EABITAT UTILIZATION BY THE AMERICAN WOODCOCK IN NORTHERN MICHIGAN Thesis for IIae Degree OI M. S. MICHIGAN STATE UNIVERSITY Dale Leslie Rabe I977 L1 1R Y IVE-1mg“) State University Wm‘ usu loAn mm Actionfiqual oppomnuy Imam " -_.:k 7 ABSTRACT HABITAT UTILIZATION BY THE AMERICAN WOODCOCK IN NORTHERN MICHIGAN By Dale Leslie Rabe Habitat utilization by woodcock was studied during the Spring of 1975 and 1976 on six quarter townships that were clearcut to approxi- mately 25%, 50%, and 75%. Standard census routes were used as a measure of the population response to the treatments. Vegetational measurements in the clearings along the routes included: species composition, density and height of the regenerating vegetation in the clearing, and the ‘for- est community types bordering the clearing. Woodcock habitat utilization was greatest where the interspersion of field and forest is maximized. Aspen communities were preferred to oak and pine as diurnal habitat. The proximity of suitable diurnal habitat was more important than the gross vegetational features of a clearing in determining the use of a particular clearing as a singing ground. Diurnal habitat preference and vegetational structure were studied on 31 plots of four community types between 19 April and 21 October 1976. 'WoodCock were located with the aid of a pointing dog. The vegetational characteristics measured at each flush site include: canopy height, basal area, number of trees, saplings and shrubs, horizontal density, soil moisture, and distance to the nearest clearing. Discriminant analysis was used to compare the structural characteristics of: nest sites, non- flying broods, flying broods, and solitary woodcock during the breeding Dale Leslie Rabe season. Habitat utilized and not utilized by solitary woodcock during the spring, early and later summer in immature aspen, aspen, mixed deciduous, and alder community types was analyzed to identify important structural variables. A preference by woodcock for the aspen community type occurred during the summer, an increased use of immature aspen was noted during the spring and fall migration periods, and there was a steady increase in the use of alder until the fall migration. The density of understory vegetation of utilized woodcock habitat is very consistent and predictable between all the community types. Woodcock nests tend to be located in areas of relatively open habitat and broods have a tendency to move to denser habitat which is similar to that used by solitary wood- cock during the spring. Although woodcock shift to habitats with a greater number of saplings during the summer, the structure of the under- story vegetation remains constant. Woodcock also show a preference to use diurnal habitat which is located close to an opening. HABITAT UTILIZATION BY THE AMERICAN WOODCOCK IN NORTHERN MICHIGAN By Dale Leslie Rabe A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Fisheries and Wildlife I977 ACKNOWLEDGMENTS This study was a cooperative effort between the Department of Fisheries and Wildlife at Michigan State University, the Michigan Depart- ment of Natural Resources, and the U.S. Fish and Wildlife Service. Fund- ing was provided by the U.S. Fish and Wildlife Service through the accel- erated research program, contract no. 14-16-0008-628. I am indebted to my major advisor, Dr. Harold H. Prince, Depart- ment of Fisheries and Wildlife, and Carl L. Bennett, Michigan Department- of Natural Resources, for their assistance in developing the project design and editing of the final manuscript. I also wish to thank the committee members, Dr. Leslie W. Gysel, Department of Fisheries and Wild- life, and Dr. Donald L. Beaver, Department of Zoology, for their contri- butions to the study. Dr. John L. Gill, Department of Dairy Science,was very helpful in supplying statistical advice. Personnel from the Wild- life Division, Michigan Department of Natural Resources, and especially the staff at the Houghton Lake Wildlife Research Station, provided materials and manpower necessary to the project. My sincere appreciation is extended to those persons. Special thanks are in order for Richard Bennett and Albert Bourgeois for the dedicated assistance in the data collection. Finally, I wish to thank my parents, Gwen and Elmer, whose encouragement and assistance have been instrumental throughout my life. ii TABLE OF CONTENTS LIST OF TABLES .......................... LIST OF FIGURES ................. ‘ ......... INTRODUCTION ........................... STUDY AREA ............................ METHODS .............................. Singing Ground Study .................... Diurnal Habitat Study ................... Data Analysis ....................... RESULTS ......... I ..................... Census Routes ....................... Diurnal Habitat Selection ................. DISCUSSION . . . .‘ ........................ SUMMARY .............................. LITERATURE CITED ......................... iv vi Table- 1. LIST OF TABLES Summary of twelve habitat structure variables measured at each woodcock flush site. . . .............. 10 Correlation matrix of untransformed habitat variables (N = 336) .......................... 14 Analysis of singing ground census routes. Dependent variable is the number of displaying male woodcock per research unit ........................ 18 Summary of discriminant analysis on seven habitat para- meters contrasting utilized and non—utilized clearings as singing fields by displaying male woodcock, including the normalized discriminant function coefficients, scaled eigenvector coefficients, and Chi-square statistic. . 22 Time spent searChing and number of woodcock contacted, divided into four utilization classes, by community for spring period (19 April to 7 June) 1976 ........... 23 A comparison of habitat utilization by four groups of woodcock during the spring, expressed by three discrim- inant functions (I, II, III). The percent of among group variation is a measure of the importance of each function. The amount of group overlap is expressed by the percent of plots correctly classified, with 100% indicating no overlap. Largest differences between group means indicate the groups being discriminated. Variables contributing the greatest separation between groups in a function are identified by the largest absolute discriminant function coefficient ......... 25 Group means i S.E. and univariate F—ratio of untrans- formed habitat variables by site utilization. Habitat plots were statistically equalized to remove sampling bias between community types (F, df1 = 3, dfz = 113) . . . . 28 iv 8. IO. 11. A comparison of habitat utilization by solitary woodcock for three seasonal periods, expressed by two discriminant functions (I, II). The percent of among group variation is a measure of the importance of each function. The amount of group overlap is expressed by the percent of plots correctly classified, with 100% indicating no over- lap. Largest differences between group means indicate the groups being discriminated. ,Variables contributing the greatest separation between groups in a function are identified by the largest absolute discriminant function coefficient ......................... 32 Group means i S.E. and univariate F-ratio of untrans- formed habitat variables for solitary woodcock over three seasonal periods. Vegetation plots were weighted to equalize searching time in each community type. (F, df1 = 2, dfz = 173) ................... 36 A comparison of the habitat utilization by solitary woodcock in four community types, expressed as three discriminant functions (I, II, III). The percent of among group variation is a measure of the importance of each function. The amount of group overlap is ex- pressed by the percent of plots correctly classified, with 100% indicating no overlap. Largest differences between group means indicate the groups being discrim- inated. Variable contributing the greatest separation between groups in a function are identified by the largest absolute discriminant function coefficient ..... 37 Group means +S. E. and univariate F- ratio (F df 172) for untransformed habitat variables 0}: intary woodcock by community type. Underscored values indicate variables that were most important in discriminating between utilized and non-utilized plots within a community type. The symbol indicates whether the non-utilized plots were less than (<) or greater than (>) utilized values. Bracketed values indicate the number of non-utilized plots that were discrim- inated on the basis of that variable ............ 40 LIST OF FIGURES Figure . 1. Location of study area in northern lower Michigan and woodcock singing ground census routes on each research unit ............................. 4 2. Number (X i S.E.) of displaying male woodcock per census route for 1975 and 1976 based on three counts per route. Percent clearing along each route is listed below unit name ............................. 16 3. Regression analysis on the number of singing male wood- cock and the amount of clearing along the census routes for 1975 and 1976 ....................... 20 4. Graphical representation of utilization group means in standardized (mean of 50, standard deviation of 1) discriminant space. Scaling of axes is proportional to the among group variation accounted for by each function. Arrows indicate the direction of increase for the major discriminating variables ............ 26 5. Mean flush rate (number per hour) of solitary woodcock by community type from May through October of 1976 ...... 29 6. Graphical representation of seasonal group means in standardized (mean of 50, standard deviation of l) discriminant space. Scaling of axes is proportional to the among group variation accounted for by each function. Arrows indicate the direction of increase for the major discriminating variables ............ 33 7. Graphical representation of community type group means in standardized (mean of 50, standard deviation of 1) discriminant space. Scaling of axes is proportional to the among group variation accounted for by each function. Arrows indicate the direction of increase for the major discriminating variables ............ 38 INTRODUCTION The niche has been described in theory as an n-dimensional hyper- volume enclosing the complete range of conditions that are necessary for an organism to successfully replace itself (HutChinson 1957). Each dimension of the space represents a unique environmental variable (Pianka 1974). Habitat is one of the more important dimensions of an animal's niche. Lack (1944, 1949) developed the theory that interspecific compe- tition of birds has led to adaptations which cause individuals to select favorable environments for survival. Baker (1938) used the terms proxi- mate and ultimate to distinguish two types of environmental factors which influence habitat selection. Hilden (1965) defined food, shelter, and structural and functional characteristics of the species as ultimate factors. He further suggested that landscape, terrain, other animals, and nest, song, lookout, feeding, and drinking sites are some of the proximate factors that birds use. In studieshe conducted, Hilden found that in some avian species habitat selection is induced by proximate factors and finally selected on the basis of suitable ultimate factors. The theory of habitat selection by a species implies the existence of some underlying continuity in the factors that individuals utilize. An apparent contradiction of this premise is found in woodcock (Philohela minor) literature. Many of the studies on woodcock have emphasized the diversity of habitats that woodcock use. Marshall (1958) found that 'the presence of shrubs and woody vegetation was the only common characteristic 2 of the singing grounds he studied. Maxfield (1961) was unsuccessful in using plant species as an indicator of woodcock singing grounds. Mendall and Aldous (1943) and Sheldon (1967) reported finding woodcock nests in open fields, conifer plantations, mixed hardwoods of various ages, and even blueberry fields. Wenstrom (1973) found that woodcock broods in Minnesota exhibited preferential use of aspen, birch, mixed deciduous, and mixed conifer-deciduous cover types at various times during the breeding season. Dunford (1971), Mendall and Aldous (1943), and Wishart (1973) documented a seasonal habitat shift from upland deciduous commu- nities during the spring to lowland communities during the summer or in drought conditions. One of the reasons for the apparent variability in woodcock hab- itat is that many studies examined the vegetational composition and not the structural characteristics of the habitat. Some more recent investi- gations (Bourgeois 1976, Wenstrom 1973, Wishart 1973) have begun to examine woodcock habitat from a structural aSpect. The purpose of this study is to describe and compare some of the structural characteristics of woodcock habitat that determine the underlying continuity in the , A structure of both the breeding and diurnal habitats. STUDY AREA This study was conducted within the boundaries of the Houghton Lake State Forest, which is located in the north-central part of Michigan's lower peninsula (Fig. 1). Most of the work was done on six quarter township research units (3 X 3 miles) which are part of a large scale habitat manipulation study being supported by the Michigan Depart— ment of Natural Resources. Five of these units are located in Roscommon County and one is located in Kalkaska County. Data were also collected on selected areas around these units and in Missaukee County. The topography is characterized by rolling uplands and flat out- wash plains. The entire area is interspersed with numerous marshes and bogs. The soils of the uplands are mostly textured, sandy types with small amounts of loams, silt loams, and clay loams (Veatch 1924). Or- ganic soils predominate in the lowland areas. The important plant com- munities of the uplands are aspen (Populus spp.), oak (Quercus spp.), maple (Ager_§pp,), and pine (Eiflu§_§pp,). The lowland communities are mostly white cedar (Thuja occidentalis), spruce (Eigea_§pp,), and to a lesser extent larch (Larix_§pp,), and alder (Alnu§_§pp,). Agriculture and lumbering dominated the region in the late 1800's and early 1900's. Today many abandoned fields remain in the area, and the upland forests are mostly second growth hardwoods. Areas of various sizes within each quarter township research unit were clearcut until a total treatment of 25%, 50%, or 75% was attained f [yr [/w i“ Russell ,_ .. .. K_E_Y. _ Kalkaska m untreated land IE Co. (forested) [:1 manipulated londII 5mm" (clear cut) m 9'5"!" '0nd Missaukee Co. — census route === gravel road M-SS "=3 poor dirt road ---- trail MICHIGAN K = Kalkaska Co. M8 Missaukee Co. R: Roscommon Co. Lower PeninSula Study area in northern lower Michigan ll~ Townline/ ,11 m \;xl‘; _ L_\ Figure l.--Location of study area in northern lower Michigan and wood- cock singing ground census routes on each research unit. 5 (Fig. 1). Each treatment level was replicated twice. The size of clear- ings ranges from 10 to over 1,500 ha, and is largely related to the treatment level of the unit. Cutting on the 25% units was done during 1972. The majority of the cutting on the 50% and 75% units was completed by 1974 and 1975, respectively. Because cutting was done on the basis of a ranked order of comnunity types, the remaining vegetational commu- nities on the units vary considerably in type and relative composition. Aspen was the first to be cut, followed by oak and maple. The lowland conifers (cedar, spruce, etc.) were always retained intact. The shape of clearings is variable and frequently follows old community boundaries. METHODS Singing Ground Study , Woodcock singing ground censuses were conducted on all of the re— search units between 25.April_and 17 May in 1975 and 1976. The censuses were used as an index of population size in order to evaluate woodcock response to the three treatment levels. Two, 4-mi1e census routes were used to sample each unit. Each route had ten listening stations at 0.4 mile intervals. Since the maximum distance that a woodcock can be heard peenting on the ground is 0.2 miles (Duke 1964), an effort was made to. keep the two census routes separated by at least 0.4 miles. At the same time an effort was made to keep the routes located in the interior por- tions of the units to reduce possible interference from surrounding areas. The total amount of clearing along the tworoutes on each unit was commensurate (i 5%) with the prescribed treatment for that unit. Evening counts along each route were replicated three times in 1975 and 1976. Routes within a unit were run simultaneously whenever possible. Starting time and weather conditions for censusing followed the guide- lines established for the federal census routes (Blankenship 1957, Duke 1964, Goudy 1960, Westfall 1954). The number of woodcock heard peenting in a 2-minute period was recorded for each listening station. The sequence of sampling was determined by random selection of routes within a replicate group. Counts of the two routes within a unit were summed for the data analysis. 7 Since woodcock were never heard at stations in forested areas, the seven vegetational parameters were recorded only at listening sta- tions which occurred in a clearing. The dominant species of regenera- tion in the clearing was classified as either aSpen or oak-maple. The average height of regeneration in the clearing was visually estimated and classified as: low (0-2 m), medium (2-4 m), or high (>4 m). The density of regenerating vegetation was also categorized as: sparse (0-25%), open (26-50%), moderately dense (51-75%), or dense (76-100%). In addition, a record was made of all community types bordering the clearing and within 250 m from the listening station. Diurnal Habitat Study In 1976, a study was designed to evaluate diurnal habitat use by woodcock. A total of thirty—one search plots of four community types were chosen from areas throughout the described study area. The commu- nity types used in this study were selected because of the different structural characteristics each possess, and because woodcock are known to use these community types to some extent. Eight of the plots were in immature aspen, eight were in aSpen, eleven were in mixed deciduous, and three were in alder. The immature aspen, aspen, and mixed deciduous types were all aspen dominated communities which differ primarily in age. The immature aspen plots were less than 10 years old and were comprised of seedling sized stems (52.5 cm. dbh) of a fairly high density. TheSe areas were the result of regeneration after recent forest clearcutting. The aspen type on the other hand, ranged from 10-20 years and was com- posed primarily of sapling sized trees (2.5 to 12.7 cm. dbh). These plots occasionally contained a light mixture of conifers or other 8 hardwoods along with the aspen. Because of natural thinning, this habi- tat type is somewhat more open than the immature aspen. Mixed deciduous, which was the oldest and most open habitat type, is dominated by pole sized timber (12.7 to 22.9 cm. dbh) and varied in composition from nearly pure aspen to nearly an even mixture of aspen and other hardwoods. Conifers were rare in these plots. Finally, the alder communities were older than 10 years and were composed entirely of alder. All 31 plots were searched seven times from April to October 1976. Each searching period was designed to correspond to some event in the chronology of the woodcock throughout the time it is in Michigan. The first searching period (19 April to 20 May) encompasses the breeding season. The Second and third searching periods (28 June to 19 July and 4 August to 16 August) are chronologically related to the postbreeding dispersal of the juveniles, and the initiation of the molting period. The fourth searching period (11 September to 14 September) is associated I with premigratory staging activities. The fifth period (16 September to 22 September) follows the start of woodcock hunting season and relates habitat selection to hunting pressure. Finally, the last two searching periods (2 October to 8 October and 18 October to 21 October) were an attempt to evaluate habitat use during migration. The amount of tine spent searching a plot ranged from 0.5 to 1.0 hours depending on the size of the contiguous habitat. One hour was the maximum time spent searching regardless of its size. Flush rates were conputed by dividing the total searching time (hours) into the total number of woodcock contacted. Nest sites and broods were counted as a single contact. Constancy in effort was maintained by using the same searching crew and pointing dog. Searching was not attempted during 9 rainy weather or when temperatures went above 27°C. Searching of plots was systenetic, and every effort was made to avoid counting reflushed birds. A vegetational analysis was made of all woodcock flush sites during the first three searching periods. During the breeding season all woodcock contacts were classified as: nests, non-flying broods, flying broods, or solitary birds. After the breeding season, woodcock become indistinguishable when contacted and all birds were classified as solitary birds. Additional searching was done during the spring to supplement the vegetational analysis of nest and broad sites found on the searching plots. All searching effort was classified by community type, total searching time, and number of woodcock contacted (by class). Twelve structural measurements of the habitat (Table l) were taken at each indicated woodcock flush site. Six randomly selected sites were also sampled in each search plot where less than two woodcock were contacted during the first three searching periods. All vegetation measurements were made within 5 days after contact with the bird, except during the spring when horizontal density measurements were postponed, until after leafout to allow seasonal comparisons of habitat. All vege- tation sampling was centered about the flush site. The tree-sapling vegetation plot (7.5 X 15 m) was arbitrarily oriented in a north-south direction. If the plot transected a community boundary, it was rotated until it was contained in a homogeneous community type. 0f the 336 plots sampled, about 20% required rotating, and only one needed to be moved. The shrub plot (2 X 7.5 m) was oriented perpendicular to the tree-sapling plot. Canopy height was measured with a Haga altimeter. A Biltmore stick was used to measure the diameter of all trees. Trees, saplings, and shrubs were counted only if they were rooted within the plot. Percent 10 Table 1.--Summary of twelve habitat structure variables measured at each woodcock flush site. Variable name Description Canopy height* Basal Area* Trees* Saplings* Shurbs* Ground cover** Density (.5)** Density (1.5)** Density (2.5)** Min. density** Soil moisture** Distance to edge* Average height of overstory canopy. Basal area of all trees greater than 2.5 cm dbh in a .01 ha plot. Number of trees greater than 7.6 cm dbh in .01 ha plot. Number of trees 2.5 to 7.6 cm dbh in .01 ha plot. Number of woody plant stems greater than .45 m in height and less than 2.5 cm dbh in a .003 ha plot. Percent of ground covered with herbaceous plants less than .45 m in height, measured by 50 point intercepts. Average horizontal obstruction of vegetation from ground level to .5 m (%). Average horizontal obstruction of vegetation .5 to 1.5 m above ground level (%). Average horizontal obstruction of vegetation 1.5 to 2.5 m above ground level (%). Minimum horizontal obstruction of vegetation O to 2.5 m above ground level (%). Moisture content of soil measured as percent of dry weight. . - Distance from flush site to clearing greater than .1 ha (m). *Transformed by JX+.5 **Transformed by arcsin /R 11 ground cover was calculated by counting the number of point intercepts on a 50-point grid that contacted herbaceousplants and multiplying by two. All horizontal density measurements were made with a density board marked off in 0.5 m intervals. The board was rotated at the flush site and obstruction measurements were made from the corners of the tree- sapling plot. Soil moisture was determined from six, 15 cm core samples extracted around the flush site. The distance to the nearest clearing was measured (in meters) from the flush site to the opening in the over- story canopy larger than 0.1 ha. Data Analysis A partially nested analysis of variance model was employed in the experimental design to test the effects of time and amount of clearing on the number of singing male woodcock, and the effect of time and community type on the flush rate of woodcock in diurnal habitat.’ In each case, the dependent variable was tested for normality and heter- ogeneous variance using the Kolomogorov-Smirnov and Bartlett's test, respectively (Sokal and Rohlf 1969). The number of displaying woodcock was found to be normally distributed (P>.05) and have homogeneous var- iances between treatment groups (P>.05), so a parametric analysis of variance was used to test the main effects and interactions. The flush rates of woodcock in diurnal habitats were not normally distributed (P<.05), and transformatiOns were unsuccessful in normalizing the data. As a result, a Kruskal-Wallis one-way analysis of variance was used to test the effects of habitat, and a Friedman two-way analysis of variance was used to test the effect of time (Siegel 1956). The interactions could not be evaluated with these non-parametric tests. 12 (For analyzing the structural features of woodcock habitat, dis- criminant analysis was used to statistically distinguish between var- ious groups (community types, seasons, and utilization classes). The discrimination is accomplished by measuring a number of parameters (p) for which all of the groups (9) are believed to differ (Lachenbush 1975). The discriminant analysis plots each case (individual within a group) in a p-dimensional space. A maximum of n (g-l) linear discriminant func- tions are used to separate the groups (Cooley and Lohnes 1971). The n orthogonal functions are computed in such a way as to maximize the among group variation for each function. The absolute value of the discrim- ‘inant function coefficient is a measure of the relative discriminating power of each variable in a function (Green 1971). The group means (location of a typical group number) gives the relative location of the ' groups in the n-dimensional space. The percent of among group variation isa measure of the relative importance of each function to the discrim- inant analysis. Examination of the correctly classified, known-group memberships is a good indication of the overall discriminating ability of the analysis. A stepwise discriminant analysis procedure (Nie et a1. 1975) was used for all habitat comparisons (except singing ground vegetation) to eliminate those variables which did not contribute significantly to the discriminating power of the functions. The stepwise method used in this study is designed to maximize the overall group separation. The minimum criteria for entry into the analysis was Set at F = 2.80, which roughly corresponds to the P = .10 significance level. Discriminant function coefficients of habitat variables that could individually account for 20% or more of the discriminating power were considered important in 13 that function. It was necessary to transform all habitat variables (Table 1) in order to meet the as$umption of normality (Kolomogorov- Smirnov, P >.O5). Variables expressed as percentages were transformed with arcsin /7, and all other variables were transformed with‘/Y7E75.' Horizontal density (0 to .5 m) was the only variable not normalized by the transformation. Discriminant analysis is robust enough to tolerate such moderate deviations from the underlying assumptions (Nie et a1. 1975). The correlation matrix of habitat variables (Table 2) indicates that all correlation larger than .14 were significant (P <.01), and the largest correlation was .84, which occurred between the horizontal den- sity (.5 to 1.5 m) and minimum density (0 to 2.5 m). Even this corre- lation is not large enough to cause any problem in computing the dis-. criminant functions or interpreting the results (Nie et a1. 1975). All statistical procedures were run on the Michigan State Uni- versity CDC 6500 computer using a programmed statistical package. 00.. 0 00 00000000000. 0000 00 00000000 00 00 0.0-0.00 00 0.-0v 00000000 .0 000002 02 00v 0000000: 000m 2m 0000000 _000000Lo: mo 0000000 000000000: Pa 0000» we 000202 02 00 0.0-00 0000000 , As 0.0.0.0 000 L2,00 000000 00 0000 00000 00 000000000: 5:50:02 o: 0000ch 000000000: No mazcsm 00 000502 :m “0000: 000000 :0 000000000 000000: 14 00. .00.- .00.- .00.- .00.- 00.- .00.- 00.- .00. .00. .00. 00 .00.- 00.- 00.- 00.- 00.- 00. 00.- 00.- 00.- 00. :0 .00. .00. .00. .00. .00. 00. .00.- .00.- .00.- 0: .00. .00. 00. .00. .00. .00.- .00.- .00.- 00 .00. .00. .00. 00. .00.- .00.- .00.- 00 .00. .00. 00. .00.- .00.- .00.- 00 .00. .00. .00.- 00.- .00.- 00 00. .00.- .00.- .00.- :0 .00.- .00.- .00.- 02 .00. .00. 02 .00. <0 00 00 :0 02 00 00 00 00 I0 02 02 00 I0 .Aomm u 2V 00—00000> 000000: 00500.0:000:: $0 xwgume :o0pmpmccounu.m @0000 RESULTS Census Routes The number of displaying male woodcock using the research units increased significantly (P = .013) between 1975 and 1976 (Fig. 2). There was however, a disproportionate increase between the 25%, 50%, and 75% treatments as evidenced by the moderately significant (P = .120) interaction between treatment and census year (Table 3). The 25% treat- ment units showed a 40% increase in woodcock use between the 2 years, while the 50% and 75% treatments increased by 114% and 105% respectively. The woodcock use between treatment levels showed a marginally signifi- cant (P = .097) response, but the individuality of routes (research units) within a treatment level was found to be more significant (P = .001). Examination of the within route variance relative to the between route variance reveals why that is the case (Fig. 2). The significance (P = .082) of the interaction, route within treatment X census year, is caused by a slight decrease in woodcock use of the Townline unit in 1976, while the M-18 unit increased substantially. To determine where signi- ficant (P <.05) differences occurred, treatment levels were tested by census year using Tukey's multiple range test (Sokal and Rohlf 1969). The results for both 1975 and 1976 showed no difference in the number of displaying woodcock between the 50% and 75% treatments, while the 25% treatment was significantly lower than the other two. There was a significant (P <.05) curvilinear relationship between the amount of clearing along census routes and the number of displaying 15 16 .050: 00:: 30—00 000000 00 00:00 2000 0:000 @0000000 0000000 .0000; 000 000000 00000 no 00000 0000 0:0 0000 000 00:00 030000 000 xuouuooz 0005 0000000000 00 0.0.m 0 xv L0nE:z-.N 000000 l7 é§ h\\\\§$, 3'5: 02 I: 3 ........... //§ NNNNNNNNNN NNNNNNNN )IOOOOOOM :10 aaewnw |'-///////////>/,§ RESEARCH UNIT 18 emf? .VN LOLLm Fmsvwmmm mmo. 00m.m. 00.00 m 0000 030000 x 000500000 000003 00300 00—. 000.0 mo.mm N 0000 030000 x 0000000 00 00305< 000000000000 003-030 m_o. . mom.mm mo.omm — 030000 mo 000> Poo. 0N0.o— 00.00 m 000500000 00000: 00300 030000 000. 000.0 00.000 0 00000000000 00000 000000 00000 0000000 00 00305< 0000000 000: 00000000000m 0 000300 0002 00 00000000> 00 000300 .0003 00000000 000 50000003 0005 0000000000 00 000530 000 00 0000000> 000000000 .000300 030000 003000 0000000 00 0000000<-.m 00000 l9 male woodcock for both 1975 and l976 (Fig. 3). The equation accounted for 41.8% and 77.4% of the variation for 1975 and 1976, respectively. It also predicts a maximum number of displaying males at 52% and 58% clearing along census routes for the twoyears. A discriminant analysis was used to compare various habitat para- meters related to the use of clearings as singing grounds by one or more woodcock (Table 4). The presence of an established aspen community bordering the clearing was the most significant (X2 = 20.0l, P <.00l) parameter associated with utilized clearings. This parameter accounted for 33.5% of the total discriminating power of the function. Although height of the vegetation in the clearing was the second most significant (X2 = 4.72, P <.l0) parameter, it did not discriminate as well as the two variables describing the dominant species of regeneration in the clearings, which were statistically insignificant. Oak and pine commu- nities bordering the clearings were the least significant and least powerful parameters in the function. Diurnal Habitat Selection Eleven woodcock nests, l0 non-flying broods, 25 flying broods, and 76 solitary birds were located during 71.9 hours of searching in four plant communities between l9 April and 7 June 1976 (Table 5). Chi- square analysis was used to test habitat preferences of flying broods, solitary birds and all utilization classes combined. Woodcock do not appear to select between habitats (P $.10 for all comparisons). Analysis was not performed on nests and non-flying broods because of the small sample sizes. Since no habitat preference was apparent, the number of vegetation plots in each community type and utilization class was statistically 20 .0000 000 0000 000 000300 030000 000 00000 00000000 00 003050 000 000 00000003 0005 0000000 00 000530 000 00 00000000 0000000000--.m 003000 21 3.0 00300 030200 020.2 02.50030 00 00 00 00 . 00 00 _ 0 _ 0 0 _ V‘ I are. "N0 QNN.NV l NXNOO. l XmVN. H > 4 4 I Q I . .I mhmw I w - 1. I/ _. -00 < < :3 ‘ VNN. H N.— IGF GON.O¢ I NXOwO. I XQNO.N H > ION 4 ORG? 1 INN INF {or xooooooM :IO uaawnw IQN 22 .00000030 000005000000 000 00 0000000> 000 00 000030000000 0>00000000 .000.v 0 00 000000050000 00v 00. 0.0 000.- 00000000 000000000 000035500 0000 000 00. 0.0 000.- 00000000 000000000 000035500 000 000 00.0 0.0 000.- 00000000 00 0000000m0>.00 0000000 000 00.0 0.00 000. 00000000 00 0000000m0> 00 00000: 000 00. 0.00 000.- 00000000 00 0000000 00000500 00 0000< 000 00.0 0.00 000.- 00000000 :0 0000000 00000500 00 0.005-000 000000.00 0.00 000.0 00000000 . 000000000 000035500 0000< A000 000300-000 00000000000 00000000000 000050000 ”3000000000030 000000 00000030 000005000000 ‘ .000000000 000300-000 000 .000000000000 00000>00000 000000 .000000000000 00000030 000005000000 0000—05000 000 00003—000 .00000003 0005 0000000000 00 000000 0000000 00 000000000 00000003-000 000 00000003 00000000000 0000050000 0000000 00>00 00 00000000 000005000000 00 >00553m--.0 0000» 23 00 0 o . o 0.0 0000< 0_ 00 0 0 . 0.00 030300000 0000: mm 00 m 0 0.0m c000< 00 o 0 o 0.0 00000 00000550 0000000m 00000 000000 00000 000000-002 0002 0000000m 00:0: 0000 000005500 00000 0000000_00= .0000 00:00 0 00 0000< 000 000000 000000 00» 000005500 00 .0000000 0000000000: 0000 00:0 0000>00 .000000000 00000002 00 0005:: 0:0 000000000 00000 0505--.m 00000 24 equalized to remove sampling bias. Discriminant analysis yielded three significant functions that accounted for 43.0%, 38.3%, and l8.7% of the among group variation (Table 6). The functions were able to classify correctly 75.0% of the nest sites, 43.8% of the non-flying_broods, 62.1% of the flying broods, and 54.4% of the solitary birds. Minimum density of vegetation (0 to 2.5 m) was the most important habitat variable in Function 1. Function II is dominated by canopy height and basal area. The important variables in Function III are canopy height, vegetative density (.5 to l.5 m), and minimum density of vegetation (0 to 2.5 m). These functions are combined in a three dimensional representation of habitat volume (Fig. 4). According to this representation, the under- story density is greater in areas utilized by Solitary birds than those used by nesting birds and broods. Nest sites, on the other hand, are distinguished from broods and solitary woodcock by a more open habitat' with higher canopy and smaller basal area (Table 7). Mean flush rates of solitary woodcock in four plant communities ranged from 0 to 7.3 birds per hour over a 6 month period (Fig. 5). Chi— square analysis revealed that there was a significant relationship (X2 = 7.22, P <.Ol) between the absence of an opening near the search plot and the absence of woodcock using the plot. Consequently, only the 26 plots near openings were used in the analysis of flush rates. Al- though statistical significance between community types at P <.05 was not achieved for any of the sampling periods, probabilities of 0.15, 0.l2, and 0.07 for differences between community types were calculated for the May, June, and August sample periods. During the spring period (19 April to 20 May) woodcock use of immature aspen, aspen, and mixed deciduous was essentially equal, but use of alder was noticeably lower. 25 .000.v 0 00 00000000000 000000030 000* mno.u mmm.n m0m.n coo. wvo.u mmp.0u mmo.n omN.u mum. 00m.: 000.: mmm.u 000. . Nm0.u mom. mmm. Fmo.0n 000.- 000. 000.0 mmm. , 00 0o 000000>00 00000000 .0 0o 00000 000000000000 00000000 000005000000 000000000000 0.0m mwm. mmm.n mmo.—u _.No 000. 0mm.u wwm. m.m0 ooo.n 000. om_.u o.mm .mpv. me.0 wmo. AF 00 000000>00 00000000 .0 00 00020 00000 000005000000 000000000000 00 00005 00000 0000000000: 0.00 0.00 0.00 000 0000:0000 00000000> 00000 00050 00 0000000 0000 00 00000000 00000005 0000 0000000 .00: 00.00 0000000. 00.00 0000000 00.0 0000000 00>00 000000 000000 00000000 00000 0000 00000 000000 000000 00000000 0000 00000000 00000 000000 00000 000000-002 0002 .00000 0000000000 000000000 000 00 0 00000 00 0000000 000000000 000005000000 .0000000 -0000 00000000 000005000000 00000000 0000000 000 00 0000000000 000 00000020 0 00 000000 0003000 0000000000 00000000 000 000000000000 00000000> .0000005000000 00000 000000 000 00000000 00005 00000 0003000 00000000000 0000000 .00000>0 00 0000000000 0000 0003 .0000000000 000000000 00000 00 0000000 000 00 0000000x0 00 00000>0 00000 00 000050 000 .00000000 0000 00 0000000050 000 00 0000005 0 00 00000000> 00000 00050 00 0000000 000 .0000 .00 .00 000000000 000005000000 00000 00 0000000x0 .000000 000 000000 00000003 00 000000 0000 00 0000000000: 0000000 00 0000000500 <--.0 00000 .00000000> 00000005000000 00005 000 000 00000000 00 000000000 000 00000000 03000< .00000000 0000 00 000 000000000 0000 -0000> 00000 00050 000 00 000000000000 00 00x0 00 0000000 .00000 000005000000 A0 00 000000>00 00000000 .00 00 00050 00~000000000 00 00005 00000 00000~00000 00 00000000000000 000000000--.0 000000 27 000mm Oz.>..u_-zoz 000207///// 0‘ 000mm 02.>._.... 0‘ >¢<.:.._Om lav 1m? Ion l_m tum III 'NOIlONfld ALanaa‘NIw-0- (Fl)AllSN30->- .LHDIBH AdONVO <‘ 28 .00.v 0 00 000000000000 00._ 0.0 0 m.m~ mwm 0 N.00 N.0 0 0.00 0.0 0 m.NN 0000 00 00000000 00., 0.0 0 m.mm 0.0 0 0.00 0.0 0 0.0m 0.N 0 0.0m 00000005 0000 000.0 . 0.0 0 0.00 0.0 0 0.00 0.N.0 0.00 0.0 0 0.00 0000000 .002 000.0 0.0 0 0.00 0.0 0 0.00 0.N 0 0.00 0.0 0 0.0m Am.mv 0000000 00.0 0.0 0 0.00 m.N H 0.N0 0.N 0 0.00 0.N 0 0.00 00.00 0000000 00.0 0.0 0 0.00 0.N 0 0.00 0.0 0 0.00 0.0 0 0.00 00.0 0000000 moJN 0.N 0 0.00 0.0 0 0.00 0.N 0 0.00 0.0 0 0.0m 00>00 000000 000.0 0.0 0 0.00 0.0 0 0.0m 0.0 0 0.00. 0.0 0 m.mm 000000 000.00 0.0 0 0.00 0.0 0 N.N0 m.m 0 0.0m 0.m 0 0.00 00000000 000.0 0.0 0 0.0 0.0 0 0.0 . 0.0 0 0.0 0.0 0 0.0 00000 *0N.00 0.0 0 0.00 . 0.0 0 0.00 0.0 0 0.0— 0.0 0 0.00 0000 00000 000.0 0.0 0 0.00 0.0 0 N.__ 0.0 0 0.0 0.0 0 0.00 000000 000000 00000-0 0000000m 00000 00000 0002 0000000> 000000 000000-002 . .00: u 000 .0 u 0.6 .0 00000 000005500 0003000 0000 00000500 0>0500 00 000000000 0000000000000 0003 00000 000000: .00000000000 0000 00 00000000> 0000000 0050000000000 00 00000-0 000000>000 000 .m.m 0 00005 00000--.0 00000 29 .0000 00 0000000 0000000 002 0000 0000 000000000 00 00000003 00000000 00 00000 000 0000000 0000 00000 000zus.0 000000 30 0>o.z 0.000 Zmaw< mm3h<22_ whamw 0:3 F.0:< P P>JDfi pmzafi wDODnZUwD Dmx.2 zmam< UflOH 83d xaawnN 31 In early summer (28 June to 19 July), woodcock flushes per hour increased from 1.3 to 7.3 in aspen. Although use in mixed deciduous and alder also increased moderately, immature aspen remained unchanged. In late summer (4 August to 16 August), utilization of aspen by woodcock remained sub- stantially higher than the other community types, but began to decline in September. Mixed deciduous and alder continued to increase in August and immature aspen remained the same. In September, the use of immature aspen and alder by woodcock increased, while mixed deciduous and aspen declined. From the beginning of woodcock hunting season (15 September) through October, use of all four community types by woodcock was indis- tinguishable. Comparisons were made between the structural characteristics of habitat used by solitary woodcock during the spring, and early and late summer. The discriminant analySis was statistically equalized for searching time in each community type to remove sampling bias. The num- ber of vegetation plots in each community type was weighted to reflect the mean flush rate for each sampling period. Ground cover and soil moisture were not included in the analysis because of the temporal changes which occur in all community types. The results of this analysis produced two significant (P <.001) functions which accounted for 73.9% and 26.1% of the among group variation (Table 8). These two functions classified correctly 72.5% of the spring sites, 58.5% of the early sum- mer sites, and 53.3% of the late summer sites. The most important var- iables in Function I were canopy height and the number of saplings (Fig. 6). The number of trees and distance to a clearing edge were the two most important variables in Function II. Structural differences between hatitats used in Spring and summer are based primarily on a 32 .0oo.v 0 00 00000000000 000000000 000* 000. 000.- 0000 00 00000000 I - 0000000 .002 000.- 000.- 00.00 0000000 000.- 000. . 00.00 0000000 000.- 000.- 00.0 0000000 . . 000000 000. . 000. 00000000 000.- 000. 00000 . . 0000 00000 000. 000. 000000 000000 00 00 000000>00 00000000 .0 00 00020 0000000> 000000000000 00000000 000005000000 00N000000000 0.00 000. 000. 0000 005500 0000 0.00 000.- 000. 0000 000000 00000 0.00 000. 000.0- 0000 1 000000 00 00 000000>00 00000000 .0 00 00020 .020 00000 00000 000005000000 00N000000000 00 00005 00000 00000000 0.00 0.00 000 000000000 00000000> 00000 00050 00 0000000 0000000000 000000000 00 0 00000 00 0000000 000000000 000005000000 .00000000000 00000000 000005000000 00000000 0000000 000 00 0000000000 000 00000000 0 00 000000 0003000 0000000000 00000000 000 000000000000 00000000> .0000005000000 00000 000000 000 00000000 00005 00000 0003000 00000000000 0000000 .00000>0 00 0000000000 0000 0003 .0000000000 000000000 00000 00 0000000 000 00 0000000x0 00 00000>0 00000 00 000050 000 .00000000 0000 00 0000000050 000 00 0000005 0 00 00000000> 00000 00050 00 0000000 000 .000 .00 000000000 000005000000 030 00 0000000x0 .0000000 00000000 00000 000 00000003 00000000 00.00000N00000 0000000 00 0000000500 0-1.0 00000 . .00000000> .3 . 00000000000000 00000 000 000 00000000 00 000000000 000 00000000 03000< .00000000 0000 00 000 .5 000000000 00000000> 00000 00000 000 00 000000000000 00 00x0 00 0000000 .00000 000000000000 AF 00 000000>00 00000000 .00 00 00000 000000000000 00 00000 00000 00000000 00 00000000000000 000000000--.0 000000 34 PIG-m: >a02<0 m02.4n4mchm mcwcqm anmem> .AmNP u New .N u _ce .av HQHH »H_::esou gumm cw we?» m:_;ucmmm mNHszcm op umpsmwmz mew: mpopa cowumgmmm> .muowcmq _m:ommmm mmcgp cm>o xuouvooz acmuw_om com mm_nwwcm> “muwaw: nmecowmcmcpcz mo ovumc-m mumwcm>wc3 use .m.m H mcmms a:ocu--.m mpamh 37 ._oo.v a He Hceowc_cucm meowHucsc __ou ecsogo - - - mnzcsm mmo.- coo.F- H_H.- Hm:H_QHm mom.P- moo. mu_. mwmch com. mom.- mxw. mmcm ~mmmm “mp. Nmm.- mum._ ucmwmg.xqocmu A_ $0 cowumw>mu vcmucmum .0 mo :mmzv mpnmwcm> mpcmwuwmwmou cowuucam ucmcwewLUmwu uwNHchncmum m.om mum. mmm.m- oo—.m- AFFV cmup< m.pm Npm. m-.- m__.m Aqmv msozuwomv cmx_z m.om mmo.- mo_. o_o.- ANN cmqm< N._m mmo. new. m—m.mu Awm :mqmm mczmeEH AP mo co_umw>mu acmucmpm .o mo :mmzv NzM, azoco momam pcmcwewcumwn umeanvcmpm cw mamme qzocm xpwczeeou m.m P.¢_ o.on Lo» umpcsouum :owumwgm> asocm mcoEm mo ucmucmm echHHHH_U »_Humccou HHH HH H mac—a mo ucmocma «cowbocsm pcmcwswcumwo .pcm_qu -mmoo cowuucaw acmcHEwLumwv mu:_0mom ummmcmp ms» xn vmwwwucmvw mcm cowuucsm m cw masocm cmmzumn cowpmcmamm pmmpmwcm ms» mcwuznwcpcou m_nwwcm> .umum:_EHLUmwu mcwma masocm mg» mpmu_u:H memos asocm :mmzumn mmucmcmwmwu ummmcmb .ampcm>o o: mcwumo_vcw xoo— cur: .umwmwmmm_u appumccou muopa mo ucmucma mg» »a uwmmmcaxm m? amFLm>o azocm mo pesoEm mg» .co_pu::$ comm mo mucmpcoaew mg“ co mesmmms m mw cowpmwcm> azogm macaw Ho pcmocwa web .AHHH .HH .Hv mco_uu::m pcmcwswgumwu mmccu mm nmmmwchm .mmaxu xuwczseoo Lao; cw xoouuooz acmprOm an cowuoNPPwp: umbram; any we cemwcmqsou <--.o~ m—nwp .mmpnmwcm> mewpmcw -Ewcomwu Loves mg» L0H mmmmcoc_ mo cowpomcwu mcp mpmuwvcw mzocc< .cowpuczw sumo An Low umpczouum mm :owumWLm> azocm macaw mcp on chowucoaoca m? mmxm mo mcwpmum .mumam ucmcwswcomwn A_ to cowumw>mu ucmucmpm .om mo :mmsv nmeucmccmpm cw magma azocm max“ xpwczssoo Ho cowpmgcmmmcamc Pmuwcamcw--.m mczmwm .39 mDODD .0mo Gwyn-2 a. v. ww .40 dd. 5 V6 ”0 N.“ e\V VQ / 00 e0 ave J! \ . hm. k0? J, 0 Q4 b be. zmam< 0» 32:25: on. [on \.mmo..< roe /\: [on low ION. III NOILOanl 83381 -> VBUV 'IVSVB <- 4o .mo.v a HH HCHHHHHCHHWH Hmm.H m.m H H.F_ _.m H m.m~ ANV m._ H «.mPH Amv.mqmnmuflqu Hmem 0H HUCHHHHD Hmo.~ ¢.H H m.mm A_v m.m H e._ev A_v ~.~ H m.mev Amy _.H H m.mmA HHHHCHH .cwz H_m.e_ Amy N.H H N.mkv ANV m.m H N.mmv A_v K.N H _._Hv o.e H o.m~ AH.NV HHHHCHQ No._ AFV m.m H m.mov H.m H o.oe m.~ H N.m© m.m H N.mu Am._v HHHHCHQ HON.N N.H H N.Nm m.F H “.mw m.F H m.Hm a._ H m.mm Am.v HHHHCHQ Hme.m A_V _.m H m.H®A Amy m.~ H o.mmv m.N H N.Hm o.m H N.mm Hw>ou ecsocu H_N.N m.o_ H m.mm Amy H.N H o.NNv H.N H m.N~ N.H H H.HH Haaccm HNF.mH m.o_ H m.NN_ Amy N.P H m.m_v AFV H.N H m.MMA A_v m.m H _.NmH HGCHPQHW Hm_.ue N.o H m.o 0.0 H m.m 0.0 H o.“ H.o H m._ HHHHH Hou.Nm _.P H m.__ m._ H H.mN m.o H H.m~ Amy mamumqumv Haze _Hmmm Hmm._e_ A_V.M4mumumqmv m.o H N.~F H.o H N.o_ m.o H _.m HHQHH; Haocmu ovumcum Lmup< mzozuwumn wmxwz cmam< cwamm mgsumEEH m_nwvcm> . .mpnmwcm> Hosp Ho mwmmn ms“ :0 umpm:_ -Ewcumwu mew: um:p_muo_q um~_~_p:-:oc Ho Logan: mgp mpmuvucw mm3~m> umpmxumcm .mmapm> vaw_wp: AAV cmgu memmgm Lo AVV can» mmmp mcmz mac—a nmNmpwuauco: as» gmcpmsz mmumoquP _onexm mg» .mQAQ xpw::EEou m c_:u_3 mpofiq umNPPwu:-:oc can ume__p: cmmzpon mcwpmcHs_LUmwn c_ “caucoasw “woe mcwz “may mm_am_cm> mpmuwucw mm:_m> vmcoumcmvc: .mazu apvcseeou xn xuouuooz acmHHFOm Ho mmFQmwcw> pmymnm; umEcommcmchz com ANNP u Nmu .m u Fmv .uv ovumc-u mpmwcm>wcz new .m.m H mcmms azocwuu._~ mpnmp 41 and two based on basal area and number of shrubs. Canopy height, number of trees, and the two lower horizontal density classes were the least important variables in discriminating between utilized and non-utilized plots. DISCUSSION The use of singing grounds by displaying male woodcock is influ- enced by the interspersion of diurnal habitat (forest) and singing ground habitat (fields). The response by singing males to an increased amount of clearing is a non-linear relationship. Based on the regression equa- tion, all indications are that the maximum woodcock use can be expected when there is roughly an equal (50:50) interspersion of the two habitat types. As the balance shifts toward excess forest, singing fields ap- pear to become limiting, and when the shift is in the opposite direc— tion, the diurnal habitat becomes limiting. The vegetational characteristics of diurnal and singing ground habitat are also important in determining the suitability of an area for woodcock use; The presence of an aspen community adjabent to a clearing is very important in determining if a clearing will be used as a singing field. When aspen is compared to oak or pine communities, it is rather conclusive that aspen is a preferred habitat in this region. There is some additional evidence that the height and density of vegetation in' the clearing are also important, but to a lesser degree. Sheldon (l97l) concluded that the only requirement for a singing ground is a "get away" route for the bird's aerial flight. Maxfield (l961) found that the height of vegetation in an opening tolerated by woodcock varied directly with the size of the singing field. Woodcock are known to be opportun- istic in use of openings for singing (Blankenship T957, Mendall and Aldous 1943), and even clearings with extremely dense regeneration will generally have a number of grassy pockets that meet the requirements. 42 f 43 The number of singing male woodcock on the quarter township units increased from 1975 to 1976. This increase was compared to 33 federal census routes in northern Michigan that also were censused in both years. A tetest showed no significant (P >.25) change in the number of dis- playing birds. This suggests that the increases on the units are in response to the habitat manipulations and not a regional population in- crease. Time and treatment are confounded and it is not clear if the change represents improvement in the habitat condition between the two‘ years or a delayed response to the units. Since the 50% and 75% units increased more than the 25% units, and since cutting on the 25% units was completed about 2 years before the others, it leads me to believe that the increase is a delayed response by the woodcock to the newly created habitat. The next question that arises is whether or not the structural features preferred by the male woodcock are the same as those preferred by the females during the breeding season. Comparisons were made be- tween the habitat structure selected by nesting females, non-flying broods, flying broods and solitary birds (presumably males). The dis- criminant analysis affirms that there are some distinct differences be- tween the groups. The solitary woodcock shows a distinCt tendency to. select diurnal habitat that has 50% more vegetational obstruction in the understory than sites used by nesting birds and broods. The impli- cation is that the solitary birds perfer the extra seclusion offered by the denser vegetation. This type of habitat would allow the birds to move without increasing their vulnerability to predators. Nesting fe- males, on the other hand, seem to prefer habitat with a higher canopy and smaller basal area than broods or solitary birds. After the eggs 44 hatch, there appears to be a tendency for the female to move the brood into a younger, more dense habitat. This wbuld allow the foraging chicks to move about freely with less chance of being detected by a predator. Bourgeois (l976) found similar results in analysis of nest and brood habitat in northern Michigan. Nest sites are most distinctly different from the other utilization Classes. Sites used by broods and solitary woodcock overlap more in habitat parameters preferred. There is a certain amount of shifting in habitat use by solitary woodcock throughout the seasons. In the early spring, and again after the beginning of w00dcock hunting season, there is no indication of a habitat preference. There is evidence that alder communities are avoided in the spring. .This is understandable, since alder is frequently inun- dated with water during that period. Throughout the summer months (June to September), aspen excels as a preferred habitat type. Although mixed deciduous communities are utilized by woodcock throughout all seasons, there is a great deal of variation in the degree of use between search 9 plots. This appears to be related to the greater variability of critical structural features within the mixed deciduous community type. Alder shows a progressive increase in woodcock use until mid-fall. This is probably related to a shift in food supply as Soil moisture in the various community types change. Dunford (1971) observed that woodcock moved to lowland coverts during periods of drought. Mendall and Aldous (l943) found birds in mixed and deciduous woods in the spring but in alder later in the year. Immature aspen shows its greatest use during the spring and early fall, relative to the other habitat types. It is interesting to note that the greatest use of immature aspen is associated with the arrival and departure of woodcock in northern Michigan. The use of this 45 type may be primarily related to woodcock migrations. Wenstrom (l973) found that female woodcock had a distinct preference for upland brushy habitat during the spring migrational period. Flush rates indicate that most woodcock had migrated by 1 October. The slight increase in all habitats for mid-October may be caused by late migrants moving through the area. Structural differences in habitat selected by solitary birds are related to the number of saplings and average canopy height. Summer use by woodcock tends to be in older habitats with slightly more sapling~ sized trees (2.5 to 7.6 cm. dbh). This type of structural shift would be predicted from examination of the flush rate data. The percentage of correctly classified plots is an indication that there is better discrim- ination between spring and summer than between early and late summer. The subtle structural shifts that do occur are probably related to the onset of the molting period. Some interesting relationships develop when discriminant analysis is used to contrast the structural features of solitary woodcock sites between community types and also utilized and non-utilized sites within each community type. It is evident that the discriminant functions are able to do an extremely good job of classifying the various community types. However, the variables that are important in discriminating be- tween community types are largely unimportant in discriminating between utilized and non-utilized areas.‘ There is one exception, the number of saplings is important in both contrasts. This can be attributed to the difference in growth form of sapling-sized trees in alder com- pared to aspen communities. The understory structure (saplings, shrubs, horizontal density, and ground cover) is the most important 46 feature determining the use of an area by woodcock and is consistent between all community types. All deviations from utilized parameters in immature aspen indicate that non-utilized areas tend to be younger and much denser in the under- story structure. It appears that woodcock are reluctant to penetrate expansive plots of the immature aspen type. Mixed deciduous tends to be too open in the understory to satisfy woodcock requirements. There is an indiCation that aspen may occasionally be too open in the understory, but the most prevalent problem appears to be related to the distance to a clearing. The two non-utilized aspen plots were distantly separated from any openings by other forest types. Apparently, woodcock are reluc- tant to seek out these areas. The causes of non-utilization of alder communities is less clear, but, in one case, seems to be related to ex- cessive grassy ground cover. Also, there is evidence that older stands of alder become too open in the understory. All evidence indicates that woodcotk are closely associated with specific habitat structure and the proximity to a clearing. Among the . three aspen community types, it is interesting to note that woodcock. penetration of the habitat is directly related to the age of the commu- nity and density of the understory. Even so, there seems to be a limit to the depth of penetration for each community type. SUMMARY The density of woodcock use during the breeding season is directly related to the amount of interspersion between suitable diurnal and singing ground habitat. . ‘ . In general, aspen seems to be the preferred community type in Michi- gan. There is a strong tendency towards the middle-aged stands (lo-20 years) for most of the summer and early fall. Gross vegetational features of a clearing are not as important in determining the use of a singing ground as the distance to suitable diurnal habitat. Understory vegetational density and proximity to a clearing are the most important features in determining utilized woodcock habitat. Middle-aged aspen communities most consistently meet these habitat requirements. Immature aspen and mixed deciduous communities fre- quently falloutside the acceptable range of habitat structure for woodcock. Woodcock nest sites tend to be in more open areas relative to broods and solitary woodcock. Solitary woodcock prefer areas with the highest vegetational density of any utilization class. Habitat structure related to seasonal use by solitary woodcock shows a shift towards the middle-aged aspen communities_during the summer, but no significant change in the understory structure occurs. 47 LITERATURE CITED Baker, J. R. 1938. The evolution of breeding seasons. B. R. deBeer: Evolution. Essays on aspects of evolutionary biology presented to E. S. Goodrich. pp l6l-l77. Oxford. Blankenship, L. H. l957. Investigations of the American woodcock in Michigan. Mich. Dept. Conserv. Rep. 2123. 2l7pp. Bourgeois, A. l976. Analysis of American woodcock nest and brood hab- itat in northern lower Michigan. M.S. Thesis. Mich. State Univ. 33pp. Cooley, W. W., and P. R. Lohnes. l97l. Multivariate data analysis. ,Wiley, New York. 364pp. Duke, G. E. *l964. Study on the reliability of censuses of singing male woodcock. M.S. Thesis. Mich. State Univ. 49pp. Dunford, R. D. l97l. The summer behavior of American woodcock in cen- tral Main. M.S. Thesis. Univ. Maine. 93pp. Goudy, W. H. 1960. Factors affecting woodcock spring population indexes in southern Michigan. M.S. Thesis. Mich. State Univ. 44pp. Green, R. H. l97l. A multivariate statistical approach to the Hutchinsonian niche; bivalve molluscs of central Canada. Ecology 53zl26-l31. . Hutchinson, G. E. 1957. Concluding remarks. Cold Spr. Harb. Symp. Quant. Biol. 22:415-427. Lachenbrush, P. A. 1975. Discriminant analysis. Hafner Press. l28pp. Lack, D. l944.‘ Ecological aspects of species formation in passerine birds. Ibis 86:260-286. . 'l949. The significance of ecological isolation. G. L. Jepsen, E. Mayr, and G. G. Simpson. Genetics, paleontology and evolution. Princeton. pp 299-308. Liscinsky, S. A. l972. The Pennsylvania woodcock management study. Pa. Game Comm. Bull. l7l. 95pp. Marshall, W. H. l958. Woodcock singing grounds at the Cloquet Experi- mental Forest, l947-l956. Trans. N. Amer. Wildl. Conf. pp 296- 304. . 48 49 Maxfield, H. K. l96l. A vegetational analysis of fifty singing grounds in central Massachusetts. M.S. Thesis. Univ. of Mass. 3lpp. Mendall, H. L., and C. M. Aldous. l943. The ecology and management of the American woodcock. Maine Coop. Wildl. Res. Unit, Univ. Maine, Orono. 201pp. Nie, N. H., C. H. Hull, J. G. Jenkins, K. Steinbrenner, and D. H. Brent. l975. SPSS. Statistical package for the social sciences. McGraw- Hill, New York. 675pp. Pianka, E. R. 1974. Evolutionary ecology. Harper and Row, New York. 356pp. Sheldon, W. G. l967. The book of the American woodcock. Univ. of Mass. Press, Amherst. 227pp. Siegel, S. l956. Non-parametric statistics. McGraw-Hill, New York. 326pp. ' Sokal, R. R., and F. J. Rohlf. l969. Biometry. W. H. Freeman and Co., San Francisco. 776pp. ' Veatch,.J. 0., and L. R. Schoenmann. l924. Soil survey of Roscommon County, Michigan. USDA Bur. Chem. and Soils. 27pp. Wenstrom, W. P. l973. Habitat utilization and activities of female American woodcock (Philohela minor Gmelin) in northeastern Minnesota during spring andeummer. Ph.D. Thesis. Univ. Minn., St. Paul. 203pp. Westfall, C. Z. l954. A study of woodcock behavior in their breeding grounds in central Maine. M.S. Thesis. Univ. Maine. ll7pp. Wishart, R. A. l973. The behavior and habitat requirements of the American woodcock in Quebec. M.S. Thesis. McGill Univ., Montreal, Canada. 6lpp. ' "Ill(llI'lllfl'lllllllf