|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 AN EVALUATION OF RED COCKADED WOODPECKER RESTORATION EFFORTS IN THE OCALA NATIONAL FOREST By ELIZABETH RAMIREZ 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 2012
2 2012 Elizabeth Ramirez
3 To my fianc Ben
4 ACKNOWLEDGMENTS First and foremost, I would like to thank those closest to me, my fianc, friends and family, for helping me keep my head together and for encouraging me when I needed it the most. Also, a special thanks to my supervisor, Carrie Sekerak, for supporting me throughout my thesis with her knowledge, guidance and support; to the U nited S tates Forest Service for providing financial support and all the equipment and supplies needed to complete the study; and to Ted Willis who always made sure that my education was taken care of from start to finish. Thanks go out to all of my volunteers that helped collect dat months: Tanya Martinez, Ausley Gage, Seth Cude Ann Sexton and Vanessa Aparicio Most importantly, I would like to offer my sincerest gratitude to my advisor, Dr. Holly Ober, who supported me from beginning t o end of this journey. Without her fantastic encouragement and guidance this thesis would not have been possible.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURE S ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 GENERAL INTRODUCTION ................................ ................................ .................. 12 Preferred Habitat ................................ ................................ ................................ .... 13 Study Area Description ................................ ................................ ........................... 15 Objectives ................................ ................................ ................................ ............... 16 2 SITE SELECTION AND REPRODUCTIVE SUCCESS OF RED COCKADED WOODPECKERS IN OCALA NATIONAL FOREST ................................ ............... 18 Introduction ................................ ................................ ................................ ............. 18 Methods ................................ ................................ ................................ .................. 19 Objective 1: To Determine Which Vegetation Conditions Provide High Quality Habitat for Red Cockaded Woodpeckers in the Ocala National Forest During 2010 ................................ ................................ ....................... 19 Stand scale features ................................ ................................ .................. 20 Landscape scale features ................................ ................................ .......... 21 Mana gement practices ................................ ................................ ............... 22 Spatial characteristics affecting social behavior ................................ ......... 22 Productivity ................................ ................................ ................................ 23 Logistic regression ................................ ................................ ..................... 24 Objectiv e 2: To Evaluate Relationships between Various Habitat Conditions Available on Record Over the Past Ten Years and RCW Productivity .......... 24 Results ................................ ................................ ................................ .................... 26 Objective 1: To Determine Which Vegetation Conditions Provide High Quality Habitat for Red Cockaded Woodpeckers in the Ocala Area During 2010 ................................ ................................ ................................ .............. 26 Objective 2: To Evaluate Relationships between Various Habitat Conditions Available on Record Over the Past Ten Years and RCW Productivity .......... 27 Discussion ................................ ................................ ................................ .............. 29 Objective 1: To Determine Which Vegetation Conditions Provide High Quality Habitat for Red Cockaded Woodpeckers in the Ocala Area During 2010 ................................ ................................ ................................ .............. 29 Objective 2: To Evaluate Relationships between Various Habitat Conditions Available on Record Over the Past Ten Years and RCW Productivity .......... 31
6 Management Implications ................................ ................................ ....................... 34 3 EFFECTS OF TRANSLOCATIONS ON THE RESIDENT POPULATION OF RED COCKADED WOODPECKERS IN THE OCALA NATIONAL FOREST ......... 38 Introduction ................................ ................................ ................................ ............. 38 Methods ................................ ................................ ................................ .................. 40 Objective 1: To Compare the Productivity of Clusters and Number of Potential Breeding Groups in the Vicinity of Translocations before vs after Trans location Events ................................ ................................ ..................... 40 Objective 2: To Determine Cluster Site Fidelity and Regional Site Fidelity of Translocated Red Cockaded Woodpeckers ................................ .................. 42 Objective 3: To Compare Productivity among Birds by Translocation Type (Released Singly or as Pairs), Gender, and Regions ................................ .... 42 Results ................................ ................................ ................................ .................... 43 Objective 1: To Compare the Produc tivity of Clusters and Number of Potential Breeding Groups in the Vicinity of Translocations before vs after Translocation Events ................................ ................................ ..................... 43 Objective 2: To Determine Cluster Site Fidelity and Regional Site Fidelity of Translocated Red Cockaded Woodpeckers ................................ .................. 43 Objective 3: To Compare Productivity among Birds by Translocation Type (Released Singly or as Pairs), Gender, and Regions ................................ .... 45 Discussion ................................ ................................ ................................ .............. 46 Objective 1: To Compare the Productivity of Clusters and Number of Potential Breeding Grou ps in the Vicinity of Translocations before vs after Translocation Events ................................ ................................ ..................... 46 Objective 2: To Determine Cluster Site Fidelity and Regional Site Fidelity of Translocated Red Cockaded Woodpeckers ................................ .................. 47 Objective 3: To Compare Productivity among Birds by Translocation Type (Released Singly or as Pairs), Gender, and Regions ................................ .... 49 Management Imp lications ................................ ................................ ....................... 51 4 USING DATA ON RED COCKADED WOODPECKER HABITAT SELECTION FROM OTHER NATIONAL FORESTS IN FLORIDA TO MAKE INFERENCES ABOUT HABITAT RESTORATION AND SUITABILITY ON OCALA NATIONAL FOREST ................................ ................................ ................................ ................. 60 Introduction ................................ ................................ ................................ ............. 60 Methods ................................ ................................ ................................ .................. 63 Objective 1: To Exam ine Habitat Selection by RCWs in All 3 National Forests in Florida ................................ ................................ .......................... 63 Objective 2: To Map the Location of Habitat Features Selecte d by RCWs in Other National Forests in Florida to the Region Abandoned by RCWs in ONF ................................ ................................ ................................ .............. 65 Results ................................ ................................ ................................ .................... 67 Objective 1: To Examine Habitat Selection by RCWs in All 3 National Forests in Florida ................................ ................................ .......................... 67
7 Coarse scale analysis ................................ ................................ ................ 67 Fine scale analysis ................................ ................................ ..................... 68 Objective 2: To Map the Location of Habitat Features Selected by RCWs in Other National Forests in Florida to the Region Abandoned by RCWs in ONF ................................ ................................ ................................ .............. 69 Discussion ................................ ................................ ................................ .............. 70 Objective 1: To Examine Habitat Selection by RCWs in All 3 National Forests in Florida ................................ ................................ .......................... 70 Coarse scale analysis ................................ ................................ ................ 70 Fine scale analysis ................................ ................................ ..................... 72 Objective 2: To Map the Location of Habitat Features Selected by RCWs in Other National Forests in Florida to the Region Abandoned by RCWs in ONF ................................ ................................ ................................ .............. 74 Management Implications ................................ ................................ ....................... 75 5 GENERAL CONCLUSION ................................ ................................ ...................... 86 APPENDIX A VARIABLES TESTED AGAINST R CW PRODUCTIVITY IN 2010 ......................... 91 B NATIONAL FORESTS OF FLORIDA VEGETATIVE DESCRIPTION .................. 93 LIST OF REFERENCES ................................ ................................ ............................... 95 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 102
8 LIST OF TABLES Table page 2 1 Logistic regression test to differentitate clusters with a potential breeding group (PBG) from those without, and to differentiate clusters that produced one fledgling from those that preduced more than one ................................ ...... 36 2 2 Logistic regression test to di fferentiate clusters with a PBG from those without, and to differentiate clusters that produced one fledgling from those with none ................................ ................................ ................................ ............ 36 3 1 Effects of translocation efforts on productivity of surrounding red cockaded woodpecker (RCW) clusters at two spatial scales ................................ .............. 53 3 2 Number of translocated RCWs that settled at their release site or beyond their release site, total number of RCWs found after translocation and the aver age dispersal distance traveled ................................ ................................ ... 53 3 3 Analysis of variance summary for variables describing distance traveled by translocated RCWs ................................ ................................ ............................. 54 3 4 Status of translocated RCWs based on type of translocation (paired vs single), gender and region (Riverside or Paisley ) ................................ ............... 54 3 5 Logistic regression test to differentiate PBGs producing fledglings from unsuccessful nest attempts ................................ ................................ ................ 54 4 1 Coarse scale variables that differentiated between used and available habitat for RCWs from all national forests in Florida ................................ ...................... 77 4 2 Fine scale variables that differentiated between used and available habitat for RCWs from all national forests in Florida ................................ ...................... 78 4 3 Percent coverage of each ecosystem across all three study areas in Ocala National Forest (ONF) ................................ ................................ ........................ 79 4 4 Number of RCW clusters that could be supported in Church Lake ..................... 79 A 1 Variables tested in chapter 2 for the 2010 breeding season ............................... 91 B 1 Land cover and v egetative description from various agencies ........................... 93
9 LIST OF FIGURES Figure page 1 1 Study areas within the Ocala National Forest (ONF ). ................................ ......... 17 2 1 Map of foraging ecotone buffer within Riverside Island and Paisley Woods ...... 37 3 1 Site f idelity by location and gender ................................ ................................ .... 55 3 2 Productivity after translocation on ONF ................................ ............................. 56 3 3 RCW status after translocation on ONF ................................ ............................. 56 3 4 RCW status by type of translocation on ONF ................................ .................... 57 3 5 Ecosystems and translocation distribution of Riverside Island in the northern region of the ONF ................................ ................................ .............................. 58 3 6 Ecosystems and translocation distribution of Paisley Woods in the southern region of the ONF ................................ ................................ .............................. 59 4 1 Map of the Osceola National Forest (OsNF) with small buffers illustrating the use d vs available habitat for RCW within the forest boundary ........................... 80 4 2 Ecosystem distribution of study area A in the ONF. ................................ ........... 81 4 3 Coarse scale w eighted overlay results of RCW habitat use mapped onto potential RCW habitat areas in study area A. ................................ ..................... 82 4 4 Fine scale w eighted overlay results of RCW habitat use mapped onto potential RCW habitat areas in study area A. ................................ ..................... 83 4 5 Coarse scale results of potential RCW habitat in study area A. ......................... 84 4 6 Fine scale results of potential RCW habitat in study area A. ............................. 85
10 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 AN EVALU ATION OF R ED COCKADED W OODPECKER RESTORATION EFFORTS IN THE OCALA N ATIONAL F OREST By Elizabeth Ramirez August 2012 Chair: Holly Ober Major: Wildlife Ecology and Conservation The relationship between red cockaded woodpeckers ( Picoides borealis ) and their habitat in sandhill of Ocala National Forest (ONF) is not well understood. This unique region is the l arge st continuous forest of sand pine ( Pinus clausa ) scrub ecosystem with islands of longleaf pine ( P. palustris ) wiregrass ( Aristida stricta ) sandhills. While red cockaded woodpeckers ( RCWs ) avoid sand pines, the longleaf pine habitat is ideal for RCWs when managed properly. This study indicate d a positive response in woodpecker reproduction from management implementation s such as 1 2 year rota tion fire regime to minimize the midstory density, annual translocation of single and paired birds to help bolster small populations, continuous treatment of hardwoods through mechanical and chemical applications to reduce midstory heights and installing a rtificial cavities away from scrub ecotones but close to other establish RCWs The most successful translocation s were those conducted with paired, unrelated birds in close proximity to resident RCWs. Future prospects include expanding the ONF RCW population west ward through habitat restoration and translocations The expansion area has vegetation conditions very different from those that occur where the current RCW
11 population s exist in ONF, so o ther national forests in Florida were explored throug h Geographical Information System data layers to find suitable and similar habitat that I estimated this region of ONF could sustain 32 clusters if the habitat was managed properly. Results of this research should aid in prioritizing the type and location of future management actions that will most benefit RCWs in ONF.
12 CHAPTER 1 GENERAL INTRODUCTION The Ocala National Forest ( ONF ) has over 154,000 h ectares (h a ) of forested land The diversity of the forest can be seen in the many ecosystems ranging from sand pine scrub to longleaf pine sandhill, natural flatwoods, live oak hammock s and swamps (Whitney et al. 2004) These ecosystems contain a variety of animals including several threatened and endangered species One key species is the red cockaded woodpecker (RCW) which lives in longleaf pine ecosystems Currently, there are two s eparate healthy populations of RCWs in well maintained ONF habitats A third vacant area once served as habitat for RCWs but is now unsuitable and has not had a potential breeding group since 1989 In addition, nearby lands managed by other agencies has th e potential to be restored for RCW habitat Though there are many species that could be considered when designing restoration plans, the RCWs can be viewed as an umbrella species Animals that would benefit from efforts to restore habitat for RCWs include southeastern American kestrels ( Falco sparverius paulus ) ( Gault et al. 2004), Bachman sparrows ( Aimophila aestivalis) (Conner et al. 2002) pocket gophers (Thomomys talpoides), fox squirrel ( Sciurus niger shermani ) and gopher tortoises ( Gopherus polyphemus) ( Myers and Ewel 1990). Restoring habitat for RCW s could create large areas that also provide improved habitat for these other species The long term success of RCW clusters is dependent upon many variables including the distance to oth er clusters. The farther a part cluster s are from each other, the less stable the population is ( Letcher et al. 1998, Schiegg et al. 2002 a ). With several close clusters to emigrate to, the birds have the ability to search for new vacancies If
13 other cluster s are too far, then the clusters are at risk of becoming inactive (Engstrom and Mikusinski 1998) Additionally, translocating individual or paired birds from other public lands to vacant clusters in ONF can aid in the long term success of the species in ON F but is an expensive alternative for bolstering isolated populations. Red cockaded woodpeckers face many threats that may lead to a decline in their numbers For instance, cavities can be usurped by other animals such as flying squirrels ( Glaucomys sabrin us) snakes, eleven species of birds, several types of bees and the broad headed skink ( Eumeces laticeps) (Whitney et al. 2004) A southern pine beetle ( Dendroctonus frontalis ) epidemic can have catastrophic damage on an entire forest including wiping out half the nesting trees as seen on the Sam Houston National Forest (NF) (Conner et al. 1991) In order to secure RCW population s and their habitats, these threats should be assessed and m anaged whenever possible. Preferred H abitat Red cockaded woodpeckers prefer pine stands that have park like characteristics These include older slash or longleaf pine stands of at least 60 years of age that have trees with heart rot fungus, understory with little or no hardwood encroachment, and that experience frequent low intensity fires Fire is an essential component becaus e it enhances the reproduction, growth, and maintenance of longleaf pine and native ground cover species while reducing hardwoods in the understory ( Saenz et al. 2001) The condition of the ground cover influences the abundance of prey (arthropods) for RCW s found on pine trees (Hanula and Franzreb 1998). These birds are unique in a sense that they are the only woodpeckers in Florida that nest in live trees (Whitney et al. 2004) As trees age, they may eventually become exposed to heart rot which softens an d decays the center of the tree This makes
14 excavating a cavity easier for the birds (Hooper et al. 1991) Cavities can take six months to several years to co mplete (Whitney et al. 2004) Artificial cavities can aid RCWs by reduc ing time spent excavating n ew cavities and increas ing time spent foraging for food Franz r eb (2006) suggests the foraging area for RCW clusters are around 60 ha Their diet includes spiders, ants, roaches, centipedes, and insect eggs that they find by foraging on tree bark (Whitney et al. 2004) The United States Fish and Wildlife Service (USFWS) wrote the second recovery plan for RCWs in 1985 which stated, "The survival of the red cockaded woodpecker ultimately depends on halting the loss of nesting habitat and providing adequate a creage in old growth pines in perpetuity Merely protecting existing clusters will delay extinction but not prevent it A continuing supply of old growth habitats is required to replace clusters lost or abandoned and to provide for population expansion" ( USFWS 1985) One recommendation to achieve this goal is to use a combination of site analysis, historic background from the project site, and information from nearby reference sites (Walker and Silletti 2006) Examples of current restoration treatments are mowing, chopping, thinning, installing artificial cavity inserts, planting pine seedlings, seeding groundcover chemical applications and most importantly, prescribed burning The most common treatment with longleaf pine restoration is the removal of har dwoods Hardwood reduction can be accomplished with repeated fire and mechanical treatments such as mowing or roller chopping, a chemical application such as herbicides, or the combination of all three Restoration has to be considered at multiple levels i n the forest including the canopy of pines and the ground level vegetation (Walker and Silletti 2006).
15 Study Area Description This study was conducted in Ocala National Forest which is located in Lake, Marion and Putman counties in central Florida The O NF has two distinct sub population s of red cockaded woodpeckers (Study areas B and C, Figure 1 1) and is planning to establish another RCW population in a third region (Study area A, Figure 1 1). The U nited States (USFS) Global Positioning System ( GPS ) database shows historical RCW nesting trees in study area A. Unfortunately, there has not been an active potential breeding group there since 1989 and reasons for abandonment are uncertain Generally, reasons for RCW abandonment are harvestin g of old tree s, years of fire suppression, failure to establish longleaf pine seedlings, and converting native species of pine to more fast growing timber. Stand records in ONF show historical RCW stands were re planted with mostly longleaf pines and some areas with slash pines in the 1920s. F ire suppression and over harvesting are likely causes of abandonment in the western region of Ocala National Forest. Study area A ( F igure 1 1 ) covers roughly 20,500 ha and is located on the west side of the ONF It is located west of Highway 314A and covers land both north and south of State Road 40 Figure 1 1 shows the location of historic al trees with black dots in study area A Study area B contains active red cockaded woodpecker clusters ( black dots ) in approximat ely 6,200 ha of forested sandhills located in the northern region of the forest around Lake Kerr, also known as Riverside Island. The southernmost active clusters are in approximately 5,30 0 ha of forested sandhills knows as the Paisley W oods (study area C) According to forest receives about 15 5 32 centimeters (cm) (X SD ) of rainfall annually and about 40% of that falls during the summer months of May, June and July (6 0 17 cm ) These
16 are the months when RCWs most commonly raise young in nest trees. daily temperature average d between 59 1 Fahrenheit (F) and 84 F but during the summer months average d between 68 1 F and 92 1 F Objectives The goal of this multi faceted study was to assess the efficacy of past management actions and anticipated future RCW restoration regimes for the ONF. I determined which habitat features and management actions have the greatest effect on RCW nest success. In addition, I investigated data collected over a ten year period in study areas B and C to assess the benefits of translocations. I also evaluate d the potential of the western area to serve as additional habitat for this endangered species The models I develop ed help identify the most beneficial restoration regime for study area A to create a third sub population and build a sustainable population of RCWs. Results of this study have the potential to inform future management decisions on ONF, by investigating which management actions have had the g reatest positive influence on RCW productivity the past 10 years, and identifying locations most suitable for RCWs in the region of the ONF that RCWs have abandoned. This research has the potential to aid in prioritizing the type and location of future man agement actions that will most benefit RCWs. The re were three specific research questions: (1) Which habitat characteristics and management activities have historically provide d high habitat quality for RCWs in the Ocala National Forest ? (2) How beneficial have the translocations of RCWs in Ocala National Forest been to the resident populations? ( 3 ) Can information on RCW habitat selection from other National Forests in Florida be used to prioritize the location of restoration efforts in Ocala National Forest?
17 Figure 1 1 Study areas within the Ocala National Forest include the his toric RCW clusters in Church Lake region (study area A) on the west side of the forest, current RCW clusters in Riverside Island (s tudy area B) located in the north and current RCW clusters located in Paisley Woods in the south (study area C).
18 CHAPTER 2 SITE SELECTION AND REPRODUCTIVE SUCCESS OF RED COCKADED WOODPECKERS IN OCALA NATIONAL FOREST Introduction The red cockaded woodpe (RCW) range currently spans from southeast Oklahoma to Virginia and from the Gulf Coast of Texas to South Florida (Ligon et al. 1986) with isolated pockets of potential breeding groups (PBGs). In 2000, an estimated 14,000 birds or about 5,600 group s occurred across this range which is approximately 3% of their original abundance (Chadwick 2004) The recovery plan provides recommendations on habitat management, but this information is based on range wide averages across the 11 states in the S outheast (USFWS 1985 ) At the edge of the species range, birds may behave differently, and with global climate change these individual s at the edge of the range are of vital importance due to behavioral changes that could affect fledgling success (Schiegg et al 2002 b ) T his study provides insight into the biology of RCW at the edge of the species range, in Ocala N ational Forest ( ONF ) The recovery plan is an excellent source for habitat management, but this information is not tailored to the Ocala N ationa l Forest I undertook this study to determine which habitat characteristics and management procedures contribute most to woodpecker productivity in ONF. In particular, I evaluated landscape scale features, stand scale features, management practices, and sp atial characteristics affecting the social behavior and reproductive success of the birds I developed two objectives to determine which factors were most influential in determining the productivity of red cockaded woodpeckers in the ONF
19 Objective 1: T o determine which vegetation conditions provide high quality habitat for r ed c ockaded w oodpecker s in the Ocala National Forest during 2010 Objective 2: To Evaluate Relationships Between Various Habitat Conditions Available on Record Over the Past Ten Years a nd RCW Productivity Methods Objective 1: To Determine Which Vegetation Conditions Provide High Quality Habitat for R ed C ockaded W oodpecker s in the Oc ala National Forest During 2010 N inety one active and inactive clusters were recorded on the Ocala National Forest in 2010 In ArcMAP (v 9.3.1, 2009) I drew a buffer with a radius of 0. 4 kilometers ( km ) around the centroid of each of these clusters to create an area totaling approx imately 5 0 hectares ( ha ) to characterize foraging habitat for each cluster (USFWS 1985 ) For each of the active clusters, the nest tree was considered the centroid point For the inactive clusters, the centroid points were centered on either a previously active tree or if there were no previously active trees, a suitable tree with a cavity was selected at random B uffers were uniquely identified by forest compartment and cluster number (i.e., 260 1) I selected 48 of these clusters with a stratified random selection to capture a wide range in RCW productivity and fire frequency (described below). Historical p roduct ivity was estimated for each of the 91 cluster s by calculating the number of fledglings recorded divided by the number of years a cluster was monitored I used data for each of the seven available years between 2001 and 2010 (2006, 2008 and 2009 were excl uded due to missing data) All clusters with a score of zero for l ere divided in half, giving the h h m l historical productivity for each cluster.
20 The fire frequency inside each of the 91 buffer s over the last ten years was determined in ArcMAP by drawing polygons around all fires every year from 1990 to 2009 Analysis Tools (v 3.27, 2007) I calculated the proportion of each buffer burned in a given year burned in a given year, regardless of season, I considered the buffer to be burned that year Buffers were placed into three fer burned 4 After RCW productivity and fire frequency was calculated, I selected 48 clusters, 16 from each of the high, medium, and low historical productivity categories and from 12 to 21 clusters in each fire frequenc y category so that each productivity x fire frequency combination was represented by 2 to 8 clusters Some categories were limited in the number of clusters and therefore all available clusters were used Other categories had an abundance of clusters to ch oose from so a random selection was drawn This stratified random selection ensured that I evaluated clusters subjected to a wide range of fire frequencies, and that experienced a wide range in productivity. Stand scale features I measured 10 stand scale h abitat features to characterize vegetation for each cluster Using the random point generator in Analysis Tools for ArcMAP I 5 m spacing between points Points were loaded into a Trimble Geo XT 20 08 so I could locate each Global Positioning System ( GPS ) coordinate to 3 meters ( m ) accuracy in the field At each point I measured stand level variables to represent the over story, mid story, and ground cover characteristics of the forest stand
21 For th e over story data, I used a variable radius plot with an imperial prism with basal factor 10 (1 0 f ee t 2 /ac re ) For each tree, I recorded DBH to 0.2 5 centimeters ( cm ) accuracy and species Trees per acre were derived for any Pinus elliottii and P. palustris tree s that were 3 5 cm Mid story vegetation data w ere collected using fixed area plots of 1/12 5 ha ( 5 m radius) All stems 5 1 2 cm in diameter were counted by species while a ny oak stems < 5 cm were ignored since the ir density changes from year to year due to frequent fire regimes Oak basal area was combined from the overstory and understory data to give a total basal area for oak species mid story was categorized as dominated by either oak or pine or neither, based on the findings using fix ed area plots ( FAP ) Finally, ground cover was measured using 1 m x 1 m quadrat s at locations 5 m to the north, south, east, and west of the central point and then averaged In each quadrat I recorded percent cover of six ground cover categories: wiregrass herbaceous, shrubs <1. 5 m in height, shrubs >1. 5 m in height, pine litter, and bare ground. Landscape scale features I measured three landscape scale variables: patch size, proportion of interior foraging buffer vs edge, and heterogeneity of the vegetative community in the foraging buffer Patch size was calculated based on the area of each isolated sandhill island within the sand pine scrub ecosystem using the United States Forest Service ( USFS ) Geographical Inform ation System ( GIS ) 4 km interior (>0. 4 km 4 km from scrub edge) (Figure 2 1) Within each RCW clust
22 in ArcMAP to calculate the proportion of edge To evaluate the heterogeneity of the vegetative community around each cluster, I uffers with the ecosystem layer to determine the proportion of acres of non sandhill habitat to the total acres. Management practices I assessed five variables pertaining to habitat management practices near each cluster : number of prescribed fires during the previous 10 years, number of years since the last prescribed fire number of years since a mowing or oak removal, number of artificial cavities installed, and the ratio of artificial cavities to natural cavities The USFS prescribe d fire history is col lected in polygon form for each fire To count the number of fires in Analysis each fire polygon with the buffer polygon and used excel to calculate whether 30% or 1 5 ha was burn ed in each event In addition, I used th ese data to find the year the buffer was last burned Forest Service records were used to determine the year of the last mow or hardwood removal treatment Hardwood treatment included mechanical removal of hardwoods using chainsaws, roller chopping or herbicide application. The artific i al cavity data are maintained in the attributes table of the USFS RCW layer in ArcMAP The count and proportion of cavities w as calculated in excel. Spatial characteristics affecting social behavior I determined four variables pertinent to the spatial characteristics affecting the social behavior of the birds: distance to the nearest active or inactive cluster number of active or inactive clusters with overlapping foraging areas percent of foraging area overlapped by others, and occurrence of translocation To find the distance to the
23 nearest neighbor, I used c Analysis T Nex Analysis Tools, I counted the number of overlapping buffers and proportion of overlap for each buffer Finally, I counted the presence or absence of the translocations in new clusters during the winter before the 2010 breeding season. Translocation is an opportunity to bolster small populations by relocating surplus birds from large populations to smaller declining population Productivity The 2010 breeding status and productivity of RCWs were used as the response variable s in logistic regression. I analyzed the data three separate ways: (1) to identify variables differentiat ing clusters with potential breeding groups from those without; (2) to identify variables that differentiated successful c lusters from unsuccessful clusters ; and (3) to identify variables that differentiated clusters that produced one fledgling from those that produced two The first assessment identified variables most strongly associated with the presence or absence of pote ntial breeding groups in the 48 selected clusters The second assessment included only the 35 clusters with potential breeding groups and identified variables most strongly associated with the presence or absence of fledglings The final assessment included data from only those 19 clusters that were successful in 2010 (successful clu sters are those that produced an offspring that survived at least 26 days) The response variable for the third assessment the fledgling count, showed a 1 or 2 representing the number of successful fledgling s in the nesting season
24 Logistic regression All data were analyzed with SAS (v 9.2, 2008) I conducted logistic regression using the variable selection technique recommended by Ho s mer and Lemeshow (2000) I first noted all variables with p values < 0.25 when examined in a univariate context (each variable was independently regressed on the dependent variable) Variables with p values < 0.25 were considered candidates for model inclusion Second, to elimina te the possibility of including redundant metrics in the same model, I determined which 6 0) and therefore should not occur in the same model Third I performed a backward selection in logistic regr ession with all remaining variables until the model contained only variables with Additionally, I evaluated the goodness of fit of the final model using Hosmer and Lemshow (2000) goodness of fit tests and Cox and Snell pseudo R square index to ass ess model fit (Cox and Snell 1989) Objective 2: To Evaluate Relationships between Various Habitat Conditions Available on Record Over the Past Ten Years a nd RCW Productivity The second part of the analysis used the data collected over the past 10 years among all clusters ( n = 91) The stand scale features were not available over the ten year period and were therefore not considered for this analysis. The landscape variables were the same as described for the analysis of 2010 data : patch size, proportion of foraging buffer within 40 0 m of the edge of the ecosystem and the proportion of the buffer containing ecosystems other than sandhills ( heterogeneity of the vegetative community ). This information was derived from GIS polygons created by the Forest Ser vice GIS Specialist, Kathy Bronson Management practices were also similar to those described in the analysis of 2010 data : the number of years since last burn in
25 30% of each 0. 4 km buffer surrounding a cluster and the number of times a buffer was burned in the previous 10 years Spatial characteristics affecting social behavior were the same as the analysis of 2010 data : distance in meters to the nearest cluster (active or inactive), number of buffers within 0. 4 km of each cluster the percent of each buffer that overlapped with other buffers the occurrence of a translocation, and the count of translocations per cluster per year. In addition, I assessed annual and summer rainfall Precipitation data w ere gathered on a daily basis at a weather stat ion at the F orest S ervice central location, from which I obtained measurements from January 2001 to December 2010 I calculated two precipitation characteristics: average daily precipitation ( a single average was used to represent each year ) and average da ily precipitation during the breeding season (May July) (again, a single average was used to represent each breeding season ) The breeding status and productivity of RCWs from 2001 to 2010 (2006, 2007, and 2009 were unavailable) were used as the response variable in the generalized linear mixed models (Glimmix). As described in objective one, I analyzed the data three separate wa ys: (1) to identify variables differentiating clusters with potential breeding groups from those without; (2) to identify variables that differentiated successful clusters from unsuccessful clusters; and (3) to identify variables that differentiated cluste rs that produced one fledgling from those that produced more than one. All data were analyzed with SAS (v 9.2, 2008) I conducted the analysis with Glimmix using the variable selection technique recommended by the College of Agriculture and Life Sciences (CALS) statistician James Colee The Glimmix procedure uses the Kenward Roger Degree of Freedom method which accounts for repeated
26 measures I first noted all variables with p values < 0.25 when examined in a univar iate context (each variable was independently regressed on the dependent variable) Variables with p values < 0.25 were considered candidates for model inclusion Second, I reviewed continuous variables with large ranges with lsmeans to find ways of organi zing data into a small number of biologically meaningful categories Third, to eliminate the possibility of including redundant metrics in the same model, I determined 6 0) and therefore shoul d not occur in the same model Finally, I performed a manual backward selection in Glimmix with all remaining variables by entering all variables and removing the variable with the highest p value and repeating Res ults Objective 1: To D etermine W hich V egetation C onditions P rovide H igh Q uality H abitat for R ed C ockaded W oodpecker s in the Ocala National Forest D uring 2010 In the first assessment differentiating clusters with potential breeding groups from clusters without potential breeding groups ( n = 48 ) the Hosmer and Lemeshow (2000) goodness of fit test indicated adequate model fit ( 2 = 6.39, df = 8, P = 0.6032) and the Cox and Snell R 2 = 0.36 T here were three features that were most useful in the final model : hardwood basal area (ft 2 /ac) ( P = 0.0026), % cover of bareground ( P = 0.0578), and % cover of shrubs >1. 5 m ( P = 0.0119), (Table 2 1) Results indicate that higher hardwood basal area and % bareground were associated with a lower likelihood of having an active cluster. On the other hand, higher % cover of shrubs >1. 5 m was associated with a higher likelihood of having an active cluster. In the second assessment ( n = 35 ) no variables were able to differentiat e successful clusters from unsuccessful cluste rs
27 The third assessment, differentiating clusters with one fledgling from clusters with two or more fledglings ( n = 19 ) the Hosmer and Lemeshow (2000) goodness of fit test indicated the model fit the data ( 2 = 9.06, df = 8, P = 0.3375) and the Cox and Snell R 2 = 0.45 T here were two features in the final model : % cover of shrubs <1. 5 m ( P = 0.0912) and distance to nearest neighbor ( P = 0.0 340) ( Table 2 1) F arther distance s from the nearest cluster were associated with higher likelihoods of a cluster produc ing more than one fledgling Also, higher % cover of shrubs <1. 5 m were associated with a greater likelihood that a cluster will produce more than one fledgling Objective 2: To Evaluate Relationships between Various Habitat Conditions Av ailable on Record Over the Past Ten Years a nd R C W Productivity The Pearson correlation test found strong correlations between proportion of non sandhill ecosystems (non sandhill) and proportion of interior sandhill ecosystems (interior) The proportion of interior was selected for model inclusion due to its known biological relevan ce to RCW reproduction ( Carl Petrick, personal communication) In the first assessment, differentiat ing clusters with potential breeding groups from those without ( n = 91 ), the close proximity of the generalized chi square statistic to 1( 2 = 1.02) indicates the variability in the data was properly modeled. T he model found four features helpful in differentiating these clusters: interior ( P = 0.0641), annual rainfall ( P = 0.0138 ), number of prescribed fires ( P = 0.0151), and number of translocations ( P = 0.0573) (Table 2 2 ) Results show that increasing the proportion of interior habitat and increasing the number of prescribed fires over a ten year period was associated with an i ncreased likelihood of having a PBG For every 10% increase in the proportion of the buffer that was interior rather than edge, there was a 9% increase in the odds of having a PBG Buffers where prescribed fire was applied 4 5 times per decade were almost
28 twice as likely to have a PBG than those burned only 1 3 times per decade, and were more than twice as likely to have a PBG when compared to those with more than 6 prescribed fires per decade T he model also showed that high annual rainfall was associated with a decreased likelihood of finding a PBG. Lastly, t he odds of having a PBG were five times greater when a single bird was translocated to a resident bird than when paired birds were translocated. There was no significant difference in the likelihood of having a PBG in areas where a single bird was release d and areas were no translocation s occurred The second assessment compar ing clusters successful at raising fledglings with clusters that failed to produce fledglings ( n = 66 ) the close proximity of the generalized chi square statistic to 1( 2 = 1.02) indicates the variability in the data was properly modeled Two features were significant in the final model : interior ( P = 0.0062) and number of years since the last prescribed hardwood treatment ( P = 0.0751) (Table 2 2 ) Results show increasing the proportion of interior habitat and the number of years since the last hardwood treatment was associated with an increased likelihood of having a successful PBG. For every 10% increase in the proportion of th e buffer that was interior rather than edge, there was a 14% increase in the odds of having a successful PBG. The trend for hardwood treatment is small : the relative odds of a potential breeding group being success ful increased by only 6% with every year since the last hardwood treatment. The final assessment differentiat ing between clusters with 1 fledgling and clusters with more than 1 fledgling ( n = 53 ) found no significant variables.
29 Discussion Objective 1: To D etermine W hich V egetation C ondi tions P rovide H igh Q uality H abitat for R ed C ockaded W oodpecker s in the Ocala National Forest D uring 2010 The features that differentiated clusters with potential breeding groups from those that were inactive were hardwood basal area, % cover of bare groun d and % cover of shrubs >1. 5 m Increasing the % cover of bare ground and hardwood basal area was associated with decreased odds of having a potential breeding group in a cluster. This corresponds to the well known relationships between RCW success and fre quent fire ( Ligon 1970, Robbins and Myers 1992, Costa 1995 ) Frequent fires help to develop native groundcover, create open canopies, reduce midstory oak and other hardwood species, and promote longleaf pine dominated overstory (Allen et al. 2006). High percentage of bare ground can disconnect the ground fuels and cause a mosaic burn pattern, leaving opportunity for hardwoods to encroach During fire suppression, plant succession will allow hardwoods to choke out longleaf and other groundcover species (Lo eb et al. 1992, James et al. 1997) The shorter the interval between fires the less chance hardwood s have to reach a size and density that becomes problematic to RCWs The RCW requires a low density of trees in the mid story (Loeb et al. 1992), which prov ides the birds the ability to protect their cavities from predators by giving them a clear view of potential danger On the other hand, my result indicating that increasing the percent cover of shrubs >1. 5 m was associated with an increase in the odds of having a potential breeding group is in congruence with some but not all prior studies Loeb et al. (1992) found midstory height was higher among active clusters vs inactive clusters while basal area and stem/ac were both lower The recovery plan states th at nesting cavities should have about 1 5 m of clearance around each cavity
30 tree and that midstory heights should stay below 4 m ( USFWS 1985 ) Clusters in ONF had a very limited range in percent cover of shrubs >1.5 m: 0 20% (Appendix A) Therefore, my r esult may illustrat e how birds may tolerate taller shrubs but only at low stem densit ies Another study found that hardwood midstory heights greater than 4 m had a profound negative effect on use by RCWs (Walters et al. 2002) My analysis only review ed shrubs in the ground cover layer most of the ground cover species in ONF do not grow in excess of 2 4 m and I did not in clude trees with diameters over 5 cm as they would be captured in the midstory layer Therefore, I cannot make inferences about the e ffects of the height of the midstory on RCWs in ONF The features that differentiated those clusters that were successful at producing two RCW fledgling s from those that produced only one were distance to nearest neighbor and % cover of shrubs <1. 5 m Inc reasing the distance to the nearest neighbor was associated with an increase in the odds of having two rather than one fledgling This suggests that competition may affect the success of the survival of offspring Ligon (1970) found similar results while e xtensively monitoring several pairs of RCWs near Gainesville, Florida In addition, this could suggest that more successful clusters have a larger home range size as well A similar study found that successful RCW groups used larger home ranges than unsucc essful groups (Hardesty et al. 1997) Increasing the percent cover of shrubs < 1. 5 m also was associated with increased odds of having two rather than one fledgling Other studies have reported similar patterns. Davenport et al. (2000) found highest RCW p roductivity in clusters with understory below 1.8 9 m and lowest productivity in clusters with understory above 3.2 6 m Hardesty et al. (1997) also found that smaller and shorter hardwoods increased the success rate of groups
31 However, another study found h ardwood encroachment lead to higher fledgling predation (Lennartz and Heckel 1987). Further research should be conducted on how height and density of oak species in the groundcover layer effect the habitat RCWs select or avoid. Objective 2: To Evaluate Rel ationships between Various Habitat Conditions Available on Record Over the Past Ten Years a nd RC W Productivity The features that differentiated clusters with potential breeding groups from those that were inactive over the last ten years were the proportion of interior habitat, annual rainfall, number of prescribed fires over the past ten years and the number of bi rds translocated to a cluster per year Increas ed proportion of interior habitat per cluster was associated with an increase in the odds of having a potential breeding group Increasing the amount of annual rainfall was associated with a decrease in the od ds of having a potential breeding group The translocation of 1 bird per year was associated with an increase in the likelihood of having a potential breeding group compared to translocating 2 birds per year but was similar to not translocating at all Pre scribed fires had a positive impact; an increase in the number of fires within the past decade was associated with an increase in the odds of having a potential breeding group The landscape variable that was found to be significant was proportion of inte rior Results indicate that PBGs were associated with sandhill habitat de void of sand pine scrub This is likely because sandhills typically have an open canopy with widely spaced pines and little to no midstory, whereas sand pine scrub has closed canopy a nd dense midstory, making it unsuitable habitat for nesting or foraging ( Myers and Ewel 1990, Walters et al. 2002) The selection of interior habitat over edge indicates that birds preferred sites where foraging buffers would contain no scrub Eglin Air Fo rce base has
32 seen similar patterns in RCW territories that are located on the edge of habitat boundaries; these edge clusters switch more frequently between status of active and inactive than do those clusters with several groups surrounding them and away from the edge (2010, personal communication with Carl Petrick) Similarly, active clusters were surrounded by less total edge habitat and more uniform forest cover than inactive clusters in the Red Hills RCW population located on private land in north Flor ida and south Georgia ( Cox 2001) The environmental variable that was found to be significant was annual rainfall Heavy rainfall is detrimental to productivity because it restricts the parents from flight and reduces the amount of feeding time, which is especially detrimental in the month of May when rainfall can have the most negative impact on parents with young (Conner et al. 2005) In Arkansas, helpers (other RCWs) were able to offset this negative impact of rainfall on nestlings (Neal et al. 1993) but comparable data are not available to examine these social patterns in more detail on the ONF. Other forests in Florida have seen similar declines in fledge suc cess during summers with heavy rain such as the Osceola National Forest in Lake City (2011, personal communication with Sarah Lauerman). It is worth noting that the trend I found may not be biologically meaningful because precipitation data was not availab le from multiple locations throughout each of the study areas. The management variable that was found to be significant was the number of times a foraging buffer was burned over the last ten years I found that burning areas 4 5 times over ten years was be tter than burning only 1 3 times, and burning more than 6 times was most beneficial This represents an average of 1 3 year burn regime
33 Previous research has found habitat areas with fire suppression were avoided by RCWs ( Provencher et al. 2002 Allen e t al. 2006) It may be the case that birds selected more frequently burned areas because an increase in fire frequency leads to more successful clusters by increasing the arboreal arthropod density on the boles of trees and reducing the basal area of hardw oods (Taylor and Walters 2004) providing higher food availability and ideal habitat structure The social variable that was found to be significant was number of birds translocated in to a cluster per year The benefits of translocations demonstrated in ot her regions include expanding small populations reducing potential adverse genetic consequences of small population sizes and aiding in recovery after a catastrophic event (Franzreb 2004) The results show that PBGs were more likely to be associated with translocations of an individual bird to a local solitary bird that already has an established territory than to translocations of an unrelated pair These results are similar to those reported in other studies (Edwards and Costa 2004, Walters et al. 2004) However, this result may be spurious for several reasons. First, RCW s that were translocated singly were paired with a resident bird, which gave them a higher chance of creating a potential breeding group in that same location shortly thereafter, wherea s birds released as a pair into a vacant area were less likely to both be detected the following year. In addition, the Glimmix model only considered those birds that formed a potential breeding group within the breeding season directly following transloca tion. Therefore, the model is limited by the fact that in some cases of paired releases clusters retained a single translocated bird and formed a potential breeding group after the breeding season directly following translocation.
34 In t he second part of this analysis the features that differentiated clusters that successfully produced one fledgling from those that failed to produce a fledgling over the last ten years were the proportion of interior habitat vs edge and the number of years since the last h ardwood treatment Similar to the previous findings, the odds of having at least one fledgling produced in a cluster was associated with an increase in the proportion of interior habitat Surprisingly, the odds of having at least one fledgling produced in a cluster increased as the number of years since the last hardwood treatment increased A limited number of hardwood treatments ( n = 20) were completed within active and inactive clusters during the study period Therefore, the data had a large number of n on treated areas which may have driven the results to a negative response when comparing successful vs unsuccessful clusters. Upon closer inspection of the data, I found that 47% of all active clusters that received hardwood removal treatments (7 of 15) sh owed an improvement in productivity within 8 years of application of which 4 clusters show ed improvement within one year of treatment. This suggests that the way I chose to analyze the data may have been the reason for the unexpected result Additionally research on the effects of hardwood treatment over time on RCW productivity is suggested. Management Implications To provide quality habitat for potential breeding groups of RCWs, managers should strive to reduce hardwood basal area and reduce coverage o f bare ground, implement prescribed fires on a 1 2 year rotation, and consider translocations of single birds to resident birds in interior sandhills habitat To improve the success of potential breeding groups, active management to promote RCW habitat sho uld be focused in areas that will minimize the amount of scrub ecotone and provide the maximum
35 distance between active neighbors, and utilize hardwood treatment in areas where fire may be inappropriate to maintain shrubs below 1. 5 m RCW habitats in study areas B and C on the ONF have been consistently burned on a 1 2 year rotation, effectively controlling midstory basal area An alternative to fire is hardwood treatment when areas have long been devoid of fire which provides a controlled approach to red ucing hardwood basal area and density but is significantly more expensive. Translocation has had a successful reputation in the Southeast and has shown to be successful in ONF. My research suggests that the translocation of a single bird to a resident bird is more productive than the introduction of a pair or none. This study explored several RCW habitat characteristics; some which can be managed for while others are dependent on the behavior of the birds M any of the stand scale features significant in my analyses of RCW site selection and productivity in 2010 in the Ocala National Forest aligned with the RCW recovery plan habitat guidelines. The ONF has a healthy and productive, albeit isolated habitat that is managed primarily for RCWs. Intensive mana gement practices (annual translocation, frequent prescribed fires, and cluster augmentation that avoid scrub ecotones) over time have helped maintain the population
36 Table 2 1 Parameter estimates standard error ( SE ) and adjusted odds ratios with 95% confidence intervals (CI) from the best fit models using backward logistic regression in predicting probability of a cluster hav ing a potential breeding group (PBG) or not in the first assessment and predicting the probability of producing one or more than one fledgling in the second assessment within the 2010 breeding season Adjusted odds ratio Year 2010 a ssessment Parameter Parameter estimate SE Estimate 95% CI P PBG vs n on PBG Hardwood basal area (ft^2/ac) 0.41 0.14 0.66 0.51 0.87 0.0026 % cover bareground 0.09 0.05 0.92 0.84 1.00 0.0578 % cover shrubs >1. 5 m 0.67 0.27 1.95 1.16 3.28 0.0119 1 fledgling vs > 1 fledgling Distance to nearest neighbor (m) 0. 0 1 0. 0 1 1. 0 1 1.00 1. 0 3 0.0340 % cover shrubs < 1. 5 m 0.33 0.19 1.39 0.95 2.03 0.0912 Table 2 2 Parameter estimates SE, and adjusted odds ratios with 95% confidence intervals from the best fit models using backward logistic regression in predicting probability of a cluster hav ing a potential breeding group (PBG) or not and the RCW productivity having at least one fledgling vs none within the 2001 2010 breeding season s Adjusted odds ratio 10 year a ssessment Parameter value Parameter estimate SE Estimate 95% CI P PBG vs Non PBG Interior 0.90 0.48 2.46 0.95 6.37 0.0641 Annual Rainfall 5.51 2.23 0.00 <0.001 0.32 0.0138 0.79 0.36 2.19 1.07 4.48 0.0311 Rx fire count 4 5 0.57 0.21 1.77 1.18 2.65 0.006 0 Rx fire count 1 3 . Translocate0 0.74 0.58 0.48 0.15 1.48 0.1992 Translocate 2 1.53 0.69 0.22 0.06 0.84 0.0267 Translocate 1 . PBG s uccess vs failure Interior 1.37 0.49 3.93 1.49 10.38 0.0062 H ard w oo d Trt mnt 0.06 0. 03 1.06 0.99 1.13 0.0751
37 A B Figure 2 1. Map showing a 0. 4 km foraging buffer along the inside edge of each sandhill 4 km from scrub 4 km from scrub edge). A) Riverside Island. B) Paisley W oods.
38 CHAPTER 3 EFFECTS OF TRANSLOCATIONS ON THE RESIDENT POPULATION OF RED COCKADED WOODPECKERS IN THE OCALA NATIONAL FOREST Introduction The red cockaded woodpecker (RCW) is a federally listed endangered species in forests of the southern Unites States Longleaf pine savannah habitat for this species was harvested in the 1800s and early 1900s and reduced to about 1% of its original range, resulting in a reduced population of RCWs of 3% or about 5,600 groups (Chadwick 2004) One of the management practices listed in the Red cockaded Woodpecker Recovery Plan is to increase the number of potential breeding groups through translocations (USFWS 1985). The re are four reasons the recovery p lan recommends translocations : (1) augmenting small populations below the critical threshold of 30 p otential b reeding g roups (PBG) (2) grow ing current population s to reduce isolation of groups, (3) releas ing birds to suitable habitat within historic ranges, and (4) management of genetic resources ( USFWS 1985 ). Translocations have been very successful at increasing population s izes by increasing potential breeding groups at recipient sites while having minimal effects on the donor populations ( Saenz et al. 2002 Herbez et al. 2011). Although augmenting individual birds from a densely populated area to a sparsely populated group is an important management strategy to help stabilize small pop ulations, it is an expensive strategy The estimated cost of transporting a pair of birds, according to the supervisory biologist on the Ocala National Forest (ONF) Carrie Sekerak (2010) is around $4500 This cost estimate includes inst allation of 8 arti ficial cavities or 2 clusters designated for the pair; man hours to scope out fledgling success,
39 the recipient location; and monitoring fledge success Given the high c ost of this management activity, it is important to measure the impact on the recipient populations to determine if the cost is justifiable The Ocala National Forest currently has two distinct populations of red cockaded woodpeckers The largest populati on of birds is located in the northern region of the forest known as Riverside Island and the other population (Paisley Woods) is l ocated in the south (Figure 1 1) In an effort to continue increasing the number of potential breeding groups juvenile red c ockaded woodpeckers are translocated from public lands with a surplus of birds to unoccupied territories on other public lands that are below the critical threshold of 30 potential breeding groups Every year since 19 87 the Ocala National Forest has been receiving birds from different public lands such as Camp Blanding near Starke, Florida ; Fort Stewart, Georgia ; and Francis Marion National Forest, McClellanville, South Carolina In 1993 records show a total of 7 potential breeding groups between Riverside Island population ONF was supplemented with a total of 81 translocated individuals over 13 years with the goal of reaching above the critical threshold of 30 PBGs per subpopulation Previous research suggests translocated bi rds respond differently according to gender, number of birds released per cluster, and varying habitat sizes ( USFWS 1985, Carrie et. 1999) The ONF that in many areas completely surrounds high qu ality sandhill habitat, and it has two distinct subpopulations of RCWs I undertook the study to determine which translocation techniques were most successful at contributing to the resident population s and to examine differences between the two subpopula tions Success was measured by the
40 number of new PBG formed the ability of a group to successfully fledge offspring and detection of translocated birds during the next breeding season at the cluster or within the RCW managed areas (fidelity) Specific fa ctors that were evaluated included number of birds released at a site, which region of the forest the birds were released, and gender of the released birds. I also examined information on the last known status of the bird and the age of the birds at first reproduction. The goal of this study was to determine whether translocations have had substantial positive impacts on resident red cockaded woodpeckers in the Ocala National Forest Objective 1: To compare the productivity of clusters and number of potent ial breeding groups in the vicinity of translocations before vs after translocation events. Objective 2: To determine cluster site fidelity and regional site fidelity of translocated birds. Objective 3: To compare productivity between birds released singly vs birds released as pairs between males vs females and between region s Methods Objective 1: To Compare t he Productivity o f Clusters a nd Number o f Potential Breeding Groups i n t he Vicinity o f Translocations before vs after Translo cation Events I conducted four paired t tests to evaluate whether translocations caused changes in the local number of PBGs and fledglings at two spatial scales : (1) within a 2. 1 kilometers ( km ) buffer surrounding each receiving cluster, and (2) at the 4 c lusters closest to the receiving cluster I chose these two spatial scales because each is biologically relevant to red cockaded woodpeckers. Conner et al. (1997) suggests that movements among dispersing birds range from 1. 8 km to 5. 4 km depending on age a nd gender With the proximity distances between
41 clusters I chose a 2. 1 km radius, which captures a maximum of 14 clusters with an average of 8.6 3.7 clusters In ArcMAP (v 9.3.1 20 09 ) I drew a 2. 1 km buffer around each receiving cluster to represent the average dispersal distance for females I recorded the number of fledglings produced by each cluster within each buffer and whether there was a potential breeding group present for two years before the translocation and two years after For example, when a translocation occurred in the winter of 2003, I averaged productivity of all clusters in the buffer from years 2001 and 2002 to characterize productivity prior to the translocation and averaged productivity for these clusters during years 2003 and 2004 to characterize productivity after the translocation T he recovery plan states that dispersal movement of fledgling s ranges from 1 to 4 clusters from the natal clus ter (USFWS 1985) Ther efore I also recorded the number of fledglings produced by each of the four nearest cluster s around each receiving cluster, and whether there was a potential breeding groups present for two years before the tra nslocation and two years after for the se same four nearest clusters to each receiving cluster I compare d productivity of nearby clusters before and after translocations, and compare d the number of potential breeding groups nearby translocation sites before and after translocations at both spatial scales, using paired t tests I used the translocation data from 2003, 2004, 2006, and 2007 ( n =16) along with the available data on fledglings and potential breeding groups from 2001 through 2008 to analyze whether there was a significant difference ( P 0 ) in the productivity of birds in the area surrounding each recipient cluster when comparing the two years prior to
42 translocation to the two years after For each cluster, the data were averaged together over the four years for which they were available Objective 2: To Determine Cluster Site Fidelity and Regional Site Fidelity of Translocated Red Cockaded Woodpeckers The second set of analyses determined cluster site fidelity and regional site fidelity of translocated individuals by gender region and re lease type (paired or single) I calculated the cluster site fidelity by determining the ratio of the number of translocated birds that remained at the release site during the next breeding season over the total number of recipient birds Similarly I meas ured regional site fidelity by determining the ratio of the number of individual birds that stayed within the boundaries of Riverside Island or Paisley Woods, respectively, during the next breeding season over the total number of birds that were received f or each region In addition, I measured the post release distance traveled by each translocated bird from their release site to their new territory These measurements were entered into an Analysis of Variance (ANOVA) to test for difference of means ( 0 ) between genders, regions (Riversid e Island or Paisley Woods ), and release types (paired or single). Objective 3: To Compare Productivity among Birds by Translocation Type (Released Singly or a s Pairs) Gender and Region s The third analysis compared the effectiveness of releasing a single RCW to a resident RCW vs translocating two unrelated RCWs as a pair to an unoccupied area. The response variables reviewed were a) the number of birds that settled into a cluster; b) the n umber of birds that formed a PBG; c) the number of birds that successfully contributed to the population with fledglings; and d) the number of fledglings produced during the first 4 years after translocation. These four components were reviewed against tra nslocation type (paired release vs single release) and gender of the
43 translocated birds. Logistic regression was used to identify which factors (type, gender, and region of release) differentiated translocated birds that produced fledglings from those that did not. Results Objective 1: To Compare t he Productivity o f Clusters a nd Number o f Potential B reeding G roup s i n t he Vicinity o f Translocations before vs after Translocation Events The first paired t test assessment of the translocation data revealed the re was no difference in fledgling success among clusters with in a 2. 1 km buffer of the release sites before vs after birds were moved ( P = 0.636) In contrast, the second paired t test found the number of potential breeding groups in 2. 1 km proximity to th e release site increased slightly ( P = 0.091) (Table 3 1 ) The third and fourth assessment s tested for differences in fledgling success and the number of potential breeding groups in the four closest clusters to each release site respectively. There were no differences in either fledgling success ( P = 0.280) or number of potential breeding groups ( P = 0.418) (Table 3 1). Objective 2: To Determine Cluster Site Fidelity and Regional Site Fidelity of Translocated Red Cockaded Woodpeckers The analysis of reco rds collected from 1993 through 2005 revealed 50 clusters were augmented with 81 birds (44 Males, 37 Females) with 38 (47%) successfully remaining in the forest Forest wide, t ranslocation was moderately successful at retaining birds within a cluster (14 b irds, 17%) and had greater success retaining birds within a region (Riverside Island and Paisley Woods; 24 birds, 30%) Overall dispersal distance among RCWs was 1.9 0. 3 km (Table 3 2 ) with 42% of birds (16 out of 38) dispersing >2. 1 km
44 Monitoring for gender across the forest during all available years found 22 Males (out of 44, 50% ; 10 within the cluster and 12 within the region ) and 16 Females (out of 37, 43% ; 4 within the cluster and 12 within the region ) settling into a new territory (Table 3 2 ) Monitoring within subpopulations found 24 birds in Riverside Island (51%; 13 within the cluster and 11 within the region) and 14 in Paisley Woods (41%; 1 within the cluster and 13 within the region) settling into a new territory (Table 3 2). Status of these birds was classified into four categories : mated with another translocated individual (Paired); claimed a territory alone (Solitary); mated with a resident bird (Local Mate); and undetected or lost (Missing) Of these, 4 Males settled in so litary ( 2 within the cluster and 2 within the region) vs 3 Females (1 within the cluster and 2 within the region ) (Figure 3 1 ) In addition, 1 1 Males formed a PBG with a local mate ( 5 within the cluster and 6 within the region) vs 8 Females ( 1 within the c luster and 7 within the region) There were 6 male birds that paired with another translocated bird (3 within the cluster and 3 within the region) compared to 4 females (2 within the cluster and 2 within the region) (Figure 3 1 ) There were 43 translocated birds that were never detected after their release (53%; 23 in Riverside Island and 20 in Paisley Woods) (Figure 3 1). A three way analysis of variance of dispersal distances indicat ed a significant ma in effect for the region of translocation, F 1,37 = 17. 42, P = 0.0002, with the average distance birds traveled when translocated to Paisley Woods significantly greater ( X = 3. 3 0. 5 km ) than birds translocated to Riverside Island ( X = 1.1 0. 3 km ) The main effects of gender and type and all interactions were non significant (Table 3 3)
45 Objective 3: To Compare Productivity among Birds by Translocation Type (Released Singly or a s Pairs) Gender and Region s Reviewing the cluster success, i n the first breeding season after a translocation, 9 recipient clus ters ( 18 %) formed a PBG by retaining at least one of the translocated individuals Six of these new pairs successfully produced 9 fledglings In the second year after a translocation, 1 7 recipient clusters formed PBGs ( 34 %) with 11 PBGs producing 1 5 fledgl ings I n the third year after a translocation there were 1 8 recipient clusters with PBGs (3 6 %) with 13 PBGs producing 1 3 fledglings Finally, in the fourth year, 22 recipient clusters had PBGs (44%) with 16 PBGs producing 13 fledglings (Figure 3 2). Translocation types included 58 birds released as pairs (29 Males, 29 Females) to inactive clusters and 23 single birds released (15 Males, 8 Females) in the vicinity of a resident bird Of the birds released as a pair, 27 (47%) settled into a territory T here were 20 (3 4 %) birds that formed PBGs and 17 (29%) were successful at producing 37 fledglings within 4 years of release and starting breeding at the average age of 2.2 yr Of the single bird translocations, 11 (48%) remained with a resident bird or set tled elsewhere There were 9 (3 9 %) new PBGs formed with 4 (17%) producing 13 fledglings within 4 years of release and starting breeding at age 2.5 y ea r s Reviewing patterns in gender revealed 2 2 ( 50 % ; out of 44 ) males settled in clusters with 1 7 (3 9 %) form ing PBGs Twelve (2 7 %) males contributed fledglings starting at the average age of 2 years There were 1 6 (4 3 % ; out of 37 ) females that remained at clusters with 1 2 (3 2 %) forming PBGs There were 9 (24%) successful females contributing offspring starting at the average age of 2.5 years (Table 3 4) Over the course of 13 years, there were 102 fledglings contributed to the population by 21 translocated birds (Figure 3 3). Logistic
46 regression indicated the f actors that differentiated translocated birds that produced fledglings from those that did not was region ( P = 0.0404), but not type of translocation ( P = 0.0762) or gender ( P = 0.9713). Birds translocated to Riverside Island were five times more likely t o produce fledglings than birds translocated to Paisley Woods. The status of translocated birds by type of release showed 14 paired males (4 p aired, 3 s olitary, and 7 with a l ocal m ate), 8 single males (2 p aired, 2 s olitary, and 4 with a l ocal m ate), 13 paired females (4 p aired, 4 s olitary, and 5 with a l ocal m ate), and 3 single females (3 with a l ocal m ate) (Figure 3 4) Discussion Objective 1: To Compare the Productivity of Clusters and Number of Potential B reeding G roup s in the Vicinity of Tr anslocations before vs after Translocation Events Red cockaded woodpecker offspring usually disperse within one year of fledging (Walters et al. 1988) During this time frame, biologists utilize the opportunity to translocate surplus individuals to another population that will not be able to reach the critical threshold of 30 potential breeding groups independently The statistical results show that the 16 translocations that were selected for analysis produced no significant changes in the local number of fledglings during the first two years after translocation release or at either spatial scale investigated These translocations only slightly increased the number of potential breeding groups formed within a 2. 1 km radius of the receiving cluster and had no effect within the 4 closest clusters This may in part be explained by the age of the donor birds These donor birds averaged about 6 months old ; both males and females are highly susceptible to nest failure or no nest attempt in the first 2 years (Dani els and Walters 2000) Previous research has shown that translocated females nested later in the season and with smaller clutches compared to
47 native females but t hese effects were not permanent and within a few years revealed no differences among the two groups (Levesque et al. 2004) The fact that the number of potential breeding groups increased slightly as a result of translocations within a 2. 1 km buffer around translocations provides support for the idea that tra nslocated birds may make positive contributions to resident populations over a longer period of time than the period I analyzed in both creating PBGs and potential fledglings Additional research should examine longer time periods to assess whether product ivity of translocated individuals increases as time since release increases. Objective 2: To Determine Cluster Site Fidelity and Regional Site Fidelity of Translocated Red Cockaded Woodpeckers In the second assessment, I analyzed the historical records fro m 1993 2005 to review cluster site fidelity and regional site fidelity among translocated birds and then compared region and gender effects on translocation This analysis was not limited by time or space as in the previous test. The proportion of undetect ed translocated birds in both subpopulations was about 53% over 13 years The high proportion of undetected birds in ONF could be due to detection error which is often associated with small population size, individuals that are difficult to sample, or limi ted sampling efforts (Gu and Swihar t 2004) Alternatively, it could be from birds emigrating out of the population beyond monitored areas or from predation Other studies show RCW translocations with similar to better success Hagan and Costa (2001) monito red two translocation events in Florida Fifty percent of translocated birds were lost in the 1 st year of translocation and 30% the 2 nd year Saenz et al. (2002) studied a large scale translocation across Texas, Louisiana, Arkansas and Oklahoma with 29% failure in retaining translocated individuals Similarly, Carrie et al. (1999) also found 29% failure in Texas ONF is
48 unique in its vegetation community heterogeneity For instance, Riverside Island is completely surrounded by scrub (Figure 3 5) The widt h of high quality habitat (sandhill) ranges between only 2 km 4 km and the RCW population is clustered towards the center and away from the bordering scrub ecosystem There is a high potential for birds to explore into the scrub and become undetected Whil surrounded by scrub (Figure 3 RCW Recovery Plan (USFWS 1985) as unsuitable Limited s ize of available habitat might also influence the RCW success rate in ONF. Walt ers et al. (2004) had a higher success rate (74%) than ONF in a larger habitat range 188,000+ha in Eglin Air Force Base vs 5000+ha in ONF. In addition to size, ONF has a large amount of edge that birds tend to avoid and during the beginning efforts of tr anslocations, there were low densities of resident PBGs. Translocation success may be negatively correlated to distances between neighboring resident birds (Allen et al 1993) The limited amount of habitat areas in ONF with its large amount of edge make it plausible there are birds seeking shelter in neighboring ecosystems that are not monitored for RCWs. Regional site fidelity was higher than cluster site fidelity suggesting that birds were more likely to stay within the region than within close proximit y to release sites There were also differences in retention between the two regions Paisley Woods retained less translocated individuals at their release site compared to Riverside Island Paisley Woods had a high number of available territories low density of established PBGs near the release site s and had an average distance between clusters of 1. 3 km ( 0.7 ) resulting in low cluster site fidelity but higher regional site fidelity ( F igure 3 1 ) Pasinelli and Walters (2002) found the probab i l it y of f ledg l ing dispersal was positively related to
49 the number of unoccupied territories where the higher number of available territories, the farther the birds traveled In contrast, Riverside Island had a lower availability of territories, a higher density of r esident PBGs and had an average distance between clusters of 0. 8 km ( 0.3 ) resulting in very little difference between cluster and regional fidelity though both were fairly high ( Figure 3 2 ) Carrie et al. (1999) observed a beneficial social interaction w hen releasing individuals in close proximity to resident birds A similar response was found in Riverside Island where there was a higher success with higher density of PBGs near the release sites The overall average dispersal distance among males and fe males was 1. 9 km ( 0.3) This is smaller than a study by Carrie et al. (1999) who found an average dispersal distance of 2. 8 km Walters et al. (19 8 8) found a significant difference in dispersal distances between males (4. 5 km ) and females (3. 2 km ) In ONF males and females dispersed similar distances ( 1. 8 km and 2. 0 km respectively ) (Table 3 2 ) Of the 38 translocated birds later detected, 24 settled outside the release site with 16 (42%) traveling farther than 2. 1 km The average starting age for repr oduction was 2.3 years. This suggests that the results of objective 1 may be misleading: 42% of the translocated after the 2 year timeframe the analysis reviewed. Furthe r research should be conducted to determine if there is a significant benefit to the population from birds dispersing beyond the 2. 1 km distance and 2 year time frame I evaluated with objective 1 Objective 3: To Compare Productivity among Birds by Transl ocation Type (Released Singly or a s Pairs) Gender and Region s Across both regions of ONF, the number of new PBGs and successful PBGs increased each year for 4 years following translocat ion events The number of fledglings
50 produced by the new PBGs varied among years Of the birds that established a territory, Riverside Island (51%) and Paisley Woods (41%) retained a lower percentage than Carrie et al. (1999) reported (71%) but Riverside Island (13 out of 24) had a higher cluster site fidelity than Carrie et al. (1999) reported (3 out of 12) and Paisley Woods (1 out of 14) The overall success rate of translocations on the ONF ( 47 %) is less than Franzreb (1999) and Allen et al. (1993) reported (63.2 % and 69% respectively) Retention of translocated birds was similar between types of translocation gender and region P aired birds released to inactive clusters resulted in 47% retention and single birds released to a resident bird resulted in 48% Translocated males showed 50 % retention and females 4 3 % (Table 3 4) Nest success was similar between types of was five times higher than Paisley Woods The paired release method was almost twice as successful (29%) compared to the single bird release technique (17%) Males were found with 27 % nest success vs females with 24% Average age to reproduction was lowest among paired and male birds A study by Allen et al. (1993) found that translocations were most successful in South Carolina when matching an experienced, adult female (3 out of 3 nested) or male (6 out 8 nested) with a resident bird They had little success translocating paired or single birds (4 out of 12 nested) into inactive clusters Investigation of the status of translocated birds found the majority of them settling with a local resident (Local Mate); this outcome was twice as common as any other status All single females were found in t his status Paired females were almost equal in all categories of status (paired, solitary, or with a local mate) Further research on
51 translocated birds, both male and female, should test whether translocating older birds results in a higher success rate than I found with translocating first year hatchlings Management Implications The positive effects of translocation are evident in both Riverside Island and Paisley Woods in the ONF Translocations have resulted in the formation of new potential breeding groups, provided solitary males the opportunity to mate with a new female and create d new solitary males for resident females These events have also resulted in a large number of offspring from the translocated birds, multiplying t he success into future generations Over a 13 year period (1993 2005), translocation s have helped increase the population from 7 to 39 PBGs (Figure 3 3) Not all events of translocation were successful The lack of neighboring birds could have led to high er failures in early years Future translocations should be completed with paired, unrelated individuals within close proximity to other resident birds Significance of translocation within a 2. 1 km radius of release sites was minimal to the resident popul ation during the first two years Additional research should be conducted at a wider range over longer time periods to determine if there is a greater impact beyond the spatial and temporal scales I investigated The habitat structure and size of resident populations might give insight to the response of translocated birds. Riverside Island is completed surrounded by scrub ecotone, has a narrower habitat structure and a denser resident population compared to Paisley Woods, and had greater success in product ivity and higher cluster site fidelity. The larger resident population could have aided in the survival of the young birds. The scrub ecotone surrounding the island is avoided by RCWs (chapter 2) and affected the
52 distance birds traveled. Future translocati ons should be in areas avoiding scrub ecotones. The success of RCWs on the Ocala National has exceeded the 5% annual 1 6 % per year (range from 1 1 to 32 % ), with an average of 6.2 birds translocated forest wide per year (range from 1 to 12 birds ) Survival of fledglings on ONF was 47% overall during the 13 years which w as similar to survival of birds translocated to sandhills in North Carolina The grew over 450% in 13 years from 7 potential breeding groups to 39. It appears that the translocation program had a positive impact on overall population growth
53 Table 3 1 Effects of translocation efforts on productivity of surrounding red cockaded woodpecker (RCW) clusters at two spatial scales. C omparisons were of number of fledglings and number of potential breeding groups (PBG) during the two years before vs two years after translocations using paired t tests Dispersal r anges Metric X SE P 95% CI 2. 1 km max (n=16) Fledglings 0.03 0.06 0.636 0.16 0.01 PBG 0.05 0.03 0.091 0.01 1.00 Closest 4 (n=16) Fledglings 0.12 0.10 0.280 0.32 0.09 PBG 0.04 0.05 0.418 0.06 0.15 Table 3 2 Number of translocated RCWs ( 1993 to 2005 ) by region gender and type that settled at their release site (cluster site fidelity) or beyond their release site but within the region (regional site fidelity), total number of birds found after translocation (total fidelity) the average dispersal distance traveled in kilometers (km) standard error (SE) and dispersal range Translocation type Cluster site fidelity Regional site fidelity Total fidelity Dispersal d istance in km ( X SE ) Range (km) Riverside ( n =47) 13 (28%) 11 (23%) 24 (51%) 1.1 0.3 0 5.8 Paisley ( n =34) 1 (3%) 13 (38%) 14 (41%) 3.3 0.5 0 7.1 Female ( n =37) 4 (11%) 12 (32%) 16 (43%) 2.0 0.5 0 3.9 Male ( n =44) 10 (23%) 12 (28%) 22 (50%) 1.8 0.3 0 7.1 Paired ( n = 58) 9 (16%) 18 (31%) 27 (47%) 1.9 0.3 0 5.8 Single ( n = 23) 5 (22%) 6 (26%) 11 (48%) 1.8 0.7 0 7.1 Forest Wide ( n = 81) 14 (17%) 24 (30%) 38 (47%) 1.9 0.3 0 7.1
54 Table 3 3 Analysis of variance summary for variable s describing distance traveled by translocat ed RCWs Difference were considered significant if P< 0.05 Source of v ariation df SS MS F value P value Type 1 2.22 2.22 0.90 0.3507 Gender 1 0. 68 0. 68 0.28 0.6029 Region 1 43.07 43.07 17.42 0.0002 Type x g ender 1 0. 28 0. 28 0.11 0.7369 Type x r egion 1 6.13 6.13 2.48 0.1258 Gender x r egion 1 2.53 2.53 1.03 0.3194 Type x g ender x r egion 1 0.89 0.89 0.36 0.5538 Error 30 74.17 2.47 Total 37 133.48 Table 3 4 Status of translocated RCWs based on t ype of translocation ( paired vs single ) gender of birds and region released Settled PBG Produced f ledgling Age to reproduction in y ea r s ( X SE ) Riverside ( n = 47) 24 (51%) 20 (43%) 16 (34%) 2.3 0.4 Paisley ( n = 34) 14 (41%) 9 (26%) 5 (15%) 2.4 0.7 Paired ( n =58) 27 (47%) 20 (3 4 %) 17 (29%) 2.2 0.4 Single ( n =23) 11 (48%) 9 (3 9 %) 4 (17%) 2.5 0.7 Male ( n =44) 2 2 ( 50 %) 1 7 (3 9 %) 1 2 ( 2 7 %) 2 1 0.4 Female ( n =37) 1 6 (4 3 %) 1 2 (3 2 %) 9 (24%) 2. 6 0.5 Forest Wide (n = 81) 38 (47%) 29 (36%) 21 (26%) 2.3 0.3 Table 3 5. Parameter estimates SE, and adjusted odds ratios with 95% confidence intervals from the best fit models using backward logistic regressio n in predicting probability of producing a n RCW fledgling Adjusted odds ratio Parameter value Parameter estimate SE Estimate 95% CI P Type 0.73 0.41 4.3 0 0.86 21.60 0.0762 Region 0.79 0.39 0.21 0.05 0.93 0.0404 Gender 0. 01 0. 37 1.02 0.24 4.40 0.9802
55 Figure 3 1 Count of red cockaded woodpeckers ( RCWs ) surviving from 1993 2005 after translocation to the Ocala National Forest (ONF) Florida, by location and gender. Counts of birds that remained in the cluster to which they were translocated (Cluster), and count found in Riverside Isla nd (RI) and Paisley Woods (PW) (Region) are shown with counts of birds found without a mate (Solitary), and found paired with a local resident (Local Mat e ) The count birds that disappeared (Missing) are represented by the second y axis Translocation type s were paired males with paired females brought to newly constructed clusters ( n = 17 each gender), males brought to solitary females (Single Males, n = 8) and females brought to solitary males (Single Females, n = 5). 0 5 10 15 20 25 30 35 40 45 50 0 2 4 6 8 10 12 14 Paired Solitary Local Mate Paired Solitary Local Mate Missing Cluster Region Number of Individuals Missing Number of Individuals Found Status of Individuals by Region and Gender Site Fidelity in Ocala National Forest by Location and Gender PW Female PW Male RI Female RI Male PW Missing RI Missing
56 Figure 3 2. Productivity of RCW p otential b reeding g roup activity after translocations on the ONF Florida, measured as counts of birds over four years after each translocation. Metrics include new potential breeding groups created within Riv erside Island and Paisley Woods (New PBGs), PBGs successfully raising young to fledgling stage (Successful PBGs), and number of young surviving to fledgling stage (Fledglings). Figure 3 3 Cumulative number of RCWs in the ONF introduced by t ranslocat ion events, fledgling s added by translocated birds (Trans Fledg), and f ledgling s produced by resident birds ( F ledgling ) T he solid line depicts the change in number of active clusters (Active) and the dashed line shows the change in number of potential breeding groups ( PBG ). 0 5 10 15 20 25 New PBGs Successful PBGs Fledglings Number of Individuals Productivity after Translocation on ONF Year 1 Year 2 Year 3 Year 4 0 10 20 30 40 50 60 Number of RCWs Year RCW Status after Translocation on ONF Translocation Trans-Fledg Fledgling Active PBG
57 Figure 3 4 Number of RCWs surviving from 1993 2005 after translocation to the ONF Florida, by gender and type of translocation (paired or single ) The number of birds that disappeared (Missing) is indicated on the s econd y axis 0 5 10 15 20 25 30 35 40 0 2 4 6 8 10 12 14 16 18 20 Paired Solitary Local Mate Missing Number of Individuals Missing Number of Individuals Found Status of RCWs RCW Status by Type of Translocation on ONF Single Female Paired Female Single Male Paired Male Overall Missing
58 Figure 3 5. Ecosystem s in Riverside Island in the northern region of the ONF Forty seven red cockaded woodpeckers were translocated in 16 newly created artificial clusters from 1993 to 2004.
59 Figure 3 6 Ecosystem s in Paisley Woods in the southern region of the ONF Thirty four RCWs were translocated in 13 newly created artificial clusters from 1995 to 2005.
60 CHAPTER 4 USING DATA ON RED COCKADED WOODPECKER HABITAT SELECTION FR OM OTHER NATIONAL FORES TS IN FLORIDA TO MA KE INFERENCES ABOUT HABITAT RESTORATION AND SUITABILITY ON O CALA NATIONAL FOREST Introduction The red (RCW) range spans across 11 states of the southeastern United States (Ligon et al. 1986). Although many researchers believe that RCW 's are most closely adapted to longleaf pine ( Pinus palustris ) ecosystems (Conner et al 2001), which now is one of the most important endangered eco systems in the nation (Simberloff 1993, Ware et al. 1993) remnant populations can be found in habitats with varying vegetative structure and species composition (USFWS 1985). Some of these habitats include shortleaf ( P. echinata ) and loblolly pine ( P. taeda ) fo rests (Wood et al. 2008), hydric slash pine ( P. ellio ttii ) flatwoods (Beever and Dryden 1992 Smith and Martin 1995), and pond pine ( P. serotina ) communities (USFWS 1985). Despite the variety of habitats they occupy, RCWs select similar vegetation features throughout their range. They prefer stands of large, old growth pines (>60 years) with minimal hardwood midstory maintained by frequent growing season fires ( USFWS 1985 Hooper and Harlow 1986 Jones and Hunt 1996 Zwicker and Walters 1999 ). In addition, t heir home range size can vary dramatically across geogr aphic regions from 14 to 225 hectares ( ha ) (Wood et al. 2008) and has been found to be negatively related to the availability of high quality habitat ( Davenport et al. 2000 ). The largest RCW population in the southeastern US occurs in northern Florida in the Apalachicola N ational F orest (ANF) contain ing approximately 650 active clusters ( James 1991, Petrick an d Krusac 2012). The clusters are located in areas composed predominantly of longleaf pine flatwoods and slash pine plantations ( Hovis and Labi sky
61 1985) with small areas of sandhills ( Petrick and Krusac 2012 ) The Osceola National Forest (OsNF) is estimated to have over 400 active clusters with 140 potential breeding groups. According to the United States Forest Service ( USFS ) forest type Geographical Information System ( GIS ) layer, the major components of their habitat consist of 40% slash pine forests and 20% bald cypress water tupelo mix; the Florida Fish and Wildlife Conservation Commission (FWC) (FWC Metadata 2004) vegetation and land cover data indicates swamp and wetland areas cover 40% of the O sNF and pinelands cover 35%. In contrast, the Ocala National Forest (ONF) contains only 75 potential breeding groups ( Petrick and Krusac 2012 ). The ONF supports two very contrasting vegetative communities often separated by sharp boundaries. The first community, longleaf pine wiregrass occurs in isolated i slands within a matrix of the second community, sand pine scrub (Kalisz and Stone 1984). Longleaf pine wiregrass communities are dominated by open stands of longleaf pine trees as well as other pine species, deciduous oaks, and a wiregrass ground cover; wh ile sand pine scrub communities are dominated by a closed canopy of sand pine ( P. clausa ) with a scrub oak understory and little herbaceous ground cover or bare ground (Myers 1985). Although RCWs' preference and adaptation to the longleaf pine ecosystem ha s been well documented (Conner et al 2001), how RCWs interact with sand pine scrub communities is poorly understood and could be a subject of future investigation. In the 1900 s fire suppression and unfavorable silvicultural practices had negative effects on the red cockaded woodpecker status and the ecosystems on which they depend ( Ligon et al. 1986 1991 ; Frost 1993; Ware et al. 1993 ; Landers et al. 1995 ; Conner et al. 2001). Loss of habitat and fragmentation of existing habitat are major
62 threats to this species. RCWs do not display movement s across large geographic distances within metapopulations and therefore require within population genetic variation (Stith et al. 1996) from an area, transloca tion of individuals to new recruitment stands may be the only option for re establishing breeding groups within historically occupied areas and may also help bolster small existing populations (Carrie et al. 1999). Rudolph et al. (1992) recommended releas ing multiple pairs of RCWs when attempting to reintroduce birds into a previously occupied habitat. Carrie et al. (1999) reported that releasing multiple pairs of RCWs in close proximity to each other and to resident groups facilitated social contact and a llowed individuals to settle and obtain mates in the area. The purpose of this study is to evaluate the suitability of potential habitat for RCWs in the area of ONF where birds have been extirpated (study area A; Figure 1 1). Due to the stark contrast betw een habitat types in this area in com parison to the sandhills currently in use by RCW populations in ONF (study areas B and C; Figure 1 1) I examine d habitat selection patterns of RCWs in other national forests in Florida to develop predictions for the geographic areas most deserving of future restoration efforts and translocations in ONF Objective 1: To examine habitat selection by RCWs in all 3 national forests in Florida. Objective 2: To map the location of habitat fea tures selected by RCWs in other national forests in Florida to the region abandoned by RCWs in ONF
63 Methods Objective 1: To Examine Habitat Selection by RCWs in All 3 National Forests in Florida I obtained GIS layers from the United States Forest Service (USFS) data base and FWC vegetation class. These layers included three categories : RCW trees (three national forests), FWC forest type (statewide), stand data (forest type and stand initiation dates; three national forests), and soils (state wide). In addition, I obtained the categories the Florida Land Use, Land Cover Classification System (FLUCCS) land cover categories and the Cooperative Research in Forest Fertilization (CRIFF) for est soil classification (Jokela and Long 1999) The recovery plan describes the average territory size of RCWs for different regional areas including central Florida, northwest Florida and Coastal Georgia, ( 129 ha 109 ha and 80 ha respectively; USFWS 198 5). I chose the largest average size, 129 ha to represent the typical RCW territory size for all three national forests in Florida to a radius of approximately 0.6 4 kilometers ( km ) I reduced the RCW tree layer to centroids of all active and inactive trees to provide a representation of areas occupied by woodpeckers both currently and historically. I created buffers around each centroid in each of the three national forests. Once the buffers were created, I wanted to differentiate between areas currently /historically in use from areas of available habitat To ensure there was no overlap in these areas I doubled the buffers around each centroid ( 1. 2 8 km radius) and ArcMAP (v 10, 20 10 ) to delete these areas from the forest
64 boundary layer. This resulted in two separate regions : buffers around current /historical nest trees (i.e., used habitat ) and available habitat I took a simple random selection of buffers around current /historical nest trees (used habitat ) in each of the three forests to sample 20% of the current RCW population of each forest. This provided a total of 30 points in Ocala National Forest, 50 points in Osceola National Forest and 160 points in Apalachicola National Forest Then I ran a random sample point generator in the available habitat region of each national forest to obtain the same number of points in available areas per forest Each data layer was intersected with the buffers from the sampling point s to generate data for a logistic regression model to test which factors differentiated between used and available habitat The FW C Global Information System ( GIS ) data on forest type was originally in a raster form. In order to intersect the data with th e sample buffers, I converted the raster into a shape file in ArcMAP and then dissolved and projected it to make it more pliable. The data created from intersecting buffers and the raster were formatted in Excel to prepare for SAS import. Due to the large ratio of variables relative to sample points collected ( 373 variables / 480 sample points) and the limited number of variables suggested per sample points for logistic regression (5 10 sample points per variable) (Peduzzi et al. 1996, Vittinghoff and McCul loch 2006), I used a hierarchical approach. First, to derive a coarse assessment of RCW habitat selection, I collapsed the FWC and USFS forest types vegetative descriptions into the 8 FNAI categories. I used the CRIFF forest soil classification to reduce t he soil layers from 149 categories to 22 Finally, I grouped age year of stands into 10 year increments, reducing the number of stand
65 variables from 124 to 15 In SAS, I completed a univariate test on each variable to find which had a p value < 0.250. Next, I ran a correlation test to determine which vegetative descriptions were always found together in a buffer per forest (P>0.600). Using logistic regression in SAS, I first tested uncorrelated variables (Appendix B) using the FNAI, stand initia tion classifications, and CRIFF to provide a coarse scale estimate of significant variables ( P > 0.100). Next, I tested uncorrelated variables using the FWC USFS forest type stand initiation classifications, and USFS soil types to provide a fine scale es timate of significant variables ( P > 0.100). Weighted overlay analysis in ArcMAP requires variables to be scaled based on suitability I used 8 categories of habitat use: very strongly avoided, strongly avoided, moderately avoided, slightly avoided, slig htly preferred, moderately preferred strong ly preferred and very strongly preferred These categories were determine d by using the inverse of th os e odds ratio s between 0 and 1 to standardize the results (Chen et al. 2010) and then order them from smalles t (most avoided) to largest (most preferred ). To represent weighted overlay scales (WOS), I scaled the results from 1 to 8 based on standardized odds ratio : < 4.72 = very strongly avoided ( WOS 1); 4.72 to 2.75 = strongly avoided ( WOS 2); 2.74 to 1.53 = moderately avoided ( WOS 3); 1.52 to 0 = slightly avoided ( WOS 4); 0 to1.52 = slightly preferred ( WOS 5); 1.53 to 2.74 = moderately preferred ( WOS 6); 2.75 to 4.72 = strongly preferred ( WOS 7); and >4.72 = very strongly preferred ( WOS 8). Objective 2: To Map the Location of Habitat Features Selected b y RCW s in Other National Forests in Florida to the Region Abandoned b y R CW s i n ONF I mapped the regions in the area of ONF RCWs have abandoned (study area A) to identify the most suitable locations for RCW reintroductions I used the ecosystem layer
66 (USFS GIS Layer) to exclude areas that will never have a chance of being managed for RCW habitat These areas include d six categories : non forested, private lands, scrub, swamp, wetlands, and water. The remaining potential habitat areas wer e sandhills and pine flatwoods. The results from the logistic regression conducted in objective one were reclassified in GIS on a scale from 1 to 8 reflecting the standardized odd s ratios Additionally, correlated variables not entered into the logistic regression were reclassified and all variables were entered into a weighted overlay analysis in ArcMAP The results were mapped within the potential habitat boundaries to highlight preferred and avoided areas. The weighted overlay analysis utilized preferred and avoided habitat areas from objective one within different layers and identified the most suitable habitat for RCWs. Correlated variables not used in the logistic regression were added to the weighted overlay analysis for additional support of suitable variables. Soils were removed from the analysis due to the lack of occurrence of significant soil variables in Church Lake. Lastly, I dissolved coarse scale WOSs 5 8 ( slightly to very strong ly preferred ) from the selected areas to calculate acreage of large connected areas of selected habitat The recovery plan ( USFWS 1985) states that high quality forested areas should provide at least 4 9 ha per cluster and low quality areas 80 120 ha Each of t he dissolved layers produced after identifying habitat features selected by RCWs in ANF and OsNF were divided by low and high quality home range size estimates to suggest the possible carrying capacity of study area A.
67 Results Objectiv e 1: To Examine Habitat Selection by RCWs in All 3 National Forests in Florida Coarse s cale a nalysis The logistic regression identified v ariables that differentiated used from available habitat at the coarse scale using FNAI natural communities combined with the FLUCCS land cover stand initiation age class, and CRIFF forest soil classification (Table 4 1). FNAI classes RCWs avoided were freshwater forested wetlands ( P = <0.0001) ( WOS 4) and hardwood forested uplands ( P = 0.0200) ( WOS 4) whereas they selected for high pine scrub ( P <0.0028) ( WOS 5) and pine flatwoods/dry prairie ( P = <0.0001) ( WOS 5) Stand initiation age classes RCW selected for were 1899 1908 ( P = 0.0036) ( WOS 5) 1909 1918 ( P = <0.0001) ( WOS 5) 1919 1 928 ( P = <0.0001) ( WOS 5) and 1929 1938 ( P =<0.0001) ( WOS 5) CRIFF soil classes selected by RCWs were BCD Savannas/Flatwoods ( P = <0.0001 ) ( WOS 5) BD Savannas/Flatwoods ( P = 0.0010) ( WOS 5) C flatwoods ( P = 0.0131) ( WOS 5) and E u plands ( P = 0.0102) ( WOS 6) The FNAI class results show the more freshwater forested wetlands and hardwood forested uplands, the less likely there will be a cluster in the area. These were both considered slightly avoided for the weighted overlay scale. In contrast, as high pine scrub and pine flatwoods/dry prairie increases, the likelihood of finding a cluster in an area increases. These were slightly preferred. The stand initiation age class results indicate the more trees older than 70 years, the more likely there will be a cluster in the area. All stand initiation age classes were scaled as slightly preferred. Increasing BCD Savannas/Flatwoods, BD Savannas/Flatwoods, C flatwoods, or E uplands increased
68 the likelihood of finding an RCW cluster. CRIFF series E was considered moderately preferred for the weighted overlay scale whereas all others were slightly preferred. Fine s cale a nalysis The second logistic regression analysis identified variables that differentiated used from available habitat a t the fine scale level, using FWC forest cover classes, USFS forest type s, stand initiation age class, and USFS soi l type s. FWC vegetation classes RCWs avoided were hardwood swamp ( P = 0.0054) ( WOS 4) and mixed wetland forest ( P = 0.0002) ( WOS 4) while pinelands ( P = <0.0001 ) ( W OS 5) and sandhills ( P = <0.0001 ) ( WOS 5) were selected for The USFS forest type RCWs selected for was longleaf pine ( P = <0.0001) ( WOS 5) pond cypress ( P = 0.0036) ( WOS 5) and slash pine ( P = 0.0954) ( WOS 5) while bottomland hardwood/yellow pine mix ( P = 0.0285) ( WOS 4) slash pine hardwood ( P = 0.0159) ( WOS 4) and forest type unknown ( P = 0.0024) ( WOS 4) were avoided Stand initiation age classes RCWs selected for were 1909 1918 ( P = 0.0033) ( WOS 5) 1919 1928 ( P = <0.0001) ( WOS 5) and 1929 1938 ( P = < 0.0001) ( WOS 5) Lastly, soil classes avoided by RCWs were B R utlege and P lummer ( P = 0.0383) ( WOS 4) G Foxworthy ( P = 0.0628) ( WOS 4) and unknown H osford ( P = 0.0017) ( WOS 1 ) whereas BCD P lummer, S apelo, and P ottsburg ( P = <0.0001) ( WOS 5) and E G oldsboro ( P = 0.0285) ( WOS 7 ) were selected for. The FWC vegetation results show that the greater the area with hardwood swamp or mixed wetland forest, the less likely there will be an RCW cluster These variables were considered slightly avoided for the w eighted overlay scale. In contrast, the greater the area with pinelands and sandhills the more likely there will be a cluster in the area. There were considered slightly preferred. The USFS forest type results indicate that increasing the amount of longlea f pine slash pine and pond cypress increased the
69 likelihood of having an RCW cluster (slightly preferred) whereas increased amounts of bottomland hardwood/yellow pine and unknown (slightly avoided) decreased the likelihood The stand initiation age class results indicate the more trees older than 70 years, the more likely there will be a cluster in the area. All stand initiation age classes were scaled at slightly preferred. Finally, increasing the amount of BCD Plummer Sapelo and Pottsburg (slightly pr eferred) and E Goldsboro (strongly preferred) soil s increased the likelihood there will be an RCW cluster In contrast, increasing the amount of areas of B Rutlege and Plummer (slightly avoided) G Foxworthy (slightly avoided) and unknown Hosford (very strongly avoided) decreased the likelihood of finding an RCW cluster Objective 2: To Map the Location of Habitat Features Selected b y RCW s in Other National Forests in Florida to the Region Abandoned b y R CW s i n ONF The region where the current popul ations of RCWs reside (study areas B and C) are 6,200 ha and 5,300 ha respectively, and approximately 90 % sandhills. In contrast, the area RCWs abandoned (study area A) is 20,500 ha with 7,400 ha of flatwoods and sandhills (the vegetation types suitable f or potential RCW management) (Figure 4 2) Study area A has a mixture of 38% private lands, along with a mosaic pattern of non forest, private lands, pine flatwoods, sandhills, scrub, swamp, water and wet prairies (Table 4 3) Variables that differentiat ed used vs available habitat at both c oarse scale and fine scale s in objective one were standardized based on their odd ratio estimates and entered into a weighted overlay analysis in ArcMAP to identify areas most suitable for future RCW introductions The se results were mapped onto study area A (Figure s 4 3 and 4 4 ).
70 To estimate the carrying capacity of potential breeding groups in study area A, I USFWS 1985) estimation for carrying capacity according to varying qualities of habitats. Areas that are considered high quality require at least 49 ha per cluster and areas that are considered low quality need 80 120 ha The coarse scale weighted overlay analysis identified 3,361 ha of potential habitat in study area A and the fine scale analysis found 2,780 ha At both scales, I isolated contiguous areas that were at least 49 ha in size to calculate the potential carrying capacity if all acres met high quality standards (Figure s 4 5 and 4 6 ). In the coarse scale, t here was 2,654 ha existing in contiguous regions at least 49 ha per polygon which would potentially provide habitat for 54 clusters if this were high quality habitat ( Table 4 4 ). At the fine scale level, there was 1 934 ha existing in contiguous regions, allowing for 39 c lusters if this were high quality habitat Next, I isolated conti g uous areas with at least 80 120 ha and found 1 737 to 2 314 ha available at the coarse scale which would provide for 14 28 clusters and 1 016 to 1 530 ha available at the fine scale which would provide for 8 19 clusters if this were low quality habitat (Table 4 4) Discussion Objective 1: To Examine Habitat Selection by RCWs in All 3 National Forests in Florida Coarse scale analysis Two FNAI natural communities were selected for : high pine / scrub and pine flatwoods/dry prairie FNAI describes high pine/scrub as hills with mesic or xeric woodlands or shrublands and pine flatwoods/dry prairie as mesic or hydric pine woodland or mesic shrubland s on flat sandy or limestone substrates These natural communit ies contain fine scale habitat feature described in the FWC and USFS
71 vegetation layers that were selected for and others that were avoided according to the univariate test Two FNAI natural communities were avoided: freshwater forested wetlands and hardwood forested uplands. These four variables were only slightly preferred or avoided on the weighted overlay scale. Other studies have observed swamps, savannas and clearcuts as avoided habitat areas when they were a part of RCW ho me range s (Porter and Labi sky 1986). Areas that have more than 10% of canopy trees with hardwoods have also been reportedly avoided (J ones and Hunt 1996 ). In contrast, RCWs foraging predominately on pines annually and throughout the breeding season have be en observed foraging on hardwoods during the non breeding season (Wood et al. 2005). In general, RCWs preferred high pines in Florida and in some areas, may use small hardwood areas for foraging. The red cockaded woodpeckers normally select habitat contai ning old growth pine trees (Jones and Hunt 1996, Zwicker and Walters 1999, Wood et al. 2005). The results found RCWs in Florida selected stands that were greater than 70 years old These four age classes over 70 years old were found slightly preferred for the weighted overlay scale. RCWs are well known for nesting in habitats that are >80 yrs and foraging during the non breeding season in trees stands that are <60 yrs (Zwicker and Walters 1999). RCWs selected for CRIFF soil classes BCD savannas/flatwoods BD savannas/flatwoods, C flatwoods and E uplands. These results correspond to RCWs well known preference for pine ecosystems. CRIFF series uplands was moderately preferred while all others were slightly preferred. The large variety of soil types se lected is likely indicative of the large area of ANF which contains many varying soil types. With the largest number of RCW clusters, it is expected that birds in ANF are
72 adaptable to different soil types. In ONF, RCWs are found primarily in areas with the CRIFF soil class G. Because this soil class has a high correlation with the FNAI category high pine scrub ( P >0.60), this soil class was not entered in the final model. Fine scale analysis At the fine scale level, RCWs avoided areas that included mixed wetland forests and hardwood swamps, and selected for areas with sandhills and pine lands These four variables were only slightly preferred or avoided for the weighted overlay scale. I t is no surprise that RCWs selected sandhills since much of their habita t is within the sandhill habitat islands in ONF, and also in ANF The OsNF is dominated by large areas of p inelands used by their resident RCW population The Forest Service delineates forest types at the stand level describing the dominant tree species t o represent the stand. The USFS forest type s selected for by RCWs were longleaf pine pond cypress and slash pine These variables were slightly preferred in the weighted overlay scale. of high pine scrub (A ppendix B) and is well understood to be a higly selected component of foraging habitat for RCWs (Nesbitt et al. 1978, DeLotelle et al. 1987, Shackelford and Conner 1997, Conner et al. 1998). One study found that longleaf pine stands were selected during fo raging 90% of the time over cypress domes 10% of the time (DeLotelle et al. 1987). On the other hand, flatwoods associated with longleaf, slash, and pond pines, were found to be used 82% of time (Nesbitt et al. 1978). Other studies monitored used of varyin g pine species and found that longleaf pines were selected over slash pine 72% of t he time vs 22% (Porter and Labi sky 1986) The USFS forest types avoided by RCWs were bottomland hardwoods/yellow pine mix slash pine/hardwoods mix and forest type unknown. These were all slightly avoided for the
73 weighted overlay scale. T he first t wo variables are associated with higher percentage of hardwoods than what is tolerated by RCWs while the other, forest type unknown, is most commonly associated with areas of open w ater. RCWs again selected for stands greater than 70 years old The USFS soil class es avoided were B Rutlege and Plummer G Foxworthy and unknown CRIFF series Hosford whereas soil classes selected were BCD Plummer Sapelo and Pottsburg and E Goldsboro E Goldsboro was strongly preferred, Hosford was very strongly avoided and all others were only slightly preferred or avoided. These soil selection pattern s are similar to those found with the coarse scale analysis where BCD and E represent a mix of savannas with flatwoods and uplands. The B CRIFF series is native to areas with depressions, stream terraces and broad wet flats which would not house suitable habitat for RCWs and were found avoided in the FWC results (hardwood swamps and mixed wetlan d forests). The G CRIFF series is associated with sandhills mixed with longleaf pines or sand pines ; it is surprising that G Foxworthy was avoided. In ONF, almost all sandhill forest s are the CRIFF G series but RCWs are only found on the A statula sand soi l type. All soil type s selected and avoided in the final fine scale model were found only in ANF. The ANF has the largest representation of RCW populations and is the only recovered population. The samples taken for this analysis represent ed a 20% sample o f all forests giving the heaviest weight to the most successful forest. Therefore, it is not surprising that the results are in f avor of soils found only in ANF.
74 Objective 2: To Map the Location of Habitat Features Selected b y RCW s in Other National Forests in Florida to the Region Abandoned b y R CW s i n ONF The vegetation in the Church Lake unit of ONF is extremely diverse (Figure 4 2 Table 4 3 ) compared to Riverside Island and Paisley Woods (study areas B and C). While the region u sed by the current population of RCW s consist s of 90% sandhills, the Church Lake unit only has 11% in sandhills Cluster l ocations for translocations can not be based on the habitat conditions currently used by RCWs in study areas B and C and therefore I co nsidered the habitat characteristics of areas used by RCWs in other national forests in F lorida. The Church Lake unit has 7,400 ha of s andhills and flatwoods ecosystems The weighted overlay analysis categorize d all variables on a scale from low to high pr eference s based on suitability of the variables determined in objective one and this was then mapped over that portion of the Church Lake area that may potentially be managed for RCWS ( sandhills and flatwoods ) The weighted overlay scale represented how RCW selected habitat features in a range from very strongly avoided to very strongly preferred The results from the first objective consistently foun d variables to be slightly preferred or slightly avoided rather than more extreme options such as mo derately, strongly, and very strongly preferred or avoided The lack of strong preferences or avoidances may have been caused by the variation in habitat s election of RCWs from one national forest to another. ANF has the largest population of RCWs and the only forest with a recovered population and therefore had the most influence in the analysis (67%; 320 sample points out of 480). OsNF was given the second most influence with 100 sample point (21%) and Ocala last with 60 sample points (13%).
75 When the resu lts were mapped onto study area A (figure s 4 3 and 4 5 ), it becomes clear that a large portion of the study area could be considered as habitat for translocations. Depending on the quality of the habitat and its connectivity a range of 8 54 clusters could be managed for. The recovery plan ( USFWS 1985) for RCWs describes the quality of habitat based on a number of variables: pine and hardwood basal area and stems per hectares; the stems per hectare and basal area of large pines; the percent of ground cover species including wiregrass and herbs; the density of hardwoods found in the midstory; and the species found in the overstory. Areas of lower quality habitat would require larger habitat range per cluster than areas of higher quality habitat to ensure prop er amount foraging areas. Given the current condition of vegetation in the Church Lake area, this region would probably be considered low quality habitat, and therefore capable of providing habitat for 8 32 clusters. Management Implications Consideration for the expansion of red cockaded wood pecker habitat in the Church Lake management unit of the Ocala National Forest (Figure 1 1; study area A) should take into account vegetative communities selected by RCWs on o the r National Forest s in Florida With the completion of future plans for RCW habitat restoration in Church Lake, there is a potential for the introduc tion of 32 clusters into the area. The restoration should be based on maintenance of old growth longleaf pi nes >70 years old found in sandhills and pineland habitats. Hardwood swamps and mixed wetland forest should be avoided when planning recruitment cluster locations Future research should explore other variables significant to RCW habitat selection. Additionally, d istance to avoided habitat in respect to ecotone tolerance was not studied here and could affect management decisions for placement of recruitment cluster s Fu rther research should
76 review areas of transitional zones between avoided and selected areas to test the tol erance of RCWs for ecotones.
77 Table 4 1. Variables that differentiated between used and available habitat for red cockaded woodpeckers ( RCWs ) from all nation al forests in Florida at a coarse scale using data from Florida Natural Areas Inventory ( FNAI ) Cooperative Research in Forest Ferti li zation ( CRIFF ) and stand initiation dates Parameter estimate standard error (SE), and odds ratios with 95% confidence intervals (CI) Adjusted odds ratio Coarse s cale Parameter value Parameter estimate SE Estimate 95% CI P FNAI class Freshwater forested wetlands 0.09 0.01 0.92 0.89 0.95 < 0 .0001 High pine/ scrub 0.02 0.01 1.02 1.01 1.04 0.00 28 Hardwood forested uplands 0.32 0.14 0.73 0.55 0.95 0.02 00 Pine flatwoo ds/ dry prairie 0.04 0.01 1.04 1.02 1.05 < 0 .0001 Stand i nitiation 1899 1908 0.06 0.02 1.06 1.02 1.11 0.00 36 1909 1918 0.07 0.01 1.07 1.04 1.09 < 0 .0001 1919 1928 0.06 0.01 1.06 1.04 1.08 < 0 .0001 1928 1938 0.04 0.01 1.04 1.02 1.05 < 0 .0001 CRIFF BCD S avannas/flatwoods ** 0.09 0.02 1.09 1.06 1.13 <0.0001 BD S avannas/flatwoods ** 0.19 0.06 1.21 1.08 1.35 0.0010 C Flatwoods** 0.02 0.01 1.02 1.00 1.03 0.0131 E Uplands** 0.63 0.25 1.88 1.16 3.04 0.0102 **Variables included in the logistic regression but not entered in the weighted overlay analysis.
78 Table 4 2 Variables that differentiated between used and available habitat for RCWs from all nation al forests in Florida at a fine scale using Florida Fish and Wildlife Conservation Commission ( FWC ) forest type United States Forest Service ( USFS ) forest type stand initiation dates and USFS soil type Parameter estimate standard error (SE), and odds ratios with 95% confidence intervals (CI) Adjusted odds ratio Fine Scale Parameter value Parameter estimate SE Estimate 95% CI P FWC f orest t ype Hardwood swamp 0.17 0.06 0.85 0.75 0.95 0.0054 Mixed wetland forest 0.17 0.04 0.85 0.78 0.92 0.0002 Pinelands 0.05 0.01 1.05 1.03 1.08 <0.0001 Sandhills* 0.07 0.02 1.07 1.04 1.11 <0.0001 USFS forest type FT46 B ottomland hardwood / yellow pine 0.1 0 0.04 0.91 0.83 0.99 0.0285 FT21 Longleaf p ine 0.07 0.01 1.07 1.05 1.1 0 <0.0001 FT23 Pond cypress 0.1 0 0.04 1.11 1.03 1.19 0.0036 FT22 Slash p ine* 0.02 0.01 1.02 1 .00 1.04 0.0954 FT14 Slash p ine /h ardwoo ds 0.07 0.03 0.93 0.88 0.99 0.0159 FT0 Unkn own 0.1 0 0.03 0.91 0.85 0.97 0.0024 Stand initiation 1909 1918 0.05 0.02 1.06 1.02 1.09 0.0033 1919 1928 0.06 0.01 1.06 1.04 1.09 <0.0001 1929 1938 0.05 0.01 1.05 1.03 1.07 <0.0001 USFS soil type B Rutlege and Plummer ** 0.04 0.02 0.96 0.93 1 .00 0.0383 BCD Plummer, Sapelo, and Pottsburg ** 0.23 0.05 1.26 1.13 1.4 0 <0.0001 E Goldsboro ** 1.2 0 0.49 3.31 1.27 8.65 0.0145 G Foxworthy ** 0.06 0.03 0.94 0.88 1 .00 0.0628 Unkn own Hosford ** 2.83 0.9 0 0.06 0.01 0.35 0.0017 *Correlated variables not included in the logistic regression but entered in the weighted overlay analysis. **Variables included in the logistic regression but not entered in the weighted overlay analysis.
79 Table 4 3 Percent coverage of each ecosystem across all three study areas in Ocala National Forest (ONF) in hectares (ha) Ecosystems Riverside Paisley Church Lake Non forested 15 (0%) 22.6 (0%) 140.3 (1%) Pine flatwoods 0 (0%) 112.5 (2%) 5043.4 (25%) Private 150.3 (2%) 125.5 (2%) 7791.1 (38%) Sandhills 5520.8 (89%) 4944.2 (93%) 2309.3 (11%) Scrub 436.4 (7%) 39.6 (1%) 781.8 (4%) Swamp 16.3 (0%) 0 .0 (0%) 2059 .0 (10%) Water 32.2 (1%) 19.2 (0%) 1158.5 (6%) Wet Prairies 0 (0%) 39.6 (1%) 1215.3 (6%) Total 6171.0 5303.2 20498.8 Table 4 4 Number of RCW clusters that could be supported in Church Lake using the variables selected in the coarse and fine scale analyse s and the r ecovery p Scale Habitat quality Area containing selected habitat features (ha) Potential number of c lusters supported Coarse High ( requires 49 ha) 2654 54 Low ( requires 80 120 ha) 1737 2314 28 32 Fine High ( requires 49 ha) 1934 39 Low ( requires 80 120 ha) 1016 1530 8 19
80 Figure 4 1. Map of the Osceola National Forest ( OsNF ) with b lack circles represent ing buffers around current and historical red cockaded woodpecker ( RCW ) sites ( used areas ) and empty circles represent ing randomly selected sites ( available areas).
81 Figure 4 2 Ecosystem distribution of Church Lake management unit (study area A) in the western region of the Ocala National Forest ( ONF )
82 Figure 4 3 Weighted overlay coarse scale results of RCW habitat use mapped onto potential RCW habitat areas in study area A (Church Lake boundary). Potential RCW habitat areas are sandhill and pine fl atwoods ecosystems.
83 Figure 4 4. Weighted overlay fine scale results of RCW habitat use mapped onto potential RCW habitat areas in study area A (Church Lake boundary). Potential RCW habitat areas are sandhill and pine flatwoods ecosystems.
84 Figure 4 5 Coarse scale results of RCW habitat use from weighted overlay analysis mapped onto study area A and dissolved to identify large contiguous blocks of potential habitat area. Connected areas >4 9 ha (highlighted in black) are recommended for consider ati on as future nesting areas.
85 Figure 4 6 Fine scale results of RCW habitat use from weighted overlay analysis mapped onto study area A and dissolved to identify large contiguous blocks of potential habitat area. Connected areas >4 9 ha (highlighted in black) are recommended for consider ation as future nesting areas.
86 CHAPTER 5 GENERAL CONCLUSION The red nited S tates with isolated pockets of potential breeding habitat I initiated this study to gain a better understanding of the biology of these birds near the southern edge of the species range, in the Ocala National Forest (ONF) I develop ed two objectives to determine the influential factors driving productivity of red cockaded wo odpeckers ( RCWs ) in the ONF : (1) examine the relationship between the 2010 breeding season and the current habitat conditions and (2) compare relationships across a ten year period (2001 2010) between RCW productivity and various habitat conditions In t he first objective, I used 48 active and inactive clusters out of 91 available based on a stratified random selection of RCW productivity and fire frequency Variables tested within each cluster included stand scale features, landscape scale features, mana gement practices, spatial characteristics and productivity which was used as the response variable Productivity was evaluated three ways: active vs inactive clusters active clusters successfully produced fledglings vs unsuccessful nest attempt and activ e clusters successfully fledg ing one vs more than one offspring Logistic regression in SAS (v 9.2 20 08 ) was used to test all variables Results demonstrated that decreasing the percent bareground and hardwood basal area increased the odds of having a pot ential breeding group Increasing the cover of shrubs >1. 5 meter ( m ) also increased the odds None of the variables investigated were successful in distinguishing between active clusters successfully producing fledglings vs unsuccessful nest attempts Incr easing the distance to the nearest neighbor and the percent cover of shrubs <1. 5 m increased the odds of having two rather than one fledgling Further research should be conducted on
87 how height and density of oak species in the groundcover layer effect the habitat RCWs select or avoid. The second objective used data from all available clusters Variables tested included landscape scale features, management practices, spatial characteristics, precipitation and productivity as the response variable Productivity was split in the same manner as above and tested using generalized linear mixed models (Glimmix) in SAS (v 9.2, 2008) which allows for repeated measures over time Results showed that increasing the proportion of interior habitat or decreasing the proportion of non sandhill habitat (correlated variables) and translocating a single bird to a single resident bird increased the odds of having a potential breeding group Increased annual rainfall decreased the odds but this variable represents the forest as a whole and would be a stronger predictor if more region s of the forest were sampled The odds of having at least one fledgling produced in a cluster increased as the proportion of interior habitat increased, and also as the number of years since last hardwood treatment increased Additionally, research on the effects of hardwood treatment over time on RCW productivity is suggested. In conclusion, habitat features of the ONF area similar to habitat described as ideal for RCWs in the recovery plan ( USFWS 1985) except for one unique feature of ONF that is influencing habitat selection not mentioned in the recovery plan. The largest sand pine scrub ecosystem located in ONF is completely surrounding an active population of RCWs while the other populati on is surrounded by other unsuitable habitat Within the interior of the habitat, birds are responding normally by increasing the number of potential breeding group over 5% annually and successfully fledgling
88 offspring. These birds are avoiding the ecotone between unsuitable habitat and longleaf pine sandhills The Ocala National Forest had only 7 potential breeding groups of red cockaded woodpeckers in 1993 Intensive translocations were implemented to bolste r an unstable subpopulation where 81 RCWs were translocated to 50 clusters from 1993 to 2005 with a 47 % success over 13 years The first objective of this portion of my study was to examine changes in productivity in fledgling success or potential breeding groups in the vicinity of the translocation at two spatial scales: 2. 1 kilometers ( km ) or 4 nearest neighbors The paired t tests revealed the 16 selected translocations had no effect on improving fledgling productivity within two years but had a slight e ffect on increasing the number of potential breeding groups ( PBGs ) at the 2. 1 km distance Neither was improved at the 4 nearest neighbor range Additional research should examine longer time periods to assess whether productivity of translocated individuals increases as time since release increases. The second objective investigated the influence of gender, age and type of translocation of the newly released birds on their site fi delity Site fidelity seemed dependent on number of available territories and resident PBGs near the release site with birds showing higher fidelity in the region with fewer available territories and more resident PBGs nearby Success of retaining translo cated birds was similar among gender and type with some difference in region RCWs on average dispersed 1. 9 km from the release site and had successful nest attempts after the age of 2. The type of translocation and region was important when comparing nest success : birds released in pairs had a 29 % success rate compared to single bird release s with a 17 % success rate and 34% of birds in Riverside Island produced offspring compared to
89 15% in Paisley Woods Because of the social characteristics of this speci es, it is recommended to continue translocation through paired release s near other resident birds. Further research should be conducted to determine if there is a significant benefit to the population from birds dispersing beyond the 2.1 km distance and 2 year time frame I evaluated with objective 1. Additionally, research on translocated birds, both male and female, should test whether translocating older birds results in a higher success rate than I found with translocating first year hatchlings. The Oca la National Forest is considering the expansion of red cockaded wood pecker habitat in the Church Lake management unit (Figure 1 1; study area A). D oing so will provide three sub populations in ONF and help drive future habitat restoration activities to support this endangered species. The history of the Church Lake area shows a historical population of RCWs extirpated in 1989. Reasons for loss are unknown but are assumed from a decline in quality habitat as areas were managed for timber harvesting. The c urrent habitat is a mosaic pattern of varying ecosystems including non forest, private lands, pine flatwoods, sandhills, scrub, swamp, water and wet prairies (Figure 4 2 ). RCW management is limited to pine flatwoods and sandhills which exists in only 7,40 0 ha of varying qualities of habitat in this region I determined which habitat variables distinguish between used and available areas in other national forests in Florida to determine the best method for selecting suitable habitat areas for future translo cations in Church Lake. Forty five percent habitat area was selected for according to RCW selection criteria determined though my analyses This potentially could result in over 50 new potential groups in the third subpopulat ion on Ocala National Forest. Further research should review areas of
90 transitional zones between avoided and selected areas to test the tolerance of RCWs for ecotones.
91 APPENDIX A VARIABLES TESTED AGA INST R ED COCKADED W OODPECKER PRODUCTIVITY IN 2 010 Table A 1 Means with standard deviation and range of each variable tested in Chapter 2 objective 1 Data collected in 2010. Model 1 tested the presence of active potential breeding groups ( PBGs ) in a cluster. Model 2 tested for nest success among act ive PBGs. Model 3 tested the scale of nest success from Model 2. Model 1 Model 2 Model 3 Variables Mean (SD) Range Mean (SD) Range Mean (SD) Range Fledgling s p roduced (Count) 0.4 (0.5) 0 1 0.5 (0.5) 0 1 1 .0 (0) 1 2 Stand scal e features Pine b asal a rea (ft^2/ac) 42.4 (15.5) 0 71 43.1 (16.6) 0 71 42.9 (19.6) 0 71 Hardwood b asal a rea (ft^2/ac) 4.6 (4.8) 0 25 3.7 (4.7) 0 25 3.1 (3.4) 0 14 Pine t rees per a cre 14.5 (6.8) 0 28 15.8 (6.2) 3 28 15.7 (6.7) 8 28 Dominant m idstory 0.5 (0.8) 0 2 0.5 (0.8) 0 2 0.3 (0.7) 0 2 Wiregrass (%) 0.2 (0.1) 0. 1 0.6 0.3 (0.1) 0.1 0.6 0.3 (0.1) 0.1 0.6 Herb (%) 0.2 (0.1) 0.0 0.3 0.1 (0.1) 0.0 0. 3 0.1 (0.1) 0. 1 0. 3 Shrub <1. 5 m (%) 0.2 (0.1) 0.1 0.4 0.2 (0.1) 0.1 0. 4 0.2 (0.1) 0.1 0.3 Shrub >1. 5 m (%) 0 .0 (0 .0 ) 0 .0 0.2 0 .0 (0 .0 ) 0 .0 0. 2 0 .0 (0 .0 ) 0 .0 0.1 Pine l itter (%) 0.4 (0.1) 0.1 0.6 0.4 (0.1) 0.1 0.6 0.4 (0.1) 0. 2 0.6 Bareground (%) 0.2 (0.1) 0.0 0.4 0.1 (0.1) 0.0 0. 4 0.1 (0.1) 0.0 0. 3 Landscape scale features Size of i slands (ha) 4493.1 (1399.5) 2386.4 6055.3 4467.1 (1360.5) 2386.4 6055.3 4730.1 (1039.9) 3957.0 6055.3 Interior (%) 0.7 (0.3) 0. 1 1 .0 0.8 (0.3) 0. 1 1 .0 0.8 (0.3) 0. 1 1 .0 N on s andhill (%) 0 (0.1) 0 0. 3 0 (0.1) 0 0. 3 0 (0.1) 0 0. 3
92 Table A 1 C ontinued Model 1 Model 2 Model 3 Variables Mean (SD) Range Mean (SD) Range Mean (SD) Range Management practices Prescribed fires (count) 4.3 (1.2) 3 8 4 .0 (0.8) 3 6 4.2 (1) 3 6 Year since last Rx fire (count) 1.3 (0.4) 1 2 1.3 (0.5) 1 2 1.3 (0.5) 1 2 H ard w oo d tr tmn ts (count) 0.2 (0.4) 0 1 0.2 (0.4) 0 1 0.3 (0.5) 0 1 Years since last h ard w oo d tr tm nt (count) 13.1 (4.0) 1 15 13 .0 (3.9) 1 15 12.5 (4.5) 1 15 Artifi cial cavities (count) 5.6 (2.9) 1 19 6 .0 (3.2) 1 19 5.1 (1.7) 1 7 Ratio of a rtificial cavities to n aturals (%) 0.7 (0.2) 0. 1 1 0.7 (0.2) 0.1 1 0.7 (0.3) 0.1 1 Spatial characteristics Nearest n eighbor (m) 0.7 (0.2) 0.4 1.4 0.7 (0.2) 0.4 1.3 0.7 (0.2) 0.4 1.2 Overlapping neighbors (count) 1.1 (0.8) 0 3 1.2 (0.9) 0 3 1.2 (0.9) 0 3 Overlap ping neighbors (%) 0.1 (0.2) 0 0.7 0.2 (0.2) 0 0.5 0.2 (0.2) 0 0.5 Translocations (count) 0 .0 (0.1) 0 1 0 .0 (0) 0 0 0 .0 (0) 0 0
93 APPENDIX B NATIONAL FORESTS OF FLORIDA VEGETATIVE DESCRIPTI ON Table B 1 Vegetative description from Florida Natural Areas Inventory (FNAI) Florida Land Use, Land Cover Classification System (FLUCCS) Florida Fish and Wildlife Conservation Commission (FWC) 2003 and the U nited S tates (USFS) Forest Type FNAI FWC USFS forest type Freshwater forested wetlands Bay swamp 68 Sweetbay / swamp tupelo/ red maple Freshwater forested wetlands Bottomland hardwood forest 46 Bottomland hardwood/ yellow pine Freshwater forested wetlands Bottomland hardwood forest 61 Swamp chestnut oak/ cherry bark oak Freshwater forested wetlands Bottomland hardwood forest 65 Overcup oak water hickory Freshwater forested wet lands Cypress swamp 23 Pond cypress Freshwater forested wetlands Cypress swamp 24 Baldcypress Freshwater forested wetlands Cypress swamp 67 Baldcypress/ water tupelo Freshwater forested wetlands Cypress swamp 79 Slash pine/ cypress Freshwater forested w etlands Hardwood swamp Freshwater forested wetlands Mixed wetland forest Freshwater non forested wetlands Freshwater marsh/ wet prairie 98 Undrained flatwoods Freshwater non forested wetlands Open water Freshwater non forested wetlands Shrub/ brushland 99 Brush species Freshwater non forested wetlands Shrub swamp Hardwood forested uplands Hardwood h ammocks and f orest 47 White oak / black oak/ yellow pine Hardwood forested uplands Hardwood h ammocks and f orest 53 White oak/northern red oak/ hickory Hardwood forested uplands Hardwood h ammocks and f orest 58 Sweetgum/ yellow poplar Hardwood forested uplands Hardwood h ammocks and f orest 62 Sweet gum/ oak Hardwood forested uplands Hardwood h ammocks and f orest 77 Oak hammock Hardwood forested uplands Hardwood h ammocks and f orest 97 Live oak Hardwood forested uplands Mixed pine/ hardwood forest 14 Slash pine/ hardwood
94 Table B 1 Continued FNAI FWC USFS forest type Hardwood forested uplands Mixed pine/ hardwood forest 18 Pond pine hardwood Ha rdwood forested uplands Mixed pine/ hardwood forest 19 Sand pine hardwood Hardwood forested uplands Mixed pine/ hardwood forest 26 Longleaf pine / hardwood Hardwood forested uplands Mixed pine/ hardwood forest 40 Hardwood / pond pine Hardwood forested uplands Mixed pine/ hardwood forest 44 Southern red oak / yellow pine Hardwood forested uplands Mixed pine/ hardwood forest 48 Northern red oak / hickory yellow pine Hardwood forested uplands Mixed pine/ hardwood forest 64 Laurel oak willow oak High pine/scrub Xeric oak scrub 57 Scrub oak High pine/scrub Sand pine scrub 34 Sand pine High pine/scrub Sand pine scrub 49 Bear oak/southern scrub oak/ yellow pine High Pine/Scrub Sandhill s 21 Longleaf pine Pine flatwoods/dry prairie Dry prairie Pine flatwoods/dry prairie Pinelands 22 Slash pine Pine flatwoods/dry prairie Pinelands 25 Yellow pine Pine flatwoods/dry prairie Pinelands 31 Loblolly pine Pine flatwoods/dry prairie Pinelands 36 Pond pine Ag (FLUCCS) Improved pasture Ag (FLUCCS) Row/field crop s Urban (FLUCCS) Extractive Urban (FLUCCS) High impact urban Urban (FLUCCS) Low impact urban Barren (FLUCCS) Bare soil/clearcut Barren (FLUCCS) Sand/beach
95 LIST OF REFERENCES Allen, D. H., K. E. Franzreb, and R. F. Escano. 1993. Efficacy of translocation strategies for red cockaded woodpeckers. Wildlife Society Bulletin 21:155 159. Allen, J. C., S. M. Krieger, and J. R. Walters. 2006. Associations of breeding birds with fire influenced and riparian upland gradients in a longleaf pine ecosystem. Auk 123:1110 1128. ArcMAP. 2009. v 9.3.1. Environmental Systems Research Institute, Inc Redlands, CA ArcMAP. 2010. v 10. Environ mental Systems Research Institute, Inc Redlands, CA Beever, J. W., and K. A. Dryden. 1992. Red cockaded woodpeckers and hydric slash pine flatwoods. Transactions of the North American Wildlife and Natural Resources Conference 57:693 700. Carrie, N. R., R. N. Conner, D. C. Rudolph, and D. K. Carrie. 1999. Reintroduction and postrelease movements of red cockaded woodpecker groups in eastern Texas. Journal of Wildlife Management 63:824 832. Chadwick, A. N. 2004. South Carolina's Safe Harbor program for red cockaded woodpeckers. Pages 180 184 in R. Costa, and S. J. Daniels, eds Red cockaded woodpecker: road to recovery. Hancock House, Blaine, Washington, USA. Chen, H., P. Cohen and S. Chen. 2010. How big is a big odds ratio? Interpreting the magnitudes of odds ratios in epidemiological studies. Communications in Statistics 39:4, 860 864. Conner, R. N., D. C. Rudolph, D. L. Kulhavy, and A. E. Snow. 1991. Causes of mortality of red cockaded woodpecker cavity trees. Journal of Wildlife Management 55:531 537. C onner, R. N., D. C. Rudolph, R. R. Schaefer, and D. Saenz. 1997. Long distance dispersal of red cockaded woodpeckers in Texas. Wilson Bulletin 109:157 160. Conner, R. N., D. Saenz, D. C. Rudolph, W. G. Ross, and D. L. Kulhavy. 1998. Red cockaded woodpeck er nest cavity selection: relationships with cavity age and resin production. Auk 115:447 454. Conner, R. N., D. C. Rudolph, and J. R. Walters. 2001. The red cockaded woodpecker: surviving in a fire maintained ecosystem. University of Texas Press, Aust in, TX. Conner, R. N., C. E. Shackelford, R. R. Schaefer, D. Saenz, and D. C. Rudolph. 2002. Avian community response to southern pine ecosystem restoration for red cockaded woodpeckers. Wilson Bulletin 114:324 332.
96 Conner, R. N., D. Saenz, R. R. Schaefer, J. R. McCormick, D. C. Rudolph, and D. B. Burt. 2005. Rainfall, El Nino, and reproduction of red cockaded woodpeckers. Southeastern Naturalist 4:347 354. Costa, R. 1995. Biological Opinion on the U. S. Forest Service Environmental Impact Statement for the management of the red cockaded woodpecker and its habitat on national forests in the southern region. Pp. 1 192 in USDA Forest Service, Final Environmental Impact Statement, Volume II, Management Bulletin R8 MB73. Cox, D. R. and E. J. Snell. 1989. Analysi s of Binary Data, Second Edition, London: Chapman and Hall. Cox, J. A. 2001. Red cockaded woodpeckers in the Red Hills region: a GIS based assessment. Wildlife Society Bulletin 29:1278 1288. Daniels, S. J., and J. R. Walters. 2000. Between year breeding di spersal in red cockaded woodpeckers: multiple causes and estimated cost. Ecology 81:2473 2484. Davenport, D. E., R. A. Lancia, J. R. Walters, and P. D. Doerr. 2000. Red cockaded woodpeckers: a relationship between reproductive fitness and habitat in the No rth Carolina Sandhills. Wildlife Society Bulletin 28:426 434. DeLotelle, R. S., R. J. Epting, and J. R. Newman. 1987. Habitat use and territory characteristics of red cockaded woodpeckers in central Florida. Wilson Bulletin 99:202 217. Edwards, J. W., a nd R. Costa. 2004. Range wide success of red cockaded woodpecker translocation. Pages 307 311 in R. Costa, and S. J. Daniels, eds Red cockaded woodpecker: road to recovery. Hancock House, Blaine, Washington, USA. Engstrom, R. T., and G. Mikusinski. 1998. Ecological neighborhoods in red cockaded woodpecker populations. Auk 115:473 478. Franzreb, K. E. 1999. Factors that influence translocation success in the red cockaded woodpecker. Wilson Bulletin 111:38 45. Franzreb, K. E. 2004. The effect of using a "sof t" release on translocation success of red cockaded woodpeckers. Pages 301 306 in R. Costa, and S. J. Daniels, eds Red cockaded woodpecker: road to recovery. Hancock House, Blaine, Washington, USA. Franzreb, K. E. 2006. Implications of home range estimati on in the management of red cockaded woodpeckers in South Carolina. Forest Ecology and Management 228:274 284.
97 Frost, C. C. 1993. Four centuries of changing landscape patterns in the longleaf pine ecosystem. Pp. 17 44 in S. M. Hermann, ed. The longleaf pine ecosystem: ecology, restoration, and management. Tall Timbers Fire Ecology Conference Proceedings, No. 18. Tall Timbers Research Station, Tallahassee, FL. Florida Natural Areas Inventory (FNAI). 2012 Florida Conservation Land (Flma_201203) Florid a State University. Tallahassee, Florida. [Online.] Available at www. fnai.org/gis_data.cfm Florida Fish and Wildlife Conservation Commission (FWC). 2004. Florida Vegetation and Land Cover. State of Florida. Tallahassee, Florida. [Online.] Available at http ://myfwc.com/research/gis/data maps/terrestrial/fl vegetation land cover/ Gault, K. E., J. R. Walters, J. Tomcho, L. F. Phillips, and A. Butler. 2004. Nest success of southeastern American kestrels associated with red cockaded woodpeckers in old growth lon gleaf pine habitat in northwest Florida. Southeastern Naturalist 3:191 204. Gu, W and R. Swihart. 2004. Absent or undetected? Effects of non detection of species occurrence on wildlife habitat models. Biological Conservation 116:195 203. Hagan, G., and R Costa. 2001. Rare woodpeckers reintroduced to North Florida. Endangered Species Bulletin 26:30 31. Hanula, J. L., and K. E. Franzreb. 1998. Source, distribution, and abundance of macroarthropods on the bark of longleaf pine: potential prey of the red cockaded woodpecker. Forest Ecology and Management 102:89 102. Hardesty, J. L., K. E. Gault, and H. F. Percival. 1997. Ecological correlates of red cockaded woodpecker ( Picoides boralis ) foraging preference, habitat use, and home range size in northwest Florida (Eglin Air Force Base). Final Report, Research Work Order 99, Florida Cooperative Fish and Wildlife Research Unit, University of Florida, Gainesville, USA. Ecology LLC. Toronto, ON, Canada Herbez, E.M., M. J. Chamberlain, and D. R. Wood 2011. Using a dult g roups of r ed c ockaded w oodpeckers for t ranslocations and p opulation a ugmentation. Journal of Wildlife Management 75:1568 1573. Hooper, R. G., and R. F. Harlow. 1986. Forest stands selected by foraging red cockaded woodpeckers. U.S. Forest Service Research Paper SE 259. Hooper, R. G., M. R. Lennartz, and H. D. Muse. 1991. Heart rot and cavity tree selection by red cockaded woodpeckers. Journal of Wildlife Management 55:323 327.
98 Hosmer, D. W., and S. Lemeshow. 2000. Applied Logistic Regression. Wiley, New York. Hovis, J. A., and R. F. Labisky. 1985. Vegetative associations of red cocka ded woodpecker colonies in Florida. Wildlife Society Bulletin 13:307 314. James, F. C. 1991. Signs of trouble in the largest remaining population of red cockaded woodpeckers. Auk 108:419 423. James, F. C., C. A. Hess, and D. Kufrin. 1997. Species centere d environmental analysis: indirect effects of fire history on red cockaded woodpeckers. Ecological Applications 7:118 129. Jokela, E. and A. Long. 1999. Using soils to guide fertilizer recommendations for southern pines. University of Florida Cooperative E xtension Service. Jones, C. M., and H. E. Hunt. 1996. Foraging habitat of the red cockaded woodpecker on the D'Arbonne National Wildlife Refuge, Louisiana. Journal of Field Ornithology 67:511 518. Kalisz, P. J. and E. L. Stone. 1984. The longleaf pine i slands of the Ocala National Forest, Florida: a soil study. Ecology 65:1749 1757. Landers, J. L., D. H. Van Lear, and W. D. Boyer. 1995. The longleaf pine forests of the southeast: requiem or renaissance? Journal of Forestry 93:39 44. Lennartz, M. R., and D. G. Heckel. 1987. Population dynamics of a red cockaded woodpecker population in Georgia Piedmont loblolly pine habitat. Pages 48 55 in R. R. Odom, K. A. Riddleberger, and J. C. Ozier, eds Proceedings of the third southeastern nongame and endangered wildlife symposium. Georgia Department Natural Resources, Game and Fish Division, Atlanta, G A USA. Letcher, B. H., J. A. Priddy, J. R. Walters, and L. B. Crowder. 1998. An individual based, spatially explicit simulation model of the population dynamics of the endangered red cockaded woodpecker, Picoides borealis Biological Conservation 86:1 14. Levesque, G. C., C. F. Bailey, and W. G. Montague. 2004. Effects of translocation on nesting success of novice breeders. Page 337 in R. Costa, and S. J. Daniels eds Red cockaded woodpecker: road to recovery. Hancock House, Blaine, Washington, USA. Ligon, J. D. 1970. Behavior and b reeding b iology of the r ed cockaded woodpecker. The Auk 87:255 278. Ligon, J. D., P. B. Stacey, R. N. Conner, C. E. Bock, and C. S. Adkisson. 1986. Report of the American Ornithologists' Union Committee for the conservation of the red cockaded woodpecker. The Auk 103:848 855.
99 Ligon, J. D., W. W. Baker, R. N. Conner, J. A. Jackson, F. C. James, D. C. Rudolph, P. B. Stacey, and J. R. Wal ters. 1991. The conservation crisis the red cockaded woodpecker: on the road to oblivion? Auk 108:200 213. Loeb, S W. Pepper, and A. T. Doyle. 1992. Habitat c haracteristics of a ctive and a bandoned r ed cockaded woodpecker colonies. Southern Journal of Applied Forestry 16:120 125. Myers, R. L. 1985. Fire and the dynamic relationship between Florida sandhill and sand pine scrub vegetation. Bulletin of the Torrey Botanical Club 112:241 252. Myers, R. L. and J. J. Ewel. 1990. Ecosystems of Florida Univers ity of Central Florida, Orlando FL Neal, J. C., D. A. James, W. G. Montague, and J. E. Johnson. 1993. Effects of weather and helpers on survival of nestling red cockaded woodpeckers. Wilson Bulletin 105:666 673. Nesbitt, S. A., D. T. Gilbert, and D. B. B arbour. 1978. Red cockaded woodpecker fall movements in a Florida flatwoods community. Auk 95:145 151. Pasinelli, G., and J. R. Walters. 2002. Social and environmental factors affect natal dispersal and philopatry of male red cockaded woodpeckers. Ecolog y 83:2229 2239. Peduzzi, P., J. Concato, E. Kemper, T. R. Holford, and A. R. Feinstein. 1996. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 49:1373 1379. Petrick, C. J. and D. Krusac. 2012. Nati onal Forests in Florida RCW Review and Audit. US Forest Service. Porter, M. L., and R. F. Labisky. 1986. Home range and foraging habitat of red cockaded woodpeckers in northern Florida. Journal of Wildlife Management 50:239 247. Provencher, L., N. M. Gob ris, L. A. Brennan, D. R. Gordon, and J. L. Hardesty. 2002. Breeding bird response to midstory hardwood reduction in Florida sandhill longleaf pine forests. Journal of Wildlife Management 66:641 661. Robbins, L. E., and R. L. Myers. 1992. Seasonal effects of prescribed burning in Florida: a review. Tall Timbers Research Station, Miscellaneous Publication Number 8 Tallahassee, FL. Rudolph, D. C., R. N. Conner, D. K. Carrie, and R. R. Schaefer. 1992. Experimental reintroduction of red cockaded woodpeckers. Auk 109:914 916. S AS for Windows. 20 10 v 4.3 C ary : SAS Institute Inc.
100 Saenz, D., R. N. Conner, D. C. Rudolph, and R. T. Engstrom. 2001. Is a "hands off approach appropriate for red cockaded woodpecker conservation in twenty first century landscapes? Wildlife Society Bulletin 29:956 966. Saenz, D., K. A. Baum, R. N. Connor, D. C. Rudolph, and R. Costa. 2002. Large scale translocation strategies for reintroduced red cockaded woodpeckers. Journal of Wildlife Management 66:212 221. Schiegg, K., G. Pasinel li, J. Walters, and S. Daniels. 2002 a Inbreeding and experience affect response to climate change by endangered woodpeckers. Proceedings of t he Royal Society of London 269:1153 1159 Schiegg, K., J. R. Walters, and J. A. Priddy. 2002 b The consequences of disrupted dispersal in fragmented red cockaded woodpecker Picoides borealis populations. Journal of Animal Ecology 71:710 721. Shackelford, C. E., and R. N. Conner. 1997. Woodpecker abundance and habitat use in three forest types in eastern Texas. Wil son Bulletin 109:614 629. Simberloff, D. 1993. Species area and fragmentation effects on old growth forests: Prospects for long leaf pine communities in Hermann, S.M. e d s Proceedings of the 18th Tall Timbers Fire Ecology Conference, The Long leaf Pine Eco system: Ecology, Restoration and Management. Tall Timbers Research Station, Tallahassee, FL 1 13. Smith, E., and R. Martin. 1995. Red cockaded woodpecker distribution and status in Louisiana. Pages 452 456 in D. L. Kulhavy, R. G. Hooper, and R. Costa, ed s Red cockaded woodpecker: recovery, ecology and management. Center for Applied Studies in Forestry, College of Forestry, Stephen F. Austin State University, Nacogdoches, Texas, USA. Stith, B. M., J. W. Fitzpatrick, G. E. Woolfenden, and B. Pranty. 1996. Classification and conservation of metapopulations: a case study of the Florida Scrub Jay. Pp. 187 215 in D. R. McCullough, ed. Metapopulations and wildlife conservation. Island Press, Washington DC. Taylor, T. B., and J. R. Walters. 2004. Arthropod communities: a review of possible links between fire history and red cockaded woodpeckers. Pages 653 656 in R. Costa, and S. J. Daniels, eds Red cockaded woodpecker: road to recovery. Hancock House, Blaine, Washington, USA. U.S. Fish and Wildlife Service 1985. Red cockaded woodpecker recovery plan. U.S. Fish and Wildl ife Serv ice Southeast Region Atlanta, GA USA Vittinghoff, E. and C. E. McCulloch. 200 6 Relaxing the rule of ten events per variable in logistic regression and cox regression. American Jo urnal of Epidemiology 165:710 718.
101 Walker, J. and A. Silletti. 2006. The Longleaf Pine Ecosystem Springer New York Walters, J. R., K. E. Gault, B. W. Hagedorn, C. J. Petrick, L. F. Phillips, Jr., J. Tomcho Jr., and A. Butler. 2004. Effectiveness of re cruitment clusters and intrapopulation translocation in promoting growth of the red cockaded woodpecker population on Eglin Air Force Base, Florida. Pages 325 334 in R. Costa, and S. J. Daniels, eds Red cockaded woodpecker: road to recovery. Hancock House Blaine, Washington, USA. Walters, J. R., S. J. Daniels, J. H. Carter, III, and P. D. Doerr. 2002. Defining quality of red cockaded woodpecker foraging habitat based on habitat use and fitness. Journal of Wildlife Management 66:1064 1082. Walters, J. R., P. D. Doerr, and J. H. Carter, III. 1988. The cooperative breeding system of the red cockaded woodpecker. Ethology 78:275 305. Ware, S., C. C. Frost, and P. Doerr. 1993. Southern mixed hardwood forest: The former longleaf pine forest. Pages 447 493 in W. H. Martin, S. G. Boyce, and A. C. Echternacht, eds Biodiversity of the Southeastern United States: Lowland Terrestrial Communities. Wiley, N.Y. Whitney, E., D. B. Means, and A. Rudloe. 2004. Priceless Florida Natural Ecosystems and Native Species : Pi neapple Press Sarasota, FL Wood, D. R., L. W. Burger Jr., a nd F. J. Vilella. 2005. Red cockaded Woodpecker (Picoides borealis) behavior in a Mississippi loblolly shortleaf pine forest. Journal of the Mississippi Academy of Sciences 50:168 176. Wood, D. R ., F. J. Vilella, and L. W. Burger Jr. 2008. Red cockaded woodpecker home range use and macrohabitat selection in a loblolly shortleaf pine forest. Wilson Journal of Ornithology 120:793 800 Zwicker, S. M., and J. R. Walters. 1999. Selection of pines for foraging by red cockaded woodpeckers. Journal of Wildlife Management 63:843 852.
102 BIOGRAPHICAL SKETCH as a student has been a long and prosperous journey. As an forestry undergraduate student, she has had the pleasure of traveling around the country during professional conferences, exploring forests and gaining a new perspective. Originally from F lorida, she never knew it smelled like Chris tmas all year long in a Douglas fir forest and enjoyed light sn ow fall while wearing light clothing because of the difference in relative humidity. As a wildlife grad uate student, her mentality changed to a more independent, research oriented and task focused mind set Still in Florida, she realized the importance of habitat to threatened and endangered wildlife species She developed her education to broaden her horizon of habitat restoration and management Elizabeth applied her thesis work to the current red cockaded woodpecker program on Ocala National and then bey ond the boundaries of her study area to connect with other agencies and continue d building on the mission of conservation.