ILLINOI PRODUCTION NOTE. University of Illinois at Urbana-Champaign Library Large-scale Digitization Project, 2007.

Transcription

1 LLNO S UNVERSTY OF LLNOS AT URBANA-CAMPAGN PRODUCTON NOTE University llinis at Urbana-Champaign Library Large-scale Digitizatin Prject, 27.

2

3 ./A S5 c A) U Zi 1948 LLNOS NATURAL STORY SURVEY CENTER FOR WLDLFE ECOLOGY \....- Z ^,..;"; -^ 4 1^ *;/*'.Y Wd Duck nvestigatins W- 118-R Final Reprt t llinis Department Natural Resurces Prepared by: Aarn P. Yetter, Stephen P. avera, Christpher S. ine, and Michelle M. rath Submitted by: Stephen P. avera llinis Natural istry Survey, avana 2 Nvember 1998

4

5 TABLE OF CONTENTS EXECUTVE SUMMARY... SUMMARY OF ACCOMPLSMENTS... iii vi NTRODUCTON ACKNOWLEDGMENTS... JOB STUDY AREA... MATERALS AND METODS... Survey s... Density Estimates Data Analysis... RESULTS... Survey Chrnlgy and Wd Duck Mvement Surveys and Observer Cmparisns... Princet n Mereds i a i s and Cluster Sizes... Princet n Mereds i a i s Sangan Surveys Sangan Surveys and Cluster Sizes..... Princet n Meredsia i s Sangan Csts DSCUSSON... Feasibility and Chrnlgy... Validity Assumptins... Assumptin 1... Assumptin 2... Assumptin 3... Assumptin 4... Assumptin 5... Assumptin 6... Cst s JOB STUDY AREA... MATERALS AND METODS... abitat Classiicatin

6 Surveys Cavity nspectin... Data Analysis... RESULTS Cavity Availability and Vertebrate Use... Density Cmparisns Cavity Availability and Vertebrate Use... Density Cmparisns Cavity Availability and Vertebrate Use... Density Cmparisns... Nest Density and Success, Artiicial Nest Bxes Cavity Tree Mrtality and Natural Cavity Lss... DSCUSSON Natural Cavity Densities... Nest Success Raccns and Fx Squirrels... Artiicial Nest Bxes... Tree Mrtality and Cavity Survival... Cmparisn Line Transect and Nest Density Estimates. MANAGEMENT RECOMMENDATONS... LTERATURE CTED

7 EXECUTVE SUMMARY Aerial (helicpter) line-transect (LT) methdlgy was emplyed at three study areas t estimate breeding wd duck (Aix spnsa) ppulatins in bttmland rests the llinis River valley. The wd duck indicated breeding pair (BP) density estimated rm the LT surveys at the Sanganis study site ranged rm.39 t.116 BPs/acre r each spring, Breeding wd duck densities estimated at the Princetn ( BPs/ac) and the Meredsia ( BPs/ac) sites were smaller than densities estimated at Sanganis and were als less precise. The csts the helicpter surveys with three bservers averaged $55.69/mi 2, r $236.31/bserver/survey. The density natural cavities suitable r nesting by wd ducks (suitable cavities) at the Sanganis Cnservatin Area (CA) in 1994 (.86 suitable cavities/ac) and 1997 (.76 suitable cavities/ac) were similar, indicating that cavity densities remained unchanged as a result lsses rm tree mrtality assciated with the 1993 and 1995 lds, and increases rm decay and pileated wdpecker (Drycpus pileatus) activity. Raccns (Prcyn ltr) were the primary users suitable cavities. Evidence raccn use was und in 29.3 t 35.3 percent cavities. Fx squirrels (Sciurus niger) were the ther primary inhabitants and ccupied 5.2 t 11.8 percent the mnitred cavities. A large number suitable cavities ( %) were nt used during the springs ; therere, cavity availability des nt appear t be limiting wd duck prductin at the Sanganis CA. 111

8 N dierences in wd duck nest densities were detected during springs , suggesting that the breeding wd duck ppulatin remained stable ver the 5-yr study. Wd duck nest success rates varied rm. t 1. percent during these springs. A cmbined sample 26 nests rm 1994, 1997, and 1998 prvided a simple estimate wd duck nest success 57.7 percent at the Sanganis CA. BP estimates wd ducks rm LT surveys in 1997 r each bserver ranged rm 39.8 t 73.5 percent the wd duck nest densities btained rm inspectins natural cavities; in 1998 BP estimates were 43.9 t 7.4 percent nest density values. Variability in the BP density estimates prevented precisely deining breeding ppulatins. The high variability in nest density estimates in bth 1997 and 1998 limited the ability t detect dierences between the LT and cavity inspectin methds estimating wd duck densities. We recmmend urther research evaluating the LT methdlgy r estimating densities breeding wd ducks in bttmland rests. LT surveys shuld emply transect lengths lng enugh t prvide 2-3 wd duck bservatins. Surveys incrprating this number sightings shuld prvide precise (ceicient variatin [CV] < 1%) estimates wd duck densities. Flights shuld nt be initiated when winds exceed 15-2 mph t acilitate the pilts abilities t strictly adhere t transect lines. Additinally, excessive winds create ripples n the water surace that increase the diiculty identiying lushing lcatins wd ducks. Multiple surveys shuld be lwn in spring t assess the chrnlgy migratin and the emergence tree liage V

9 that hinders detectin wd ducks. Researchers shuld cnsider the pssible disturbance eects lw-altitude aerial surveys n nntarget species, such as bald eagles (aliaeetus leuccephalus), great blue herns (Ardea herdias), great egrets (Casmerdius albus), and duble-crested crmrants (Phalacrcrax auritus), in bttmlands. Artiicial wd duck nest bxes were mnitred during June-July each year rm 1996 t 1998 at the Sanganis CA. Only a small percentage ( %) the bxes were used r nesting by wd ducks. Simple estimates nest success in the bxes ranged rm 57.1 t 1. percent. Metal bxes received the highest use ( %) each year, while nly 1.5 t 9.1 percent the plastic bxes were used during the springs 1996 t An estimate the density wd duck nests derived rm bxes was.1-.3 nests/ac bttmland rest r each spring. The mrtality 61 trees cntaining ptentially suitable wd duck nest cavities (ptential cavities) was mnitred subsequent t the Great Fld Tree mrtality at Sanganis CA appeared t have peaked ater 1996 when 55.7 percent the mnitred trees had perished. Sixty-ive natural cavities were examined annually beginning in the winter and spring By July 1998, 4 these 65 cavities (61.5%) had becme unsuitable r nesting by wd ducks. The daily survival rate r suitable cavities was with an annual survivrship 8.8 percent ( [95%C]).

10 SUMMARY OF ACCOMPLSMENTS STUDY AERAL ELCOPTER SURVEYS OF BREEDNG WOOD DUCKS N BOTTOMLAND FOREST JOB -1 Ptential Ppulatin Estimate r Breeding Wd Ducks in Bttmland Frest in llinis. We evaluated the easibility and cst r using helicpters t aerially census wd duck ppulatins in bttmland rests during springs, Tw aerial line-transect surveys were lwn each spring at three lcatins in the llinis River valley. Breeding wd duck densities were estimated r each year. Csts aerial surveys were cmpared with previus studies in ther lcatins. JOB -2 Cmparisn Aerial Surveys with Densities Wd Ducks Nesting in Natural Cavities. The density breeding wd ducks estimated in JOB -1 was cmpared with the density nesting wd ducks estimated rm inspectins natural cavities the same year. Suitable nesting cavities were mnitred r vertebrate use and wd duck nest success, 1996 t Natural cavities suitable as wd duck nest sites were surveyed during winter t increase the sample suitable cavities r mnitring. Tree mrtality was estimated rm a sample riginally identiied in vi

11 FNAL REPORT Wd Duck nvestigatins Federal Aid in Wildlie Restratin W-118-R July 1995 thrugh 3 June 1998 STUDY : AERAL ELCOPTER SURVEYS OF BREEDNG WOOD DUCKS N BOTTOMLAND FOREST OBJECTVES: Study bjectives were t: l)estimate breeding wd duck ppulatins in bttmland rests by evaluating an aerial (helicpter) census technique, and 2) cmpare aerial ppulatin estimates breeding wd ducks in bttmland rests with ppulatin densities wd ducks nesting in natural tree cavities. NTRODUCTON A majr challenge in wd duck management is the inability t estimate ppulatin sizes because its secretive nature and inhabitatin rested wetlands. Bellrse (198) stated that the wd duck is the mst diicult ducks t census and that aerial ppulatin estimates are inadequate. wever, helicpter surveys have been used r estimating breeding ppulatins since 199 in assciatin with the Black Duck Jint Venture the Nrth American Waterwl Management Plan in Maine and eastern Canada. Wd ducks are detected n these surveys, but estimates precisin are nt reprted. Likewise, helicpters are currently used in Wiscnsin t survey a variety waterwl in marsh habitats. Sherman et al. (1992) used helicpters t census wd duck ppulatins in rested habitat, but they suggested urther evaluatin was needed t prduce reliable ppulatin estimates.

12 nrmatin n the breeding ppulatin size wd ducks is necessary t enhance management this endemic Nrth American species. Sampling thery and design used r aerial surveys wildlie ppulatins in ther habitats have been deined and need nly slight mdiicatins r use in bttmland rests. Aerial surveys have been used t mnitr many species including: manatees (Trichechus manatus)(packard et al. 1985), kangars (Macrpus spp.) (Chquent 1995), seabirds (Briggs et al. 1985), inless prpise (Nephcaena phcaenides)(yshida et al. 1998), harbr prpise (Phcena phcena)(laake et al. 1997), prnghrn antelpe (Antilcapra americana)(jhnsn et al. 1991; Pjar et al. 1995), Arican ungulates (Nrtn-Griiths 1978), whitetailed deer (Odcileus virginianus)(pietsch 1954), mule deer (O. heminus) (White et al. 1989), nrthern bbwhite (Clinus virginianus)(shupe et al. 1987), and waterwl (avera 1998). Bateman (197) reprted that helicpters were a reliable methd censusing mttled ducks (Anas ulvigula) in Luisiana castal marshes. Jhnsn et al. (1989) used helicpters t survey mttled ducks in salt marshes and und they were superir t ixed-wing aircrat. elicpter surveys waterwl in the breal rest prduced similar results t mre cstly grund cunts n the same areas (Rss 1985). Likewise, aerial bservers identiied mre wintering American black ducks (A. rubripes) than did grund bservers (eusmann 199). elicpters prvide increased visibility ver ixed-wing aircrat because slwer air speeds and mre assciated nise, which induces birds t lush. Thus, helicpters are a reasnable alternative t ixed- 2

13 wing aircrat r use in surveying wd ducks (Bateman 197, Brme 1985). n 1993, states in the Atlantic and Mississippi lyways alng with the U.S. Fish and Wildlie Service develped a wd duck management strategy t utline databases needed t eectively manage wd duck ppulatins (Kelley 1997). One gal identiied in this strategy was t assess ways t mnitr wd duck breeding ppulatins. wever, the preseasn banding and radside survey data used t achieve this gal generally have been inadequate. Therere, this study was designed t evaluate the easibility using helicpters t estimate breeding numbers wd ducks in selected bttmland rests in llinis. ACKNOWLEDGMENTS Prject W-118-R was supprted by Federal Aid in Wildlie Restratin Act (Pittman-Rbertsn), with unds administered by the U.S. Fish and Wildlie Service (USFWS) and the llinis Department Natural Resurces (DNR). The llwing DNR sta at the Sanganis Cnservatin Area cntributed assistance and advice during this prject: D. Cwen, J. pps, R. Smith, and R. Mann. The llwing llinis Natural istry Survey (NS) sta assisted with this prject: P. White and L. Andersn prvided ield assistance, K. Rat supplied technical assistance, and F. Bellrse added helpul cmments and insights. F. Jhnsn the USFWS assisted with the study design and data analysis. llinis Department Transprtatin (DOT), Divisin Aernautics, Springield, llinis, urnished aerial supprt r the prject.

14 JOB NO..1. Ptential Ppulatin Estimate r Breeding Wd Objectives: Ducks in Bttmland Frest in llinis. T evaluate the easibility and cst r using helicpters t aerially census breeding wd ducks in bttmland rest. T cmpare aerial helicpter estimates breeding wd ducks in bttmland rest at varying gegraphic lcatins. STUDY AREA Three study sites were classiied as bttmland rests the llinis River (Fig. 1) including: 1) prtins the Sanganis CA and nearby private lands (Sanganis) at Chandlerville; 2) the Meredsia Natinal Wildlie Reuge (NWR) the llinis River Natinal Wildlie and Fish Reuges and adjacent private lands (Meredsia) at Meredsia; and 3) the Princetn Game and Fish Club and surrunding bttmlands (Princetn) at ennepin. abitats n the study areas were cnsidered representative ther palustrine rested wetlands (Cwardin et al. 1979) in the llinis River valley and were selected because their vast expanses bttmland rest. abitats included at Sanganis were slughs, backwater lakes, rested pnds, and bttmland rest (L Dept. Cns. 1975). Majr tree species n the area included: silver maple (Acer saccharinum), eastern cttnwd (Ppulus deltides), willw (Salix spp), red ash (Fraxinus pennsylvanica), and American elm (Ulmus americana) (Yetter et al. 1999). A rest inventry the Meredsia NWR in 1985 indicated that silver maple (84%) and eastern cttnwd (13%) represented 97 percent the tree basal area (aley 1985). Site visits demnstrated that rest and tree

15 species cmpsitin at Meredsia and Princetn were similar t Sanganis. The Sanganis study site encmpassed prtins suthwest Masn, nrthwest Cass, and east Schuyler cunties and represented 8,15 ac bttmland habitat (Fig. 2). Study area bundaries were marked n the nrth and suth by the llinis and Sangamn rivers, respectively. Lngitude lines deined east (9 18' 39") and west (9 21' 57") bundaries. Meredsia encmpassed prtins suthwest Cass and nrthwest Mrgan cunties and cnsisted 4,8 ac bttmland habitat (Fig. 3). Study area bundaries were marked n the west and east by the llinis River and the Meredsia Lake Drainage and Levee District, respectively. Latitude lines designated nrth (39 55' ") and suth (39' 51' 26") bundaries. Princetn cnsisted 4,15 ac bttmland habitat and was lcated in sutheast Bureau Cunty (Fig. 4). Study area bundaries were marked n the east by the llinis River and by the Chicag Rck sland and Paciic Railrad tracts n the west. Nrth and suth bundaries were identiied as latitude lines 41 17' " and 41 14' 38", respectively. MATERALS AND METODS Surveys Study area bundaries and size were determined rm Natinal Wetlands nventry (NW) maps and measured using a digitizing bard and Measugraph 2.1 stware. NW data were btained rm aerial phtgraphs dated spring 1984 and spring Parallel transects were systematically spaced (White et. al. 1989) every 12" latitude r lngitude (Figs. 2-4). Seventeen, 13, and 18 5

16 transect lines were selected at Sanganis, Princetn, and Meredsia, respectively. Aerial surveys wd ducks at each study site were cnducted twice each year during April, The timing aerial surveys crrespnded with nesting activities wd ducks in llinis. Aerial surveys were initiated ater the peak spring migratin wd ducks in central llinis but bere lea emergence (Bellrse 198, eitmeyer and Fredricksn 199, Yetter 1992, Bellrse and lm 1994, Ryan et al. 1998, avera 1998). All surveys were lwn using a Bell Lng Ranger helicpter with 1 pilt and 3 bservers (let rnt [LF], let rear [LR], and right rear [RR]). elicpters and pilts were cntracted rm the Divisin Aernautics, DOT, Springield, llinis, USA. elicpters were lwn at an altitude 15 t abve grund level (AGL) t prvide suicient clearance abve bttmland timber and at grund speeds 5-64 mph (Sherman 199, R.M. Kaminski, Mississippi State Univ., pers. cmmun.). elicpters were equipped with LORAN-C t aid in the navigatin transect lines and a radar altimeter t maintain a cnstant altitude. Density Estimates Densities breeding wd ducks alng transect lines were estimated by emplying a LT apprach (Burnham et al. 198) using gruped, perpendicular distance classes and analyzed with Prgram DSTANCE versin 2.1 (Buckland et al. 1993, Laake et al. 1994). The LF bserver mnitred the prper curse and altitude the aircrat and als recrded all wd ducks within 45 t and ive distance classes (-75[1], [2], E3], [4], and 6

17 31-45[5] t) n the let side the aircrat in 1997 and 1998 (Fig. 5). n 1996, the LF bserver nly mnitred distance classes 1 and 2. Wd ducks recrded by the LF bserver in 1996 were used t determine visibility birds under the aircrat, but n density estimates were generated. The rear bservers culd nt see the 15-t wide path directly belw the aircrat uselage; therere, distance classes r the rear bservers were set 75 t n either side the transect line (Fig. 5) (Jhnsn and Lindzey 199, Buckland et al. 1993). The rear bservers recrded wd ducks within a 375-t strip n their side the aircrat and placed the birds int ur distance classes (-75[1], [2], [3], and [4] t). Wd ducks lushing rm underneath the aircrat were recrded by the rear bservers but were nt used t calculate density estimates. We established the rientatin line transects t satisy independence relative t the distributin wd ducks n the' study sites. Grund elevatin was used t identiy a density gradient wd ducks n sites because lwer elevatins were mre likely t be inundated during high water perids in spring. Because the suthern edge Sanganis and the western edges Meredsia and Princetn were at higher elevatins abve mean sea level (MSL), we established transect lines perpendicular t this inerred density gradient (White et al. 1989, Buckland et al. 1993: , Yshida et al. 1998). Transects were riented in a nrth-suth directin at Sanganis and an east-west directin at Meredsia and Princetn (Figs 2-4). A grund bserver was placed near the halway pint alng 1 transects at Sanganis during a survey n 2 April This 7

18 bserver recrded the number ducks that lushed bere the arrival the helicpter in rder t determine i wd ducks were leaving transects prir t detectin by aerial bservers. Ater the passage the aircrat, the grund bserver mved via all terrain vehicle (ATV) t the next transect lcatin. All bservers recrded wd ducks detected n transects with hand-held tape recrders. During the secnd survey each study site in 1997 and all surveys in 1998, the LR bserver estimated the percentage the transect inundated with water by denting when the aircrat was ver wet r dry grund. The time needed t cmplete each transect, as measured rm the audi tapes, was used t determine the average velcity the aircrat alng transect lines. Observers recrded wd ducks in distance classes as they were detected (pairs, mixed sex lcks, single sex lcks, single sex, and unknwn sex). T avid cnusin, nly bservatins wd ducks and nt ther species were recrded in distance classes. Surveys were cnducted n days with gd visibility and with winds < 25 mph (U.S. Fish Wildl. Serv. and Can. Wildl. Serv. 1987). n rder t place a wd duck bservatin (cluster) int its respective distance class, reerence lines were marked n the helicpter windws using wax pencils. A secnd set reerence marks were made n a string munted rm the dr t the ceiling. Aligning the reerence marks and lines insured that the bservers' heads were in the prper psitin when a wd duck cluster was sighted (Nrtn-Griiths 1978, Jhnsn et al. 1989, Jhnsn and Lindzey 199, Buckland et al. 1993, and Yshida 8

19 1998). Upn reaching a right angle t the sighting lcatin, the bserver wuld assign the cluster t its respective distance class. Reerence marks were generated mathematically and validated prir t LT surveys using grund measurements (Nrtn- Griiths 1978). The ttal number BPs wd ducks identiied in distance classes by each bserver was determined by summing bservatins segregated pairs, tris (pair and extra male), lne males, males in bachelr grups. 4, and lne emales (ammnd 1969, Stewart and Kantrud 1972, U.S. Fish Wildl. Serv. and Can. Wildl. Serv. 1987, Yetter 1992, Bellrse and lm 1994). Grups unknwn wd ducks were classiied as BPs accrding t the minimum number they culd represent. Fr example, ne unknwn wd duck was classiied as ne BP because it was either a lne male r lne emale, bth which represented a pair. A cluster tw unknwn wd ducks was gruped as ne BP because they may have been a pair as ppsed t tw males r tw emales. A cluster three unknwn wd ducks was cnsidered as ne BP because they were likely a tri. Fur unknwn wd ducks were cnsidered as tw BPs because they culd have been tw pairs rather than ur males r emales. wever, bservatins ur unknwn wd ducks were rare and nly ccurred n 11 ccasins ver the 18 individual surveys. Data Analysis Data were cmputerized and analyzed using the Statistical Analysis System (SAS) versin 6 (SAS nst. nc. 1988) and Prgram DSTANCE versin 2.1 (Laake et al. 1994). A Pearsn prduct-

20 mment crrelatin (Prc CORR, SAS nst. nc. 1988) was used t identiy any buildup wd ducks bserved n transects as surveys prgressed rm ne side a study area t the ther. Tukey/Kramer pst hc multiple cmparisn tests (Prc GLM, SAS nst. nc. 1988) were emplyed t determine i dierences existed in the mean cluster size wd ducks recrded amng the distance classes by each bserver in 1997 and All tests were cnsidered signiicant when Ps.5. The density wd duck BPs alng transect lines was calculated using prgram DSTANCE. DSTANCE generated densities based n the number BPs bserved in each distance class alng transects. The wd duck BP density was estimated r each survey and bserver using the rmula: D=n ()/2L, where n was the number BPs bserved, L was the ttal length all A transects sampled, and () was the estimated prbability density unctin perpendicular distance classes rm the transect line evaluated at distance zer. Density estimates were calculated r each bserver because the varying detectin prbabilities amng bservers. BP densities and standard errrs were dubled r each bserver and survey because bservers nly viewed ne side (M) each transect. Tw mdels were itted t the perpendicular distance data t estimate (): 1) the unirm key unctin with a csine series expansin (Furier Series mdel); and 2) the hal-nrmal key unctin with a csine series expansin. The mdel that best it the shape criterin utlined by Burnham et al. (1979) and Buckland et al. (1993) with the smallest CV (Jhnsn and Lindzey 1

21 199) and/r smallest Akaike nrmatin Criterin (AC) value (Buckland et al. 1993) was selected. RESULTS Survey Chrnlgy and Wd Duck Mvement Observers nted the numbers Canada geese (Branta canadensis), mallards (Anas platyrhynchs), blue-winged teal (A. discrs), green-winged teal (A. crecca), American cts (Fulica americana), and duble-crested crmrants identiied incidentally t wd ducks during aerial LT surveys. Discussins ater the lights by bservers were used t assess the chrnlgy the spring waterwl migratin n that day and lcatin. The large number migrant waterwl and waterbirds bserved n surveys cnducted n April 1996 indicated that the spring waterwl migratin was nt yet cmplete, even thugh grund surveys cnducted at Chautauqua NWR, apprximately 2 miles nrtheast the Sanganis CA, suggested that the wd duck migratin was essentially ver by 9-12 April (Fig. 6). Because these late migrants, the irst survey each area in 1996 was nt used t calculate wd duck BP densities. Similarly, the secnd survey at each lcatin was used in 1997 t prevent inclusin migrant wd ducks. n 1998, the irst survey at each lcatin was used t estimate wd duck BP densities because the limited visibility resulting rm lea emergence encuntered during the secnd surveys. Aerial LT surveys were systematically designed t begin n ne side a study area (east bundary at Sanganis and suth bundary at Princetn and Meredsia) and prceed west r nrth t 11

22 the ppsite side with transects spaced every 12" latitude r lngitude. T determine whether wd ducks were being mved r herded rm ne side the study area t the ther, a crrelatin analysis was used t cmpare the ttal number wd ducks bserved by each bserver n each transect with the transect number. A signiicant crrelatin was identiied in nly ne 54 tests (3 yr x 6 surveys/yr x 3 bservers). A crrelatin (rx =.582, P =.37) was identiied in the RR bserver's data r the 23 April 1998 survey at Princetn, and was likely the result the lw number wd ducks detected during this survey (Table 1). Cnsequently, crrelatin analyses did nt substantiate that wd ducks bserved n ne transect were recrded again n subsequent transects. The grund bserver at Sanganis did nt detect any mvement waterwl prir t the arrival the aircrat n eight the ten transects mnitred during 2 April The grund bserver nted n three ccasins that as the helicpter passed, lushing ducks immediately returned t the water near their riginal departure lcatin. Althugh we acknwledge sme wd ducks avided detectin by aerial bservers, we believe the number bservatins missed because early lushes was minimal Surveys and Observer Cmparisns.--Six LT surveys were cnducted during April between 9: a.m. and 5: p.m. CST (Table 1). The number transects at each lcatin varied smewhat rm the number planned due t uel cnstraints and availability 12

23 the aircrat. Fiteen and 16 transects at Sanganis, 12 and 13 transects at Princetn, and 16 and 18 transects at Meredsia were lwn during the irst and secnd surveys at each lcatin, respectively (Table 1). The number wd ducks detected by rear bservers during surveys varied rm 79 t 438 and the number BPs varied rm 42 t 21. Originally, it was thught that all wd ducks lcated directly belw the aircrat wuld lush s that the rear bservers culd identiy wd ducks lying int the irst distance class. This pssibility was tested by cmparing the bservatins the LF bserver with the simultaneus bservatins recrded by the LR bserver. Cmparisns bservatins alng transect lines indicated that nt all wd ducks bserved directly belw the aircrat lushed, and thse that lushed, nt all were identiied by the LR bserver. Sme birds lushed t the right while thers lew parallel t the transect either in rnt r behind the aircrat. Thse wd ducks that did nt lush directly t the let r at an angle ahead and t the let the appraching aircrat culd nt be bserved by the LR bserver. The LF bserver recrded 55 wd duck bservatins in the secnd distance class (76-15 t) during the secnd survey all study sites cmbined. O these 55 bservatins, the LR bserver recrded 46 (83.6%). The LR bserver recrded 25 additinal wd duck bservatins in this distance class that were nt recrded by the LF bserver. These additinal recrds prvided a ttal 8 wd duck bservatins, which the LR bserver detected 71 (88.8%). The varying number wd duck bservatins recrded 13

24 by the rnt and rear seat bservers was attributed t dierences in visibility rm these psitins. As a result, the 15-t strip under the aircrat was excluded rm analyses r rear bservers, and density estimates were generated r each bserver because dierences in sighting prbabilities. Princetn.--The number wd duck BPs detected by each bserver during the 27 April survey at Princetn was lw (Table 1). Cnsequently, utput statistics prvided by Prgram DSTANCE were aected. Mdels it the data prly r the LR bserver (Fig. 7), and AC values were the same r each mdel (Table 2). wever, the unirm mdel prvided the smallest CV value (26.9%) and prvided a mean.39 BPs/ac. The wd duck BP data r the RR bserver was best represented by the unirm mdel (Fig. 7). This mdel als prvided the smaller AC and CV values, yielding a density.25 BPs/ac (Table 2). Meredsia.--A limited number wd duck BP bservatins were detected by rear seat bservers during the 27 April survey at Meredsia (Table 1). The hal-nrmal mdel (Fig. 8) was used r the LR bservers data; hwever, neither mdel prvided a gd it t the data. The density estimate r the LR bserver was.55 BPs/ac (Table 3). The unirm csine mdel (Fig. 8) it the data btained by the RR bserver and indicated a density.26 BPs/ac, but the CV was 28.9 percent (Table 3). Sanganis.--Rear bservers identiied a greater number wd duck BPs (99-111) during the 22 April survey at Sanganis than r the secnd surveys at bth Princetn and Meredsia 14

25 (Table 1), and Prgram DSTANCE achieved a better it t each bservers data (Fig. 9). The hal-nrmal mdel best it the detectin curve r the data generated by bth rear bservers. BP density estimates r the LR and RR bservers were.116 and.84 BPs/ac, respectively (Table 4). Density estimates r bth bservers had CV values belw 2 percent, indicating better levels precisin when cmpared with density estimates r Princetn and Meredsia (Tables 2 and 3) Surveys and Cluster Sizes.--Six LT surveys were cmpleted during April 1997 and were lwn between 8:5 am and 3:6 pm CST (Table 1). The secnd survey at Sanganis was interrupted near the halway pint because mechanical diiculties. That survey was cmpleted the llwing aternn. The number wd ducks recrded by bservers ranged rm wd ducks, and the number BPs ranged rm (Table 1). abitat cnditins were drier at Princetn (39% the transect area was inundated) than Meredsia (76% inundated) and Sanganis (68% inundated) during the secnd survey each site in The estimate rm Meredsia was misleading because much this cverage was rm pen water prtins transects ver Meredsia Lake. The majrity the bttmland rest at Meredsia was dry, which was similar t Princetn, and subsequent wd duck bservatins were lw. The cmparisn the mean cluster sizes between distance classes revealed nly minr discrepancies (Tables 5-7). We identiied a dierence in cluster size between the distance 15

26 classes in nly ne instance acrss all surveys and bservers. The mean cluster size bserved by the LR bserver during the 13 April survey at Sanganis was smaller in distance class three than in distance class ur (Table 7). This dierence was in part caused by tw large grups wd ducks (six and eight) bserved in the utermst distance class. When these tw bservatins (utliers) were mitted rm the data set, n signiicant dierences were detected. Therere, BPs rather than clusters were used t estimate ppulatins lcally breeding wd ducks. Princetn.--Density estimates rm this study site may be biased due t the limited number wd duck bservatins (14-25 BPs) recrded during each survey (Table 1). Mdels prly it the detectin curve r the LF bserver's data (Fig. 1) rm the secnd survey at Princetn. The density estimate generated by the unirm-csine mdel was.15 BPs/ac r the 22 April survey (Table 2). Despite smaller percent CV and AC values r the unirm mdel (Table 2), the hal-nrmal mdel best it the data recrded by the LR bserver during the secnd survey (Fig. 1). The estimated density by the LR bserver was.18 BPs/ac. The it the detectin curve and the AC value suggested the hal-nrmal mdel best represented the data recrded by the RR bserver (Fig. 1, Table 2) even thugh the percent CV values were smaller r the unirm mdel. The density estimate generated r the RR bserver by the hal-nrmal mdel was.23 BPs/ac. 16

27 Meredsia.--The lw number wd duck bservatins (16-3 BPs) limited the reliability LT surveys at Meredsia (Table 1) (Burnham et al. 198). Bth mdels yielded similar results using data cllected by the LF bserver (Table 3). The halnrmal mdel (Fig. 11) generated a density estimate.45 BPs/ac. Analysis the LR bserver's data rm 21 April with the hal-nrmal mdel prvided a density estimate.35 BPs/ac (Fig. 11, Table 3). Data cllected by the RR bserver was best represented by the hal-nrmal mdel (Fig. 11, Table 3). A density estimate.4 BPs/ac was btained during the 21 April survey at Meredsia by the RR bserver. Sanganis.--Bth mdels it the LF bserver's data r the 2-21 April survey (Fig. 12, Table 4). Percent CV values varied, but AC values indicated that the hal-nrmal mdel best represented the data. The density estimate r the LF bserver's data was.39 BPs/ac. The hal-nrmal mdel achieved a reasnable it the detectin curve t the LR bserver's data during the secnd survey (Fig. 12). The AC value als suggested a better it the hal-nrmal mdel, and the estimated density was.66 BPs/ac (Table 4). Data cllected by the RR bserver prvided a gd it the detectin curve rm bth mdels, but the AC value indicated the hal-nrmal mdel better represented the data (Fig. 12, Table 4). The crrespnding density estimate generated r the RR bserver was.72 BPs/ac Surveys and Cluster Sizes.--Six LT surveys were lwn in April between 9:2 am and 4:57 pm CST. 17 The number wd ducks

28 recrded by bservers during each survey ranged rm 5-34, and the number BPs ranged rm (Table 1). Princetn (64-75% transect area was inundated) was drier than Meredsia (97-1% inundated) and Sanganis (93-95% inundated) during bth surveys, but all three study areas hsted higher river stages than in 1996 and wever, ewer numbers wd ducks were recrded (Table 1) by bservers in 1998 than in 1996 and The cmparisn the mean cluster sizes between distance classes revealed nly minr discrepancies (Tables 8-1). Observed cluster sizes wd ducks between the distance classes did nt vary during either survey at Sanganis (Table 1). Dierences were detected r the LF and RR bserver during surveys at Princetn and Meredsia; hwever, a limited number wd ducks were detected during these surveys by all bservers, and n wd ducks were detected in sme distance classes (Tables 8-9). Therere, BPs rather than clusters were used t estimate ppulatins lcally breeding wd ducks. Princetn.--The hal-nrmal mdel best it the data r each bserver during the 15 April survey (Fig. 13). The limited number wd duck bservatins (Table 1) again hampered estimates wd duck densities (Table 2). Estimates r the LF and LR bservers were similar;.57 and.58 BPs/ac, respectively. wever, estimates were mre precise r the LF bserver (CV = 21.%) than r the LR bserver (CV = 3.%) (Table 2). The density estimated r the RR bserver was.4 BPs/ac, and it had a higher CV value (35.2%). 18

29 Meredsia.--The lw number BP bservatins (4-15 BPs) recrded limited the reliability LT surveys at Meredsia (Table 1). The hal-nrmal mdel best it the data r each bserver althugh the it r the RR bserver's data was pr (Fig. 14). Density estimates r all bservers ranged rm BPs/ac, and precisin was lacking with CV values ranging rm percent (Table 3). Sanganis.--The hal-nrmal mdel again prvided the best it the detectin curve r all bservers data during the 14 April survey (Fig. 15), and AC values r the hal-nrmal mdels were smaller than the unirm mdels r each bserver (Table 4). The density estimate wd ducks btained by the LF bserver during the 14 April survey was.56 BPs/ac with a CV value 15.6 percent. Prgram DSTANCE generated a similar density r the LR bserver (.69 BPs/ac) and had a small CV value (17.7%). The RR bserver detected a smaller number wd ducks (Table 1) than the LF and LR bservers, and the crrespnding density estimate was lwer (.43 BPs/ac) with a higher CV value (22.3%; Table 4). Csts n rder t evaluate the cst cnducting aerial LT surveys bttmland habitat, we rented a helicpter and pilt including uel rm the DOT r a cst $85/passenger/hr. The average cst r the six surveys each year was $2,976. The area the three study sites ttaled mi 2. Because each study site was lwn twice each spring, the ttal area inventried was 19

30 53.44 mi 2. Bttmlands the llinis River valley were surveyed with three bservers r a cst $55.69/mi 2, r $18.56/mi 2 /bserver. The Sanganis area represented mi 2 and it was inventried r $55.69/mi 2, r $78.93/survey and $236.31/bserver/survey. DSCUSSON Feasibility and Chrnlgy Aerial LT sampling wd ducks during spring may generate ppulatin estimates in bttmland rests. wever, bttmland rests in llinis large enugh t inventry with this methd are limited. Observatins wd duck clusters at Princetn and Meredsia were nt numerus enugh t prvide the precise density estimates required r management recmmendatins; nevertheless, they may depict trends. Althugh Meredsia, Princetn, and Sanganis were amng the largest tracts bttmland rests remaining in the llinis River valley, the minimum number wd duck bservatins (> 4; Burnham et al. 198) needed t estimate ppulatin densities with LT mdels was nly achieved at Sanganis. When selecting study sites, spring inundatin and habitats avided by wd ducks (ie., pen water) shuld be cnsidered. Fr example, the rests at Meredsia and Princetn were lded just during the 1998 surveys whereas Sanganis generally cntained water. n additin, the lw number wd ducks bserved at Princetn and Meredsia as cmpared with Sanganis (Table 1) resulted rm their smaller size (i.e., smaller transect lengths). Sanganis cntains a myriad histric stream beds, swales, pnds, and slughs that 2

31 hld water during lwer river stages and prvide laing and raging sites r wd ducks. Waterwl management units at Sanganis CA cntained water and verall cnditins at Sanganis prvided habitat mre cnducive t LT surveys than either the Princetn r Meredsia sites. Aerial LT sampling shuld be evaluated elsewhere in areas cntaining vast expanses lded bttmland rests(> 6, ac) where bservatins wd duck clusters can exceed the minimum needed t estimate densities using LT methdlgy. Burnham et al. (198) recmmended a minimum 4 bservatins r LT surveys and suggested 6-8 wuld be preerable. White et al. (1989) indicated that 2 bservatins mule deer were needed t achieve a < 1 percent level precisin during aerial helicpter surveys in nrthwestern Clrad. Based n data cllected during the 22 April 1996, 2-21 April 1997, and 14 April 1998 surveys at Sanganis, ttal lengths transects n each survey shuld be increased rm 3-7 times t achieve CV values 1 percent (Burnham et al. 198:35-36, Kelley 1996:33). This level precisin is nt pssible because this entire study area was systematically cvered by the transects. We suggest that an even greater number bservatins (-3) are necessary t achieve a 1 percent level precisin during aerial LT surveys wd ducks in palustrine rested wetlands. The chrnlgy surveys was critical because lea emergence was rapid and visibility t the rest lr was reduced within a ew days. On 14 April 1998, visibility was adequate; hwever, bservers nted that by 22 April leaemergence severely limited visibility. Kelley (1996) nted 21

32 decreased visibility assciated with lea-ut while cnducting grund LT surveys wd duck ppulatins in Missuri. Validity Assumptins Several assumptins have been established t ensure unbiased estimates density when using LT sampling thery (Burnham et al. 198:14,3; Buckland et al. 1993:29-37; Guthrey 1988). Assumptin 1.--All bjects n the transect line are detected, g(o) = 1. This assumptin was vilated because the LR bserver nly identiied (83.6%) wd duck clusters recrded by the LF bserver in the irst distance class during the secnd survey all study sites in Therere, density estimates r the LR bserver were ptentially biased and apprximately 16 percent lw due t these missed bservatins (Buckland et al. 1993:3). This situatin culd be crrected by having 2 bservers cllectively view the same side the transect line t increase sightings in the irst distance class, thus ensuring g(o) = 1. Assumptin 2.--Objects d nt mve prir t detectin. We presume this assumptin was satisied because wd ducks that lushed upn arrival the helicpter culd be placed in their riginal lcatin via ripples n the water. Buckland et al. (1993:32) suggested recrding the lushing lcatin in this instance because it is the lush that leads t the detectin. The speed the helicpter as well as ur lw altitude (in sme instances < 25 t abve the canpy) allwed bservers t detect clusters bere substantial mvement by wd ducks ccurred. The 22

33 grund bserver during the 2 April 1997 survey at Sanganis als substantiated ur cnclusin that wd duck mvement prir t the arrival the aerial bservers was minimal. Mvement wd ducks in respnse t the aircrat varied in directin. Sme lushed a ew eet and landed, sme lushed away rm the transect, sme lew twards the transect, thers lew parallel with the helicpter, and still thers dve beneath the water nly t resurace within a ew eet. Many wd ducks remained in their riginal lcatin while swimming rapidly in a tight circle. Buckland et al. (1993:34) suggested bias wuld be trivial i incrrect distances were recrded < 5 percent the time due t animal mvement in respnse t the bserver. We cncluded that undetected wd duck mvement was minimal during ur surveys. Assumptin 3.--Distance measurements are exact. We believe this assumptin was satisied because data were gathered in gruped perpendicular distance classes; therere, vilatins this assumptin shuld have ccurred nly near the distance class brders. Our methd determining distances by aligning tw sets reerence marks minimized errrs in distance determinatins (Nrtn-Griiths 1978, Jhnsn et al. 1989, Jhnsn and Lindzey 199, Buckland et al. 1993, and Yshida 1998). Reerence marks were validated prir t each survey using knwn distances n land. elicpters were equipped with radar altimeters s that transects culd be lwn at the prper altitude at all times, and pilts navigated transect lines using LORAN-C. The LF bserver mnitred the radar altimeter and 23

34 LORAN-C unit t ensure transects were lwn at the prper altitude and adherence was maintained t the transect line. The LR bserver als spt-checked the altitude during each survey. Assumptin 4.--Sightings are independent events. The mean cluster size wd ducks detected during LT surveys in 1997 and 1998 was generally <2 (Tables 5-1), which indicated that mst bservatins were either single males r pairs wd ducks. n a ew instances, grups BPs lushed at the same time; hwever, in these inrequent situatins the BPs usually led in separate directins, thus increasing the prbability the lushes being independent events. Mixed-sex grups wd ducks that did nt separate int pairs upn lushing were cnsidered migrants and were nt cunted as BPs. We surmise that this assumptin was nt vilated. Assumptin 5.--ndividual animals are nt cunted mre than nce. We identiied a buildup BPs n successive transects in nly ne the surveys and by nly ne bserver; hwever, a limited number wd ducks were sighted during this survey. The grund bserver during the 2 April 1997 survey at Sanganis als indicated that lushed wd ducks immediately returned t their riginal lcatin ater the helicpter had passed. We think vilatin this assumptin was minimal. Assumptin 6.--Guthrey (1988) suggested that the prbability sighting a cluster animals shuld be independent grup size. This assumptin was tested using a cmparisn the mean cluster sizes amng the distance classes r individual bservers 24

35 during each survey in 1997 and 1998 (Tables 5-1). We detected signiicant dierences in nly a ew instances, mst which ccurred when a limited number wd duck bservatins were recrded. We cnclude that this assumptin was met. Csts We estimated ppulatins breeding wd ducks in bttmland rests the llinis River valley with helicpters and three bservers r a cst $55.69/mi 2, cnsiderably less than the $29/mi 2 r helicpter surveys bttmland timber with ne bserver and $259/mi 2 r grund LT surveys in lded bttmland rests in Mississippi (Sherman et al. 1992). Shupe et al. (1987) reprted a smewhat similar cst $27.2/mi 2 r helicpter surveys nrthern bbwhite with tw bservers in Texas rangeland. These researchers reprted that helicpter surveys were less expensive than using a Lincln ndex (markrecapture; $11.1/mi 2 ). we used tw bservers instead three, csts r aerial surveys wuld be reduced t $37.12/mi 2, which mre clsely resembled estimates rm Texas (Shupe et al. 1987). Fixed-wing aircrat with ne bserver have been used t survey mallards and wd ducks in rested wetlands r a cst $32.38/mi 2 (Sherman et al. 1992). wever, helicpters er advantages (decreased velcity, increased visibility, and maneuverability) ver ixed-wing aircrat r waterwl surveys in emergent and rested wetlands (Jhnsn et al. 1989, Sherman et al. 1992). 25

36 JOB NO Cmparisn Aerial Surveys with Densities Wd Objectives: Ducks Nesting in Natural Cavities. T examine whether helicpter surveys breeding wd ducks relate with nesting densities wd ducks determined rm natural cavity surveys at Sanganis Cnservatin Area. T cntinue mnitring natural tree cavities suitable r use by nesting wd ducks at Sanganis Cnservatin Area. T determine nesting success wd ducks in natural tree cavities at Sanganis Cnservatin Area. The number suitable cavities identiied in sample plts in decreased due t extensive tree mrtality caused by the extreme lding in 1993 and 1995 the llinis and Sangamn rivers (Yetter et al. 1999). Cnsequently, bttmland rest at the Sanganis CA was resurveyed r cavities during December 1996-April 1997, and anther density suitable cavities was generated. Nest success and ther inrmatin were derived rm the entire sample suitable cavities identiied during bth the and the cavity surveys. STUDY AREA The natural cavity study area encmpassed prtins suthwest Masn, nrthwest Cass, and east Schuyler cunties (Fig. 16) and included 9,476 ac the Sanganis CA. Sanganis CA lies at the cnluence the llinis and Sangamn rivers and is a state-wned reuge and public hunting area. Sanganis CA was created in 1948 when the state llinis purchased several private duck clubs. The largest these clubs was the Sanganis Gun Club rm which the area received its name (ll. Dept. Cns. 1975). Over the years, ther land purchases have expanded 26

37 Sanganis CA t its current size apprximately 1,3 ac. abitats n the area were cnsistent with the Sanganis study site described in Jb.1. MATERALS AND METODS abitat Classiicatin Wetland and upland habitats n the study area were classiied using NW data stred n the llinis Gegraphic nrmatin System (GS), DNR, Springield, llinis, USA. NW data were btained rm aerial phtgraphs dated spring NW data were grund-truthed r accuracy and identiicatin tree species within varius habitat types. Wd duck nesting habitat was deined as any palustrine rested wetland within the Sanganis CA regardless water regime and/r special mdiiers. Frested/scrub-shrub, rested/emergent, scrub-shrub, and scrub-shrub/emergent wetland habitats were excluded rm sampling because the dminant trees grwing in these habitats (determined rm grund truthing) were willw saplings that were nt large enugh t prduce cavities suitable r nesting wd ducks. Surveys Natural cavities suitable as wd duck nest sites were initially identiied in when 86 suitable cavities were lcated (Yetter et al. 1999). This sample cavities decreased t 43 by spring 1996 due t tree mrtality caused by extensive lding in 1993 and Therere, urther sampling bttmland timber was cnducted in , t increase the sample size suitable cavities. The same techniques and 27

38 criteria were used t inspect and identiy suitable cavities (Yetter et al. 1999) s that direct cmparisns culd be made between the tw cavity density estimates. Ninety-seven and 58 sample pints were selected r tree cavity investigatins (Figs ) during and , respectively. Study area bundaries were drawn n NW maps and placed n a digitizer. Latilng crdinates were randmly selected and lcated n a digitizing bard using Measugraph 2.1 stware. Only thse crdinates selected within desired habitats (palustrine rested wetland) were utilized. Apprximately tw percent and ne percent the palustrine rested wetlands at Sanganis CA were surveyed r suitable wd duck nest cavities in and , respectively. Sample pints were lcated in bttmland timber with a glbal psitining system (GPS) and NW maps. ATVs and a jn bat were used r transprtatin. Circular plts (1.24 ac) (Bkhut 1986) centered n each sample pint were marked using range tree paint. All trees within the 1.24-ac plts were searched by tw bservers with binculars. Grund surveys were cnducted at Sanganis CA r ptential cavities ater lea all in 1992 and 1993 and again in Trees cntaining ptential cavities were marked with tree paint and a numbered aluminum tag. Tree and cavity variables enabling bservers t relcate ptential cavities r subsequent inspectin were recrded including: tree species, dbh, status (dead r alive), height, and lcatin within the plt and entrance rientatin and height. All trees having ptential cavities were ascended in and t determine i 28

39 the cavities were actually suitable as wd duck nest sites. All suitable cavities were inspected ater the nesting seasn each spring t determine their use by wd ducks and ther vertebrates (Gigstead 1938, Bkhut 1986, Bellrse and lm 1994). Cavity nspectin Natural cavities were examined r suitability using a mdiied versin the single rpe, rpe-walking system (Mntgmery 1982, Meredith and Martinez 1986, Nadkarni 1988, Warild 199, Padgett and Smith 1992, Stanback and Kenig 1994) and with climbing spikes and saety belt. Varius methds placing a climbing rpe ver a supprt branch in the cavity tree were emplyed. The best methd was utilizing a cmpund bw equipped r bw ishing (Weier 1966, Greenlaw and Swinebrad 1967). Ater shting a ish arrw ver a supprt branch abve the cavity, a heavy nyln string was tied t the ishing line (Munn 1991). Fllwing the remval the arrw, the ishing line was retrieved thereby pulling the heavier nyln string ver the branch. The nyln string was then tied t a climbing rpe and pulled ver the branch and anchred. Natural cavities were cnsidered suitable as wd duck nest sites i they had entrance dimensins at least 2.5 x 3.5 in (Grice and Rgers 1965), platrm dimensins at least 5 x 7 in, and were nt mre than 197 in deep (Bellrse et al. 1964, Bkhut 1986). Cavities were classiied as unsuitable i they held water, cntained excessive debris, were t shallw t cnceal the incubating hen (Rbb and Bkhut 1995), r were 29

40 hllw t the grund (F.C. Bellrse, ll. Nat. ist. Surv., pers. cmmun.). An instrument r cavity inspectin was cnstructed rm tw 6 in sectins 2 in PVC pipe and a right angle PVC cupler. A mirrr was attached inside the right angle cupler, and a small lashlight was attached t ne end the device. With this instrument, researchers culd inspect cavities r internal dimensins and evidence use. Cavities, whse platrms were nt visible r diicult t inspect r evidence nesting activity, were examined by lwering adhesive tape n a weighted string (Nagel 1969, Bkhut 1986, Rbb and Bkhut 1995). Thus, any nest material rm the platrm wuld adhere t the tape and culd be examined. Nests were cnsidered successul i they hatched at least ne egg, and nest success was determined rm eggshells and membranes (Stewart 1957, Bellrse and lm 1994). Vertebrate use suitable cavities was determined by the presence hair, eathers, r scats. Data Analysis All data were cmputerized using Ltus stware Release 5. r Windws, and analyzed using SAS (SAS nst. nc. 1988). The estimated wd duck BP density btained rm spring LT surveys at Sanganis (Jb.1.) was cmpared t the wd duck nest densities btained rm natural cavity investigatins at Sanganis CA with tw sample t-tests (inkle et al. 1988:259, Zar 1996:129). A tw sample t-test (Prc TTEST, SAS nst. nc. 1988) was used t cmpare the 1994 and 1997 densities suitable cavity, and a X 2 gdness--it test was used t cmpare wd 3

41 duck nest success rates (Prc FREQ, SAS nst. nc. 1988). We tested r dierences in wd duck nest densities amng the years ( ) using Tukey/Kramer pst hc multiple cmparisn tests. All statistical tests were cnsidered signiicant when P5.5. The Mayield methd was used t determine the annual lngevity suitable cavities (Mayield 1975). Cavity mrtality (a suitable cavity becming unsuitable r wd duck nesting) was assumed t be the midpint between ur cavity visits. Cavity expsure was deined as the number days between visits. A 95% cnidence interval r the estimated annual cavity survival rate was calculated accrding t Jhnsn (1979). RESULTS 1996 Cavity Availability and Vertebrate Use.--O the 86 riginal suitable cavities, 14 (16.3%) were n lnger available t wd ducks prir t the 1996 nesting seasn, 15 (17.4%) trees with cavities were n lnger climbable during inspectins in spring 1996, and 14 (16.3%) cavities were classiied as nt suitable ater the 1996 inspectins. The remaining 43 (5.%) cavities were lcated in stable trees and available r use by wd ducks. Late spring lding by the llinis River inundated 9 the remaining 43 cavities, urther reducing the sample t 34 suitable cavities. O these 34 suitable cavities, 15 (44.1%) had n evidence vertebrate use, 12 (35.3%) had been ccupied by raccns prir t inspectin, and 4 (11.8%) had evidence x squirrel use. Only 3 (8.8%) cavities were used r nesting by 31

42 wd ducks, yielding a density.25 nests/ac (SE =.14, Cg +.29) bttmland rest. All three nests were successul, and all were lcated in cavities excavated by pileated wdpeckers. Density Cmparisns.--The wd duck BP density estimated rm the LT surveys at Sanganis n 22 April 1996 by the LR bserver (T =.116 BPs/ac, CV = 18.%, n = 45, Jb.1.) was signiicantly greater than the 1996 wd duck nest density (_ =.25 BPs/ac, SE =.14, n = 97, Jb 1.2.) bserved during natural cavity investigatins (t = 6.22, 15 d, P <.5)(Fig. 19). BP densities generated by the RR bserver (5 =.84 BPs/ac, CV = 17.2%, n = 23) were als greater than the 1996 wd duck nest density (t = 4.4, 14 d, P <.5) Cavity Availability and Vertebrate Use.--Thirty-eight suitable cavities were identiied in the 58 sample plts yielding a suitable cavity density.76 cavities/ac bttmland rest (SE =.13, C, ), which was similar t the.86 cavities/ac (SE =.9, Cs +.19) und in 97 sample plts in (t =.631, 153 d, P =.529). n , 86 suitable cavities were identiied (Yetter et al. 1999). This number steadily decreased every spring thrugh Only 34 (39.5%) the riginal 86 suitable cavities were inspected during June and July The ther cavities and/r trees were n lnger climbable (2.9%), allen r lgged (8.1%), r n lnger suitable (26.7%). Tw cavities 32

43 (2.3%) were discarded due t inaccessibility, and tw cavities (2.3%) culd nt be lcated, presumably a result tree all. Tw additinal suitable cavities were identiied during the reinspectin ptential cavities in tw sample plts initially surveyed in These cavities were nt included in suitable cavity r nest density estimates in rder t maintain independence amng the data sets. wever, these additinal cavities were included in the sample available t wd ducks during spring Therere, a ttal 74 suitable cavities was mnitred during June/July Evidence vertebrate use was und in 56.8 percent the 74 suitable cavities. Raccns were the primary users (32.4%) the inspected cavities (nt including 1 cavity cntaining a wd duck nest destryed by a raccn). Wd ducks nested in 1.8 percent the suitable cavities, and x squirrel evidence (hair r nesting material) was identiied in 6.8 percent cavities. The density wd duck nests at Sanganis CA was.98 nests/ac bttmland rest (SE =.35, Cgs ±.7). Five the wd duck nests were lcated in cavities created by pileated wdpeckers, tw wd ducks nested in cavities rmed by limb rt, and ne hen nested in a hllw snag with bth a tp entrance (bucket) and a pileated wdpecker entrance. Six the wd duck nests were successul (75%) and tw were depredated (25%), ne each by a raccn and pssibly a x squirrel. Density Cmparisns.--The wd duck BP densities estimated rm LT surveys varied rm.39 BP/ac t.72 BP/ac (Table 4) and were similar (P >.5) t the 1997 wd duck nest density 33

44 ( =.98 nests/ac, SE =.35, n = 58) bserved during natural cavity investigatins (Fig. 2). Whereas n dierences were und between the aerial LT and cavity inspectin estimates, the high variability in the nest density estimate may have prevented the detectin any dierences Cavity Availability and Vertebrate Use.--Thirty-ne the 38 cavities (81.6%) identiied in the 58 sample plts in were still suitable as wd duck nest sites. Tw cavities (5.3%) were n lnger climbable, tw cavity trees (5.3%) had allen, ne cavity each was ull debris (2.6%), held water (2.6%), r had an expsed platrm (2.6%). Twenty-ive the 86 cavities (29.1%) identiied in remained suitable as wd duck nest sites during spring The ther cavities and/r trees were n lnger climbable (19.8%), allen r lgged (12.8%), r n lnger suitable (33.7%). Fur cavities (4.7%) were discarded due t inaccessibility. Tw additinal suitable cavities were identiied during the reinspectin ptential cavities in tw sample plts initially surveyed in These cavities were nt included in suitable cavity r nest density estimates in rder t maintain independence amng the data sets but were included in the sample available t wd ducks during spring Therere, a ttal 58 suitable cavities were mnitred r vertebrate use during June-July Evidence vertebrate use was und in 31 (53.4%) the 58 suitable cavities. Raccns were the primary users 17 (29.3%) 34

45 inspected cavities (three cavities with wd duck nests destryed by raccns and ne cavity cntaining a hatched wd duck nest with evidence raccn use were nt included). Wd ducks nested in nine (15.5%) suitable cavities, x squirrel hair r nesting material was identiied in three (5.2%) cavities, ne cavity was used by an unknwn mammal, and ne cavity was ccupied by a nesting screech wl (Otus asi) with 5 chicks. The density wd duck nests at Sanganis CA was.98 nests/ac bttmland rest (SE =.35, C95 ±.7). This density was a minimum estimate because the large number suitable cavities that were lcated in trees n lnger stable r climbing and the availability artiicial wd duck nest bxes n the area. Seven the nine wd duck nests were lcated in cavities created by pileated wdpeckers, ne wd duck nested in a cavity rmed by limb rt, and ne hen nested in a hllw snag with bth a tp entrance (bucket) and a pileated wdpecker entrance. Six the wd duck nests were successul (66.7%) and three were destryed by raccns. Density Cmparisns.--The wd duck BP densities estimated rm LT surveys during 14 April 1998 varied amng bservers rm.43 BP/ac t.69 BP/ac (Table 4) and were similar (P >.5) t the 1998 wd duck nest density (Z =.98 nests/ac, SE =.35, n = 58) resulting rm natural cavity investigatins (Fig. 21). The high variability (272% CV) in the nest density estimate may have limited the ability t detect any dierences. 35

46 Nest Density and Success, N dierences (F = 1.72; 4,46 d; P =.146) in wd duck nest densities were identiied during springs, , indicating the breeding wd duck ppulatin remained stable ver the 5-yr study. Wd duck nest success rates varied rm. t 1. percent during the springs and a X 2 gdness--it test indicated that nest success rates diered amng years (X 2 = 11.92, 4 d, P =.18). wever, the number mnitred suitable cavities was lwer in 1995 and 1996 when nly three and ive wd duck nests were und in natural cavities. these tw years were mitted rm analyses, n dierences resulted in the wd duck nest success rates r 1994, 1997, and 1998 (X 2 = 3.47, 2 d, P =.177). Therere, a cmbined estimate nest success during these three years was 57.7 percent (n = 26). Artiicial Nest Bxes Frm 98 t 113 artiicial wd duck nest bxes were inspected during June-July each year rm 1996 t 1998 at the Sanganis CA (Table 11). Only a small percentage ( %) the bxes were used by wd ducks each spring. n 1996, nest success in the artiicial bxes was at least 68.8 percent with ne hen still incubating when last inspected. Nest success in nest bxes ell t 57.1 percent (4 7 nests hatched) in 1997, but was 1 percent in 1998 when all 7 nest attempts were successul. Nest bxes were cnstructed mainly plastic (Ducks Unlimited, nc.) and metal with a ew wden bxes. Metal bxes received the highest use by wd ducks ( %) each year, while nly 1.5 t 9.1 percent the plastic bxes were used 36

47 during the springs 1996 t A simple estimate wd duck nest density derived rm nest bxes was.1-.3 nests/ac bttmland rest during the springs Cavity Tree Mrtality and Natural Cavity Lss The mrtality 61 trees cntaining ptentially suitable wd duck nest cavities was mnitred ater the Great Fld 1993 (Fig. 22). n early (January-April) 1994, nly 1.6 percent the trees were dead; hwever, many were shwing signs stress. By July 1994, 11.5 percent the cavity trees had perished. A recrd ld in the spring 1995 exacerbated mrtality when 5.8 percent the mnitred trees were dead. Mrtality in the bttmland rest at Sanganis CA appeared t have reached a plateau ater 1996 when 55.7 percent the mnitred trees were dead. Survival the riginal 86 suitable cavities was determined in 1998 rm 65 cavities that were lcated in trees still stable enugh t climb and that were mnitred annually since the winter and spring Frty the 65 natural cavities (61.5%) became unsuitable r nesting by wd ducks. The daily survival rate r suitable cavities was with an annual survivrship 8.8 percent ( [95%C]). DSCUSSON Natural Cavity Densities The natural cavity density estimates btained at Sanganis CA in 1994 (.86 suitable cavities/ac; Yetter et al. 1999) and 1997 (.76 suitable cavities/ac) were similar indicating that cavity densities have nt changed rm tree mrtality assciated 37

48 with the 1993 and 1995 ld, decay, and pileated wdpecker activity. Nest Success Success rates rm the cmbined sample 26 nests in 1994, 1997, and 1998 was 57.7 percent; this value was cmparable t the 63.6 percent estimate success rm a sample upland and bttmland nesting wd ducks in suthern llinis (Ryan et al. 1998) but was greater than that previusly und in central llinis (39.9% [Bellrse et al. 1964]; 31.3% [Shake 1967]). Nest success at Sanganis was greater than studies reprted r Gergia (44.4% [Almand 1965]), Missuri (33.3% [Weier 1966]), and suthcentral ndiana (36.4% [Rbb and Bkhut 1995]). Raccns and Fx Squirrels Raccns were the primary users suitable natural cavities. Evidence raccn use was und in 29.3 t 35.3 percent suitable cavities. Fx squirrels were the ther primary inhabitant cavities; hwever, use by squirrels was lwer ( %). Similar rates raccn and x squirrel use were bserved in 1994 and 1995 in natural cavities at Sanganis CA (Yetter et al. 1999). Rbb and Bkhut (1995) bserved lwer cavity use rates by raccns (18.5%) but higher rates r x squirrels (22.7%) in suthcentral ndiana. wever, because 43.2 t 46.6 percent suitable cavities were nt used during the springs 1996 t 1998, the number used by raccns, x squirrels, and ther vertebrates des nt appear t be limiting wd duck prductin. 38

49 Artiicial Nest Bxes Wd duck use artiicial nest bxes was lw and ranged rm 6.2 t 16.3 percent. This value was cmparable t ccupancy rates in the nrthern-tier states the Mississippi Flyway, which averaged 16.1 percent, but was lwer than the 43.3 percent use rate by wd ducks in central states (including llinis) the Mississippi Flyway (Sulliere 199). The percentage natural cavities ccupied by wd ducks each year at Sanganis CA ranged rm 8.8 t 15.5 percent, indicating an abundance suitable natural cavities n the area. The percentage preerred but unused pileated wdpecker cavities (5-69%) each year als suggested that natural cavities were nt limiting wd duck prductin at Sanganis CA. Nest bx prgrams n the area appear unjustiied because the cmparable wd duck nest success rates in artiicial nest bxes (75.9%) and natural cavities (57.7%), the high cst ($25-$12) prducing a lighted juvenile wd duck rm a nest bx (Sulliere 1986), and the abundance natural cavities. Tree Mrtality and Cavity Survival Estimates tree mrtality in 1995 at Sanganis CA were 42.7 percent (Yetter et al. 1999). Cntinued mnitring a sample these trees indicated mrtality resulting rm the 1993 and 1995 lds reached a peak 55.7 percent in Yin et al. (1994) und mrtality rates 37.2 percent n Pl 26 the Upper Mississippi River the year llwing the Great Fld 1993; mrtality increased t 45.6 percent by August 1995 (Rbert J. Csgri, NS, persnal cmmunicatin). Sme shrt- 39

50 term eects lding n tree mrtality at Sanganis CA have been realized; hwever, the extent lng-term eects may nt be knwn r several decades. Cmparisn Line Transect and Nest Density Estimates Wd duck BP densities btained rm LT surveys in 1997 and 1998 by all three bservers were similar t nest densities btained rm natural cavity inspectins. BP density estimates btained by rear seat bservers in 1996 were greater than nest densities determined rm natural cavities. The lwer nest density bserved in 1996 cmpared with BP densities rm LT surveys was likely the result the reduced sample natural cavities rather than an actual dierence in estimates rm these tw methds. BP estimates wd ducks rm LT surveys in 1997 r each bserver ranged rm 39.8 t 73.5 percent wd duck nest densities btained rm natural cavities; in 1998 BP estimates were 43.9 t 7.4 percent nest density values. Variability in the BP density estimates rm LT surveys precluded deining ppulatins precisely, even thugh CV values were belw 2 percent r 5 the 6 estimates in 1997 and Als, the high variability in nest density estimates in bth years (CV = 272%) limited detectin dierences between these methds estimating wd duck density. We expected BP densities rm LT surveys t be greater than nest densities bserved rm cavities because the inclusin pssible late migrants, upland nesting wd ducks laing in bttmlands, and male-biased sex ratis (Bellrse and lm 1994). 4

51 Ryan et al. (1998) reprted 82 percent nests attempted by radi-cllared hens in suthern llinis were lcated in upland rests. Nests in uplands were und as ar as 2.3 miles rm capture sites and.9 miles rm the nearest wetland (Ryan et al. 1998). wever, we und BP densities t be larger than nest densities nly in 1996 when the sample suitable cavities mnitred was small (n = 34). Pst hc multiple cmparisn tests indicated that wd duck nest densities in cavities were similar during springs 1994 thrugh MANAGEMENT RECOMMENDATONS We recmmend urther research evaluating LT methdlgy t estimate the densities breeding wd ducks in bttmland rests. Multiple bservers shuld cperatively mnitr the same side transect lines t guarantee that all wd ducks in the irst distance class are detected (ie., ensuring g(o) = 1). Surveys shuld be cnducted in areas where the length transect lines allws the detectin 2 t 3 wd duck clusters. Flights shuld nt be initiated when winds exceed 15-2 mph s that pilts can strictly adhere t transect lines. Additinally, excessive winds create ripples n the water surace which increases the diiculty crrectly identiying lushing lcatins. Observers need t speak ludly and clearly int tape recrders t vercme excessive engine nise in the helicpter uselage. Multiple surveys shuld be lwn in spring t assess the chrnlgy the migratin and the emergence tree liage that hinders the detectin wd ducks. Finally, researchers shuld cnsider the pssible negative eects lw-altitude 41

52 helicpter surveys n nntarget species, such as bald eagles, great blue herns, great egrets, and duble-crested crmrants. These species were incubating r brding their nestlings during ur surveys. LTERATURE CTED Almand, J.D A cntributin t the management requirements the wd duck (Aix spnsa) in the Piedmnt Gergia. M.S. Thesis, University Gergia, Athens. 58 p. Bateman,.A An evaluatin the helicpter r censusing waterwl in Luisiana's castal marshes and ricelands. Prc. Annu. Cn. Sutheast. Assc. Game and Fish Cmm. 24:9-13. Bellrse, F.C Ducks, geese, and swans Nrth America. 3rd ed. Stackple Bks, arrisburg, PA. 54p. _ and D.J. lm Eclgy and management the wd duck. Stackple Bks, Mechanicsburg, PA. 588p., K.L. Jhnsn, and T.U. Meyers Relative value natural cavities and nesting huses r wd ducks. J. Wildl. Manage. 28: Bkhut, T.A Best management practices r creating and maintaining wd duck habitat n Natinal Wildlie Reuges in Regin 3: a wrkshp. Ohi Cp. Fish and Wildl. Res. Unit, Ohi State Univ., Clumbus. 13 p. Briggs, K.T., W.B. Tyler, and D.B. Lewis Aerial surveys r seabirds: methdlgical experiments. J. Wildl. Manage. 49: Brme, L.S Sightability as a actr in aerial survey bird species and cmmunities. Aust. Wildl. Res. 12: Buckland, S.T., D.R. Andersn, K.P. Burnham, and J.L. Laake DSTANCE sampling: estimating abundance bilgical ppulatins. Chapman & all, Lndn. 446p. Burnham, K.P., D.R. Andersn, and J.L. Laake Rbust estimatin rm line transect data. J. Wildl. Manage. 43: , and Estimatin density rm line transect sampling bilgical ppulatins. Wildl. Mn p. 42

53 Chquent, D Species- and habitat-related visibility bias in helicpter cunts kangars. Wildl. Sc. Bull. 23: Cwardin, L.M., V. Carter, F.C. Glet, and E.T. LaRe Classiicatin wetlands and deepwater habitats the United States. U.S. Dept. nter., Fish and Wildl. Serv. FWS/OBS-79/31. 13p. Gigstead, G Wd ducks in the llinis River bttms. Trans. N. Amer. Wildl. Cn. 6: Greenlaw, J.S., and J. Swinebrad A methd r cnstructing and erecting aerial-nets in a rest. Banding 38: Bird- Grice, D., and J.P. Rgers The wd duck in Massachusetts. Massachusetts Div. Fisheries and Game. Final reprt, Federal Aid in Wildlie Restratin Prject N. W-19-R. 96 p. Guthrey, F. S Line transect sampling bbwhite density n rangeland: evaluatin and recmmendatins. Wildlie Sciety Bulletin 16: aley, G Frest inventry Meredsia Natinal Wildlie Reuge. Unpublished Reprt t U.S. Fish Wildl. Serv. llinis Dept. Cns. 26p. ammnd, M.C Ntes n cnducting waterwl breeding surveys in the nrth central states. Pages in Saskatn Wetlands Seminar. Can. Wildl. Serv. Rep. Ser. 6. avera, S. P Waterwl llinis: status and management. llinis Natural istry Survey Special Publicatin N. 21, Champaign, llinis, USA. n Press. eitmeyer, M.E., and L.. Fredricksn Abundance and habitat use wd ducks in Ming Swamp sutheastern Missuri. Pages in L.. Fredricksn, G.V. Burger, S.P. avera, D.A. Graber, R.E. Kirby, and T.S. Taylr, eds. Prc Nrth Am. Wd duck Symp., St. Luis, MO. eusmann,.w Evaluatin air/grund cunt cmparisns r wintering American black ducks. Wildl. Sc. Bull. 18: inkle, D.E., W. Wiersma, and S.G. Jurs Applied statistics r the behaviral sciences. 2nd ed. ughtn Milin C., Bstn, MA. 682p. llinis Department Cnservatin Sanganis Cnservatin Area. ll. Dep. Cnserv. Pamphlet N. 25M

54 Jhnsn, B., and F. Lindzey Guidelines r estimating prnghrn numbers using line transects. Unpublished reprt. Wyming Game and Fish Dept. and Wyming Cp. Fish and Wildl. Res. Unit. 38p. _,, and R.J. Guenzel Use aerial line transect surveys t estimate prnghrn ppulatins in Wyming. Wildl. Sc. Bull. 19: Jhnsn, D Estimating nest success: the Mayield methd and an alternative. Auk 96: Jhnsn, F.A., K.. Pllck, and F. Mntalban Visibility bias in aerial surveys mttled ducks. Wildl. Sc. Bull. 17: Kelley, J.R.,Jr Line-transect sampling r estimating breeding wd duck density in rested wetland. Wildlie Sciety Bulletin 24: Wd duck ppulatin mnitring initiative: Final Reprt. Atlantic Flyway Cuncil, Mississippi Flyway Cuncil, and U.S. Fish and Wildlie Service. Administrative Reprt. Laurel, Maryland. USA. 248pp. Laake, J.L., J. Calambkidis, S.D. Osmek, and D.J. Rugh Prbability detecting harbr prpise rm aerial surveys: estimating g(). J. Wildl. Manage. 61:63-75., S.T. Buckland, D.R. Andersn, and K.P. Burnham DSTANCE user's guide V2.1. Clrad Fish & Wildl. Res. Unit, Clrad State Univ., Frt Cllins, CO. 84p. Mayield,.F Suggestins r calculating nest success. Wilsn Bull. 87: Meredith, M., and D. Martinez Vertical caving. 2nd ed. Lyn Equipment C., Dent, Sedbergh, Cumbria. 8 p. Mntgmery, N.R Single rpe techniques: a guide r vertical cavers. 3rd ed. The Sydney Spelelgical Sciety, Sydney, Australia. 12 p. Munn, C.A Trpical canpy netting and shting lines ver tall trees. J. Field Ornithl. 62: Nadkarni, N Use a prtable platrm r bservatins trpical rest canpy animals. Bitrpica 2: Nagel, R.E Predatin n eggs in simulated nests and tree cavity abundance in wd duck nesting habitat. M.S. Thesis. wa State Univ., Ames. 94 p. Nrtn-Griiths, M Cunting animals. Arican Wildl. Fund., Nairbi, Kenya. 139p. 44

55 Packard, J.M., R.C. Summers, and L.B. Barnes Variatin visibility bias during aerial surveys manatees. J. Wildl. Manage. 49: Padgett, A., and B. Smith On rpe: Nrth American vertical rpe techniques r caving, search and rescue, and muntaineering. Natinal Spelelgical Sciety, untsville, AL. 341 p. Pietsch, L.R White-tailed deer ppulatins in llinis. ll. Nat. ist. Surv. Bil. Nte. N pp. Pjar, T.M., D.C. Bwden, and R.B. Gill Aerial cunting experiments t estimate prnghrn density and herd structure. J. Wildl. Manage. 59: Rbb, J.R., and T.A. Bkhut Factrs inluencing wd duck use natural cavities. J. Wildl. Manage. 59: Rss, R.K elicpter vs. grund surveys waterwl in the breal rest. Wildl. Sc. Bull. 13: Ryan, D.C., R.J. Kawula, and R.J. Gates Breeding bilgy wd ducks using natural cavities in suthern llinis. J. Wildl. Manage. 62: SAS nstitute, nc SAS/STAT User's Guide Release. Versin 6. SAS nst. nc., Cary, NC. 1,28p. Shake, W Starling-wd duck interrelatinships. M.S. Thesis. Western llinis Univ., Macmb. 46 p. Sherman, D.E Line-transect sampling and indices t estimate wintering wd ducks and mallards in Mississippi green-tree reservirs. M.S. Thesis, Mississippi State Univ., Starkville, MS. 122p. SR.M. Kaminski, and B.D. Lepld Ptential indices mallard and wd duck abundance in rested wetlands during winter. Wildl. Sc. Bull. 2: Shupe, T.E., F.S. Guthery, and S.L. Beasm Use helicpters t survey nrthern bbwhite ppulatins n rangeland. Wildl. Sc. Bull. 15: Sulliere, G. J Cst and signiicance a wd duck nest-huse prgram in Wiscnsin: an evaluatin. Wildlie Sciety Bulletin 14: _ 199. Review wd duck nest-cavity characteristics. Pages in L.. Fredricksn, G.V. Burger, S.P. avera, D.A. Graber, R.E. Kirby, and T.S. Taylr, eds. Prc Nrth Am. Wd Duck Sympsium, St. Luis, MO. 45

56 Stanback, M.T., and W.D. Kenig Techniques r capturing birds inside natural cavities. J. Field Ornithl. 65:7-75. Stewart, P.A The wd duck, Aix spnsa and its management. Ph.D. Dissertatin. Ohi State Univ., Clumbus. 353 p. Stewart, R.E., and.a. Kantrud Ppulatin estimates breeding birds in Nrth Dakta. Auk 89: U.S. Fish Wild. Serv., and Can. Wildl. Serv Standard perating prcedures r aerial waterwl breeding grund ppulatin and habitat surveys in Nrth America. U.S. Fish Wildl. Serv., Washingtn, D.C.; and Can. Wildl. Serv., Ottawa, Ontari. 93p. Warild, A Vertical: a technical manual r cavers. Spelelgical Research Cuncil Ltd., Bradway, Australia. 128 p. Weier, R.W A survey wd duck nest sites n Ming Natinal Wildlie Reuge in sutheast Missuri. M.S. Thesis. Univ. Missuri, Clumbia. 78 p. White, G.C., R.M Bartmann, L.. Carpenter, and R.A. Garrtt Evaluatin aerial line transects r estimating mule deer densities. J. Wildl. Manage. 53: Yetter, A.P Ppulatin densities and use palustrine wetlands by breeding waterwl in nrtheastern llinis. M.S. Thesis, Suthern llinis Univ., Carbndale, L. 196p., S.. avera, and C.S. ine Natural cavity use by wd ducks in llinis. J. Wildl. Manage. 63:in press. Yin, Y., J.C. Nelsn, G.V. Swensn,.A. Langrehr, and T.A. Blackburn Tree mrtality in the Upper Mississippi River and ldplain llwing an extreme ld in Pages 39-6 in Natinal Bilgical Service, llinis Natural istry Survey, wa Department Natural Resurces, and Wiscnsin Department Natural Resurces. Lng Term Resurce Mnitring Prgram 1993 Fld Observatins. Natinal Bilgical Service, Envirnmental Management Technical Center, Onalaska, W, Dec LTRMP 94-SO11. Yshida,., K. Shirakihara,. Kishin, M. Shirakihara, and A. Takemura Finless prpise abundance in Omura Bay, Japan: estimatin rm aerial sighting surveys. J. Wildl. Manage. 62: Zar, J Bistatistical analysis. 3rd. ed. Prentice- all, nc., Upper Saddle River, NJ. 899 p. 46

57 Figure 1. Study areas r aerial line transect surveys breeding wd ducks in llinis during April,

58 { ±7' h L 1~ /r Miles LAND COVER igh Density Urban Medium Density Urban ]--- Lw. i....u.a [ _ Lw Density Urban [--] Rw Crp S-Z1 Small Grains Orchards/Nursuries SUrban Grass jj Grassland Wds SOpen Wds Pine Water 17 Shallnw Mrcsh Deep marsh Bttmland Water, shallw 1_] Barren i Figure 2. The Sanganis study site near Chandlerville, llinis, depicting transect lines that were lwn with a helicpter t estimate breeding ppulatins wd ducks during April, The study site included prtins Sanganis Cnservatin Area and adjacent private lands.

59 & 6 r r. ;~ ;, **ii 1, ::ii?));" L..A,, r:"ll ii r i. --a i: t., :,,..,i t^ - - t ' ". r. -- % K~i - Z 4 Miles LAND COVER igh Density Urban SMedium Density Urban _ ] Lw Density Urban [ _ Rw Crp SSmall Grains Orchards/Nursuries Urban Grass SGrassland SWds SOpen Wds Pine Water SShallw Marsh Deep marsh SBttmland Water, shallw _ Barren Figure 3. The Meredsia study site near Meredsia, llinis, depicting transect lines that were lwn with a helicpter t estimate breeding ppulatins wd ducks during April, The study site included prtins the Meredsia Natinal Wildlie Reuge and adjacent private lands.

60 2 2 4 Miles i / /Rails LAND COVER Sih Densitv Medium density L w Density [ Rw Crp Small Grains Orchards/Nursuries i Urban Grass [ Grassland Wds Open Wds Pine Water [L Shallw Marsh Deep marsh Bttmland j1 Water, shallw Barren Figure 4. The Princetn study site near ennepin, llinis, depicting transect lines that were lwn with a helicpter t estimate breeding ppulatins wd ducks during April, The study site included prtins the Princetn Game and Fish Club and surrunding bttmlands.