Contents lists available at ScienceDirect Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco Review and synthesis The in fl uence of prescribed fi re on wild turkeys in the Southeastern United States: A review and synthesis Gregory T. Wann a, ⁎ , James A. Martin a , Michael J. Chamberlain a a Warnell School of Forestry and Natural Resources, University of Georgia, 180 E. Green St., Athens, GA 30602, United States A R T I C L E I N F O Keywords: Habitat selection Meleagris gallopavo Movement Pine-grassland Prescribed fi re Return interval Survival A B S T R A C T The pine-grassland ecosystems once prevalent in the Southeastern United States were dependent on frequent fi re events to maintain plant communities and avoid succession to hardwood and shrub-dominant communities. The use of prescribed fi re has replaced naturally occurring fi res produced from lightning strikes to maintain re- maining pine-grassland systems, and to expand and promote restoration into reclaimed areas. Currently, pre- scribed fi re is a widely accepted management tool promoted by both state and federal wildlife and land-man- agement agencies, and is assumed to be bene fi cial for both game and non-game species. However, a comprehensive set of guidelines related to use of prescribed fi re for promotion of wildlife is lacking for most species, including species whose dependence on fi re is presumed to be critical. We reviewed available literature on prescribed fi re and its in fl uence on wild turkeys ( Meleagris gallopavo ) in the Southeastern distribution of its range. We reviewed extant literature relative to historical use of prescribed fi re for upland gamebird manage- ment, and focused on documented e ff ects of fi re on life-history characteristics of wild turkeys, including habitat selection, demography, and movement. The literature supported preferential use of areas burned in the previous 3 years, with avoidance of areas lacking a recent fi re history. Fire return intervals between 2 and 3 years were generally supported in the literature as optimal to reduce woody shrub encroachment and maintain an her- baceous understory. However, areas infrequently burned, such as mature hardwood forests, provide important habitat during fall and winter, and provide important roosting sites. Contrary to misperceptions among the public, growing-season fi re appears to pose little direct risk to wild turkey nests and poults, but research on this topic is limited and only recently initiated. We lack a collective knowledge of the most appropriate spatial scale and extent of prescribed fi res for wild turkeys, and a single set of values for these metrics that can be applied throughout the Southeastern U.S. is likely to be unrealistic given variations in local plant communities and landscape composition. Non-target species should be carefully considered before implementing prescribed fi re regimes targeted speci fi cally towards wild turkeys, as such fi re regimes may not be optimal for other species. 1. Introduction Naturally occurring fi re is responsible forshaping vegetative com- munities in many ecosystems throughout North America, including the Southeastern pine-grassland systems where fi re frequency has been altered through fi re suppression and landscape changes (Waldrop et al., 1992; Glitzenstein et al., 1995, 2003; Van Lear et al., 2005). Once ecosystems previously maintained by wild fi res are lost, rehabilitating them to historic conditions requires implementation of management plans that use prescribed fi re and consistent fi re events to maintain desired plant communities (Waldrop and Goodrick, 2012). Prescribed fi re has been used as a primary management tool to bene fi t many avian species endemic to ecosystems historically maintained by naturally occurring fi res (Saab et al., 2005; Russell et al., 2009), including red- cockaded woodpecker ( Leuconotopicus borealis ) and Bachman ’ s sparrow ( Peucaea aestivalis ) in pine-grasslands of the Southeast (James et al., 1997; Plentovich et al., 1998; Russell et al., 2009) and prairie chickens ( Tympanuchus spp.) in the Great Plains (Kirsch, 1974). Prescribed fi res in the Southeastern U.S. have also been used to promote conditions favorable for game species such as northern bobwhite ( Colinus virgi- nianus ; e.g., McGrath et al., 2017), white-tailed deer ( Odocoileus virgi- nianus ; Masters et al., 1993), and wild turkey ( Meleagris gallopavo ; Little et al., 2014). Prescribed fi re is now a primary tool used by natural re- source managers, and is increasingly used on state and federal lands (Waldrop and Goodrick, 2012). Nonetheless, challenges remain in im- plementing prescribed fi re due to negative perceptions about fi re, and https://doi.org/10.1016/j.foreco.2019.117661 Received 22 July 2019; Received in revised form 12 September 2019; Accepted 30 September 2019 ⁎ Corresponding author. E-mail address: greg.wann@uga.edu (G.T. Wann). Forest Ecology and Management 455 (2020) 117661 Available online 11 November 2019 0378-1127/ © 2019 Elsevier B.V. All rights reserved. T because of logistical challenges of conducting prescribed fi res at large spatial extents necessary to manage pine-grassland ecosystems (Brennan et al., 1998). Furthermore, direct e ff ects (e.g., nest mortality) of prescribed fi re on wildlife are still poorly understood, necessitating work to synthesize existing studies on e ff ects of prescribed fi re on wildlife. Fortunately, substantive research focusing on responses of wildlife to prescribed fi re has accumulated over the past 3 decades (e.g., Fontaine and Kennedy, 2012). Hence, reviewing and synthesizing ex- tant literature is bene fi cial to inform wildlife management agencies charged with using prescribed fi re to manage forest and grassland communities. We focus on the wild turkey ( Meleagris gallopavo ; here- after turkey), a species whose range in the Southeastern U.S. is often within fi re-managed ecosystems on public lands. We restrict our in- ference to the fi re ecology of turkeys and pine-grassland dominant systems in the Southeastern U.S., because little information exists out- side of this region, and because prescribed fi re is a common approach used to manage vegetation communities in pine forests of the Southeast (Johnson and Hale, 2002). The wild turkey is the largest galliform in North America and has a distribution ranging from southern Canada to southern Mexico (McRoberts et al., 2014). Turkeys inhabit a wide range of cover types in the Southeastern portion of their range, and habitat use varies sea- sonally (Miller and Conner, 2007). In general, both forests and treeless cover types are important determinants of turkey space use (Porter, 1992). Turkey populations were threatened with extirpations by the early 20th century resulting from over harvesting and habitat loss. Their historic distribution has since been restored due to restocking and harvest regulations by state and federal agencies (Kennamer et al., 2001). However, since 2000, populations have exhibited declines in both abundance (Eriksen et al., 2015) and productivity (Byrne et al., 2015). State wildlife agencies throughout the U.S. are actively trying to ensure sustainable populations of turkeys, given the economic and so- cial importance of turkeys as a game species (Southwick Associates, Inc. 2003). Therefore, management strategies that can enhance turkey po- pulations by improving habitat conditions are actively pursued by wildlife managers. In the Southeast, pine-grassland ecosystems are critical in supporting turkeys and other wildlife species. Historically, longleaf pines ( Pinus palustris P. Miller) were once dominant in savanna and forest communities throughout the Southeastern U.S. (Bridges and Orzell, 1989; Peet and Allard, 1993). Today, < 3% of presettlement old-growth pine habitat remains (Frost, 1993), due primarily to fi re suppression (Brockway and Lewis, 1997). The in fl uence of fi re on turkeys has been of interest since the 1930s. Herbert Stoddard wrote frequently on using fi re to manage northern bobwhite (Way, 2008), and Stoddard ’ s career spanned a period when the U.S. Forest Service was promoting and practicing complete fi re suppression (Way, 2006). Stoddard viewed the use of prescribed fi re as essential to maintain productive habitats for game by promoting early successional plant communities conducive for foraging, nesting, and brooding (Stoddard, 1935, 1963). He also recognized that areas lacking frequent fi re events quickly developed closed canopies and understories too dense to be used by turkeys (Stoddard, 1963). By the 1970s, use of fi re as a management tool for turkeys began being reported by biolo- gists (e.g., Lewis and Harshbarger, 1976). Previously, e ff ects of fi re on turkey habitat were only mentioned in passing in the literature and were rarely the primary focus of scienti fi c articles. Herein, we focus on the in fl uence of prescribed fi re on turkeys to synthesize the extant literature and identify areas where lack of knowledge is prohibitive to making informed management re- commendations. Our objective is to provide state and federal agencies charged with managing turkeys on landscapes with prescribed fi re an assessment of how fi re in fl uences turkey populations, and identify key uncertainties in best management practices. We have organized the paper into 4 sections. First, we describe fundamental concepts in fi re ecology as they relate to land management practices. We discuss dor- mant and growing season fi res, fi re return intervals, and spatial extent of burns. Second, we synthesize extant literature detailing in fl uences of prescribed fi re on 3 components of turkey life-history, including habitat use, demographic responses, and movement. Third, we discuss in- formation needs to guide and motivate future research evaluating how prescribed fi re in fl uences turkey populations. We o ff er suggestions re- lative to reporting results in a consistent manner to ease comparisons among studies. Fourth, we provide practical guidance on managing pine-grassland habitats for turkeys given what is known from extant literature. 2. Fire management Turkeys were historically abundant throughout the Southeastern U.S. (Mosby and Handley, 1943; Kennamer et al., 2001), occupying ecosystems largely composed of fi re adapted forests and grasslands (Frost, 1993). Hence, turkeys are presumably well adapted to regularly occurring fi res in pine-grassland habitats. A common goal among state agencies is to create and manage historic pine-grassland conditions, which relies on understanding fi re ecology. Currently, no region-wide statistics have been published on the extent to which prescribed fi re is applied across the landscape. In the 13 states that comprise the Southeastern U.S., the U.S. Forest Service classi fi es nearly 94 million acres of “ fi re forest ” type habitats, which are composed of pine and mixed-pine stands (data extracted from EVALIDator v1.8.0.00). Of these 94 million acres, roughly 6 – 9 million acres (6.4 – 9.6%) were burned in 2017 using prescribed fi re. This estimate was based on sur- veys administered to state forestry agencies (M. Melvin, personal communications). If the average time between fi re events (i.e., ‘ fi re- return interval ’ , detailed below) is 3 years, then an estimated 19 – 29 million acres are managed using prescribed fi re, accounting for 20 – 31% of the 94 million acres of fi re forest communities in the Southeastern U.S. Therefore, a solid understanding of fi re management practices is important given the extent to which they are used. The practical im- plementation of fi re management plans depends on land managers making 3 decisions, including when to burn, the frequency of burns ( fi re-return intervals), and the spatial extent (scale) of burns. The in- tensity at which a fi re burns is also an important component of fi re ecology, but is highly dependent on site-speci fi c conditions and beyond the scope of this review. 2.1. Season of burn Historically, fi res across the Southeastern U.S. were ignited by lightning strikes and indigenous humans. There has been considerable debate over the frequency each of these sources contributed to fi res in North America prior to European settlement (Robbins and Myers, 1992), and in regards to timing when fi res were most prevalent. Fires caused by lightning strikes were undoubtedly most common in the growing season (spring and summer; ∼ April – August) when vegetation was driest and lightning strikes frequent (Komarek, 1964). In contrast, fi res started by indigenous peoples were likely most frequent in the dormant season (late summer through early spring; ∼ September – - March), which remains common today as most managers prefer to burn during this time (Stanturf et al., 2002). Dormant season burns are often preferred by managers because they are easier to control, and lower- temperature fi res should reduce scorch and mortality of mature trees (Robbins and Myers, 1992, but see Sparks et al., 2002). Likewise, dryer air and lower winds that follow wetting rains during the dormant season produce desirable conditions for fi re control and smoke disper- sion (Cronan et al., 2015). From a wildlife standpoint, dormant season burns were initially thought to reduce risks to ground- or shrub-nesting gamebirds and songbirds because these fi res preceded breeding activities (Robbins and Myers, 1992; Knapp et al., 2009). For example, in the Southeast turkeys begin nesting activities in March with peaks in incubation occurring in April (Williams and Austin, 1988; Little et al., 2014; Yeldell et al., G.T. Wann, et al. Forest Ecology and Management 455 (2020) 117661 2 2017a; Wood et al., 2018a). Although the dormant season is the pri- mary period when prescribed fi res are conducted, there has been building momentum over the past few decades to incorporate more growing season burns into management planning (Cox and Widener, 2008). There are several reasons for this scheduling shift. First, objec- tives in pine-grassland systems are to achieve understories dominated by forb and grass species with little woody vegetation. However, many hardwood saplings, such as oak ( Quercus spp.) and sweetgum ( Liqui- dambar styraci fl ua ) are not easily killed by fi res during the dormant season (e.g., Boyer, 1993; Glitzenstein et al., 1995). Conversely, hard- wood saplings are more susceptible to fi re mortality during the growing season, particularly after leaf out (Farrar, 1998). Second, historic con- ditions of pine-grassland systems may be more achievable when fi res occur in the growing season due to e ff ects of fi re on vegetation struc- ture. Naturally occurring fi res (i.e., non-human produced) were most abundant during the growing season, so it seems reasonable that this is when they should occur if the goal is to recreate historic conditions (Robbins and Myers, 1992). Third, studies examining the in fl uence of growing-season fi res on breeding birds indicate their e ff ects on nest loss may be either minor or insigni fi cant (Cox and Widener, 2008), likely due to prescribed fi re only a ff ecting a small percentage of the landscape at any given time (for an example with turkeys, see Kilburg et al., 2014; Wood et al., 2018a). Likewise, wild turkeys often renest and frequently initiate clutches in stands not scheduled for prescribed fi res when fi re return intervals exceed 2 years (Yeldell et al., 2017a; Wood et al., 2018a). However, studies examining e ff ects of growing season fi re on non-cavity nesting passerines in the Southeast have been restricted to only a few species, including Henslow ’ s sparrow ( Ammodramus hen- slowii ; Thatcher et al., 2006) and Bachman ’ s sparrow ( Aimophila aesti- valis ; Tucker et al., 2004; Cox and Jones, 2007). More work is needed before general conclusions can be drawn about in fl uences of growing season fi res on a more comprehensive list of ground-nesting or shrub- nesting birds. 2.2. Fire-return interval Frequency of fi re is an important component of habitat management plans in pine grasslands, and the fi re-return interval, severity, and timing of fi res, determines vegetative response and resulting plant community composition (Thaxton and Platt, 2006). Understory vege- tation becomes dominated by hardwoods when fi res are excluded from pine-grasslands, and shading eventually limits grasses and herbaceous plants. Frequent fi re return intervals (e.g., 1 – 3 years) are necessary to maintain desirable herbaceous ground conditions associated with pine- grassland communities (Waldrop et al., 1992; Brockway and Lewis, 1997; Glitzenstein et al., 2012). This fi re-return interval range is com- monly reported for maintaining pine-grassland plant communities. Turkeys use a variety of habitat types throughout their annual cycle, which requires land managers to maintain a range of vegetative con- ditions. For example, nesting females select dense vegetation with high visual obstruction from predators (Badyaev, 1995; Streich et al., 2015), often near openings in forested and shrub habitats (Byrne et al., 2015). Conversely, brood habitat generally contains relatively less understory vegetation that facilitates foraging by poults and female vigilance (Burk et al., 1990; Jones et al., 2005; Spears et al., 2007). Concealment cover is an important attribute of brood habitat, particularly when poults are < 2 weeks old and roost on the ground (Metzler and Speake, 1985; Speake et al., 1985). Such conditions can generally be found in areas 2 years after burns (Wood et al., 2018a), but once poults begin roosting in trees, areas that have not been burned in 3 or more years may be used (Wood et al., 2018a). During the non-breeding season, turkeys often use hardwood forests in bottomlands, drainages, and swamps (Porter, 1992; Miller and Conner, 2007), which are typically not ex- posed to regular fi re events. Therefore, variable habitat conditions re- quired by turkeys throughout their annual cycle depend on fi re history (Fig. 1), and fi re-return intervals outside of the commonly cited 1 – 3 year range are needed in a portion of the individual home range. 2.3. Scale of fi re Previous studies examining fi re e ff ects on turkeys have focused primarily on season of burn and fi re-return intervals, but the scale of fi re is rarely discussed or su ffi ciently reported. The scale of fi re man- agement can be thought of as forming a hierarchy (Fig. 2). The top-most level is the total area being managed (i.e., total management area), which is comprised of both fi re-free and fi re-managed habitats. Note that our use of total management area here is synonymous with the more typically used term, study area. Basically, it is the geographic area within which population(s) are studied. The second level of the hier- archy is the area within the total management area that is subjected to fi re (i.e., fi re managed area) over some de fi ned temporal period (e.g., several years to several decades). De fi ning the temporal period of fi re history is important, otherwise the entire management area would likely be considered a fi re management area given the historic pre- valence of fi res in the Southeast. The third level of the hierarchy is the sum of the area burned each year (i.e., annual burn area). Finally, the fourth level of the hierarchy is the area of individual burns (i.e., burn compartment area). Note that the area of each level of the hierarchy is nested within the area of the next highest level. The term fi re rotation is standard and commonly used to describe the amount of time required for an area equal to a de fi ned area of interest to burn. We developed these terms and de fi nitions, with the exception of fi re rotation, to ex- plicitly aid with synthesis and development of this paper, but they are not necessarily standard terms used by federal and state land or wildlife management agencies. Very little information exists on the appropriate scale of fi re for managing turkey habitat. Stoddard (1963) recommended mana- ging ∼ 33% of turkey range as fi elds and pastures based on his ex- perience in Georgia, although it is unclear how he arrived at this number. Managing this percentage of turkey range in early successional plant communities would translate to a fi re management area of roughly 33% of the total management area based on our de fi nitions, but does not provide information on the annual burn area. Speake et al. (1975) recommended that spring and summer habitat should include 12 – 25% of “ well dispersed ” openings based on turkeys studied in Ala- bama and Kentucky. This recommendation appears to be based on dispersed openings that resulted in the lowest distances of spring movements into nesting and brood-rearing habitats. If openings are maintained by prescribed fi re, this recommendation translates to a fi re management area that is 12 – 25% the size of the total management area. Similar to Stoddard (1963), information was not provided to determine an annual burn area. Hurst (1978) recommended that “ one-third of a given forest compartment should be burned annually. ” Here, the sta- tistic provided is not particularly helpful, but probably translates to an annual burn area that is 33% of the fi re management area. These 3 reported statistics highlight problems caused by a lack of common de- fi nitions available for studies evaluating how fi re e ff ects turkeys and other wildlife. The scale at which prescribed fi re is applied has implications for maintaining the amount of treeless cover types, given that fi re is a primary tool used to maintain treeless, early successional plant com- munities. The average fi re management areas have been reported in most studies, although only recent studies reported the average burn compartment area. In 5 studies reporting average burn compartment area, areas ranged from 10 to 485 ha, with an average of 90 ha (area of burned patches presented in Martin et al., 2012; Kilburg et al., 2014; Little et al., 2014; Yeldell et al., 2017c; Wood et al., 2018a). Several studies reported only the fi re management area over the period of study (Campo et al., 1989; Sisson et al., 1991; Stys et al., 1992). In these studies, the fi re management areas presumably followed management plans, but determining the area optimal for turkey management has not yet been addressed. Furthermore, the size of the average burn G.T. Wann, et al. Forest Ecology and Management 455 (2020) 117661 3 compartment area is often confounded or not distinguished from the annual burn area, making it di ffi cult to determine which is driving turkey responses. 3. Fire e ff ects Herein, we review direct e ff ects of prescribed fi re on turkeys and provide generalized conclusions, while also identifying knowledge gaps. We partition these e ff ects into 2 categories most easily estimated in studies collecting individual-level data (e.g., VHF- and GPS-marked birds): e ff ects in fl uencing how turkeys use the landscape (habitat use and movement), and e ff ects directly in fl uencing population growth (nest and brood mortality). The in fl uence of fi re on turkey behavior and demographic rates are of primary interest because they ultimately provide information on what habitat conditions are needed to support turkeys, and if these habitat conditions in fl uence whether individuals survive and reproduce. We reviewed relevant studies using search en- gines (Google Scholar and Web of Science) and targeted search terms (e.g., “ Meleagris gallopavo ” and “ fi re ” ), reviewed conference proceed- ings (National Wild Turkey Symposium, Southeastern Association of Fish and Wildlife Agencies) and the Tall Timbers Research Bulletin, and identi fi ed and reviewed literature cited in studies located using the aforementioned methods. 3.1. Habitat use The hierarchical process of behavioral responses by individuals that leads to disproportionate use of habitats is termed habitat selection (Jones, 2001), whereas habitat use is simply descriptive of where ani- mals are and is the end result of habitat selection. There is an as- sumption that habitats selected for by individuals provide fi tness ben- e fi ts over those selected against (Martin, 1998). Understanding habitat selection is important because it provides information on areas in need of protection or restoration (for a practical example see Aldridge and Boyce, 2007). Habitat selection may vary by season and demographic class (i.e., age and sex), and often varies by life stage (e.g., pre- breeding, nesting, brood rearing). Habitat studies have been extensively reported for turkeys, but many only described habitat use rather than selection, the di ff erence being that habitat selection studies analyzed location data in a used-available context compared to other studies, which simply described habitats where turkeys were found. We sum- marize fi ndings from habitat selection studies with respect to turkeys in landscapes managed with prescribed fi re. In the context of prescribed fi re, there are 2 primary questions of interest relating to how it in fl u- ences habitat selection. First, do turkeys preferentially select or avoid areas with a burn history? In either case, understanding the time scale and at what successional stages these choices occur is important. Second, how do these habitat preferences vary by season and Fig. 1. A series of photographs monitoring time-since- fi re in fl uences on vegetation structure at a forest stand located at the Jones Ecological Research Center at Ichauway in Southeastern Georgia, USA. A prescribed fi re occurred on 21 October 2003, and the fi rst photograph was taken shortly thereafter in November, followed by photos taken at the same time in subsequent years through 2008. G.T. Wann, et al. Forest Ecology and Management 455 (2020) 117661 4 demographic class? Most habitat selection studies on turkeys have focused on females because of interest in reproduction, although studies have included marked males (Godwin et al., 1992; Martin et al., 2012). We separated studies into pre-nesting (period before egg laying), nesting (laying through incubation), brooding (post-hatching through fl edging), and non-breeding (fall and winter) seasons, because this was the most generalizable way it could be reported from the literature. These life- history divisions varied across studies. At one extreme, 3 distinct per- iods were used to distinguish habitat selection throughout the calendar year (Martin et al., 2012), but at the other extreme, 7 distinct categories were used for the breeding season alone (Stys et al., 1992). Moreover, most studies used fi xed seasonal windows (e.g., Little et al., 2014) by de fi ning the start date of the egg-laying period as the median of all fi rst reported egg-laying events. Yeldell et al. (2017a, 2017c) used seasonal windows that were unique to each female since high-resolution GPS marks allowed researchers to determine precise dates of each life-his- tory event for each female. While these di ff erences were sensible for objectives of each study, they posed comparative challenges for synthesis, so we attempted to coarsely consolidate fi ndings. In some cases, studies could not be directly compared to others because their fi ndings were not reported in a comparable way (e.g., not providing information on habitats selected, but providing information on fi re-re- turn-intervals selected, and vice versa). Pre-nesting – The pre-nesting period is often de fi ned as occurring from roughly the beginning of March until onset of laying. Habitat selection may be particularly important during this time period due to e ff ects on fecundity (Badyaev, 1995; Chamberlain and Leopold, 2000). Cover types selected during pre-nesting generally lack understory and mid-story cover. Most recent studies noted that female turkeys selected hardwood forests during pre-nesting (Martin et al., 2012; Little et al., 2016; Yeldell et al., 2017c; Wood et al., 2018b), whereas Kilburg et al. (2015) also found that transitional zones between treeless areas and forests were selected by females. Little et al. (2016) found that shrub- scrub communities were avoided in Georgia, whereas Yeldell et al. (2017c) found that mature pine forests were avoided in Louisiana. Only Fig. 2. A hypothetical example of the hierarchical scales of a prescribed fi re management plan. The full grid in each panel represents the total management area in 3 di ff erent years. The fi re management area is represented by light yellow (a). Prescribed fi re is applied within the fi re management area during the fi rst year, and the sum of these burn compartment areas is equal to the annual burn area (b). By the second year 66% of the fi re management area has been burned (c), and by the third year all the fi re management area has been burned. This example represents a fi re management area that is ∼ 30% of the total management area, with an annual burn area that is ∼ 33% of the fi re management area. A management plan with a 3-year fi re return interval and a 3-year fi re rotation could produce similar results. (For interpretation of the references to colour in this fi gure legend, the reader is referred to the web version of this article.) G.T. Wann, et al. Forest Ecology and Management 455 (2020) 117661 5 one study reported on habitat preferences of males during pre-nesting; Martin et al. (2012) found that males in Georgia did not preferentially select any habitat types during this period. However, habitats were sometimes selected based on their fi re histories during pre-nesting. Martin et al. (2012) reported that females selected hardwood drain cover types burned in the previous 2 years. Kilburg et al. (2015) re- ported that females selected edges between forests and treeless areas burned during the same growing season in North Carolina, and Yeldell et al. (2017c) reported that females avoided mature pine forests not burned (during either dormant or growing season) in the previous 2 years. Other studies examined habitat selection of turkeys in areas man- aged with prescribed fi re, but the successional stage of selected cover types was not reported. Godwin et al. (1992) found that both juvenile and adult males in Mississippi preferentially selected forested cover types in spring (dates of which corresponded roughly to pre-nesting), but fi re history of forested cover types selected was not reported. In Mississippi, Palmer and Hurst (1998) found that females used areas with lower ground cover height, more herbaceous vegetation, and less woody vegetation, conditions found in areas burned 0 – 1 year prior. Females used areas burned > 2 years prior less than their availability. Across studies, other cover types with varying fi re-return intervals were not reported as being selected for or against. Therefore, it appears that forested edges along treeless cover types with relatively recent fi re histories ( ≤ 2 years) were selected by females during pre-nesting. Nesting – Habitat selection during nesting is perhaps the most im- portant factor in fl uencing fecundity of female turkeys as reproductive success is a primary driver of population sustainability (Vangilder, 1992; Palmer et al., 1993; Roberts and Porter, 1996; Thogmartin and Johnson, 1999; Park et al., 2001). In contrast to pre-nesting, females typically select habitats with more shrub cover during nesting (Streich et al., 2015). However, Davis et al. (1995) found that females in South Carolina did not preferentially select any cover type for nesting, but did fi nd that nearly 90% of nests were located in stands subjected to peri- odic prescribed burning (33% in stands burned < 1 year before nesting, 28% in stands burned 1 – 2 years before nesting, 5% in stands burned 2 – 3 years before nesting, and 23% in stands burned > 3 years before nesting). Kilburg et al. (2014) reported that transitional zones between pine and hardwood forests were selected by nesting females, whereas Little et al. (2016) and Martin et al. (2012) both reported that hardwood forests were selected during nesting. Little et al. (2016) also reported that nesting females selected shrub-scrub cover types and pine forests, whereas Martin et al. (2012) found that females avoided pine forests. Little et al. (2014) reported that nesting females selected areas burned the same season (within the previous 2 months). Martin et al. (2012) reported that nesting females selected areas burned in the pre- vious 2 years, and Wood et al. (2018b) reported nesting females in Georgia preferred treeless cover types and mature pine forests, the latter of which was preferentially selected for when fi re occurred within the previous year. Yeldell et al. (2017a, 2017c) also reported that ma- ture pine forests were preferred, and stands burned 2 years prior were selected (Yeldell et al., 2017a). Mixed oak-pine forests burned the previous year were also selected during nesting (Yeldell et al., 2017c), in addition to mature pine forests burned 2 years prior (Yeldell et al., 2017a). Across studies, females exhibited greater plasticity in habitat selection during nesting than pre-nesting. Notably, habitats burned ≤ 2 years prior were often selected by nesting females. Stys et al. (1992), in a study of radio-marked females in Mississippi pine plantations, reported that older forests were preferred for nesting (stands burned < 1 year at time of nesting were avoided, whereas those burned the prior 1 – 6 years were preferred). These results were similar to Burk et al. (1990), who found that nesting females in Mis- sissippi pine plantations generally selected habitats burned an average of 3 years prior. Conversely, previous contemporary studies all occurred in areas where pine plantations were absent or only a minor percentage of available habitat. Therefore, in non-plantation pine forests of the Southeast, habitats burned within the previous 2 years appear to be important for nesting females. Brood-rearing – The brood-rearing period is a dynamic time when females must make habitat selection decisions based on changing re- source requirements of poults as they age and become able to roost in trees (Williams and Austin, 1988; Healy, 1992). During the day, un- derstories that allow poults to move freely and secure arthropods are important (Hurst, 1978; Metzler and Speake, 1985; Wood et al., 2018b). Little et al. (2016) reported that females preferred hardwood forests, pine forests, and shrub-scrub cover during brood-rearing, while avoiding treeless cover types. Martin et al. (2012) found that females selected hardwood forests and avoided pine forests during brood- rearing. Sisson et al. (1991) also found that females in Georgia avoided pine forests during brood-rearing, but in contrast to Little et al. (2016), noted that females selected for treeless cover types. Wood et al. (2018b), in agreement with Martin et al. (2012) and Little et al. (2016), found that brooding females selected hardwood forests. They also found that treeless cover types, young pine forests, and mature pine forests were selected for during brood-rearing. Broods also roosted at sites not recently burned (3 – 6 years post- fi re), but selected diurnal use areas burned 2 years prior. Yeldell et al. (2017c) reported that females se- lected against hardwood forests and mature pine forests burned > 3 years prior during brood-rearing. Campo et al. (1989) reported brood-habitat use in managed pine forests in east Texas and noted greater selection by females with broods in burned areas, but the burn history was not provided. Regardless, a certain percentage of the landscape with a 2-year fi re-return interval in forested cover types appears to produce vegetative conditions selected for by brooding fe- males. However, forests lacking such a fi re-return interval are also likely important to broods once roosting in trees occurs at 2 weeks of age (Barwick et al., 1970; Phalen et al., 1986). Collectively, these re- sults indicate general selection for forests (hardwood and pine) and treeless cover types with relatively recent burn histories (gen- erally ≤ 2 years). Non-breeding – During the non-breeding season, insects, soft mast, and leafy vegetation become less prevalent in the diet of turkeys, while hard mast begins to dramatically increase (McRoberts et al., 2014). Not surprisingly, hardwood forests are recognized as being critical non- breeding habitat to wild turkeys, because of foraging resources and availability of roosts (Chamberlain and Leopold, 2000; Miller et al., 2000). Unfortunately, most contemporary studies did not examine ha- bitat use in areas with frequent burns during the non-breeding season, although Little et al. (2016) reported that hardwood forests and pine forests were selected for during this time, whereas shrub-scrub cover types were selected against. Yeldell et al. (2017c) reported that hard- wood forests, mixed forests, and treeless cover types were selected, but these cover types had no recent fi re history. Other periods – One study of GPS-marked female turkeys in Louisiana examined the relationship between use of recently burned areas and presence of escape cover (Yeldell et al., 2017b). The authors estimated the probability that turkeys would use burned areas within 250 days post- fi re. Turkeys readily used recently burned stands and were sometimes found in burned stands the same day a burn occurred. Use of burned areas increased up to 141 days after fi res occurred, before declining, and there was a relationship between the size of a burn and turkey use. Turkey space use within burned areas declined as distance to unburned areas increased, suggesting that turkeys favored edges of burned and unburned areas that could serve as escape cover. The strength of this relationship declined as time-since- fi re increased. The authors concluded that smaller burn units were likely to be bene fi cial for turkeys because they provide more edge cover in the form of shrubs and trees for concealment. 3.2. Nest and brood survival Direct e ff ects of prescribed fi re on turkeys is of great interest G.T. Wann, et al. Forest Ecology and Management 455 (2020) 117661 6 because of potential links to fi tness. This interest has prompted recent studies, although some authors essentially dismissed the potential for direct e ff ects of prescribed fi re on turkeys (Brennan et al., 1998). Direct e ff ects of fi re on turkeys can be separated into potential e ff ects on nests and poults, as direct e ff ects on adults is highly unlikely given their mobility. Nests – Moore et al. (2005, 2010) summarized previous studies conducted in the upper coastal plain of South Carolina using radio- marked females to monitor nests in the presence of prescribed fi re (studies conducted by Moore et al., 2002; Carlisle, 2003). Prescribed fi re was applied annually to 1.3% of the total management area during the growing season, and 7.5 – 10% during the dormant season (i.e., an- nual burn area was 8.8 – 11.3% of total management area). The authors stated that fi res were planned on a 3 – 5 year rotation (not fi re-return interval) to enhance conditions for red-cockaded woodpeckers. Of 22 females monitored, 2 (9%) had nests destroyed by prescribed fi re, and one female renested after her fi rst nest was destroyed. The authors concluded that the population was likely to be minimally a ff ected by growing season prescribed fi res given the small areas burned during this period. Some of the most comprehensive studies on e ff ects of fi re on turkey nests have been conducted in southern Georgia. Little et al. (2014) detailed in fl uences of fi re on nests on a site with fi re return intervals ranging from 1 to 3 years. The authors monitored 78 nests and found that only 5 were exposed to presc