Wildfe Restoration via Forest Management i Fire-Supresd Longleaf Pine Sandhils by David A. Sten A disertation submited to the Graduate Faculty of Auburn University in partial fulfilment of the requientsor the Dgree of Doctor of Philosophy Auburn, Alabam ecber 12, 2011 Keywords: Aspidoscelis sxlineatus, Bird, Burn, Pinus palustris, Squamate, Restoration Ecology Copyright 2011 by David A. Sten Approved by Craig Guyer, Co-chair, Profesor, Department of Biological Siencs Lora L. Smth, Co-chair, Asocat Scintt, Josph W. Jones E Rearch Center . Mike Conner, Co-ir, Sist, J. Jcolls Nanete E. Chadwik, Ase Profsor, Departent of Biologicl Siencs Js B. Grand, Aociat, School of Forestry & Wildlf 1 ii Abstrac 2 3 4 The onc-extnsive longleaf pine (Pinus palustris) ecosystem of the southeastern United 5 Staes ha ben reducd to a fraction of its hitoric extnt. A fire-adaptd system, mny 6 remining fragments have been fire-suppresed and invaded by hardwood tres, particularly oaks 7 (Quercus sp.). This change in species compositon alters the habita and is to the detriment of 8 wildlfe asmblages aociated with longlaf pine forest. Fire surrogates and prescribed 9 burning have been suggestd as potntil mnagement straegis to restore fire-suppresed and 10 hardwood-invaded longleaf pine forest toarget conditions; due to the unique efects of fire, it is 11 generaly suggested that prescribed burning should follow applicaton of any hardwood removal 12 treament. To detrmine whetr fire surrogates folled by presribed burning afectd wildlfe 13 popultions and asemblages, we sampld for birds and reptiles within 20 experimntl sites and 14 six referenc sites. Experintl sites were initaly subjctd to either mechanial hardwood 15 removal followed by fire, herbicide applicton followed by fire, presribed burning lone, or 16 reained in a fire-suppresed stae (i.e., controls). Folling initial tretment, al sites 17 experiencd over a decade of prescribed burning on an approxitely two-year intrval. We 18 valuatd the effects of a given treatment by comparison of wildlf popultions and asmblages 19 on treatment sits to those on refrenc sits initialy and also after over a decade of prescribed 20 burning. If conditions asociatd with a given tretment wre indistinguishable from those of 21 refenc sites, w cidered this a evidence tha manageent objectives wre mt. Over the 22 long-trm, applicaton of herbicide followd by prescribed burning was the only method tha 23 iii restored bird asemblages to the referenc condition, although species positvely asociated with 24 longleaf pine in refrence condition responded positvely to al tretmnts. Ocupancy 25 probabilits for thes pecis on al treatment sits were indistinguishable from those on 26 refenc sites by t conclusion of the study. Initialy, rptile asmblages within treatmnt sites 27 tread with prescribed burning alone were most simlr to thos of referenc sites; fire 28 surrogates did not imediatly provide an observed benefit. A the conclusion of the study, 29 reptil asblages at l sits were indistinguishabl from those on referenc sites except for 30 asmblages on sites reated with herbicide, suggesting herbicide applicaton was reltively 31 inefective at restoring reptil asemblages. A mark-recapture study of the six-lned racerunner 32 (Aspidoslis sexlineatus) also identifd prescribed burning as efctive. Initialy, bundancs on 33 site treatd with prescribed burning alone, as wel as on its treatd with mechanial hardwood 34 removal followed by fire, were comparabl to abundances within referenc sits. Over time, 35 abundancs atl sits re crable tohose on refre sits. Overal, efcti restoration 36 of wildlfe populations and asemblages in fire-suppresed longleaf pine sandhils wa achieved 37 and prescribed burning over approxitely a decade was generaly sufficent to achieve this 38 result. In general, there ws litle obsrved benefit or need to employ fire surrogats prior to 39 prescribed burning. 40 41 42 43 44 45 iv Acknowledgmnts 46 47 The research described herein has been a collaborative efort in every sens; I have 48 benefited from considerable support dedicatd to resolving both the personal and profesional 49 isue asociated with producing any diserttion. Special thanks are due to Craig Guyer and 50 Lora Smith; their insights, constructive criticsm, and support were instrumental in guiding this 51 disertation and my experienc as doctoral student. Mike Conner, Nanet Chadwick, and 52 Jms Grand provided much helpful guidance as members of my commite. Aknowledgmnts 53 specif to each portion of my research apper at the end ofost chaptrs. 54 Thank you to everyone upon whom I have leaned. The cumulative contributions of the 55 individuals and organizations involved in this work are significnt and I appreciate the 56 opportunity to have my name on the final product. I bear responsibilty for any errors that appear 57 within ts document. 58 59 60 61 62 63 64 65 66 v Table of Contents 67 68 69 Abstrac.........................................................................................................................................ii 70 knowledgmnts........................................................................................................................iv 71 List of Tabls.............................................................................................................................viii 72 ist of Figures...............................................................................................................................x 73 Chapter 1 General Introduction....................................................................................................1 74 The Longlaf Pine Ecosystem.......................................................................................... 75 Role of Fire in Influencing Comunity Compositon...............................................2 76 Fire Suppresion................................................................................................................4 77 Restoration of Fire-adapted Systems................................................................................5 78 Study Area........................................................................................................................7 79 Conceptual Framework.....................................................................................................9 80 Disrtation Outline........................................................................................................11 81 Literaure Cited...............................................................................................................12 82 Chaptr 2 Concptualizng Communities a Individuals with implicatons for conservation and 83 restoration ecology..........................................................................................................26 84 85 Introduction.....................................................................................................................27 86 What is a Community?....................................................................................................29 87 t is an Individual?....................................................................................................31 88 A Framework for Identifying Communities a Individuals............................................33 89 Boundaris...................................................................................................................... 90 vi An Ostensive Definiton.................................................................................................37 91 Communitis are Cohesive.............................................................................................39 92 There Are No Instances..................................................................................................40 93 Applicatons of the Evolutionary Community Concept..................................................42 94 Exoti Speies................................................................................................................. 95 Climat Change...............................................................................................................44 96 Refrence Conditions......................................................................................................45 97 Community Resilenc....................................................................................................46 98 Conclusion......................................................................................................................47 99 Acknowledgmnts...........................................................................................................48 100 Literaure Citd............................................................................................................... 101 Chaptr 3 Restoration of Avian Populations and Asemblages: Long-term Efects of Fire 102 Surrogae and Prescribed Burning.................................................................................61 103 104 Introduction.....................................................................................................................62 105 Methods...........................................................................................................................63 106 Result.............................................................................................................................72 107 Discusion.......................................................................................................................75 108 Aknowledgmnts...........................................................................................................81 109 Literaure Citd...............................................................................................................82 110 Chaptr 4 Restoration of Reptile Asmblages: Long-term Efects of Fire Surrogates and 111 Precribed Burning........................................................................................................108 112 113 Introduction................................................................................................................... 114 vii Methods.........................................................................................................................110 115 Result...........................................................................................................................116 116 Discusion.....................................................................................................................119 117 Aknowledgmnts.........................................................................................................124 118 Literaure Citd.............................................................................................................125 119 Chaptr 5 Six-lined Racerunner, Aspidoscelis sxlineatus, Population Size and Survivorship: 120 Longterm Efts of Fire Surrogat and Prescribed Burning....................................146 121 122 Introduction...................................................................................................................147 123 Methods.........................................................................................................................149 124 Result...........................................................................................................................154 125 Discusion.....................................................................................................................157 126 Aknowledgmnts.........................................................................................................161 127 Literaure Citd............................................................................................................. 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 vii List of Tables 143 144 145 Chapter 3 146 Tabl 1 Models used to evaluate occupancy probabilites for selct bird species detced 1994- 147 2010 to detrmine how their populations reponded to hardwood removal on fire- 148 supprsd longleaf pi sandhils.................................................................................92 149 150 Table 2 Oak and lf pine basl are in treatment and referenc sites before and after oak 151 removal..........................................................................................................................93 152 153 Table 3 P-values aociated with multi-response permutation procedure on pairwse comparisons 154 of treatentnd refrence s (1994, 1998-1999, and 2009-2010)............................95 155 156 Table 4 Bird specis identifd as having a significant asociaton with treatment or referenc 157 sites for al three study periods, Egln Ar Force Ba, Florida.....................................96 158 159 Table 5 Top models explaining occupancy paterns of slct bird species within fire-suppresed 160 longleaf pine sndhils undergoing oak remova, 1994-2010........................................99 161 162 Table 6 Probability of occupancy (and standard errors) for American kestrel and northern 163 bobwhie obsrved on longlef pine sndhils on Eglin Air Force Ba, 1994-2010..101 164 165 Chapter 4 166 Tabl 1 Tre density within hardwood-removal sites and referenc sites, Sant Rosa and 167 Okaloosa Countes, Eglin Air Force Ba, Florida......................................................134 168 169 Table 2 Totl cptures of reptils by treatnt and referenc sites on Eglin Air Force Bas, 170 1997-1998 and 2009-2010...........................................................................................136 171 ix Table 3 P-values aociated with multi-response permutation procedure on pairwse comparisons 172 of treatmentnd refrence s (1997-1998 and 2009-2010).....................................139 173 174 Table 4 Prcnt indicator values for species significantly asociated with a particular tretment 175 on Eglin Air Force Ba, 1997-1998...........................................................................141 176 177 Chapter 5 178 Tabl 1 Model comparison table for POPAN capture-mark-recapture analysis aesing efects 179 on capture probability (p), entry probability (pent) and apparent survival (phi) on 180 Aspidoscelis sxneatus populations in longlef pine sndhilsubjctd to various 181 hardwood removal straegis on Egln Air Forc Base betwen 1997-1998...............172 182 183 Table 2 Model comparison tabl for POPAN capture-mark-recapture analysis aesing efects 184 on capture probability (p), entry probability (pent) and apparent survival (phi) on 185 Aspidoscelis sxneatus populations in longlef pine sndhilsubjctd to various 186 hardwood removal straegis on Egln Air Forc Base betwen 2009-2010...............174 187 188 Table 3 Sx ratios and age clas of Aspidoscelis sxlineatus populations in longleaf pine 189 sandhils subjectd to varioushardwood remova straegis on Eglin Air Forc Base176 190 191 Table 4 Survivorship estimates (phi, and stndard errors) and 95% confidence intrvals for 192 Aspidosceli sxlneatus populions in longleaf pine sandhils subjctd to various 193 hardwood reova tretnts on Eglin Air Forc Base..............................................177 194 195 Table 5 Tree density within hardwood removal sites, Sant Rosa and Okaloosa Counties, Eglin 196 Air Forc Base, Florida.................................................................................................178 197 198 Table 6 Number of marked Aspidoscelis sxlineatus captured within each hardwood removal 199 study site, Snt Rosa and Okaoosa Counties, Eglin Ar Force Bas, Florida, and 200 correponding gross popultion estimt (and stndard errors) of 95% confidence 201 intrvals........................................................................................................................180 202 x 203 204 205 206 207 List ofFigures 208 209 210 Chapter 1 211 Figure 1 Restoration scenario wherein reintroduction of a natural disturbance regime restore a 212 degrad s to a previous conditon along a linear trajecory....................................21 213 214 Figure 2 Conceptualizaton of vari restoration scenarios potntialy leading to ecological 215 recovery........................................................................................................................22 216 217 Figure 3 Conceptualizaton of restoration scenario specif to this diertation, prior to hardwood 218 remova trement.......................................................................................................23 219 220 Figure 4 Conceptualizaton of restoration specif to this diertation. Gren represent fire- 221 suppresd control. Sie tha experiencd hardwood removal are not consided 222 recovered.......................................................................................................................24 223 224 Figure 5 Conceptualizaton of restoration specif to this diertation. After al sites recived 225 managent, they arecovered. The study lacks a true control for 2009-2010. The 226 hollow circle represent our expetaion regarding the positon of a control, should one 227 exist...............................................................................................................................25 228 229 Chaptr 2 230 231 Figure 1 New communities can arise from a variety of diferent proces. Transitons 232 (represnted here a fuzzy bars in thebsnc ofsct geologic events) may occur 233 siply becaus comunitis change constantly over tim (A). Whetr such change (X 234 to Z) is of relvance depends on t scle resrchers desgnated as importnt. In (B) 235 Community W transitons into two communiti (X and Y) following biogeographic 236 divergence (e.g., the division of a cy following a shiftng river channel). If a 237 biogeographic convergence vent merged cunites X a, they would form 238 new comunity Z. Ts changes are likely of ecological reevanc. In (C) community 239 X transitons to a new sta a reult of ether a natura transiton (e.g., succesion) or 240 se anthropogenic disurbance. Transions may also ocur as a reult of reparing 241 cmunity degradation (fuzzy gray bar in D). It i phiosophicaly impossibl to 242 manage a comuniy that has changed to a new community such that i once again 243 beces part of the originalommunity (se txt under ?An Osensive Dfiniton). 244 Howver, it i possible to recreate the srucure and function of the original community 245 (represntd by X1). In al of t abovecnarios, the scale of relvanc to the 246 reearcher may alow for a cunity to expeencom degree of change over tie 247 xi while remaining the same individual............................................................................59 248 249 Figure 2 The number of specis in a given area is a subset of the species that occur in a larger 250 area. Trefore, for exaple, only a st of al make a l 251 semblge (D) unique wil be unique to a smar community (A). Conversely, mny 252 speci within a smal asblage (A) wil not be unique to this habit type but wil be 253 unique to a larger ae (B). The nesed naure of specie amblages across 254 spatial sles suggest the resercr i reponsible for desgnatng the relevant sale 255 when identifying t unique specis amage of a given area.................................60 256 257 Chapter 3 258 Figure 1 Non-metric diensional scaling ordination of bird asemblages observed on fire- 259 suppresd longlaf pinendhils on Egln Air Forc Ba, 1994 and 1998-1999...102 260 261 Figure 2 Non-etric diensional scaling ordination of bird aseblages observed on longleaf 262 pine sandhils followng hardwood removal on Eglin Air Forc Ba, 1998-1999 and 263 2009-2010...................................................................................................................103 264 265 Figure 3 Relationship betwn probability of occupancy and year of study for Bachman?s 266 sparrow pre-reamnt (1994; A), and three years (1998-1999; B) and fourten yers 267 (2009-2010, C) following hardwood removal on firesuppresed longlf pine sndhils, 268 Eglin Air Force Bas, Florida.....................................................................................104 269 270 Figure 4 Relationship betwn probability of occupancy and year of study for brown-headed 271 nuthach pre-reament (1994; A), and three years (1998-1999; B) and fourten yers 272 (2009-2010, C) following hardwood removal on firesuppresed longlaf pine sndhils, 273 Eglin Air Forc Bas, Florida.....................................................................................105 274 275 Figure 5 Relationship betwen probability of occupancy and year of study for red-headed 276 woodpecker pre-reatmnt (1994; A), and three yers (1998-1999; B) and fourten years 277 (2009-2010, C) following hardwood removal on fire-suppresed longlef pine sndhils, 278 Eglin Air Force Bas, Florida.....................................................................................106 279 280 Figure 6 Relationship betwn probability of occupancy and year of study for blue grosbeak pre 281 tremnt (1994; A), and three years (1998-1999; B) and fourten years (2009-2010, C) 282 following hardwood removal on fire-suppresed longlef pine sndhils, Eglin Air Force 283 Base, Florida...............................................................................................................107 284 285 Chapter 4 286 287 Figure 1 Relative proportion of species captured in treatment and referenc sites on Eglin Air 288 Force Bas, 1997-1998..............................................................................................142 289 290 Figure 2 Relative proportion of species captured in treatent and referenc sites on Eglin Air 291 Force Bas, 2009-2010...............................................................................................143 292 xii 293 Figure 3 Non-metric diensional scaling of treatment and referenc sites for 1997-1998 and 294 20092010, Eglin Ar Force Bas, Snt Rosand Okaloosa Counti, Florida........144 295 296 Figure 4 Canonical correspondenc biplot for reptiles cptured in 1997-1998, Eglin Air Force 297 Base, Snt Rosa and Okaloosa Count, Forida.....................................................145 298 299 Chapter 5 300 Figure 1 Man population sizes (gross population and standard errors) of Aspidoscelis 301 sexlineatus in longlaf pine sandhils subjecd to various hardwood remova straegis 302 on Egln Air Forc Bas in 1997-1998 (A) and 2009-2010 (B).................................182 303 304 Figure 2 Mean number of marked adults (and standard errors) of Aspidoscelis sxlineatus in 305 longlf pine sndhils subjecd to various hardwoodremoval stragi on Eglin Air 306 Forc Base in 1997-1998 (A) and 2009-2010 (B)......................................................183 307 308 Figure 3 an number of marked juveniles (and standard errors) of Aspidoscelis sxlineatus in 309 longlef pine sndhils subjctd to various hardwood removal stragi on Egln Air 310 Forc Bas in 1997-1998 (A) and 2009-2010 (B)......................................................184 311 312 1 Chapter 1 General Introduction Abstract. Within ts chapter, I introduce the longleaf pine community and describe important mechanis that maintn(ed) this systm?s intgrity. I provide a relevant bakground to sucding cpters, including the role of disturbance (i.e., fire) in maintning the longleaf pine community, how fire-suppresion has degraded this habita type, and how restoration eforts have ateptd to introduce prescribed burning or other straegis that mimc the efects of fire. I describe the study design of my disertation and include a concptual fraework desribing how I gauged t succes of restoration eforts. Key words: burn, longlaf pine, prescribed fire, rstoration. THE LONGLEAF PINE ECOSYTEM The longleaf pine (Pinus palustris) hitoricaly ranged through the Coastal Pin of the Soutastern Unitd Stes, from North Carolina south to peninsular Florida and west to eastern Texas (Frost 1993, Ware et al. 1993). The pine forest within ts region historicaly contined a relatively open understory and tree canopy. A fire-maintned system, the forest cins numerous species endemic to the habit. As a consequence of fire-suppresion, conversion to off-sit pine planttions, and fragmentation, the extnt of longlaf pine ecosystem has been reduced considerably. In the absnc of fire, hardwood trees become established and reduc habita quality for some longleaf pine endemics. To restore hardwood-invaded forest to an open 2 canopy pine forest, managers often remove hardwood trees in concert with introduction of presribed burning. THE ROL OF FIRE INFLUENCING COMUNITY COMPOSITION In many systems, species have evolved in the presenc of natural disturbances such as fire. Consequently, thes sis may avoid injury or mortality from the dist through morphological or behavioral adapttions (Lotan et al. 1985, Russel et al. 1999). In fire-adapted systes, fire my even be required to facilte succesful reproduction and recruitment for som specis (Clewl 1989, Brewer and Plt 1994). Fire fulfils many rols in the systems in which it occurs; for example, Brockway et al. (2005) described the benefits of fire to a fire-adapted system (in this ca, the longlef pine forest) a, ?(1) maintning the physiognomic charactr of longlef pine bunchgras ecosystems by excluding invasive plants that are poorly adapted to fire, (2) preparing a sedbed favorabl for the regeneration of longlef pine sedlings, (3) reducing the density of understory vegetion and thus providing microsits for a varity of herbaceous plants, (4) stimulating increased sd production by native grase, (5) releasing nutrints imobilized in acumultd phytomas for recycling to the infertil soil and subsquently more rapid uptake by plants, (6) improving forage for grazing, (7) enhancing wildlfe habita, (8) controlling harmful insects and pathogens, and (9) reducing fuel evels and the wildfre hazrd.? Lightning strikes are thought to be one of the primary sources of ignition of natural fires (e.g., Rorig and Ferguson 1999); Native Americans also used fire to manage landscpes (Delcourt and Dlcourt 1997). The frequency of natural fires i thought to vary dending on the 3 system, for exaple in subalpine forest fire isnfrequent (300-400 years; Romme 1982), as compared to boreal forest (69-132 years; Bergeron et al. 2001), nd ponderosa pine forest (1- 125 years; Vebln et al. 2000). Longlef pine forest experienc frequent fire reltive to these other forest types (2-10 years, Ware 1993). In systms where fi occurs very infrequently, such as t subalpine forest, these events are likely to result in forest detruction leading to succesion. In fst wre fi frequently occurs, such as longleaf pine forest, ls fuel aumulats betwen fires, thus fire intensity is generaly low. In thes caes, frequent fires generaly do not result in mortality of native species or shifts in community compositon. The intermdiate disturbance hypothesi (Connel 1978) suggest diversity wil be highest a intdit lvels of disturbance (although empirical evidence does not always support this prediction, e.g., Collins 1992). This high level of diversity is thought to result from a mixture of habita speialst and generalist persistng in a gin area; an intermdiate lvel of disturbance represent a compromise betwen habita types thereby mking the are suitbl for a lrger number of specis. Howver, herein w are interestd in t response of species semblages aociated with a particular habita (i.e., thos that evolved in the presnc of one another, Chapter 2). Therefore, in addition to asemblage-wide analyses, it i also important to evaluate prescribed fire in t context of its efct on aseblages tha evolved with is preenc (.g., Stn et al. 2010). Fire can influence species compositon by fulfilng natural history requirements, thereby maintning species amblages (Gilam and Plat 1999). For example, fire facilts succesful reproduction for longlef pine trees a wel as wiregras (Aristda sp.), both haratristic speies of laf pi forest. Fire exposs soil which is eential for establishment of longlaf pine sedlings (Bruce 1951). In addition, fire removes plnt species 4 tha are potentialy competiors with longleaf pine sedlings (Boyer 1993). In turn, longleaf pine trees are oftn struck by lightning and srve asn ignition source for fires (Plat e al. 1988). Thes fires, when ty occur during the growing season, stimulat wiregras to flower and produce viable sd (Mulligan and Kirkman 2002). These complex proces tha maintn popultions of crtain plnt species, in addition to t wildlf specis that thrive with frequent fire (e.g., Mushinsky 1985, Tucker et al. 2004), suggest a unique eosystem (i.e., a community). The mechanism by which wildlfe benefit from fire may vary; gopher tortois (Gopherus polypheus) benefit from forage plants that requi open ares created and maintned by fire (Yager et al. 2007) and they, along with other reptiles such as the six-lned racerunners (Aspidoscelis sxlineatus), prefer open areas for thermoregulation (Mushinsky 1985). FIRE SUPRESION Fire fquency in North America generaly decreased following Columbian setlment and in particular following the mid-1800?s (e.g., Cuttr and Guyete 1994). In the southeastrn United Stes, fire suppresion was due largely to land use changing to crop farming, pasture, and planttion (Frost 1993, Vn Ler et al. 2005) nd the percption that fire ws inconsistent with preferred land managemnt. One of the consequences of fire suppresion was acumulation of coars woody debris and liter. With this increasd fuel load, forest experiencd iresed potential for catsrophic fires (Varner et al. 2005), which are generaly greatr in intensity than ven fire-dapted speies my tolerat. Another consquenc of fire suppresion s a change in species compositon in response to changes in habita structure (e.g., Gilam and Plat 1999). Frequent burning discourages the 5 establishment of species not adapted to persist in the presenc of fire. When fire is removed, specis rihnes in area is likely to increase initialy as fire-snsitve species colonize. These olonizing species my eventualy out compet the native asemblage, resulting in a decreas in species aociatd with the ancestral condition. For exapl, fire mintns the compositon and stability of the vegetaive community in savannas (Beckage et al. 2009), including longleaf pine forest (Mitcl et al. 2006). RESTORAION OF FIRE-ADPTE SYSTEM The imediat goal of many restoration eforts in fre-adapted systems i not restoration of native asblges per se, but reduction of fuel loads and the potntial for catsrophic wildfre (Agee and Skinner 2005, Schwilk et al. 2009). Once reduction of excesive fuel loads has been achived, frequent and relatively low-intensity fires should maintn this reduced fuel lvel. Howr, reintroducing fire to a long-unburned area my have unintended consquencs, such as excsive mortality of native species (.g., Vrner et al. 2005). As a reult, i is occasionaly necesry to reduce fuel loads via mens other than fire. Reintroduction of natural disturbanc regims i often a goal of restoration eforts, though this straegy alone may underestiate what is necsry to restore a functioning system (e.g., Suding et al. 2004). For example, onc hardwoods beome established in longleaf pine forest, fire alone may be insufficnt to kil mature haood tres (Wldrop et al. 1992). Due to concerns asocited with fuel loads and inability of fire alone to restore fst sructure and function, various fire surrogates have been developed (e.g., Provencher et al. 2001a,b). 6 Fire surrogates include chemical (i.e., herbicide) applicaton and feling and girdling (i.e., mechanial removal) of oak trees. Vrious studies have been atmptd to detrmine the efect of fire surrogates on vegetaion (Brockway et al. 1998, Provencher et al. 2001a), trees (Provencher et al. 2001b), amphibins and reptiles (Greenberg et al. 1994, Lit e al. 2001), smal mals (Grenberg et al. 2006) and birds (Provencher et al. 2002). However, fire has unique efects on an ecosystem (Brockway et al. 2005). As a consequence, mchanial removal or herbiide ppliton alone is generaly considered insufficnt to restore fire-dapted systes (Menges and Gordon 2010). What has emerged from previous studies i that fire surrogates may quickly move a community towards a desired condition or enhanc the efects of ubsquent burning, but fire is likey necesry to restore and maintn that condition (.g., Brockway and Outcalt 2000). Therefore, som have recommended a restoration straegy that includes fire surrogates initialy, followed by reintroduction of frequent fire, for long-trm manageent (Menges and Gordon 2010). There have been limted opportunities to quantify the efcts of this stragy, as it requires long-term monitoring. Howver, Outcalt and Brockway (2010) documentd efective restoration of vegetaion communites via this method. To determine the efcts of this retoration straegy on wildlfe requires large-scl, ong-trm controlld study (Block et al. 2001). My resrch atempted to deteine how wildlfe asmblages aociated with he longleaf pine forest repond to difrent methods of habit restoration. My study is a continuation of a project initiaed in 1994, in which fire-suppresed longleaf pine forest on Eglin Air Force Bas (EAFB) wre subjectd to diferent mans ofhardwood removal (i.e., fire surrogates including herbicide appliaton and mchanil hardwood reoval, as wl as fire alone). The initial study examined short-term efets of these treatments on forest sructure and 7 wildlfe relative to control sites and referenc sites (e.g., Lit e al. 2001, Provencher et al 2001a, b, 2002, 2003). A randomizd-block design was usd to asign treatments to sites in this study. After this inital tretment in the mi-1990s, al tretment sits, including controls, received prescribed fire on a two-thre year rottion until I collctd additional data in 2009-2010. Independent of the study design, several longleaf pine stnds that appered to represent a fire- maintned and natural forest (i.e., a desired future condition) at EAFB were slctd as refenc sites. A noted in Provencher et al. (2001a), referenc sites wre, ?hosen on the basi of the following crita, which indicate t original condition of sandhils: an uneven age distribution of P. palustris; presenc of old-growth P. palustris; abundance of largely herbaceous understory species intrsped with bare ground; a sparse midstory; pres of Picoides borealis (a charatristic bird scies); and a history of frequent growing season fires (Myers 1990)?. I gauged restoration sucs by comparing wildlfe asmblages within sites that experienc habita mnipulation to those within refrenc sits. STUDY AREA EAFB consist largely of a forested miltary reservation (approximately 188,000 ha) located in Sant Ros, Okaloosa and Walton countis, Florida, United Sts. Oficaly stblished in 1935, a large portion of EAFB?s current extent was formd by the addition of the former Choctawhatchee Ntional Forest in 1940. The miltry mison of EAFB has changed considerably over t last cntury; today most acivity relates to the, ?development, tsting, procurement and support of air-delivered weapons? (SAIC 2009). 8 In addition to serving as the site for al things related to any conventional and non- conventional weaponry usd by t Unitd Staes Air Forc, EAFB has a land-use history that includes considerabl exploitaion of the forest now within its confines. For exampl, longleaf pine trees wre harvested in the lter portion of the 19 th century; consequently, there is reltively lit old-growth forest remaining. Howver, EAFB stil contains the greats extent of remaining ol-th longleaf pine forest sndhils (SIC 2009). Mny pine trees wre tapped for turpentine until the 1930?s (SAIC 2009). Until 1989, forest managemnts typified by fire- suppresion, which generaly degraded the quality of longlef pine forest. The, ?primary objective of t Air Force Natural Resources Program is to ensure continued aces to lnd airspace required to acomplish the Air Force mison while maintning thes reources in a healthy condition? (SAIC 2009). EAFB contains vast extnts of natural habitas, t majority of which is longleaf pine sandhils. The current mnagement philosophy of EAFB is baed on guidelines outlind within t Eglin Integratd Natural Resources Management Plan (SAIC 2009) and is baed on ecosystm manageent and biodiversity consrvation, rather than a focus on timber harvest and silviculture, which typified the management philosophy for much of the 20 th century. Management plans are cated in consultion with the United Staes Fih and Wildlf Srvice and the Florida Fish and Wildlfe Conservation Commison, aong other organizations, to ensure land use isn complianc with fderal law in relation to protectd species. Eforts to reforest areas on EAFB in the middle of the 20 th century were typified by establishment of sand pine (Pinus clausa) and slash pine (Pinus eliotti) planttions. Land mnagers began to plnt longleaf pine sedlings by approximately 1980 (SAIC 2009). Today, forest mament activis pertint to longleaf pine forest include removal of sand pine, 9 conversion of pine planttions to longleaf pine, thinning of forest to recreate uneven-aged stand, and salvage of timber (SAIC 2009). Restoration activies include plnting of longlef pine and herbicide and mechanial removal of undesirabl tres. Longleaf pine forest on EAFB are burned frequently as a reult of mison (i.e., miltary) activiy as wl as fire program that conducts precribed burns covering over 28,000 ha eh yer (SAIC 2009). Prescribed fires a ignited on the ground and through aerial ignition from helicopters. The vast majority of EAFB is burned on a fire-turn interval < 10 yers (SAIC 2009) with much of t longlef pine forest d more frequently. CONCEPTUAL FRAMEWORK The goal of ecological restoration is typicaly to move a degraded site(s), via manageent, to a stae comparable to what existed before dedation (National Reserch Council 1992). Restoration may be considered as a proces (Hobbs and Cramer 2008), refrring to management activies changing the conditions on a given site, or a goal, wherein a target sate is achived. Hreaftr, I generaly use t trm restoration when refrring to t proces, and the term reovery to identify when restoration was efective at replicatng the target sat. Evaluating whetr t goals of resttion have ben met requis consideration of appropriate endpoints (Palmer et al. 1997). These endpoints may be characterizd by conditions on refrenc sit, which should srve as represntation of the ancestral condition (White and Walker 1997). If conditions on a restored site become indistinguishabl from those on refrenc sites, it i reasonable to suggest recovery has ocurred. If conditions on a previously degraded sit are distinc from those at referenc sites, by my definiton, recovery has not occurred. 10 Within ts diertation, I evaluate whetr management objectives were mt based on structural endpoints (Plmr et al. 1997), which include specis rihnes, amblge compositon, or population size. I did not measure functional endpoints, which include the abioti and biotic proces typical of refrenc conditions. I have made the asumption that requid ecologial procs are present if the structural endpoints of a site that experiencd restoration are indistinguishable from thos at referenc sites. The simplest explanation of how a degraded sit becoms cparable to a referenc site includes a linear moveent to a recovered stae (Figures 1, 2). Experimntal evide provides some support for the asumption that sructural endpoints can be reached after reintroduction of historic disturbance regimes (Mitsch and Wilson 1996) including those of fire-adapted systems (Copeland et al. 2002). Howver, eological restoration may not procd linerly (Suding et al. 2004, Figure 2) or conform to simple preditions (Hilderbrand et al. 2005). For example, communities may transiton to an altrnat sable sta following habita degradation (Figure 2). In thes cs, retoration may require surpasing ecological thresholds to re-establish ancestral fedback mehanism (e.g., Mrtin and Kirkman 2009) before native asmblges my bcom restablished. T restoration methods used in this study (i.e., prescribed burning alone, mechanial hardwood removal, and herbicide applicaton) were all intended to reduce hardwood density. Therefore, in addition to asuming that he difrent tatmnt sites wre comparable prior to inital tretment (Figure 3), I ase rdwood reoval tretents are functionaly simlar to ech other before repeated-prescribed fire was introduced as wl as ftily silr to ech otr after repeated-fire ws introduced (Figures 4, 5). In other words, although the hardwood removal tretmnts may vary in their relative efectivenes (Provencher et al. 2001a,b), I suggest 11 the various treatments are unlikely to send wildlfe asmblages on disparate jctories that would confound intrpretaion of structural endpoint diferencs betwn treatmnt sits and refenc sites (Figure 4). Therefore, if a treatment contains structural endpoints significantly difrent than thos at referenc sites, I asu this to men recovery was not achieved, rather than raise the potential that the treatmnt isn a transitonal or alternative st that requires a longer period of tim before creting conditions indistinguishabl from referenc conditions. To confirm this aumption would require continuous monitoring as community sructure changed in response to management. Although the study design initialy included fire-suppresed controls, al control sites were burned periodicaly following the conclusion of the initial study. Therefore, the long-trm study lacks a true control, which limts inferential powr. To mke inferencs regarding whetr trements were efective, I asumed that if control sites wre difrent from referencs in 1997- 1998, they would ha been difrent from referencs in 2009-2010 (Figure 5). In addition, within each chapter I atmpt to deonstra that if conditions on treatment sits were diferent from those on refrencs after initial tretment, but were indistinguishabl aftr long-trm prescribed burning was applid, it is due to change at treatment sits and not referencs. I therefore infer that recovery has been achieved if conditions on treaent sits are indistinguisbl from conditions on refrenc sites (Figure 5). DISSERTATION OUTLINE Chapter One present an introduction to the longleaf pine forest community as wel as the relevant concpts and asumptions embedded within t succding chapters of this dirtation. 12 Chapter Two is an atempt to philosophicaly demonstrae that communities are natural entites; thereby establishing that is approprite to use comunitis a targets for restoration. Chaptr Three describes how bird asemblages reponded to the initil mnagement followed by a decade of presrid burning. Individual species hypotsized to respond strongly to restoration were selctd for population level ocupancy modeling. Chaptr Four focuses on the respons of reptils to restoration. In addition to determining how the compositon of reptile asmblages changed in response to restoration, w wre able to link capture rates to specif habit fetures to identify potntial mchanism behind this change. This chaptr identifed A. sexlineatus as species aocited with the refenc condition of longleaf pine forest. W conductd a mark- reapture study of this specis to detrmine how popultions changed after habita restoration nd over time (Chapter Five). LITERATURE CITED Agee, J. K., and C. N. Skinner. 2005. Basic principles of forest fuel reduction treatments. Forest Ecology and Management 211:83-96. Beckage, B., W. J. Plt, and L. J. Gross. 2009. Vegetaion, fire, and fedbacks: a disturbance- mediatd model of savannas. American Nturalist 174:805-818. Bergeron, Y., S. Gauthier, V. Kfka, P. Lfort, and D. Lesiur. 2001. Natural fire fquency for the eastern Canadian boreal forest: consequencs for sustainable forestry. Canadian Journal of Forest Reerch 31:384-391. 13 Block, W. M., A. B. Franklin, J. P. Ward, Jr., J. L. Ganey, and G. C. White. 2001. Design and implentation of monitoring studies to evaluate the succes of ecological restoration on wildlf. Restoration Ecology 9:293-303. Boyer, W. D. 1993. Regenerating longleaf pine with natural seding. Pp. 299?309. In: Hermann, S. M. (ed.), The longleaf pine ecosystm: ecology, restoration and managemnt, Procedings, 18th Tl Timbers Fire Eology Conferenc, Tal Tibers Research, Inc., Talhas, Florida. Bradshaw, A. D. 1984. Ecological principles and land reclamtion practie. Landscpe Planning 11:35?48. Brewer, J. S., and W. J. Plat. 1994. Efects of fire season and soil fertilty on clonal growth in a pyrophilic forb, Pityopsis graminifolia (Astrac). Amrican Journal of Botany 81:805- 814. Brockway, D. G., K. W. Outcalt, and R. N. Wilkins. 1998. Restoring longleaf pine wiregras ecosystems: plant cover, diversity and biomas following l-rat hexazinone applicaton on Florida sndhils. Forest Ecology and Mnagement 103:159-175. Brockway, D. G., and K. W. Outalt. 2000. Restoring longlaf pine wiregras ecosystems: hexazinone applicaton enhances efcts of prescribed fire. Forest Eology and Management 137:121-138. Brockwy, D. G., K. W. Outcalt, D. Tomczak, E. E. Johnson. 2005. Restoration of longleaf pine ecosystems. Gen. Th. Rep. SRS-83. U.S. Department of Agricul- ture, Forest Srvic, Southern Resarch Staion, Ashevile, NC. Bruce, D. 1951. Fire resitnce of longlaf pine sedlings. Journal of Forestry 49:739-740. 14 Clewl, A. F. 1989. Natural history of wiregras (Aristida strica Michx., Gramineae). Ntural Areas Journal 9:223?233. Collins, S. L. 1992. Fire fquency and community heterogeneity in talgras prairie vegetaion. Ecology 73:2001-2006. Connel, J.H. 1978. Diversity in tropical rain forest and coral reefs. Scienc 199:1302-1310. Copeland, T. E., W. Sluis, and H. F. Howe. 2002. Fire season and domina in an Illinois tlgras prairie restoration. Restoration Ecology 10:315-323. Cutter, B. E., and R. P. Guyete. 1994. Fire fquency on an oak-hickory ridgetop in the Misouri Ozarks. American Midland Nturalist 132:393-398. Delcourt, H. R., and P. A. Delcourt. 1997. Pre-Columbian Ntive American use of fire on southern Appalachin landspes. Conservation Biology 11:1010-1014. Frost, C. C. 1993. Four centuris of changing landscpe paterns in the longleaf pine ecosystem. Pp. 17-43 In S. Hrmann (ed.) Procedings Of the Talimbers Fire Ecology Conferenc, No. 18, The longlaf Pine Ecosyste; Ecology, Restoration and Managemnt. Talhase, FL: Tlimbers Resarch Staion. Gil, F. S., and W. J. Plt. 1999. Efects of long-term fire exclusion on tree species compositon and stand structure in an old-growth Pinus paltris (Longlaf pine) forest. Plant Ecology 140:15-26 Greenberg, C.H., D. G. Neary, nd L. D. Harris. 1994. Efect of high-intensity wildfre and silvicultural tretmnts on reptile communities in sand-pine scrub. Conservation Biology 8: 1047-1057. 15 Greenberg, C. H., D. L. Otis, and T. A. Waldrop. 2006. Response of white-footd mice (Promyscus leucopus) to fire and fire surrogate fuel reduction treatmnts in a southern Appalahin hardwood forest. Forest Ecology and Manageent 234:355-362. Hilderbrand, R. H., A. C. Wats, and A. M. Randle. 2005. The myths of restoration ecology. Ecology and Society 10:19. [online] URL: http:/ /w.ecologyandsity.org/vol10/is1/art19/. Hobbs, R. J., and V. A. Cramer. 2008. Restoration ecology: interventionist approaches for restoring and maintning ecosystem function in the fac of rapid environmental chnge. Annual Review of Environmnt and Resources 33:39-61. Lit, A. R., L. Provencher, G. W. Tanner, and R. Franz. 2001. Herpetofaunal responses to restoration treatments of longlef pine sandhils in Florida. Restoration Ecology 9:462- 474. Lotan, J., J. Brown, and L. Neuenschwander. 1985. Role of fire in lodgepole pine forest, pp. 133-152 in D. Baumgartner et al. (eds) Lodgepol pine the specis and its management Symposi Procedings. Washington Stae University, Pullman. Martin, K. L., and L. Kirkmn. 2009. Mnagemnt of ecologicl thresholds to re-establish disturbance-maintned herbaceous wetlands of the south-eastrn USA. Journal of Applid Ecology 46:906-914. Menges, E. S., and D. R. Gordon. 2010. Should mechanial tretments and herbicides be used a fire surrogates to manage Florida?s uplands? A review. Florida Scientst 73:147-174. itchel, R. J., J. K. Hirs, J. J. O?Brien, S. B. Jck, and R. T. Engstrom. 2006. Silviculture that sustains: the nexus betwn silviculture, frequent prescribed fire, and conservation of 16 biodiversity in longleaf pine forest of the southeastern United Staes. Canadian Journal of Forest Reearch 36:2723-2736. Mitsch, W. J., and R.F. Wilson. 1996. Improving the succes of wetland creation and restoration with know-how, time, and self-design. Ecologial Applictons 6:77-83. ulligan, M. K., and L. Kirkman. 2002. Burning influences on wiregras (Aristda beyrichiana) restoration plntings: natural sedling recruitment and survival. Restoration Eology 10:334-339. Mushinsky, H. R. 1985. Fire and the Florida sandhil herpetofaunal community: with special atention to responses of Cnemidophorus sexlineatus. Herpetologica 41:333-342. Ntional Resarch Council (NRC). 1992. Restoration of aquatic eosystems: sienc, technology, and the public. National Academy Pres, Washington, D.C. Outalt, K. W., and D. G. Brockwy. 2010. Structure and compositon changes following restoration treatments of longleaf pine forest on the Gulf Coastal Pin of Alabam. Forest Ecology nd Managemnt 259:1615-1623. Palmer, M. A., R. F. Ambrose, and N. L. Poff. 1997. Ecological theory and community restoration ecology. Restoration Ecology 5:291-300. Plat, W. J., G. W. Evans, and S. L. Rathbun. 1988. The population dynamics of a long-lived conifer (Pinus palustris). American Nturalist 131:491-525. Provencher, L., B. J. Herring, D. R. Gordon, H. L. Rodgers, K. E. M. Galey, G. W. Tanner, J. L. Hardesty, and L. A. Brennan. 2001a. Efects of hardwood reduction tchniques on Longlef Pine sandhil vegetaion in northwst Florida. Restoration Eology 9:13-27. Provencher, L., B. J. Herring, D. R. Gordon, H. L. Rodgers, G. W. Tanner, J. L. Hardesty, L. A. Brennan, and A. R. Lit. 2001b. Longleaf pine and oak responses to hardwood reduction 17 techniques in fire-suppresed sandhils in northwest Florida. Forest Ecology and Management 148:63-77. Provencher, L., N. M. Gobris, L. A. Brennan, D. R. Gordon, and J. L. Hardesty. 2002. Breeding bird response to midstory hardwood reduction in Florida sandhil ongleaf pine forest. Journal of Wildlf Management 66:641-661. Provencher, L., A. R. Lit, and D. R. Gordon. 2003. Predictors of species rihnes in northwest Florida longleaf pine sandhils. Conservation Biology 17:1660:1671. Romme, W. H. 1982. Fire and lndscape diversity in subalpine forest of Yelowstone National Park. Ecological Monographs 52:199-221. Rorig, M. L., and S. A. Ferguson. 1999. Characteristic of lightning and wildland fire ignition in the Pacif Northwst. Journal of Applid Mtorology 38:1565-1575. Russel, K. R., D. H. Van Ler, and D. C. Guynn, Jr. 1999. Prescribed fire efects on herpetofauna: reviw and management iplicatons. Wildlfe Socity Bulltin 27:374- 384. Schwilk, D.W., J. E. Kely, E. E. Knapp, J. McIver, J. D. Bailey, C. J. Fetig, C. E. Fiedlr, R. J. Harrod, J. J. Moghaddas, K. W. Outalt, C. N. Skinner, S. L. Stphens, T. A. Waldrop, D. A. Yussy, and A. Youngblood. 2009. The national fire and fire surrogate sudy: efcts of fuel reduction methods on forest vegetaion structure and fuels. Ecological Appliatons 19:285?304. Scienc Aict International Corporation. 2009. Eglin Integratd Natural Resources Management Plan. Prepad for the Department of the Air Force, Eglin Air Forc Bas, Florida. 18 Steen, D. A., A. E. Ral-McGe, S. M. Hermann, J. A. Stiles, S. H. Stiles, and C. Guyer. 2010. Efects of forest managemnt on aphibins and reptils: generalist species obscure trends among native forest aociates. Open Environmental Sciencs 4:24-30. Suding, K. N., K. L. Gross, and G. R. Housmn. 2004. Altrnative sts and positve fedbacks in restoration ecology. TRENDS in Ecology and Evolution 19:46-53. Tucker, J. W., Jr., W. D. Robinson, and J. B. Grand. 2004. Influence of fire on Bachman?s sparrow, an endemic North American songbird. Journal of Wildlfe Management 68:1114-1123. Van Ler, D. H., W. D. Carroll, P. R. Kapeluck, and R. Johnson. 2005. History and restoration of the longleaf pine-grasland ecosystem: iplicatons for species at risk. Forest Ecology and Mnagemnt 211:150-165. Vrner, J. M., III, D. R. Gordon, F. E. Putz, and J. K. Hiers. 2005. Restoring fire to long- unburned Pinus palustris ecosystems: novel fire efcts and consequences for l- d ecosystems. Retoration Ecology 13:536-544. Vebln, T. T., T. Kitzberger, and J. Donnegan. 2000. Climatic and human influences on fire regimes in ponderosa pine forest in the Colorado Front Range. Ecological Applicatons 10:1178-1195. Waldrop, T. A., D. L. White, and S. M. Jones. 1992. Fire regimes for pine-grasland communities in the southeastern United Staes. Forest Ecology and Mnagement 47:1095-1210. Ware S., C. Frost, and P. Doerr. 1993. Southern mixed hardwood forest: the former longleaf pine forest. In W. H. Martin, S. G. Boyce, and A. C. Echternacht [eds.], Biodiversity of 19 the southeastern United Staes: lowland terrestrial communities, 447?493. John Wiley and Sons, Nw York, Nw York, USA. White, P. S., and J. L. Walker. 1997. Approximating nature?s variation: selcting and using refnce informtion in restoration ecology. Restoration Ecology 5:338-349. Yager, L. Y., C. D. Hise, D. M. Epperson, M. G. Hinderliter. 2007. Gopher tortoise reponse to habita mnagemnt by prescribed burning. Journal of Wildlfe Managemnt 71:428-434. 20 Figure 1. Restoration scenario wherein reintroduction of a natural disturbance regime restore a degrad site to a previous condition along a linear trajectory. Figure is taken directly from Suding et al. (2004; Figure 2). Figure 2. Conceptualizaton of diferent restoration scenarios potentialy leading to ecological recovery. Axis labels in Figures 2-5 are adapted from Bradshaw (1984). Figure 3. Conceptualizaton of restoration scnario specif to this diertation, prior to hardwood removal treatmnts. Figure 4. Conceptualizaton of restoration specif to this diertation. Green represent fire- suppresed controls. Sites that experiencd hardwood removal are not considered recovered. Figure 5. Conceptualizaton of restoration specif to this diertation. After al sites reived restoration, they are presumed to have been restored to the refrenc condition. The study lacks a true control for 2009-2010. The hollow circl represent our expectaion regarding t positon of a control, should one exist. 21 Figure 1. 22 Figure 2. 23 Figure 3. 24 Figure 4. 25 Figure 5. 26 Chapter 2 Conceptualizng Communitis a Individuals with Implicatons for Conservation and Restoration Ecology Abstract. Recent work has uggested that conservation eforts such as retoration ecology and invasive speis eradication are largely value-driven pursuits, a opposed to sienc-driven. Additionaly, changes to global cimte are forcing ecologist to consider if and how collections of species wil migrate, and whetr or not we should be asitng such movements. Within ts haptr, I propos a philosophical framwork for addresing these iues by identifying an ecological community as a natural entity (i.e., an individual). Esntial to making this oncptualizton (termd the Evolutionary Community Concept, ECC) pplied isdentifcation of a unique collction of species that interact and have co-volved in a given geographi are. I first esablish that communitis should be considered entites by examining the in light of the various qualites other entites, such as taxonomic speis and areas of endeism, have been shown to poss. I tn map out the implitons of ECC for a number of global consrvation isue. Specifaly, this fraework alows us to establish a biological and scienc-driven rationale for restoring ecosystms to referenc conditions and removal of exoti species, and the ECC has implicatons for how we viw shifts in species ablages due to climat change. In addition, conceptualizng a community as an individual advances our understnding of various ecological cpts, such as reilenc. Key words: individual, class, community, exotic, nvasive species, rtoration 27 INTRODUCTION Ecological restoration is the, ??proces of repairing damage cused by humans to the diversity and dynamis of indigenous ecosystms? (Jackson et al. 1995). Although the field of restoration ecology is baed on scientfi principles, retoration goals may be influencd, for example, by ethis, morals, or astheis (Higgs 1997). Without estblishing a given indigenous cosyst as natural ecological entity (i.e., an individual), there is no basi for demonstraing whetr restoration eforts are driven by goals that reflect t evolutionary history of the system being repaired. In other words, if the target condition of a given restoration ecology efort is an artifcal construct (i.e., a clas), mnaging a damaged ecosystem so that i moves towards this ondition my result in reconstructed syste with misng parts or a failed reconstruction altogether. Restoration eforts often focus on species amblages within a given area, and these asemblages are often considered communitis. However, comunities ha been suggestd to have no, ??intrinsic evolutionary or ecological purpos?? and therefore its not valid to, ??invoke any eologial (or evolutionary) rational to establish particular restoration goals? (Davis and Slobodkin 2004). The logical consequence of this philosophy is to conclude that tempting to restore communities i a value-driven pursuit based on our judgments and indepent of natural laws (Choi 2007). Some ha contstd this point, suggesting restoration ecology is not soley value-driven (Wintrhalder et al. 2004); however, to convincingly demonstrae that the goals of restoration ecology are based on natural laws requires etablishing tha restoration targets are natural entites. 28 Herein, I argue that an ecological community can be conceptualized as a unique asemblage of species t ocurs in a given geographic are and is connetd by interspecif nd abiotic intrations. Just as species consist of multiple parts (i.e., organism), a community may be made up of multipl parts, (e.g., forest paches, iolated wtlands). Given this onceptualizton, which I term the Evolutionary Community Concpt (ECC) we my establish boundaris around communitis and describe how they fulfil the critea of natural ecologicl enties that exist indepent of anthropogeni naming conventions. This exercise i analogous to the wl-tread discusion regarding whetr species are individuals or clas (below) but the topic has not been explored in-depth in relation to eological communities. Designating a community as a natural entity is a philosophical exercis and operational dificultis are omnipresent when applying philosophical notions to biological enties (Frost and Hils 1990); however, it ismportant not to confuse community conceptualizton with community delineation (as for species, de Queiroz 2007). Based on the ECC, it is dificult to determine the spatil boundaris betwn communities. In addition, one consequence of concptualizng a species an individual is tha s organism do not qualify as a specis (e.g., an individual incapabl of breeding cannot be a meber of any species under the Biological Species Concept; Baum 1998). Simlrly, som species amblages do not qualify as a community a crtain scles. In any case, it is unlikely that any one conceptualizaton wil provide a cpletly stisfying answr on how to best conceptualize a community, as evidenced by the plethora of concepts used to define a species. Howver, a philosophical discusion of the topic may help ensure our clasifcation systm is an acurate representtion of natural proces and tha reserch questions and conservation and restoration goals are properly formulatd. As mentioned by Ghiselin (2002) in refrenc to species, appreciatng that they are individuals, 29 ??can help us to clarify the roles of history on the one hand and the laws of nature on the other in evolutionary biology.? The ECC can be applied to extant semblages. For example, longleaf pine (Pinus palustris) forest once dominatd the coastl plin of the southeastrn Unitd Stes. Typified by a reltively open canopy, sandy soils, and frequent fire, tre a mny specis that evolved within ts forest system. The environmental conditions of the region alowed for a unique species ablage to persist. Comparisons of species lit from a random sampling of omparabl-sized ares across the planet wil fnd intrating species such as longleaf pine trees, red-cockaded woodpekers (Pioides borealis) gopher tortois (Gopherus polyphemus) ong others, ocurring in sympatry more oftn than expected by chance. Population fluctuaions in one species are likely to influence the otr species in the asmblage. Under the ECC, this asmblage cn be recognizd as a community (i.e., t longlef pine community). WHAT IS A COMUNITY? Although the concept of a community is frequently invoked, there has been litle examinaton into how to difrentiate aong communities or identify tm as ecological entites. Consquently, some have suggested they are of litl iportance (Ricklefs 2008). Perhaps this argument can be atributd to the complexity of these systms, whih tnds to preclude developmnt of general laws (Laton 1999). Howver, despite their complexity, the relevanc of the concept of communities to current ecological and consrvation-orintd problems ss to necsita their investigaon and inclusion in the eological siencs (Siberloff 2004). To enhance t context of community-orientd studies, it sems obligatory to establish whetr 30 communities arerbitrary designations (i.e., clase) or entits with diagnostic properties (i.e., individuals). Progres in scienc may be hampered when multiple definitons are alowed to proliferat (McCoy and Shrader-Frehete 1992, Mikkelson 1997). Howver, there a numrous definitons for eological communitis. For example, definitons include those that sres dominant species (e.g., Riklefs 1990), interactions (e.g., Wilbur 1972, Holt 1977), or staistcl properties (Fild et al. 1982, Clarke 1993). Some resarchers have presentd more refined definitons (e.g., Looijn and Andel 1999) to enhanc precison for addresing ecological questi, while others have argued that a very general definiton for what constiutes a community wil suffic for most studies (Futh et al. 1996, McGil 2010). However, common definitons of communities struggle to capture their unique nature. Dominant specis or interactions, for two exapls, ay be considered components of an ecological community, and multiple communities my even share such cnts. Therefore, current definitons suggest comunitis are clas. The inconsistencis among current community definitons may be tributed to an ecological dihotomy described by Los (1996), in discusing proxite and ultimat approahes to cmunity eology. Proximate approahes concern themslve with, "the proces occurring within comunities and the efect thos procs ha on community structure" (Los 1996). However, the species prent in a given area a not solely influenced by current forces and may be a function of the evolution of a particulr lineage in a given area (Helmus et al. 2007, Cavender-Bares et al. 2009). Ultimate approahes to community ecology, which aknowledge that evolutionary lineages are likely to be spatialy constrained, come closr to helping us concptualize communities something more than just a colletion of arbitrary 31 species. Such an approach is defined by Los as, "involv[ing] study of why communities have partiular organization and why diferencs exit bewen communities [ephasi ine]." WHAT IS AN INDIVIDUAL? Conceptualizng units in biology as individuals or clase has been discused in depth regarding specis (e.g., Ghiselin 1987), and more recently regarding areas of endemis (Crother and Murray 2011). Therefore, I do not delve deply into a review of these concepts. In short, individuals are ecological entites that exist because of their evolutionary history whereas clse are actualy groups of entits. Clas may not represnt natural kinds; therefore tir us my be limted when atempting to understnd evolutionary relationships (Ghiselin 2002). Individuals, in this contxt, are entites with a definite loction in spac and exist for a finite period of time. Individuals can be singl things (e.g., Luke Skywalker) or can be composd of multipl parts (e.g., Auburn University faculty or Panthera tigrs). In contrast, clse are abstrac constructs that cannot be atributed to a discrete tim or place, and they have mmbers tha aresigned to the cls on the basi of defining propertis. For example, ?unirsity faculty? is a clas to which any number of people may belong if they mt the criton of being demic stf at university, and the concpt itslf i not rooted to any particular plce or time. Speies in the clas sens refers to al groups of organism tha mt some crit (e.g., Biological Specis Concpt) to warrant designation as such, and it would be reasonable to talk about a kind of sies. Species in the individual sens is a particular thing (e.g., Panthera tigrs), and it would not make sns to discus a kind of Panthera tigrs. A noted by Crotr and Murray (2011) in reltion to areas of endemis, designating communitis a natural entites 32 requis acepting species a individuals; the reader is referred elsewhere to make this cae (Hennig 1966, Ghislin 1974, 1981, 1987, Hull 1976, Wily 1980, Bernier 1984, Holsinger 1984, Kitcher 1984, Mishler and Brandon 1987, Ereshefky 1992, Frost and Kluge 1994, Baum 1998, de Queiroz, 1999, Colman and Wiley 2001, Mayden 2002, Brogard 2004, Rieppel 2007, Reydon 2009). For something to be an individual, several critea must be met: 1) it must have temporal and spatial boundaris, 2) definitons must be craftd ostnsively, 3) there mt be cohesines in response to change, and 4) there can be no instances of this entity; it must represent a unique entity (e.g., Ghiselin 1974, Hull 1976, Frost and Hils 1994, Crother and Murray 2011). Therefore, if a community is an individual, it must be discovered through some proces of identifying its boundary, inferring its origin, and determining its ultimate deis. A mntioned earlir, I suggest that wihin a given geographical are, there is likely a specis aemblage comprised of species tt are unlikely to ocur togetr elsewre. Esentialy, this group of speis can be considered as an area of endemis (Crother and Murray 2011). I believe the rationale usd to intify areas of eis a individuals i relevant here but it ismportant to add that communities difer becuse the structure and compositon of comunities are influenced by intrspecif, as wl asbiotic, interactions (Fontaine et al. 2011). I argue that this group of coevolved and interacting speies, whih is unique to a given area, is a community, and under this conceptualizaton, w can consider communities as individuals. 33 A FRAMEWORK FR IDENTIFYING COMUNITIES A INDIVIDUALS BOUNDARIES Eldredge (1985 p. 162) states that, ?some ecologist?take strong isue with the suggestion that communities can be construed as individuals. The problem ss to come from the apparent lack of definitve boundednes to such entites?. It is dificult to delineat discrete boundary surrounding an asemblage in a finite spac without identifying arbitrary thresholds for particular variables, such as the density of a given speies or interaction levels betwen to or more speies. This i analogous to a population of one scis with varying genotypes, i.e., to what degre do two groups have to difer in their geneti make-up before they are considered separat species? To ha any applicaton, thresholds deliting communities hould have biologial and evolutionary relevanc. Although some may argue for a specif threshold beyond which individuals are considered sparate entits (e.g., Highton 1989), these tsds can be onsidered arbitrary (Frost and Hils 1990). It is most consistent with som concptualiztons (e.g., the Phylogenetic Speies Concept; Cracraft 1983, 1987) to suggest any evidence of a unique evolutionary liage is sufficnt to identify something as an individual (e.g., Young and Crother 2001). Biogeographical paterns in species rihnes and asemblage compositon may help demonstrae that spatil boundaris exit around a unique asbl of species. For example, an are?s biological uniquenes may be inferred after using null models to demonstrae that observed paterns difer from random expectaions (e.g., theid-domain efect, Colwl and Ls 2002). It is necsary to have some a prior designation of the spatil boundaries of areas so 34 tha paterns of species rihnes may be compared for these analyses; areas my be defined by a grid systm (e.g., Hawkins and Diniz-Filho 2002), politcal boundaris (e.g., Means and Siberloff 1987), or elevation (e.g., McCain 2004). pecies rihnes alone tels us litle about asemblage compositon, which may alow us to distinguish betwen areas with simlar specis rihnes paterns. If a group of speies occurs in sympatry more frequently than expected (e.g., as defined by null models, Goteli 2000), this suggest the area is subject to eologial or evolutionary forces reulting in a particular species asemblage. If ts sm speies occur together more frequently in a given geographil area than they do in other geographi areas, te areas my be considered discrete. In addition to co- occurrence analyses, parsimony analysis of endemicty (Morrone 1994) is a mthod of identifying areas with unique species compositons. Defining boundaries prior to analysis doe not alow us to identify the scale of forces influencing paterns of specis compositon. Therefore, demonstraing that patrns of specis richnes or compositon are not random does not inform us regarding the spatial extent of the are influenced by the same biogeographical proces. However, if given ares poss unique paterns of specis rihnes or diferent speis compositons, there must be a spatial boundary beyond whih these fatures are no longer unique. These boundaries exit, but we are limted in our ability to delinet them. This i not necesarily a waknes of the ECC outlined here; boundaries may be fuzzy wn charatrizing something as an individual (e.g., Ghiselin 2002, Crother and Murray 2011). In fact, i is likely folly to asume precisely delineatd boundaries acurately represent a natural entity; alowing a certain degre of boundary fuzzis when delineting boundaris isnot a concesion to our limtions at identifying their extent but ratr a more acurate characterization of the entity in question (Baum 1998). 35 Ecotones are generaly considered, ?transitonal areas betwen adjacent eological systems? (Riser 1995) and may posse atributes of two or more systms. Som res that could be defined as cotones, such as the intrtidal zone of the Pacif Northwest, United Staes, are probably beter considered as an independent community under the ECC, as they contin a charateristic st of species that are highly adapted to the syste. However, ecotones should be oncptualizd at multipl sals (Gosz 1993), therefore it may also make snse to consider some ecotones a the fuzzy boundaries betwen communities (e.g., riparin areas), and others as simply a function of species-pecif atributs (e.g., microhabita transiton zones betwen patches colonized by alopathi plants). Because eotones could either be considered ommunitis, the fuzzy boundary betwen communitis at larger scale, or as reulting from a proces occurring within a community, y concept subsumes that of the eotone. At some point in tie, due to shifting patrns in specis compositon (e.g., due to extirpation or stochasticy), co-occurrence patrns may ceas to be diferent from nearby geographic areas. Speies diagnosti to a community would at that point stop interacting, or the nature of the interactions could shift outside the bounds of the distribution by whih they were previously charatrized. Conversely, at some point in time, random species amblages in a given area cn becom non-random and difrent from other areas. Sis within such ares would likely begin interacting and shaping the evolutionary trajectoris of one another. So, while it is operationaly difiult to pinpoint precise beginnings or ends, it is theoreticaly plausibl that one could asign temporal boundaries to a unique group of species within a spatil are. Changes in species amblages lading to such boundaris may be atributed to several causes. Over ecologial ti-scle, the habits in a given areay change due to nthropogeni climte change or sucsion (Gleason 1926). Direct anthropogenic disturbances 36 may also influence a species amblage. For example, due to habita loss and land conversion, the longleaf pine forest, and asocited species, of the southestern United Stes have been reducd to a fraction of their historic extnt. Over geologic tim, cliat patrns or geomorphology wil become unsuitable for species within a given are. Species wil ether adapt or bece extirpated. Biogeography, and its influenc on evolutionary lineages (Wiley 1988) must also be considered when conceptualizng temporal community boundaries. Individual phylogenies of species are influencd by vicarinc and dispersal events, and thes indivil nis may ultimatly ile community asembly (Webb et al. 2002; Figure 1). It is also the cse that such biogeographic events may act directly on the incipient community, rather than being propagated through speies. For example, the creation of a river, or sparation of teconi plates could split a community, a divergenc of ecologicl significance. Simlarly, the removal of geographic barriers wil aow two communities to converge. It follows from the ECC that a given area wil poss multiple communities over geological time as changing climtes altr habit suitbility for a given suite of specis. Since speies and intractions wil be replacd over time, succesion wil aso reult in multipl communitis. Thus, a community cn transiton into another community (i.e., a branching event is not required to result in new comunities). A an analogy, if a species changes considerably over geologic time its logical to consider the oldest organism a one sis and the youngest a another (Sipson 1961) and aknowledge that t initial species wnt extinct a some point in time (Hull 1976). Although we argue that communities may eventualy transiton into diferent individuals (Figure 2), the scal of relevanc to most ecologicl studies suggest communitis can change to some degree through tim yet reain their identity. Alowing for a certain degree of c is not 37 necesarily a problem. For exaple, it is generaly aceptd that each organism i an individual. Over the cours of an organis?s lif, it may undergo relatively drasti changes, such as in the case of a tadpole developing into a frog or a cterpilr into a butterfly. Although the organism hanges, it esnc, perhaps best conceived as it genotype, remains the same (Hull 1976). Each organism cae into existenc som short yet fuzzy aount of tie before it ws born, or hatched, and ech organism case to exist some fy yet short amount of time aftr it dies, a it is broken down and the parts aresimlatd into other organis. Silarly, unique asmblages of co-evolved and interacting species my change, for example as when popultions of competiors, or ofpredator and prey, fluctuae in abundanc. Fluctuaions in the frequency of natural disturbance my also occur. Although it is most philosophicaly straightforward to consider a community a new individual as soon as it changes atl, if organism and species an hange and stay the sae ivil, why cannot communities? So long as the unique oevolved asemblge of specis and their asocited intractions are extant and functional, a community reins the sae individual. AN OSTEIV DEFINITION As noted above, certain species wil co-occur together within a given area more often than expectd by chanc and more oftn than they c-occur tther elsewhere. Thes specis are often considered specialst of a given habita with limted geographic distributions. Indiator specis analysis (Dufr?ne and Legendre 1997) my be a usful mans of identifying harateristic speies of a given area. We can point to these species and therefore diagnose communitis bad on their presenc. For example, at t sal of a forest sand, the presnc of 38 longleaf pine, gopher tortoise, red-cockaded woodpeckers, and wiregras in a given area is sufficnt to identify that this area is unique tohe Coastl Pin of the Southeastern United Staes. In sum, we can define communities by identifying characteristic speis, i.e., identify communities ostnsively. W must revisit the isue of scale. If our area of interest wa the planet Earth and we wished to compare t species amblge of plnet Erth to neighboring plts, hen every species i an indicator of Earth. As the focal scle decrease, widespread species wil begin to stop playing a role in what mkes a given are uniqu (Figure 2). For exampl, the gopher tortoise i an indicator of planet Earth, the continent of North Aerica, the Coastal Pin of southeastrn United Stes, and the longleaf pi forest, but not the pitr plnt bogs that may occur within longlaf pine forest. Beyond the scale of the longleaf pine forest, the gopher tortoise i not useful for diferentiatng betwen ares. Thus, again, the relevant scale is reliant on the decison of the resarcher. This mater of scle may shed some light on controversial subjects in ecology. Neutral theory (Hubbel 2001) suggest cmunitis may be comprised of asmblages of organism arising from forces indepent of species intractions. Simlarly, Gleson (1927) argued that the structure of a particular asemblage is due largely to pioneering specis; thee species become established due to their dispersl abilites, rather than becaus they belonged to any disret ntity. Tse ideas, t least on smal tporal and spatil sles, appear to run counter to some of community ecology?s most basic underpinnings (Chase and Libold 2003). Examining an aseblage atn inappropriate sl may encourage mislading interpretaion. Continuing our longlf pine forest exampl, quantifying species compositon within forest snds of a few hectares ech ay reveal that the species within eah stnd appear random. Howevr, at a larger 39 scale, the species characteristic of longleaf pine forest are diferent than those tha appear in a ponderosa pine, Pinus ponderosa, forest, or in the Sonoran desrt. At this scal, species asemblges are not random, they are distinc. One may argue that t unique species amblage of a given community could be relocated to another location and thereby creat another part of that cty. Silarly, one my suggest a community cn be restored following habita conversion tha resulted in a change to a new cty. However, it ismportant to consider the evolutionary origin of the speies in a given asemblage as wl the interactions betwen t species. Once an evolutionary lineage dirges into sparate lineages, t parts of the difrent lineages can never again be the sme. They have experincd diferent evolutionary histories. A an analogy, an organis cannot di and tn arise a the sam organism again. A specis cannot become extinct and then be resurrectd through an independent evolutionary linege (Hull 1976). You cannot take t components of a community (i.e., the unique asembla) into an environment with simlar abioti conditions and recreate that comunity without breking spatial and tporal boundaries; therefo, the reloctd asemblageould be an indepent and new community. COMUNITIES ARE COHESIV Communities are comprised of species. Thus, the isue is how to conceptualize a group of species reponding to change as cohesive unit. This topic was recently discusd in refrence to ares of endemis (Crother and Murray 2011). The unique speis amblage of a given area (se An Ostensiv Definiton) likely engages in important ierscif interactions that sustin 40 the identity of a particular community. These intractions may promote co-volution and community sructure and dynaics (Johnson and Stinchcombe 2007). For exaple, within the community asociated with pine forest in the Coastal Pin of the southeastrn Unitd Staes, longleaf pine tres are conduits for lightning strikes that ignite a highly flmble understory (Plt e al. 1988). The resulting ground fires a necesary for reproduction of other species (e.g., wiregras, Mulligan and Kirkman 2002) and mintn habitat suitable for others (e.g., gopher tortoise, Yager et al. 2007). Gopher tortoise, through the procs of burrow creation, provide structure important to other specis (e.g., Jackson and Milstrey 1989, Kinlw and Grasmueck, in pres). The estblishment of one or more of these species facilted the persistenc of additional species. In addition, a change, such as gradual limt change that alrs the abiotic conditions in a given area, wil lkely reduce habita suitability for one or more speies. Due to the influence of interspecif intractions, mny species within the unique asmblage are likely to respond; this reponse may be manifestd in hanges in abundance. Thus, species within a geographic area my respond cohesively to change and therefore fulfil this criteon to be considered an individual. THERE A NO INSTANCES If we recognize a community as an individual, for example, the longleaf pine forest community (whih consist of multiple parts), there cannot be anotr lf pi fst t. I have discused how communities my be spatialy and temporaly bounded, based on paterns of co-ocurrenc of characteristic speis. I have identifd how these communitis may be defined by the prese of a unique asemblage of species and how ts spcies repond 41 cohesively to change. It is dificult to conceive of how there may be multiple instances of a ommunity tt fulfil these crita. If a community is bounded by specif limts (.g., climatic, physiological) and thos limts help charaterize a comunity, along with a suite of specilst species that provides the ostensive definiton, another indepent community cannot share the sam lits and same specis compositon. This delineation is no diferent from sying tt indepent evolutionary trajetoris delineate betwen species. Certain stuaions are problmatic for this concptualizaton, such as new islands. A new island is subject to the cliatic influences of that particulr region. Abiotic fators interact wih erly colonizrs to facilte the persist of these pioneering species. At this point communities are best charactrizd in the proximat sns (Los 1996), since they are a function of a species? ability to colonize an are, rather than a function of t myriad of interactions that constiute community ecology. It could be unlikely, at the earliest sges, to have speies preent that had developed iportant ierspecif relationships, unls they emigratd from habitas where ty previously co-ocurred. Very quickly, however, biotic factors wil play importnt roles in influencing which specis persist. A this point, the speies ompositon and abundanc in the area a function of its unique adaptive and evolutionary past and they can qualify as a community under the ECC. Since t species that colonize islands originated elswhere, tre a unlikely to be endemic speis on very young islands. Nverthels, there is likely to be a unique asemblage comprised of scies that are good disperse, and a unique suite of species may be sufficnt to alow us to consider an asemblage an individual (Crother and Murray 2011). However, if this unique suite of specis appered on multiple new islands due to their dispersal cpabilits and 42 perhaps not initialy influenced by interspecif interactions, it does not qualify as a single community under my concptualizaton. APLICATIONS OF THE VOLUTIONARY COMUNITY CONCEPT EXOTIC SPECIES Perhaps the most relevant applicaton of the ECC concerns exotic speies (i.e., a species living outside its native range, Hunter 1996). Some specis becom invasive by influencing the unique species amblage of a given area (e.g., Frits and Rodda 1998); isi species managent is oftn driven by a desire to rid a particular are of species demed daaging to the native species or communities (e.g., brown tree snakes, Boiga irgularis, in Gua) but this type of managent has ben criticzd as potntialy xenophobic or based primarily on ethics (e.g., Brown and Sax 2005). This critsm i likely encouraged by the fact that identifation of communities ha heretofore been subjective (Siberloff et al. 2003) and did not sufficently diferentiat betwen specis considered nati versus those that are considered introducd (e.g., Fauth et al. 1996). Howver, if communities are spatialy and tmporaly bounded and consist of unique asemblage of species and their asocited intractions, then exotic speies threaten their continuity. Because humans influence the planet on a scale rger than any other single species (Vitousek et al. 1997), it is reaonabl to ctegoriz humn activiy as ditinc from other biotic procs. Species physicly moved by humans or whose movements were failted through infrastructure, such as imported decorative plnts, invertbratsithin balst wtr, or escaped 43 pets, are not components of native communities. The proximate cuse of invasion by many scies i clearly direct human intervention and tir presnc in area is not due to the ommunity?s unique evolutionary lineage. Since exotic speies my result in the functional extirpation of a native species, a wel as the functional extincton of interactions betwen native specis (e.g., Riciardi and Simberloff 2009), ty may result in the demis of the original ommunity. Consequently, under the ECC, eforts to eradicate xotic speies are justifed while assisted migration eforts (McLahln et al. 2009) are not. On the otr hand, humn activies may drasticaly alter native asemblages, for example by reducing densites of dominant predators (e.g., Friedlnder and DeMartini 2002), and also by changing abiotic paramtrs, such as in the case of global cimat change (e.g., Walther et al. 2002). In these circumstances, a species my coloniz an are becuse it represent suitble habita wn it previously did not. Exampls include coyotes colonizing the eastrn coast of the United Stes to fil the nic of extirpated wolf populations or birds shifting ranges in concordanc with climate changes (Tingly et al. 2009). In these cas, the species have not invaded an area becus humans physicaly aided their dispersl, ratr ty are using their own dispersal cpabiltes to respond to changes in t habita. In these circumstances, eradication cmpaigns are not an efctive management tool as the are in question has becom an extensi of their native range. Many exotic speies ither do not become established or establish populations without noticebly influencing native species (Wilamson et al. 1986, Mnchester and Bullock 2000), som have argued that the presnc of exotic speies in a given area my even have conservation benefits (Schlapefer et al. 2011). For exampl, exotic speiesy fil the role of extinct organism. In this c, although co-evolution was not a factor in an exotic speis' role within a 44 community, its role isndistinguishable from those that arise from co-evolutionary proces. If we regard the intrspecif interactions a specis partakes in a the defining component of its identity, we may reogniz thes exoti sies a components of communities. However, if we regard identity as a function of unique evolutionary trajectoris and spatio-poral boundaris, as outlined in this eay, then exotic speis can never be components of communities. This dichotomy has importnt ipliatons for the debate regarding whetr restoration of ecological proces may be more iportnt than t specis usd to restore tm (e.g., Pleistocne re- wilding of North America; Donlan 2005). CLIMATE CHNGE I lay out an argument here that a subset of species within area comprise a unique asemblge, are strongly intracting, and are consistntly present within a given cmunity ype nd not elsewhere. Once w can identify a community as an individual, it is thee species that help us difrentiat betwn communities. The ECC has imediat implicatons for how to view changing global dynaics. For exapl, cliate change is expeted to led to range shifts among individual species (Parmesn and Yohe 2003), which may in turn lad to community disebly (Thuilr 2004). If one viws communities simply as the groups of species reiding within a given area, the efects of cliate change may be mitgaed by complex landspes, hich wil lkely continue to harbor a diversity of species (Anderson and Frre 2010). However, if we reognize the importance and unique nature of intrspecif interactions, we may be ls optimstic regarding how comunities wil fare in response to anthropogeni-driven climte 45 change, as interacting species may have varying abilites to adapt and persist (e.g., Parmesn 2006). REFENCE CODITIONS Many restoration eforts are gauged by comparison to referenc communities. However, current definitons for communities characterizd by dominant specis, interactions, or staistcl properties are often inconsistnt with the goals of restoration eology. In the United Stes, for exampl, restoration ecology is often primarily concerned with returning degraded communitis into a condition consistent with the species compositon and abundance that may be expected prior to setlment of this area by Europeans. It is thought that thes amblges, whih wil alwys include som degree of natural varition (White and Walker 1997), likely best represent the aestral condition. The ECC, which posits the communities are individuals due to unique specis amblages, evolutionary histories, and interspecif intractions, offers a scientfic rationale for this approach. Disruption of natural disturbance regimes within a given community may encourage the proliferati a species preent at low lvels. Although these species are not exotic, they may disrupt the continuity of a community. For example, fire-suppresion of longleaf pine forest alows oak trees to increase in abundance, resulting in a change in the habit structure and a reduction in habita quality for other specis (Mitchel et al. 2006). This change may eventualy result in a transiton to a new community. Trefore, eforts to restore natural disturbance regimes and manage species to lvels that best typify a community are warranted, as they wil aintn a community that exist due to natural proces. 46 Restoration ecologist often strive to replicate the species compositon and abundance derived from a unique evolutionary history and us dominant sis, interactions, or staistcl properties a secondary metrics to evaluate succes. For example, much has ben discused regarding the relative mrits of focusing on one speis for consrvation eforts vers a suit of species (e.g., Lmbeck 1997, 2002; Lindenmayer et al. 2002) or rather, perhaps most omprehensive, on a comunity level (Simberloff 2004). However, t ultimate goal is always the same, i.e., to restore, or at last consrve in some form, the group of specis in a given are tha best represent whats found in the area due to evolutionary proces. I have argued here tt the evolutionary origin of a community is an important component of itsdentity because this origin faciltes intrspecif interactions betwen species unlikely to co-occur together elswhere. In addition, individuals have tmporal boundaris. Following this logi, once a given community has transitoned into another comunity (for example, through habita degradation and/or disruption of normal disturbance regimes), it is philosophicaly impossible to change this community such that i becoms a part of the original community (Figure 1). Operationaly though, it is posible to create a cmunity that is functionaly and structuraly identical to the target community. COMUNITY RESILIENCE Resilenc refers tohe time required for a system to return to its equilibrium following disturba (Pim 1984). Unls w alow a comunity to experienc some change while remaining the sae individual, the concept of ecological resil is dificult to appreciate. Specifly, if w define communitis bad solely on their structure and/or function, as i 47 acomplished by most current definitons, then a community often cannot be resilent, because once its structure and/or function changes it i no longer a mber of the sam class. For example, if we define a longleaf pine community as any P. palustris dominated- forest hat is burned at lst once very two yers (by designating definable propertis, w establish the longlef pine community as clas), then t forest no longer is a longleaf pine community onc two years has paed without a fire. Although a forest that has ben fire- suppresed for a fe yers il lkely appear somewhat diferent tn a forest t was burned more frequently, this i due primarily to fluctuaions in the densites of species thatere always present; I suggest i esenc reins the same. Even after a P. palustris dominatd forest i fire- suppresd (i.e., disturbed) for decades, retoration of fire alone is sufficent to alter the structural components of the forest (e.g., vegetion, bird and reptile populations) such that ty are indistinguishable from fst that have been burned regulrly (Outclt and Brockway 2010, Chapters 3, 4, 5). Over this time priod, it makesore sens to conceive of a longlef pine community as an individual changing over time and in respons to disturbanc than it does to onceive of a forest switching clase depending on the structural and functional components of a gin dfiniton. However, onc the unique scies amblage begins to change through extirpation and colonization, t original community has ced to exist and can never return to an equilibrium. CONCLUSION I have made a case that communities may fulfil the critea necesary to be considered individuals. Furthermore, I have described how communitis my fulfil thes critea due to 48 their unique evolutionary history. In doing so, I have built upon the work of Los (1996), who identifed a dichotomy in how communities are conceptualized and my conceptualizaton complmnts work emphasizng the iportanc of historical influences in current community structure (e.g., Los 1996, Ricklefs 2008, Cavender-Bares et al. 2009). If the cponents of a community result from historial forcs, it i likely most appropriate to consider tse forcs when conceptualizng what a community is. Retoration ecology goals nd eological questions should be focusd on the unique species amblage of a given area as wel as the asocited interactions. I argue that tse components help conceptualize a community, a commonly invoked entity. ACKNOWLEDGMNTS C. Murray provided helpful discusion and asited in conceptualizaton of figures. K Barret, C. Guyer, K. Smith, L. Sith, M. Conner, N. Chadwick, J. Grand, C. Anderson, J. Maerz, C. Murray, J. Stiles, S. Stiles, D. Alix, C. Romagos, J. Goesling, D. Laurencio, M. Wines nd B. Folt provided commnts on earlier versions of the manuscript. E.P. Cox and Auburn University librarians provided asitnc obtaining referencs. LITERATURE CITED Anderson, M. G., C. E. Ferre. 2010. Conserving the stage: climate change and the geophysical underpinnings of specis diversity. PLoS ONE 5:e11554. 49 Baum, D. A. 1998. Individuality and the existenc of species through time. Systeatic Biology 47:641?653. Bernier, R. 1984. The species a an individual: facing esentialsm. Systeatic Zoology 33:460? 469. Brogaard, B. 2004. Species a individuals. Biology and Philosophy, 19:223?242. Brown, J. H., and D. F. Sax. 2005. Biological invasions and scientfic objetiviy: reply to Casey et al. (2005). Austral Eology 30:481-483. vender-Bares, J., K. H. Kozak, P. V. A. Fine, and S. W. Kembel. 2009. The merging of community ecology and phylogetic biology. Ecology Lters 12:693-715. Chase, J.M. 2003. Community asembly: when should history matr? Ocologia 136:489-498. s, J. M. and M. A. Leibold. 2003. Ecological Nihes. T University of Chicgo Pres, Chicago, IL. 212 pp. Choi, Y. D. 2007. Restoration ecology to the future: a cal for new paradigm. Restoration Ecology 15:351-353. Clarke, K. R. 1993. Non-parametric multivarite analyses of changes in community sructure. Australian Journal of Eology 18:117-143. Coleman, K. A., and Wiley, E. O. 2001. On species individualism: a new defens of the species- as-individuals hypothesi. Philosophy of Sinc 68:498?517. Colwel, R. K., and D. C. Les. 2002. The mid-domain efect: geometric constraints on the geography of specis rihnes. TREE 15:70-76. Cracraft, J. 1983. Sies concepts and speciaton analysis. Current Ornithology 1:159-187. raft, J. 1987. Specis cpts and the ontology of evolution. Biology and Philosophy 2:63- 80. 50 Crother, B. I., and C. M. Murray. 2011. Ontology of areas of endemis. Journal of Biogeography 38:1009-1015. Davis, M. A., and L. B. Slobodkin. 2004. The scienc and values of restoration ecology. Restoration Ecology 12:1-3. de Queiroz, K. 1999. The general lineage concept of species and the defining properties of the species catgory. Species: new intrdisciplnary esys (ed. by R.A. Wilson), pp. 49?89. MIT Pres, Cambridge, MA. De Queiroz, K. 2007. Species concepts and species delimtaion. Systematic Biology 56:879- 886. Donlan, J. 2005. Re-wilding North America. Nture 436:913-914. ufr?ne, M., and P. Legendre. 1997. Speis aemblages and indicator species: the need for a flexibl asymmtrical approach. Eologicl Monographs 67:345?366. Eldredge, N. 1985. Unfinished synthesi: biolal hierarchies and modern evolutionary thought. Oxford University Pres, New York, NY. Ereshefky, M. 1992. Units of evolution: esays on the nature of species. MIT Pres, Cambridge, MA. Fauth, J. E., J. Bernardo, M. Camara, W. J. Resetaris Jr., J.Van Buskirk, and S.A. McCollum. 1996. Simplifying the jrgon of community ecology: a conceptual approah. The Aerican Nturalist. 147:282-286. Field, J.G., K.R. Clrke, and R.M. Warwick. 1982. A practial straegy for analysing multispecies ditribution paterns. Marine Eology - Progres Sris, 8:37-52. 51 Fontaine, C., P. R. Guimar?es, Jr., S. K?fi, N. Loeuile, J. Memott, W. H. van der Putten, F. J. van Ven, and E. Th?bault. 2011. The ecological and evolutionary implicatons of mrging difrent types of networks. Eology Letrs:in pres. Friedlander, A. M., and E. E. DeMartini. 2002. Contrast in density, size, and biomas of reef fishe betwen the northwstrn and the main Hwain islands: the efects of fishing down apex predators. Marine Ecology Progres Seris 230:253-264. Frits, T. H., and G. H. Rodda. 1998. The role of introducd species in the degradation of island ecosystems: a case hitory of Guam. Annual Review of Eology and Systematics 29:113- 140. Frost D. R., and D. M. Hils. 1990. Species in concept and practie: herpetological applicatons. Herpetologica 46:87-104. Frost, D. R., and A. G. Kluge. 1994. A consideration of epistemology in systematic biology, with special referenc to species. Cladistic 10:259?294. Ghiselin, M. T. 1974. A radial solution to the species problem. Systeatic Zoology 23:536-544. hisli. T. 1981. Categoris, life, and thinking. The Behavioral and Brain Sciens 4:269? 313. Ghiselin, M. T. 1987. Species concepts, individuality, and objectiviy. Biology and Philosophy 2:127-143. Ghiselin, M. T. 2002. Species concepts: the basi for controversy and reconcilation. Fish and Fisheries 3:151-160. Gleason, H. A. 1926. The individualistc concept of the plant asociaton. Bulletin of the Torrey Botanicl Club 53:7-26. Gleason, H. A. 1927. Further views on the succesion concept. Ecology 8:299-326. 52 Gosz, J. R. 1993. Ecotone hierarchies. Ecological Applicatons 3:369-376. oteli, N. J. 2000. Null model analysis of speies co-ocurrence patrns. Ecology 2606-2621. Hawkins, B. A., and J. A. F. Diniz-Filho. 2002. The mid-domain efct cannot explain the diversity gradient of Nearctic birds. Global Ecology and Biogeography 11:419-426. Helmus, M .R, K. Savage, M. W. Diebel, J. T. Maxted, and A. R. Ives. 2007. Separating the determinants of phylogenetic community sructure 10:917-925. Higgs, E. S. 1997. What is good eologial restoration? Conservation Biology 11:338-348. ighton, R. 1989. Biochemical evolution in the slimy salnders of the Plethodon glutinosus complex in t estern United Staes. Part I. Geographic protein variation. Illinois Biological Monographs 57:1-78. Holt, R.D. 1977. Predation, apparent competion and the structure of prey communities. Theoretical Population Biology, 12:197?229. Hubbel, S. P. 2001. The Unified Nutral Theory of Biodiversity and Biogeography. Princeton University Pres. Princton, NJ. Hull, D. L. 1976. Are species realy individuals? Systematic Zoology 25:174?191. unter, M. L., Jr. 1996. Fundamntls of Conservation Biology. Blackwel Scienc. Cambridge, Maschusets. 482 pp. Jackson, D.R., and Milstrey, E.G., 1989. The fauna of Gopher Tortoise burrows. In: Diemr, J.E., D.R. Jackson, J.L. Landers, J.N. Lyne, and D.A. Wood. (eds.). relocation symposium procedings, Tech. Rep. No. 5. Talhase, FL: Florida Gmes and Fresh Water Fish Commison, Nongame Wildlfe Program Tchnial Report #5. pp. 86-98. Jackson, L. L., N. Lopoukhine, and D. Hilyard. 1995. Ecologilstoration: a definiton and omments. Retoration Ecology 3:71-75. 53 Johnson, M. T. J., and J. R. Stinchcombe. 2007. An emerging synthesi betwen community ecology and evolutionary biology. TRENDS in Ecology and Evolution 22:250-257. Kinlaw, A., and M. Grasmueck. In pres. Evidence for and geomorphologic consequences of a reptilan ecosyste engineer: the burrow cascde initiaed by the gopher tortois. Geomorphology: in pres. Kitcher, P. 1984. Species. Philosophy of Scienc 51:308?333. Lambek, R. J. 1997. Focal species: a multi-speis umbrela for nature conservation. Conservation Biology 11:849-856. Lambeck, R. J. 2002. Focal species and restoration ecology: response to Lindenmayer et al. Conservation Biology 16:549-551. Lindenmayer, D. B., A. D. Manning, P. L. Smith, H. Possingham, J. Fischer, I. Oliver, and M. A. McCarthy. 2002. The focal-species approach and landscpe restoration: a critique. Conservation Biology 16:338-345. Lawton, J. H. 1999. Are there general laws in ecology? Oikos 84:177-192. os, J. B. 1996. Phylogetic perspectives on community ecology. Ecology 77:1344-1354. Looijen, R. C., and van Andel, J. 1999. Eological cties: conceptual problems and definitons. Perspectives in Plant Eology, Evolution and Systmatics 2:210?222. Manchester, S. J. and J. M. Bullock. 2000. The impacts of non-native speies on UK biodiversity nd the efectivenes of control. Journal of Applied Ecology 37:845-864. ayden, R. L. 2002. On biological species, pecies concpts, and individuation in the natural world. Fish and Fisheries 3:171?196. McCain, C. M. 2004. T mid-domain efect applied to elevational gradients: species rihnes of smal mals in Cost Ric. Journal of Biogeography 31:19-31. 54 McCoy, E. and K. Shrader-Frechete. 1992. Community ecology, scale, and the instability of the stability concept. PSA: Procdings of the Biennial Meting of t Philosophy of Scienc Asociton. 1992:184-199. McGil, B. J. 2010. Towards a unificaton of theories of biodiversity. Ecology Letrs 13:627- 642. cLahln, J. S., J. J. Helmann, and M. W. Schwartz. 2007. A framework for debate of asited migration in an era of clite change. Conservation Biology 21:297-302. Mikkelson, G.M. 1997. Methods and mtaphors in community ecology: the problem of defining stability. Perspectives on Scienc 5:481-498. ishler, B. D. and Brandon, R. N. 1987. Individuality, pluralism, and the phylogenetic speies concept. Biology and Philosophy 2:397?414. Mithel, R. J., J. K. Hiers, J. J. O?Brien, S. B. Jack, and R. T. Engstrom. 2006. Silviculture that sustains: the nexus betwn silviculture, frequent prescribed fire, and conservation of biodiversity in longleaf pine forest of the southeastern United Staes. Canadian Journal of Forest Reearch 36:2723-2736. Morrone, J. J. 1994. On the identifcation of areas of endemis. Systematic Biology 43:438-441. ulligan, M. K., and L. Kirkmn. 2002. Burning influences on wiregras (Aristda beyrichiana) restoration plantings: natural sedling recruitment and survival. Restoration Eology 10:334-339. Outcalt, K. W., and D. G. Brockway. 2010. Structure and compositon changes following restoration treatments of longlef pine forest on the Gulf Coastal Pin of Alabam. Forest Ecology and Managemnt 259:1615-1623. 55 Plat, W. J., G. W. Evans, and S. L. Rathbun. 1988. The population dynamics of a long-lived conifer (Pinus palustris). American Nturalist 131:491-525. Palmer, M. A., R. F. Ambrose, and N. L. Poff. 1997. Ecological theory and community restoration ecology. Restoration Ecology 5:291-300. Parmesn, C. 2006. Eological and evolutionary responses to recent climate change. Annual Review of Ecology, Evolution and Systematics 37:637-669. Parmesn, C., and G. Yohe. 2003. A globaly coherent fingerprint of climate change impacts cross natural systems. Nature 421:37-42. Pim, S. L. 1984. The complxity and stability of ecosystems. Nature 307:321-326. Reydon, T. A. C. 2009. Species and kinds: a critique of Rippel?s ??one of a kind?? acount of species. Cladisti 25:660?667. Riciardi, A., and D. Simberloff. Asited colonization is not a viable conservation straegy. Trends in Ecology and Evolution 24:248-253. Ricklefs, R. E. 1990. Eology. 3 rd edition. W.H. Freeman, New York. klfs, R. E. 2008. Disntegration of the ecologicl comunity. American Nturalist 172:741- 750. Rieppel, O. 2007. Species: kinds of individuals or individuals of some kind. Cladistic 23: 373? 384. Riser, P. G. 1995. The staus of the scienc examining ecotones. BioScienc 45:318-325. Schlapfr, M. A., D. F. Sx, and J. D. Olden. 2011. The potential consrvation value of non- native species. Conservation Biology 25:428-437. Simberloff, D. 2003. Confronting introduced species: a form of xenophobia? Biological Invasions 5:179-192. 56 Simberloff, D. 2004. Community ecology: ist time to move on? The American Nturalist 163:787-799. Simpson, G. G. 1961. Principles of animal txonomy. Columbia University Pres, New York. Thuiler, W. 2004. Paterns and uncrtintes of pecies? range shifts under climat change. Global Change Biology 10:2020-2027. Tingley, M. W., W. B. Monahan, S. R. Beisnger, and C. Moritz. 2009. Birds track their Grinelian niche through a century of climate change. Procedings of the National Acdemy of Sienc 106:19637-19643. Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. Melio. 1997. Human dominaton of Earth?s ecosystems. Scienc 277:494-499. Walther, G-R, E. Pt, P. Convey, A. Menzl, C. Parmesn, T. J. C. Beebe, J-M. Fromentin, O. Hoegh-Guldberg, and F. Bairln. 2002. Ecologicl responss to recnt climat change. Nature 416:389-395. Webb, C. O., D. D. Ackerly, M. A. McPek, and M. J. Donoghue. 2002. Phylogenies and community eology. Annual Reviw of Ecology and Systematics 33:475-505. hite, P. S., and J. L. Walker. 1997. Approximating nature?s varition: selting and using refnce informtion in restoration ecology. Restoration Ecology 5:338-349. Wilbur, H.M. 1972. Competion, predation, and the structure of the Ambystoma-Rana sylvatica community. Ecology 53:3-21. iley, E. O. 1980. Is the evolutionary species fiton? ? A consideration of clase, individuals, and historical entites. Systemati Zoology 29:7680. Wiley, E. O. 1988. Virianc biogeography. Annual Review of Ecology and Systematics 19:513-542. 57 Wilamson, M. H., K. C. Brown, M. W. Holdgate, H. Kornberg, R. Southwood, and D. Mollison. 1986. The analysis and modeling of British invasions. Philosophical Transactions of t Royal Society of London. Seris B, Biological Siencs 314:505-522. Winterhalder, K., A. F. Clewl, and J. Aronson. 2004. Values and sienc in ecological restoration- respons to Davis and Slobodkin. Restoration Ecology 12:4-7. Yager, L. Y., C. D. Heis, D. M. Epperson, M. G. Hinderliter. 2007. Gopher tortoise reponse to habita mnagemnt by prescribed burning. Journal of Wildlfe Managemnt 71:428-434. Young, J. E., and B. I. Crother. 2001. Alozyme evidence for the sparation of Rana areolata and Rana capito and for t resurrection of Rana svosa. Copeia 2001:382-388. 58 Figure 1. New communities can arise from a variety of diferent proces. Transitons (represntd here as fuzzy bars in the absnc of discret geologic events) may occur simply because communities change constantly over time (A). Whetr such change (X to Z) is of relvanc depends on the sale resrchers designatd as important. In (B) community W transitons into two communitis (X and Y) following biogeographic divergence (e.g., the division of a cunity following a shifting river channel). If a biogeographic convergence event merged communities X and Y, they would form a new community Z. These changes are likey of ecological revanc. In (C) community X transitons to a new sta as reult of either a natural transiton (e.g., succesion) or se anthropogenic disturbance. Transitons may also occur as a reult of repairing community degradation (fuzzy gray bar in D). It is philosophicly impossible to manage a cty that has changed to a new community such that i once again becoms a part of the original community (se txt under ?An Ostensive Definiton). Howver, it is posible to recreate t structure and function of the original community (representd by X 1 ). In al of the above snarios, the scale of relevanc to t researcher may alow for a community to experienc some degree of change over tim while remining t sme individual. Figure 2. The number of specis in a given area is a subst of the specis that occur in a larger are. Trefore, for example, only a subset of al the species that make a large asemblge (D) unique wil be unique to a salr community (A). Conversly, mny species within a sal asemblage (A) wil not be unique to this habita type but wil be unique to a larger asemblge (B). The nested nature of species amblges across spatial scles suggest the resarcher is responsibl for designating the relvant scale when identifying the unique species amblage of a given area. 59 Figure 1. 60 Figure 2. 61 Chapter 3 Restoration of Avian Populations and Asemblages: Long-term Efects of Fire Surrogates and Prescribed Burning Abstract. Removal of hardwood trees together with reintroduction of fire has been suggested as a method of restoring fire-suppresd longleaf pine (Pinus palustris) forest; howver, wildlf respons to this retoration method is not wl documentd. We examined how bird populations and asemblages in fire-suppresd longleaf pine-sandhils reponded to echanial removal or herbicide applicton or fire alone to reduc hardwood levels. Individual treatnts were ompared to untreated controls and referenc sites. Aftr initial tretment, al sitsre managed with prescribed fire, on an approximatly two-year intrval, for over a decade. Non-etric dimensional saling ordinations suggested that avin asemblages on sites that experincd any form of hardwood removal difered from asmblages on both fire-suppresd sites and refenc site in the 3-4 years after reatnt. After >10 yers of prescribed burning on al sits, only asmblages at sis tretd with herbicide were indistinguishable from asemblages at referenc sites. Species indicative of referenc sites becam evenly distributd aong al tretnts by the end of the study. Oupancy modeling of individual species highly asociated with refenc sites also demonstraed ireasing homogeneity across treatmnts over tim. Overal, athough we documentd long-trm and variable asmblage-level change, our results indicate occupancy for birds considered longleaf pine specilsts ws silar at treatment and refenc sits after over a deca of prescribed burning, regardles of initil method of hardwood removal. Key words: longleaf pine, non-metric dinsional scaling, occupancy modeling, prescribed fire, red-cockaded woodpecker. 62 INTRODUCTION The longleaf pine (Pinus palustris) forest once ranged throughout the southeastern United Staes but has declid considerably due to land us conversion and suppresion of frequent fire (Wre et al. 1993). In the absenc of a natural fire regime (fire every 1-10 years, Myers 1990), hardwood trees (.g., Oak, Quercus spp.) become established in the midstory (Mitchel et al. 2006). Thes tres alter forest compositon, generaly degrading t habita for longleaf pine asocites (Mans 2006). Restoration is a management objective for many longleaf-pine forest (Brockway et al. 2005) and is generaly ateptd by reoving hardwood tres and reintroducing fire. Several methods of hardwood removal are commonly used, including mechanial removal (i.e., fling and girdling), applicaton of herbicides, fire, or a combination of thes methods. These restoration straegis are typialy evaluated by measuring vegetaion respons (.g., Provencher et al. 2001a,b), and fauna are generaly asumd to respond to changes in the habita (i.e., become pasively restored, Scott e al. 2001). The inital efects of restoration on wildlfe may become ls pronounced over time (.g., Hanowski et al. 2007) but its generaly thought that periodi burning is sufficnt to maintn initl response of longleaf pi forest to hardwood removal. For example, mhanicl removal of hardwood tres coupld with reintroduction of fire is benefical for bird species aociated with pine-grasland ecosystems and this management is likely suffient to manage their populations (Cram et al. 2002, Provencher et al. 2002b). However, fire my need to be applied repetedly over long-tie periods to achieve efective restoration of southern pine forest (Waldrop et al. 1992). Therefore, long-trm studis are esential to acurately characteriz 63 wildlfe response to restoration activies (Zdler and Calawy 1999, George and Zack 2001), including change in abundance (Purcl et al. 2005). Birds play vitl rols in forest a predators, consumers, and sed disperse (Means 2006). This funal group may be sensitve to landscpe-scal habita change (.g., McGrigal and McComb 1995, Drapeu et al. 2000, Lindenmyer et al. 2002); its therefore important to understand how birds respond to forest management and restoration. Aseblage-level study my identify general trends in how wildlfe responds to habita change (Luck and Dily 2003, Bennet e al. 2004). However, measure of asmblage structure may obscure species-pecif and population-level trends (Ms et al. 2009). To determine how avin populations and assemblges repond to forest retoration of fire-suppresd longleaf pine sndhils over long tie scals, we investigaed asmblage-level response in breding birds after hardwood removal and again aftr al sits recived prescribed fire for over tn years. We usd this analysis to inform selction of speies losely asociated with referenc sites and examined changes in occupancy for thes scis over tim. If avin populations and aseblages on treatment sits were indistinguishable from those on refernc sites, w asumd mnaent objctives wre met. METHODS Study Site and Experimental Design This study took place on Eglin Air Force Bas, Okaloosa, Snta Ros, and Walton Counties, Florida, U.S. W focusd our study on fire-suppresed longlef-pine sndhils. Most of 64 the reatment sits had no records of having been burned since 1973, when record keeping began (B. Wils, Jackson Guard, pers. comm.). Tn sits experid burns of varying extnt from unknown cuses betwen 1977 and 1989. The study was baed on a randomized block design to asign hardwood removal treatments to 24 sites, ech 81 ha in area and asigned to six blks (Provencher et al. 2001). Mthods of hardwood removal applied in 1995 included burning (Burn), herbicide applicaton (Herbicide), or feling-girdling (Mchanial). There was alo control, whih reeived no tretmnt. Six refrenc sites (alo 81 ha in size) wre also designated. Thes refrence sits had been subjectd to a fire fquency over a long time-span simlar to t natural disturba regim and wre slectd as a representation of the ancstral condition and a target of restoration eforts (Whit and Walker 1997). More details regarding refenc sitelction can be found in Provencher et al. (2001a). The burn tretment ws applied betwen April-June 1995, herbicide (ULW, hexazinone, 1.68 kg of active ingredint/ha, Gonzalz 1985) was applied in early May 1995, and mechanial removal was conducted betwen June and November of 1995 Hrbicide and Mechanial sites were subjectd to a prescribed burn in 1997. More details on the treatments arevailble in Provencher et al. (2001a,b). After 1999, al sites recived comparable management, which included prescribed fire on a 2-3 year rottion, but no additional hardwood reoval or herbiide applicaton. Beause prescribed fire was applied to al sites following the initial experimentl tretments, w have approahed t analysis in the contxt of two phase. T first phas eployed a randomized block design with multiple treatments plus refrenc sits. After 1999, al treatment sits, including those that wre originaly considered controls (i.e., fire-suppresd longlaf pine sandhils) wre subjcted to prescribed burning but no additional forest management. Referenc 65 site wre stil considered representative of the desired condition. For clarity, we refr to treamnts acording to naming conventions designated during the initil phas of the project. Tre Basal Area We calculted basal are density for longleaf pine and al oak (Quercus spp.) trees for each site using dat ollcted in 1995, 1998, and 2009-2010. We considered individual pine trees ! 4 cm diaetr at breast hight (dbh) as a component of the overstory and those " 4 cm dbh as components of the midstory. We considered an individual oak tree a component of the overstory if it was !6.3 cm dbh and a component of the midstory if it was l. Dat on individual trees were collectd in subsites, summed, and divided by total smpled are to generate basal are density.. In generating mean values for 2009-2010, we excluded one block and a singl referenc site that experiencd additional management activis outide of this study. Avian Sampling To maximize the likelhood of indepence, al avian smpling in treatment sites occurred in the corners furtst from other treatmnt sites. Spling within refrenc sites urred within t centr of the site (se figure in Provencher et al. 2002b). Al samples wre colletd betwen approximately 0545 h and 1000 h. T order of sites sapled within a given morning was varid to rduc bis aocited with time; however, we wre unabl to sample site in random order because of occasional restricions on acs to sits because of miltry training activies. Four tretmnt sites or 2-4 referenc sites were sampled in aorning unles 66 aces was retriced due to miltary training. Two observers visited a site during each smpling ocsion and walked along parale transects 250 m apart from each other and approxitely 450 m long. 1994 Sampling (Pre-Tatment) Al sites wre visited four times betwen 4 May and 18 July 1994, prior to hardwood removal treatmnts (Provencher et al. 2002b). Ech time a treatent sit was viited, two observers conducted eight minute point counts approxitly 200 m apart along the transects (four total point counts each visit) and recorded al detecd birds. Efort was doubled on refenc sites, which resulted in eight point counts on four transects per site. 1998-1999 Sampling Al sites wre visited six tmes each betwen 1 May and 30 June in 1998 and again in 1999 (12 total smpls). In contrast to pre-tremnt dat collection, each observer conducted only one point count per visit (the point was at eiher t beginning or end of a transect, varying by visit). In addition, observers walked an entire transect (450 m) and recorded al birds detecd. Walking a transect took approximtely 22 minuts. With the addition of the eight-minut point count, ech obsrver sapled birds for approximately 30 minutes per site (Provencher et al. 2002b). 67 2009-2010 Sampling We atmpted to sample four blocks and three refrence sits four times each betwen 27 May and 13 July of 2009. Excptions include one Mchanial site that was spld only thre times, a referenc site that received a single visit, and a referenc site tts sampled twice. Five blocks and five refrenc sits wre sampld thre tims each betwn 11 My and 18 June of 2010. Four transects werealked in referenc sites in 2009, otherwise sampling methods replicated thos usd in 1998-1999. Ordination We treated ech point count asn indepent sample for the pre-treament data, such tha four smpls wre cated per visit. When necesry, w randomly reoved from consideration half of the point counts conductd on refrence sites to make smpling efort omparable tohat of treatment sites. For both study periods folowing hardwood reoval treamnts, w pooled detcions from both observers collectd within a transect and point count, such that each tim a sit was viited one sampl was cretd. We removed the first two samples in each of 1998-1999 from consideration to mke dat from thes years comparable tohat of the other study periods. We created a presenc/absenc matrix wre if a species wa detcd within a sample it was given a score of ?1?, wheres species not deted in a given smple wre asigned a score of ?0?. Theref, a species could have sored a maximum of 16 detcions in a given site for the pre-treament study period, eight for 1998-1999 and seven for 2009-2010. 68 We usd non-metric diensional scaling (NMDS), which is a nonparametric ordination (Clarke 1993) useful for graphialy demonstraing diferencs in aseblages bad on species identity and an index of abundance (.g., Kennedy et al. 2010). W conducted two NMDS ordinations with Bray-Curtis (Sorenson) distancs. The first ordination included pre-treament and 1998-1999 data. The second ordination iluded t 1998-1999 and 2009-2010 dat. As some sites were not smpld in every time period, we conducted sparate ordinations to facilte cparisons. Staistcl significanc was detrmined by comring obsrved stres to that obtained by Monte Carlo simultions. We usd a multi-response permutation procedure (MRPP, Mielke and Berry 2001) to test the hypothesi that avian asmblages did not difr betwen treamnts and referenc sites. For each ordination, we reoved species detced in only one sple to reduc the impact of rare and rarely detcd species. Although rare speis may be important to include in some analyses (e.g., Cao et al. 1998), removing rare scies i a common straegy within NMDS (e.g., Kreutzwisr et al. 2005). We also did not include two aquatic specis, the great blue heron (Ardea herodias) and common loon (Gavia imer). Ordinations and MRPP were completd using PC-ORD 4.0 (McCune and Meford 1999). Ordition graphs were prepared with SigmaPlot (Systa software, Sn Jose, CA) aicrosoft PowerPoint 2008. If the MRP indicted no significnt diferenc betwn a given treatmnt and referenc site in eitr of the study periods following hardwood removal, we considered this evide that the treatment was efective at restoring the avian aseblage. Treatnt sites significantly diferent than refrenc sits were suggested to be inefctive at restoring the avin asemblage. 69 Indicator Species Analysi We identifed indicator species for the diferent tatments and referenc sites using methods described by Dufr?ne and Lgendre (1997). This analysis considered the number of detcions and exclusivity of each species to sites within a treatment. Indicator species wre asigned a value of 0-100. A 100 would indicat a species w obsrved in al sits of a given trement and no other sites (Dufr?ne and Legendre 1997). W used the matrices decribed in the ordination sction to identify indicator specis. Staistcl significanc ws detrmined with 1000 Monte Carlo simulations. Inditor sies analyses wre completd within PC-ORD 4.0 (McCune and Meford 1999). As part of Eglin Air Force Bas?s recovery plan for red-cockaded woodpeckers, artifcal cavities wre instaled in pine tres betwen the 1998-1999 and 2009-2010 study periods (K. Gult, pers. comm. Jckson Guard). Therefore, we cannot interpret any change in their staus as an indicator speies aftr 1999 as due to t restoration mthods usd in this study. Red-ded woodpekers are klptoparasite of red-cockaded woodpecker cavities (USFW 2003) and may also have benefited from instaltion of artifal cvities; howver, this benefit was likely reltily smal, compared that of red-cockaded woodpeckers, hence, w interpret change in paraetrs asocited with this speies a relevant to hardwood removal treatmnts. Ocupancy Modeling The species w selctd for occupancy modeling included those identifed as indicators (as determined with indiator speies analysis, above) of refere conditions in 1994. Of these 70 species, w excluded red-cockaded woodpeckers and blue jays (Cyanocita cristata). We xcluded red-cockaded woodpekers due to the additional mnagement this species recived and eld blue jays due to their generalist habita use and widespread distribution. To stndardize t methodology across study periods, we usd only point count data and made ech visit (.e., sampling occsion) equivalent to the sum of the detecions from two point counts. In 1994, eight point counts were conductd in each refrenc sit per visit; we randomly removed four point counts. Sinc four point counts were conducted during each visit in 1994 (and only two for the following study periods), w removed point counts conducted in the middle of the transect (half of al point counts in 1994) from analysi. In 2009, four point counts wre conductd in eah referenc site; w randomly selctd wo of these for analysis. We again removed the first two sampls of 1998-1999. W poold data such that each tim a sit was visited, one saple ws generated. As a reult, we generated four sples for the pre-trement data, eight smpls for the 1998-1999 smpling period, and sven sampls for t 2009 and 2010 smpling period. To model occupancy, we usd the multi-seaon model (MacKenzi et al. 2003) in Program PRESENCE (Hines 2010). In contrast to the single sson model (MacKenzi et al. 2002), the multi-seaon model aows for changes in occupancy within a site by distinguishing betwen primry spling periods, betwen which ocy may change, and scondary sapling periods, in which the population is considered closed to imigration, emigration, or extincton. We defined t pre-trement data (1994), imdiat post-reatent (1997-1998) and long-trm post-reatment (2009-2010) as our three priry spling periods. Each visit whin a primary spling period was considered a scondary sampliriod. 71 We modeled occupancy in treatment and referenc sites separatly for each species. Our interest wa in deting changes in pecis occupay; therfore, w considered detion probability a nuisance parametr. We first modeled detcion probability for each species and used the combination of covariats that best predictteibility based on Akaike?s Information Criteria (AIC), in succesive ocupancy models. Models used to evaluat detecion probability in tretmnt site included 1) constant detecability over al thre study periods, 2) varying detecability by treatnt type, 3) varying tbility by treatment type and each secondary smpl, and 4) varying detecability by treatment type and priry sapling period. Models used to evaluate detcion probability in refrenc sites included 1) constnt detecability over al thre study periods, 2) varying detecability by secondary sampling period and 3) varying detecability by primary spling period. W evaluated five occupancy models for each species in treatment sites, thee models representd sveral hypothese (Table 1) regarding how bird popultions may respond to hardwood removal. We evaluatd two occupancy models for each species in referenc site and used the combination of covariates producing the best eimte of detion probability for each specis to model this paramtr within occupancy models for that species. Models were ranked using AIC and we considered models with !AIC values < 2 as important (Burnham and Anderson 2002). When more than one model had IC values < 2, we usd model averaging to estimate occupancy probability. No formalethod exist for detrmining goodnes-offit for ulti-son models. Therefore we usd the singl season model (MacKenzi et al. 2002) for the post-reatment data (1998-1999) with occupancy (") s a function of tretmnt type and detecion probability varying by survey and treatment type to acount for unmeasured 72 heterogeneity (e.g., Adams et al. 2011). We conducted this analysis for data asociated with treamnt sites only. RESULT Tre Basal Area Oak basal are generaly decreased following treatment (Table 2). However, midstory oaks in Mechanil sites increased aftr initial tretent to lvels higher than obsrved in pre- treamnt conditions. In Control sites, oak basl area decreased over time. Longleaf pine basal re ws similar among treatents over time, but basl are in treatnt sites had not approached that of referenc sites by the end of the study. Ordination A two-dimensional solution was the best fior the 1994 and 1998-1999 data with a final stre of 17.91 and an itability of 0.0005 after 200 iteraions (stre was le than expected by chance; P = 0.03; Figure 1). Referenc sits, locatd within the middl of Axis 1 in 1994, moved slightly aong this axi betwn 1994 and 1998-1999. With one exception, control sites alo moved slightly aong Axis 1 betend 1998-1999 but wre sparated from Refrence site on Axis 2. Al sites that experincd some form of hardwood removal in 1995 moved considerably along Axis 1 and approached refrencs site along Axis 2 (Figure 1). 73 A three-dimnsional solution was the best fior the 1998-1999 and 2009-2010 data with a final stres of 11.29 and an instability of 0.004 after 200 iteraions (stres was le than expected by chance; P = 0.03; Figure 2). Control sits moved ciderably along Axis 2. These sit displayed the greats degree of change from 1998-1999 to 2009-2010, which was not unexpected since t mnagemnt regie they received during this time shifted more drasticaly than other treatmnt sites (i.e., they had not reid a hardwood-reoval treatnt or prescribed burning by 1998-1999). Tre was considerable variation in the spatil arrangement of Burn, Mechanial, and Herbicide sites but they appeared to be generaly converging to the centr of Axis 1 and the bottom of Axis 2. Referenc sites were significantly diferent from treatment sites in 1994, whereas no diferencs wre detcd among tretnts (Tabl 3). Following hardwood removal, Control and refree sites were distinc from each other and al other treatment sites. In 2009-2010, refenc sitsre disti f al tretments except for Herbicide sits, and Herbicide sites wre diferent from Controls and Mechanial sits (Tabl 3). Identifcation of Indicator Species Eight species wre positvely asociated with referenc sites in 1994; eight species wre also positvely asociatd with Mechanil sit thre years post-reatment (Table 4). Al other trements had fewr, or no, indiator species (Table 4). Only two specis wre asociated with the sa treatnt for both study periods following hardwood removal. 74 Ocupancy Modeling For occupancy modeling, we slcted six species that were positvely asociated with refenc sites in 1994: American kestrel (Falo sparvrius), Bachman?s sparrow (Pucaca aestivals), blue grosbeak (Passrina caerulea), brown-headed nuthath (Sita pusila), northern bobwhite (Colinus virginianus) and red-heded woodpecker (Melanerpes erythroceophalus, Tabl 5). Goodnes of fit-est for t 1998-1999 data did not provide evinc for any unexplined heterogeneity. Ocupancy of American kestrel and northern bobwhite in treatment sites was best explained by models that alowd occupancy to vary by primary spling period. American kestrel occupancy was considerably lower in treatent sites than in referencs, but thes values were simlar fter hardwood removal (Tabl 6). Northern bobwhit occupay remained relatively high throughout the duration of the study. Estimated occupancy probabilites for Bachman?s sparrow, brown-headed nuthatch, red- headed woodpker, and blue grosbeak exhibited simlr paterns through time (Figures 3-6). T most important models for ech specis included tratnt as covariat (Tbl 5). Ocupancy probabilites for al four sies wre lower in tretment sites than in referenc sites prior to hardwood removal. In the 2-3 years folling hardwood reoval and in general, occupancy probabilites for tse species in Mechanial and Herbicide sites were simlar to those of refere sites. By the end of the study howver, occupancy probabilits in al tretent sites wre simlar to thos in referenc sites for al four speies. 75 DISCUSION Controlled experiments are the most efective means of determining how wildlfe asemblages repond to ecological restoration (Block et al. 2001). Yt, i is dificult to xperintly apply treatments at scale applicable to many wildlfe speies due to their long- lives and spatial ecology. For the few controlld studis that exist, most take plc over relatively smal teporal and spatial scls (Bennet and Adams 2004). Our study, whih incorporates a lndscpe-scale experimentl design and spans more than a decade, reveald that hardwood reduction n a longlaf pine forest may benefit avian asmblges and, specifaly, populations of species positvely asociated with site in referenc condition. Our result are consistnt with Mas et al. (2009), in that asemblage lvel dirsity may be a poor proxy for an individual species? reponse to habita cnge. Trends documentd herein would appear to suggest that appliaton of herbicide followed by frequent prescribed burns was the most efective method for incresing the simlarity of avian asemblages to those obsrved at refrenc sits. Howr, in-depth consideration of species positvely asocitd with longleaf pine in refenc sites suggested any of the methods of hardwood removal used in this study (including burning alone) together with long-trm prescribed burning was likely sufficent to recover populations of these species. Hrdwood removal together with reintroduction of fire within fire-suppresed longleaf pine sandhils i likely to benefit avian species aociated with the refenc habita. Howver, complet eradication of hardwood trees my be to the detrimnt of even longlef pine specialst (Prkins et al. 2008). We did not identify thresholds of hardwood density required to sustin the 76 species w identifed as indicators of referenc conditions, though it may be worthwhile to xplore the concpt (Gu?nete and Vilard 2005). Ordination and Indicator Species Eight species wre significant indicators of referenc sites during the pre-treament period (Table 4), including four speies identifed elswhere as longleaf pine specilst: red- cockaded woodpecker, Bachman?s sparrow, brown-aded nuthatch, and northern bobwhite (Engstrom 1993, Means 2006), and three species that prefer open woodlands (American kestrel, red-headed woodpecker and blue grosbeak; Ingold 1993, Smalwood and Bird 2002). Interestingly, blue jays were also significntly asociated with referenc sites; thi is countrintuitive due to their general use of many habits and pehant for oak trees (Tarvin and Woolfenden 1999). Although we expectd al tretments to have had simlar bird asmblges prior to hardwood removal in 1995, three specis (Downy Woodpecker, Picoides pubescens, Northern Cardinal, Cardinalis cardinalis, and Pileatd Wr, Dyocopus pilatus) wre significantly asociated with site that would eventualy become Control sites (Table 4). However, gin that thes thre specis were not positvely asociatd with Control sites after hardwood removal, we asume this aociaton did not confound our interpretaions. The multi- response permutation procdure provided support for this aumption and suggested that al treamnts wre comparable to each other and distinc from referenc sites prior to rdwood reoval. Three years after treatment applicaton, there was a clear distincon betwen avin asemblages on sites that experincd hardwood removal nd smblages on Control sites. This 77 suggest al three mthods of hardwood removal were efective at altering the bird asemblage from those that inhabit fire-suppresed sandhils, corroborating earlir analyses (Provencher et al. 2002b). Howver, bird asemblages at refnce sites were also ditinc from thos on hardwood removal sites, suggesting that rdwood removalas insufficent to restore the avian asemblage to the refenc condition. Although Mechanial sites clustered together in 1998-1999, they were not distinc from other sites. Howver, eight bird specis wre positvely asociated with Mchanial sites, in contrast to only one species in Controls and two in refrenc sites (Table 4). With the exception of blue grosbeak, a sis of open woodlands, thes species are not thos that weould expect to necesrily use either pine or hardwood-dominated forest more than any other species. W suggest the trends w identifed are temporary and result from disturbance unique to fling and girdling trees (i.e., mechanial reoval). The specis positvely asocited with Mechanial sites may be responding to short-term changes in inset communities brought on by kiling adult oak tres and leaving the slash (.g., Aul?n 1991). W viw the bird asemblages t these sits a transitonal. Sinc bird asemblages at these sits reembld those of refrence sits by t end of the study, we suggest occupancy declined for the majority of specis identifd as indicator species n Mechanial sites in 1998- 1999 while longlef pine sialst remained. By 2009-2010, bird asmblges ore closely resembld that of refrence sits. However, dditional monitoring of these sits would have ben necsary to confirm this hypothesi. Mechanial removal of trees wa initialy as efective at reducing oak overstory density as appliaton of herbicides (Tabl 2 and Provencher et al. 2001b). Both methods are thought to quickly advance restoration, as compared to fire alone (Menges and Gordon 2010). However, 78 although herbicide applicaton prohibits reprouting of oaks (Brockway et al. 1998), mechanial removal may atualy encourage oak resing, at least in the absenc of prescribed fire (Provencher et al. 2001b). Although we did not find evidence to suggest birds positvely asocited with Mechanial sites wre exclusive to thes areas, our analysis corroborates work suggesting thes sites have reltively high avian species rihnes (Provencher et al. 2003). Provencr et al. (2002b) suggested that lhough mhanical hardwood removal (and also herbicide applicaton) may benefit bird aseblages in the short-term, additional management, such as preribed fire, is necesary to maintn these trends. There were eight speis aocited with rfrenc sites in the initial survey, but only two (red-cockaded woodpeckers and red headeoodpeckers) were positvely asociated with these site three years later. Fiften years into t study, none of the original indictor specis wre stil asocitd with refrenc sites (although Misippi kites, which previously had revealed no relationship, were; Table 4). This suggest hardwood removal in treatment sites incresd the simlrity of bird asmblges on treatment sites tohose of reference sites over the long-term, to the extent that they were indistinguishabl by the conclusion of the study. T asociaton betwn Misippi kits and referenc sites i atributed to a 2009 nest on one referenc site. Population Levl Efects of Restoration Prior to hardwood removal, there were sveral species with relatively high occupancy probabilites only in referenc sites. By t end of the study, ocupancy probabilites for these specis had generaly ireasd and become relatively uniform across al sites. This suggest tha, for birds positvely asocited with longlf pine forest in referenc condition, burning 79 alone over a long period of time is sufficent to increase occupancy probability on previously fire-suppresed sits to levels typial of refree sits. Mhanial removal of hardwood trees or herbicide applicaton aclrated the obsrved response. This finding was further supportd by the long-term change in ocupancy probability a control sits, which reeived prescribed fire after t first phase of the study. American kestrel, blue grosbeak, red-headed woodpecker, Bachman?s sparrow and brown-heded nuthatch responded positvely to hardwood removal. For the lter three species, occupancy probabilites wre simlar to those of refenc site iediatly following treaments at the Mchanial and Herbicide sites, which may have influenced the interpretaion by Provencr et al. (2002b) that thes treatments are reltily efective. Red-aded woodpecker and blue grosbeak had reltively high occupancy probabilits prior to tretment, hih is lily a function of low detecion probabilites. These species wre detecd infrequently within several difrent tatments and it is dificult to onfirm that a speies i absent ift has a low detcion probability. Northern bobwhite, although detecd more often in referenc sites prior to treatment (Table 4), were likely presnt in al sits in every study period (Tabl 6). This species i of consrvation concrn and population declines ha been atributed to habita degradation and fire suppresion (Brennan 1991). Our result suggest that, alhough northern bobwhite abundance may be greater in referenc sites than ipre-treament fire-suppresed longlaf-pine sandhils, the specis w presnt in al treatment sites even prior to hardwood removal. Blue grosbeak ws an indictor of referenc sites in the initial phase of the study, which is consistent with is known habita prefres (Engstrom et al. 1984), but this species xhibited trends incistentith those of other bird species for which we modeled occupancy. Although 80 ocupancy probability for blue grosbeak at treatment sits generaly increased over time, probabilites at treatment sits were distinguishabl from those of refere sits. Howver, our ability to model occupancy for this spcies wa limted becaus it was detced at nearly ery site. Synthesi Avian asemblages t formerly fire-suppresed longleaf pine sandhils became indistinguishabl from those on rfrenc sites only aftr applicton of herbiide followd by over a decade of prescribed burning. Howver, for species highly asociated with the ancestral condition of this habita, occupancy probabilites on treatmnt sites generaly becam omparable to those on refere sites over the long-term, regardles of initil method of hardwood removal. Overal, our study demonstrad difrent tmporal and treatmnt responses to restoration on the population and aseblage level in birds. These shifts my be ongoing, for example, idstory oak density a Burn and Mchanial sites appear to be increasing relative to lvels idiately after treatment (Table 1). If oak density continues to irese, w might expect to obsrve declines in the occupancy probability of longleaf-pine specialst. Applicaton of herbicide likey prohibited resprouting of oaks for a longer period of time than mehanil removal (Table 2), alowing for the bird asemblage to gradualy transiton to one comparable to that of rfrenc sites. Although herbicide applicton appeared to be the most efective long-trm straegy for moving avian asemblages toward that of refenc sites, further resarch is warrantd. For example, the active ingredient of the herbicide in this study, Hexazinone, cn rech surrounding bodis of water (Nary et al. 1983); Hexazinone is generaly 81 considered safe for wildlfe (Michael et al. 1999) but limted resarch has been conducted pertaining to som groups (Berril et al. 1994, Bridges and Slitsch 2000). Given the diversity nd rarity of se wildlfe species in longleaf pine ecosystems (Means 2006) and on Eglin Air Force Bas in particular (.g., Enge 2005), w suggest caution when developing management plans which include hexazinone applicaton. If the goal is to restore avin asemblges to a condition representative of those on fire- maintned longleaf pine fst, our data suggest that applicaton of herbicides followd by long-term prescribed burning is an efective pproach. However, it may be more appropriate to focus restoration goals on a suite of indiator speies aociatd with the longleaf pine ecosystm (Lambeck 1997, 2002, Roberge and Per Anglstam 2004); these are t specis likely to be of conservation concern due to the global iperilent of this habita type. In this cae, reintroduction of burning alone over the long-trm (a relatively-inexpensive mthod, Provencher et al. 2002b) would be an appropriate approach. ACKNOWLEDGMENTS Funding was provided by the Straegic Environmental Research and Development Program (SERDP) roject Number: SI-1696. S. Pokswinski, K. Hirs, and B. Wilams provided logistic help. A. Barnet provided asitance with data mnagement. D. Sipson and M. Cent provided asitance in the field. L. Smith, C. Guyer, M. Conner, B. Hls, C. Anderson, J. Grand, N. Chadwick, J. Ditloff, C. Schmidt, D. Laurencio, J. Stiles, S. Stiles, O. Sylvester, C. McClure and S. Golladay provided comments on an erlier draft of the manuscript. C. McClure 82 provided helpful discusion of Program PRESENCE. E. P. Cox and Auburn University librarians d asitance obtaining referencs. LITERATURE CITED Adams, M. J., C. A. Pearl, S. Galvan, and B. McCreary. 2011. Non-native species impacts on pond occupancy by an anuran. Journal of Wildlfe Magement 75:30-35. Aul?n, G. 1991. Increasing insect abundance by kiling deciduous tres: a method of improving the food situaion for endangered woodpeckers. Eography 14:68-80. Bennet, A. F., S. A. Hinsley, P. E. Belamy, R. D. Swetnam, and R. Mac Nly. 2004. Do regional gradints in land-use influence richnes, compositon and turnover of bird asemblages in sml woods? Biologial Conservation 119:191-206. Bennet, L. T., and M. Adas. 2004. Asemnt of ecological efects due to forest harvesting: approaches and staistcl isue. Journal of Applid Eology 41:585-598. Berril, M., S. Bertram, L. MGilvray, M. Kolohon, and B. Pauli. 1994. Efects of low concentrations of forest-use pesticdes on frog embryos and tdpols. Environmental Toxiology and Chemistry 13: 657?664. Block, W. M., A. B. Franklin, J. P. Ward, Jr., J. L. Ganey, and G. C. White. 2001. Design and implentation of monitoring studies to evaluate the succes of ecological restoration on wildlf. Restoration Ecology 9:293-303. Brennan, L. A. 1991. How can we reverse the northern bobwhite population decline? Wildlfe Society Bulletin 19:544?555. 83 Bridges, C. M., and R. D. Semlitsch. 2000. Varition in pesticde toleranc of tadpoles among and within specis of Ranidae and paterns of amphibian decline. Consrvation Biology 14:1490-1499. Brockway, D. G., K. W. Outcalt, nd R. N. Wilkins. 1998. Restoring longleaf pine wiregras ecosystems: plant cover, diversity and biomas following l-rat hexazinone applicaton on Florida sndhils. Forest Ecology and Mnagement 103:159-175. Brockway D. G., K. W. Outcalt, D. J. Tomzak, and E. E. Johnson. 2005. Restoration of longleaf pine ecosystems. Gen. Th. Rep. SRS-83. Ashevile, NC. U.S. Department of Agriulture, Forest Srvice, Southern Research Staion. Burnham, K.P., and D. R. Anderson. 2002. Model sction and multi-model inferenc, second edition. Springer-Verlag, New York, New York, USA. Clarke, K. R. 1993. Non-paramtric multivariat analyses of changes in community sructure. Australian Journal of Eology 18:117-143. Cao, Y., D. D. Wilms, and N. E. Wilams. 1998. How important are ra species in aquatic comunity ecology and bioasent? Linology and Ocenography 43:1403-1409. Cram, D. S., R.E. Masters, F. S. Guthery, D. M. Engle, and W. G. Montague. 2002. Northern bobwhite popultion and habita respons to pine-grasland restoration. Journal of Wildlf Management 66:1031-1039. Drapeau, P., A. Leduc, J. Giroux, J. L. Savard, Y. Bergeron, and W. L. Vickery. 2000. Lndscape-scal disturbances and changes in bird communities of boreal mixed-wood forest. Eologil Monographs 70:423-444. Dufr?ne, M., and P. Legendre. 1997. Species amblages and indicator species: the need for a flexibl asymmtrical approah. Eologicl Monographs 67:345?366. 84 Enge, K. M. 2005. Herpetofaunal drift fenc surveys of stephead ravines in the Florida panhandl. Southestern Naturalist 4:657-678. Engstrom, R. T. 1993. Charactristic mmals and birds of longleaf pine forest. Pages 127-138 In S. M. Hermann, editor. Procedings of the Talimbers Fire Ecology Conferenc. Tal Timbers Resrch Staion, Talhase, FL, USA. Engstrom, R. T., R. L. Crawford, and W. W. Baker. 1984. Breeding bird populations in relation to changing forest sructure following fire exclusion: a 15-year study. Wilson Bulletin 96:437-450. George, T. L., and S. Zack. 2001. Spatial and temporal considerations in restoring habita for wildlfe. Restoration Eology 9:272-279. Gonzalez, F. E. 1985. Hexazinone and sulfometuron methyl herbicides for pine forestry. Procdings of the Southern Wed Socity 38:146-156. Gu?nete, J., and M. A. Vilard. 2005. Thresholds in forest bird response to habita alteraion as quantitave targets for conservation. Conservation Biology 19:1168?1180. Hanowski, J., N. Dnz, and J. Lind. 2007. Breding bird response to riparian forest management: 9 years post-harvest. Forest Ecology and Managemnt 241:272-277. Hines, J. E. 2010. PRESENCE3-Software to estite patch ocupancy and related parametrs. USGS-PWRC. http:/.mbr-pwrc.usgs.gov/software/prese.html. Ingold, J. L. 1993. Blue Grosbeak (Guiraca caerulea ). In The Birds of North Aerica, No. 79 (A. Poole and F. Gil, Eds.). Philadelphi: The Acdemy ofatural Scincs; Washington, D.C.: The Americn Ornithologist' Union. 85 Kennedy, C. M., P. P. Marra, W. Fagan, and M. C. Nel. 2010. Landscpe matrix and species trais mediat responses of Neotropical resident birds to forest fragmenttion in Jami. Ecologicl Monographs 80:651-659. Kreutzweisr, D. P., S. S. Capel, and K. P. Good. 2005. Efects of fine sedimnt inputs from a logging road on stream insect communities: a large-sal experintal approach in a Canadin headwter stre. Aquatic Eology 39:55-66. Lambeck, R. J. 1997. Focl species: a multi-speies umbrela for nature conservation. Conservation Biology 11:849-856. Lambeck, R. J. 2002. Focal species and restoration ecology: response to Lindenmayer et al. Conservation Biology 16:549-551. Lindenmayer, D. B., R. B. Cunningham, C. F. Donnely, H. Nix, and B. D. Lindenmayer. 2002. Efects of forest fragmentation on bird asemblages in a novel landscpe context. ologial Monographs 72:1-18. Luck, G., and G. C. Dily. 2003. Tropical countryside bird asemblages: richnes, compositon, foraging difer by landscpe context. Ecological Applictons 13:235-247. Mas, B., D. D. Putra, M. Wltert, Y. Clough, T. Tsharntke, and C. H. Schulze. 2009. Six years of habita modificaton in a tropical rainforest margin of Indonesia do not afct bird diversity but endemi forest speies. Biologicl Conservation 142:2665-2671. MacKenzi, D. I., J. D. Nichols, G. B. Lachmn, S. Droege, J. A. Royle, C. A. Langtim. 2002. Estimating site ocupancy rates when detecion probabilits are ls than one. Ecology 83:2248-2255. 86 MacKenzi, D. I., J. D. Nichols, J. E. Hines, M. G. Knutson, and A. B. Franklin. 2003. Estimating site ocupancy, colonization, and local extincton when a species i detecd iperfectly. Ecology 84:2200-2207. McCune, B. and M.J. Meford. 1999. PC-ORD. Multivariate Analysis of Ecological Dt, Version 4. MjM Software Design, Gleneden Bech, OR, USA. 237pp. cGarigal, K., and W. C. McComb. 1995. Relationships betwen landscpe structure and breeding birds in the Oregon coast range. Ecological Monographs 65:235-260. Means, D. B. 2006. Vertbrate faunal diversity of longlef pine ecosystems. Pages 157-213 In: S. Jose, E. Jokela, and D. L. Miler, editors. The laf pi et: ecology, silviculture, and restoration. Springer, New York, USA. Menges, E. S., and D. R. Gordon. 2010. Should mchanial tretments and herbicides be used as fire surrogates to manage Florida?s uplands? A review. Florida Scientst 73:147-174. ichael, J. L., E. C. Wbber, Jr., D. R. Bayne, J. B. Fischer, H. L. Gibbs, and W. C. Sesock. 1999. Hexazinone disipaton in forest ecosystems and ipacts on aquatic communities. Canadian Journal of Forest Reearch 29:1170-1181. Mielke, P. W., and K. J. Berry. 2001. Prmuttion methods: a distance function approach. Springer-Verlag, New York, New York, USA. itchel, R. J., J. K. Hirs, J. J. O?Brin, S. B. Jack, and R. T. Engstrom. 2006. Silviculture that sustains: the nexus betwen silviculture, frequent prescribed fire, and conservation of biodiversity in longlaf pine forest of the southeastern United Staes. Canadian Journal of Forest Reearch 36:2723-2736. Myers, R. L. 1990. Scrub and high pine. Pages 150?193 in R. L. Myers and J. J. Ewel, editors. Ecosystems of Florida. University of Central Forida Pres, Orlando, Florida, USA. 87 Neary, D. G., P. B. Bush, and J. E. Douglas. 1983. Of-site movement of hexazinone in stormflow and baseflow from forest wtersheds. Wd Scinc 31:543-551. Perkins, M. W., L. M. Conner, and M. B. Howz. 2008. The importae of hardwood trees in the longleaf pine forest ecosystem for Sherman?s fox squirrels. Forest Ecology and Managemnt 255:1618-1625. Plentovich, S., J. W. Tucker, Jr., N. R. Holler, and G. E. Hil. 1998. Enhancing Bachman?s sparrow habita vi management of red-cockaded woodpeckers. Journal of Wildlfe Management 62:347-354. Provencher, L., B. J. Hrring, D. R. Gordon, H. L. Rodgers, K. E. M. Galey, G. W. Tanner, J. L. Hardesty, and L. A. Brennan. 2001a. Efects of hardwood reduction tchniques on Longlef Pine sandhil vegetaion in northwst Florida. Restoration Eology 9:13-27. Provencher, L., B. J. Herring, D. R. Gordon, H. L. Rodgers, G. W. Tanner, J. L. Hardesty, L. A. Brennan, and A. R. Lit. 2001b. Longleaf pine and oak responses to hardwood reduction techniques in fire-suppresed sandhils in northwest Florida. Forest Ecology and Management 148:63-77. Provencher, L., N. M. Gobris, and L. A. Brennan. 2002a. Efects of hardwood reduction on winter birds in northwest Florida longleaf pine sandhil forest. Auk 119:71-87. Provencher, L., N. M. Gobris, L. A. Brennan, D. R. Gordon, and J. L. Hardesty. 2002b. Breeding bird response to midstory hardwood reduction in Florida sandhil longleaf pine forest. Journal of Wildlf Management 66:641-661. Provencher, L., A. R. Lit, and D. R. Gordon. 2003. Predictors of species rihnes in northwest Florida longleaf pine sandhils. Conservation Biology 17:1660:1671. 88 Purcel, K. L., S. R. Mori, and M. K. Chase. 2005. Design considerations for examining trends in avin abundance using point counts: exampls from oak woodlands. Condor 107:305- 320. Roberge, J-M, and P. Angelstam. 2004. Usefulnes of the umbrela species concept as conservation tool. Conservation Biology 18:76-85. Sott, T. A., W. Wehtj, and M. Whtje. 2001. The need for straegic planning in pasive restoration of wildlfe populations. Restoration Ecology 9:262-271. Smalwood, J. A., and D. M. Bird. 2002. American Kestrel (Falco sparverius). In The Birds of North America, No. 602 (A. Pool and F. Gil, eds.). The Birds of North America, Inc., Philadelphi, PA. Tarvin, K. A. and G. E. Woolfenden. 1999. Blue Jay (Cyanocita cristata), The Birds of North America Online (A. P, Ed.). Ithac: Cornel Lab of Ornithology. U.S. Fish and Wildlfe Srvice. 2003. Reovery plan for the red-cockaded woodpecker (Picoides borealis): scond revision. U.S. Fish and Wildlfe Srvie, Atlant, GA. US. Waldrop, T. A., D. L. White, S. M. Jones. 1992. Fire regims for pine-grasland communities in the southeastern Unitd Staes. Forest Ecology and Manageent 47:1095-1210. are S., C. Frost, and P. Doerr. 1993. Southern mixed hardwood forest: the former longleaf pine forest. In W. H. Martin, S. G. Boyce, and A. C. Echternacht [eds.], Biodiversity of the southeastern United Stes: lowland trrestrial communitis, 447?493. John Wiley and Sons, Nw York, Nw York, USA. White, P. S., and J. L. Walker. 1997. Approximating nature?s variation: selcting and using refnce informtion in restoration ecology. Restoration Ecology 5:338-349. 89 Zedlr, J. B. 1993. Canopy architecure of natural and planted cordgras marshes: selcting habita evaluation crita. Eological Applictons 3:123-138. Zedlr, J. B., and J. C. Calwy. 1999. Traking wetland restoration: do mitgaion sites follow desired trajectoris? Restoration Ecology 7:69-73. 90 Figure 1. Non-metric diensional scaling ordination of bird asemblages observed on fire- suppresed longlaf pine sandhils on Eglin Air Force Bas, 1994 and 1998-1999; 1 = 1994, 2 = 1998-1999. Figure 2. Non-metric diensional scaling ordination of bird asemblages observed on longleaf pine sandhils following hardwood removal on Eglin Air Forc Bas, 1998-1999 and 2009-2010; 1 = 1998-1999, 2 = 2009-2010. Figure 3: Relationship betwen probability of occupancy and year of study for Bachman?s sparrow pre-trement (1994; A), and three years (1998-1999; B) and fourten yers (2009-2010, C) following hardwood reoval on fire-suppresed longleaf pine sandhils, Eglin Air Force Base, Florida. Lack of numerical convergenc and an inability to compute variance-covarianc matrix suggest sndard errors should be interpretd with caution. Figure 4: Relationship betwen probability of occupancy and yer of study for brown-headed nuthatch pre-trement (1994; A), and three years (1998-1999; B) and fourten years (2009- 2010, C) following hardwood removal on fire-suppresed longleaf pine sandhils, Eglin Air Force Bas, Florida. Progra PRESENCE was unabl to produc stndard errors surrounding occupancy at Herbicide sites in B and C. Figure 5: Relationship betwn probability of occupancy and year of study for red-headed woodpecker pre-trement (1994; A), and three years (1998-1999; B) and fourten yers (2009- 2010, C) following hardwood removal on fire-suppresed longleaf pine sandhils, Eglin Air Force Bas, Florida. igure 6: Relationship betwen probability of occupancy and year of study for blue grosbeak pre-treament (1994; A), and thre years (1998-1999; B) and fourten years (2009-2010, C) 91 following hardwood removal on fire-suppresed longleaf pine sandhils, Eglin Air Force Bas, Florida. 92 Table 1. Models used to evaluate occupancy probabilites for selct bird species detced from 1994-2010 to determine how their popultions responded to hardwood removal on fire-suppresd longleaf pine sandhils. Treatment Ocupancy Models Hypothese !(PRD), " (P),p(x)* Ocupancy and colonization varied by primary spling period (TRT + P), " (TRT + PRD), p(x) y aoltirid by priry spliriod and treatment type ! + PRD), # (T + P), p(x) Ocupancy and extincton varied by primary spling perind tretnt t , "(TRT + P), (TRT + PRD), p(x) Colonization axtiton rats vary by priry sampliriod and treatment type and are based on initial occupancy !, "(TRT), #(T + PRD), p(x) Colonization vries by treatment type and extincton rates vary by primary spling period and tretment type, both are basd on initial occupay Referenc Ocupancy Models !(.), " (.),p(x) Ocupancy and colonization rates are constant (PRD), (P),p(x) y aoltion rats vary by primry sapling period 93 Table 2. Oak and longleaf pine basal are in treatment and referenc sites before and after oak removal. Al units are m 2 /ha (stndard error). 1994 1998-1999 2009-2010 Pinus palustris midstory Burn 0.10 (0.04) 0.04 (0.02) 0.04 (0.01) Control 0.08 (0.01) 0.07 (0.01) 0.02 (0.01) Herbicide 0.06 (0.02) 0.04 (0.01) 0.29 (0.08) Mhanial 0.09 (0.02) 0.04 (0.01) 0.08 (0.02) Referenc 0.04 (0.02) 0.03 (0.02) 0.13 (0.05) Pinus palustris overstory Burn 10.20 (2.06) 9.50 (1.94) 11.22 (2.14) Control 7.19 (0.78) 7.63 (0.90) 8.86 (1.26) Herbicide 9.39 (2.22) 9.56 (2.22) 10.18 (1.65) Mhanial 10.00 (2.06) 9.64 (2.25) 11.25 (1.77) Referenc 17.62 (1.91) 17.92 (1.97) 18.71 (2.64) Quercus sp. midstory Burn 1.53 (0.65) 0.43 (0.16) 0.81 (0.30) Control 1.08 (0.09) 1.13 (0.13) 0.66 (0.20) Herbicide 0.76 (0.19) 0.04 (0.01) 0.15 (0.03) Mhanial 0.92 (0.18) 0.08 (0.05) 1.58 (0.26) Referenc 0.09 (0.03) 0.12 (0.09) 0.08 (0.07) Quercus sp. overstory Burn 14.26 (3.40) 8.18 (2.52) 7.09 (2.26) 94 Control 11.54 (1.27) 10.53 (1.55) 5.08 (1.61) Herbicide 13.54 (2.97) 2.73 (0.13) 0.11 (0.07) Mhanial 13.04 (2.12) 4.50 (2.52) 6.42 (5.43) Referenc 4.88 (1.30) 2.73 (0.25) 1.42 (0.82) 95 Table 3. P-values aociated with multi-response permutation procedure on pairwse comparisons of tretmnt and referenc sits (1994, 1998-1999, and 2009-2010). Bold indicates a significant diferenc betwn groups. 1994 Burn Control Mechanial Herbicide Referenc Burn 0.55 0.94 0.85 0.0006 Control 0.86 0.21 Mechanial 0.81 0.0007 Hrbiide 0.002 1998-1999 Burn Control Mechanial Herbicide Referenc Burn 0.01 0.10 0.25 0.003 Control 0.0009 0.001 0.0005 Mechanial 0.16 0.0006 Hrbiide 0.04 2009-2010 Burn Control Mechanial Herbicide Referenc Burn 0.36 0.54 0.05 0.04 Control 0.93 0.02 0.01 Mechanial 0.01 Hrbiide 0.58 96 Table 4. Bird specis identifed as having a significant asociaton with treatment or referenc sites for al three study periods, Eglin Air Forc Bas, Florida. Percnt Indicator Value Treatment of Mxium Asociaton Species Burn Control Mechanial Herbicide Referenc P-value 1994 Referenc American Kestrel 18 0 0 0 54 0.006 Bahmn's Sparrow 0 1 4 60 0.002 Brown-headed Nuthatch 2 0 0 0.009 Blue Grosbek 2 8 11 5 51 0.003 Jay 14 23 17 17 29 0.007 Northern Bobwhite 9 10 8 13 50 0.001 Red-cockaded Woodpecker 7 1 3 4 60 d Hear 0 1 81 0.001 Control Downy Woodpecker 6 43 16 3 2 0.016 97 Northern Cardinal 19 35 23 10 2 0.047 Pileatd Woodpecker 17 34 8 17 4 0.048 1998-1999 Control Eastern Titmouse 22 31 16 18 11 0.001 Mechanial Blue Grosbeak 15 2 37 23 14 0.036 Brown Thrasher 8 11 42 12 5 0.004 Carolina Wren 19 20 36 1 0.043 Chimney Swift 3 9 38 12 3 0.04 Eastern Bluebird 6 1 41 35 5 0.023 strn Towhee 13 6 48 5 0 0.008 Indigo Bunting 6 0 50 2 0.01 Summer Tanager 5 5 41 17 2 0.027 Referenc Red-cockaded Woodpecker 12 0 10 22 39 0.007 d Hear 24 20 16 37 0.004 2009-2010 Control Eastern Titmouse 24 35 27 10 3 0.001 98 Mechanial Eastern Towhee 27 28 37 2 3 0.018 Hrbiide Brown-aded Nuthatch 20 9 12 30 23 0.02 Referenc Misippi Kite 0 0 0 0 67 0.022 99 Table 5. Top models explaining occupancy paterns of selct bird species within fire-suppresed longleaf pine sandhils undergoing hardwood removal, 1994-2010. Species Site Model AIC !AIC Weight Likelhood Par. -2*Loglike Amrian Kestrel TRT "(PRD),#(P),p(TRT) 255.79 0 0.94 1.00 9 237.79 REF (.),(.),p(SURV) 103.08 0.90 21 61.08 Blue Grosbeak TRT ",#(TRT + PRD),$(TRT + PRD),p(TRT + P) 468.64 0 0.46 1.0 17 434.64 "(PRD),#(P),p(TRT +PRD) 469.15 0.51 0.36 0.77 11 447.15 REF (.),(.),p(SURV) 120.59 0 0.88 1.0 21 78.59 Bachman's Sparrow TRT ",#(TRT + PRD),$(TRT + PRD),p(TRT + SURV) 339.23 0 0.97 1.00 33 273.23 REF "(.),#(.),p(S) 119.01 0.70 21 77.01 (PRD),(P),p(SURV) 120.68 1.67 0.30 0.43 24 72.68 100 Brown-headed Nuthatch TRT !,"(TRT + PRD),#(TRT + PRD),p(TRT + SURV) 294.13 0 0.79 1.00 33 228.13 REF !(.),"(.),p(.) 111.22 0.91 3 105.22 Northern Bobwhite TRT !(PRD),"(P),p(TRT + PRD) 544.44 0 0.97 1.00 11 522.44 REF (.),(.),p(.) 134.42 0 0.99 3 128.42 Red-headed Woodpecker TRT !,"(TRT),#(T + PRD),p(TRT + PRD) 410.66 0 0.54 1.00 16 378.66 TRT !,"(TRT + PRD),#(TRT + PRD),p(TRT + P) 410.98 0.32 0.46 0.85 17 376.98 REF !(.),"(.),p(.) 121.8 0 0.98 1.00 3 115.8 101 Table 6: Probability of occupancy (and standard errors) for American kestrel and northern 1 bobwhit observed on longleaf pine sndhils on Eglin Air Forc Bas, 1994-2010. 2 1994 1998-1999 2009-2010 American Kestrel Tretnt 0.18 (0.12) 0.85 (0.13) 0.7 (0.17) Referenc 0.83 (0.12) 0.83 (0.12) 0.83 (0.12) Northern Bobwhite Treatment 0.99 (0.12) 0.97 (0.0) 1.0 (0.0) Refrenc 1.0 (0.001) 1.0 (0.001) 1.0 (0.001) 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 102 18 19 20 21 22 23 24 25 26 27 28 103 29 30 31 32 33 34 35 36 37 38 39 40 104 41 42 43 44 45 !"#$% "% "#$% "#&% "#'% "#(% )% )#$% )#&% *+,-% ./-0,/1% 23456-7461% 83,9747:3% ;3<3,3-43% =44+>6-4?% @,360A3-0% B% !"#$% "% "#$% "#&% "#'% "#(% )% )#$% )#&% *+,-% ./-0,/1% 23456-7461% 83,9747:3% ;3<3,3-43% =44+>6-4?% @,360A3-0% *% !"#$% "% "#$% "#&% "#'% "#(% )% )#$% )#&% *+,-% ./-0,/1% 23456-7461% 83,9747:3% ;3<3,3-43% =44+>6-4?% @,360A3-0% .% 105 46 47 48 !"#$% "% "#$% "#&% "#'% "#(% )% )#$% )#&% *+,-% ./-0,/1% 23456-7461% 83,9747:3% ;3<3,3-43% =44+>6-4?% @,360A3-0% B% !"#$% "% "#$% "#&% "#'% "#(% )% )#$% )#&% *+,-% ./-0,/1% 23456-7461% 83,9747:3% ;3<3,3-43% =44+>6-4?% @,360A3-0% *% !"#$% "% "#$% "#&% "#'% "#(% )% )#$% )#&% *+,-% ./-0,/1% 23456-7461% 83,9747:3% ;3<3,3-43% =44+>6-4?% @,360A3-0% .% 106 49 50 51 !" !#$" !#%" !#&" !#'" (" (#$" )*+," -.,/+.0" 12345,6350" 72+863692" :2;2+2,32" <33*=5,3>" ?+25/@2,/" A" !" !#$" !#%" !#&" !#'" (" (#$" )*+," -.,/+.0" 12345,6350" 72+863692" :2;2+2,32" <33*=5,3>" ?+25/@2,/" )" !" !#$" !#%" !#&" !#'" (" (#$" )*+," -.,/+.0" 12345,6350" 72+863692" :2;2+2,32" <33*=5,3>" ?+25/@2,/" -" 107 52 53 54 !" !#$" !#%" !#&" !#'" (" (#$" )*+," -.,/+.0" 12345,6350" 72+863692" :2;2+2,32" <33*=5,3>" ?+25/@2,/" A" !" !#$" !#%" !#&" !#'" (" (#$" )*+," -.,/+.0" 12345,6350" 72+863692" :2;2+2,32" <33*=5,3>" ?+25/@2,/" )" !" !#$" !#%" !#&" !#'" (" (#$" )*+," -.,/+.0" 12345,6350" 72+863692" :2;2+2,32" <33*=5,3>" ?+25/@2,/" -" 108 Chapter 4 55 Restoration of Reptil Asemblages: 56 Long-term Efects of Fire Surrogats and Prescribed Burning 57 58 Abstract. Atempts to restore fire-suppresed and hardwood-invaded longleaf pine forest 59 are common but the long-trm efects on wildlf are raly measured. We employed a 60 lndscape-scale, randomized-block design to identify how reptil asblages initialy 61 responded to restoration treatmnts including removal of hardwood trees vi feling and girdling, 62 herbicide applicaton, or prescribed burning alone. Then, we xamined reptil asmblages fter 63 al sites experiencd more than a decade of prescribed burning at 2-3 year return intervals. Data 64 wre collctd concurrently at referenc sites chosen to represent trget conditions for 65 restoration. Reptils reponded to the greats extnt, initialy, to prescribed burning but reptile 66 asemblages at al sites, including referenc sites, wre generaly indistinguishable by the end of 67 the study. Thus, w suggest prescribed burning in longleaf pi forest over long tim-periods is 68 an efective straegy for restoring reptile asmblages to the referenc condition. 69 Key Words: Aspidosclis sexlineatus, longlaf pine, non-mtric dinsional scaling, prescribed 70 fire, squamates, Tantila coronata. 71 72 INTRODUCTION 73 74 The longleaf pine ecosystem of the southeastern United Staes wa once extnsive 75 (Landers et al. 1995) but is now highly iperild (Noss 1988). A fire-daptd habita, longleaf 76 pine forest not lost to human development or land-use conversion may become degraded due to 77 fire suppresion (Ns 1989). Hrdwood tres (e.g., Oaks, Quercus sp.) oftn eventualy 78 109 dominate forest in which fire has been excluded, altering forest sructure and compositon 79 (Mitchel et al. 2006). 80 Restoration of longleaf pine forest typicaly includes reintroduction of frequent fire 81 (Brockway et al. 2005). Howver, public aepte of prescribed fire is mixed (e.g., Shindler 82 and Tomn 2003, Brunson and Evans 2005). In additon, fire aloneay be inefctive at 83 restoring the functions of highly degrad ecosystems (Brockway et al. 2005). Consequently, fire 84 surrogates have been developed, including herbiides and mechanil removal of hardwood 85 tres. Atmpts ha ben undertaken to determine the reltive eftivenes of these surrogates 86 at reducing fuel loads (McIver et al. 2009) as wl ast restoring the getaion to t ancstral 87 condition (Brockway et al. 1998). However, the efects of fire surrogates on wildlfe popultions 88 are les wl known (Russel et al. 1999). 89 Hrbicides and mchanial mens of hardwood removal are unlikey to replicate the 90 ecologial efets of frequent, prescribed burning in longleaf pine forest (Menges and Gordon 91 2010); although they may be useful tools in restoring conditions necesary to reintroduce fire 92 into these forest (Provencher et al. 2001a, Brockway et al. 2005). It is generaly suggestd that 93 ts fire surrogates may quickly alter a forest towrds a desired condition, and tha this cnge 94 can be maintned or enhanced through subsequent applicatons of prescribed fire (Brockway et 95 l. 2005, Outclt and Brockway 2010). Howver, it my be neesary to apply fire over long 96 time-periods to move the habit to a condition comparabl to tha which was observed prior to 97 European setlment (Waldrop et al. 1992). 98 Longlaf pine forest contain a rich diversity of vertebrat animals (Means 2006) and 99 forest management may have considerable efcts on asociatd wildlfe (Russl et al. 2004, 100 Van Ler et al. 2005). The magnitude of thes efets i not often quantifd, perhaps due to the 101 110 considerable chalenges aociated with acurately characterizing relevant parametrs (e.g., 102 Block et al. 2001, Gardner et al. 2007). For exampl, although it is likely necsary to study 103 wildlfe response to mnagement on a long teporal scale (Zdler and Calwy 1999, 104 Cunningham et al. 2007), most investigaons typily lst only a f years (e.g., Bennet and 105 Adas 2004, Greenberg and Waldrop 2008, Kilpatrick et al. 2009, Sten et al. 2010). 106 Smal reptils may be abundant in suitable habitas, comprising a considerable 107 component of the vertebrat biomas (e.g., Bullock and Evans 1990). Many reptils occur largely 108 in longleaf pine forest, to the extent that several are considered specilst of this habit (Guyer 109 and Baily 1993, Means 2006). Consquently, this group may be partiularly sensitve to forest 110 mnageent (.g., Grenberg et al. 1994, Todd and Andrews 2008) but its dificult to predict 111 how thy may respond (Lindenmayer et al. 2008). Within ts study, we usd ordination 112 techniques and simlarity and diversity indices to examine how reptil asmblages varied among 113 fire-suppresed longlef pine sandhils treatd with prescribed fire and fire surrogats (herbicide 114 and mechanial hardwood removal) and how repeated presrid burning afectd these initial 115 patrns. 116 117 METHODS 118 119 Study Site 120 121 This study took place on Eglin Air Force Bas, Okaloosa and Sant Rosa Counties, 122 Florida, U.S. We focusd our study on fire-suppresd longlef-pine sndhils. A randomizd 123 block design was usd to asign hardwood removal method treatments to 16 sites, each 81 ha in 124 111 are and arranged in four blocks (Lit e al. 2001, Provencher et al. 2001a). Six of these sits had 125 experiencd a single burn betwen 1977 and 1989; sie this burn frequency is l than the 126 natural fire fquency (i.e., every 1-10 years, Myers 1990), w treat them als fire-suppresed. 127 Al other treatment sits had not been burned since at last 1973 (wn record-keeping began, B. 128 Wilas, Jckson Guard, pers. comm.). Methods of hardwood removal included burning (Burn), 129 herbicide appliaton (Herbicide), or fling-girdling (Mecnial) and Controls, which 130 experiencd no tretmnt in 1995 (below). Four 81 ha refrence sites were also deignated. 131 Refre sitelction is decribed in Provencher et al. (2001a); sitsre selctd as 132 representations of the ancestral condition and a target condition of restoration eforts (White and 133 Walker 1997). 134 135 Treatmnt Applicaton 136 137 Initial hardwood removal treaments occurred in 1995. Burn treaments were applied in 138 April-June. Herbicide, (ULW, hexazinone, 1.68 kg of active ingredient/ha, Gonzlz 1985) was 139 applied in early May and mechanial hardwood removal was conductd betwen June and 140 November. Herbicide and Mnicl sites were subjectd to a prescribed burn in 1997. More 141 details on treatnts are available in Provencher et al. (2001b). After treatment applicaton, al 142 site recived comparable mnagemnt, which iluded prescribed fire on a 2-3 yer rottion but 143 no additional targetd reoval of hardwoods or applicaton of herbicide. One referenc site 144 received herbicide applicaton betwen 1997-2009; thus, we excluded data collctd from this 145 sit during 2009-2010. 146 147 112 Reptil Trapping 148 149 Drift fenc arrays (Campbel and Christman 1982) were placed at the centr of each of 16 150 treament sits and four refrence sites to capture represntatives of Squamat and Tstudines, 151 hereftr, reptiles. Fncs wre made ofluminum flshing and sixten 19 L pitfls were placed 152 along the fncs of each array (30 m totl of flashing per array). In the initial study, arraysre 153 smpled from My-August 1997 and from April-ugust 1998 (Lit 1999, Lit e al. 2001); arrays 154 wre reoved in 1998. In the second phase of the study, we reinstaled rrays in the same 155 location at each site and reptils wre trapped from May-Sptmber 2009 and My-August 2010. 156 For 2009 and 2010, we added box traps to the centr of the rrays as part of a separat sudy 157 (Burgdorf et al. 2005, Stn et al. 2010), but usd t same number and positon of pitfal traps 158 per array as in the original study. 159 All reptiles wre individualy marked in 1997-1998 but due to low recapture rates of most 160 species (.g., astrn fnce lizard, Sceloporus undulatus, 7.4%, broad-heded skink, Plstiodon 161 latips, 6%, litle brown skink, incla lateralis, 0%) and low recapture rates for these 162 animls in general (.g., Todd and Andrews 2008), we only individualy marked A. sxlineatus in 163 2009-2010. We suggest daa used in our analyss (i.e., the number of cptures, irrespective of 164 recapture staus) are comparabl to those usd in otr comparisons of capt rates (.g., 165 MCoy and Mhinsky 1999, Mathews et al. 2010). We did not convert overal captures to 166 captures per trap night becuse trapping efort was tndardizd across al tretments within each 167 study period (e.g., Lit 1999). W excluded box trap cptures from the analysis snc this method 168 was not usd in the initial study. 169 170 113 Vegetation Data 171 172 Vegetaion data were collectd in 1994, 1998, 2009 (treatment sits only) and 2010 173 (refrenc sits only). To quantify groundcover vegetaion and tre density, we collcted data 174 within 16 subplots at Sep 10 in each site (se study design in Provencher et al. 2001a,b). In 175 1994-1998, it was necsary to use dat from Stp 50 within one Burn site. Vegetaion cover 176 clase (1-5%, 5-25%, 25-50%, 50-75%, 75-95%, 95-100%) for four ground cover vegetaion 177 tgoris (i.e., Gras, Woody Liter, Fine Liter) wre converted to midpoints to crete mn 178 percent cover for each site. 179 Midstory trees wre distinguished from overstory trees bad on their diametr at breast 180 height (dbh). A pine treas considered overstory if it was ! 4 incs (10.16 c) dbh. An oak 181 tree was considered overstory if it was ! 6.3 inches (16 cm) dbh. We calulted the mean basal 182 are (m 2 /ha) of midstory and overstory trees for each site. 183 184 Reptil Diversity 185 186 We calculted the Morisita-Horn simlarity index for al reptiles at each site with 187 Estimate S softwre version 8.2 (Colwel 2009). We slcted this particulr simlrity index 188 becus its staistcly robust and relatively insitve to low speies rihnes and saple 189 sizes (Mgurran 2004). We first derid simlarity values betwen refrenc sites in 1997-1998 190 and again for 2009-2010. Each site within a study period was then compared to the mean 191 refenc simlarity index for that sudy period. We clcuted t Shannon index (Mgurran 192 114 2004) to quantify diversity for each site in both study periods. This index is commonly used to 193 describe reptile dirsity (e.g., Grenberg et al. 1994, Michael et al. 2008). 194 W usd a before-aftr control-impact sudy design (Stwart-Oten et al. 1986) to 195 compare reptile simlrity and diversity with serat least squares mns analyses of variance. 196 e cred silarity and dirsity on fire-suppresd Controls and Burn, Mchanil, and 197 Hrbicide treatments to imlarity and diversity on refrence sites in 1997-1998. We also 198 ompared silrity and diversity on treatments in 1997-1998 to simlarity and diversity on 199 treaments in 2009-2010 to determine if reptil aseblages difered following a decade of 200 prescribed burning. Finaly, w compared simlarity and diversity on al treament sites to tha of 201 refrence sites in 2009-2010. Our alpha level for al analyses wa 0.10. 202 203 Non-metric Diensional Scaling 204 205 We conducted a single non-metric diensional scaling ordination, based on Bray-Curtis 206 (Sorenson) distancs, such that each sit appeared in the ordition twic, onc basd on the 207 1997-1998 dat and again based on 2009-2010 dat. We usd a multi-response permutation 208 procedure (MRPP, Mielke and Berry 2001) to detrmine whetr a particular tretment (or 209 refnc site) was distinc from the otr treatments within a given time period. Staistcl 210 significa was determined with Monte Carlo siulations. Analysis wa iplentd with PC- 211 ORD v. 4.25 (McCune and Meford 1999). Ordination graphs were prepared with SigmaPlot 212 (Systa software, Sn Jos, CA) and Microsoft PowerPoint 2008. 213 We asumed that control sites in 1997-1998 wre representative of the pre-treament 214 condition at al treatmnt sites prior to hardwood removal. If the MRPP indicated no significant 215 115 diferenc betwen a treatment and referenc sites, w interpretd this to mean that the treament 216 resultd in conditions indistinguishabl from thos of refnc sites. If the MRPP reveld a 217 signifiant diferenc betwen conditions on treament and refre sits, we considered the 218 trement as inefctive. 219 220 Indicator Species Analysi 221 222 We identifed indicator reptile species by quantifying the relative exclusivity and 223 abundanc of each speis to a partiular tretmnt (Dufr?ne and Legendre 1997). We compared 224 tretment (or referenc) only to other tretents within a study period. Staistcl significance 225 was detrmined with 1000 Monte Carlo simulations. Analysis wa completd with PC-ORD v. 226 4.25 (McCune and Meford 1999). 227 228 Canonical Correspondence Analysi 229 230 To determine if reptile abundance was significantly asociated with treatment type or 231 refenc sites whil acounting for varition in habit haractristic, w conductd sparate 232 canonical correspondene analysis (CCA, ter Braak 1986) for eah study period with specis 233 ptured at lest tn tims. CCA is a form of multivarite regresion useful for identifying 234 relationships betwen abundance data and environmntal variables (Palmr 1993). Within a 235 CCA, least squares regresion is conducted of sitecores (dependent varible, derived from 236 weightd specis abundance data) against environmntal variables (ipendent variable). In this 237 manner, eah site recives a sitcore based on the regresion equation (LC scores, Plmr 238 116 1993). An advantage of this technique is that is unafectd by correlated environmental 239 variables or skewed distributions (Palmer 1993) and may identify reltionships other than those 240 tha are unimodal (tr Braak and Vrdonschot 1995). The analysis alows production of a biplot 241 t graphs sites and species in ordination spae acording to their asociton with 242 environmntal variabls. Important environmntal variables may be graphed onto the biplot as 243 vectors, the length of which represent their reltive importnc (Methrata and Link 2006). 244 Environmntal dat included in t CCA included vegetaive ctgories of Gras, Woody 245 Liter, Fine Liter, Ok Midstory, Pine Midstory, and Oak Orstory. Count data wre square- 246 root transformd and environmental variables wre log-transformed prior to analysis (Palmer 247 1993). Staistcl significanc ws detrmined via Monte Carlo simulations of eigenvalues and 248 species-nvironment correltions. Analysis w compltd with PC-ORD v. 4.25 (McCune a 249 Mford 1999). 250 251 RESULT 252 253 Vegetation Data 254 255 Oak density decreased initialy at the three hardwood removal treaments (Table 1). Burn, 256 Control, and Herbiide tremnt sites had lowr in oak overstory basl are in 2009-2010, while 257 oak basal are increased at Mechanial sites betwen 1997-1998 and 2009-2010. Oak midstory 258 decresed at Control and Hrbiide sits bet- -2010 while itncreased at 259 Burn and Mchanial sites. 260 261 117 Reptil Diversity 262 263 We recorded 1775 captures of 16 reptile species in 1997-1998 and 1648 captures of 19 264 reptil speis in 2009-2010 (Tble 2). Simlarity (Morisita-Horn index) interactd significantly 265 betwen treatment and time (F 4,1 = 2.20, P = 0.093). In 1997-1998, simlarity indies at 266 Refrenc sits wre difrent than Herbicide (P = 0.05) and Control sites (P = 0.0006). These 267 trends are likely influenced heavily by two speies; the relative proportion of A. sexlineatus and 268 southeastern crowned snakes, Tantil coronata, ws lower and higher, respectively, in Control 269 and Hrbicide sites (Figure 1). 270 In 2009-2010, simlarity did not difer betwen treatments (Figure 2), simlarity changed 271 significantly at Controls (P = 0.0006) and Hrbicide (P = 0.06) sites betwen 1997-1998 and 272 2009-2010. Cumulatively, this suggest that Burn and Mechanial tretmnts were efective at 273 replicatng the ancestral condition shortly after treatmnt appliton (i.e., by 1997-1998). 274 Betwen this tim period and 2009-2010, the reptile asblages at Control and Herbicide sites 275 changed significantly to become indistinguishabl from thos on refnce sits. W documntd 276 no sifint hanges in Shannon?s diversity index (F 4,1 = 0.52, P = 0.72). 277 278 Non-metric Diensional Scaling 279 280 A two-dimensional solution was the best fior the data, with a final stres of 9.3 and an 281 instability of 0.00009 after 55 iteraions. T stres was le than expected by chance (P = 0.03; 282 Figure 3). For 1997-1998, the MRPP indicated that Controls, Menial, and Hrbicide sites 283 were indistinguishable (Table 3). Referenc sits were distinc from al tretments, a wre Burn 284 118 site. This suggest that Mechanial and Herbicide treatments did not alter the reptile 285 asmblages such tt they wre difrent from aseblges at sies tha experincd no 286 hardwood reoval. Reptile asmblages at Burn sites likely represntd an intermdiate 287 condition, diferent from thos on Control sites but stil distinguishable to thos of Refrenc 288 site. Reptil asmblages at Herbicide sits wre distinc from thos of refnces in 2009-2010; 289 otherwise there were no difrencs (Table 3). 290 291 Indicator Species Analysi 292 293 Three species wre significantly asociated with a particular tretment in 1997-1998 294 (Table 4). Aspidoslis sxlineatus ws positvely asociated with refrenc sits, ring-necked 295 snake, Diadophis punctatus, was positly asiateith Control sites, and eastrn fenc 296 lizard, S. undulatus, was positvely asociated with Burn sites. No significnt indictor species 297 were identifed in any of the treatments in 2009-2010, indicating a relatively uniform distribution 298 of specis aross treatments. 299 300 Canonical Correspondence Analysi 301 302 For the 1997-1998 data, 35.5% of the species ditribution variance was explained by the 303 first two axes (Figure 4). Eigenvalues for Axis 1 and 2 were significnt (P = 0.03 and 0.09, 304 respectively). Important habita variables explaining variation on Axis 1 included Fine Liter 305 (intraset correlation of -0.78). Specis with CCA scores > 0.5 from 0 on this axi included scarlt 306 snake, Cemophora coccinea (-0.53), and smooth earth snake, Virginia valeriae (-0.51). 307 119 Important variables explaining variation on Axis 2 included oak midstory (intraset correlation of 308 0.67) and oak overstory (intraset correltion of 0.86). Species with scores > 0.5 from 0 on axis 2 309 included green anole, Anolis carolinensis (0.55) and C. cocinea (-0.53). Eigenvalues for the 310 2009-2010 data wre not significantly diferent than expected by chance, suggesting axes were 311 not efctive at explining speies ditribution variancs. 312 313 DISCUSION 314 315 We demonstrae that applicton of prescribed fire resulted in increased imlarity of 316 reptil asblages on trement sits to asemblages on refrenc sits in the short-term, 317 corroborating Lit e al. (2001). In the long-trm, repeted us of prescribed fire was efctive at 318 altering asemblages at all treament sits such that they became indistinguishable from those on 319 refncs sit. Thus, we conclude that long-term presribed burning is an efctive mthod of 320 restoring reptile asmblages in fire-suppresd longleaf pine forest. 321 Basd on silrity indices, reptile asmblges at sies treated with prescribed fire alone, 322 as wel as those treated with mhanial hardwood reoval, wre indistinguishable from 323 smblages on refnc sites in 1997-1998 while asmblages on Control sites and sits treated 324 with herbicide were not. Non-mtric diensional scaling for the same ti period suggestd 325 Burn sites contained reptile asblages difrent from those at otr treatent sits but also 326 distinguishabl from thos on refenc sites. The NMDS also suggestd that smblages on 327 Mecnial sites wre indistinguishabl from those on Control and Herbicide sites. Both 328 analyss wre consistent in identifying asemblagesrol arbii sits a having 329 significantly difrent asmblages from thos at referenc sites in 1997-1998, corroborati 330 120 previous analyses (Lit e al. 2001). Lit e al. (2001) suggested hat some species benefit from 331 habita hetrogeneity, which may be relatively low in both Hrbicide and Control sits. 332 Herbicide sites experiencd a reduction in ground cover vegetaion following herbicide 333 appliton and a reduction in woody debris due to prescribed burns, whereas Control sites 334 contined a high percentage of liter and woody debris (Lit e al. 2001). 335 Regardls of the initil reative efectivenes of the three treatments, our result were 336 generaly consistent in suggesting reptil asmblages atl tretnts wre indistinguishabl 337 from those at refrenc sites by 2009-2010 (with the exception of the NMDS distishing 338 asemblages on Herbicide sits from those on refenc sits). Since reptile asmblages 339 responded quickly following the prescribed burn treatment and asmblages atl sites 340 eventualy beame indistinguishable from those of refrenc sites, we see no long-trm benefit to 341 mchanil or herbicide removal of rdwoods. Prescribed fire alone was sufficent to recover 342 reptile asmblages of the longleaf pine ecosystem over the long-term, as ha ben obsrved 343 aong vetion communitis in longleaf pine forest elewr (Outcalt and Brockway 2010). 344 Based on the result of the NMDS ordination, reptil asmblges at sites treted with 345 herbicide and over a decade of prescribed burning were distinc from thos at refrenc sits. 346 Trefore, it appears that this treatment was reltively inefective at restoring reptile 347 asemblages. Herbicide applicton ws highly efecti at reducing of oak overstory density, to 348 the point that density levels were lower than at refrenc sites. Hardwood trees are important to 349 certain wildlfe specis aocitd with the longleaf pine ecosystm (Perkins et al. 2008), thus, it 350 is possibl that the low hardwood densites at Hrbicide sites wa detrintal to reptiles. 351 Although t limted resarch examining wildlfe respons to hexazinone and reltd products 352 suggest i generaly sfe (Berril et al. 1994, Michal et al. 1999, but se Wan et al. 1988), it is 353 121 also possible that his herbicide had a long-lasting and negative efect on reptiles either directly 354 or on their prey. 355 Although some species likely benefited from hardwood removal, particularly A. 356 sexlineatus, we suggest the asmblage lvel change we documentd is due lrgely to the decline 357 of hardwood-asocited species. For exaple, D. punctatus, although observed only rarely, was 358 an indictor of Controls in 1997-1998 but was not detecd in 2009-2010 despit increased 359 trapping efort. Diadophis punctatus prefers ares with abundant undisturbed liter nd detrius 360 (Perison et al. 1997), as doe Scincela latralis (Conant and Collins 1998), which also declined 361 in obsrved numbers betwen the two study periods. Both species are likely to avoid frequently 362 burned landscpes (Wilgers and Horne 2006). 363 Canonial correspondence analysis identifed potential mechanism behind the 364 asemblage-level change. Fine litr, oak midstory, and oak overstory were iportant variables 365 in explining specis ditribution paterns in 1997-1998 (Figure 4). Sinc Control sites had 366 relatively high levels of Fine Litr and Oak Midstory (Fi 4), these variabls re likely 367 importnt to sral species which delined in relative proportion betwn the two study periods. 368 The CCA suggestd that Virginia valeriae and Cemophora coccinea ere positvely 369 asocited with fine liter cover. Cemophora coccinea also had a negative reltionship with oak 370 tre density, suggesting this snake prefers relatively open canopy habita with abundant fine 371 litter. Both fine liter cover and oak density were positvely asocited with Control sites and are 372 likey to be altred considerably following hardwood removal and reintroduction of fire. We also 373 observed a decline in the relative number of captures of T. coronata (Tble 2, Figures 1 and 2), 374 another speies that my select lndscapes baed on microhabit fetures (Smlitsch et al. 375 1981). Cumulatively, our data suggest that smal snakes deline in abundance at fire-suppresed 376 122 sites following hardwood removal and reintroduction of frequent fire. Todd and Andrews (2008) 377 observed t declines aong this poorly known group of snakes occur in response to timber 378 hast in pi plnttions and suggested that the declines were due largely to reduction in 379 canopy cover and liter density. Our result from natural onglf pine stnds apper to 380 orroborate his finding. 381 Anolis carolinensis wa also observed les frequently in the second study period. In 1997- 382 1998, this species wa positvely asociatd with midstory oak tres, whih likely offer suitable 383 perching habit for this arboreal species (Irschik et al. 2005). Since frequent burning reducs 384 midstory oak density, A. carolinensis populations may decline following a reduction in this 385 habita feture. On the otr hand, the speciesy shift habita use to larger and taler oaks in 386 the absnc of midstory oaks, making tm ls susceptibl to cpture in terrestril traps. 387 Aspidoscelis sxlineatus in fre-suppresd habitas benefit from restoration including 388 hardwood removal and reintroduction of fire (Mushinsky 1985, Perry et al. 2009). We 389 documentd considerable shifts in the relative proportion of this specis betwen the two study 390 periods (Figures 1 and 2). Initialy, A. sexlineatus capture rates on al treatmnt sits difered 391 from those on referenc sites (Tbls 2 and 4). However, the relative proportion of A. sxlineatus 392 at al sits wa simlar aftr repeated prescribed fire over t long-term. Thus, frequent fire is 393 likey to benefit this pecies (Mushinsky 1985) as it does for other reptil species highly 394 asocited with the longlaf pine ecosystem (.g., Yager et al. 2007). It ismportant to note that 395 several reptil species aocited with the longlef pine ecosyste, such as gopher tortois, 396 Gopherus polyphemus, indigo snake, Drymarchon corais, eatrn diamond-backed ratlesnake, 397 Crotal adamante, southern hognosed snake, Hetrodon sius, pinesna, Pituophis 398 melanoleucus, and mimc glas lizard, Ophisaurus mimcus (Guyer and Bailey 1993, Means 399 123 2006) were either undetecd or captured only rarely given our sampling methodology. It is 400 unknown wtr t trends we documentd are applicable to this group. 401 Evaluating reptile asmblage response to forest retoration based solly on the first few 402 years following treatnt my be an inappropriat time scal for reptils. The potential 403 importnce of time sinc treatment is deonstrad by contrasting reptile asmblges at Burn 404 sites with thos at Hrbicide and Mchanial sites. Al three treatmnts recived fire before 405 reptil sampling was initiaed in 1998; howver, Burn sites recived fire in 1995 while 406 Mechanil and Herbicide sites were burned early in 1997. The disparate reptile asmblages 407 obsrved among the treatment sits suggest ime sinc burn may influenc the reptil 408 aseblage. 409 Due to paterns of change in vegetaion, even this study may not have been conducted on 410 a time scale necsary to detec long-term trends in reptile asmblge respons to the treatmnts 411 and subsequent reiroduction of frequent fire. For exapl, sites that experiencd mechanil 412 removal of hardwood trees initialy experiencd a considerabl decline in oak density (Table 1); 413 however, by 2009-2010, oaks had rebounded to the extent that heir overstory deity 414 approached levels oberved at Controls in 1997-1998. This patern is likely due to increased 415 resprouting following mchanial removal (Provencher et al. 2001b). Continued monitoring of 416 these sits may document a gradual increase in oak density and a transiton of the reptile 417 asmblge towrds one more asocitd with hardwood-dominated habitas. 418 It isportant to consider how heterogeneity in detecion probability influences capture 419 rates when mking inferencs about relative abundancs (Mazrolle et al. 2007). In fct, i is 420 likey more appropriat to asume detcion probabilites are unequal when comparing 421 populations (McKenzi and Kndal 2002). Although there a methods to integrat variation in 422 124 detecion probability to generate estimates of relative abundance (.g., Royle and Nichols 2003), 423 they may not be efective at smal sple sizs or low detcion rats (Stn et al. 2011, Chapter 424 5). Most specis within ts study wre detcd infrequently and in low numbers. 425 Greenberg et al. (1994) suggested that disturbance in general, rather than a specif 426 forest-retoration treatment, ay be important in maintning reptile communities aociated 427 with frequently burned habits. Since our study design did not include long-trm monitoring of 428 site treated only with mechanial removal of hardwood trees or herbicides, we are unable to 429 detrmine if continued disturbance of this type would have had the same efcts a frequent fire. 430 In any case, given the uncertainty regarding long-term trends in oak density at Mehanical sites 431 and the difrences in reptil asemblages betwn Hrbicide and referenc sits, reintroduction 432 of frequent fire is the only manant straegy w can reommnd without cavet for efective 433 restoration of small reptile asmblages. Opportunely for lnd managers, prescribed burning is 434 the most inexpensive hardwood reoval treatment evaluated in this study (Provencher et al. 435 2002). 436 437 ACKNOWLEDGMNTS 438 439 Funding was provided by the Straegic Environmental Research and Development 440 Program (SERDP) roject Number: SI-1696. S. Pokswinski, K. Hirs, and B. Wilams provided 441 logistic help. A. Barnet provided asitance with data mnagement. L. Sith, C. Guyer, M. 442 Conner, J. Grand, C. Anderson, and N. Chadwick provided input on earlier drafts. O. Sylvester 443 provided helpful discusion on stistcal analyses and reviewd an erlir draft of the 444 manuscript. D. Simpson, M. Baragona, M. Cent and M. Betzhold provided asitance in t field. 445 125 Numerous volunters asited with trap instaltion. E. P. Cox and Auburn University librarians 446 provided asitanc obtaining refrences. 447 448 LITERATURE CITED 449 450 Bennet, L. T., and M. Adams. 2004. Asemnt of ecological efects due to forest 451 harvesting: approaches and staistcl isue. Journal of Applid Eology 41:585-598. 452 Berril, M., S. Bertram, L. MGilvray, M. Kolohon, and B. Pauli. 1994. Efects of low 453 concentrations of forest-use pesticdes on frog embryos and tdpols. Environmental 454 Toxiology and Chemistry 13: 657?664. 455 Brockway, D. G., K. W. Outcalt, and R. N. Wilkins. 1998. Restoring longleaf pine wiregras 456 ecosystems: plant cover, diversity and biomas following l-rat hexazinone applicaton 457 on Florida sndhils. Forest Ecology and Mnagement 103:159-175. 458 Brockway, D. G., K. W. Outcalt, D. Tomzak, E. E. Johnson. 2005. Restoration of longleaf pine 459 ecosystems. Gen. Th. Rep. SRS-83. U.S. Department of Agricul- ture, Forest Srvic, 460 Southern Resarch Staion, Ashevile, NC. 461 Brunson, M. W., and J. Evans. 2005. Badly burned? Efects of an escaped prescribed burn on 462 social aceptability of wildland fuels treatmnts. Journal of Forestry 103:134?38. 463 Bullk, D. J., and P. G. H. Evans. 1990. The distribution, density and biomas of terrestrial 464 reptiles in Dominica, West Indies. Journal of Zoology 222:421-443. 465 Burgdorf, S. J., D. C. Rudolph, R. N. Conner, D. Saenz, and R. R. Schaefr. 2005. A succesful 466 trap design for capturing large terrestril snakes. Herpetologial Reviw 36:421-424. 467 126 Clarke, K. R. 1993. Non-parametric multivariate analyses of changes in community sructure. 468 Australian Journal of Eology 18:117-143. 469 Colwel, R. K. 2009. EstimateS: Staistcl estimation of species rihnes and shared species 470 from saples. Vrsion 8.2. User' Guide and appliaton published at: 471 http:/purl.oclc.org/estimats. 472 Conant, R., and J. T. Collins. 1998. Reptiles and amphibians: estern/cntral North America. 473 Houghton Mifln Company, Nw York, USA. 616 pp. 474 Cunningham, R. B., D. B. Lindenmayer, M. Crane, D. Michael, and C. MacGregor. 2007. 475 Reptile and arboreal mrsupil response to replantd vegetaion in agriultural 476 landscpes. Ecologicl Applicatons 17:609-619. 477 Dufr?ne, M., and P. Legendre. 1997. Speies amblages and indicator species: the need for a 478 flexibl asymmtrical approach. Eologicl Monographs 67:345?366. 479 Gardner, T. A., J. Barlow, and C. A. Peres. 2007. Paradox, presumption and pitfals in 480 conservation biology: the importanc of habit change for amphibins and reptiles. 481 Biological Conservation 138:166-179. 482 Gonzalez, F. E. 1985. Hxazinone and sulfometuron methyl herbicdes for pine forestry. 483 Procdings of the Southern Wed Socity 38:146-156. 484 Greenberg, C. H., and T. A. Waldrop. 2008. Short-term response of reptiles and amphibians to 485 prescribed fire and mechanil fuel reduction in a southern Appalachin uplnd 486 hardwood forest. Forest Eology and Management 255:2883-2893. 487 Greenberg, C. H., D. G. Neary, and L. D. Hrris. 1994. Efct of high-intensity wildlfire and 488 silvicultural tretmnts on reptile communities in sand-pine scrub. Conservation Biology 489 8:1047-1057. 490 127 Guyer, C., and M. A. Bailey. 1993. Amphibians and reptiles of longleaf pine communities. Pages 491 139-158 in S.M. Hrmann, editor. Procedings of the Talimbers Fire Ecology 492 Conferenc. No. 18, The Longleaf Pine Ecosystem: ecology, restoration and 493 managent. Tal Timbers Resrch Staion, Talhase, Florida, USA. 494 Hines, J. E. 2010. PRESENCE3-Software to estimte patch ocupancy and related parametrs. 495 USGS-PWRC. http:/.mbr-pwrc.usgs.gov/software/prese.html. 496 Irschik, D. J., B. Vanhooydonck, A. Herrel, and J. Meyers. 2005. Intraspecif correlations 497 among morphology, performanc and habita us within a green anole lizrd (Anolis 498 carolinensis) population. Biologicl Journal of the Linnean Socity 85:211-221. 499 Kilpatrik, E. S., T. A. Wldrop, J. D. Lanham, C. H. Greenberg, and T. H. Contreras. 2010. 500 Short-term efects of fuel reduction tretents on herpetofauna from the southeastern 501 Unitd Stas. Forest Scienc 56:122-130. 502 Landers, J. L., D. H. Vn Lar, W. D. Boyer 1995. The longleaf pine forest of the Southeast: 503 requiem or renaisnce? Journal of Forestry 93: 39-44. 504 Lindenmayer, D. B., J. T. Wood, C. MacGregor, D. R. Michael, R. B. Cunningham, M. Crane, 505 R. Montague-Drake, D. Brown, R. Muntz, and D. A. Driscoll. 2008. How predictable are 506 reptile responses to wildfre? Oikos 117:1086-1097. 507 Lit, A. R. 1999. Hrpetofaunal responses to longleaf pine sandhil restoration treatments on 508 Eglin Air Force Bas, Florida. M.S. thesi, University of Florida. 130 pp. 509 Lit, A. R., L. Provencher, G. W. Tanner, and R. Franz. 2001. Herpetofaunal responses to 510 restoration treatments of longlef pine sandhils in Florida. Restoration Ecology 9:462- 511 474. 512 128 Magurran, A. E. 2004. Measuring biological diversity. Blackwel Publishing, Malden, MA. 256 513 pp. 514 athews, C. E., C. E. Moorman, C. H. Greenberg, and T. A. Waldrop. 2010. Response of 515 reptiles and amphibins to repeatd fuel reduction tretments. Journal of Wildlf 516 Managent 74:1301-1310. 517 acKenzi, D. I., and W. L. Kendal. 2002. How should detecion probability be incorporated 518 into estimates of relative abundance? Ecology 83:2387-2393. 519 Mazeroll, M .J., L. L. Bailey, W. L. Kndal, J. A. Royle, S. J. Converse, and J. D. Nichols. 520 2007. Making great lps forward: acounting for detcability in herpetologial field 521 studies. Journal of Herpetology 41:672-689. 522 McCoy, E. D., and H. R. Mushinsky. 1999. Habit fragmentation and the abundances of 523 vertebrats in the Florida scrub. Ecology 80:2526-2538. 524 cCune, B. and M. J. Meford. 1999. Multivariate Analysis of Ecological Dt. Version 4.25. 525 Mj Software, Glneden Beach, Oregon, U.S.A. 526 cIver, J., A. Youngblood, and S. L. Stephens. 2009. The national Fire and Fire Surrogate 527 study: ecological consequencs of fuel reduction methods in seasonaly dry forest. 528 Ecologial Applit 19:283?284. 529 Means, D. B. 2006. Vertbrate faunal diversity of longleaf pine ecosystems. Pages 157-213 In: S. 530 Jose, E. Jokela, and D. L. Miler, editors. The lf pi et: ecology, 531 silviculture, and restoration. Springer, New York, USA. 532 Methrata, E. T., and J. S. Link. 2006. Seasonal variation in groundfish habita asociatons in the 533 Gulf of Maine-Georges Bank region. 326:245-256. 534 129 Michael, J. L., E. C. Webber, Jr., D. R. Bayne, J. B. Fischer, H. L. Gibbs, and W. C. Sesock. 535 1999. Hexazinone disipaton in forest ecosystems and ipacts on aquatic communities. 536 Canadian Journal of Forest Reearch 29:1170-1181. 537 Michael, D. R., R. B. Cunningham, and D. B. Lindenmayer. 2008. A forgotten habita? Granite 538 inslbergs conserve reptile diversity in fragmentd agricultural landscpes. Journal of 539 Applied Ecology 45:1742-1752. 540 Mielke, P. W., and K. J. Berry. 2001. Permutation methods: a distance function approach. 541 Springer-Verlag, New York, Nw York, USA. 542 itchel, R. J., J. K. Hirs, J. J. O?Brien, S. B. Jack, and R. T. Engstrom. 2006. Silviculture that 543 sustains: the nexus betwn silviculture, frequent prescribed fire, and conservation of 544 biodiversity in longleaf pine forest of the southeastern United Staes. Canadian Journal 545 of Forest Reearch 36:2723-2736. 546 Mushinsky, H. R. 1985. Fire and the Florida sandhil herpetofaunal community: with special 547 atention to responses of Cnemidophorus sexlineatus. Herpetologica 41:333-342. 548 yers, R. L. 1990. Scrub and high pine. Pages 150?193 in R. L. Myers and J. J. Ewel, editors. 549 Ecosystems of Florida. University of Central Forida Pres, Orlando, Florida, USA. 550 Noss, R. F. 1988. The longleaf pine landscpe of the Southeast: almost gone and almost 551 forgotten. Endangered Species Update 5:1-8. 552 Noss, R. F. 1989. Longleaf pine and wiregras: keystone components of an endangered 553 ecosystem. Ntural Ares Journal 9:234-235. 554 Outalt, K. W., and D. G. Brockway. 2010. Structure and compositon changes following 555 restoration treatments of longlef pine forest on the Gulf Coastal Pin of Alabam. 556 Forest Ecology and Managemnt 259:1615-1623. 557 130 Palmer, M. W. 1993. Putting things in even beter order: the advantages of canonical 558 correspondence analysis. Ecology 74:2215-2230. 559 Perison, D., J. Phelps, C. Pavel, R. Kelison. 1997. The efects of timber harvest in a South 560 Carolina blackwter bottomland. Forest Ecology and Manageent 90:171-185. 561 Perry, R. W., D. C. Rudolph, and R. E. Thil. 2009. Reptile and amphibian responses to 562 restoration of fire-maintned pi woodlands. Restoration Ecology 17:917-927. 563 Perkins, M. W., L. M. Conner, and M. B. Howze. 2008. The importance of hardwood trees in 564 the longleaf pine forest ecosystem for Sherman?s fox squirrels. Forest Ecology and 565 Managemnt 255:1618-1625. 566 Provencher, L., B. J. Herring, D. R. Gordon, H. L. Rodgers, K. E. M. Galey, G. W. Tanner, J. L. 567 Hardesty, and L. A. Brennan. 2001a. Efects of hardwood reduction tchniques on 568 Longlef Pine sandhil vegetaion in northwst Florida. Restoration Eology 9:13-27. 569 Provencher, L., B. J. Herring, D. R. Gordon, H. L. Rodgers, G. W. Tanner, J. L. Hardesty, L. A. 570 Brennan, and A. R. Lit. 2001b. Longleaf pine and oak responses to hardwood reduction 571 techniques in fire-suppresed sandhils in northwest Florida. Forest Ecology and 572 Management 148:63-77. 573 Provencher, L., N. M. Gobris, L. A. Brennan, D. R. Gordon, and J. L. Hardesty. 2002. Breeding 574 bird response to midstory hardwood reduction in Florida sandhil longleaf pine forest. 575 Journal of Wildlf Management 66:641-661. 576 Royle, J. A. and J. D. Nichols. 2003. Estiating abundance from repeated presenc-absenc data 577 or point counts. Eology 84:777-790. 578 Russel, K. R., D. H. Van Ler, D. C. Guynn, Jr. 1999. Prescribed fire efects on herpetofauna: 579 review and mnagemnt iplicatons. Wildlfe Society Bulltin 27:374-384. 580 131 Russel, K. R., T. B. Wigley, W. M. Baughman, H. Hanlin, and W. M. Ford. 2004. Responses 581 of southeastern amphibians and reptiles to forest mnagement: a review. In: Rauscher, 582 H.M., Johnsn, K.(Eds.), Southern Forest Scienc: Past, Presnt, and Future. Soutrn 583 Research Staion, Ashevile, NC, pp. 319?334. 584 Semlitsch, R. D., K. L. Brown, and J. P. Caldwel. 1981. Habit utilzation, seasonal activiy, 585 and population size structure of the southeastrn crowned snake Tantila coronata. 586 Herpetologicl 37:40-46. 587 Shindlr, B., and E. Toman. 2003. Fuel reduction straegis in forest communities: A 588 longitudinal alysis of public support. Journal of Fstry 101:8?16. 589 Sten, D. A., C. Guyer, and L. L. Smith. A case study of relative abundance in snakes. In 590 Measuring and Monitoring Biodiversity: Standard Methods for Reptils (R. W. 591 cDirmid, M. S. Foster, C. Guyer, J. W. Gibbons, and N. Chernoff, eds.), Smithsonian 592 Instiuton Pres: in pres. 593 Sten, D. A., A. E. Ral-McGe, S. M. Hermann, J. A. Stiles, S. H. Stiles, and C. Guyer. 2010. 594 Efects of forest managemnt on aphibins and reptils: generalist species obscure 595 trends among native forest aociates. Open Environmental Sciencs 4:24-30. 596 Sten, D. A., L. L. Sith, and M. A. Baily. 2010. Suggestd modifitons to terrestrial box 597 traps for snakes. Herpetological Review 41:320-321. 598 Stewart-Oten, A., W. W. Murdoch, and K. R. Parker. 1986. Environmental impact asemnt: 599 ??psudoreplicaton?? in time? Ecology 67: 929?940. 600 ter Braak, C. J. F. 1986. Canonical correspondence analysis: a new eigenvector technique for 601 multivariate direct gradient analysis. Ecology 67:1167-1179. 602 132 ter Braak, C. J. F, 1994. Canonical community ordination. Part I: Basic theory and linear 603 methods. Ecosienc 1:127-140. 604 ter Braak, C. J. F., and P. F. M. Verdonschot. 1995. Canonical correspondence analysis and 605 relted multivariate mthods in aquati ecology. Aquati Scies 57:255-289. 606 Todd, B. D., J. D. Wilson, C. T. Winne, and J. W. Gibbons. 2008. Aspet of the ecology of the 607 earth snakes (Virginia valeriae and V. striatula). Southeastern Naturalist 7:349-358. 608 Waldrop, T. A., D. L. Whit, S. M. Jones. 1992. Fire regimes for pine-graslnd communities in 609 the southeastern United Staes. Forest Ecology and Manageent 47:1095-1210. 610 an, M. T., R. G. Wts, and D. J. Moul. 1988. Evaluation of the acut toxicity to juvenile 611 pacif salmonids of hexazinone and its formulated products: Pronone 10G, Vlpar L, and 612 their crriers. Bulletin of Environmental Contminaton and Toxicology 41:609-616. 613 White, P. S., and J. L. Walker. 1997. Approxiting nature?s variation: selting and using 614 refnce informtion in restoration ecology. Restoration Ecology 5:338-349. 615 ilgers, D. J., and E. A. Horne. 2006. Efets of diferent burn regimes on tal gras prairie 616 herpetofaunal species diversity and community compositon in the Flint Hils, Kanss. 617 Journal of Herpetology 40:73-84. 618 Van Ler, D. H., W. D. Carroll, P. R. Kapeluck, and R. Johnson. 2005. History and restoration of 619 the longleaf pine-grasland ecosystem: iplicatons for species at risk. Forest Ecology 620 and Mnagemnt 211:150-165. 621 Yger, L. Y., C. D. Heis, D. M. Epperson, M. G. Hinderliter. 2007. Gopher tortoise reponse to 622 habita mnagemnt by prescribed burning. Journal of Wildlfe Managemnt 71:428-434. 623 Zedlr, J. B., and J. C. Calawy. 1999. Tracking wetland restoration: do mitgaion sites follow 624 desired trajectoris? Restoration Eology 7:69-73. 625 626 133 Figure 1. Relative proportion of species captured in treatment and referenc sites on Eglin Air 626 orce Bas, 1997-1998. Species captured ! 5 ties re not included in graph. Specis codes are 627 provided in Table 2. 628 Figure 2. Reltive proportion of species captured in treatment and referenc sites on Eglin Air 629 orce Bas, 2009-2010. Species captured ! 5 ties re not included in graph. Specis codes are 630 provided in Table 2. 631 Figure 3. Non-mtric diensional scaling of treatment and referenc sites for 1997-1998 and 632 2009-2010, Eglin Air Forc Base, Snt Ros and Okaloosa Countis, Florida. 1 = 1997-1998, 2 633 = 2009-2010. 634 Figure 4: Canonical correspondence biplot for reptiles captured in 1997-1998, Eglin Air Force 635 Base, Sant Ros and Okaloosa Countis, Florida. B = Burn, C = Control, H = Herbicide, M = 636 Mchanil and R = Referenc. 637 638 134 Table 1. Tree density within hardwood-removal sites and referenc sites, Sant Rosa and 638 Okaloosa Countis, Eglin Air Force Bas, Florida. One refre sit ws not included in 2009- 639 2010 summries. Al units are m 2 /ha (standard error). 640 1994 1998-1999 2009-2010 Pinus palustris midstory Burn 0.13 (0.05) 0.05 (0.02) 0.05 (0.02) Control 0.10 (0.02) 0.07 (0.01) 0.01 (0.01) Herbicide 0.09 (0.02) 0.04 (0.01) 0.28 (0.10) Mhanial 0.10 (0.02) 0.03 (0.01) 0.07 (0.02) Referenc 0.03 (0.01) 0.02 (0.01) 0.13 (0.06) Pinus palustris overstory Burn 12.78 (1.85) 12.01 (1.72) 12.93 (1.66) Control 7.88 (0.93) 8.71 (0.93) 10.09 (0.40) Herbicide 11.84 (2.35) 12.01 (2.41) 11.36 (1.50) Mhanial 12.15 (2.43) 11.14 (3.16) 11.79 (2.18) Referenc 16.15 (2.34) 16.65 (2.69) 18.12 (4.74) Quercus sp. midstory Burn 0.79 (0.16) 0.22 (0.11) 0.56 (0.21) Control 1.07 (0.13) 1.23 (0.19) 0.72 (0.24) Herbicide 0.56 (0.14) 0.02 (0.01) 0.14 (0.04) Mhanial 0.87 (0.08) 0.09 (0.07) 1.59 (0.33) Referenc 0.11 (0.03) 0.17 (0.13) 0.11 (0.11) 135 Quercus sp. overstory Burn 10.08 (2.45) 5.41 (2.79) 5.22 (1.65) Control 10.10 (1.34) 9.36 (1.97) 3.76 (1.19) Herbicide 9.08 (1.27) 0.40 (0.15) 0.04 (0.02) Mhanial 11.74 (1.73) 2.18 (1.22) 7.82 (6.78) Referenc 4.93 (1.93) 2.93 (0.33) 0.93 (0.64) 641 136 Table 2. Total cptures of reptiles by treatment and referenc sites on Eglin Air Force Bas, 642 1997-1998 and 2009-2010. Trapping efort within a year ireasd in 2009-2010 and one 643 refenc site was excluded from study (se Mthods). 644 Control Burn Hrbicide Mechanial Referenc Total Anolis carolinensis (ACAR) 1997-1998 18 20 1 1 10 50 2009-2010 5 3 2 3 14 Aspidoscelis sexlineatus (ASEX) 1997-1998 106 200 101 197 338 942 2009-2010 224 297 228 233 232 1214 Cemophora coccinea (CCOC) 1997-1998 3 1 1 6 1 12 2009-2010 6 3 3 4 3 19 Coluber constricor 1997-1998 0 1 3 1 0 5 2009-2010 0 1 3 Coluber flagelum 1997-1998 0 0 0 0 0 0 2009-2010 1 1 2 4 Diadophis punctatus 137 1997-1998 3 0 0 0 0 3 2009-2010 0 0 Gopherus polyphemus 1997-1998 0 0 0 0 0 0 2009-2010 1 1 Hetrodon platyrhinos 1997-1998 0 0 0 0 0 0 2009-2010 1 1 2 Lampropeltis elapsoides 1997-1998 0 0 1 1 0 2 2009-2010 1 0 0 1 Micrurus fulvius 1997-1998 0 0 1 0 0 1 2009-2010 0 0 Nerodia fasciata 1997-1998 0 1 1 0 1 3 2009-2010 0 0 1 0 1 Plestiodon egregius (PEGR) 1997-1998 7 8 2 2 4 23 2009-2010 3 5 1 4 17 Plestiodon laticeps 138 (PLAT) 1997-1998 22 6 11 10 3 52 2009-2010 8 14 7 4 4 37 Regina rigida 1997-1998 0 0 0 0 0 0 2009-2010 1 1 Sceloporus undulatus (SUND) 1997-1998 13 50 16 29 30 138 2009-2010 42 56 28 26 49 201 Scincela lateralis (SLAT) 1997-1998 29 22 10 18 15 94 2009-2010 10 9 4 3 8 34 Sistrurus milarius 1997-1998 2 1 0 1 0 4 2009-2010 0 1 1 5 Storeria occipitomaculata 1997-1998 1 0 0 1 0 2 2009-2010 0 1 Tantila coronata (TCOR) 139 1997-1998 128 55 89 111 49 432 2009-2010 15 15 23 19 15 87 Terapene carolina 1997-1998 0 0 0 0 0 0 2009-2010 1 1 2 Virginia valeriae (VAL) 1997-1998 4 0 2 2 4 12 2009-2010 0 1 1 0 0 2 645 646 647 648 649 650 651 140 Table 3. P-values aociated with multi-response permutation procedure on pairwse 652 comparisons of tretmnt and referenc sits (1997-1998 and 2009-2010). Bold indicats a 653 significant diferenc betwen groups (! = 0.10) 654 655 Burn Control Mechanial Herbicide Referenc 1997-1998 Burn X 0.01 0.008 0.01 0.034 Control 0.460.240.02 Mechanial X 0.3 0.09 Hrbiide X 0.02 2009-2010 Burn X 0.440.47 0.690.77 Control 0.530.77 0.19 Mechanial X 0.9 Hrbiide X 0.08 656 657 658 659 660 661 662 663 664 665 666 667 668 141 Table 4: Prcent indicator values for reptile species ignificantly asociated with a particular 669 tremnt on Eglin Air Force Bas, 1997-1998. Bold indites a significnt asociaton with a 670 particular tretent. 671 Burn Control Mechanial Herbicide Referenc P-value Aspidoscelis sxlineatus 21 11 21 11 36 0.007 Diadophis punctat 0 75 0 0 0 0.025 Sceloporus undulatus 36 9 21 12 22 0.015 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 142 688 689 Figure 1. 690 691 692 693 694 695 696 697 698 699 700 !"# $!"# %!"# &!"# '!"# (!"# )!"# *!"# +!"# ,!"# $!!"# -./01.2# 341/# 5617898:6# ;69<=/89=2# >6?616/96# @ 6 1 9 6 / 0 # . ? # A = BC 2 6 # DDEF# G-H># AIJK# AFEG# @FEG# @LM># EALN# --H-# E-E># 143 701 702 Figure 2. 703 704 705 706 707 708 709 710 711 712 713 714 !"# $!"# %!"# &!"# '!"# (!"# )!"# *!"# +!"# ,!"# $!!"# -./01.2# 341/# 5617898:6# ;69<=/89=2# >6?616/96# @ 6 1 9 6 / 0 # . ? # A = BC 2 6 # D-E># AFGH# AIJD# @IJD# @KL># JAKM# --E-# J-J># 144 715 716 717 Figure 3. 718 719 720 721 722 723 724 725 726 145 727 728 Figure 4. 729 730 731 732 733 734 735 736 146 Chapter 5 737 Six-lined Racerunner, Aspidoscelis sxlineatus, Population Size and Survivorship: Long-term 738 Efets of Fire Surrogates and Prescribed Burning 739 740 Abstract. Fire surrogates and prescribed burning are management tools for the 741 restoration of fire-suppresd and hardwood-invaded longlef pine forest. To evaluate how 742 popultions of a common squaate, the six-lned racrunner, Aspidoscelis sxlineatus, 743 responded to forest retoration, w conducted a mark-reapture study of populations in formerly 744 fire-suppresed longleaf pine forest exposd to prescribed fire or fire surrogates (i.e., 745 mechanial or herbicide-fciltaed hardwood removal) as wel as in untreatd control sites and 746 refnce sites. After initil tretmnt, al sites wre exposd to over a decde of prescribed 747 burning with an average interval of approximatly two years. Our population analysis (POPAN) 748 produced counter-intuitive result, which we atribute to uncertinty resulting from low sample 749 siz and detcion probabilites. Howver, the man number of marked adults and juvenils at 750 site treated with prescribed burning and sites treted with mechanil removal of hardwoods 751 was comparabl tohe mean number of marked adults and juvenils at referenc sites over 752 relatively short-tim scls. Over the long-term, mean numbers of marked individuals observed 753 t al tretents wa not diferent from t mr ofarked indivils observed at 754 refenc sits. W conclude that prescribed burning over long time scals i an efctive 755 approach for restoring A. sexlineatus populations in fire-suppresd longlef pine sandhils. 756 Key Words: Before-After-Control-Impact, Longleaf Pine, Mark-recapture, Pinus palustris, 757 POAN, Prescribed Fire, Reptile, Survival 758 759 147 INTRODUCTION 760 761 Longleaf pine forest, which once spanned throughout the coastal plin of the 762 southeastrn Unitd Staes (Ware 1993), contain diverse vertbrate asmblges (Means 2006). 763 Tse forest hitoricly had a sparse cnopy of pines with a divers herbaceous understory 764 maintned by frequent wildfres that occurred every 1-10 years (Myers 1990). However, due to 765 fire suppresion, hardwood trees have beome established in the midstory of many formr 766 longleaf pine-grasland habitas. This ha functioned to reduce habita quality for mny species 767 asocitd with the ancestral condition (Mitchel et al. 2006). 768 Interest in restoring ecological communities ha increased a natural habitas are lost 769 outright (Hobbs and Harris 2001) or degraded by disruption of natl disturbance regimes (.g., 770 Nowacki and Abrams 2008). Restoration methods for fire-suppresed longlaf pine forest 771 include diret reoval of hardwoods via mchanial mens or applicaton of herbicides. 772 However, due to the unique efects of fire, burning is likely an esentil component of any 773 succsful longleaf pine forest retoration efort; if mechanial removal or herbicides are 774 employed, they should be used together with eventual reintroduction of fire (Brockway et al. 775 2005). 776 The short and long-term efects of these difrent measure on the structure of longleaf 777 pine forest can be readily apparent; howver, t efects of habit change on wildlf are not 778 wel known (e.g., Grdner et al. 2007). Previous resarch has suggested applicton of herbicides 779 or mchanial hardwood removal, which some consider to be fire surrogats, may have varied 780 short-term efects on wildlfe asblages (.g., Lit e al. 2001) but long-term use of prescribed 781 148 burning may be necesary to replicate refnce conditions (Chapters 3, 4), as ha been 782 demonstraed among vegetion communitis (Outcalt and Brockway 2010). 783 Respons of reptiles to habita mnageent and restoration is generaly studied at the 784 asemblage level (Grenberg et al. 1994, 2008, Russl et al. 2002, Renken et al. 2004, Lynaud 785 nd Bucher 2005, Mathews 2010), including within longleaf pine forest (e.g., Lit 2001, Smith 786 and Risler 2010, Chapter 4). However, without careful atntion to what constiutes a target 787 semblage, general trends may be obscured becuse reptiles are diverse group (Barret nd 788 Guyer 2009) and habita asocitons of individual specis may difr (Stn et al. 2010a). In 789 addition, simlarity is generaly quantifed by comparison of raw counts an approximation of 790 bundance, overlooking variation in detcion probabilites (MacKenzi et al. 2006, Mzeroll et 791 al. 2007). Consquently, asemblge-level study may obscure trends among species highly 792 sensitve to forest managent (.g., Mas et al. 2009, Chapter 3). 793 Smal squates arebundant in longleaf pine forest and play important roles in the 794 ecosyste (Mens 2006); in addition, they my respond to habit restoration in reltively short 795 tim sals (.g., Trainor and Woinarski 1994, Bateman et al. 2008, Letink et al. 2010). 796 Therefore, focus on a squamate my be a useful proxy for wildlf community response to 797 restoration. Six-lned racrunners, Aspidosclis sxlineatus, are widely distributed across North 798 America (Fitch 1958, Hardy 1962), but prefer open and xeric habitas, particularly longleaf pine 799 forest, within the southestern United Stas (Guyer and Bailey 1993). Aspidoselis sxliatus 800 has been found to respond positvely to frequent burning (Mushinsky 1985, Chaptr 4). 801 Therefore, A. sexlineatus may be an appropriate focal species for monitoring the succes of 802 restoration eforts in longlef pine forest (Blk et al. 2001). 803 149 Due to long generation times or delayed responses to vegetaion changes, long-term 804 studies may be necesary to charactriz wildlfe res to habit c (Congdon et al. 805 1993, Brooks t al. 1999, Helm et al. 2006). In addition, imediat response to restoration may 806 not be reflective of long-trm paterns (Chapters 3, 4). Within ts study, w usd a randomized 807 block design and quantifed A. sxlineatus population sizes, while acounting for variation in 808 capture probability, to detrmi how the species reponded to eologial restoration over a 15- 809 yer period. Since acurate abundance estimats may be dificult to generate for squamates that 810 have low detion probabilits (Stn 2010, Sten et al. 2011), we also compared the mn 811 number of marked adults and the mean number of marked juvenils within each treatment to 812 refenc sites. If the structural endpoints wesured here (i.e., popultion estits, number 813 of marked adults and juveniles) wre indistinguishable betwen treatment and referenc sites, w 814 infer mnagement objectives wre met; otherwis, w suggest a tretnt was inefctive. 815 816 METHODS 817 818 Study Site 819 820 This study took place in fre-suppresed longleaf pine sandhils on Eglin Air Force Bas, 821 Sant Rosa and Okaloos Countis, Florida, U.S. A randomized block design was usd to asign 822 hardwood removal method treatments to 24 81-ha sites (six blks, Provencher et al. 2001). With 823 the exception of six sts that experincd a single burn betwen 1977 and 1989, al sites had 824 ben fire-suppresed since at last 1973 (when records began). Mthods of hardwood removal 825 included burning (Burn), herbicide applicaton (Herbiide), or feling-girdling (Mechanial) and 826 150 a control, which experiencd no hardwood removal. Six 81 ha referenc sites were also 827 designated. Proveher et al. (2001a) describe critea for selction of referenc sits; they were 828 selctd as representative of a natural longleaf pine forest, bad on forest sructure, disturbanc 829 regim, and the presnce of characteristic wildlf, and were the target condition of restoration 830 eforts. 831 832 Treatmnt Applicaton 833 834 Hardwood removal occurred in 1995. Burn sites were burned April-June. Herbicide was 835 applied in erly May, and mehanial hardwood removalas conducted betwen J and 836 November. Herbicide and Mcnil sites were subjectd to a prescribed burn in 1997. After 837 treaent appliaton, al sites reived comparabl management, whih included prescribed fire 838 on an pproximtely 2-year rottion but no additionalchanial hardwood removal or 839 applicton of herbicides. 840 841 Vegetation Data Collection 842 843 Vegetaion data were collectd in treatment sites and referenc sites in 1998. Dat for 844 treamnt sites were collctd again in 2009 and refrence sits wre resampld in 2010. 845 Spling for vegetaion ocurred along transects in each sit (as decribed in Provencher et al. 846 2001a, b). Vetion ws characterizd by cover clse (1-5%, 5-25%, 25-50%, 50-75%, 75- 847 95%, 95-100%) and convertd to midpoints. Thes midpoints were usd to generate man 848 151 percent cover for al sites. Oak midstory (trees < 16 cm, diaetr at breast height) basal are 849 (m 2 /ha) for each site ws alo collectd. 850 851 Aspidoscelis sxlineatus Trapping 852 853 Drift fenc arrays (Campbel and Christman 1982) were placed in four treatment blocks 854 (one drift f in each of 16 treatent sit) and four rfrenc sits. However, one refrence 855 site was treted with herbiide following the initial smpling, which fl outsid of the study 856 design, thus it was excluded from analysis. Traps were placed at the centr of each site. 857 Aspidoscelis sxlineatus were trapped in drift fnc and pitfl arrays as deribed in Lit (1999) 858 in 1997 and 1998 (Lit e al. 2001). Each array contained 30 m of 50 cm tal-galvanized stl 859 flashing and 16 19-L pitfal traps. Arrays were plced in the sae loctions and liards trapped 860 from May-Septmber 2009 and May-August 2010. In 2009 and 2010, we added box traps 861 (Burgdorf et al. 2005, Sten et al. 2010b) at the centr of arrays but usd the same number of 862 pitfal traps per array. 863 Aspidoscelis sxlineatus !500 mm snout-vent length were considered juveniles (slightly 864 smaler than the siz of reproductively active feals, a reportd for Arkansas animals, Trauth 865 1983). Al otr liards were charaterizd as ml or femal based on a single scondary sexual 866 charateristic, .e., blue coloration on mals (Conant and Collins 1998). Lizards wre 867 individualy marked by toe clip and relesd (those that escaped before receiving a mark were 868 not included in analysis). Clipping toes may influenc repture probability or survivorship of 869 some vertbrates (Murray and Fuller 2000), and can interfre with normal behavior of climbing 870 lizards (.g., Bloch and Irschik 2004). However, toe-clipping did not afect running speed of a 871 152 ground-dweling skink (Borges-Land?ez and Shine 2003) or of A. sexlineatus elsewhere (Dodd 872 1993). Therefore, we asumed this mthod of marking did not influenc the paramtrs of 873 interest wihin ts study. 874 875 Analysi 876 877 We usd the R 2.9.1 (R Development Core Team 2009) package RMark 1.9.6 (Lake and 878 Rexstand 2008) to build POPAN models (Schwrz and Arnason 1996) in Program MARK 6.0 879 (White and Burnham 1999) to estimateonthly apparent survival (phi), capture probability (p), 880 probability of entry io the popultion (pent) and population size (N) separatly for 1997-1998 881 and 2009-2010. The POPAN model asumes a superpopultion from which trapped individuals 882 re smpled, which may be appropriat when not al individuals arevailable for capture during 883 every survey (Wilms et al. 2011). POPAN is modified from the Jolly-Sber model (Pollock et 884 al. 1990, Schwarz and Arnason 1996). Jolly-Seber models are commonly used in wildlfe 885 popultion analyses and the POPAN paramtrization may be wel-suitd for lizards (.g., 886 Wiederhecker et al. 2003, Graceva et al. 2008). Only a subst of traps were open in May 1997, 887 and sinc POPAN models cannot handle unequal sampling efort among sits, we removed May 888 1997 from consideration in analysis. 889 We included gras/edge and bare ground coverages a covariates when building models. 890 Both of thes vegetaive charactristic may represent important structural fatures for A. 891 sexlineatus (e.g., Mushinsky 1985, Lit e al. 2001, Lindenmyer et al. 2008). We also included 892 midstory oak total basl area in the models because reduction of this component of the habita 893 was the objective of the experimental tretmnts. Although additional vegetaion data were 894 153 colletd, previous analyses did not identify a correlation betwen A. sexlineatus capture rates 895 and vegetaive charactristic (Chapter 4), so they were excluded from model building. 896 Although we considered phi and N paramtrs of primary interest, acurate estimates of 897 phi and N depend on appropriate specifation of p and pent. Therefore, we developed an a 898 prior set of models to explore the best mens of modeling p and pent. For both p and pent, we 899 considered the potential influence of t habita variables decribed above (bare ground, 900 gras/edge, and midstory oak), treatmnt, and sex/ge. For phi, the candidate model set 901 considered models including tretent and sx/a efects. We specifd N only as varying by 902 site. W calcultd the standard error of N by summing variancs from gross initial population 903 stimts (the popultion of animal present during the first trapping sesion), adding these values 904 to overal net superpopulation estits, and converting to standard errors for each N estimate. 905 We usd a sequentil framework to generate estimtes for p and pent and ranked models 906 with Akaike?s Information Critria corrected for sal spl siz (AICc, Burnham and 907 Anderson 2002). When generating models for p or pent, other parametrs were held constant in 908 their most-parametrizd iteraion in the candidate model set. T best models etimting p nd 909 pent were then usd in the model set to estimt phi (Tabls 1 and 2). Within any model set, 910 models that failed to estimateultipl paraetrs (likely due largely to overparametrization) 911 were excluded from the st. W considered phi estimates a significntly difrent if 95% 912 confidence intrvals did not overlap. We atptd to as fit using the goodnes of fit test in 913 U-CARE 2.3.1 (Choquet e al. 2009), but lacked sufficent data to estimate ! (varianc inflation 914 factor, a term which indicates overdisperion). As a reult, w insted investigaed the 915 robustnes of the model sts by manipulating ! from 1.0 (no dispersion) to 3.0 (extrem 916 dispersion; Devris et al. 2003). 917 154 We usd a before-after control-impact sudy design and analysis of variance (Stwart- 918 Oaten et al. 1986) to compare the 1) gross population size, 2) number of mrked adults and 3) 919 number of marked juveniles betwen treaments and over tim with SAS 9.2 (SS Instiute, Inc. 920 Cary, NC). Comparisons of a prior intrest wrehetr population sizes and number of 921 marked adults and juveniles within treatment sitesere indistinguishabl from those of referenc 922 site for both study periods and whetr thes paratrs within treatments changed over tim. 923 W st our alpha level to 0.05. 924 e asumd that A. sexlineatus population sizes and survivorship at treatment sits prior 925 to hardwood removal in 1995 wre comparable to thos w observed on control sits in 1997- 926 1998. We considered A. sexlineatus populations to be restored if population size estimates or the 927 number of marked adults and juveniles at given treatment did not difer from thos on 928 refenc sites. 929 930 RESULT 931 932 We had 712 captures of 584 individual A. sexlineatus in 1997-1998 and 1075 captures of 933 846 individuals in 2009-2010. Among ivils captured i-1998, the proportion of 934 females ranged from 0.25 in Control sites and 0.44 in Burn sites and t proporti juveniles 935 ranged from 0.15 in Control sites and 0.29 in Burn sites (Tabl 3). The proportion of femal 936 lizards cptured in 2009-2010 ranged from 0.45 in Control sites to 0.54 in Burn sites. and the 937 proportion of juveniles ranged f 0.15 in Herbicide sites to 0.26 in Mechanial sit (Table 3). 938 The model that best explained variation in capture probability (p) for the 1997-1998 939 period was one tt alowed this varible to vary by bare ground. For the 2009-2010 period, the 940 155 best supported specifation of p alowed p to vary by gras/edge, bare ground, and oak 941 midstory (Tabls 1 and 2). Estimatd cpture probability was 0.14 (standard error = 0.02) in 942 1997-1998 and 0.21 (standard error 0.02) in 2009-2010. The best model explaining variation in 943 probability of entry into the population (pent) for the 1997-1998 period alowed this varible to 944 vary by sex and age and treatment. For the 2009-2010 period, the best supportd specifation of 945 pent alowd pent to vary by sx and age of t individual animal (Tbles 1 and 2). The best 946 models explaining survivorship for both study periods alowed this paratr to vary by 947 treaent and sex (Tables 1 and 2). 948 The best-upportd model (phi(sex/age+treatment)) in these sts wa unchanged for ! 949 values up to 1.5 for the 2009-2010 dat st and up to 1.25 for t 1997-1998 dat set. However, 950 this model remained within 4 AICc units of the adjusted best-upported model (phi(sx/age)) 951 until ! values of 3.0 for the 2009-2010 set, and of 2.75 for the 1997-1998 st. This indictes 952 some snsitviy to changes in !, although without an estimate of the actual ! value for thes sts, 953 w are unable to say the degree to which (if any) overdispersion atly ocurred. 954 With regard to gross population estimates, there was no significant ieraction betwen 955 treament and time (F 4,1 = 0.76, P = 0.56). Mn popultion size at referenc sit (76.89) was 956 significntly salr than Burn sites (167.63, P = 0.04). No difres wre detecd betwen 957 refenc sites and Control (102.01, P = 0.56), Herbicide (116.67, P = 0.36), or Mhanical 958 (122.62, P = 0.29) sites in 1997-1998 (Figure 1). In 2009-2010, mean population size at 959 refenc sites (132.38) was not significantly diferent from Burn (132.88, P = 0.99), Control 960 (94.66, P = 0.38), Herbicide (77.23, P = 0.21) or Mchanial sites (122.95, P = 0.65; Figure 1). 961 We did not obsrve signifiant diferencs in survivorship betwn treatments or age/sex clase 962 (Tabl 4). 963 156 There was no significant ieraction betwen treatment and time (F 4,1 = 1.45, P = 0.24) 964 for t number of mrked adults. In 1997-1998, the mn number of marked adults on referenc 965 site (38) was not significantly diferent than on Burn (23.25, P = 0.06) or Mechanial (23.75, P 966 = 0.06) but lrger than on Control (13, P = 0.002) and Herbicide (13.75, P = 0.003). In 2009- 967 2010, the number of marked adults on referenc sits (37) did not difer from those on Burn 968 (39.25, P = 0.76), Control (29.75, P = 0.34), Hrbicide (32.5, P = 0.55) or Mchanial sites 969 (27.75, P = 0.22; Figure 2). 970 With regards to the number of marked juveniles, there was no significant ieraction 971 betwen treatment and time (F 4,1 = 0.89, P = 0.49). In 1997-1998, the men number of marked 972 juvenils on refrenc sits (10.33) was not significantly diferent than on Burn (9.5, P = 0.80) or 973 Mechanial (5, P = 0.12) but higher than on Control (2.25, P = 0.02) and Herbicide (3.5, P = 974 0.046). In 2009-2010, the mean number of marked juveniles on referencs (10) was not 975 significantly diferent than Burn (7.25, P = 0.41), Control (5.75, P = 0.21), Herbicide (5.75, P = 976 0.21), or Mhanical (10, P = 1.0; Figure 3). 977 In summry, long-term prescribed burning did not interact wih a specif hardwood 978 removal treatent to result in diferent gross population estimats or in the number of marked 979 adults or juvenils, rather, burning afcted A. sexlineatus silrly for al treatments. In 1997- 980 1998, the mean gross population siz at Burn sits wa lrger than at referenc sits. In 2009- 981 2010, tn grosstion sie atl treatments comparabl to that of referenc sites. 982 The mean number of marked adults and juvenils at Burn and Mechanial sites wa 983 indistinguishable from that of referenc sites in 1997-1998 and al tretmntsre 984 istisbl f referencs in 2009-2010. 985 157 The long-term efects of hardwood removal on vegetaion structure varied by treatment 986 (Table 5). Oak densits dereased in al treatent sites following initial tretmnt and reined 987 reltively high in Controls. Howver, al tretmnts experiencd gradual increase in midstory 988 oak density, a trend most pronounced in Mechanial sites. 989 990 DISCUSION 991 992 Aspidoscelis sxlineatus isan indicator of longleaf pine forest in referenc condition 993 (Chapter 4) and this species my ply n importnt rol in the ecosystem. Our rsult suggest 994 efctive restoration of A. sxlineatus populations may be ahieved following restoration of fire- 995 suppresed longleaf pine sandhils. Our findings based on relatily traditional measure of 996 abundanc (i.e., the number of marked adults and juvenils) suggest prescribed burning resulted 997 in efective restoration of A. sexlineatus populations on relatively short-time sals (a did 998 mhanial removal of hardwoods followed by prescribed fire). Or the long-trm, prescribed 999 burning in al treatents reulted in numbers of animals omparable to t number of animals 1000 observed on refrnc sits. In this sens, our findings corroborat multi-axa, aseblge-level 1001 analyss on the same study site indicating prescribed burning is an efective, and perhaps 1002 necesary, method of restoring fire-suppresed longleaf pine sandhils for wildlfe (Chapters 3, 1003 4). Thes findings also corroborate sudis conductd elswhere on the specis? repons to 1004 habita restoration (e.g., Mushinsky 1985, Greenberg et al. 1994). Although al treatments 1005 eventualy resulted in numbers of A. sexlineatus indistinguishable from referenc conditions, 1006 plots treated with prescrid burning alone and those treated with mchanial removal of 1007 hardwoods quickly ahieved this reult (! 4 years). Due to the added cost of mechanial 1008 158 hardwood removal, we recommend reintroducing prescribed burning to fire-suppresed longleaf 1009 pine sandhils to restore A. sxlineatus populations, as ha been recommended elswhere for the 1010 entire reptile asmblage (Chapter 4). 1011 Abundance values alone may not be appropriate as comprehensive indices of how 1012 populations respond to habit change (e.g., Todd and Rothermel 2006). For exampl, in many 1013 cse the number of individuals required to constiute a minimum viable population is unknown 1014 and abundance values may not reflect the efective population siz; this limts the use of these 1015 values when quantifying wildlf respons to habita restoration (Salwood 2001). Dnsity also 1016 cannot be asumed to be positvely related tbit quality; mesurement of population 1017 dynamics i likely more informati (Vn Horne 1983). Howver, although we do not know the 1018 number of individuals required to represent a minimum viable population in A. sxlineatus, we 1019 asume the number of individuals obsrved at referenc sits are representtive of an ideal or 1020 trget condition. 1021 Previous research identifying changes in A. sexlineatus abundance in relation to 1022 prescribed fire fquency suggested tt increas in abundance wre atributble primarily to 1023 imigration (Mushinsky 1985). Our study sites wre relatively large (81-ha) and our traps were 1024 located in the centr of each site, suggesting imigration is unlikely to be the primary 1025 mchanis resulting in the trends w observed. Given that we obsrved as mny juveniles in 1026 Burn and Mechanial plots ae did in refrence sits in 1997-1998, we suggest that relatively 1027 high numbers of A. sexlineatus caught in thes areas i due largely to either higher rates of 1028 succesful reproduction or incresed fcundity. Aspidoscelis sxlineatus mature relatively 1029 quikly (i.e., ~ one year of age; Clark 1976); therefore an increas in the number of succesful 1030 reproductive ents my quickly increse t number of sexualy mture adults. 1031 159 Simlarities and diferencs in our gross population estimates of A. sexlineatus among 1032 site are dificult to intrpret. In 1997-1998, we did not detec a difrenc in gross population 1033 siz and survivorship betwen Controls, which had ben fire-suppresed for decades, and 1034 refenc sites, which wre fi-maintned and representd the ancstral condition. Thus, we are 1035 unabl to make inferencs regarding how gross population estimates of A. sexlineatus changed in 1036 response to forest management; our result suggest fire-suppresion of longlaf pi forest i 1037 not to the detriment of A. sxlineatus populations. 1038 Our gross population estimates of A. sexlineatus herein difered from estimates preentd 1039 in most previous squamte sudis in that ours wre derived from a relatively rigorous mrk- 1040 recapture analysis that incorporated hetrogeneity in detecion probability. Integration of 1041 deteion probability may elucidat biological patrns that wre not otherwis apparent 1042 (MacKnzie et al. 2006; Mzeroll et al. 2007) and it is posible that A. sexlineatus are more 1043 resilent to fire-suppresion than previously indicated. In general, squamats are thought to 1044 maintn relatively stable populations over tim (Shoer 1985), and our population estimates 1045 re consistent with this trend. However, mark-recapture analyses may fre poorly at esiting 1046 population parametrs when capture probabilites are < 0.30 (Whit e al. 1982), and we 1047 estimted cpture probabilites of 0.14 in 1997-1998 and 0.21 in 2009-2010. In addition, w 1048 recorded a relatively smal number of individuals (Table 6), and this may increase bis and 1049 uncertainty of estites derived from mark-recpture studis. 1050 As a reult of smal population sizes and low capture probabilites for A. sexlineatus, 1051 model-based etitors my be a poor mthod for estimting population siz (Mnkens and 1052 Anderson 1988). For example, although only one anial ws cptured in one of our Control sites 1053 in 1997-1998, likely indicating a relatively poor-quality habita, we estimated there were nearly 1054 160 58 individuals preent in the population (Table 6). Thus, we conclude that alhough it is 1055 important in principl to incorporate hetrogeneity in detcion probabilites when quantifying 1056 abundances, low detecion probabilits confound a researcher?s ability to derive reasonable 1057 estimts, a has ben obsrved among other terrestril squamtes (i.e., snakes, Sten 2010, 1058 Stn et al. 2011). Future eforts to derive estimats of sat popultion siz basd on mark- 1059 recapture techniques should include multiple trapping arrays within a site to achieve capture 1060 probabilits high enough to facilte asociatd anales. 1061 Habit restoration my not be sufficent to recover a population that is already in decline 1062 (Schrott e al. 2005). Presumably A. sexlineatus populations cn persist a reltively low levels 1063 even in poor-quality habits, such as thos that ypify fire-suppresed longleaf pine sandhils 1064 (i.e., our control sites in 1997-1998). We therefore suggest the specis i unlikely to be extirpated 1065 in longleaf pine sandhils following invasion of hardwood trees. W are unable to determine if 1066 the popultions we smpled wre supplemntd by emigration following treamnt and by 1067 extnsion, whetr it is necsary to consider the landscpe matrix and neighboring population 1068 deites in future restoration eforts; however, this i an importnt consideration in determining 1069 how populations of smal squaates repond to habita restoration (Mushinsky 1985). 1070 Mushinsky (1985) described increased abundance of A. sexlineat in frequently-burned 1071 habitas and Greenberg et al. (1994) notd a higher a of the species in habitas that were 1072 disturbed in a manner that mimcked some efcts of fire, as compared to mature pine forest tt 1073 were infquently burned. On our study sit, the speies wa previously identifed as an 1074 important driver of asemblage-level change on multipl time-scal in respons to prescribed 1075 fire (Lit e al. 2001, Chaptr 4) and an indicator of longlaf pine forest in referenc condition 1076 (Chapter 4). Herein, our data suggest A. sexlineatus popultions may becom indistinguishable 1077 161 from those of sites in referenc condition through applicaton of prescribed burning. Thus, we 1078 conclude prescribed burning is an efective straegy for restorati A. sexlineatus populations 1079 in fre-suppresed longleaf pine sandhils. 1080 1081 ACKNOWLEDGMNTS 1082 1083 Funding was provided by the Straegic Environmental Research and Development 1084 Program (SERDP) roject Number: SI-1696. S. Pokswinski, K. Hirs, B. Wilams provided 1085 logistic help. D. Simpson, M. Baragona, M. Cent, and M. Betzhold provided asitnce in the 1086 field and with data mnageent. C. Guyer, L. Smith, M. Conner, J. Grand, N. Chadwick, and C. 1087 Anderson reviewd erlir drafts. G. Morris and R. Katz provided helpful asitance with 1088 Program MARK. E. P. Cox and Auburn University librarins provid asit obtining 1089 refencs. 1090 1091 LITERATURE CITED 1092 1093 Balinger, R. E. 1976. Evolution of life history straegis: implicatons of recruitment in a lizard 1094 population following density manipulations. Southwestrn Nturalist 21:203-208. 1095 Barret, K., and C. Guyer. 2008. Diferentil responses of amphibians and reptiles in riparian and 1096 stream habitas to land us disturbances in wstrn Georgi, USA. Biological 1097 Conservation 141:2290-2300. 1098 162 Bateman, H. L., A. Chung-MacCoubrey, and H. L. Snel. 2008. Impact of non-native plant 1099 reoval on lizards in riparin habitas in the southwestrn United Staes. Retoration 1100 Ecology 16:180-190. 1101 Block, W. M., A. B. Franklin, J. P. Ward, Jr., J. L. Ganey, G. C. White. 2001. Design and 1102 implentation of monitoring studies to evaluate the succs of ecological restoration on 1103 wildlf. Restoration Ecology 9:293-303. 1104 Bloch, N., and D. J. Irschik. 2004. Toe-clipping dramaticly reduces clinging performance in a 1105 pad-bearing lizard (Anolis carolinensis). Journal of Herpetology 37:293-298. 1106 Borges-Lnd?ez, P. A., and R. Shi. 2003. Influence of toe-clipping on running speed in 1107 Eulamprus quoyi, an Australian scincd lizard. Journal of Herpetology 37:592-595. 1108 Brockway, D. G., K. W. Outcalt, D. Tomzk, E. E. Johnson. 2005. Restoration of longleaf pine 1109 ecosystems. Gen. Th. Rep. SRS-83. U.S. Department of Agricul- ture, Forest Srvic, 1110 Southern Resarch Staion, Ashevile, NC. 1111 Brooks, T. M., S. L. Pim, and J. O. Oyugi. 1999. Time lag betwen deforestaion and bird 1112 extincton in tropical forest fragments. Consrvation Biology 13:1140-1150. 1113 Burgdorf, S. J., D. C. Rudolph, R. N. Conner, D. Saenz, and R. R. Schaefr. 2005. A succesful 1114 trap design for capturing large terrestril snakes. Herpetologial Reviw 36:421-424. 1115 Burnham, K.P., and D. R. Anderson. 2002. Model section and multi-model inferenc, second 1116 edition. Springer-Verlag, New York, Nw York, USA. 1117 Campbel, H. W., and S. P. Christman. 1982. Field techniques for herpetofaunal community 1118 analysis. In N. J. Scott, Jr. (ed.), Herpetologial Communities, pp. 193-200. U.S. Fish and 1119 Wildlfe Srvie, Wildlfe Resarch Report 13. 1120 163 Choquet, R., J. D. Lebreton, O. Gimenez, A. M. Reboulet, and R. Pradel. 2009. U-CARE: 1121 Utiles for performing goodnes of fit test and manipulating CApture-REcapture data. 1122 Ecography. 32: 1071-1074 1123 Clark, D. R., Jr. 1976. Ecological observations on a Texas population of six-lned racerunners, 1124 Cnemidophorus sexlineatus (Reptila, Lcrtila, Tidae). Journal of Herpetology 1125 10:133-138. 1126 Conant, R., and J. T. Collins. 1998. Reptiles and amphibians: estern/cntral North America. 1127 Houghton Mifln Company, Nw York, USA. 616 pp. 1128 Congdon, J. D., A. E. Dunham, and R. C. van Loben Sels. 1994. Demographics of common 1129 snapping turtles (Chelydra serpentina): implicatons for consrvation and manageent of 1130 long-lived organism. Arican Zoologist 34:397-408. 1131 Conner, L. M., S. B. Castleberry, and A. M. Derrick. 2011. Efects of mesopredators and 1132 prescribed fire on hispid cotton rat survival and caus-speifortality. Journal of 1133 Wildlfe Management 75:938-944. 1134 Cuelar, O. 1993. On competion and natural history of coexisting parthenogenetic and bisexual 1135 whiptail lzrds pp. 345-370 In J. Wright and L. J. Vit (eds.), Biology of Whiptail 1136 Lizrds (Genus Cnemidophorus). Oklahom Museum of Natural History, Normn, 1137 Oklahom, USA. 417 pp. 1138 Devris, J. H., J. J. Cita, M. S. Lindberg, D. W. Howertr, and M. G. Anderson. 2003. Breeding- 1139 seaon survival of mlard femals in the prairi pothole region of Canada. Journal of 1140 Wildlfe Management 67:551?563. 1141 Dodd, C. K., Jr. 1993. The efcts of toeclipping on sprint performance of the lizard 1142 Cnemidophorus sexlineatus. Journal of Herpetology 27:209-213. 1143 164 Fitch, H. S. 1958. Natural history of the six-lned racerunner (Cnemidophorus sexlineatus). 1144 University of Knsas Publicatons Museum of Natural History 11:11-62. 1145 Gardner, T. A., J. Barlow, and C. A. Peres. 2007. Pradox, presumption and pitfals in 1146 conservation biology: the importanc of habita change for amphibins and reptiles. 1147 Biological Conservation 138:166-179. 1148 Graceva, G., P. Bombi, L. Luiseli, and M. A. Bologna. 2008. Do demographic aspect of 1149 neighbouring lizard populations difer? A case study with the comon wal lizrd, 1150 Podarcis muralis. Amphibi-Reptila 29:443-448. 1151 Greenberg, C. H., D. G. Neary, and L. D. Hrris. 1994. Efects of high-intensity wildfre and 1152 silvicultural tretmnts on reptile communities in sand-pine scrub. Consrvation Biology 1153 4:1047-1057. 1154 Greenberg, C. H., and T. A. Waldrop. 2008. Short-term responses of reptiles and amphibians to 1155 prescribed fire and mechanil fuel reduction in a southern Appalachin upland 1156 hardwood forest. Forest Eology and Management 255:2883-2893. 1157 Hardy, D. F. 1962. Ecology and behavior of the six-lned racrunner, Cnemidophorus 1158 sexlineatus. The University of Kanss Scienc Bulletin 43:3-73. 1159 Helm, A., I. Hanski, and M. P?rtel. 2006. Slow respons of plant species rihnes to habita loss 1160 and fragmentation. Ecology Lters 9:72-77. 1161 Hobbs, R. J., and J. A. Hrris. 2001. Restoration ecology: repairing the Earth?s ecosystems in the 1162 new milennium. Restoration Ecology 9:239-246. 1163 Lake, J. & Rexstad, E. (2008) RMark ? an alternative approach to building linear models in 1164 MARK. Program MARK: ?A Gentl Introduction?, 7th ed. (eds E. Cooch and G. C. 1165 White), pp. C-1?C-115. Availabl at: http:/w. phidot.org/software/mrk/docs/book. 1166 165 Letink, M., G. Norbury, A. Cree, P. J. Seddon, R. P. Duncan, and C. J. Schwarz. 2010. Removal 1167 of introduced predators, but not artifcal refuge supplemntation, increse skink survival 1168 in coastal duneland. Biological Conservation 143:72-77. 1169 Leynaud, G. C., and E. H. Bucher. 2005. Restoration of degraded Chaco woodlands: efects of 1170 reptile asmblages. Forest Ecology and Management 213:384-390. 1171 Lindenmayer, D. B., J. T. Wood, C. MaGregor, D. R. Michal, R. B. Cunningham, M. Crane, 1172 R. Montague-Drake, D. Brown, R. Muntz, and D. A. Driscoll. 2008. How predictable are 1173 reptile responses to wildfre? Oikos 117:1086-1097. 1174 Lit, A. R. 1999. Hrpetofaunal responses to longleaf pine sandhil restoration treatments on 1175 Eglin Air Force Bas, Florida. M.S. thesi, University of Florida. 130 pp. 1176 Lit, A. R., L. Provencher, G. W. Tanner, and R. Franz. 2001. Herpetofaunal responses to 1177 restoration treatments of longlef pine sandhils in Florida. Restoration Ecology 9:462- 1178 474. 1179 Mas, B., D. D. Putra, M. Waltert, Y. Clough, T. Tscharntke, and C. H. Schulze. 2009. Six years 1180 of habita modificton in a tropical rainforest margin of Indonesia do not afct bird 1181 diversity but endemi forest speies. Biologicl Conservation 142:2665-2671. 1182 MacKenzi, D.I, J.D. Nichols, J.A. Royl, K.H. Pollock, L.L. Baily, and J.E. Hines. 2006. 1183 Ocupancy estimation and modeling: inferring paterns and dynamics of species 1184 ocurree. Elsvier, Burlington, Maschusts, USA. 1185 Masot, M., J. Clobert, T. Pilorge, J. Lecomt, R. Barbault. 1992. Density dependence in the 1186 common lizard: demographi consquences of a density manipulation. Ecology 73:1742- 1187 1756. 1188 166 Mathews, C. E., C. E. Moorman, C. H. Greenberg, and T. A. Waldrop. 2010. Responses of 1189 reptiles and amphibins to repeatd fuel reduction tretments. Journal of Wildlf 1190 anagent 74:1301-1310. 1191 Mazeroll, M. J., L. L. Bailey, W. L. Kendal, J. A. Royle, S. J. Converse, and J. D. Nichols. 1192 2007. Making great lps forward: acounting for detcability in herpetologial field 1193 studies. Journal of Herpetology 41:672-689. 1194 Means, D. B. 2006. Vertbrat faunal diversity of longleaf pine ecosystems. Pages 157-213 In: S. 1195 Jose, E. Jokela, and D. L. Miler, editors. The lf pi et: ecology, 1196 silviculture, and restoration. Springer, New York, USA. 1197 Menkins, G. E, Jr., and S. H. Anderson. 1988. Estimation of smal-al population size. 1198 Ecology 69:1952-1959. 1199 ihener, W. K. 1997. Quantitavely evaluating restoration experiments: reearch design, 1200 staitcl analysis, and dat management considerations. Restoration Ecology 5:324? 1201 377. 1202 Mitchel, R. J., J. K. Hiers, J. J. O?Brien, S. B. Jack, and R. T. Engstrom. 2006. Silviculture that 1203 sustains: the nexus betwn silviculture, frequent prescribed fire, and conservation of 1204 biodiversity in longleaf pine forest of the southeastern United Staes. Canadian Journal 1205 of Forest Reearch 36:2723-2736. 1206 Murray, D. L., and Fullr, M. R. 2000. A critical review of the efects of marking on the biology 1207 of vertebrats. In Research Techniques in Animal Ecology (ed. F. T. Boitni L), pp. 15- 1208 64. Nw York: Columbi University Pres. 1209 Mushinsky, H. R. 1985. Fire and the Florida sandhil herpetofaunal community: with special 1210 atention to responses of Cnemidophorus sexlineatus. Herpetologica 41:333-342. 1211 167 Myers, R. L. 1990. Scrub and high pine. Pages 150?193 in R. L. Myers and J. J. Ewel, editors. 1212 Ecosystems of Florida. University of Central Forida Pres, Orlando, Florida, USA. 1213 Nowaki, G. J., and M. D. Abrams. 2008. The demise of fire and ?mesophiction? of forest in 1214 the eastern United Staes. BioScienc 58:123-128. 1215 Outcalt, K. W., and D. G. Brockwy. 2010. Structure and compositon changes following 1216 restoration treatments of longleaf pine forest on the Gulf Coastal Pin of Alabam. 1217 Forest Ecology and Managemnt 259:1615-1623. 1218 Pollock, K. H., J. D. Nichols, C. Brownie, and J. E. Hines . 1990. Staistcl inferenc for 1219 capture-reapture experiments. Wildlf Monographs 107:1?97. 1220 Provencher, L., B. J. Herring, D. R. Gordon, H. L. Rodgers, K. E. M. Galey, G. W. Tanner, J. L. 1221 Hardesty, and L. A. Brennan. 2001a. Efects of hardwood reduction tchniques on 1222 Longlef Pine sandhil vegetaion in northwst Florida. Restoration Eology 9:13-27. 1223 Provencher, L., B. J. Herring, D. R. Gordon, H. L. Rodgers, G. W. Tanner, J. L. Hardesty, L. A. 1224 Brennan, and A. R. Lit. 2001b. Longleaf pine and oak responses to hardwood reduction 1225 techniques in fire-suppresed sandhils in northwest Florida. Forest Ecology and 1226 Management 148:63-77. 1227 R Developmnt Core Ta. 2009. R: a language and environment for staistcl 1228 computing. R Foundation for Stistcl Computing,Vinna, Austri. 1229 http:/w.R-project.org/. Acesd 13 March 2010 1230 Renken, R. B., W. K. Gram, D. K. Fantz, S. C. Richter, T. J. Miler, K. B. Ricke, B. Russel, and 1231 X. Wang. 2004. Efects of forest managemnt on aphibians and reptiles in Misouri 1232 Ozrk forest. Consrvation Biology 18:174-188. 1233 168 Russel, K. R., H. G. Hanlin, T. B. Wigley, and D. C. Guynn, Jr. 2002. Responses of iolated 1234 wtland herpetofuna to upland forest management. Journal of Wildlf Managemnt 1235 3:603-617. 1236 Schoener, T. W. 1985. Are lizard population sizes unusualy constant through time? Aerican 1237 Naturalist 126:633-641. 1238 Schrott, G. R., K. A. With, and A. W. King. 2005. Demographic limtaions of the ability of 1239 habita restoration to rescue declining populations. Conservation Biology 19:1181-1193. 1240 Schwarz, C. J., and A. N. Arnason. 1996. A general methodology for the analysis of capture- 1241 recpture experiments in open populations. Biomtrics 52:860-873. 1242 Scott, T. A., W. Wehtj, and M. Wehtj. 2001. The need for straegic planning in pasive 1243 restoration of wildlfe populations. Restoration Ecology 9:262-271. 1244 Smalwood, K. S. 2001. Linking habit resttion to meaningful units of animal deography. 1245 Restoration Ecology 9:253-261. 1246 Smith, W. H., and L. J. Risler. 2010. Quantifyng disturbance in trrestrial communities: 1247 abundance-biomas cparisons of herpetofauna closly track forest succsion. 1248 Restoration Ecology 18:195-204. 1249 Sten, D. A. 2010. Snakes in the gras: secretive natural histories defy both conventional and 1250 progresive staistc. Herpetologial Conservation and Biology 5:183-188. 1251 Sten, D. A., A. E. Ral MG, S. M. Hermnn, J. A. Stiles, S. H. Stiles, and C. Guyer. 2010a. 1252 Efects of forest management on aphibians and reptils: generalist species obscure 1253 trends among native forest aociates. Open Environmental Sciencs 4:24-30. 1254 Sten, D. A., L. L. Sith, and M. A. Baily. 2010b. Suggestd modifitons to terrestrial box 1255 traps for snakes. Herpetological Review 41:320-321. 1256 169 Sten, D. A., C. Guyer, and L. L. Smith. 2011. A case study of relative abundance in snakes. In 1257 Measuring and Monitoring Biodiversity: Stndard Methods for Reptils (R. W. 1258 cDirmid, M. S. Foster, C. Guyer, J. W. Gibbons, and N. Chernoff, eds.), Smithsonian 1259 Instiuton Pres: in pres. 1260 Stewart-Oten, A., W. W. Murdoch, and K. R. Parker. 1986. Environmental impact asemnt: 1261 ??psudoreplicaton?? in time? Ecology 67: 929?940. 1262 Todd, B. D., and B. B. Rotherml. 2006. Aseing quality of clearcut habitas for amphibians: 1263 efcts on abundances versus vital rats in the southern toad (Bufo tersti). Biologicl 1264 Consrvation 133:178-185. 1265 Trainor, C. R., and J. C. Z. Woinarski. 1994. Responses of lizards to three experimental fire 1266 regimes in the savanna forest of Kakadu National Prk. Wildlf Resarch 21:131-148. 1267 van Baaln, M., V. Kr?in, P. C. J. van Rijn, and M. W. Sabelis. 2001. Alternative food, 1268 switching predators, and the persistenc of predator-prey systems. Arican Nturalist 1269 157:512-524. 1270 Van Horne, B. 1983. Density as a misleading indicator of habita quality. Journal of Wildlfe 1271 Managemnt 47:893-901. 1272 Ware S., C. Frost, and P. Doerr. 1993. Southern mixed hardwood forest: the former longleaf 1273 pine forest. In W. H. Martin, S. G. Boyce, and A. C. Echternacht [eds.], Biodiversity of 1274 the southeastern United Stes: lowland trrestrial communitis, 447?493. John Wiley 1275 and Sons, Nw York, Nw York, USA. 1276 White, P. S., and J. L. Walker. 1997. Approximating nature?s variation: selcting and using 1277 refnce informtion in restoration ecology. Restoration Ecology 5:338-349. 1278 170 White, G. C. and K. P. Burnham. 1999. Program MARK: survival estimation from populations 1279 of marked anials. Bird Study 46 (Supplent):120?138. 1280 iederhecker, H. C., A. C. S. Pinto, M. S. Paiva, and G. R. Colli. 2003. The demography of the 1281 lizard Tropidurus torquatus (Squamt, Tropiduridae) in a highly seasonal neotropical 1282 svanna. Phyllomedusa 2:9-19. 1283 Wilams, K. A., P. C. Frederick, and J. D. Nichols. 2011. Use of the superpopulation approach 1284 to estiate breding population size: an exampl in asynchronously breeding birds. 1285 Ecology 92:821-828. 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 171 Figure 1. Mean population sizes (gross population and standard errors) of Aspidoscelis 1300 sexlineatus in longlef pine sandhils subjectd to various hardwood removal stragis on Eglin 1301 Air Force Bas in 1997-1998 (A) and 2009-2010 (B). 1302 Figure 2. Man number of marked adults (and standard errors) of Aspidoscelis sxlineatus in 1303 longleaf pine sndhils subjectd to various hardwood removal straegis on Eglin Air Force 1304 Bas in 1997-1998 (A) and 2009-2010 (B). 1305 Figure 3. Mean number of marked juveniles (and standard errors) of Aspidoscelis sxlineatus in 1306 longleaf pine sndhils subjectd to various hardwood removal straegis on Eglin Air Force 1307 Bas in 1997-1998 (A) and 2009-2010 (B). 1308 1309 1310 172 Table 1: Model comparison table for POPAN capture-mark-recapture analysis aesing efects 1310 on cpture probability (p), entry probability (pent) and apparent survival (phi) on Aspidoslis 1311 sexlineatus populations in longleaf pine sandhils subjectd to various hardwood removal 1312 stragies on Eglin Air Forc Bas betwen 1997-1998. Table includes number of paraetrs 1313 (K), model wights (relative likelhood of models in the st), and diferenc in Akaike?s 1314 information criteon corrected for smal sple siz (!AIC c ). 1315 Model no. Model K !AIC c Model weight Efects on pa 1 p(bare ground) 37 0.00 0.46 2 (gras sedge) 1.19 0.25 3 p(sx/age) 39 2.66 0.12 4 (constnt) 36 3.37 0.09 5 p(oak midstory) 37 4.00 0.06 6 (sex/age + treatment) 43 7.60 0.01 7 p(tretnt) 40 9.64 0.00 Efects on pentb 1 pent(sx/age + treatment) 43 0.00 1.00 2 nt(constnt) 36 12.67 0.00 3 pent(treatment) 40 13.85 4 nt(gras sdge) 37 14.87 0.00 5 pent(oak midstory) 6 nt(bare ground) 37 14.95 0.00 173 7 pent(gras sedge + bare ground + oak midstory) 39 19.47 0.00 Efects on phic 1 phi(sx/age + treatment) 37 0.00 0.86 2 (sex/) 33 3.61 0.14 3 phi(treatmnt) 34 26.64 0.00 4 (constnt) 30 34.01 a Additional parametrs modeled as phi(sex/age + treatment)pe(sex/age + treatment)N(site) 1316 b Atil paratrsld as (sx/ + tretnt)p(sx/ + tretnt)(~sit) 1317 c Additional parametrs modeled as p(bare ground)pent(~sex/age + treatment)Nite) 1318 1319 1320 1321 1322 1323 174 Table 2: Model comparison table for POPAN capture-mark-recapture analysis aesing efects 1323 on cpture probability (p), entry probability (pent) and apparent survival (phi) on Aspidoslis 1324 sexlineatus populations in longleaf pine sandhils subjectd to various hardwood removal 1325 stragies on Eglin Air Forc Bas betwen 2009-2010. Table includes number of paraetrs 1326 (K), model wights (relative likelhood of models in the st), and diferenc in Akaike?s 1327 information criteon corrected for smal sple siz (!AIC c ). 1328 Model no. Model K !AIC c Model weight Efects on pa 1 p(gras sdge + bare ground + oak midstory) 39 0.00 0.95 2 p(gras sedge) 37 6.13 0.04 3 (sx/age) 39 15.47 0.00 4 p( sex/ + treatment) 43 16.99 5 (oak midstory) 37 32.43 0.00 6 p(treatment) 40 34.17 7 (bare_ground) 37 45.67 0.00 8 p(constant) 36 46.11 Efects on pentb 1 pent(sx/age) 39 0.00 0.9040 2 nt(constnt) 36 6.64 0.0327 3 pent(sex/age + treatment) 43 7.18 0.0249 4 nt(bare ground) 37 8.81 0.0110 175 5 pent(gras sedge) 37 8.81 0.0110 6 nt(oak midstory) 7 pent(treatment) 40 10.83 0.0040 8 nt(gras sdge + bare ground + oak midstory) 39 13.18 0.0012 Efects on phic 1 phi(sx/age + treatment) 35 0.00 0.90 2 (sex/) 31 4.36 0.10 3 phi(treatmnt) 32 33.90 0.00 4 (constnt) 28 45.68 a Additional parametrs modeled as phi(sex/age + treatment)pe(sex/age + treatment)N(site) 1329 b Atil paratrsld as (sx/ + tretnt)p(sx/ + tretnt)(~sit) 1330 c Additional parametrs modeled as p(gras sedge + bare ground + oak midstory) 1331 pent(sex/age)N(sit) 1332 1333 176 Table 3. Sx ratios and age clase of Aspidoscelis sxlineatus populations in longleaf pine 1333 sndhils subjectd to three hardwood removal treaments on Eglin Air Force Bas. Individuals 1334 tha escaped without reciving a mark are not included. 1335 Burn Control Herbicide Mechanial Referenc 1997-1998 Female 41 13 20 30 38 Male 52 39 35 65 76 % Female 0.44 0.25 0.36 0.32 0.33 Juvenile 38 9 14 20 31 % Jnil 0.29 0.15 0.20 0.17 0.21 2009-2010 Female 85 54 64 58 51 Male 72 65 66 53 60 % Female 0.54 0.45 0.49 0.52 0.46 % Juvenil 0.16 0.16 0.15 0.26 0.21 1336 1337 1338 1339 1340 1341 1342 1343 177 Table 4: Survivorship estimates (phi, and standard errors) and 95% confidence intrvals for 1344 Aspidosclis sexlineatus popultions in longlef pine sandhils subjectd to various hardwood 1345 removal treatmnts on Eglin Air Force Bas. 1346 1997-1998 2009-2010 phi LCL U phi LCL U Femal Burn 0.77 (0.05) 0.67 0.86 0.81 (0.03) 0.73 0.87 Control 0.92 (0.03) 0.83 0.97 0.84 (0.04) 0.76 0.90 Herbicide 0.83 (0.05) 0.70 0.91 0.93 (0.03) 0.86 0.97 Mhanial 0.91 (0.04) 0.79 0.97 0.81 (0.04) 0.72 0.87 Referenc 0.87 (0.05) 0.76 0.94 0.86 (0.04) 0.75 0.92 Mal Burn 0.86 (0.03) 0.80 0.91 0.82 (0.04) 0.74 0.88 Control 0.96 (0.02) 0.91 0.98 0.85 (0.03) 0.77 0.90 Herbicide 0.90 (0.03) 0.82 0.95 0.93 (0.02) 0.87 0.97 Mhanial 0.95 (0.02) 0.89 0.98 0.82 (0.04) 0.74 0.88 Referenc 0.93 (0.02) 0.87 0.96 0.87 (0.04) 0.77 0.93 Juvnil Burn 0.73 (0.07) 0.58 0.84 0.65 (0.07) 0.50 0.78 Control 0.90 (0.05) 0.77 0.96 0.70 (0.07) 0.54 0.82 Herbicide 0.79 (0.08) 0.60 0.91 0.85 (0.05) 0.72 0.93 Mhanial 0.89 (0.06) 0.72 0.96 0.65 (0.07) 0.52 0.77 Referenc 0.85 (0.06) 0.68 0.93 0.73 (0.09) 0.53 0.87 1347 178 Table 5. Tree density within hardwood removal sites, Sant Rosa and Okaloosa Counties, Eglin 1347 Air Forc Bas, Florida. One referenc site was not included in 2009-2010 summris. Al units 1348 are m 2 /ha (standard error). 1349 1994 1998-1999 2009-2010 Pinus palustris midstory Burn 0.13 (0.05) 0.05 (0.02) 0.05 (0.02) Control 0.1 (0.02) 0.07 (0.01) 0.01 (0.01) Herbicide 0.09 (0.02) 0.04 (0.01) 0.28 (0.10) Mhanial 0.10 (0.02) 0.03 (0.01) 0.07 (0.02) Referenc 0.03 (0.01) 0.02 (0.01) 0.13 (0.06) Pinus palustris overstory Burn 12.78 (1.85) 12.01 (1.72) 12.93 (1.66) Control 7.88 (0.93) 8.71 (0.93) 10.09 (0.40) Herbicide 11.84 (2.35) 12.01 (2.41) 11.36 (1.50) Mhanial 12.15 (2.43) 11.14 (3.16) 11.79 (2.18) Referenc 16.15 (2.34) 16.65 (2.69) 18.12 (4.74) Quercus sp. midstory Burn 0.79 (0.16) 0.22 (0.11) 0.56 (0.21) Control 1.07 (0.13) 1.23 (0.19) 0.72 (0.24) Herbicide 0.56 (0.14) 0.02 (0.01) 0.14 (0.04) Mhanial 0.87 (0.08) 0.09 (0.07) 1.59 (0.33) Referenc 0.11 (0.03) 0.17 (0.13) 0.11 (0.11) 179 Quercus sp. overstory Burn 10.08 (2.45) 5.41 (2.79) 5.22 (1.65) Control 10.10 (1.34) 9.36 (1.97) 3.76 (1.19) Herbicide 9.08 (1.27) 0.40 (0.15) 0.04 (0.02) Mhanial 11.74 (1.73) 2.18 (1.22) 7.82 (6.78) Referenc 4.93 (1.93) 2.93 (0.33) 0.93 (0.64) 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 180 Table 6. The number of Aspidoscelis sxlineatus marked within each hardwood removal and referenc site, Sant Rosa and Okaloosa 1383 Countis, Eglin Air Force Bas, Florida and corresponding gross popultion estiates (and standard errors) and 95% confidence 1384 intervals. 1385 1997-1998 2009-2010 Site Treatment Marked Individuals Population Estimate 95% LCL 95% UCL Marked Individuals Population Estimate 95% LCL 95% UCL 1ANE Hrbicide 26 215.22 (17.07) 181.77 248.67 49 98.95 (6.43) 86.35 111.56 W Control 23 199.17 (14.68) 170.39 227.94 33 77.56 (6.44) 64.94 90.18 1ASE Burn 38 182.21 (15.28) 152.27 212.16 29 99.03 (10.63) 78.20 119.85 Mechanial 13 129.08 (12.04) 105.48 152.67 20 61.82 (6.98) 48.14 75.50 1CE Refrence 48 71.35 (6.02) 59.55 83.15 44 119.56 (12.82) 94.43 144.68 1CW fre 67 55.84 (5.28) 45.49 66.20 51 149.43 (17.22) 115.68 183.18 2ANE Mechanial 29 77.81 (7.37) 63.37 92.25 23 64.58 (6.75) 51.34 77.82 W Control 11 97.54 (12.00) 74.02 121.06 19 51.51 (6.14) 39.48 63.54 2ASE Burn 33 42.97 (5.93) 31.33 54.60 45 113.53 (9.29) 95.32 131.73 Herbicide 12 45.56 (5.76) 34.28 56.84 30 59.92 (4.35) 51.40 68.44 181 3ANE Herbicide 13 75.47 (8.41) 58.98 91.96 39 80.53 (5.71) 69.34 91.72 W Control 26 53.81 (6.85) 40.39 67.24 48 150.03 (14.55) 121.51 178.55 3ASE Mechanial 47 101.97 (8.76) 84.80 119.14 62 193.36 (17.74) 158.58 228.13 Burn 34 168.64 (12.81) 143.55 193.74 56 164.45 (14.30) 136.42 192.48 3CN Referenc 30 103.48 (10.11) 83.65 123.30 46 128.14 (14.04) 100.62 155.66 4ANE Mchanial 26 182.96 (15.04) 153.48 212.44 130.74 (11.76) 107.68 153.79 W Control 1 57.51 (11.49) 34.99 80.04 42 99.53 (7.86) 84.13 114.92 4ASE Burn 26 276.71 (21.64) 234.29 319.13 56 154.52 (12.89) 129.25 179.79 Herbicide 18 130.44 (11.57) 107.75 153.12 35 69.75 (5.18) 59.60 79.89 1386 1387 1388 1389 1390 1391 1392 1393 182 1394 1395 1396 1397 1398 1399 1400 !" #!" $!!" $#!" %!!" %#!" &'()" *+),(+-" ./(01213/" 4/256)126-" 7/8/(/)2/" 4/ 6)"9 + :'-6,1 + )";1 < / " =(/6,>/)," ?" !" #!" $!!" $#!" %!!" %#!" &'()" *+),(+-" ./(01213/" 4/256)126-" 7/8/(/)2/" 4/ 6)"9 + :'-6,1 + )";1 < / " =(/6,>/)," &" 183 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 !" #" $!" $#" %!" %#" &!" &#" '!" '#" #!" ()*+" ,-+.*-/" 01*234351" 61478+348/" 91:1*1+41" 61 8+"68* ; 1 5 "< 5 )/.= " >*18.?1+." <" !" #" $!" $#" %!" %#" &!" &#" '!" '#" #!" ()*+" ,-+.*-/" 01*234351" 61478+348/" 91:1*1+41" 61 8+"68* ; 1 5 "< 5 )/.= " >*18.?1+." (" 184 1412 1413 1414 1415 1416 1417 1418 1419 1420 !" #" $" %" &" '!" '#" '$" ()*+" ,-+.*-/" 01*234351" 61478+348/" 91:1*1+41" 61 8+"68* ; 1 5 "< ) = 1 +3 /1 > " ?*18.@1+." A" !" #" $" %" &" '!" '#" '$" ()*+" ,-+.*-/" 01*234351" 61478+348/" 91:1*1+41" 61 8+"68* ; 1 5 "< ) = 1 +3 /1 > " ?*18.@1+." (" 185 Appendix I: Species ncluded in non-metric diensional scaling ordinations (Chapter 3). 1421 Species Code Sis Specis Code Species AMCR American Crow GCFL Great Crestd Flycatcher AMKE rin Kestrel GHOW ret Horned Owl BACS Bachman's Sparrow HAWO Hairy Woodpecker RS rn Swlow INBU Indigo Bunting BGGN Blue-gray Gnatcaher LOSH Loggerhead Shrike BHU Brown-headed Nuthatch MIKI Misippi Kite BLGR Blue Grosbek MODO ourning Dove JA Jay NOBO Northern Bobwhite BRTH Brown Thrasher CA ortrn Cardinal BWA Broad-winged Hawk NOMO Northern Mockingbird CACH Carolina Chickadee OROR Orchard Oriole RW roli Wren PIWA Pine Warblr CEDW Cedar Waxwing O iletd Woodpecker CHS Chimney Swift PUMA Purpl Martin COGD Comon Ground Dove RBWO Red-belied Woodpecker R mrackle RCW -cockaded wr CONI Comon Nighthawk REVI Red-eyed Vireo YE mon Yelowthroat RHWO -headed Woodpecker CWWI Chuck-wil's Widow RSHA Red-shoulred Hawk DOWO Downy Woodpecker RT -taile EABL Eastern Bluebird SUTA Summr Tnager 186 EAKI Eastern Kingbird TUVU Turkey Vulture EAME strn Madowlrk WEVI White-yed Vireo EASO Eastern Screeh-owl WITU ild Turkey EATO strn The WOTH ood Thrush ETTI Easteitmouse YBCU Yelow-biled Cuckoo FICR Fish Crow YSFL l-shaftd Flicker 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 187 Appendix II. UTM coordinates for centr of site sampled in 2009-2010. Referenc site 3CS was 1439 not included in analyses pertaining to the ti period following 1998-1999 and iluding 2009- 1440 2010. 1441 Block Treatment X Y 2CE Refrenc -86.7828 30.4711 2CW fere -86.7944 30.4713 1CE Refrenc -86.8433 30.5084 1CW fere -86.854 30.5083 1 Burn -86.8202 30.5506 Mechanial -86.8476 30.5601 1 Hrbiide -86.8325 30.5705 Control -86.8429 30.5734 2 Herbicide -86.8178 30.5893 Burn -86.8084 30.5941 2 Control -86.8208 30.5994 Mechanial -86.8118 30.603 3CN Refrence -86.7672 30.5976 3CS fre -86.7577 30.5796 3 Burn -86.742 30.6049 Mechanial -86.7259 30.6062 3 Control -86.7287 30.6149 Herbicide -86.7167 30.618 4 Control -86.7136 30.6408 188 4 Mechanial -86.6935 30.6236 Hrbiide -86.7047 30.6158 4 Burn -86.686 30.6171 6 Herbicide -86.2869 30.6504 Burn -86.2844 30.6411 6 Control -86.2707 30.6445 1442 1443 1444 1445 1446