roads and edges, and in forested habitats were 11

more successful than those that were not (Seiss et al. 1990). Badyaev (1995) found that eastern wild turkey hens preferred cover types that had lower overstory densities and fewer trees of all classes, and successful nests better concealed incubating hens by having denser lateral and vertical cover while having lower densities of large trees. Seiss et al. (1990) found no selection for burn age by nesting hens, but Hon et al. (1978) found hens in Georgia selected for recently burned areas, while Exum et al. (1987) found hens in Alabama selected areas not recently burned. Research suggests nest success as the most important factor affecting wild turkey population growth and ultimate size (Seiss et al. 1990, Roberts and Porter 1996), habitat often drives nest success, and habitat selection is a hierarchical process where birds select features at different scales (Johnson 1980, Lazarus and Porter 1985, Thogmartin 1999). Therefore, to better manage the unique Florida wild turkey subspecies, further information is needed to understand habitat determinants of nest success and how management practices such as roller-chopping and prescribed burning affect Florida wild turkey hen nest site selection and success. Study Objectives My objectives for this project were to: 1) determine what nest site characteristics influence hens nest site selection at different spatial scales, 2) discern how habitat affects nest success, and 3) evaluate how nest site selection relates to success. Study Sites I conducted this study on two sites in south-central Florida from 2008-2010. The first site was Three Lakes Wildlife Management Area (WMA), located in Osceola County, Florida. Data collection was limited to the 6,273 ha Quail Enhancement Area, where managers conducted frequent prescribed burning and roller-chopping. Three Lakes 12

WMA consists primarily of pine flatwoods, though there are also intermingled hammocks, swamps, and wet and dry prairies (Florida Natural Areas Inventory 2010). The state of Florida owns Three Lakes WMA, and the Florida Fish and Wildlife Conservation Commission (FWC) executes management and allows the public to hunt the property for white-tailed deer (Odocoileus virginianus), feral hog (Sus scrofa), northern bobwhite, small game, and wild turkey. The second site was Longino Ranch, located in Sarasota County, Florida. Longino Ranch encompasses approximately 4,040 ha, with 2,020 ha used for the production of cattle, sod, and citrus. The remaining 2,020 ha are in pine flatwoods, wet and dry prairies, and oak-cabbage palm hammocks (Florida Natural Areas Inventory 2010). Longino Ranch conducts prescribed burns and roller-chopping, but not on the scale of Three Lakes WMA. Longino ranch historically managed solely for timber, but now manages for both timber and cattle. The ranch managers operate deer, feral hog, and wild turkey hunts for the owning family. 13

CHAPTER 2 METHODS Data Collection I prepared capture sites (n = 20-35/year; January-February) within the Quail Enhancement Area on Three Lakes WMA and within the boundaries of Longino Ranch. I baited each capture site with cracked corn or three-grain scratch feed, and prepared rocket-nets at sites only after confirming use by female turkeys through the presence of tracks and excrement. I used rocket nets to capture turkeys from January to early March each year from 2008-2010 on both study sites (Bailey et al. 1980). Upon firing nets, I secured captured turkeys, and subsequently placed each into cardboard boxes specifically designed to contain wild turkeys. Then, I fitted each captured hen with standard numbered metal leg bands and backpack-style radio transmitters with mortality switches (ATS transmitters, model A1540, 69-80 grams [not including harness material]: weighing <3.5% of birds body weight). I aged, weighed, and administered a dose of vitamin E to each captured hen in an attempt to offset the stress associated with capture. Finally, I released turkeys within 45 minutes of capture at the capture location. I located radioed hens remotely by triangulation (White and Garrot 1990) using radio receivers and hand-held three element Yagi antennae from pre-established telemetry stations (n ~ 100-250 depending upon site and year) using the peak method (Fuller et al. 2005). To locate radioed hens, I recorded one azimuth from 3 distinct telemetry stations within 15 minutes to reduce error associated with long-distance movements of the radioed hen (Fuller et al. 2005). I entered recorded azimuths into the program Location of a Signal (LOAS; Ecological Software Solutions 2010) to map the estimated location of tracked hens and generate error polygons. I located hens 3 times weekly from early 14

March until July 15 th each year 2008-2010. When observations indicated that a hen had initiated a nest and begun incubation (i.e., was found repeatedly in the same location), I recorded a nesting attempt if it could be confirmed by homing in on the nesting hen (Tirpak et al. 2006). I monitored active nests daily via telemetry, and when the hen was away from her nest, confirmed the status of the nest visually, careful to keep disturbance to a minimum in nesting areas. I considered nests that hatched 1 egg successful (Tirpak et al. 2006). Once a nest fate had been determined, I measured habitat characteristics at multiple scales at nest sites and random locations. To characterize the microhabitat (i.e., the area within 7 m of the nest bowl), I measured lateral cover (i.e., horizontal visual obstruction), vegetation cover, shrub height, basal area, and tree stem density in a 7 m radius plot centered on the nest bowl (Table 1, Figure 1). I measured basal area of hardwood, coniferous, and palm species separately from the center using a standard 10-BAF prism (Higgins et al. 2005). Only trees that measured >11.43 cm (4.5 in) diameter-at-breast-height (dbh) were considered (Sparks et al. 2002). Stem counts of tree species >2.54 cm (1 in) dbh within the plot were tallied as hardwood, conifer, or palm species. I measured lateral cover by visually estimating total cover (%) of a 36 cm x 90 cm cover board placed at three equally spaced points along the perimeter of the 7 m plot (Higgins et al. 2005). To classify cover densities, I recorded estimates in one of six cover classes (i.e., 1 = 0-3%, 2 = 4-12%, 3 = 13-25%, 4 = 26-50%, 5 = 51-75%, 6 = 76-100%). When estimating cover obstruction, I viewed the cover board from the center point of plots at standing height (1.7 m), and used the mean of readings taken for analysis. I measured vegetation cover and shrub height below 1.5 m using three 7 m transects radiating from the plot center (Krebs 15

1999). Canopy cover of saw palmetto below 1.5 m was estimated by line-intercept divided by the total length of transects, expressed as a percentage (Higgins et al. 2005). To determine intercepts, I ignored intercepts <1 cm, while small openings <20 cm within individual plants or gaps <10 cm between individual plants were included as cover. On each transect, I recorded species and height of the tallest shrub (up to 150 cm) of the tallest point of shrub intercept (including shrubs and trees <2.54 cm dbh; Higgins et al. 2005). For a simple estimate of total cover of all shrubs except saw palmetto, I summed the individual shrub estimates on all three transects for analysis. After recording the characteristics at the nest site, I recorded characteristics in a plot located a random distance and direction from the nest within the same habitat patch. To characterize vegetation at the patch level (i.e., area 0.25 ha around the nest bowl), I recorded vegetation characteristics in a circular plot 28 m in radius centered on the nest bowl (Figure 2). I measured lateral cover, vegetation cover, shrub height, basal area, and tree stem density (Table 1). Point-centered habitat characteristics were measured in four 7 m radius circular plots, one centered on the nest site and three adjacent plots equally spaced around the center plot at 21 m, one on each of three transects run outwards at 120º (Krebs 1999), using the methods described for the microhabitat. I measured vegetation cover and shrub height below 1.5 m using three 28 m transects radiating from the center plot at the same angles as the adjacent plots. Canopy cover of saw palmetto below 1.5 m was estimated by line-intercept methods, calculated as the accumulated length intercepted by living and standing dead parts of palmetto divided by the total length of transects, expressed as a percentage (Higgins et al. 2005). In 7 m intervals along each transect, I recorded the height (up to 150 cm) and 16

species of the tallest point of shrub intercept (including shrubs and trees <2.54 cm dbh; Higgins et al. 2005). I averaged all shrub measurements for analysis. To characterize nesting habitat at the landscape level, I used 95% fixed kernel home ranges generated for radioed hens using the Home Range Tools extension in ArcGIS (Environmental Systems Research Institute 2009; Rodgers et al. 2007) to obtain the median home range size for each study site and year. I censored hens with <30 locations. To define the study areas by site and year, I used the Create Minimum Convex Polygons extension in Hawth s Analysis Tools (Beyer 2004; Schad 2009) to create minimum convex polygons around all hen locations at each study site as generated by LOAS with an estimated error of <10 ha. To delineate habitat cover types, I downloaded and imported Florida Natural Areas Inventory (FNAI) Cooperative Land Cover Map shapefiles (Florida Natural Areas Inventory 2010) into ArcGIS. I also downloaded United States Geological Survey (USGS) orthophoto quadrangles to create shapefiles delineating landscape features such as roads and water features that were absent from the FNAI shapefiles. I projected the nest sites discovered during the three years of the project into ArcGIS and buffered each with a circular buffer equivalent to median home range size of birds for each site and year to establish landscape level use areas (Tirpak et al. 2010). To establish availability, I divided the study area size for each site and year by the median home range size for that site and year. This provided the number of home ranges that could fit into each study area. To obtain random points and establish availability, I arbitrarily multiplied the number of home ranges that could fit into each study area by five to increase sample and study area coverage. I used Hawth s Analysis Tools to generate 17

random points within each study area according to this formula and buffered each with a circular buffer equal in area to the median home range size for each site and year (Tirpak et al. 2006). I created a total of 370 random points and corresponding buffers, which ranged from 25-105 per study area per year. This method of establishing availability provided nearly total coverage of each study area. I developed four suites of variables (i.e., habitat, management, landscape, habitat treatment) for landscape level analyses (Johnson s level 2; Johnson 1980). I determined habitat from the FNAI land cover shapefile and represented the area of each particular habitat type found within the buffer of each nest point, both random and actual (Table 2). I used the ArcGIS intersect function to merge actual and random buffers with the habitat shapefile to determine the area of each habitat within each buffer. Five habitat types (i.e., developed, bottomland forest, successional hardwood forest, sandhill, xeric hammock) represented less than one percent of each study area and were combined into one category, which I labeled other. Additionally, I created new categories to combine similar habitat types, including clearing and unimproved pasture into abandoned clearing, and hydric hammock and mesic hammock into hammock (Table 2). I divided management into two treatments (i.e., prescribed burning and roller-chopping) and separated each treatment into five distinct age categories defined as: 1) treatment application <6 months prior to nest initiation, 2) treatment application 6 months 2 years prior to nest initiation, 3) treatment application >2 years prior to nest initiation, 4) no record of recent management, or 5) management records incomplete/unavailable (Williams 1991; Table 2). To establish management age for random home ranges used in selection analyses, I used median initiation dates for the 18

respective area and year. Study site managers provided records of management history and shapefiles were created according to these records. I then intersected this layer with buffer layers to obtain areas of different management ages within each buffer. The landscape category contained variables dealing with the area of and distance to road, habitat edge, and water (Table 2). I mapped together paved roads, dirt roads, firebreaks, and paths visible from aerial photographs, reasoning that if they were large enough to be detected with aerial photography, they were large enough to be traveled by turkeys and therefore could affect turkey behavior. Then, I applied a 2.5 m buffer to all roads because this most closely resembled average width of roads present within study sites as per my field experience. I intersected these landscape attributes with nest buffers to acquire areas of each within buffers. Finally, I used the near feature in ArcGIS to obtain distances from each nest to the nearest feature of each variable within this suite (Table 2). In the final landscape level variable category, I combined both habitat type and management history to create habitat treatment variables that denoted particular management for several habitats. Habitats included were those that site managers targeted for management. To accomplish this, I used the identity function in ArcGIS to combine the FNAI habitat layer and the management layer into one. I intersected this new layer with the buffers around the nests and random points to obtain areas of habitat with treatment histories (Table 2). Analysis To analyze how Florida wild turkey hens selected nest sites and how habitat affected success, I used an information-theoretical approach and logistic regression in SYSTAT 12.0 (SYSTAT 2007). I used case-control logistic regression to compare habitat variables present within the vegetation plots and their associated random points at the 19

microhabitat level (i.e., characteristics from the 7m radius plot centered on the nest bowl; Table 1). For patch level (i.e., vegetation characteristics within 0.25 ha area surrounding nest bowls; Table 1) selection analyses, I used case-control logistic regression to compare habitat variables from nest plots and their respective random points. To quantify selection at the landscape level (i.e., landscape attributes present within simulated circular home ranges around nest sites and random points; Table 2), I used logistic regression to compare the habitat present within simulated home ranges for site and year to habitat within equally sized random home ranges generated across study areas annually. I created models featuring each individual variable present at all three levels (Table 1, Table 2), models containing combinations of these variables, and a null model. Based upon prior knowledge, project goals, and my own field experience, I also created a priori models containing combinations of variables. I evaluated models using Akaike s Information Criterion (AIC C ) adjusted for small sample size (n/k<40), and considered models with AIC C 2 supported (Burnham and Anderson 2002). To rank model and variable importance, I used Akaike weights (w), and adjusted coefficients and odds ratios of competing models (Burnham and Anderson 1998). When 95% confidence intervals for variables within supported models overlapped with zero, I considered them to have a weak effect on the dependent variable, and only indicate a trend. Finally, I examined both the best model from each landscape level category (e.g., management), and also the best models from all landscape level categories combined to determine which had the greatest effect on wild turkey hen nest site selection. 20

I used logistic regression to compare habitat of successful and unsuccessful nests at the microhabitat level (i.e., characteristics from the 7m radius plot centered on the nest bowl; Table 1), patch level (i.e., characteristics from the 0.25 ha area around each nest; Table 1), and landscape level (i.e., landscape attributes present within simulated circular home ranges around nest sites; Table 2). I used methods as listed above for selection analyses to determine important factors influencing to nest success. 21

Table 2-1. Variable names, abbreviations, and their definitions used in a priori models to predict nest habitat selection and success at microhabitat and patch levels for Florida wild turkey hens in south Florida, 2008-2010. Variable Abbreviation Description Conifer basal area BAC Conifer basal area m 2 /ha Hardwood basal area BAH Hardwood basal area m 2 /ha Palm basal area BAP Palm basal area m 2 /ha Total basal area BAT Total basal area m 2 /ha Conifer stems STC Conifer stems no./ha Hardwood stems STH Hardwood stems no./ha Palm stems STP Palm stems no./ha Total stems STT Total stems no./ha Saw palmetto density SD Saw palmetto density % Visual obstruction VO Visual obstruction % Shrub height SHT Shrub height cm 22

Table 2-2. Variable categories, names, abbreviations, and their definitions used in a priori models to predict nest habitat selection and success at the landscape level for Florida wild turkey hens in south Florida, 2008-2010. Variable Category Variable Abbreviation Description Habitat Abandoned clearing A Ha of abandoned clearing Agriculture Ag Ha of agriculture Basin swamp BS Ha of basin swamp Baygall BG Ha of baygall Bottomland forest BF Ha of bottomland forest Clearing C Ha of clearing Depression marsh DM Ha of depression marsh Dome swamp DS Ha of dome swamp Dry prairie DP Ha of dry prairie Hammock H Ha of hammock Hydric hammock HH Ha of hydric hammock Improved pasture IP Ha of improved pasture Mesic flatwoods MF Ha of mesic flatwoods Mesic hammock MH Ha of mesic hammock Other O Ha of other Pine plantation PP Ha of pine plantation Sand hill SH Ha of sand hill Scrub S Ha of scrub Scrubby flatwoods SF Ha of scrubby flatwoods Shrub bog SB Ha of shrub bog Successional SHF Ha of successional hardwoods forest hardwood forest Unimproved pasture UP Ha of unimproved pasture Upland hardwood UHF Ha of upland hardwood forest forest Wet flatwoods WF Ha of wet flatwoods Wet prairie WP Ha of wet prairie Xeric hammock X Ha of xeric hammock Landscape Distance to edge DEDGE Distance to nearest habitat edge m Distance to roads DROAD Distance to nearest road m Distance to nearest edge DRD_DEDGE Distance to nearest habitat edge or road m Distance to water DWATER Distance to nearest water body m Road ROAD Total amount of road ha Edge EDGE Amount of habitat edge ha Edge total RD_EDGE Total amount of habitat edge and roads ha Water WATER Total amount of water ha Management Burn1 B1 Ha of area burned <6 months Burn2 B2 Ha of area burned between 6 months and 2 years Burn3 B3 Ha of area burned >2 years Burn4 B4 Ha of area with no recent burn history Burn5 B5 Ha of area with incomplete burn history Chop1 C1 Ha of area roller chopped <6 months Chop2 C2 Ha of area roller chopped between 6 months and 2 years Chop3 C3 Ha of area roller chopped >2 years Chop4 C4 Ha of area with no recent roller chopping history 23

Table 2-2. Continued. Variable Category Variable Abbreviation Description Habitat Treatment Chop5 C5 Ha of area with incomplete roller chopping history Dry prairie1 DP1 Ha of dry prairie burned <6 months Dry prairie2 DP2 Ha of dry prairie burned 6 months - 2 years Dry prairie3 DP3 Ha of dry prairie burned >2 years Dry prairie4 DP4 Ha of dry prairie with no recent burn history Dry prairie5 DP5 Ha of dry prairie with incomplete burn history Mesic flatwoods1 MF1 Ha of mesic flatwoods burned <6 months Mesic flatwoods2 MF2 Ha of mesic flatwoods burned 6 months - 2 years Mesic flatwoods3 MF3 Ha of mesic flatwoods burned >2 years Mesic flatwoods4 MF4 Ha of mesic flatwoods with no recent burn history Mesic flatwoods5 MF5 Ha of mesic flatwoods with incomplete burn history Pine plantation1 PP1 Ha of pine plantation burned <6 months Pine plantation2 PP2 Ha of pine plantation burned 6 months - 2 years Pine plantation3 PP3 Ha of pine plantation burned >2 years Pine plantation4 PP4 Ha of pine plantation with no recent burn history Pine plantation5 PP5 Ha of pine plantation with incomplete burn history Scrubby flatwoods1 SF1 Ha of scrubby flatwoods burned <6 months Scrubby flatwoods2 SF2 Ha of scrubby flatwoods burned 6 months - 2 years Scrubby flatwoods3 SF3 Ha of scrubby flatwoods burned >2 years Scrubby flatwoods4 SF4 Ha of scrubby flatwoods with no recent burn history Scrubby flatwoods5 SF5 Ha of scrubby flatwoods with incomplete burn history Wet flatwoods1 WF1 Ha of wet flatwoods burned <6 months Wet flatwoods2 WF2 Ha of wet flatwoods burned 6 months - 2 years Wet flatwoods3 WF3 Ha of wet flatwoods burned >2 years Wet flatwoods4 WF4 Ha of wet flatwoods with no recent burn history Wet flatwoods5 WF5 Ha of wet flatwoods with no recent burn Wet prairie1 WP1 Ha of wet prairie burned <6 months Wet prairie2 WP2 Ha of wet prairie burned 6 months - 2 years Wet prairie3 WP3 Ha of wet prairie burned >2 years Wet prairie4 WP4 Ha of wet prairie with no recent burn history Wet prairie5 WP5 Ha of wet prairie with incomplete burn history Chop_dry prairie1 DPC1 Ha of dry prairie roller chopped <6 months Chop_dry prairie2 DPC2 Ha of dry prairie roller chopped 6 months - 2 years 24

Table 2-2. Continued. Variable Category Variable Abbreviation Description Chop_dry prairie3 DPC3 Ha of dry prairie roller chopped >2 years Chop_dry prairie4 DPC4 Ha of dry prairie with no recent roller chopping history Chop_dry prairie5 DPC5 Ha of dry prairie with incomplete roller chopping history Chop_mesic MFC1 Ha of mesic flatwoods <6 months flatwoods1 Chop_mesic flatwoods2 MFC2 Ha of mesic flatwoods roller chopped 6 months - 2 years Chop_mesic flatwoods3 MFC3 Ha of mesic flatwoods roller chopped >2 years Chop_mesic flatwoods4 MFC4 Ha of mesic flatwoods with no recent roller chopping history Chop_mesic flatwoods5 MFC5 Ha of mesic flatwoods with incomplete roller chopping history Chop_scrubby SFC1 Ha of scrubby flatwoods <6 months flatwoods1 Chop_scrubby flatwoods2 SFC2 Ha of scrubby flatwoods roller chopped 6 months - 2 years Chop_scrubby flatwoods3 SFC3 Ha of scrubby flatwoods roller chopped >2 years Chop_scrubby flatwoods4 SFC4 Ha of scrubby flatwoods with no recent roller chopping history Chop_scrubby flatwoods5 SFC5 Ha of scrubby flatwoods with incomplete roller chopping history Chop_wet flatwoods1 WFC1 Ha of wet flatwoods <6 months Chop_wet flatwoods2 WFC2 Ha of wet flatwoods roller chopped 6 months - 2 years Chop_wet flatwoods3 WFC3 Ha of wet flatwoods roller chopped >2 years Chop_wet flatwoods4 WFC4 Ha of wet flatwoods with no recent roller chopping history Chop_wet flatwoods5 WFC5 Ha of wet flatwoods with incomplete roller chopping history Chop_wet flatwoods2 WFC2 Ha of wet flatwoods roller chopped 6 months - 2 years 25

Figure 2-1. Schematic of vegetation sampling plot used to record vegetation characteristics and measurements to predict nest habitat selection and success at the microhabitat level for Florida wild turkey hens in south Florida, USA, 2008-2010. 26

Figure 2-2. Schematic of vegetation sampling plot used to record vegetation characteristics and measurements to predict nest habitat selection and success at the patch level for Florida wild turkey hens in south Florida, USA, 2008-2010. 27

CHAPTER 3 RESULTS During the three years of study, I captured and radioed 142 hens on the two study sites. I discovered 67 nests, 27 of which were successful. At Three Lakes WMA, I discovered 8, 8, and 14 nests in 2008, 2009, and 2010, respectively. I found 10, 10, and 17 nests in 2008, 2009, and 2010, respectively, at Longino Ranch. The leading cause of nest failure was depredation (n = 24), though nests also failed due to predation of the hen on the nest (n = 8) and abandonment (total n = 7; due to habitat management n = 3, due to observer interference n = 1, unknown cause n = 3). Habitat management efforts accounted for three cases of nest abandonment through prescribed fire (n = 2) and logging (n = 1). One nest was established nearing the study s terminus and not monitored to fate. I censored this nest and nests failing due to management or observer interference (n = 5) from both selection and success analyses because no data regarding vegetation characteristics could be recorded (i.e., vegetation characteristics hens selected were destroyed, or at minimum, radically changed after a prescribed fire) and these nests failed due to artificial causes not dependent upon hens selection decisions. Selection At the microhabitat level of selection, I found three supported models (Table 3). The most supported model contained palm and conifer stem density and saw palmetto density. Saw palmetto density was the only variable for which the 95% confidence interval for the parameter estimate did not overlap with zero and indicated that hens selected nest sites with a greater amount saw palmetto (Table 4). The other models indicated that turkeys also selected for a greater density of palm stems. Trends 28

suggested that hens selected against increasing conifer and hardwood stem densities (Table 4). Six models were supported at the patch level, with the most supported model containing palm and hardwood stem densities (Table 5); however all 95% confidence intervals for parameter estimates overlapped with zero which limited interpretation (Table 6). Trends indicated that while hens selected higher densities of palm stems, they also selected for more open areas, manifested by lower hardwood, conifer, and total stem and saw palmetto densities, and lower levels of visual obstruction. The habitat category for landscape level selection contained two supported models, with the best model containing the habitat types agriculture, dry prairie, mesic flatwoods, and wet flatwoods (Table 7). Agriculture, dry prairie, and mesic flatwoods had 95% confidence intervals of parameter estimates not containing zero, and suggested that hens selected for greater amounts of each; while trends indicated that hens also selected for scrubby flatwoods and wet flatwoods (Table 8). In the landscape category, there were two supported models (Table 7). The best model contained the variables distance to road and distance to water. Both had 95% confidence intervals of estimates that did not overlap with zero, and suggested that hens selected sites further from roads and water (Table 8). Trends indicated that turkeys also selected sites that were located nearer to habitat edges (Table 8). There were six supported models at the landscape level of selection in the management category (Table 7). The best model contained the variables denoting areas burned 0.5-2 years ago, unburned, and unchopped; though only the 95% confidence interval of the estimate for the unchopped did not overlap with zero (Table 8). This 29

suggested that hens selected for areas that had not received any roller-chopping application. Trends for other parameter estimates indicated that hens selected against sites burned >6 months prior, but selected for sites chopped >6 months prior (Table 8). The habitat treatment category had two supported models, and the most supported model contained unburned dry prairie, scrubby flatwoods, and mesic flatwoods, mesic flatwoods burned 0.5-2 years ago, mesic flatwoods chopped 0.5-2 years ago, and unchopped mesic flatwoods (Table 7). All parameters had confidence intervals containing zero except unchopped mesic flatwoods, which suggested that hens selected for greater amounts of this habitat treatment type (Table 8). Trends suggested that hens selected for unburned scrubby flatwoods, unchopped and mesic flatwoods chopped 0.5-2 years ago, and against unburned dry prairie and mesic flatwoods and mesic flatwoods burned 0.5-2 years ago (Table 8). When I compared the best models from each of the landscape level categories, only models from the management category were supported (Table 9), with the models containing burned 0.5-2 years ago, unburned, and unchopped variables. Additionally, I found that only the unchopped parameter had a 95% confidence interval of the estimate that did not overlap with zero (Table 8, Table 9), suggesting that hens selected areas with greater amounts of unchopped vegetation. Trends indicated hens selected for areas unburned or chopped >6 months ago, while avoiding burns >6 months old. Success At the microhabitat level, nine models examining habitat differences between successful and unsuccessful nests were supported (Table 10). The most supported model contained only total basal area, which had a 95% confidence interval not overlapping zero and indicated that successful nests were associated with a lower total 30

basal area than unsuccessful nests (Table 11). Other supported models suggested that nest success was associated with a lower conifer basal area and higher saw palmetto density. Additionally, trends indicated that hens selecting areas with greater visual obstruction, hardwood basal area, and lower palm, conifer, and total stem density, and hardwood and conifer basal area were more likely to succeed (Table 11). Patch level nest success had three supported models (Table 12). The most supported model included only palm basal area, but parameters within all models had 95% confidence intervals that overlapped with zero, limiting interpretation (Table 13). Trends indicated that successful nests had greater palm stem density and lower total and palm basal area than unsuccessful nests (Table 13). The habitat category at the landscape level had four supported models, with the most supported model containing scrubby flatwoods and wet flatwoods, but all parameter 95% confidence intervals contained zero, limiting interpretation (Table 14, Table 15). Trends suggested that when compared with unsuccessful nests, successful nests were more often associated with scrubby flatwoods, and less often with wet flatwoods and dry prairie (Table 15). In the landscape category, the null model had the most support, though there were eight other supported models. As the null model had the most support, all results in this category must be interpreted very conservatively. All parameters had 95% confidence intervals overlapping zero, but trends suggested that successful nests were located further from roads, habitat edge, and water than unsuccessful nests, while having more area of each within the home range (Table 15). 31

Within the management category, I found five models supported at landscape level for success (Table 14). Both parameters within the most supported model had 95% confidence intervals that did not overlap with zero, and suggested that nests in areas that contained more burns 0.5-2 years old and fewer chops 0.5-2 years old were more likely to succeed (Table 15). All variables present in other models had 95% confidence intervals that overlapped with zero, but trends suggested that successful nests had more burns 0.5-2 years old, but less area chopped >6 months ago and not chopped, and burns <6 months old and unburned (Table 15). The habitat treatment category of landscape level nest success had two supported models, with the most supported model containing unburned dry prairie and mesic flatwoods chopped 0.5-2 years ago (Table 14). All parameters in both models had 95% confidence intervals that overlapped with zero, but trends indicated that successful nests were more often associated with unburned dry prairie, and less with dry prairie burned 0.5-2 years ago and mesic flatwoods chopped 0.5-2 years ago when compared to unsuccessful nests (Table 15). When I compared results among categories at the landscape level, six models had support (Table 16). These models came from the habitat treatment, management, and habitat categories. The most supported model contained unburned dry prairie and mesic flatwoods chopped 0.5-2 years, but estimates of both parameters had 95% confidence intervals that overlapped with zero. Of the supported models, only two parameters had 95% confidence intervals not containing zero, and suggested that hens selecting areas with more burns and chops of age 0.5-2 years had greater success than those not associated with these treatments (Table 15). 32

Table 3-1. A priori models, number of variables (K), second-order Akaike s Information Criterion corrected for small sample size (AIC c ), distance from the lowest AIC c (ΔAIC c ), and model weights (w i ) used to predict nest habitat selection at the microhabitat level for Florida wild turkey hens in south Florida, 2008-2010, USA. Model K AIC C AIC C w i STP,STC,SD 3 59.18 0.00 0.31 STP,STH,SD 3 59.29 0.11 0.29 STP,SD 2 59.52 0.34 0.26 BAT,STT,SD 3 62.01 2.83 0.07 BAT,STT,SD,VO 4 63.75 4.57 0.03 STC,STH,SD 3 65.07 5.89 0.02 STC,SD 2 66.11 6.92 0.01 STC,BAH,SD 3 67.66 8.48 4.42E-03 BAT,SD 2 68.11 8.93 3.52E-03 STT,SD 2 68.18 9.00 3.40E-03 BAC,BAH,SD,VO,STC,STH 6 71.41 12.22 6.79E-04 SD,VO 2 70.83 11.65 9.04E-04 SD 1 71.98 12.79 5.10E-04 BAC,BAH,SD 3 73.04 13.86 3.00E-04 BAC,SD 2 74.12 14.93 1.75E-04 BAC,STH,SD 3 74.47 15.29 1.47E-04 BAT,STT 2 74.90 15.72 1.18E-04 STP,STT 2 75.86 16.68 7.30E-05 BAT,STT,VO 3 77.07 17.89 3.99E-05 STP,STC,STH 3 78.09 18.90 2.41E-05 STT 1 77.97 18.79 2.54E-05 STC 1 78.95 19.76 1.56E-05 STP,STC 2 79.70 20.52 1.07E-05 BAC,STC 2 79.85 20.67 9.95E-06 STT,VO 2 80.10 20.92 8.78E-06 STP,STH 2 80.58 21.40 6.90E-06 BAT 1 80.65 21.47 6.67E-06 BAH,STH 2 81.34 22.16 4.73E-06 STH 1 81.30 22.11 4.83E-06 STP 1 81.32 22.13 4.78E-06 STP,BAH 2 81.67 22.48 4.02E-06 STP,BAT 2 81.71 22.52 3.94E-06 BAC,BAH,STC,STH,SD,VO 6 83.14 23.96 1.92E-06 SHT 1 81.83 22.65 3.69E-06 BAT,BAC 2 82.01 22.83 3.38E-06 BAT,VO 2 82.10 22.92 3.23E-06 BAT,BAH 2 82.16 22.98 3.13E-06 BAH 1 82.05 22.87 3.31E-06 BAP 1 82.27 23.09 2.96E-06 STP,BAC 2 82.70 23.52 2.39E-06 BAC,STH 2 82.97 23.78 2.10E-06 VO 1 82.83 23.64 2.25E-06 STP,VO 2 83.16 23.97 1.91E-06 BAC 1 83.17 23.99 1.89E-06 STP,BAP 2 83.40 24.21 1.69E-06 BAC,BAH 2 83.57 24.39 1.55E-06 NULL 0 83.79 24.68 1.34E-06 33

Table 3-2. Parameter estimates, 95% confidence intervals (CI), and odds ratios (OR) for variables used in supported a priori models to predict nest habitat selection at the microhabitat level of Florida wild turkey hens in south Florida, USA, 2008-2010. 95% CI Model Variable Estimate Lower Upper OR STP,STC,SD STP 0.191-0.178 0.560 1.210 STC -0.014-0.034 0.005 0.986 SD 0.066 0.027 0.106 1.069 STP,STH,SD STP 0.293 0.050 0.535 1.340 STH -0.010-0.025 0.005 0.990 SD 0.065 0.026 0.104 1.067 STP,SD STP 0.279 0.029 0.529 1.321 SD 0.065 0.027 0.104 1.067 34

Table 3-3. A priori models, number of variables (K), second-order Akaike s Information Criterion corrected for small sample size (AIC c ), distance from the lowest AIC c (ΔAIC c ), and model weights (w i ) used to predict habitat selection at the patch level for Florida wild turkey hens in south Florida, 2008-2010, USA. Model K AIC C AIC C w i STP,STH 2 70.63 0.00 0.19 STP,STT 2 70.76 0.13 0.18 STP,STH,SD 3 71.75 1.13 0.11 STP,VO 2 72.02 1.39 0.09 STP 1 72.12 1.49 0.09 STP,STC,STH 3 72.57 1.94 0.07 STP,BAC 2 72.75 2.12 0.07 STP,BAT 2 73.15 2.52 0.05 STP,BAP 2 73.78 3.16 0.04 STP,SD 2 74.22 3.59 0.03 STP,STC 2 74.23 3.60 0.03 STP,BAH 2 74.25 3.63 0.03 STP,STC,SD 3 76.29 5.66 0.01 STH 1 79.74 9.12 1.97E-03 BAC,STH,SD 3 80.70 10.08 1.22E-03 BAH,STH 2 80.99 10.36 1.06E-03 STC,STH,SD 3 81.48 10.86 8.26E-04 BAC,STH 2 81.72 11.10 7.32E-04 STT 1 82.04 11.42 6.25E-04 BAP 1 82.31 11.68 5.47E-04 BAT 1 82.39 11.77 5.25E-04 BAT,STT 2 82.62 12.00 4.68E-04 VO 1 82.95 12.33 3.96E-04 BAT,VO 2 83.52 12.90 2.98E-04 STT,SD 2 83.53 12.91 2.96E-04 SD 1 83.54 12.91 2.96E-04 SHT 1 83.56 12.94 2.92E-04 BAC 1 83.61 12.98 2.85E-04 BAT,STT,SD 3 83.65 13.02 2.80E-04 BAH 1 83.70 13.07 2.73E-04 STC 1 83.72 13.09 2.71E-04 STT,VO 2 83.74 13.11 2.68E-04 NULL 1 83.86 13.24 2.52E-04 BAT,SD 2 83.87 13.24 2.51E-04 BAT,STT,VO 3 84.33 13.70 1.99E-04 BAT,BAC 2 84.39 13.77 1.93E-04 BAT,BAH 2 84.50 13.88 1.83E-04 STC,SD 2 84.94 14.32 1.46E-04 SD,VO 2 85.10 14.47 1.36E-04 BAC,SD 2 85.21 14.58 1.28E-04 BAC,BAH 2 85.55 14.92 1.08E-04 BAC,STC 2 85.61 14.98 1.50E-04 BAT,STT,SD,VO 4 85.95 15.33 8.84E-05 BAC,BAH,STC,STH,SD,VO 6 86.28 15.65 7.51E-05 BAC,BAH,SD,VO,STC,STH 6 86.28 15.65 7.51E-05 STC,BAH,SD 3 86.97 16.34 5.32E-05 BAC,BAH,SD 3 87.23 16.60 4.67E-05 35

Table 3-4. Parameter estimates, 95% confidence intervals (CI), and odds ratios (OR) for variables used in supported a priori models to predict nest habitat selection at the patch level of Florida wild turkey hens in south Florida, USA, 2008-2010. 95% CI Model Variable Estimate Lower Upper OR STP,STH STP 0.045-0.020 0.110 1.046 STH -0.001-0.003 0.000 0.999 STP,STT STP 0.048-0.019 0.115 1.049 STT -0.001-0.003 0.000 0.999 STO,SH,SD STP 0.043-0.022 0.108 1.044 STH -175.000-0.004 0.000 0.998 SD -0.008-0.022 0.007 0.993 STP,VO STP 0.055-0.019 0.130 1.057 VO -0.016-0.039 0.006 0.984 STP STP 0.047-0.018 0.112 1.048 STP,STC,STH STP 0.046-0.020 0.112 1.047 STC -0.001-0.003 0.002 0.999 STH -0.002-0.003 0.000 0.998 36

Table 3-5. A priori models, number of variables (K), second-order Akaike s Information Criterion corrected for small sample size (AIC c ), distance from the lowest AIC c (ΔAIC c ), and model weights (w i ) used to predict habitat selection at the landscape level Florida wild turkey hens in south Florida, 2008-2010, USA. Category Model K AIC C AIC C w i Habitat AG,DP,MF,WF 4 205.69 0.00 0.39 AG,DP,MF,WF,SF 5 206.58 0.89 0.25 AF,DP,MF,WF,SF,DS 6 208.32 2.63 0.11 AG,DP,DS,MF,SF,WF 6 208.32 2.63 0.11 DS,MF,SF,WF,AG 5 208.32 2.63 0.11 AG,DP,MF 3 211.76 6.07 0.02 AG,DP,DS,MF,SF,BS 6 213.26 7.57 0.01 AG,DP,DS,MF,SF 5 213.44 7.76 0.01 MF,DP,SF,WF 4 217.38 11.69 1.14E-03 DP,DS,MF,SF,WF 5 219.39 13.70 4.16E-04 AG,MF 2 219.77 14.08 3.44E-04 MF,SF,WF 3 220.53 14.85 2.37E-04 SF,S,SH,MF,WF,DP 6 221.27 15.58 1.62E-04 MF,WF 2 223.07 17.39 6.59E-05 MF,DP 2 224.27 18.59 3.62E-05 MF,SF,DP 3 225.37 19.69 2.09E-05 DS,DP,MF 3 226.20 20.51 1.38E-05 SF,DS,DP,MF 4 227.38 21.70 7.64E-06 DS,MF 2 228.66 22.98 4.03E-06 SF,S,SH,MF,DP 5 229.17 23.48 3.13E-06 MF 1 231.33 25.65 1.06E-06 SF,S,SH,MF 4 231.96 26.27 7.76E-07 SF,DS 2 239.47 33.78 1.81E-08 DP,WF,SF 3 240.50 34.81 1.08E-08 SF,DS,DP 3 240.93 35.24 8.75E-09 DS 1 242.48 36.79 4.03E-09 DS,DP 2 242.79 37.10 3.45E-09 BS 1 247.48 41.80 3.30E-10 DP,WF 2 247.72 42.03 2.94E-10 SF,WF 2 248.55 42.87 1.93E-10 AG,DP 2 251.07 45.39 5.48E-11 AG,WF 2 253.01 47.32 2.08E-11 O 1 256.46 50.77 3.71E-12 WP 1 256.58 50.89 3.50E-12 DP,SF 2 257.90 52.21 1.80E-12 DM 1 264.18 58.49 7.82E-14 SF 1 266.68 60.99 2.24E-14 WF 1 270.19 64.50 3.88E-15 DP 1 273.56 67.87 7.19E-16 PP 1 275.12 69.43 3.29E-16 AG 1 280.37 74.69 2.38E-17 IP 1 280.76 75.08 1.96E-17 H 1 281.34 75.66 1.46E-17 MH 1 283.30 77.62 5.50E-18 SB 1 283.35 77.67 5.36E-18 UHF 1 284.39 78.71 3.19E-18 BG 1 284.55 78.87 2.94E-18 A 1 284.97 79.28 2.39E-18 S 1 287.16 81.47 8.00E-19 37

Table 3-5. Continued. Category Model K AIC C AIC C w i Habitat BF 1 287.43 81.74 7.00E-19 UP 1 288.25 82.56 4.64E-19 X 1 291.19 85.50 1.07E-19 C 1 292.81 87.13 4.74E-20 HH 1 294.59 88.91 1.94E-20 SH 1 295.72 90.04 1.10E-20 O 1 297.58 91.90 4.36E-21 SHF 1 301.29 95.60 6.83E-22 NULL 0 303.44 97.75 2.33E-22 Landscape DROAD,DWATER 2 182.59 0.00 0.69 DROAD,DEDGE,DWATER 3 184.59 2.00 0.26 RD_EDGE,DROAD 2 189.99 7.40 0.02 EDGE,DROAD 2 190.10 7.50 0.02 DRD_DEDGE,WATER 2 191.07 8.48 0.01 ROAD,DROAD 2 193.24 10.64 3.38E-03 ROAD,EDGE,DROAD,DEDGE 4 193.98 11.39 2.33E-03 RD_EDGE,DWATER 2 194.33 11.74 1.96E-03 DRD_DEDGE,EDGE 2 196.53 13.93 6.53E-04 RD_EDGE,DRD_DEDGE 2 196.59 14.00 6.30E-04 DROAD,DEDGE 2 198.20 15.61 2.82E-04 DROAD 1 199.96 17.37 1.17E-04 ROAD,DWATER 2 202.15 19.56 3.92E-05 RD_EDGE 1 202.34 19.75 3.56E-05 EDGE 1 202.77 20.18 2.88E-05 RD_EDGE,DEDGE 2 203.07 20.48 2.48E-05 EDGE,DEDGE 2 203.25 20.65 2.27E-05 DRD_DEDGE,ROAD 2 203.79 21.20 1.72E-05 DWATER 1 204.00 21.41 1.55E-05 RD_EDGE,WATER 2 204.35 21.76 1.30E-05 ROAD,DEDGE 2 213.40 30.81 1.41E-07 ROAD 1 216.03 33.43 3.80E-08 ROAD,WATER 2 217.93 35.34 1.47E-08 DRD_DEDGE 1 218.23 35.63 1.26E-08 DEDGE 1 237.60 55.00 7.87E-13 WATER 1 284.25 101.66 5.82E-23 NULL 0 303.44 120.85 3.97E-27 Management B2,B4,C4 3 170.87 0.00 0.17 B2,C4 2 171.36 0.50 0.15 B2,B3,B4,C4 4 171.89 1.03 0.11 B2,B4,C2,C4 4 172.15 1.28 0.09 B2,B3,C4 3 172.25 1.38 0.09 B2,B3,B4,C2,C3,C4 6 172.48 1.62 0.08 B1,B2,C4 3 173.35 2.48 0.05 B3,B4,C3,C4 4 173.57 2.71 0.04 B1,B2,B3,B4,C3,C4 6 173.77 2.90 0.04 B4,C4 2 173.62 2.76 0.04 B1,B2,B3,B4,C4 5 173.85 2.99 0.04 B3,B4,C4 3 174.20 3.33 0.03 C4 1 174.64 3.78 0.03 B3,C4 2 174.79 3.92 0.02 B1,B4,C4 3 175.63 4.76 0.02 B1,C4 2 176.64 5.78 0.01 C1,C2,C3,C4 4 179.85 8.99 1.94E-03 38