Habitat Use and Survival of Preflight Wild Turkey Broods
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Research Article Habitat Use and Survival of Preflight Wild Turkey Broods BRIAN L. SPEARS,1,2 Department of Range, Wildlife, and Fisheries Management, Box 42125, Texas Tech University, Lubbock, TX 79409-2125, USA MARK C. WALLACE, Department of Range, Wildlife, and Fisheries Management, Box 42125, Texas Tech University, Lubbock, TX 79409-2125, USA WARREN B. BALLARD, Department of Range, Wildlife, and Fisheries Management, Box 42125, Texas Tech University, Lubbock, TX 79409-2125, USA RICHARD S. PHILLIPS, Department of Range, Wildlife, and Fisheries Management, Box 42125, Texas Tech University, Lubbock, TX 79409-2125, USA DERRICK P. HOLDSTOCK, Department of Range, Wildlife, and Fisheries Management, Box 42125, Texas Tech University, Lubbock, TX 79409-2125, USA JOHN H. BRUNJES, Department of Range, Wildlife, and Fisheries Management, Box 42125, Texas Tech University, Lubbock, TX 79409-2125, USA ROGER APPLEGATE, Kansas Department of Wildlife and Parks, P.O. Box 1525, Emporia, KS 66801-1525, USA MICHAEL S. MILLER, Texas Parks and Wildlife Department, Pampa, TX 79065, USA PHILLIP S. GIPSON, Kansas Cooperative Fish and Wildlife Research Unit, 205 Leasure Hall, Kansas State University, Manhattan, KS 66506, USA ABSTRACT Wild turkey (Meleagris gallopavo) broods spend the first several days of life on the ground until poult flight capabilities are attained. This is a critical period of wild turkey life history, with poult survival ranging from 12% to 52%. We measured vegetation in plots used by Rio Grande wild turkey (M. g. intermedia) preflight broods at 4 sites in southwest Kansas and the Texas Panhandle, USA, to determine microhabitat selection for ground roosting and to determine if microhabitat was related to poult survival. Hens selected ground-roost locations with more visual obstruction from multiple observation heights than random sites. Plots surrounding ground roosts had 1) greater visual obstruction; 2) increased tree decay; 3) higher percent grass, shrub, litter, and forb cover; and 4) lower percent bare ground cover than random sites. Grass, shrubs, and downed trees appeared to provide desired cover for ground-roosting broods. Poult survival increased with age of poult, size of brood, and density of shrubs 1–2 m tall. Plots used by broods ,10 days old with above average survival contained more visual obstruction and shrubs than plots used by broods 10–16 days old with above average survival, signifying a shift in habitat use by successful broods as poults attain flight abilities. Density of shrubs 1–2 m tall in brood-use areas appears to be important for poult survival to 16 days of age on southern Great Plains rangeland habitats. Ground-level vegetative cover appears to be a significant factor in preflight poult survival. Provisions of ground-level vegetative cover should be considered during wild turkey brooding periods where increased poult survival is desired. (JOURNAL OF WILDLIFE MANAGEMENT 71(1):69–81; 2007) DOI: 10.2193.2005-676 KEY WORDS brood, ground roost, habitat, Meleagris gallopavo intermedia, poult, Rio Grande wild turkey, survival. The first 2 weeks posthatch are the most critical period for Phalen et al. 1986, Campo et al. 1989, Sisson et al. 1991, wild turkey (Meleagris gallopavo) survival, with observed Edelmann et al. 2001). Habitat parameters that facilitate survival rates ranging from 12% to 52% during this period predator avoidance may increase preflight poult survival (Vander Haegen et al. 1988, Sisson et al. 1991, Peoples et al. (Porter 1980, Hurst et al. 1996). Everett et al. (1980), 1995, Miller et al. 1998b, Paisley et al. 1998). Several factors Speake et al. (1985), and Peoples et al. (1995) suggested that may affect preflight poult survival, including predation, habitat management for wild turkeys should focus on weather, starvation, and disease (Hurst et al. 1996, Roberts facilitating preflight poult survival by maintaining high- and Porter 1998, Rolley et al. 1998). Mammalian predation quality brood habitat. appears to be the single greatest factor in preflight poult Habitat use by wild turkey broods has been studied across mortality (Speake et al. 1985, Palmer et al. 1993, Roberts a wide geographic area (Williams et al. 1973, Metzler and and Porter 1998). Speake 1980, Pack et al. 1980, Porter 1980, Hennen and Survival rates reportedly increase after poults reach Lutz 1996). Brooding hens appear to select for specific approximately 2 weeks of age (Speake et al. 1985, Vangilder habitat composition (Porter 1980, Healy 1985, Hennen et al. 1987, Peoples et al. 1995, Hubbard et al. 1999). By 1999, Edelmann et al. 2001) and habitat types. Lack of this point, poults are able to escape ground predators by available vegetation structure that allows preflight poults to flushing to trees and roosting in trees at night. These escape predation may lead to low preflight poult survival observations suggest that ground escape and roost cover are rates. Correlating vegetation structure with preflight poult essential for poult survival during the first 2 weeks of life. survival rates is essential in identifying desirable habitat Previous studies indicated that turkey broods select for cover management plans. However, previous studies of wild types with specific tree density, tree size, shrub density, turkey preflight broods simply used locations of broods as minimum vegetation height, canopy, percent ground cover, an index of habitat preference and quality, and relatively few and visual obstruction 1 m in height (Pack et al. 1980, studies of preflight poult brood habitat use examined 1 E-mail: brian_spears@fws.gov relationships between survival rates and vegetative param- 2 Present address: Upper Columbia Fish and Wildlife Office, 11103 eters (e.g., Porter 1980, McCabe and Flake 1985, Schemnitz E. Montgomery Drive, Spokane, WA 99206, USA et al. 1985, Phalen et al. 1986, Campo et al. 1989, but see Spears et al. Wild Turkey Brood Habitat Use and Survival 69
Metzler and Speake 1980). Assessment of habitat use by broods may change depending on landscape organization (Pack et al. 1980, Phalen et al. 1986, Sisson et al. 1991, Hennen 1999) or determination of habitat availability (McClean et al. 1998). Thus, use may not always be an accurate index of habitat quality (Van Horne 1983) or brood resource preference. Lack of information on daily poult survival in broods 14 days old may further obscure habitat–poult survival relation- ships. Unfortunately, few studies have examined preflight poult survival 2 weeks posthatch. Rio Grande wild turkey (M. g. intermedia) poult survival appears to increase at 9–10 days (Hennen 1999, Spears et al. 2005), not at 14 days as reported for other wild turkey subspecies. Timing of flight attainment by Rio Grande wild turkey poults corresponds with timing observed in other subspecies (Williams 1974, Healy 1992). This suggests that daily survival rates 2 weeks posthatch, undocumented in other subspecies, should correspond with what Hennen (1999) and Spears et al. (2005) observed in the Rio Grande subspecies. We examined relationships between microhabitats used and survival of preflight Rio Grande wild turkey poults at 4 sites in southwest Kansas and the Texas Panhandle, USA. We hypothesized that specific habitat characteristics in- creased preflight poult survival. Our objectives were to 1) describe ground roosts and distances moved by ground- roosting poults, 2) determine timing of first brood tree- roost, 3) determine vegetation characteristics selected by ground-roosting broods, and 4) determine correlations Figure 1. Rio Grande wild turkey preflight poult study sites, Kansas and Texas, USA, 2000 and 2001. (CNG, Cimarron National Grassland; between poult mortality rates and the physiognomy of GHWMA, Gene Howe Wildlife Management Area; MWMA, Matador habitat utilized within the first 2 weeks posthatch. Wildlife Management Area) STUDY AREA area was dry at the surface, except for occasional flooding. We conducted our study at 4 sites (Fig. 1). The first site was in Precipitation was usually ,41 cm per year and was southwest Kansas, where Rio Grande wild turkey population concentrated from April to September (United States Forest numbers have declined since the mid-1980s (Kansas Depart- Service, unpublished data). The other 3 study sites were ment of Wildlife and Parks, unpublished data). The south- located in the Rolling Plains of Texas. This region was west Kansas study site was centered in the Cimarron National characterized by 55–76 cm of annual rainfall. Elevation Grassland (CNG) in Morton County, but it included parts of ranged from 242 m to 909 m. Topography was gently rolling Stevens County, Kansas, and Baca County, Colorado, USA. hills dissected by narrow intermittent stream valleys flowing The CNG was located within the High Plains province of the northwest to southeast. Great Plains (Cable et al. 1996). Elevation ranged from 955 The first Texas site was centered on the Matador Wildlife m to 1,121 m. Topography included rock cliffs, sand dunes, Management Area (MWMA), where turkey populations grassy fields, and the Cimarron River basin. Sand sagebrush were thought to be declining (Texas Parks and Wildlife, (Artemisia filifolia) prairie was adjacent to the river corridor. unpublished data). The MWMA was 11,410 ha and located Grasses in the area included sand bluestem (Andropogon north of Paducah in Cottle County. The MWMA was hallii), blue grama (Bouteloua gracilis), sideoats grama located in the Mesquite Plains subregion of the Rolling (Bouteloua curtipendula), sand dropseed (Sporobolus cryptand- Plains. The area was traversed by the confluence of the rus), sand lovegrass (Eragrostis trichodes), prairie sandreed Middle and South Pease rivers. Mesquite (Prosopis glan- (Calamovilfa longifolia), and buffalo grass (Buchloe dacty- dulosa)–grass savanna dominated, but sand sagebrush, loides). These grasses, as well as sagebrush, four-wing saltbush shinnery oak (Quercus havardii), sand plum (Prunus (Atriplex canescens), rabbitbrush (Chrysothamnus sp.), snake- angustifolia), acacia (Acacia spp.), netleaf hackberry (Celtis weed (Gutierrezia sarothrae), and plains yucca (Yucca glauca) reticulata), honey mesquite (Prosopis juloflora var. glandulosa), dominated the fields and hills adjacent to the Cimarron River and redberry juniper (Juniperus pinchotii) were also common. corridor. Grasses, cottonwood (Populus deltoides), and tam- Native grasses included bluestems and gramas. Cottonwood, arisk (Tamarix chinensis) groves dominated the river basin western soapberry (Sapindus drummondi), and hackberry (Cable et al. 1996). A major part of the Cimarron River in this (Celtis occidentalis) lined riparian corridors (Hodge 2000). 70 The Journal of Wildlife Management 71(1)
The second Texas site was on the Gene Howe Wildlife under approved Texas Tech University Animal Care and Management Area (GHWMA), where Rio Grande wild Use Protocols (no. 99917 and no. 01173B). We captured 45 turkey numbers were apparently at high density and adult hens, 15 juveniles, and 15 adult gobblers at each site in increasing. The GHWMA consisted of 2,138 ha and was 2000 and 2001 and equipped them with backpack-mounted located in Hemphill County, Texas, along the Canadian transmitters (AVM Instruments, Livermore, CA; Advanced River, also in the Mesquite Plains subregion of the Rolling Telemetry Systems [ATS], Isanti, MN). We continued Plains. Vegetation was similar to the MWMA, with trapping efforts from January through mid-March or until meadows of native grasses and cottonwood, soapberry, target numbers were reached. Transmitters had a mortality common hackberry, black walnut (Juglans nigra), western signal delay of 4.5 hours (AVM) or 8 hours (ATS). We soapberry, sumac (Rhus aromatica), and tamarisk groves in sexed, aged (Pelham and Dickson 1992), and weighed the riparian corridor. A narrow transition zone separated captured birds. We recorded temperatures with a rectal riparian corridors and sandhills (Hodge 2000) dominated by thermometer, banded birds with aluminum leg bands, and sand sagebrush. released processed birds on-site. The third Texas site was on the Salt Fork of the Red River We located transmitter-equipped birds 2 times per week in eastern Donley and western Collinsworth counties in the following trapping to determine general movement patterns, Texas Panhandle. This site appeared to have a high-density mortality, and onset of nesting. We determined incubation turkey population that was thought to be increasing (Texas by the continuous location of a hen at a specific point Parks and Wildlife, unpublished data). Seventy-nine percent (Miller et al. 1998a, Keegan and Crawford 1999). We of Donley County and 55% of Collinsworth County were triangulated incubating hens on nests to within 50 m rangeland. Elevation ranged from 632 m to 995 m. Annual (Everett et al. 1980) and monitored individual hen rainfall averaged between 52 cm and 55 cm, with the incubation periods to determine hatch date. majority of precipitation occurring from April to October. We located up to 7 transmitter-equipped hens with Vegetation consisted of cottonwood, honey locust (Gleditsia successful nests at each site each year through walk-in triacanthos), black locust (Robinia pseudo-acacia), Texas triangulations as soon as possible after hatching and flushed sugarberry (Celtis laevigata), and netleaf hackberry trees in them from their brood. We hand-captured up to 5 poults riparian areas, and rolling hills of yucca, grama grasses, per brood (Peoples et al. 1995, Hubbard et al. 1999), leaving bluestem, post oak (Quercus stellata), and snakeweed range- 1 poult of each brood for the hen to return to while we land (McEwen et al. 1973, Williams and Crump 1980). processed captured poults (Peoples et al. 1995). We did not Cattle grazed all 4 areas during this study. At CNG, the flush broods during extreme heat (.358 C) or within 2 national grasslands were rotationally grazed whereas private hours of sunset. We placed poults in an insulated cooler lands were used seasonally or yearlong. Stocking rates were with either a hot water bottle to serve as a heat source during not available, but use was generally moderate (40–50% of cool weather or with moist towels for comfort during standing crop). Riparian areas were not excluded from warmer parts of the day. We removed captured poults .35 grazing. Cattle grazing occurred on selected parts of the m from the capture location to prevent the hen from hearing GHWMA throughout the study period. Cattle use varied peeping from poults that were being processed. We weighed markedly with the diverse patterns of ownership at this site and equipped poults with model 384 transmitters (ATS). but, where grazed, use was moderate to heavier (up to 60% Transmitters weighed 1.8 g, were 10.5 mm long, 20 mm estimated) in some riparian areas. Cattle grazing also wide, 4 mm thick, and had a 10-cm antenna. We shaved a occurred on the MWMA throughout the study period. patch matching the footprint of the transmitter with an No riparian areas were excluded from grazing unless that electric razor on each poult’s back and attached a transmitter pasture was deferred for that year. Cattle were rotated with Supergluet (Bowman et al. 2002) in 2000 and among pastures to achieve moderate use, although use was Superglue or Superglue gel in 2001, both cyanoacrylate heavier around windmills and in riparian areas. The Salt adhesives. As soon as transmitter attachment was dry Fork consisted of several large ranches that primarily used (approx. 45–60 min), we relocated and flushed brooding deferred rotation systems with their cow–calf operations. hens and returned equipped poults to the brood (Hubbard et Grazing access to riparian areas was allowed. Stocking rates al. 1999). were unknown and highly variable, although use was usually We conducted walk-in locations on instrumented broods light to moderate. With grazing rotations, deferments, or 1–4 times per day. We discretely approached the trans- landownership patterns, most turkeys had ungrazed pastures mitter-equipped hen to within 30 m until we could visually available to them throughout the nesting period. Grazed locate her. Once we located the hen, we used radiotelemetry pastures provided areas with herbaceous and woody cover to check for the presence of each transmitter-equipped poult for turkey nesting at all 4 sites. known to be with that brood. When we could not easily visually locate broods (e.g., broods were hiding in dense METHODS brush), we estimated their location to within 10 m by We prebaited locations on study sites during winter months circling them at a distance .30 m and estimating their with whole kernel corn and milo. We used drop and rocket location through triangulations of radio signals. nets (Baldwin 1947) to capture juvenile and adult turkeys We made every effort to ensure that locations represented Spears et al. Wild Turkey Brood Habitat Use and Survival 71
those of undisturbed broods. Observations of telemetered recording the lowest visible 0.1-m band on the visual turkeys during walk-in locations suggested that common obstruction pole. We indexed total hiding cover, or total responses of hens and broods were to either freeze obstruction, by recording total number of visible 0.1-m immediately when an observer was detected or to continue bands. We considered bands visible if any portion could be to move slowly away. When a brood appeared to change seen. locations due to observer presence, we terminated observa- We quantified ground cover along each of the plot’s 10 4- tions and chose another brood for relocation during that m transects that had been used to estimate visual portion of the day. For broods that appeared to be moving obstruction. We determined ground cover (categorized as prior to detection, we placed locations at the point of first grass, forb, crop, shrub, litter, bare ground, or other) by visual contact. sighting through an ocular tube with crosshairs at one end We located brood ground roosts at the beginning of each held vertically. We recorded the ground cover category 10 day by locating and observing ground-roosting hens. We times at 40-cm intervals along each 4-m transect between made telemetry triangulations to within 10 m when we the sighting and visual obstruction pole. could not visually observe hens. We flagged locations of We defined shrub cover as woody-stemmed vegetation ground roosts for vegetation measurements to be made after .0.5 m in height and ,10 cm in diameter. We indexed brood departure. We documented each turkey hen or poult shrub cover by recording number of shrubs within a 2 3 20- location with Universal Transverse Mercator coordinates. m belt transect that was walked with a 2-m long pole held We tracked broods in the above manner until all poults in a horizontally at 0.5 m from the ground along the 20-m brood began to roost in trees. centerline transect. We identified shrub stems that inter- We located and visually inspected instrumented poults sected the pole to species and placed individual plants into that were separated from their brood to determine their fate height classes of 0.5–1 m, 1–2 m, 2–4 m, 4–6 m, or .6 m. (Peoples et al. 1995). Poult transmitters had a limited range We classified woody stems 10 cm diameter at breast (10–100 m), and we found transmitters with preyed-upon height that intersected the pole as trees, and classified woody poult remains several hundred meters from known brood stems ,0.5 m in height as ground cover. locations during our study. Transmitters from these poults We classified trees .10 cm diameter as overstory. To could easily be lost without extensive searches. We did not quantify overstory, we recorded the species and diameter at document any transmitter failures during this study. Bow- breast height (or diam below trunk splitting) for all trees man et al. (2002) reported no failures and .29-day originating within plots. Following Thomas et al. (1979), retention times for this transmitter model attached with we recorded tree decay class for each tree within the plot. these methods on pen-raised poults, and Spears et al. (2002) We also recorded number of coarse woody debris (single reported a mean retention time of 20.4 days for these large branches, logs, or fallen trees originating outside of the transmitters and this technique on wild turkey poults. We plot) for each plot. therefore considered poults equipped with transmitters that In addition to the above measurements, we measured were lost as unlikely to survive on their own and most likely visual obstruction of each actual ground-roost location with to have been carried off by predators. Our survival rate the visual obstruction pole. We recorded visual obstruction 4 estimates may be biased low if these poult losses were in fact m from the ground roost at 0.25-m, 0.5-m, and 1-m height transmitter failures. intervals from all 4 cardinal directions. We considered To determine vegetation characteristics of brood-use areas, readings from the 1-m height standard plot readings. we measured vegetation characteristics within 10 3 20-m Readings from the 0.25-m and 0.5-m heights more closely plots at ground-roost, habitat-use, and random locations. simulated head height and thus natural vision height of Ground-roost plots were centered on the overnight brood- possible mammalian predators such as raccoons (Procyon roosting locations. Habitat-use plots consisted of any plot in lotor), foxes (Vulpes spp.), bobcats (Lynx rufus), or coyotes which a radioequipped poult was observed, including (Canis latrans). Poles from which observations were taken daytime ground-roost plots. We centered plots on poult or were marked at 0.25-m intervals to standardize height brooding hen locations and oriented them north–south observations. We averaged measurements from each height along the 20-m axis. to determine overall ground-roost visual obstruction. We We indexed vegetative visual obstruction with a 1-m tall defined vegetative species utilized for roosting cover as the vertical visual obstruction pole 2.5 cm in circumference species of individual plant under which a brood or poult was marked with 0.1-m vertical bands (Robel et al. 1970). We located. In the case of nonliving material, we recorded cover attached a 1-m pole to the visual obstruction pole by a 4-m in categories such as natural debris pile, man-made debris string to standardize readings (Robel et al. 1970). We pile, or dead standing material. observed the visual obstruction pole from a perpendicular We made identical plot and ground-roost vegetation distance of 4 m and a height of 1 m. We recorded 10 visual measurements in 10 3 20-m paired random plots located 50 obstruction readings along a 20-m centerline transect in m from each ground-roost or habitat plot in a random each plot, each 2 m apart, alternating sides of the centerline. cardinal direction. We treated the center of each ground- We measured ground-level hiding cover, which we roost random plot as a random ground-roost location. We determined as the lowest height of 100% obstruction, by made vegetation measurements at all ground-roost and 72 The Journal of Wildlife Management 71(1)
random sites within 1 week following observation at a differences where data violated the assumption of equal ground roost and within 2 weeks following observation at variances. other brood locations. We used paired t-tests to evaluate Except where noted, we used a significance level of 0.05 differences between ground roosts and paired random plots for all calculations. We give means 6 standard error. We to identify selection of specific characteristics in which used Microsoft Excel and StatisticaÓ (Statsoft, Inc., Tulsa, broods ground roosted. OK) or SAS (Cantor 1997) for all computer analyses. We used proportional hazards regression (SAS procedure PHREG; Cantor 1997) to detect attributes of hens, poults, RESULTS and habitat that influenced risks to poult or brood survival. Ground Roosts Explanatory variables we tested included hen age, hen Ground roosts typically consisted of a nest-like bowl of weight (at time of hen capture), brood size, poult weight (at flattened grass or soil scratched out by the hen. Observations time of poult capture), poult age, and visual obstruction. We of the shape of roost bowls and poult dropping patterns examined both individual poult and overall brood survival, suggested poults roosted underneath the hen’s tail, but defined as 1 poult within a brood surviving to tree poults also appeared to roost underneath the wing or breast roosting. For these analyses, we considered all poults within feathers. Distances between consecutive ground roosts a brood to be the same age. We also defined 2 survival averaged 614 6 145 m (range ¼ 0–2,733 m). We observed periods of 0–9 days and 10–16 days posthatch within which 2 instances where a brood used a ground-roost spot twice, to conduct separate analyses based on Spears et al. (2005). including the reuse of a nest bowl the first day after hatch at We interpreted risk ratios following Allison (1995:117) and MWMA. Broods continued to ground roost on all study present differences at the a ¼ 0.10 level. areas for several days after some poults in the brood were We tested poult survival–habitat correlations separately observed flying to trees, suggesting that hens will ground within the 0–9-day and 10–16-day posthatch periods. We roost until all members of the brood are able to do so. used poult survival days (PSD) in each period as an index of Broods first tree-roosted 11–16 days posthatch (x̄ ¼ 13.714 survival for each brood in each period. We calculated PSD 6 0.438 days, n ¼ 14). We observed broods tree-roosting for the first time on large, horizontal trunks or limbs 3–5 m as the mean number of days transmitter-equipped poults in high, by themselves or with other roosting broods. a particular brood survived. A poult in the 0–9-day survival We measured variables at 420 ground-roost and associated period survived 0 days if killed the day of hatch and 10 days paired random plots across the 4 study sites: 58 roost and 56 if alive at the end of the ninth day. A poult in the 10–16-day random plots from 8 broods on GHWMA, 95 roost and 95 period survived 0 days if killed on the tenth day posthatch random from 10 broods on the CNG, 24 roost and 24 and 7 days if alive at the end of the 16th day. We random from 7 broods on MWMA, and 34 roost and 34 normalized PSD using a modified arcsine square-root random from 8 broods on the Salt Fork. Actual ground- transformation (Zar 1999). We assessed normality and roost locations at all 4 sites had more ground-level visual heterogeneity of variances with Shapiro–Wilk (Shapiro and obstruction (determined by the lowest visual obstruction Wilk 1965) and Hartley’s tests (Hartley 1940). pole dm observed) and more total visual obstruction We performed multivariate forward stepwise regression (P (determined by the total number of obstruction pole dm ¼ 0.15 to remove or enter; Pope and Webster 1972) to observed) than random points at all 3 observation heights (P determine important habitat variables for predicting PSD in , 0.001; Table 1). Plots surrounding ground roosts also had each survival period. We used a general linear model more ground-level visual obstruction than random plots multivariate analysis of variance to test whether relation- (lowest obstruction band visible x̄ ¼ 2.7 vs. x̄ ¼ 2.0; P , ships between PSD and habitat variables in resulting models 0.001), more total visual obstruction (x̄ ¼ 7.9 obstruction were the same among sites. Where selected variables did not bands visible vs. x̄ ¼ 8.7; P , 0.001), less bare ground (x̄ ¼ differ, we pooled data among sites. Where variables differed 6.7% vs. x̄ ¼ 12.9%; P , 0.001), and more trees (x̄ ¼ 3.2 vs. among sites, we tested differences within each site x̄ ¼ 1.8; P ¼ 0.028) and overstory canopy (x̄ ¼ 4.2 canopy hits separately. vs. x̄ ¼ 2.6; P ¼ 0.013) than random plots (Table 1). There To further examine survival–habitat relationships, we used were also study site–specific differences in overnight t-tests to assess habitat characteristic differences in plots ground-roost site selection (Table 2). Brooding hens on used by broods 0–9 days old with below average and above CNG used ground-roost locations with more ground-level average PSD. We also assessed survival–habitat relation- visual obstruction at 0.5-m observation heights (lowest ships for broods 10–16 days old with t-tests comparing obstruction band visible x̄ ¼ 3.6 vs. x̄ ¼ 2.5; P , 0.001) and habitat variables in plots used by broods 10–16 days old with plots with more coarse woody debris (x̄ ¼ 12.4 vs. x̄ ¼ 7.4; P below average and above average PSD. Finally, we used t- ¼ 0.011) and trees in a more decayed state (tree decay index tests to determine differences in habitat use between age x̄ ¼ 3.2 vs. 2.7; P ¼ 0.044) than at paired random plots. classes of broods by comparing broods with above average Brooding hens on MWMA also used ground-roost PSD 0–9 days to those with above average PSD 10–16 days locations with more ground-level visual obstruction at the posthatch. We used Welch’s approximate t-tests and 0.5-m observation height (lowest obstruction band visible x̄ associated degrees of freedom (Zar 1999) to determine ¼ 3.5 vs. x̄ ¼ 2.5; P ¼ 0.001). Brooding hens on Salt Fork Spears et al. Wild Turkey Brood Habitat Use and Survival 73
Table 1. Habitat values measured in plots used as ground night-roosts by preflight Rio Grande wild turkey broods versus random plots at 4 sites in southwest Kansas and the Texas Panhandle, USA, May–August, 2000 and 2001. Used plots Random plots Variable x̄ SE x̄ SE t25 P a Roost 1-m lowest dm visible 2.837 0.130 1.890 0.088 6.293 ,0.001 Roost 1-m total dm visibleb 7.621 0.171 8.882 0.121 7.245 ,0.001 Roost 0.5-m total dm visiblec 5.966 0.216 7.900 0.161 9.598 ,0.001 Roost 0.25-m lowest dm visibled 3.812 0.137 3.218 0.155 4.530 ,0.001 Roost 0.25-m total visiblee 4.172 0.253 6.557 0.202 9.291 ,0.001 Plot lowest dm visiblef 2.683 0.105 2.026 0.096 6.056 ,0.001 Plot total dm visibleg 7.907 0.150 8.655 0.163 5.681 ,0.001 Canopy hits (x/20)h 4.166 0.611 2.631 0.426 2.681 0.013 No. of shrubs 4–6 mi 0.128 0.071 0.094 0.048 0.375 0.711 No. of shrubs .6 mi 0.126 0.118 0 0 1.066 0.297 % grass coverj 0.469 0.036 0.473 0.032 0.154 0.879 % bare groundj 0.067 0.011 0.129 0.016 4.618 ,0.001 % forb coverj 0.205 0.022 0.204 0.017 0.042 0.970 % litter coverj 0.119 0.015 0.104 0.013 1.400 0.174 No. of treesk 3.222 0.593 1.789 0.460 2.338 0.028 Mean tree dbh (cm) 24.708 3.336 24.101 2.685 0.142 0.855 Mean tree height (m) 5.000 0.602 5.482 0.694 1.216 0.238 Mean height lowest branch (cm) 94.815 14.298 88.590 13.781 0.015 0.511 a Mean lowest decimeter visible (all or part) on obstruction pole at ground-roost location from 1-m observation height, 4 m from pole. b Mean total decimeters visible (all or part) on obstruction pole at ground-roost location from 1-m observation height, 4 m from pole. c Mean total decimeter visible (all or part) on obstruction pole at ground-roost location from 0.5-m observation height, 4 m from pole. d Mean lowest decimeter visible (all or part) on obstruction pole at ground-roost location from 0.25-m observation height, 4 m from pole. e Mean total decimeters visible (all or part) on obstruction pole at ground-roost location from 0.25-m observation height, 4 m from pole. f Mean lowest decimeter visible (all or part) on 10 obstruction pole readings in plots surrounding ground-roost locations from 1-m observation height, 4 m from pole. g Mean total decimeters visible (all or part) on 10 obstruction pole readings in plots surrounding ground-roost locations from 1-m observation height, 4 m from pole. h Number of overstory canopy hits at 20 evenly spaced intervals on perimeter of plots surrounding ground-roost locations. i Plot sample within 2 3 20-m belt transect along centerline. j 100-point sample within plot. k Total in plot. used ground-roost locations with more ground-level visual during this period, with Salt Fork having higher mean PSD obstruction at the 0.5-m observation height (lowest (8.917 6 0.48, n ¼ 9 broods) than GHWMA (6.085 6 obstruction band visible x̄ ¼ 4.1 vs. x̄ ¼ 2.4; P ¼ 0.015), 1.082; n ¼ 8 broods, t15 ¼ 2.489, P ¼ 0.025) and CNG and plots with more shrubs 0.5–1 m tall (x̄ ¼ 112.4 vs. x̄ ¼ (7.225 6 0.522; n ¼ 10 broods, t17 ¼ 2.371, P ¼ 0.030). 54.6; P ¼ 0.030) and more total number of shrubs (x̄ ¼ 140.0 Mean PSD for MWMA (8.385 6 0.491; n ¼ 8 broods) was vs. x̄ ¼ 62.7; P ¼ 0.022) than at paired random sites (Table in-between and did not differ from Salt Fork (t15 ¼ 0.773, P 2). ¼ 0.451) or from CNG and GHWMA (F2,23 ¼ 2.350, P ¼ 0.118). Overall PSD 10–16 days posthatch averaged 5.63 6 Survival and Survival–Habitat Relationships 2.1 days (n ¼ 24 broods). Means did not differ among study Four percent of transmitter-equipped poults (5 of 115, from sites during this period (F3,18 ¼ 0.302, P ¼ 0.823). 3 broods) were lost. For inclusion in survival analyses, we categorized these poults as mortalities following their last Percentages of transmitter-equipped poults surviving the known location. Risk of brood mortality (all telemetered entire 16-day period were 27% (9/33) on GHWMA, 29% poults dying) decreased with hen weight at time of initial (10/34) on CNG, 40% (10/25) on MWMA, and 36% (10/ capture ([0.635 1] 3 100 ¼ 39.5%/kg of hen wt; P ¼ 28) on Salt Fork. 0.0002), brood size (5.6%/added poult; P ¼ 0.0295), and We measured vegetation characteristics in 442 habitat-use brood age (6.0%/d; P , 0.0001; Table 3). Individual plots used by 35 broods 0–9 days old: 89 plots on poults had decreased risks of mortality with increased hen GHWMA, 205 on CNG, 74 on MWMA, and 74 on Salt weight at time of initial capture (50.4%/each additional kg Fork. We measured vegetation characteristics in 142 plots of hen wt; P , 0.0001), brood size (6.3% for each used by 22 broods 10–16 days old: 40 plots on GHWMA, additional poult in brood; P ¼ 0.0271), and poult age 66 on CNG, 15 on MWMA, and 21 on Salt Fork. (8.4% for each additional d of age; P , 0.0001). Broods Using the Cox proportional hazards model, increased 10–16 days old had 66.7% less (P ¼ 0.0001) risk of density of shrubs 1–2 m decreased risk of brood mortality mortality than those 0–9 days old. 1.4% per shrub stem (P ¼ 0.0210; Table 3). Increased Overall PSD 0–9 days posthatch averaged 7.66 6 0.372 density of shrubs 1–2 m also decreased risk of poult days (n ¼ 35 broods). Mean PSD differed among study sites mortality 1.4% per shrub stem (P ¼ 0.0070; Table 4). 74 The Journal of Wildlife Management 71(1)
Table 2. Habitat values measured in plots used as ground roosts by preflight broods of Rio Grande wild turkey versus random plots at 4 sites in southwest Kansas and the Texas Panhandle, USA, May–August, 2000 and 2001. Used plots Random plots Sitea Variable x̄ SE x̄ SE df t P b GHWMA Roost 0.5-m low dm visible 3.706 0.359 3.004 0.259 4 1.667 0.171 Coarse woody debrisc 0.934 0.615 0.452 0.372 4 1.903 0.130 No. of shrubs 0.5–1 md 25.102 10.982 14.153 4.708 4 1.812 0.144 No. of shrubs 1–2 md 5.534 3.702 4.680 1.948 4 0.394 0.714 No. of shrubs 2–4 md 0.745 0.512 0.763 0.324 4 0.030 0.977 Total no. of shrubsd 32.371 14.988 19.633 6.416 4 1.659 0.173 Mean % shrub covere 0.185 0.064 0.113 0.035 4 1.637 0.177 Mean tree decayf 1.337 0.184 1.0 0 4 2.141 0.253 CNG Roost 0.5-m low dm visible 3.638 0.327 2.494 0.176 9 3.215 ,0.001 Coarse woody debris 12.382 2.209 7.392 1.327 9 3.180 0.011 No. of shrubs 0.5–1 m 2.469 0.712 1.629 0.402 9 0.627 0.087 No. of shrubs 1–2 m 0.254 0.126 0.188 0.074 9 0.394 0.714 No. of shrubs 2–4 m 0.316 0.140 0.141 0.064 9 1.298 0.226 Total no. of shrubs 2.987 0.825 1.672 0.444 9 2.188 0.056 Mean % shrub cover 0.039 0.010 0.024 0.005 9 1.843 0.098 Mean tree decay 3.164 0.247 2.664 0.334 9 2.381 0.044 MWMA Roost 0.5-m low dm visible 3.462 0.216 2.50 0.127 4 8.603 0.001 Coarse woody debris 0.937 0.324 1.35 0.768 4 0.791 0.473 No. of shrubs 0.5–1 m 9.133 2.142 9.02 2.193 4 0.605 0.961 No. of shrubs 1–2 m 8.533 1.011 4.633 1.300 4 2.410 0.074 No. of shrubs 2–4 m 2.240 0.326 1.233 0.347 4 1.881 0.133 Total no. of shrubs 19.907 2.758 15.233 2.435 4 1.281 0.270 Mean % shrub cover 0.147 0.029 0.114 0.026 4 0.976 0.384 Mean tree decay 2.372 0.114 2.411 0.216 4 0.259 0.808 Salt Fork Roost 0.5-m low dm visible 4.113 0.329 2.357 0.368 5 3.655 0.015 Coarse woody debris 2.442 1.143 2.031 1.030 5 0.606 0.571 No. of shrubs 0.5–1 m 112.422 32.709 54.556 21.554 5 2.989 0.030 No. of shrubs 1–2 m 21.694 7.936 7.506 3.982 5 1.838 0.125 No. of shrubs 2–4 m 1.611 0.970 0.533 0.299 5 1.684 0.153 Total no. of shrubs 135.950 38.621 62.650 22.195 5 0.024 0.022 Mean % shrub cover 0.261 0.071 0.165 0.053 5 1.946 0.109 Mean tree decay 0.935 0.462 2.034 0.176 5 1.946 0.146 a Gene Howe Wildlife Management Area (GHWMA), Cimarron National Grasslands (CNG), Matador Wildlife Management Area (MWMA), Salt Fork of the Red River (Salt Fork). b Mean lowest decimeter visible (all or part) on obstruction pole at ground-roost location from 0.5-m observation height, 4 m from pole. c Number of coarse woody debris in plot. d Plot sample within 2 3 20-m belt transect along centerline. e 100-point sample within plot. f For all trees in plot. Number of shrubs 1–2 m tall and mean diameter at breast the model (F2,30 ¼ 6.69, P ¼ 0.004; Table 5). The height of trees explained 32% of variation in 0–9-day PSD. relationship between 0–9-day PSD and individual variables Both variables were positively correlated with PSD, and a of number of shrubs 1–2 m (F3,19 ¼ 1.036, P ¼ 0.399) and significant positive relationship existed between PSD and diameter at breast height (F3,19 ¼ 0.926, P ¼ 0.447) did not differ among sites. The relationship between 0–9-day PSD and the combined variables was not different among sites Table 3. Cox proportional hazards model of Rio Grande wild turkey brood survival (1 poult surviving) for (n ¼ 33) broods observed at 4 sites in southwest Kansas and the Texas Panhandle, USA, May–August, 2000 and Table 4. Cox proportional hazards model of Rio Grande wild turkey poult 2001. survival for broods (n ¼ 33) observed at 4 sites in southwest Kansas and the Texas Panhandle, USA, May–August, 2000 and 2001. Hazard Variable na df Levels v2 P ratio Hazard Variable na df Levels v2 P ratio Likelihood ratio 557 7 94.2661 ,0.0001 Hen age 557 1 0.8394 0.3596 0.901 Likelihood ratio 557 5 151.5157 ,0.0001 Hen wt 557 1 13.9360 0.0002 0.635 Hen wt 557 1 32.8181 ,0.0001 0.497 Brood size 557 1 4.7378 0.0295 0.944 Brood size 557 1 4.8850 0.0271 0.937 Brood age 557 1 27.6282 ,0.0001 0.940 Brood age 557 1 44.3041 ,0.0001 0.916 Shrub density 557 1 0.5–1.0 m 2.8674 0.0904 1.002 Shrub density 557 1 0.5–1.0 m 3.3704 0.0664 1.002 557 1 1.0–2.0 m 5.3252 0.0210 0.991 557 1 1.0–2.0 m 7.2664 0.0070 0.986 557 1 2.0–4.0 m 0.3351 0.5627 1.009 557 1 2.0–4.0 m 0.3755 0.5400 0.986 a a We based sample size upon telemetry locations of broods. We based sample size upon telemetry locations of broods. Spears et al. Wild Turkey Brood Habitat Use and Survival 75
Table 5. Multiple regression coefficients and standard errors of habitat plots were sand sagebrush on CNG; sand sagebrush, sand variables in order of importance for 0–9-day-old Rio Grande wild turkey poult survival at 4 sites in southwest Kansas and the Texas Rolling Plains, plum, shinnery oak, sumac, and honey mesquite were most USA, during May–August, 2000 and 2001.a common at the Texas sites. Mottes of shinnery oak often completely concealed broods but allowed poults to move Standard Variable Coeff. SE coeff. SE t28 P about freely. Although mean diameter at breast height was included in Intercept 6.038 0.150 7.469 ,0.001 the model explaining PSD, trees did not appear to be No. of shrubs 1–2 m 6.038 0.150 0.602 0.031 3.656 0.001 important to poult survival on their own. Number of trees Tree dbh (cm) 0.032 0.005 0.207 0.031 1.257 0.219 and mean diameter at breast height of trees in brood plots a did not differ between broods with below and above average Model: R2 ¼ 0.323, F2,28 ¼ 6.685, P , 0.004. PSD 0–9 days posthatch (Table 6). Neither mean diameter at breast height (F3,27 ¼ 2.07, P ¼ 0.13) nor mean number of (F9,19 ¼ 1.471, P . 0.25). Plots used by broods ,10 days trees (F3,31 ¼ 1.68, P ¼ 0.19) in brood-use plots differed old with above average PSD had more shrubs 1–2 m than among sites. With the exception of honey mesquite, trees in those used by broods ,10 days old with below average PSD these areas were generally limited to riparian areas and (x̄ ¼ 8.3 vs. x̄ ¼ 2.2; P ¼ 0.005; Table 6). windbreaks. Trees most commonly found in brood-use plots Among all plots measured, shrubs 1–2 m tall were more were plains cottonwood at CNG; common hackberry, black common at Salt Fork (x̄ ¼ 4.974 6 0.941) than at walnut, and western soapberry at GHWMA; netleaf GHWMA (x̄ ¼ 2.700 6 0.408; F1,1604 ¼ 4.916, P ¼ hackberry and honey mesquite at MWMA; and Texas 0.027), CNG (x̄ ¼ 0.305 6 0.058; F1,694 ¼ 31.741, P , sugarberry, netleaf hackberry, honey mesquite, and post oak 0.001), and MWMA (x̄ ¼ 2.794 6 0.307; F1,544 ¼ 4.025, P at Salt Fork. ¼ 0.045). The mean number of total shrubs was also higher Forward stepwise regression on habitat variables resulted in plots measured at Salt Fork (x̄ ¼ 32.508 6 4.542) than at in a number of shrubs 2–4 m tall being the only variable GHWMA (x̄ ¼ 13.436 6 1.737; F1,604 ¼ 15.385, P , included in the model predicting 10–16-day PSD (y ¼ 6.505 0.001), CNG (x̄ ¼ 3.224 6 0.304; F1,694 ¼ 53.541, P , 0.386 [shrubs 2–4 m]; R2 ¼ 0.164). However, simple 0.001), and MWMA (x̄ ¼ 9.576 6 1.746; F1,544 ¼ 18.666, P regression analysis showed no significant relationship , 0.001). Mean number of shrubs 1–2 m tall within plots between number of shrubs 2–4 m and PSD during this used by broods was greater at Salt Fork (x̄ ¼ 5.274 6 1.171) period (F2,20 ¼ 3.93, P ¼ 0.061). Furthermore, the than at GHWMA (x̄ ¼ 2.759 6 0.452; F1,444 ¼ 3.933, P ¼ relationship between 10–16-day PSD and number of shrubs 0.048) and at CNG (x̄ ¼ 0.349 6 0.074; F1,513 ¼ 22.468, P 2–4 m did not differ among sites (F6,14 ¼ 1.651, 0.10 . P . , 0.001). Most common shrubs 1–2 m tall in brood-use 0.25). The most common shrub 2–4 m in plots used by Table 6. Differences between habitat variables in plots used by Rio Grande wild turkey broods ,10 days old with below average and above average poult survival days (PSD) on 4 sites in southwest Kansas and the Texas Rolling Plains, USA, May–August, 2000 and 2001. Below average PSD Above average PSD Variable x̄ SE x̄ SE t32 P Welch’s t df P Lowest dm visiblea 2.302 0.111 2.497 0.105 1.227 0.102 Total dm visibleb 8.414 0.130 8.092 0.134 1.624 0.096 % grass coverc 0.469 0.033 0.432 0.034 0.732 0.873 % shrub coverc 0.107 0.022 0.144 0.024 1.048 0.211 % bare groundc 0.101 0.020 0.072 0.010 1.426 0.163 % forb coverc 0.192 0.022 0.227 0.018 0.211 0.235 % litter coverc 0.130 0.024 0.124 0.014 0.257 0.799 No. of shrubs 0.5–1 md 10.783 4.938 36.467 14.846 1.642 24 0.114 No. of shrubs 1–2 md 2.194 0.607 8.338 1.896 3.086 23 0.005 No. of shrubs 2–4 md 0.910 0.268 1.509 0.303 1.361 0.183 No. of shrubs 4–6 md 0.139 0.059 0.215 0.739 31 0.465 Total no. of shrubsd 13.914 5.214 46.663 16.264 1.917 23 0.067 No. of treese 2.540 0.508 3.206 0.497 0.890 0.380 Mean tree decayf 2.114 0.317 2.211 0.209 0.267g 0.792 Mean tree dbh (cm) 24.070 4.323 23.944 2.611 0.025g 0.980 a Mean lowest decimeter visible (all or part) on 10 obstruction pole readings in plots surrounding ground-roost locations from 1-m observation height, 4 m from pole. b Mean total decimeters visible (all or part) on 10 obstruction pole readings in plots surrounding ground-roost locations from 1-m observation height, 4 m from pole. c 100-point sample within plot. d Plot sample within 2 3 20-m belt transect along centerline. e Total in plot. f For all trees in plot. g df ¼ 29. 76 The Journal of Wildlife Management 71(1)
Table 7. Differences between habitat variables in plots used by Rio Grande wild turkey broods 10–16 days old with below average and above average poult survival days (PSD) on 4 sites in southwest Kansas and the Texas Rolling Plains, USA, May–August, 2000 and 2001. Below average PSD Above average PSD Variable x̄ SE x̄ SE t20 P Welch’s t df P a Lowest dm visible 2.508 0.182 2.224 0.112 1.313 0.205 Total dm visibleb 8.231 0.169 8.518 0.148 1.082 0.292 % grass coverc 0.480 0.064 0.449 0.055 0.307 0.762 % shrub coverc 0.132 0.045 0.095 0.024 0.766 0.101 % bare groundc 0.035 0.013 0.091 0.019 0.718 0.094 % forb coverc 0.253 0.042 0.195 0.028 1.109 0.281 % litter coverc 0.101 0.132 0.169 0.026 2.316 19 0.031 No. of shrubs 0.5–1 md 42.405 22.700 10.977 5.535 1.345 5 0.231 No. of shrubs 1–2 md 5.064 2.125 2.348 0.874 1.421 0.171 No. of shrubs 2–4 md 2.036 0.922 0.549 0.147 1.593 5 0.169 No. of shrubs 4–6 md 0.550 0.238 0.174 0.095 1.786 0.089 Total no. of shrubsd 50.055 24.986 14.024 6.334 1.398 5 0.215 No. of treese 2.628 0.548 4.533 0.953 1.733 19 0.098 Mean tree decayf 2.271 0.301 2.198 0.186 0.209g 0.837 Mean tree dbh (cm) 28.801 5.360 29.068 4.311 0.035g 0.973 a Mean lowest decimeter visible (all or part) on 10 obstruction pole readings in plots surrounding ground-roost locations from 1-m observation height, 4 m from pole. b Mean total decimeters visible (all or part) on 10 obstruction pole readings in plots surrounding ground-roost locations from 1-m observation height, 4 m from pole. c 100-point sample within plot. d Plot sample within 2 3 20-m belt transect along centerline. e Total in plot. f For all trees in plot. g df ¼ 19. broods 10–16 days old was tamarisk, composing 38% of all did not tree-roost until all poults in a brood were capable of shrubs 2–4 m present, followed by mesquite (22.5%), reaching a suitable roosting spot. Delayed tree-roosting western soapberry (15%), hackberry (7%), shinnery oak would subject poults capable of roosting in trees to greater (6%), and woolybucket bumelia (Bumelia lanuginosa; 6%). risks of predation on the ground, enhancing the need for The only habitat variable that differed between plots used by high-quality ground-roost habitat until all poults can tree- broods with below and above average PSD 10–16 days roost. posthatch was percent litter cover, with more litter being Brooding hens at 3 of our sites selected ground-roosting present in plots used by more successful broods (x̄ ¼ 17% vs. areas with increased hiding cover 0.5–2 m tall. This cover on x̄ ¼ 10%; P ¼ 0.031; Table 7). our study sites was provided mostly by shrubs. Brooding Successful broods appeared to shift habitat use as poults hens at CNG also selected areas with more coarse woody began to attain flight abilities. Plots used by ground- debris and decayed trees. Fallen trees were in a more decayed roosting broods 0–9 days old with above average PSD had state than standing trees, and thus the presence of fallen more visual obstruction (x̄ ¼ 8.1 obstruction bands visible vs. trees increased the average index of tree decay for a plot. x̄ ¼ 8.5; P ¼ 0.050), more shrubs 1–2 m (x̄ ¼ 8.6 vs. x̄ ¼ 2.3; Selection for roost areas with increased woody debris and P ¼ 0.007), and more shrubs 2–4 m (x̄ ¼ 1.5 vs. x̄ ¼ 0.5; P ¼ tree decay suggests that brooding hens also sought out fallen 0.007) than did ground-roosting broods 10–16 days old branches and trees as ground-roosting cover. with above average PSD (Table 8). Number of shrubs 1–2 m tall was the most important habitat variable explaining poult survival 0–9 days posthatch DISCUSSION at all of our study sites. Shrubs of this height category were Rio Grande wild turkey broods in our study appeared to sufficiently large to provide hiding cover for both poults and seek out specific habitat components providing ground-level the brooding hen, and provided vegetation into which horizontal visual obstruction, both for ground roosting and poults could escape. Radioequipped broods located in and use throughout the day. These components appeared among shrubs of this height were often completely particularly important in facilitating poult survival ,10 concealed. Mean PSD 0–9 days posthatch at Salt Fork days after hatch, the period prior to general poult flight was greater than at GHMWA and CNG. Correspondingly, attainment during which poult mortality is the highest. number of shrubs 1–2 m in areas used by broods was higher Average time to first brood tree-roost in our study was at Salt Fork than any other site. This correlation further between 13 days and 14 days posthatch, similar to that suggests the importance of the presence of shrubs in this observed in other subspecies (Barwick et al. 1970). Broods height category to poult survival during this time period. continued to ground roost after individual poults within the Rio Grande and eastern (M. g. silvestris) wild turkey brood were observed flying to trees. This suggests that hens preflight broods in other studies used areas with greater Spears et al. Wild Turkey Brood Habitat Use and Survival 77
Table 8. Differences between habitat variables in plots used by Rio Grande wild turkey broods 0–9 days old with above average poult survival days and broods 10–16 days old with above average poult survival days on 4 sites in southwest Kansas and the Texas Rolling Plains, USA, May–August, 2000 and 2001. 0–9 d old 10–16 d old Variable x̄ SE x̄ SE t34 P Welch’s t df P a Lowest dm visible 2.467 0.105 2.221 0.108 1.617 0.115 Total dm visibleb 8.101 0.140 8.518 0.148 2.033 0.050 % grass coverc 0.429 0.036 0.449 0.055 0.330 0.743 % shrub coverc 0.148 0.025 0.095 0.376 1.777 0.142 % bare groundc 0.072 0.011 0.091 0.019 0.0931 0.358 % forb coverc 0.226 0.019 0.195 0.028 0.953 0.347 % litter coverc 0.124 0.015 0.167 0.026 1.570 0.126 No. of shrubs 0.5–1 md 37.812 15.544 10.978 5.535 1.626 23 0.117 No. of shrubs 1–2 md 8.648 1.967 2.348 0.874 2.927 25 0.007 No. of shrubs 2–4 md 1.549 0.316 0.549 0.147 2.874 26 0.007 No. of shrubs 4–6 md 0.218 0.088 0.174 0.095 0.346 0.732 Total no. of shrubsd 48.368 17.005 14.024 6.334 1.893 24 0.071 No. of treese 3.331 0.506 4.533 0.953 1.114 23 0.277 Mean tree decayf 2.293 0.203 2.198 0.186 0.336g 0.739 Mean tree dbh (cm) 23.481 3.831 29.068 4.311 1.143g 0.262 a Mean lowest decimeter visible (all or part) on 10 obstruction pole readings in plots surrounding ground-roost locations from 1-m observation height, 4 m from pole. b Mean total decimeters visible (all or part) on 10 obstruction pole readings in plots surrounding ground-roost locations from 1-m observation height, 4 m from pole. c 100-point sample within plot. d Plot sample within 2 3 20-m belt transect along centerline. e Total in plot. f For all trees in plot. g df ¼ 32. herbaceous vegetative height and visual obstruction 1 m were located hiding up to 15 m from hiding brooding hens, (Metzler and Speake 1980, Healy 1985, Hennen 1999). suggesting that hens and poults hid in the immediately Trees and shrubs have been observed to create the majority available vegetation upon being alerted. Moreover, preflight of ground cover at this height in other wild turkey habitats poults did not have the ability to move large relative (Schmutz et al. 1990). distances to find cover within the time between being Average diameter at breast height of trees in brood use alerted by the hen and being located by the observer. areas was positively correlated with poult survival 0–9 days Components other than available habitat, such as weather posthatch, but was not different between sites with different (Roberts and Porter 1998) and hen ability to raise a brood, survival rates or between broods with below average and may also affect poult survival. We were not able to above average PSD. Increased diameter at breast height does determine important weather variables, such as rainfall, increase visual obstruction and thus ground-level hiding accurately in specific brooding areas and were therefore cover. We observed broods ground roosting against large tree unable to determine effects on survival. No difference bases and under or against downed logs. Hen coloration existed in our study in poult survival to 16 days of age blended well with logs and tree bases, and trees that hens between juvenile and adult hens (Spears 2002), suggesting ground roosted against often provided 100% visual obstruc- inexperience in raising a brood was not a factor in poult tion 1 m from 2 cardinal directions. survival. Hen weight can be used as an index of health. An Shrubs in the 2–4-m height category appeared unsuitable increase in hen weight at initial capture (1–2 winters prior to for poult hiding cover. Most shrubs 2–4 m in plots used by that breeding season) appeared to affect poult survival (P , ground-roosting broods 10–16 days of age were tamarisk, 0.0001). However, this effect in our model is related to followed by mesquite and soapberry. Tamarisk shrubs change per kilogram in hen weight. Hens averaged 4.8 kg at consisted of dense stems that did not provide space at time of capture. One kilogram would represent a change of ground level in which poults could hide, whereas mesquite approximately 20% of hen body weight. and soapberry trunks did not provide cover at ground level Studies of eastern wild turkeys suggested that brooding into which poults could enter. hens selected clearings (Sisson et al. 1991) for their Hens in habitats with dense brush were often able to increased level of invertebrate availability for poults (Hill- observe the researcher and warn poults to hide before the estad and Speake 1970, Hurst and Stringer 1975, Martin researcher located the brood. This problem is unavoidable in and McGinnes 1975). Our study and others support an a study such as ours and may bias brood locations toward alternate hypothesis: preflight broods use clearings for areas with more visual obstruction. However, we observed predator avoidance, facilitated by their ground-level vege- poults of this age most commonly hiding immediately in tation components. First, adequate insect abundance for situ when warned by the hen. Transmitter-equipped poults poults is available in wooded areas. Harper et al. (2001) 78 The Journal of Wildlife Management 71(1)
found that biomass of Hexapoda, insects preferred by turkey predator avoidance. However, certain lower-quality vegeta- poults, was not different among forest stands of different tive species (i.e., tamarisk) may replace that which could ages and openings, and found twice as many invertebrates in provide desired cover. forest strata than in openings. Hurst and Stringer (1975) Low-growing shrubs appeared to provide the optimal observed that average animal percent intake by poults ,1 escape services for preflight poults in the southern Great week old was higher in pine forest and hardwood cover types Plains habitats we studied. Managers desiring to increase than in forest openings. Martin and McGinnes (1975) Rio Grande wild turkey preflight poult survival in these found that, although insect abundance was higher in forest areas should manage for increased shrubs as hiding cover in clearings than forest, poults consumed the same amount of brooding areas from May through the end of July, the peak insects in both cover types. Healy (1985) found that poults periods of hatching for Rio Grande wild turkey populations. obtained adequate insect intake in forested areas if ground- Shrubs we observed providing high-quality hiding or escape cover dry weight was 40–300 g/m2. Second, we found that cover included shinnery oak and sand sagebrush. Shrubs we ground-level vegetation characteristics were directly related observed providing little or no services to preflight poults to poult survival. Relationships between poult survival and included tamarisk, mesquite, and western soapberry. Tam- ground-cover characteristics in forest stands and openings arisk and mesquite were especially notorious for invading was not examined in some studies that concluded broods and dominating shrub habitats in these areas. Activities to selected forest openings or pastures for insect abundance control or remove these shrubs should be considered. Upper (e.g., Hillestad and Speake 1970, Martin and McGinnes limits of ground-level vegetation appear to exist for wild 1975). Clearings often provided ground-level vegetative turkey poult productivity (Healy 1985). This needs to be structure that allowed brooding hens to detect predators and examined for Rio Grande wild turkey broods. cover that facilitated poult predator avoidance (Pack et al. Previous habitat management suggestions for wild turkey 1980, Sisson et al. 1991, Harper et al. 2001). Preflight broods included maintaining a diverse mosaic of high- broods also selected for wooded cover types when they quality habitats (Metzler and Speake 1980, Campo et al. provided understory vegetative structure similar to that of 1989). Relationships between survival and habitat parame- clearings. Rio Grande wild turkey broods in Colorado and ters in our study support these recommendations. We Kansas selected areas containing wooded riparian areas observed a shift in habitat type use by successful broods not when this cover type was not readily available (Schmutz et yet tree-roosting occurring between 0–9 days and 10–16 al. 1990, Hennen 1999). Broods 1–14 days old in days after hatch. Successful broods ,10 days old used plots Mississippi, USA, selected for bottomland hardwood cover with more ground-level visual obstruction and shrubs 1–2 m types, which had ‘‘moderately dense herbaceous ground than successful broods 10–16 days old. At 10–16 days of age, cover that resembled that of an open field habitat’’ (Phalen most poults can fly to trees. This ability may reduce the need et al. 1986:402). Hubbard et al. (2001) indicated that for ground-level cover, but increases the importance of increasing woody cover in an area occupied by transmitter- roosting trees. Interspersion of shrubs 1–2 m with trees in equipped poults increased survival. Most woody cover on brood-use areas appears optimal for poult survival to 16 days their study site included a ‘‘diverse vegetative understory,’’ posthatch in these habitats. which ‘‘may have contributed to decreased poult mortality’’ Forested areas may contain ground-level vegetation (Hubbard et al. 2001:169). characteristics that adequately facilitate preflight wild turkey In areas where forest stands do not include high-quality poult survival. These areas may already constitute high- ground-level habitat, preflight turkey broods may ultimately quality foraging areas and hiding cover for preflight broods, use clearings as a refuge from predation, not for their relative as well as a variety of vertical habitat structure (i.e., mix of insect abundance, especially in areas where high predator shrubs and trees) for use by broods of varying ages and flight densities occur in wooded habitats (Priest et al. 1995). Use abilities. Thus, forest clearings may not necessarily be a of clearings for predator avoidance or insect abundance can required land-cover type in high-quality turkey brood only be assessed through examination of relationships habitat. Managers of forested areas should consider under- between forest and clearing vegetative parameters, poult story characteristics present and the services they provide to survival, and insect abundance in different cover types. turkey broods prior to clearing areas for the sake of Knowledge of actual reasons for selection of specific habitat providing clearings. types may allow managers to make better decisions regarding habitat management for turkeys based on ACKNOWLEDGMENTS individual area characteristics. We thank T. Barnett and numerous field technicians that aided in data collection. Texas Parks and Wildlife Depart- MANAGEMENT IMPLICATIONS ment, Kansas Department of Wildlife and Parks (Federal Ground-level vegetation structure appears to be a key factor Aid Grant W-54-R), the National Wild Turkey Federation, in facilitating wild turkey poult survival. In terms of wild and the Texas Chapter (Superfund) of the National Wild turkey ecology, physiognomy may be more important than Turkey Federation provided equipment, logistical help, and floristics; a variety of species and vegetation types may funds. Employees from Texas Parks and Wildlife Depart- provide the hiding and escape cover necessary for poult ment, Kansas Department of Wildlife and Parks, and Spears et al. Wild Turkey Brood Habitat Use and Survival 79
You can also read