ROOSTING HABITAT OF MERRIAM'S TURKEYS IN THE BLACK HILLS, SOUTH DAKOTA

This file was created by scanning the printed publicatin. Errrs identified by the sftware have been crrected; hwever, sme errrs may remain. ROOSTING HABITAT OF MERRIAM'S TURKEYS IN THE BLACK HILLS, SOUTH DAKOTA MARK A. RUMBLE, USDA, Frest Service, Rcky Muntain Frest and Range Experiment Statin, 501 E. St. Je, Rapid City, SD 57701 Abstract: Lack f rst habitat (trees -40 cm diameter breast height [dbh] and _18 m2/ha basal area) can limit ppulatins f Merriam's turkeys (Meleagris gallpav merriami). The Black Hills regin has relatively large ppulatins f Merriam's turkeys, yet trees _40 cm dbh are uncmmn. Cnsequently, I studied rsting habitat f this subspecies in a hierarchical manner t quantify rst habitat requirements in an area f apparent limited suitable rst habitat. Little r n selectin fr rsts ccurred amng macrhabitats. Basal area at rst sites averaged 19-25 m2/ha. Winter and summer (excluding hens with pult) rst sites were mre similar than rst sites selected by hens with pults r randm sites. Vegetative characteristics at rst plts shwed trends tward trees with larger dbh, lwer tree density (stems/ha), and higher basal area (m2/ ha). Rst trees averaged 35 cm dbh, but trees >23 cm dbh were used. Rst trees had layered hrizntal branches and ften large dbh, but large dbh was nt a prerequisite fr rst trees. Timber management practices in the Black Hills that mdify the frest belw 21 m2/ha and decrease the number f 25-35 cm dbh trees will reduce rsting habitat fr Merriam's turkeys. J. WILDL. MANAGE. 56(4):750-759 Rsts are apparently imprtant t sustaining ppulatins f turkeys (Beker and Sctt 1969, Mackey 1984, Kilpatrick et al. 1988). Merriam's turkeys abandned areas in Arizna where basal area at rst sites was reduced t 16.8 m2/ha (73 ft2/acre) (Sctt and Beker 1975). Bryant and Nish (1975) partially attributed nnuse f pinn-juniper (Pinus edulis-juniperus spp.) habitats t the lack f suitable rst sites. Rst trees selected by Merriam's turkeys typically have been large (>40 cm dbh), mature, r vermature (large diameter ld trees with flat tps and large hrizntal branches) pndersa pine (P. pndersa) (Hffman 1968, Beker and Sctt 1969, Phillips 1980). Narrwleaf cttnwd (Ppulus agustiflia), Engelmann spruce (Picea engelmannii), white fir (Abies cnclr), and Duglas-fir (Pseudstuga menziesii) als are used fr rsting (Hffman 1968, Mackey 1984, Lutz and Crawfrd 1987). Trees >40 cm dbh are uncmmn in the Black Hills, but the area supprts large and sustaining turkey ppulatins. My bjective was t describe, in a hierarchical manner, the rsting habitat f Merriam's turkeys in an area where large (?40 cm dbh) trees were in shrt supply. This research was supprted by the USDA Frest Service, Rcky Muntain Frest and Range Experiment Statin; Black Hills Natinal Frest; Natinal Wild Turkey Federatin; and Suth Dakta Game, Fish and Parks. Dr. A. J. Bjugstad (deceased) prvided initial advice and encuragement. Technical assistance was prvided by R. A. Hdrff, T. R. Mills, C. D. Oswald, K. J. Thrstensn, K. L. Jacbsn, and L. J. Harris. M. P. Green vlunteered his time thrughut this study, and R. L. Taylr allwed access t his prperty. Appreciatin is extended t W. C. Aney, J. G. Dicksn, C. B. Edminster, R. W. Hffman and 2 annymus referees fr review f manuscript drafts. STUDY AREA AND METHODS I cnducted this study fr 5 years (Mar 1986- Jan 1991) in the central Black Hills f Suth Dakta. Elevatin ranges frm apprximately 1,300 t 1,800 m abve sea level. Mst f the land was managed by the Black Hills Natinal Frest, Pactla Ranger District. Private hldings assciated with ranch peratins ccurred in sme meadws and several private hmes and cabins ccurred within the study area. Vegetatin f the study area was pure pndersa pine frest (84%), with meadws and aspen (P. tremulides)-birch (Betula papyrifera) vegetatin in drainages. Sme mntypic aspen stands ccurred n nrthern expsures. Bur ak (Quercus macrcarpa) and white spruce (P. glauca) cmprised <1% f the study area. Sme turkeys in the Black Hills used ranch feed lts and suburban husing develpments fr winter feeding; thers remained in the frest thrughut winter r until deep snw frced them t use ranches (Petersen and Richardsn 1975). My research was cnducted n birds in the latter categry. I trapped turkeys frm late February t early March each year f the study using alpha-chl- 750

J. Wildl. Manage. 56(4):1992 MERRIAM'S TURKEY ROOSTING HABITAT * Rumble 751 ralse (Williams 1966), drp nets, r rcket nets. Eighty-eight individual turkeys (63 females and 25 males) were fitted with back-pack munted radi transmitters having a mass f apprximately 108 g. The fewest number f marked birds in any year was 13. Actual number f birds in the field at any time was a functin f survival (up t 4-5 yr) and annual mrtality. During the first 3 years, I lcated apprximately 1 rst/mnth; during the last 2 years, apprximately 1 rst/week was lcated. Turkeys are gregarius with sme mingling and exchange amng flcks. Usually >1 radimarked bird was lcated at a rst. During winter, birds ccasinally used the same rst repeatedly. Individual rst sites were included in the sampling nly nce, and c-ccurrence f radi-marked birds at a rst site was treated as 1 bservatin. I selected rsts frm different flcks t ensure adequate representatin f all radi-marked birds. Over the curse f the study, 158 rsts were lcated. Habitat Descriptins Macrhabitats. -I determined turkey rsting habitat using a hierarchical apprach t habitat selectin (Jhnsn 1980). Macrhabitats were the lwest level f habitat delineatin and crrespnded t third-rder habitats (Jhnsn 1980). Macrhabitats were numerically identified gegraphical units, the bundaries f which were defined by watershed tpgraphy (ridges and drainages) r distinct changes in vegetatin type. Typically, these were 4- t 32-ha land units; althugh smaller size macrhabitats were delineated if distinct vegetatin types such as aspenbirch r meadws culd be identified n 1:24,000 aerial phtgraphs. I assigned private lands in the study area t macrhabitats based n interpretatin f aerial phtgraphs; bundaries f adjacent macrhabitats were extended if the vegetatin type was cntinuus, r new bundaries were assigned if changes in vegetatin were apparent. I delineated 9 macrhabitat categries and 513 gegraphical units fr my study. Macrhabitats were described vegetatively based n dminant species f vegetatin, diameter breast height (dbh), and verstry canpy cver (Buttery and Gillam 1983). Dminant vegetatin type included pndersa pine, aspen-birch, ak, spruce, and meadws. Dbh categries included 2.5-22.9 cm and >22.9 cm; verstry canpy cver included 0-40, 41-70, and 71-100%. I estimated verstry canpy cver f macrhabitats frm the fllwing relatinship develped fr the Black Hills: verstry canpy cver(%) = 2.23*basal area(m2/ha) - 1.94 (Bennett 1984). These prcedures fr describing macrhabitats are used by Rcky Muntain Regin Natinal Frests fr mdelling wildlife habitat relatinships. Micrhabitats. I evaluated micrhabitat at rsts at 3 scales f reslutin (rst site, rst plt, and rst tree), which represented furth and higher rders f habitat selectin (Jhnsn 1980). Vegetatin measurements at micrhabitats included basal area (m2/ha), density f trees (stems/ha), average dbh, slpe, aspect, and verstry canpy cver. Three plts were sampled at each rst site: ne at the rst, and ne 30 m in each directin n the cntur. Basal area, tree density, and dbh were recrded fr trees using a 10-factr prism t determine sample trees at each plt. Aspect was determined frm a dwn-hill cmpass bearing, percent slpe was estimated with a clinmeter, and verstry canpy cver (%) was estimated with a spherical densimeter (Lemmn 1956, Griffing 1985). During the last 2 years f the study, I expanded micrhabitat data cllectin t include silvicultural prescriptin, time since cut, stand structure, lcatin n the slpe, fuels (tns [M]/ ha), and rst tree characteristics. I classified prescriptins as: n cut, n evidence f past cutting at the rst site; clearcut, mst r all trees remved in an area pattern (ccasinally sme small diameter trees wuld be left); cmmercial thin, evidence that past cutting had remved several t mst mature trees (included past selective harvest); shelterwd seed cut, mst trees harvested with remaining few trees f mature seed-prducing size (basal area <9 m2/ha); and precmmercial thinning, small diameter trees cut and left in the frest. Time since cut was categrized as <2 years, 2-5 years, and >5 years. I classified stand structure as multistry r nt based n the presence f 2 r mre layers t the tree canpy. Lcatin f rst n the slpe was determined amng 6 exclusive categries; ridge, tp 25%, upper 25%, lwer 25%, bttm 25%, and bttm. I estimated fuels fr 2.5 t 7.6-cm and >7.6-cm diameter categries based n pictrial guides develped fr the Black Hills (USDA Fr. Serv., Rcky Mt. Reg. 1982). Rst tree height, rst height, and spacing between branches at rst height were determined frm cular estimates. The same micrhabitat sampling scheme was

752 MERRIAM'S TURKEY ROOSTING HABITAT * Rumble J. Wildl. Manage. 56(4):1992 used at 220 randm sites within the study area. Randm sites were lcated based n stratified randm sampling f macrhabitats (described abve). Data pertaining t psitin n slpe, tree height, rst height, and branch spacing were nt cllected at randm sites. Analyses Habitat selectin f hens with pults is distinct frm ther turkeys until pults are apprximately 12 weeks ld (Rumble 1990). Stratificatin f data by sex r age (subadult-adult) f turkeys during winter was impssible because mixed flcks frequently rsted tgether, and I culd nt crrectly identify all birds. As a result, I used the fllwing categries in analyses: winter, summer (excluding hens with pults), and hens with pults. Macrhabitats. -Chi-square tests attempt t fit a specified mdel t a data set, and significance indicates deviatins frm the expected fit under the mdel. The mdel in each test was n interactin amng factrs f the cntingency table, and hyptheses were stated in a psitive sense (Steel and Trrie 1980:496-499). I used Chi-square tests f independence t test hyptheses that macrhabitats used fr rsting by turkeys (excluding hens with pults) were similar amng seasns and that macrhabitats used fr rsting by hens with pults were similar t ther turkeys. Chi-square gdness-f-fit tests crrected fr cntinuity (Cchran 1963:57) were used t test hyptheses that selectin f rsts by Merriam's turkeys amng macrhabitats was similar t randm expected use. In the verall Chi-square tests, I cmbined ak, spruce, and aspen habitats t reduce the number f cells with <5 expected bservatins. These, and ther macrhabitats were cnsidered separately with Bnferrni cnfidence intervals arund prprtinal use t determine which macrhabitats were selected disprprtinately frm expected use (Neu et al. 1974, Byers et al. 1984). Micrhabitats.-I weighted randm data t accunt fr deviatins frm prprtinal sampling f macrhabitats in the analyses. Analysis f variance prcedures were used t test the hypthesis that vegetative characteristics at rst sites did nt differ frm randm sites. I analyzed data that did nt adhere t hmgeneity f variance assumptins amng grups using Welch's test, which is recmmended when variances and sample sizes are nt equal (Milliken and Jhnsn 1984). Paired t-tests were used t test hyptheses that basal area, tree density, dbh, and verstry canpy cver did nt differ between rst plts and plts 30 m away. I used Chi-square tests fr independence t test hyptheses that aspect, prescriptin, time since cut, and stand structure were similar amng rsts selected during winter, summer, and by hens with pults. Chi-square gdness-f-fit tests were used t test hyptheses f independence f these variables at rsts frm randm sites. I used Bnferrni cnfidence intervals (Neu et al. 1974) and standardized Chi-square residuals with a Bnferrni crrectin t the Z-statistic (if bserved use was zer; Msteller and Parunak 1985) t determine categries f these variables that deviated frm randm. Terminlgy used in this study fllws Jhnsn (1980) and Thmas and Taylr (1990). Habitat use implies utilizatin that was nt cmpared t availability. Habitat selectin implies use that was nt cmpared t availability. Habitat preference required differential selectin f resurces given equal availability. Tests f hyptheses fr aspect, time since cut, and stand structure cnsidered all pssible respnses and significant deviatins frm randm inferred preference r avidance fr these variables. Statistical significance in this study was determined at a < 0.10. Csts f Type II errrs in the analyses were weighed against attaining higher prbability cnfidence intervals. Because this research was directed tward prviding managers with infrmatin regarding the effects f frest management n turkeys, Type II errrs wuld be equivalent t incrrectly suggesting turkeys use habitats randmly. Type II errrs wuld result in lack f management in frest ecsystems t enhance r maintain turkey habitat. RESULTS Macrhabitats The use f macrhabitats by turkeys did nt differ (P = 0.82) between summer (excluding hens with pults) and winter. Therefre, these data were pled fr further tests. Macrhabitats used fr rsting by hens with pults differed (P = 0.03) frm thse used by ther turkeys. Selectin f macrhabitats fr rsting by turkeys during summer and winter differed frm randm (P = 0.04). Hwever, few differences were apparent when prprtinal use f individual macrhabitats was cmpared t prpr-

J. Wildl. Manage. 56(4);1992 MERRIAM'S TURKEY ROOSTING HABITAT * Rumble 753 Table 1. Selectin f macrhabitats fr rst sites by Merriam's turkeys in the Black Hills, Suth Dakta, 1986-91. Number f rsts Percent canpy Habitat Dbh cver Prprtinal area Hens with pultsa Summer/winterb Meadw 0.1016 2 le Pine 2.5-22.5 cm 0-40 0.0701 1 14 41-70 0.1677 8 24 >70 0.2173 4 24 Pine >22.5 cm 0-40 0.0498 1 2 41-70 0.2083 19 33 >70 0.1239 1 18 Otherd 0-100 0.0616 3 3 a Pults were?12 weeks ld. b Summer and winter were pled because n differences (P 0.10) ccurred. c Habitats selected less (P < 0.10) than expected. d Includes all dbh and verstry categries f aspen, ak, and spruce vegetatin types. tinal area, and selectin f meadws less than available was the nly statistically significant deviatin (Table 1). Macrhabitats selected by hens with pults were nt different frm randm (P = 0.16). Micrhabitats Rst Sites.-During summer, basal area was higher (P = 0.01) at rst sites selected by hens with pults than at rst sites selected by ther turkeys (Table 2). Basal area at winter and summer rsts and randm sites did nt differ (P > 0.10). Density f trees did nt differ (P _ 0.10) between summer and winter rst sites, but tree density at these rsts was lwer (P < 0.10) than at rst sites f hens with pults and at randm sites. Average dbh f trees at winter rst sites was greater (P _ 0.10) than at rst sites selected by hens with pults; average dbh at summer rsts did nt differ (P < 0.10) frm that f winter r pult rsts. Average dbh at randm sites was less (P < 0.10) than at all rst sites selected by turkeys. Slpe was similar amng all rst sites but greater (P < 0.001) than at randm sites. Overstry canpy cver was greater (P < 0.10) at winter rsts than at randm sites, r rsts selected by hens with pults. Small diameter fuels (2.5-7.6 cm) were nt different (P < 0.10) amng all rst sites, but were greater (P = 0.02) at randm sites than at sites selected fr winter rsts. N differences (P = 0.27) were fund fr large diameter fuels (>7.6-cm diameter). Chi-square analysis indicated that aspect, 50 60 ez Rst * Randm 040 (j) 10 1 c U1) Nrth East Suth West CL NO CT SC PT s Aspect Prescriptin 30-30 20-60- 50-40- ~ 30.70 5). < 2 a2r yr 2-5 25y yr 5 > yr 5 yr Single stry Multistry Time Tiesnecut since Stand structure Fig. 1. Distributin f aspect (n = 158), prescriptin (n = 95), CL = clearcut, NO = n evidence f timber harvest, CT= cmmercial thin, SC = shelterwd seed cut, PT = precmmercial thin), time since cut cut (n (n = = 44) 44) and and stand structure (n (n = 95) = 95) at rst and randm sites in the Black Hills, Suth Dakta, 1986-91. Asterisk indicates rst and randm sites differed (P (P I _ 0.10).

(0 754 MERRIAM'S TURKEY ROOSTING HABITAT * Rumble J. Wildl. Manage. 56(4):1992 Table 2. Micrhabitat characteristics (R + SE) f rst sites and randm sites in the Black Hills, Suth Dakta, 1986-91.a Habitat characteristicb Winter Summer Hens with pults Randm Basal area (m2/ha) 22.4? 1.1ABe 19.4? 1.1A 25.9 + 1.7B 21.1 + 0.8AB Tree density (trees/ha) 598.2 + 78.1A 534.4 + 97.1A 973.3? 142.8B 1,001.6 + 75.1B Diameter breast height (cm) 26.9 + 0.8A 26.6 + 0.9AB 23.3 + 1.1B 19.7 + 1.6C Slpe (%) 28.2 + 1.4A 30.3 _ 2.2A 29.4? 2.2A 22.9 + 0.9B Overstry canpy cver (%) 58.3 + 1.4A 53.1 + 2.0AB 52.6? 2.6AB 46.9? 1.5B Fuels 2.5-7.6 cm [tns(m)/ha] 2.6? 0.5A 2.4 + 0.4AB 2.4? 0.8AB 3.8? 0.2B Fuels >7.6 cm [tns(m)/ha] 6.0? 0.9A 5.2? 0.9A 7.2 + 2.5A 7.0 + 0.3A an = 71, 48, and 39 fr winter, summer, and brds, respectively, fr basal area t verstry canpy cver, n = 49, 34, and 6 fr winter, summer, and brds, respectively, fr fuels variables, n = 220 fr all variables frm randm sites. Different sample sizes ccurred because variables were added after the first 2 years. b Cnversins frm metric t English units are: basal area (m2/ha) = 0.2291*ft2/acre; density (trees/ha) = 2.471*trees/acre; dbh (cm) = 2.54*inch; and fuels (metric tns/ha) = 2.24*tns/acre. crw means with different letters differ (P _ 0.10); ANOVA r Welch's test. prescriptin, time since cut, and stand structure did nt differ (P > 0.40) amng rsts selected during summer, winter, r by hens with pults. Thus, these data were pled fr cmparisns with randm sites. Mst (42%) rsts were lcated n slpes facing nrtheast t sutheast (Fig. 1). Hwever, n (P = 0.24) preference relative t aspect was nted. Of rsts lcated in areas f prir timber harvest activity, mre ccurred in stands that had been cmmercially thinned (selectively cut) and fewer ccurred in precm- 40 30 0) 20 a 10 * l L) a, 1 30)I 1 2 3 4 5 6 7 8 9 C - Pstitin n slpe Fig. 2. Distributin f Merriam's turkey rst trees per site (n = 158) and lcatin n slpes (n = 95) in the Black Hills, Suth Dakta, 1986-91. mercially thinned stands (P < 0.10) than at randm. Turkeys tended t select rsts in stands with n evidence f timber harvest (57%) r that had been harvested at >5 years prir (78%). Neither f these latter variables differed frm randm (P > 0.10). Merriam's turkey avided (P = 0.0008) rsting in even-aged r single stry stands. The number f rst trees per site shwed a significant negative expnential distributin (r2 = 0.97, P < 0.001) (Fig. 2). Apprximately 80% f the rst sites had <3 rst trees, with nearly 40% having nly 1 rst tree. Mre (P < 0.001) rst trees/site were fund at winter rsts (I + SE = 3.2? 0.2) than summer rsts (?? SE = 2.2? 0.3); bth f which were greater than sites selected by hens with pults (f + SE = 1.5? 0.1). Eighty-five percent f all rsts were lcated n the upper half f slpes r n ridges. Rst Plts.-There was a trend tward greater basal area, lwer density f trees, and larger dbh at plts cntaining rsts versus thse n adjacent plts (Table 3). Basal area was higher at plts cntaining the rst tree than plts 30 m away at summer rsts (P = 0.08) and rsts selected by hens with pults (P = 0.06). During winter, density f trees at rsts was lwer (P = 0.03) than n adjacent areas. Dbh at rst plts during winter and summer was greater (P = 0.04) than n adjacent plts. Hens with pults selected rsts with greater (P = 0.002) verstry canpy cver than n adjacent plts. Otherwise n differences in verstry canpy cver were evident at rsts versus adjacent plts. Rst Trees.-Mean dbh f rst trees (?? SE = 37 + 0.5 cm) was nt different amng summer, winter, r hen with pult rsts (P = 0.21). Hwever, the distributin f dbh fr trees

D Cr) t t-- 0 J. Wildl. Manage. 56(4):1992 MERRIAM'S TURKEY ROOSTING HABITAT * Rumble 755 m 20 Winter.) 16-0;1 O 12-0 I l. 0 +1 +1 +1 +1 0O01 0) 40 Summer?..3 s O CC20 ni 20l- )-- r O ~ +1+1l -I+1+ C) cd d' co 0 C 8 24 Hens with pults S20 c" U C._ C)J 16-.3 +1+1+1+1I 4- (00 O L E 6 a e6cj 10~ 01 10 I 10 i VI 23 25 28 30 33 36 38 41 43 46 48 51 53 56 Dbh (cm) E C', Sl S t~ 0, Fig. 3. Dbh f rst trees at winter (n = 80), summer (n = 39), and hen with pult (n = 39) rsting sites f Merriam's turkeys in the Black Hills, Suth Dakta, 1986-91. V) n. C '13 (N 0001 10 0. 00 selected fr rsts was visibly different (Fig. 3). N differences in branch spacing (P = 0.80) r tree height (P = 0.13) ccurred amng rst categries. Mean (?SE) branch spacing and tree height were 0.9? 0.1 m and 27.0? 0.4 m, respectively. Hens with pults rsted higher (P = 0.03) in trees (13.4 + 0.5 m) than ther turkeys during winter (11.5? 0.5 m). Summer rsts averaged 12.6? 0.6 m frm the grund. 0 E 'i.h 03.d +1+1+1+1 1- C) C)l CO 01001 CD <10')1 < ~cd0 CU SE "'S -0 U +1- S.- DISCUSSION Macrhabitats Similar use f macrhabitats between summer and winter fr rsting by turkeys in my study cntrasted with results f Lutz and Crawfrd (1987) wh reprted decreased use f mixed cnifer habitats fr rsting frm winter (92%) t summer (59%). These differences may have resulted frm snw accumulatins in the muntains f Oregn that wuld frce turkeys t migrate t lwer elevatins during winter. Turkeys in my study did nt demnstrate large elevatinal migratins, prbably because snw accumulatins ver 20 cm were infrequent and 4) 0 F- 0 rj ci I r 01 C- O 11 'U. -1. >de U~ c3c- i E, 0 " -0a l 0)0 CC)P SOK

756 MERRIAM'S TURKEY ROOSTING HABITAT * Rumble J. Wildl. Manage. 56(4):1992 shrt term, and because nly a 500-m change in elevatin was pssible. Outside f infrequent use f islated trees in meadws, n patterns f selectin fr rsting were evident at the macrhabitat scale f reslutin. Lutz and Crawfrd (1987) and Mackey (1984) als reprted few deviatins frm expected use f lw reslutin habitats fr rsting by turkeys. Micrhabitats Rst Sites.-The ability t relate vegetative characteristics f rsts amng studies is difficult because f incnsistencies in the variables reprted. Basal area at rst sites in my study was similar t that reprted in Arizna (Sctt and Beker 1975), but was substantially less than the 33 m2/ha reprted fr turkeys rsting in Duglas-fir in Washingtn (Mackey 1984). Overstry canpy cver was lwer than reprted by Mackey (1984), but was greater than reprted by Lutz and Crawfrd (1987). Patterns f similarity r differences between basal' area and verstry canpy cver are expected since these 2 variables are psitively crrelated (Bennett 1984, Hver and Wills 1984:411). Densities f trees at rst sites in my study were greater than thse reprted by Hffman (1968) fr rst sites in Clrad. Merriam's turkeys usually select rst sites n mderately steep (20-30%) slpes (Jnas 1966, Lutz and Crawfrd 1987, this study), but ccasinally use relatively gentle slpes (5%) (Sctt and Beker 1975). N differences in percent slpe were fund at rst sites in my study. Hwever, slpes at summer rsts have been reprted t be mre gentle than winter rsts (Hffman 1968, Lutz and Crawfrd 1987). Mst turkeys in my study rsted near the tp f slpes r n ridges, which was cnsistent with ther studies (Lutz and Crawfrd 1987, Schemnitz et al. 1985). Mre rsts in my study had easterly aspects, but patterns did nt deviate frm randm. Beker and Sctt (1969) pstulated that first mrning light was imprtant in selectin f easterly aspects at turkey rsts. Pndersa pine, the mst cmmn tree selected fr rsting by Merriam's turkeys, is assciated with sites deficient in rainfall (Fwells 1965), typical f eastern slpes f muntain ranges in the western United States. As a result, easterly aspects wuld predminate in the landscape and culd, in part, explain the selectin f eastern aspects fr rst sites. Nrthwestern winds and weather patterns als have been suggested as factrs determining selectin f easterly aspects fr rsts (Jnas 1966, Beker and Sctt 1969). Weather may, in part, influence lcatin f rsts relative t aspect. Mre rsts with western and nrthern aspects were lcated n lwer prtins f the slpes cmpared t rsts with eastern r suthern aspects in my study, which wuld decrease expsure t prevailing nrthwest winds. Turkeys in the Black Hills selected sites fr rsts that had limited r n recent timber activity. The prescriptin at mst sites selected fr rsts with evidence f cutting was a cmmercial thin under current management, r selective cuts under previus management. The primary difference between these prescriptins is that a cmmercial thin is designed t release suppressed trees in a stand fr increased grwth. Typical timber harvest prescriptins fr the Black Hills are mre drastic (i.e., verstry remval r shelterwd seed cut). Seed cut prescriptins (abut 9 m2/ha basal area remaining) wuld be well belw the average basal area at rst sites. Overstry remval, the final stage f 3-step shelterwd prescriptins, remves all the verstry if the stand has advanced regeneratin (Hver and Wills 1984:215). Even-age management gals f pndersa pine in the Black Hills are fr stands f 14-18 m2/ha (60-80 ft2/acre) Grwing Stck Levels (GSL). GSL are tree stcking levels in basal area crrected fr future grwth t a standard dbh f 25 cm (10 inches). GSL f 18 fr trees 18 cm (7 inches) dbh wuld have less than 16 m2/ha (70 ft2/acre) basal area (Bldt and Van Duesen 1974). Mderate timber harvest activity in Arizna did nt affect rst site selectin by Merriam's turkeys, but timber harvest that left 18 m2/ha basal area caused turkeys t abandn rsts (Sctt and Beker 1975). N rsts were lcated in secnd grwth timber in Clrad (Hffman 1968). There was a statistical preference fr multistry stands fr rsts in my study, and rsts in Oregn were primarily lcated in multistry stands (Lutz and Crawfrd 1987). Hwever, the direct effects n turkey rsting behavir f even-age management are still unknwn. Fewer rst trees per site ccurred in my study than reprted elsewhere fr Merriam's turkeys. Other regins f the west averaged 5-13 rst trees/site with up t 37/site, and winter rsts usually have mre trees/site than summer rsts (Beker and Sctt 1969, Lutz and Craw-

J. Wildl. Manage. 56(4):1992 MERRIAM'S TURKEY ROOSTING HABITAT * Rumble 757 frd 1987, this study). The highest number f rst trees per site recrded in my study was 9. Large winter flcks, typical f mst Merriam's turkey range did nt ccur in the ppulatin I studied and culd accunt fr reprted differences in rst trees per site. Traditinal rsts with large flcks ccur in sme areas f the Black Hills usually at lw elevatins and in assciatin with ranch peratins r husing develpments where turkeys btain dmestic grains. Rst Plts.-Vegetative characteristics surrunding the rsts demnstrated few statistical deviatins frm adjacent plts, but sme patterns were evident: basal areas were higher, densities f trees were lwer, and average dbh was greater. Turkeys usually enter rst trees directly frm belw (Mackey 1984), r frm pen areas n the uphill side (Hffman 1968, Beker and Sctt 1969). These characteristics at rsts may represent a cmbinatin f larger diameter trees selected fr rsting (see belw) and the relative pen areas frm which turkeys fly int rsts. Rst Trees. -Large diameter trees ( 50-cm dbh) have been reprted fr turkey rsts in the literature (Beker and Sctt 1969, Mackey 1984, Lutz and Crawfrd 1987). This has led sme managers and researchers t pstulate that turkeys need large diameter trees fr rsts. My data suggest that diameter is nt the criterin fr selectin f rst trees by turkeys. Thirtysix percent f all rst sites had 1 r mre trees 2.5 cm dbh larger than the tree selected fr rsting and 16% had 1 r mre trees 7.6 cm larger dbh than the tree selected fr rsting within the rst plt. Trees selected fr rsts in my study had layered hrizntal branches spaced at intervals that allwed easy access by turkeys, and were typical f phtgraphs accmpanying previus research (Jnas 1966, Hffman 1968, Beker and Sctt 1969). Easy access t rsts was a factr in selectin f flat, layered cnfiguratin f rst trees by eastern wild turkeys (Kilpatrick et al. 1988). Layered hrizntal branches are typical f trees in the Black Hills that are apprximately 100 years ld (G. Gire, Black Hills Nat. Fr., Rapid City, S.D., pers. cmmun.). Rst trees in Clrad averaged 160 years (Hffman 1968) and thse in Oregn were >300 years ld (Lutz and Crawfrd 1987). Reduced thermal expenditure has been suggested as a mechanism fr the selectin f certain rst tree characteristics by Merriam's turkeys (Mackey 1984). Thermregulatry benefits f rsting in trees are mre prbable frm reduced wind velcities than frm reduced heat lss due t the verhead canpy (Kelty and Lustick 1977, Walsberg 1985). Lcatin f rsts n easterly aspects wuld result in sme prtectin frm wind (Hffman 1961) and rsting clse t the tree trunk wuld maximize benefits f reduced wind velcity (Walsberg and King 1980, Pekins et al. 1991). Hwever, many rsts f Merriam's turkeys were lcated n ridges (Beker and Sctt 1969, my study), and mst turkeys bserved in rst trees were n limbs away frm the trunk. The subject f thermregulatry benefits frm rsting in trees appears t be unreslved. Walsberg and King (1980) fund n substantial energy cnservatin fr American rbins (Turdus migratrius) rsting 1-2 m frm the trunk in Duglas-fir and suggested that prtectin frm predatrs may be f greater imprtance than thermregulatry ecnmy. Thmpsn and Fritzell (1988) fund thermregulatry benefits t ruffed gruse (Bnasa umbellus) rsting in red-cedar (Juniperus virginiana), mstly due t imprved radiative heat balance. Walsberg (1986) reprted that night rsts in dense tree canpies imprved the thermregulatry balance f phainpepla (Phainpepla nitens), mstly by reducing wind velcity. Tree heights in my study were similar t ther studies (with larger dbh trees) suggesting slwer grwth rates f trees in thse areas, and prbably clser spacing f branches. Natural limb pruning ccurs slwly in pndersa pine (Fwells 1965), and lder trees may be required fr rsting in regins where tree grwth is slw. MANAGEMENT IMPLICATIONS Rst site selectin by Merriam's turkeys was nt evident at macrhabitat levels f reslutin typical f current Frest Service mnitring plans. Habitat cnditins necessary t meet rsting habitat requirements f Merriam's turkeys shuld cme frm active management. Timber management that results in stands with trees <25 cm dbh and Grwing Stck Levels <22 m2/ha will reduce the availability f rsting habitat fr Merriam's turkeys. If management fr Merriam's turkeys is cnsidered a pririty, silvicultural prescriptins that maintain prtins f the frest at basal areas >21 m2/ha (90 ft2/acre) with trees 25-35 cm average dbh

758 MERRIAM'S TURKEY ROOSTING HABITAT * Rumble J. Wildl. Manage. 56(4):1992 will need t be develped. Rsting habitats shuld be dispersed thrughut the frest and can be included in winter habitats (Rumble 1990). In terms f Frest Service management criteria, stands f pndersa pine >22.5 cm dbh and 70-100% verstry canpy cver wuld meet these criteria. Timbered stands managed t prvide rsting habitats fr turkeys shuld include trees n the upper third f the slpe with layered hrizntal branches, spaced at 0.9-m intervals, in the upper half f the tree. These frest and tree characteristics may be partially related t dbh f the tree. LITERATURE CITED BENNETT, D. L. 1984. Grazing ptential f majr sils within the Black Hills f Suth Dakta. M.S. Thesis, Suth Dakta State Univ., Brkings. 199pp. BOEKER, E. L., AND V. E. SCOTT. 1969. Rst tree characteristics f Merriam's turkey. J. Wildl. Manage. 33:121-124. BOLDT, C. E., AND J. L. VAN DUESEN. 1974. Silviculture f pndersa pine in the Black Hills: the status f ur knwledge. U.S. Fr. Serv. Res. Pap. RM-124. 45pp. BRYANT, F. C., AND D. NISH. 1975. Habitat use by Merriam's turkey in suthwestern Utah. Prc. Natl. Wild Turkey Symp. 3:6-13. BUTTERY, R. F., AND B. C. GILLAM. 1983. Frest ecsystems. Pages 43-71 in R. L. Hver and D. L. Wills, eds. Managing frested lands fr wildlife. Cl. Div. Wildl. in cperatin with U.S. Fr. Serv., Rcky Mt. Reg., Denver. 459pp. BYERS, C. R., R. K. STEINHORST, AND P. R. KRAUS- MAN. 1984. Clarificatin f a technique fr analysis f utilizatin-availability data. J. Wildl. Manage. 48:1050-1053. COCHRAN, W. G. 1963. Sampling techniques. Jhn Wiley and Sns, Inc., New Yrk, N.Y. 413pp. FOWELLS, H. A. 1965. Silvics f frest trees f the United States. USDA Agric. Handb. N. 271. Washingtn, D.C. 762pp. GRIFFING, J. P. 1985. The spherical densimeter revisited. Suthwest Habitater. Vl. 6, N. 2, U.S. Fr. Serv., Reg. 3. Albuquerque, N.M. 2pp. HOFFMAN, D. M. 1968. Rst sites and habits f Merriam's turkeys in Clrad. J. Wildl. Manage. 32:859-866. HOFFMAN, R. S. 1961. The quality f winter fd f blue gruse. J. Wildl. Manage. 25:209-210. HOOVER, R. L., AND D. L. WILLS, editrs. 1984. Managing frested lands fr wildlife. Cl. Div. Wildl. in cp. with U.S. Fr. Serv., Rcky Mt. Reg., Denver. 459pp. JOHNSON, D. H. 1980. The cmparisn f usage and availability measurements fr evaluating resurce preference. Eclgy 61:65-71. JONAS, R. 1966. Merriam's turkeys in sutheastern Mntana. Mnt. Game and Fish Dep. Tech. Bull. N. 3, Helena. 36pp. KELTY, M. P., AND S. I. LUSTICK. 1977. Energetics f the starling (Sturnus vulgaris) in a pine wds. Eclgy 58:1181-1185. KILPATRICK, H. J., T. P. HUSBAND, AND C. A. PRIN- GLE. 1988. Winter rst site characteristics f eastern wild turkeys. J. Wildl. Manage. 52:461-463. LEMMON, P. E. 1956. A spherical densimeter fr estimating frest verstry density. Fr. Sci. 2: 314-320. LUTZ, R. S., AND J. A. CRAWFORD. 1987. Seasnal use f rst sites by Merriam's wild turkey hens and hen-pult flcks in Oregn. Nrthwest Sci. 61:174-178. MACKEY, D. L. 1984. Rsting habitat f Merriam's turkeys in suth-central Washingtn. J. Wildl. Manage. 48:1377-1382. MILLIKEN, G. A., AND D. E. JOHNSON. 1984. Analysis f messy data: vlume 1-designed experiments. Van Nstrand Reinhld, C. New Yrk, N.Y. 473pp. MOSTELLER, F., AND A. PARUNAK. 1985. Identifying extreme cells in a sizeable cntingency table: prbabilistic and explratry appraches. Pages 189-224 in D. C. Haglin, F. Msteller, and J. W. Tukey, eds. Explring data tables, trends, and shapes. Jhn Wiley and Sns, Inc., New Yrk, N.Y. 527pp. NEU, C. W., C. R. BYERS, AND J. M. PEEK. 1974. A technique fr analysis f utilizatin-availability data. J. Wildl. Manage. 38: 541-545. PEKINS, P. J., F. G. LINDZEY, AND J. A. GESSAMAN. 1991. Physical characteristics f blue gruse winter use-trees and rst sites. Great Basin Nat. 51:244-248. PETERSEN, L. E., AND A. H. RICHARDSON. 1975. The wild turkey in the Black Hills. Tech. Bull. N. 6, S.D. Game, Fish and Parks, Pierre. 51pp. PHILLIPS, F. 1980. A basic guide t rst site management fr Merriam's turkeys. Wildl. Digest, Abstr. N. 12. Ariz. Game and Fish Dep., Phenix. 6pp. RUMBLE, M. A. 1990. Eclgy f Merriam's turkeys (Meleagris gallpav merriami) in the Black Hills, Suth Dakta. Ph.D. Diss., Univ. Wyming, Laramie. 169pp. SCHEMNITZ, S. D., D. L. GOERNDT, AND K. H. JONES. 1985. Habitat needs and management f Merriam's turkey in suthcentral New Mexic. Prc. Natl. Wild Turkey Symp. 5:199-231. SCOTT, V. E., AND E. L. BOEKER. 1975. Eclgy f Merriam's wild turkey n the Frt Apache Indian Reservatin. Prc. Natl. Wild Turkey Symp. 3: 141-158. STEEL, R. D. G., AND J. H. TORRIE. 1980. Principles and prcedures f statistics: a bimetrical apprach. McGraw-Hill, Inc., New Yrk, N.Y. 633pp. THOMAS, D. L., AND E. J. TAYLOR. 1990. Study designs and tests fr cmparing resurce use and availability. J. Wildl. Manage. 54:322-330. THOMPSON, F. R., AND E. K. FRITZELL. 1988. Ruffed gruse winter rst site preference and influence n energy demands. J. Wildl. Manage. 52:454-460.

J. Wildl. Manage. 56(4):1992 MERRIAM'S TURKEY ROOSTING HABITAT * Rumble 759 USDA FOREST SERVICE, ROCKY MOUNTAIN RE- GION. 1982. Pht series fr quantifying frest residues in the Black Hills. U.S. Fr. Serv., Rcky Mt. Reg. AFM 831. Denver, Cl. 80pp. WALSBERG, G. E. 1985. Physilgical cnsequences f micrhabitat selectin. Pages 389-413 in M. L. Cdy, ed. Habitat selectin in birds. Academic Press, Orland, Fla. 558pp. 1986. Thermal cnsequences f rst-site selectin: the relative imprtance f three mdes f heat cnservatin. Auk 103:1-7., AND J. R. KING. 1980. The thermregulatry significance f the winter rst-sites selected by rbins in eastern Washingtn. Wilsn Bull. 92:33-39. WILLIAMS, L. E. 1966. Capturing turkeys with alpha-chlralse. J. Wildl. Manage. 30:50-56. Received 9 September 1991. Accepted 23 April 1992. Assciate Editr: Mrrisn.