IN FO R M A T IO N TO USERS This material was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the original submitted. The following explanation of techniques is provided to help you understand markings or patterns which may appear on this reproduction. 1. The sign or "target” for pages apparently lacking from the document photographed is "Missing Page(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting thru an image and duplicating adjacent pages to insure you complete continuity. 2. When an image on the film is obliterated with a large round black mark, it is an indication that the photographer suspected that the copy may have moved during exposure and thus cause a blurred image. You will find a good image of the page in the adjacent frame. 3. When a map, drawing or chart, etc., was part of the material being photographed the photographer followed a definite method in "sectioning" the material. It is customary to begin photoing at the upper left hand corner of a large sheet and to continue photoing from left to right in equal sections with a small overlap. If necessary, sectioning is continued again — beginning below the first row and continuing on until complete. 4. The majority of users indicate that the textual content is of greatest value, however, a somewhat higher quality reproduction could be made from "photographs" if essential to the understanding of the dissertation. Silver prints of "photographs" may be ordered at additional charge by writing the Order Department, giving the catalog number, title, author and specific pages you wish reproduced. 5. PLEASE NOTE: Some pages may have indistinct print. Filmed as received. Xerox University Microfilms 300 North Zeeb Road Ann Arbor, Michigan 48106 I I I 75-20,835 FITCH, JohnH., 1944BEHAVIORAL HABITAT SELECTION AS A FACTOR IN THE DISTRIBUTION OF WOODLAND DEERMICE, PERCMYSOJS MANICULATUS GRACILIS, IN THE KINGSTON PLAITS OF NORTHERN MICHIGAN:-Michigan State University, Ph.D., 1975 Zoology Xerox University Microfilms, Ann Arbor, M ichigan 48106 BEHAVIORAL HABITAT SELECTION AS A FACTOR IN THE DISTRIBUTION OF WOODLAND DEERMICE , PEROMYSCUS MANICULATUS GRACILIS , IN THE KINGSTON PLAINS OF NORTHERN MICHIGAN By John H. Fitch. A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Zoology 1975 ABSTRACT BEHAVIORAL HABITAT SELECTION AS A FACTOR IN THE DISTRIBUTION OF WOODLAND DEERMICE, PEROMYSCUS MANICULATUS GRACILIS, IN THE KINGSTON PLAINS OF NORTHERM MICHIGAN By John H. Fitch The purpose of this study was to investigate the relationship between localized variations in habitat occurrence and behavioral habitat selection of the woodland deer mouse, Peromyscus maniculatus gracilis, in the Kingston Plains of northern Michigan. In preliminary small mammal trapping studies, P. m. gracilis were captured in both small isolated woodlots and surrounding open lichen grass habitats within an area of 28.3 square kilometers com prising the Kingston Plains. Burt (1946) reported that adult breeding populations of P. m. gracilis were usually restricted to forest habitats; however, in the Kingston Plains, densities were similar in both forest and open habitats. Harris (1952) reported that P. m. gracilis captured from a site 20 miles south of the Kingston Plains chose laboratory models simulating forest habitats. The first experiment was designed to test the hypothesis that individuals occurring in the vicinity of John H. Fitch forest and open habitat interfaces within the Kingston Plains were partitioned by habitat type. Extensive live- trapping and recapture studies were conducted over a twoyear period in a grid of 3.16 hectares which overlapped two habitat interfaces. The number of individuals moving within habitats differed significantly from the number of individuals moving between habitats in zones of habitat interface, indicating that the majority of P. m. gracilis were partitioned by habitat type. Densities of predators and interspecific competitors were very low in both forest and open habitats within the Kingston Plains. Habitat partitioning might be explained by either of the following: (1) one habitat might be optimal but when densities are high, both optimal and suboptimal habitats might be occupied; or (2) each habitat was optimal to those individuals occurring within it. The second experiment was designed to test the hypothesis that indi­ viduals occurring in the vicinity of habitat interfaces would move to adjacent similiar rather than different habitats in which densities had been artificially reduced. Identical numbers of individuals moved from forest and open habitats to areas of lower density. Of 32 individuals moving to areas of lower density, 26 moved to similiar habitats and only 6 moved to different habitats, indicating that density manipulations could modify home ranges but not patterns of habitat occurrence in a majority of individuals. John H. Fitch On the basis of trapping and recapture data from the grid, three groups of P. m. gracilis could be distin­ guished: those occurring in forest; those occurring in open habitats; and a few individuals which moved between the habitats. The third experiment was designed to test the hypothesis that individuals would select cue sets in the laboratory which modeled those habitats in which they occurred in the study grid. Behavioral habitat selection was evaluated on the basis of seven dependent variables which were measured within pens containing forest and open habitat models. Group means of individuals captured from forest and those captured from open habitats differed signi­ ficantly in the direction hypothesized from one another and from theoretical random means for all seven variables. In contrast, the group means of individuals captured in both habitats did not differ significantly from theoretical random means for any of the seven variables. Thus, patterns of habitat occurrence were directly related to behavioral habitat selection in this experiment. Changes in selective pressures resulting from ecological disturbances in the Kingston Plains evidently brought about two distinct patterns of habitat occurrence and behavioral habitat selection in less than 90 years. The results of this study were discussed in relation to previous studies of behavioral habitat selection and specific recommendations for future research were presented. To my wife, Dennie, and my two daughters, Mary and Elizabeth ii ACKNOWLEDGEMENTS I wish to thank the many people who, in one way or another,, have contributed substantially to this research project. My major professor, Dr. Rollin Baker f has been most helpful in providing research space and equipment, finan­ cial support, advice, and encouragement at crucial times during the study. The other members of my guidance com­ mittee, Dr. James C. Braddock, Dr. John A. King and Dr. Peter G. Murphy, have provided a blend of encouragement, enthusiasm, constructive criticism and guidance without which this study surely would have suffered. Many fellow graduate students also provided encour­ agement, advice and useful criticism so necessary to develop a research project. Jerry Hall, Carl Quertermus and Chris Thoms were especially helpful in this regard, I was especially fortunate to have the aid of three undergraduates, Craig Weisse, Susan Woodley and John Zmiejko, on my research project. The combination of teach­ ing, assistance and learning was most invigorating for all of us. Susan Woodley deserves special thanks for the long hours that she spent assisting in the preparation of this manuscript. General assistance with the statistical design was provided by Maurice Barrone. In addition, the_advice of Dr. Charles Cress in the statistical analyses of Experiments 1 and 2 and the assistance of John Gill in the statistical analysis of Experiment 3 are gratefully acknowledged. This project would not have been possible without financial aid from the following: Michigan State University Graduate School Grant No. 11-5873, Michigan State University Special Grant No. 71-3337 and a grant from the Foundation for Environmental Education. The Department of Zoology, tbe Biological Science Program, and The Museum provided finan­ cial support in the form of teaching and research assistantships which is also gratefully acknowledged. John Ozoga, research biologist with the Michigan Department of Natural Resources, provided much encouragement and advice regarding the field portion of this project. Louis Verme, director of the Michigan Department of Natural Resources' Cusino Wildlife Research Station, kindly made the facilities of the station available to me. Several people accompanied me on trips to the Kingston Plains and were of valuable assistance in field research. Foremost among those were my wife Dennie, James Spalding and Michael and Margaret Jaeger. My sincere thanks go to Gary Connor of the Psychology Research Electronics Laboratory for invaluable assistance in planning and constructing the electrical apparatus for the laboratory portion of the project. Last but hardly least, I am very grateful for my family's continuing patience, assistance and encouragement in this endeavor. TABLE OF CONTENTS Page LIST OF T A B L E S .......................................... viii LIST OF F I G U R E S ........................................... xii INTRODUCTION ............................................. Criteria for Habitat Selection ................... Processes of Habitat Selection ................... Mechanisms of Habitat Selection ................ Modification of Behavioral Habitat Selection . . Critique, Purpose and Experimental Design . . . STUDY A N I M A L ......................... 1 4 5 10 11 13 17 RESEARCH A R E A ............................................ 20 General Description ............................. 20 Ecological Associations ......................... 25 Study G r i d ......................................... 33 EXPERIMENT 1: HABITAT OCCURRENCE ...................... 49 Introduction ...................................... 49 P r o c e d u r e ............................................ 50 R e s u l t s ............................................ 55 EXPERIMENT 2: DENSITY MANIPULATION ................... 58 Introduction ...................................... 58 P r o c e d u r e ............................................ 59 R e s u l t s ............................................ 64 EXPERIMENT 3: HABITAT SELECTION ...................... 67 Introduction ...................................... 67 Habitat Modeling and Preference Measurement . . 68 Test A p p a r a t u s ......................................71 P r o c e d u r e ............................................ 83 R e s u l t s ............................................ 84 Page D I S C U S S I O N ............................................... 98 S U M M A R Y ................................................. 106 LIST OF R E F E R E N C E S ........................................108 vii LIST OF TABLES Table Page 1. Species of mammals occurring in the vicinity of the Kingston P l a i n s ......................... 32 2. Common plant species and species diversity listed by habitat category. Species listed were found in at least one of the 4m2 plots inspected; those marked by an asterisk were dominant in at least half the plots inspected for that specific habitat cate­ gory. Species diversity = number of individuals/number of species/meters2 . . . 42 3. Ground level incident light intensities in study grid on October 10, 1973. Measure­ ments are in foot candleu n i t s ................ 45 4. Chronology of field activities in the Kingston Plains ............................. 47 Summary of trap sites and individual captures in open, transitional, and forest habitats in the study grid for both trapping seasons . 53 Chi square analysis of distribution of individuals captured in forest and open habitats in subgrids for both trapping s e a s o n s ................ 56 5. 6. 7. Chi square analysis of distributions of individuals moving within and between habitats for forest and open quadrants of both s u b g r i d s ............................... 56 8. Chi square analysis of distributions for numbers of individuals moving between habitats and numbers of individuals moving within habitats .................... viii 57 Table 9. 10 . Page Summary of trap stations and numbers of individuals removed and tested in short term and long term phases of the density manipulation experiment ................... . 6 2 Numbers of individuals remaining in test area and numbers of individuals moving to adja­ cent areas of lower density for both test areas in short term removal study . . . . 64 11. Chi square analysis of distributions for numbers of individuals moving to areas of lower density from forest and open test a r e a s ...................................... 65 12. Chi square analysis of distributions for numbers of individuals moving to a similar habitat of lower density and numbers of individuals moving to a different habitat of lower d e n s i t y ................................66 13. Summary of long term removal study indicating numbers of individuals remaining in test areas and those moving to removal areas with different and similar habitats . . . . 66 14. Dependent variables used to measure preference for cue sets modeling forest and open h a b i t a t s ......................................... 71 15. Listing of vegetation comprising models of open habitat and forest habitat in enclosures . . 81 Summary of group means of individuals occurring in forest habitat, open habitat and both habitats for seven dependent variables . 85 Chi square analysis of distributions of individuals captured from forest and open habitats that selected cue sets modeling these habitats ............................. 86 Results of a 1 sample T test of within group habitat preferences for 6 time related variables for animals captured in forest, open and both habitats. All data have been transformed to arcsin values . . . . 88 16. 17. 18. Table Page 19. Analysis of variance table for percent of time spent orienting to the open habitat model prior to release from the runway (runway time) .................................. 91 20. Tukey's test of differences among means for runway t i m e ...................................... 91 21. Analysis of variance table for percent of time spent orienting to the open habitat model subsequent to runway release and prior to habitat choice (initial choice time) . . . 92 22. Tukey’s test for differences among means for initial choice t i m e ............................ 92 23. Analysis of variance table for percent of time spent in the open habitat model compartment during the first night (first night compart­ ment t i m e ) ......................................93 24. Tukey's test for differences among means for first night compartment t i m e ...................93 25. Analysis of variance table for percent of time spent in the open habitat compartment during the second night (second night compartment t i m e ) ............................................ 94 26. Tukey's test for differences among means for second night compartment time ................ 94 27. Analysis of variance table for percent of time spent in the open habitat compartment during the third night (third night compartment t i m e ) ............................................ 95 28. Tukey's test for differences among means for third night compartment t i m e .................. 95 29. Analysis of variance table for average percent of time spent in the open habitat compartment during the three nights tested (average night compartment time) ............................. x 96 Table . 30. 31. Page Tukey's test for differences among means for average night compartment time ............. 96 Summary of results of Tukey's test for differences among means for all 6 time dependent variables ................... 97 xi 8 LIST OF FIGURES Page Figure 1. 2. 3. 4. 5. 6. 7. 8 . 9. 10 . 11. General map of Michigan showing the location of the Kingston Plains and the Cusino Wildlife Refuge ......................... 21 Map of the Kingston Plains showing woodlots, lakes and the location of the study grid 26 Study grid showing open habitat and isolated ................................... woodlot 29 Region of habitat interface between forest and open habitats in the study grid . . . . 29 Map of study grid showing trap sites and ............................. habitat types 35 Open habitat characterized by lichen-grass associations in the study grid. Stumps and partially buried logs served as shelter areas for P. m. gracilis . . . . 37 Large stump in open habitat in Kingston Plains study grid. The humus layer has been removed by successive fires and erosion . . . . 37 Forest habitat characterized by maple-birchbeech and pine associations in study grid . 40 Trap site in forest habitat showing Sherman live trap and partially buried tin can which served as a semipermanent marker . 40 Map of subgrids and quadrants used in the habitat partitioning study (Experiment 1) . 51 Map of removal and test areas for density manipulation study (Experiment 1) . . . 60 xii Page Figure 12. 13. 14. 15. 16. 17. Experimental enclosures for habitat selection ................... studies (Experiment 3) . 72 Floor plan of experimental enclosures indicating the position of center runways and causeways, nestboxes, and food and water supplies ............. . 74 Recording apparatus for the experimental enclosures. All microswitch data is recorded by the Esterline Angus Event Recorder. The clocks measure total time spent in each compartment and the counters record activity wheel revolutions ................................ . 76 Center runway and causeway extending into compartment modeling forest habitat . . 76 Compartment modeling forest habitat. Natural vegetation from isolated woodlots in the Kingston Plains was used in this compartment . 78 Compartment modeling open habitat. Natural vegetation from open lichen-grass associa­ tions in the Kingston Plains was used in this compartment ......................... 78 xiii . INTRODUCTION Animal populations exhibit non~random patterns of spatial distribution, especially in regions of habitat heterogeneity. Spatial distributions of taxa may often be associated with discrete elements of vegetational physiognomy. Klopfer (1969) defines this association as habitat correlation. Three major categories of proximal factors have been hypothesized to account for habitat correlation (Klopfer, 1962). First, physiological and morphological factors may allow taxa to occur in certain types of habitats while precluding their existence in others. Factors in this category vary from a specific range of temperature tolerances (Bursell, 1960) to specific morphological constraints such as beak size or shape in birds which may limit diets to specific foods (Klopfer, 1969). Second, factors associated directly with the habitat type such as predator and/or competitor densities may prevent taxa from utilizing specific habitats well, 1959, 1965; Turner, 1961). Kettlewell (Kettle- (1965) reported that moths of several species were most frequently collected from color matched substrates, not because such sites were actively selected but because moths on other 1 2 backgrounds were more susceptible to predation. Third, habitat partitioning might be due to a preference for specific habitats based upon a behavioral selectivity of environmental cues (Lack, 1937; Wecker, 1963, Klopfer, 1969). Behavioral habitat selection may be operationally defined as a perceptual selectivity of environmental cues resulting in either an attraction for or an avoidance of specific habitat types. The term "selection" is used in this context only to imply a behavioral preference for a given habitat. The purpose of this paper is to review conceptual and experimental bases of behavioral habitat selection and to present a study in which the relationship of habitat correlation and behavioral habitat selection is investigated in a single taxon. Elton (1927) reported that species-specific habitat preferences could influence spatial distribution. He sug­ gested that such preferences could lead to habitat parti­ tioning, population isolation and eventually even to sympatric speciation. Although Elton presented no evidence nor any experimental means of testing his hypotheses (Mayr, 1947), his ideas were seminal in that they provided an initial conceptual base and encouraged further research. Lack (1933) hypothesized that psychological preferences of specific environmental cues could explain patterns of spatial distribution in birds. He had observed that species of birds frequented specific habitats while other habitats, 3 which appeared to be equally suitable, were not utilized. Thorpe (1939) reported that habitat preferences of adult fruit flies (Drosophila melanogaster) could be modified by controlled habitat conditioning of larvae. Thorpe (1945) hypothesized that "habitat imprinting," early experience in specific habitats, could modify subsequent habitat preference and spatial distribution, Harris (1952) and Wecker (1963) investigated habitat preference and imprinting experimentally by allowing indi­ viduals to chose between two available habitats. The results of habitat selection tests involving two subspecies of deer mice, Peromyscus maniculatis bairdii and P. m. gracilis, led Harris (1952) to propose that the perceptual selection of environmental cues was an inherited trait. Wecker (1963) manipulated experiential and genetic compo­ nents of habitat selection in P. m. bairdii and concluded that early experience could reinforce but not change subse­ quent innate patterns of habitat selection. Two levels of evidence are necessary to establish that behavioral habitat selection directly influences spatial distribution. First, characteristic behavioral and distributional patterns must exist in order to claim behavioral habitat selection. Second, the processes and mechanisms governing these patterns should be established. 4 Criteria for Habitat Selection Klopfer (1969) notes that a positive correlation between species occurrence and habitat type is not sufficient evidence to claim a behavioral selection or preference of habitats. Distributional and behavioral evidence necessary to establish that members of a taxon "prefer" a habitat are listed below: (1) Individuals of taxa must consistently occur within specific habitat types. Such patterns of habitat occurrence have been established in a number of species (Brown, Lieberman and Dengler, 1972; Klopfer, 1969; Krebs, 1972; Rosenzweig, and Winakur, 1969). (2) Individuals of taxa must remain within specific habitat types and/or must be absent from other habitat types regardless of changes in such constraints as intra and interspecific competition. These constraints have been experimentally manipulated (Sheppe, 1961; Rosenweig, 1973) but such a pattern has not been established. Studies com­ paring habitat occurrence and selection of specific taxa occurring on islands where interspecific competition is reduced and on mainlands where it is increased indicate that either habitat selection does not limit distribution (Klopfer, 1962) or that habitat selection may undergo adaptive changes (Grant, 1970). (3) Individuals of taxa must be attracted to those habitats in which they occur and/or avoid other habitats. « 5 I This criterion has been tested by presenting individuals with a choice of several artificial habitats simulating natural habitats. Behavioral preferences in choice situa­ tions have been established in a number of species includ­ ing insects amphibians (Russell and Rao, 1942) fish (Norris, 1963), (Wiens, 1970), reptiles (Sexton, 1958), birds (Klopfer and Hailman, 1965), and mammals (Harris, 1952; Wecker, 1963). These three criteria have not yet been firmly established for any single taxon. Until such criteria can be established, it is premature to claim that behavioral habitat selection is sufficient to modify habitat occurrence. Processes of Habitat Selection The processes governing behavioral patterns of habitat selection may be divided into the following cate­ gories: (1) capabilities to perceive environmental cues and habitats, (2) perceptual selectivity environmental cues and, (preference) of (3) factors determining or influ­ encing perceptual selectivity. Von Uexkull (1921) suggested that only those elements of an environment which are actually perceived (merkwelt) are of any relevance. Because each species differs in per­ ceptual capabilities, merkwelts should be species specific. Perception of environmental cues has been investigated in two different ways. First, sensory modalities have been directly measured by recording sensory output (Granit, 1955; Maturana et al., 1960). Second, movement and form 6 discrimination have been studied (Rahman, Rahman and King, 1968; Sutherland, 1960). Although such analyses have not been related to specific aspects of environmental cue dis­ crimination (Klopfer and Hailman, 1965) , sensory modalities most important in cue detection have been established. Vision is considered to be the most important sensory modality in the orientation of terrestrial mammals except echolocators (Munn, 1950) and in birds (Klopfer, 19 69). However, olfaction has been implicated in orientation to environmental cues in insects (Schoonhovern, 1968) and fish (Hasler, 1966). Both approaches can investigate perceptual abilities; but neither approach can establish selective perception of stimuli. Perceptual selectivity has been studied by creating models consisting of specific cues or groups of cues which evoke specific behavioral responses; overt preferences for such cues may then be evaluated. Experimental approaches range from orientation studies in which artificial cue sets are manipulated under controlled laboratory conditions (Joslin, 1971) to studies which test preference to natural or simulated natural cue sets in habitat enclosures (Harris, 1952; Wecker, 1963; Klopfer and Hailman, 1956). Such studies have been productive because specific sensory modalities, preferred cue sets and habitat distribution may be related using this approach (Hagen, 1967); and the determinants of perceptual selectivity may be investigated under different sets of environmental stimuli. The most productive approach has been to analyze overt preferences for cue sets simulating natural habitats under controlled conditions. A major difficulty, however, ‘ has been the development of reliable dependent variables which accurately measure overt preferences. The use of artificial cue sets that are not within the range of natural cues encountered may evoke supernormal responses. Although supernormal stimuli are of general interest, they are of little value in establishing or predicting perceptual selectivity of natural environmental cue sets. Most experimental cue sets have involved visual stimuli. Factors determining or influencing perceptual selectivity include genetic and experiential variables, developmental state and, physiological state. The relative importance of genetic and experiential factors has been evaluated by comparing overt preferences for cue sets among groups of individuals given early experience with different specific cue sets. Genetic influences are inferred if overt preferences for specific cue sets remain constant regardless of experiential manipulation. Genetic influ­ ences have not yet been specifically defined although it would be possible and highly desirable to do so. Harris (1952) and Hagen (1967) interbred taxa of mammals and fish respectively in which genetic components of habitat selec­ tion were thought to differ. In both studies, individuals of the hybrid filial generation were intermediate to parental generations in the overt habitat preferences they exhibited. The following general conclusions have been 8 established by studies involving genetic and experiential influences: (1) Early experience with appropriate cue sets simulating natural habitats can reinforce and sometimes modify subsequent selection of environmental cues (Quertemus, 1972 with fish; Wiens, 1970 with amphibians; Sargent, 1956 with birds; and Grant, 1970 with mammals). However, genetic influences are sometimes more important than experiential influences (O’Hara, 1974 with amphibians; Klopfer, 1965 with birds; and Wecker, 19 63 with mammals). (2) Wecker (1963) reported changes in the motiva­ tional state of perceptual selectivity in P. m. bairdii that had been under different selection pressures (laboratory conditions) for 12 to 20 generations. Careful breeding experiments are needed to substantiate these results because relatively inbred strains of subjects have been utilized in many experimental studies of habitat selection. (3) There is still insufficient evidence to accept or reject Thorpe's (1945) hypothesis of "habitat imprinting." Although Hasler (1966) has related spawning site selection to olfactory imprinting in Pacific salmonids in fresh water 9 streams/ it is difficult to separate genetic and experiential influences- Wiens (1970) reported habitat imprinting in frogs. However, critical periods which are usually associated with imprinting have not been established by either Hasler or Wiens. (4) Genetic and/or experiential changes in percep­ tual selectivity may be related to specific developmental and/or physiological states. Taylor (1972) and O'Hara (1974) have provided evidence that perceptual selectivity of environ­ mental cues changes during metamorphosis in amphibians. Evidence for changes in perceptual selectivity in relation to physiological state, in this case reproduction, has been presented by Thoms (1974) for fish and by Hilden (1965) and Sargent (1965) for birds. Additional research is needed to investigate other changes in perceptual selectivity corresponding to physiological state. In general, studies of determinants demonstrate that perceptual selectivity is based upon a complex interaction of physiological and developmental factors operating within genetic and experiential frameworks. Several types of mechanisms have been proposed to relate these determinants. 10 Mechanisms of Habitat Selection Several hypotheses regarding the underlying mechanisms of behavioral habitat selection have been proposed. Thorpe (1963) proposed a goal directed innate mechanism based upon an "internal visualization" of an ideal stimulus situation to which all others could be compared. Such a stimulus situation could be defined by a complex of configurational stimuli that has yet to be experimentally established. Complexes of stimuli have been related to the formation of search images (Tinbergen, 1960) and to imprinting (Klopfer and Hailman, 1965) . However, both search image and imprinting involve experiential factors which interact with learning, and learning seems of minor importance in the habitat selection of most species. Sale (1969) suggested a non-goal directed, negative feedback mechanism in which appetitive exploration continues until an appropriate consummatory stimulus situation is attained. When the correct consummatory stimulus situation is attained, drive and, therefore, appetitive exploration is reduced. According to his hypothesis, the amount of explora­ tory or investigative behavior should be greater in non­ selected habitats, and other types of behavior such as feeding should be greater in selected habitats. obtained by Wecker Results (1963) and by Sale (1969) support this hypothesis; however, other studies such as Klopfer (1965) and Quertemus (1972) do not. Sale's model provides no new 11 conceptual frame upon which investigations of the nature and components of the selection mechanism may be based. Does behavioral habitat selection involve both the attraction to and the avoidance of specific stimulus situations? Sale's model would predict that attraction is most important because appetitive exploration simply con­ tinues until the appropriate habitat is encountered. However, there is evidence that some stimulus situations are actively avoided (O'Hara, 1974). Mechanisms of habitat selection are poorly understood, perhaps in part because the range of variation in behavioral habitat selection within a single taxon has not been adequately determined. Modification of Behavioral Habitat Selection In what way may behavioral patterns of habitat selection be modified? Overt preferences for specific cue sets have been modified experimentally in two ways. First, preferences have been changed in some species by manipulating experience with specific cue sets (Wecker, 1963). Second, preferences have been modified by genetic manipulation, either by interbreeding individuals of two taxa which differ in habitat selection (Harris, 1952) or by drastically alter­ ing selective pressures acting upon a taxon through many generations (Wecker, 1963). How may behavioral habitat preferences be altered under natural conditions? Spontaneous changes in either 12 experiential factors or in gene frequency as a result of hybridization of two taxa rarely occur under natural conditions. However, selective pressures due to such variables as predation and interspecific competition may vary considerably through the ranges of taxa. Few taxa exhibit uniform habitat occurrence throughout their ranges. MacArthur (1965) reported that increased interspecific competition led to more specific habitat partitioning in birds. He also noted that decreased competition on islands can lead to greater ubiquity in habitat occurrence. and Phillips Ozoga (1964) and Grant (1970) have noted similar patterns in mammals. As was noted previously, such patterns do not necessarily indicate changes in habitat selection. Kluvyer and Tinbergen (1960) contend that species have an optimum habitat which will only support a given density; once such density levels are reached, less optimal habitats will be utilized. These habitats may not be preferred; possibly they are simply not avoided. One implication of such an interpretation is that individuals will move to more optimal habitats when permitted to do so. An experi­ mental means of testing this hypothesis is to artificially reduce either inter or intraspecific densities in adjacent habitats and observe resulting changes in habitat occurrence. Sheppe (1961) removed an isolated population of Peromyscus maniculatus oreas from a large ravine and noted that P. m. artemisiae from adjacent hillsides moved into that habitat. Ozoga and Phillips (1964) observed that P. m, gracilis was 13 more ubiquitous in habitat occurrence when populations were high; when populations were low, only forest habitats ordinarily occupied by the taxon were utilized. Few studies relate variations in habitat occurrence with behavioral habitat selection in single species dis­ tributions. Miller (1973) reported that prairie deer mice (Peromyscus maniculatus bairdii) from South Dakota selected a simulated forest habitat corresponding to their natural habitat; in Michigan, where P. m. bairdii occur only in grassland habitats, individuals selected a simulated grass­ land habitat. However, no studies have yet related varia­ tions in local habitat distribution with behavioral patterns of habitat selection. Critique, Purpose and Experimental Design There are several important reasons for the lack of conceptual unity between the causes and effects of behavioral habitat selection. As Pittendrigh (1958) suggested, causal relationships are virtually impossible to establish because of the difficulty in separating behavioral "selection of" and "adaptation to" specific habitats. Behavioral patterns of what is termed "habitat selection" are so variable in their causes and effects that an inclusion under any general heading would perhaps be a misnomer. In any case, the specific behavioral patterns under study should be carefully defined. Another difficulty, noted by Krebs (1972), is that the significance of behavioral habitat selection in 14 animal distribution has yet to be demonstrated. One reason may be that variations in patterns of habitat dis­ tribution within taxa have not yet been well documented (Rosenzweig and Winakur, 1969). In addition, the three criteria necessary to demonstrate behavioral habitat selection have not been established for any single taxon. Therefore, studies should not be based on the priori assumption tuat behavioral habitat selection is a signifi­ cant factor in the spatial distribution of animal taxa. In the present study, the variations in habitat occurrence within a single taxon is used as the independent variable to evaluate the following questions: (1) to what extent can variations in local habitat use be related to differences in behavioral habitat selection, and (2) can behavioral and distributional evidence necessary to claim behavioral habitat selection be documented in a single taxon? The experimental design formulated to answer these questions is based upon the following: (1) Deer mice of the subspecies Peromyscus maniculatus gracilis generally occupy forested habitats throughout their range (Burt, 1946) . (2) Harris (19 52) established that both wild caught P. m. gracilis and laboratory reared individuals from the Cusino Wildlife Refuge in Michigan's Upper Peninsula selected 15 artificial habitats representing forest over those which modeled open habitats. (1) In the Kingston Plains, the writer discovered a population of P. m. gracilis occurring in a disturbed, open habitat characterized by grass and lichen associations, only 22 miles from the Cusino Wildlife Refuge. In pilot studies, P. m. gracilis from the Kingston Plains selected cue sets modeling open habitats; P. m. gracilis from Cusino, tested as a control, selected cue sets modeling forest habitats. (2) Preliminary trapping studies indicated that P. m. gracilis also occupied small isolated forest groves within the matrix of open habitats in the Kingston Plains. On the basis of observed differences in the percep­ tual selectivity of environmental cues existing between P. m. gracilis inhabiting forest habitats at Cusino and those inhabiting open habitats in the Kingston Plains, the following experimental hypotheses were developed regarding variations in habitat occurrence in the Kingston Plains: (1) Individuals are partitioned by habitat type along interfaces between forest and open habitats in the Kingston Plains. 16 (1) Individuals will continue to be partitioned by habitats even if densities are artifi­ cially reduced in the adjacent habitat. (2) Individuals will select test habitats simulating those in which they were partitioned. These specific hypotheses will be tested as Experiments 1, 2 and 3 respectively in this paper. Experimental procedures and results are described separately for Experiments 1, 2 and 3. Tests of these hypotheses will establish relation­ ships between habitat occurrence and behavioral habitat selection in a unique ecological situation in which a greater range in habitat occurrence of the population may or may not lead to increased plasticity in individual habitat preferences. STUDY ANIMAL The woodland deer mouse, Peromyscus maniculatus gracilis, was selected as the experimental species in this study for the following reasons: handled and marked in the field; (1) it is easily trapped, (2) it can be kept in captivity without great difficulty; (3) previous studies have already established that this taxon would respond to artificial cue sets simulating its natural habitat; and (4) this taxon shows some variation in local habitat occurrence. This subspecies of cricetid rodent is widely dis­ tributed throughout northeastern United States and south­ eastern Canada, occurring from Pennsylvania north to southern Quebec and from western Manitoba east to Maryland (Hall and Kelson, 1959). In Michigan, P, m. gracilis occurs throughout the Upper Peninsula, on many islands in Lake Michigan and as far south as Missaukee County in the Lower Peninsula (Burt, 1946). This subspecies occupies most forested habitats within its range but occurs in greatest abundance in upland deciduous forests of maple, birch and beech (Dice, 1925). 17 18 Klein (1960) reported' that P. m. gracilis in New York were restricted to woodland habitats even in ecotonal areas where forest and grassland habitats were adjacent. However, there are several exceptions to the pattern of forest habitat occurrence in P. m. gracilis; in each exception, pressures of predation and/or interspecific competition are reduced. A population of P. m. gracilis on Beaver Island in Lake Michigan apparently utilizes a variety of habitats including open grassland when densities are high (Ozoga and Phillips, 1964). Hatt et a l . (1948) reported similar variation in habitat occurrence on South Manitou Island, also in Lake Michigan. The writer discovered a population of P. m. gracilis occupying an open, disturbed habitat characterized by lichen-grass associations and charred stumps in the Kingston Plains of Michigan's Upper Peninsula (Figure 6). Small, isolated, forest groves in the Kingston Plains were also occupied by gracilis Harris (Figure 3). (1952) investigated habitat selection in this subspecies by allowing individuals from forest habitats in the Cusino Wildlife Refuge to chose between artificial cue sets simulating forest and grassland habitats respec­ tively. He discovered that both wild caught adults and laboratory bred and reared young selected artificial habitats simulating their natural forest habitat. In a pilot study, the writer investigated habitat selection in the population of P. m. gracilis from the open Kingston Plains habitat. Individuals were allowed to chose between 19 habitats simulating forest and open habitats; P. m. gracilis from Cusino were used as controls. Individuals from the Kingston Plains selected the open habitat pre­ dominately while those from Cusino selected the forest habitat. In the present study, individuals of this sub­ species were utilized in trap and recapture studies in the field as well as in experimental studies on habitat selec­ tion under controlled laboratory conditions. Individuals were handled with 12 inch forceps equipped with rubber tips and were marked by means of a standard toe-clipping system. Those individuals used in habitat selection tests were housed in the Michigan State University Museum laboratories in clear plastic cages measuring 12.7 cm x 27.9 cm x 15.2 cm with removable wire lids. Cage bottoms were covered with wood shavings and cotton was provided as nesting material. Food (Purina Mouse Breeder's Chow) and water were supplied ad libitum. RESEARCH AREA The Kingston Plains was chosen as the research area for this study on the basis of the following initial observa­ tions: (1) high population densities of P. m. gracilis were found in open, atypical habitats for this taxon, individuals captured in these open habitats selected laboratory cue sets simulating open habitats in preference to cue sets simulating forest habitats in pilot experiments; (2) P^. m. gracilis were also found in abundance in isolated forest groves in the Kingston Plains; (3) densities of potential predators of P. m. gracilis were low in the Kingston Plains; and (4) potential interspecific competitors of gracilis were reduced in numbers and species in both open and forest habitats in the Kingston Plains. General Description The Kingston Plains is situated in the northeast corner of Alger County (T48, R15W) approximately 7 miles south of Lake Superior in the Upper Peninsula of Michigan (Figure 1). The Plains consists of approximately 10.9 square miles (28.3 square kilometers) of upland lichen grass Figure 1 — General map of Michigan showing the location of the Kingston Plains and the Cusino Wildlife Refuge. 22 fete Royal® LAKE SUPERIOR i/% g e r County LOCATION OF RESEARCH AREAS M ICHIGAN I * Kingston Rains 2- Cusino Wildlife Refuge Figure 1 23 associations, isolated forest groves and small lakes. This area, like most of the eastern Upper Peninsula, is underlain by Paleozoic sediments. With the retreat of the Wisconsin Glacier, most of the eastern Upper Peninsula was inundated by Lake Algonquin and old glacial moraines were reduced by lake wave action. As the Great Lakes receded to current levels, drifting and disturbed drainage patterns created extensive swamp and marsh areas (Kelley and Farrand, 1967). The histc”*” -'■on Plains is characterized by a series of ec .resulting from exten- sive logging and operations, the J forest consistinc .nsively logged (Jenkins, 1942). i occurred and logging from 1888 to 1890, operations were cone salvaged. ially burned lumber was Although most of the slash material was consumed by the fire, many snags, stumps, and a few young pines were not destroyed (Figures 6 and 7). Jenkins (1942) suggested that at least one major fire occurred between 1890 and 1910 and the eastern portion of the Plains was burned again in August of 19 36. No major fires have been reported subsequent to this date. The Kingston Plains is currently under the jurisdiction of the Michigan Department of Natural Resources. Between 1951 and 1957, the Department attempted to improve habitats in the Plains in several ways. The eastern third of the Plains was allotted to :n habitat improvement. A total were planted in this area he two-»; woody cover (Amman, 195 7}. ;r (La thy r us sylvestris) were pi,mu several pothole lakes and a loo plantings to provide food and ;• 19 57). The central third o f * :* control area and the west-. < Forestry Division. By .195 d pines had been planted m th:a have been successful in some : m- have failed; survival of tie: tpoor (Van Etten, 19 57). since 1957 do : o (Switzenburq, port. Repeated fires have :o;u. exposing a layer of rub icon sr. . lain by brown sand extending • (Veatch, Schoenmann and Lesh, „.3 acidic, ranging in pH from 3.of soil samples taken by the wr ito; The climate of the K m vs Lake Superior. The Plains is ;o tures and deep snow character :z : Lake Superior. The mean annual fluctuations between a mean maxi minimum of 32.2. Annual percipi 23 associations, isolated forest groves and small lakes. This area, like most of the eastern Upper Peninsula, is underlain by Paleozoic sediments. With the retreat of the Wisconsin Glacier, most of the eastern Upper Peninsula was inundated by Lake Algonquin and old glacial moraines were reduced by lake wave action. As the Great Lakes receded to current levels, drifting and disturbed drainage patterns created extensive swamp and marsh areas (Kelley and Farrand, 1967). The history of the Kingston Plains is characterized by a series of ecological disturbances resulting from exten­ sive logging and several major fires. Prior to logging operations, the Plains was an extensive white and red pine forest consisting of trees between 300 and 400 years old (Jenkins, 1942). The Plains area was extensively logged from 1888 to 1890; in 1890, a major fire occurred and logging operations were concluded after partially burned lumber was salvaged. Although most of the slash material was consumed by the fire, many snags, stumps, and a few young pines were not destroyed (Figures 6 and 7). Jenkins (1942) suggested that at least one major fire occurred between 1890 and 1910 and the eastern portion of the Plains was burned again in August of 1936. No major fires have been reported subsequent to this date. The Kingston Plains is currently under the jurisdic­ tion of the Michigan Department of Natural Resources. Between 1951 and 1957, the Department attempted to improve habitats in the Plains in several ways. The eastern third 24 of the Plains was allotted to the Game Division for wildlife habitat improvement. A total of 151,600 red and jack pines were planted in this area between 1951 and 1956 to improve woody cover (Amman, 1957). In July, 1956, flat peas (Lathyrus sylvestris) were planted along the shores of several pothole lakes and along the edges of several pine plantings to provide food and cover for wildlife (Van Etten, 1957). The central third of the Plains was designated as a control area and the western third was assigned to the Forestry Division. By 1953, a total of 418,500 red and jack pines had been planted in this area. The pine plantings have been successful in some areas but a large number of them have failed; survival of the pea plantings was reported to be poor (Van Etten, 19 57). since 1957 No plantings have been attempted (Switzenburg, pers. comm.). Repeated fires have destroyed the organic soil layer exposing a layer of rubicon sand, 4 to 6 inches deep, under­ lain by brown sand extending to a depth of 18 to 24 inches (Veatch, Schoenmann and Lesh, 19 29). The rubicon soil is acidic, ranging in pH from 3.66 to 5.70, according to pH soil samples taken by the writer in June, 1971. The climate of the Kingston Plains is influenced by Lake Superior. The Plains is within the zone of low tempera­ tures and deep snow characterizing the southern shoreline of Lake Superior. The mean annual temperature is 40.2°F with fluctuations between a mean maximum of 48.0 and a mean minimum of 32.2. Annual percipitation averages 32.15 inches 25 with a total annual snowfall which is usually in excess of 100 inches. The annual growing season, the period between killing frosts, is 155 days. Ecological Associations The Kingston Plains has over forty small pothole lakes which may disappear during dry summers as well as eight, large, permanent lakes (Figure 2). Vegetation along the shoreline of the pothole lakes is characterized by species such as: blueberry leatherleaf (Chamaedaphne calyculata) , (Vaccinium sp.) and red sorrel (Rumex acetosella) . In low, poorly drained areas in the vicinity of larger lakes, associations of tamarack (Larix laricina) and balsam fir (Abies balsamea) are common. Much of the ten square mile area constituting the Kingston Plains consists of well drained barren expanses characterized by old charred snags and tree stumps and occasional sand blow-outs. areas includes: Dominant vegetation in these hair grass (Deschampsia flexuosa) , lichens (Cladonia ap. and Cladina sp.), bracken ferns aquilinum), blueberries (Rumex acetosella) . (Pteridium (Vaccinium sp.) and red sorrel Open groves of yellow birch (Betula lutea) in association with bracken ferns are interspersed throughout the open upland areas. Although birch has not been cirtificially planted in the Kingston Plains, both white and yellow birch seem to do well in such early successional stages. Based upon a sample of 15 tree borings made in Figure 2.— Map of the Kingston Plains showing woodlots, lakes and the location of the study grid. . 27 Roads f Grid Woodlot Trails 08P Lakes Natural stam — Edge of opening 0 I kilometer KINGSTON PLAINS Part of T48N.RI5W Figure 2 Alger County, Mich. 28 June, 1972, birch in these groves are between 30 and 40 years old. However, litter layers and forest cover asso­ ciations are not well developed in these birch groves. In several isolated upland areas in the Kingston Plains, fire destruction was apparently not extensive and small woodlots were able to develop (Figure 3). These woodlots are characterized by conifer associations of white pine (Pinus strobus), balsam fir (Abies balsamea) and hemlock (Tsuga canadensis) and deciduous associations of red maple (Acer rubrum), American beech (Fagus grandifOlia) and yellow birch (Betula lutea). A definite litter layer is present in these woodlots; forest ground cover mainly dominated by red maple seedlings is also present. White pines were the oldest tree species in these woodlots. The ages of ten large white pines in the study grid were deter­ mined on the basis of tree ring counts. The mean age was 74.8 years; the maximum age, 130 years, predated the first fire and logging operations. Red maples and yellow birches sampled in the same area were between 40 and 50 years old. Most habitats in the Kingston Plains were occupied by Peromyscus maniculatus gracilis. In upland open and forest habitats, maximum densities were 3 to 4 times higher than the maximum densities reported by Blair (1941) for a deciduous forest in the Cusino Wildlife Research Area, some twenty miles away. There are several probable reasons for high densities and relative habitat ubiquity of this taxon in the Kingston Plains. Potential shelter of stumps 29 Figure 3.— Study grid showing open habitat and Isolated woodlot. Figure 4.— Region of habitat interface between forest and open habitats in the study grid. Note the sharp line of demarcation between the two habitats. t 30 jMY*» Figure 4 31 ' ' (Figure 7), snags and logs are numerous in both, open and wooded areas, and densities of potential predators are low. Large hawks and owls rarely occur in the Kingston Plains; only two red-tailed hawks (Buteo jamaicehsis) and one marsh hawk (Circus cyaneus) were sighted during the entire study and they were observed during the fall migratory period. Great horned owls (Bubo virginianus) were heard on several occasions along the periphery of the Plains but none were sighted in the Plains nor were any pellets located. hawks Sparrow (Falco sparverius) were abundant from June to late August or early September in the Plains. They nested in large snags and subsisted upon grasshoppers predominately according to an analysis of regurgitated pellets. Mammalian predators capable of subsisting upon P. m. gracilis were also scarce. Mammals occurring in the Kingston Plains and observed along the Plains periphery are listed in Table 1. Only one red fox (Vulpes fulva) was sighted in the Plains during the study and tracks were observed rarely. Coyotes (Canis latrans) were never sighted in the Plains nor were any scats or tracks observed; coyotes were heard calling from adjacent wooded areas on four occa­ sions. Both species have been bountied in Alger County for some time; Ozoga (pers. comm.) reported that several trappers have been active in the vicinity of the Kingston Plains. Black bears (Ursus americanus) sometimes venture into the Plains; on two occasions, bears were observed as they tore apart old stumps and logs. However, judging from the small TABLE 1.— Species of mammals occurring in the vicinity of the Kingston Plains. Status Abundance Masked Shrew (Sorex cinereus) resident rare Short-tailed Shrew (Blarina brevicauda) resident low Black Bear (Ursus americanus) transient medium Long-tailed Weasel (Mustela frenata) transient rare River Otter resident low Red Fox (Vulpes fulva) transient low Coyote (Canis latrans) adjacent areas not recorded Woodchuck (Marmota monax) resident common Eastern Chipmunk (Tamias striatus) resident common Red Squirrel resident medium Northern Flyinq Squirrel (Glaucomys sabrinus) resident low Beaver (Castor canadensis) resident medium Woodland Deer Mouse (Peromyscus maniculatus gracilis) resident common Red-backed Vole (Clethrionomys gapperi) resident rare Meadow Vole (Microtus pennsylvanicus) transient rare Porcupine (Erethizon dorsatus) adjacent areas not related White-tailed Deer (Odocoileus virginianus) transient low Species (Lutra canadensis) (Tamiasciurus hudsonicus) 33 percentage of dismantled stumps and logs, the effect of this predator upon gracilis populations is probably slight. The long-tailed weasel (Mustela frenata), another potential predator of P. m. gracilis, is extremely rare; only one individual was captured during the study. Potential interspecific competitors such as the red-back vole (Clethrionomvs gapperi) and the meadow vole (Microtus pennsylvanicus) may restrict the distribution of P. m. gracilis (Manville, 1949; Ozoga and Verme, 1968) . Only seven C. gapperi were captured during the study in the Kingston Plains; this species was commonly found in moist, lowland forests rather than upland habitats (Burt, 1946) . The open habitats of the Kingston Plains should support populations of Microtus; however, only one Microtus was captured during the study. Perhaps they have not been able to penetrate a ten mile forest barrier between the Kingston Plains and the nearest grassland habitat where they occur in large numbers since the last major fire. Study Grid In preliminary studies, a significant number of P. m. gracilis that were captured in large expanses of uninterrupted open habitat in the Kingston Plains selected cue sets simulating open habitats in preference to cue sets simulating forest habitats. In contrast, a significant number of P. m. gracilis from uninterrupted forest habitat 34 in the Cusino Wildlife Refuge preferred those cue sets simulating forest habitats. In subsequent trapping surveys in the Kingston Plains, P. m. gracilis were captured in small isolated woodlots as well as in adjacent open habitats. Are individuals occurring in interface areas between these two habitats partitioned by habitat type or are they able to utilize both habitats? In order to evaluate movements of individual I?. m. gracilis between and within habitat interfaces, a grid of 198 Sherman live traps was established in June, 1972. were spaced at intervals of 14.1 meters Traps (45 feet) on the basis of previous trapping studies on P. m. gracilis (Blair, 1941, 1942). The grid included twenty-two rows covering 295.3 meters (945 feet) of nine traps each covering 112.5 meters (360 feet) with a total area of 3.16 hectares acres). (7.8 The grid was positioned so that an isolated woodlot and adjacent open habitat on opposite sides of the woodlot could be sampled (Figure 5). Trap sites in the study grid were placed in one of three general habitat categories: and forest habitats. open field, transition, These a priori categories were opera­ tionally defined as follows: (1) open habitat included grass-lichen-fern associations in which the trap site was more than 7 meters habitat (Figure 6); (22.5 feet) from the nearest forest (2) transition included areas of interface between open and forest habitats in which the 35 Figure 5.— Map of study grid showing trap sites and habitat types. KINGSTON PLAINS STUDY GRID A B C 123 4 567 89 - D E F G H i r 1 1 i 1 | i 1 | i 1 | i 2 i i 2 | i 1 | i 2 | 2 1 | 1 2 i _L. A B C 1 1 2 1 2 2 2 2 2 1 1 1 nr 3 2 2 3 2 3 3 3 3 3 3 3 3 3 2 3 3 3 1 I J K L M N O 1 1 "r 1 I i I 3 3 3 3 3 1 1 I 3 3 3 3 3 1 1 2 3 3 3 3 3 2 2 2 3 3 3 3 3 2 2 3 3 3 3 3 2 2 3 3 3 3 3 2 2 3 3 3 3 3 2 1 3 3 3 3 3 3 2 3 3 3 3 3 3 2 _JL.,Jl _JL l -JL,_J_I D E F G H I J K L M N O Distance between traps - 13.72 meters I- open 2- transition Figure 5 P Q R S T U V 1 .i.i. T— 1— ,— 2 1 1 1 1 1 1 1 1 III 1 1 1 1 1 I 1 1 1 III 1 1 1 1 1 1 1 1 1 I 1 1 2 1 1 | | | 1 1 1 1 1 1 1 1 1 1 1 1 i i i .L.„ L..I |— 1—1 |—2 1-3 | —4 1 —5 1 -6 | -7 1 —8 1 -9 I. P Q R S T U V S CALE: I m m . - 2m. 3- forest u * Figure 6.— Open habitat characterized by lichen^grass associations in the study grid. Stumps and partially buried logs served as shelter areas for P. m. gracilis. Figure 7.— Large stump in open habitat in Kingston Plains study grid. The humus layer has been removed by successive fires and erosion. Figure 6 Figure 7 39 trap site was less than 7 meters (22,5 feet) from both habitats (Figure 4); and (3) forest habitat included associations of maple, birch, beech and white pine with intact litter layers and forest ground cover in which the trap site was more than 7 meters (22.5 feet) from the nearest open habitat (Figures 8 and 9). Ten vegetational plots, each 4 meters square, were sampled in order to obtain relative measures of plant density and species diversity in each of these three habitat categories. Sampling was conducted in fall 1972. Common plant species and species diversity are listed in Table 2. Open habitats were dominated by hair grass (Deschampsia flexuosa), lichen (Ciadina rangiferina, Cladonia mitis and Ciadina alpestris) and bracken ferns (Pteridium aquilinum) . Blueberry (Vaccinium sp.) was common in some of the plots, especially in the vicinity of stumps and logs. Forest plots were dominated most commonly by red maple seedlings (Acer rubrum) although white pine (Pinus strobus) was dominant in one plot. Woodlots such as the one in which the study grid is located should eventually attain a maple-birch association climax. Where red maple seedlings are not dominant, club moss (Lycopodium sp.) and hair cap moss (Polytrichum commune) are common on the forest floor. All of the plant species listed in Table 2 for open and forest habitats were found in at least one of the transla­ tional plots in areas of habitat interface. Two species 40 Figure 8.— Forest habitat characterized by maple-birch-beech and pine associations in study grid. Figure 9.— Trap site in forest habitat showing Sherman live trap and partially buried tin can which served as a semipermanent marker. I 41 Figure 8 ass S S IifB it j h Figure 9 TABLE 2.— Common plant species and species diversity listed by habitat category. Species listed were found in at least one of the 4m2 plots inspected; those marked by an asterisk were dominant in at least half the plots inspected for that specific habitat category. Species diversity = number of individuals/number of species/meters2. Habitat Category Open Plant Max. 1.25 Species Min. .50 Diversity Average .932 Common Plant Species *Deschampsia flexuosa, hair grass Festuca rubra, grass Epigaea repens, trailing arbutus Rumex acetocella, red sorrel Hieracium aurantiacum, hawk weed Viola s p ., violet Rosa sp ., wild rose Gaultheria procumbers, winter green Rubus setosus, blackberry Vaccinium angustifolium, blueberry Vaccinium myrtilloides, blueberry *Pteridium aquilinum, bracken fern *Cladonia mitis , grey lichen 42 Taraxacum officinale, dandelion TABLE 2.— continued. Habitat Category Plant Max. Species Min. Diversity Average Common Plant Species *Cladina rangiferina, lichen *Cladina alpestris, lichen Forest 3.00 1.75 2.363 *Acer rubrum, red maple *Fagus qrandifolia, American beech *Betula lutea, yellow birch *Pinus strobus, white pine Tsuga canadensis, eastern hemlock Abies balsamea, balsam fir Cornus canadensis, bunchberry Coptis trifolia, goldthread Carex sp., sedge Lycopodium selago var . patens, club moss Polytrichum commune , hair cup moss Cladonia coniocraea , lichen Cladonia deformis , lichen *Cladina rangiferina , lichen TABLE 2.— continued. Habitat Category Transition Plant Max. 2.00 Species Min. 1.25 Diversity Average 1.600 Common Plant Species *Deschampsia flexuosa, hair grass Vaccinium angustifolium, blueberry ♦Acer rubrum, red maple ♦Betula lutea, yellow birch ♦Pinus strobus, white pine Lycopodium selago var . patens, club moss ♦Pteridium aquilinura, bracken fern ♦Cladonia mitis , grey lichen ♦Ciadina rangiferina, lichen ♦Ciadina alpestris, lichen 45 of grasses listed for open habitats, grey lichen, bracken ferns and red maples were most common in these plots. Habitat categories could be distinguished not only on the basis of plant association but also on the basis of plant species diversity indices. Maximum and minimum species diversity indices did not overlap for open and forest habitats. Another means of evaluating open, transitional and forest habitats is by measuring the relative amounts of incident light at ground level at trap sites in each cate­ gory. Incident light values were obtained for 75 trap sites, 25 in each habitat category, on October 10, 1973. Measurements were taken from 1 pm to 3 pm when the sky was slightly overcast. Maximum light levels did not change during this two-hour period. Incident light was measured by means of Weston light meter. in Table 3. The results are presented Maximum and minimum incident light levels for open and forest habitats do not overlap. TABLE 3.— Ground level incident light intensities in study grid on October 10, 1973. Measurements are in foot candle units. Incident Light Habitat Category Number of Trap Sites Minimum Maximum Average Open 25 3200 4000 3580.9 Transition 25 900 3200 1944.2 Forest 25 230 1800 851.2 46 In summary, habitat categories established for trap sites in the study grid may be distinguished on the basis of plant associations, plant species diversity and inci­ dent light levels. Nineteen visits were made to the Kingston Plains from June 1971 to October 1973 (Table 4). Habitat utiliza­ tion was evaluated on the basis of seventeen trapping periods in the study grid. Movements data for Experiment 1 were obtained from trapping periods 1 through 6 for season 1 and periods 10 through 19 for season 2. Removal of individuals for Experiment 2 was accomplished during the latter part of trapping period 6. Subjects for Experiment 3 were captured during trapping periods 6 and 19. Sherman live traps measuring 75 x 75 x 227 mm were used at all trap sites in the study grid. Traps were baited with a mixture of peanut butter and rolled oats. Trapping periods varied in duration from two nights to six nights depending upon time constraints and weather conditions. Traps were usually checked twice a day, in the morning and in late afternoon. the night. All but two mice were trapped during All captured mice were marked by toe-clipping, and then measured, weighed, aged, sexed and released at the point of capture. TABLE 4.— Chronology of field activities in the Kingston Plains. Trip Number Trapping Periods Number of Days Between Visits Activities 1 25-26 June 71 — Survey of vegetation 2 18-21 Sept 71 84 Small mammal habitat survey 3 27-29 May 71 4 18-20 June 72 20 2nd grid trapping period 5 7-9 July 72 17 3rd grid trapping period 6 28-31 July 72 19 4th grid trapping period 7 29 Aug-1 Sept 72 29 5th grid trapping period 8 13-18 Sept 72 12 6th trapping period (13-15 Sept) 248 Establish study grid, 1st grid trapping period 13-18 Sept 72 Removal of animals for Experiment 2 (16-18 Sept) 13-18 Sept 72 Caputre of animals for Experiment 3 (16-18 Sept) 9 13-15 Oct 72 25 7th grid trapping period, short term removal evaluation 10 27-29 Oct 72 12 8th grid trapping period, short term removal evaluation 11 27-28 April 73 180 9th grid trapping period, long term removal evaluation TABLE 4.— continued. Trip Number Trapping Periods Number of Days Between Visits Activities 12 19-20 May 73 21 10th grid trapping period, long term removal evaluation 13 9-10 June 73 20 11th grid trapping period 14 7-9 July 73 27 12th grid trapping period 15 26-28 July 73 17 13th grid trapping period 16 11-13 Aug 73 14 14th grid trapping period 17 28-30 Aug 73 15 15th grid trapping period 18 19-21 Sept 73 20 16th grid trapping period 19 10-15 Oct 73 19 17th grid trapping period, capture of animals for Experiment 3 EXPERIMENT 1: HABITAT OCCURRENCE Introduction On the basis of pilot studies, the writer estab­ lished that: (1) P. m. gracilis occupied open habitats in the Kingston Plains; Blair (1941) had already observed that populations of P. m. gracilis occurred in forest habitats in the Cusino Wildlife Refuge. (2) Individuals captured from either Cusino or Kingston selected the cue set modeling the habitat in which they occurred. Isolated woodlots within the matrix of open habitats in the Kingston Plains were also occupied by P. m. gracilis. Individuals occurring within areas of habitat interface between open and forest habitats in the Kingston Plains could either range between habitats or be restricted to one habitat type. Experiment 1 was designed to test the hypothesis that individuals occurring in areas of forest and open 50 habitat interface in the Kingston Plains tend to be parti­ tioned by habitat type. Procedure In order to test Hypothesis 1, subgrids were centered along the two zones of habitat interface in the study grid as illustrated in Figure 10. Each subgrid included 81 trap sites and was 109.7 meters square with an area of 1.2 hectares (360 feet) (2.96 acres). Subgrids were divided into four equal quadrants, two in each habitat, to measure movement between and within habitats. quadrant was 54.9 meters Each (180 feet) square with an area of .3 hectares (.74 acre). To be evaluated on an equal basis, all quadrants, regardless of habitat, should have similar densities of P. m. gracilis. In order to obtain a relative comparison of densities in different habitats, an average number of individuals captured per trap site was calculated for open, transitional and forest habitats (Table 5). The overall average number of individuals captured per trap site was 3.16. Average numbers of individuals captured per trap site in different habitats were similar; 3.10 for open, 3.13 for transition and 3.27 for forest. Only those individuals which have the potential to change habitats should be included in a test of Hypothesis 1. Therefore individuals used in this test were chosen on the basis of the following criteria: 51 Figure 10.— Map of subgrids and quadrants used in the habitat partitioning study (Experiment 1). SUBGRIDS AND QUADRANTS FOR HABITAT PARTITIONING STUDY ROMAN NUMERALS * QUADRANTS — 1 1 * 1 1 1 1 1 I I 1 1 1 1 1 2 3 1 4 5 6 7 8 ill 9 J I I j— I L 1 II II' 2 3 4 1* 5 6 IV IV* 7 III' 8 9 J— I— I— I__I A B C D E F G H I » » » JL_J «- « 1 i J K L M N O P Q R S T U V Distance between traps » 13.72 meters Figure 10 » SCALE: I mm. • 2 m. U1 tsj TABLE 5.— Summary of trap sites and individual captures in open, transitional, and forest habitats in the study grid for both trapping seasons. Habitat Number of Traps % Total Traps Number of Individual Captures % Total Individuals Average Number of Individuals per Trap 107 54.0 332 53.1 3.10 Transition 31 15.7 97 15.5 3.13 Forest 60 30.3 196 31.4 3.27 Totals 148 100.0 625 100.0 Open Average = 3.16 54 (1) Centers of activity must fall within one of the subgrids in areas of habitat interface. Centers of activity (Hayne, 1949) were cal­ culated by averaging row coordinants of capture and column coordinants of capture for all recaptures. Individuals utilized in this test were captured a minimum of three times. (.2) Radii of activity must extend from the quad­ rant in which the center of activity is located into at least one other quadrant. Radii of activity were defined as the maximum distance between capture points and the center of activity. (3) Radii of activity must be sufficiently distant from the center of activity to allow the individual to cross into a different habitat. Movement between habitats was defined as the extension of radii of activity from the quadrant in which the center of activity was located into a quadrant of a different habitat. Individuals which had radii of activity extending into both different and similar habitats relative to the center of activity were included in the category of individuals which changed habitats. Data for Experiment 1 was gathered on Trips 1 through 6 and Trips 11 through 19. The statistical hypothesis tested was that the number of individuals moving between habitats differed 55 significantly from the number of individuals moving within habitats in zones of habitat interface in the study grid. Results In order to compare numbers of individuals moving between and within habitats, it would be desirable to pool data from both subgrids and habitats. done only However, this can be if there are no significant differences in the proportions of animals captured in either subgrids or habitats. Results of a chi square test for differences between subgrids is presented in Table 6. A priori alpha significance levels were set at .05 for all chi square tests. Differences were not significant (.95 < P < .98). Differ­ ences between habitats were tested in Table 7; differences were also not significant (P > .99). Therefore, individuals from both subgrids and habitats could be pooled in order to test Hypothesis 1. A chi square testwas used to evaluate this hypothesis and its results are presented in Table 8. The number of individuals moving between habitats differed significantly from the number of individuals moving within habitats in zones of habitat interface in the study grid. Therefore, in two sample subgrids, individual P. m. gracilis tend to be partitioned by habitat. However, there seems to be a sufficient number of individuals moving between habitats to assume gene flow between habitats. 56 TABLE 6.— Chi square analysis of distribution of individuals captured in forest and open habitats in subgrids for both trapping seasons. Subgrid Item 1 2 x2 3df .95 < P Forest Open Observed 13 13 Expected 12.2 13.8 Observed 11 14 Expected 11.8 13.2 Totals 24 27 Totals 26 25 51 .20 < .98 TABLE 7.— Chi square analysis of distributions of individuals moving within and between habitats for forest and open quadrants of both subgrids. Centers of Activity Radii of Activity Item Within Habitat Observed 19 22 Expected 19.3 21. 7 Observed 5 5 Expected 4.7 5,3 Between Habitats Totals x2 3df P > .99 .04 Forest 24 Open 27 Totals 41 10 51 57 TABLE 8.— Chi square analysis of distributions for numbers of individuals moving between habitats and numbers of individuals moving within habitats. Individuals Moving Between Habitats Item Individuals Moving Within Habitats Observed 10 41 Expected 25.5 25.5 Totals 10 41 x2 ldf Totals 51 51 18. 86 P < .001 Experiment 3 was designed to investigate differences in behavioral habitat selection in individuals found in forest habitats, open habitats, and individuals moving between habitats. EXPERIMENT 2: DENSITY MANIPULATION Introduction Based upon the results of Experiment 1, the majority of individuals occurring in both forest habitats and open habitats in the subgrids were partitioned in these habitat types. Habitat partitioning might occur in the Kingston Plains for two reasons. First, perhaps one habitat is actually favored but when densities are high, both optimal and suboptimal habitats are occupied. A second possibility is that both habitats are equally optimal to those indi­ viduals occurring in them. If densities are artificially lowered in adjacent habitats, then individuals should move to a different, more optimal habitat if the first possibility is correct. However, if the second possibility is true, then individuals should move to a similar adjacent habitat of lower density. Experiment 2 was designed to test the hypothesis that individuals occurring in areas of habitat interface tend to move into similar adjacent habitats of lower density rather than different adjacent habitats of lower density. 59 Procedure In order to test Hypothesis 2, the study grid was divided into five sectors as shown in Figure 11. Both, removal areas 1 and 2 included 45 trap stations and measured 109.7 meters (360 feet) by 54.9 meters with areas of .6 hectare (1.48 acres). (180 feet) Test areas 1 and 2 and removal area 3 included 36 trap stations and measured 109.7 meters (360 feet) by 41.1 meters (135 feet) with areas of .45 hectare (1.12 acres). Densities were artificially lowered in removal areas 1, 2 and 3 by removing 21 individuals from each area as shown in Table 9. Individuals were removed during the last three trap nights of Trip 8 (16-18 September 1972) as indicated in Table 4. Individuals in forest and plains test areas could then move into similar or different adjacent habitats of lower density. The occurrence of specific individuals in test areas was established prior to the removal period on the basis of capture information obtained during Trip 7 (29 August-1 September 1973) and the first three trap nights of Trip 8 (13-15 September 1972) . There is a possibility that changes in the move­ ments of individuals subsequent to the manipulation of population densities in adjacent habitats may vary with, seasonal changes in population density. For example, individuals might move into areas of lower density in the 60 ( Figure 11.— Map of removal and test areas for density manipulation study (Experiment 1). REMOVAL AND TEST AREAS FOR DENSITY MANIPULATION STUDY A B C D E F Q H I H I i I f c Tf ~ r ~ J K L M f M O P O R S T t— T T U s~r V 1 1 2 2 3 4 3 5 REMOVAL REMOVAL TEST I 2 I 4 TEST 2 REMOVAL 5 6 6 7 7 8 e 2_J I L J g L A B C D E F G H ■1-- 1— , —L I 1 J K L M N 0 P Q R S T U V Distance between traps * 13.72 meters S C A L E : Imm. ■ 2 m. Figure 11 o\ TABLE 9.— Summary of trap stations and numbers of individuals removed and tested in short term and long term phases of the density manipulation experiment. Total Number of Individuals Areas Predominant Habitat Category Total Trap Stations Removed Short Term Long Term Removal 1 Open 45 21 0 0 Test 1 Forest 36 0 23 6 Removal 2 Forest 45 21 0 0 Test 2 Open 36 0 20 4 Removal 3 Forest 36 21 0 198 63 43 Totals 10 63 fall when pressure 1from intraspecific competition is likely to be at a maximum but return to areas previously occupied the following spring after high winter mortality. Such seasonal density related changes could obscure differences between optimal and suboptimal habitats that this experi­ ment was designed to test. To investigate this possibility, both short term and long term movements of individuals in test areas 1 and 2 were investigated. Short term movements were defined as those established by recaptures in subse­ quent trapping periods during the same season as they were removed; Trip 9 (13-15 October 1973) and Trip 10 (27-29 October 197 3). Long term movements were evaluated on the basis of recaptures during trapping periods in the next season; Trip 11 (27-28 April 1973) and Trip 12 (19-20 May 197 3). The hypothesis was that trends in short term and long term movements would be similar. In test area 1, those individuals recaptured in removal area 1 were included in the category of individuals moving to a different habitat of lower density; those recaptured in removal area 2 were defined as individuals moving to a similar habitat of lower density. In test area 2, individuals recaptured in removal area 2 moved to a different habitat of lower density, and those captured in removal area 3 moved to a similar habitat of lower density. Those individuals from test areas recaptured in both adjacent removal areas were placed in the category of individuals moving to different habitats of lower density. 64 Results In order to compare numbers of individuals moving to different and similar habitats of lower density, it would be desirable to pool data from both test areas and habitats. A general summary of numbers of individuals remaining in test areas and those moving to removal areas is presented in Table 10. Pooling of test areas and habitats can be done only if there are no significant differences in the proportions of individuals captured in test areas and those moving into adjacent removal areas of lower density between test areas. TABLE 10.— Numbers of individuals remaining in test area and numbers of individuals moving to adjacent areas of lower density for both test areas in short term removal study. Test Area Habitat Individuals Remaining in Test Area Individuals Moving into Removal Areas Totals 1 Forest 7 16 23 2 Open 4 16 20 11 32 43 Totals Results of a chi square test for differences in the numbers of individuals moving to areas of lower density from forest and open test areas are presented in Table 11. A priori alpha significance levels were set at .05 for all chi square tests. Differences were not significant 65 TABLE 11.— Chi squhre analysis of distributions for numbers of individuals moving to areas of lower density from forest and open test areas. Test Area Habitat Areas of Lower Density Item Similar Habitat Observed Different Habitat Open Totals 12 14 26 Expected 13 13 Observed 4 2 Expected 3 3 16 16 Forest Totals x2 3df 6 32 .82 .80 < P < .90 (.80 < P < .90); therefore, it is possible to pool the data from both test areas. Of a total of 32 individuals moving to habitats of lower density, 26 moved to similar habitats and only 6 moved to different habitats. These results were significantly different in a chi square test (.001 < P < .01) presented in Table 12. Unfortunately, winter mortality was high and sufficient animals for statistical tests were not recap­ tured during Trips 11 and 12, the following season. However, the results, as presented in Table 13, are in the direction hypothesized. In two test areas within the study grid, individuals moving to areas of lower density occupied similar rather than different habitats. 66. TABLE 12.— Chi square analysis of distributions for numbers of individuals moving to a similar habitat of lower density and numbers of individuals moving to a different habitat of lower density. Individuals to Similar Habitat Item Individuals to Different Habitat Observed 26 6 Expected 16 16 Totals 26 6 x2 32 32 12.5 ldf .001 < P Totals <.01 TABLE 13.---Summary of long term removal study indicating numbers of individuals remaining in test areas and those moving to removal areas with different and similar habitats. Individuals Moving into Removal Areas Test Area Habitat Individuals Remaining in Test Area Similar Habitat Different Habitat Totals 1 Forest 6 2 0 8 2 Open 4 3 0 7 10 5 0 15 Totals EXPERIMENT 3: HABITAT SELECTION Introduction The results of Experiments 1 and 2 indicated that individuals were partitioned by habitat type in the study grid and that habitat occurrence could not be modified by density reductions in adjoining different habitats. In areas of habitat interface, three categories of P. m. gracilis could be distinguished on the basis of habitat occurrence; individuals occurring in forest habitats, individuals occurring in open habitats and those individuals moving between habitats. Experiment 3 was designed to compare behavioral habitat selection in these three categories of individuals. Hypothesis 3 was that individuals would select cue sets modeling the habitat type from which they were captured in the study grid. It was also hypothesized that open habitat and forest habitat groups would differ significantly in habitat selection. 68 Habitat Modeling and Preference Measurement Habitat models used in behavioral habitat selection studies have been of two major types* Harris (1952) utilized artificial cues to model habitats; for example, strips of manila folders simulated grassland in his study. Animals were allowed to chose between two cue sets model­ ing different habitats which were located in relatively small enclosures (3.2m x 2.7m). Harris controlled all environmental variables except habitat cues in the enclos­ ures. In contrast, Wecker (1963) utilized a relatively large outdoor enclosure (30.5m x 4.9m) in which few environmental variables could be controlled. The enclosure was centered along a forest-field interface in order to allow individuals to choose between two natural habitats. Not only were the data more difficult to analyze in Wecker’s study but the results were less conclusive than those of Harris. The total area of Wecker's enclosure represented only 10% of the total home range of P. m. bairdii as estab­ lished by Blair (1942). The tendency to explore the entire enclosure area might outweigh tendencies to select a specific habitat within such a confined space. Perhaps an animal's initial preference for a specific habitat decreases with the amount of time it spends in a small enclosure. Such possible changes in habitat preference should be 69 evaluated on the basis of initial habitat choices as well as subsequent time related measures of preference. Habitat models for the present study were designed as follows: (1) Habitats were modeled in relatively small enclosures (1.83m x 1.22m) under laboratory conditions in which such environmental variables as light and temperature could be constantly controlled. (2) Habitats to be modeled were selected on the basis of trapping results in the Kingston Plains. Each of the habitats modeled were selected by different individual P/ m. gracilis in pilot studies. (3) Vegetational cues used to model each habitat were selected on the basis of their prominence and abundance in the natural habitats. Natural vegetation was used to model the habitats on a small scale relative to the enclosure size. (4) The effects of seasonality were kept constant by modeling the habitats as they appeared during a specific season. In this study, the fall season was chosen because habitat cues such as forest leaf litter could be easily modeled. Fall was also the season in which the population density of P. m. gracilis 70 was highest and test specimens were most easily obtained. All vegetation used in the models was obtained from the Kingston Plains. Only those animals captured during the fall season were tested in these models. Two types of dependent variables were chosen to measure habitat preference in this study (Table 14). In pilot studies, differences in the initial habitat choice in P. m. gracilis captured at Cusino and at Kingston were highly significant (P < .01). In the present study, the initial choice was measured by the first habitat that the animal actually entered in the enclosure and b£ two time related measures. Time related measures for the initial choice were time spent orienting to each habitat before being released from a central runway compartment and time spent orienting to each habitat after being released from the runway compartment and prior to the animal's entrance into a specific habitat. Habitats should be actively selected at night because P. m. gracilis is nocturnal. In pilot studies, daytime activity in the enclosures was minimal. Therefore the second type of dependent variable was time related and measured the amount of time spent in each habitat compart­ ment during each of three succeeding nights subsequent to the individual's initial choice. The consistency of such time related measures should provide an indication of whether habitat preference was time specific in confined enclosures. 71 TABLE 14.— Dependent variables used to measure preference for cue sets modeling forest and open habitats, Variable Unit of Measurement Runway time Percent of time spent orienting to open habitat model prior to release from runway compartment. Initial choice time Percent of time spent orienting to open habitat model subsequent to release from runway and prior to habitat choice. Initial habitat choice Habitat compartment first entered subsequent to release from runway. First night compartment time Percent of time spent in open habitat model compartment. Second night compartment time Percent of time spent in open habitat model compartment. Third night compartment time Percent of time spent in open habitat model compartment. Average night compartment time Percent of time spent in open habitat model compartment. Test Apparatus Four pens (2.44m x 1.83m x 1.52m) were constructed of plywood panels for this experiment and 17). (Figure 12, 13, 16 Each pen was divided into two equal compartments measuring 1.83m x 1.22m by a plywood partition. Adjoining compartments were connected by a plexiglass runway (24 cm x 7 cm x 7 cm) mounted in the middle of the partition, above the floor. 30 cm The runway was equipped with a treadle and mercury switch to monitor movements between compartments and to measure total time spent in each compartment. Figure 12.— Experimental enclosures for habitat selection studies (Experiment 3). Figure 12 74 Figure 13.— Floor plan of experimental enclosures indicating the position of center runways and causeways, nestboxes, and food and water supplies. E N C L O S U R E S F O R H A BI TA T S E L E C T I O N T E S T S Pen 4 Pen I Nest Box Running Wheel Open Habitat —Food Food — D Water Water 0 ’ Running Wheel Nest Box Open Habitat Wire Vertical Door •Center Partition ■ Runway Forest Habitat Nest Box Running Wheel Pen 2 ----- —NestmBox . V water m Forest Habitat Food Pen 3 s Food— •Food Running Wheel Forest Habitat —Food water q , Running Wheel Nest Box * b Water Water Q• Open Habitat ■Center Partition' Open Habitat . D Water Nest Box Running Wheel Water Forest Habftat Running Wheel Nest Bax •Food Food — Figure 13 76 | Figure 14.— Recording apparatus for the experimental enclosures. All microswitch data is recorded by the Esterline Angus Event Recorder. The clocks measure total time spent in each compartment and the counters record activity wheel revolutions. Figure 15.— Center runway and causeway extending into compartment modeling forest habitat. Note the vertical sliding plastic doors in the runway and the treadle. ip 77 ?sa ?«*i W|«0 ^«{Tp;infW -v r^,”wvt4'tffj r c '->•!. t , £,1 ~ /ft .iAti.'f »**/ \ *i ’< » j*r-0-I -* Vu t"*sub*a>vv.t * •/Ss'," p f/»u, « <» >* * '“ SW* -sffe- V Figure 15 78 < Figure 16.— Compartment modeling forest habitat. Natural vegetation from isolated woodlots in the Kingston Plains was used in this compartment. Figure 17.— Compartment modeling open habitat. Natural vegetation from open lichen-grass associations in the Kingston Plains was used in this compartment. 'r- v 4*&*: Figure 17 80 Vertical sliding doors of clear plexiglass were installed at both ends of each runway to allow individuals to visually inspect the compartments without actually entering them. When the sliding doors were closed, movements within the runway and orientation to each compartment were measured by the treadle and mercury switch. The doors were installed so they could be raised from outside the pen which minimized the disturbance to the test animal. The runway was connected to the floor of each compartment by a sloping, wire mesh causeway (Figure 15). Mirrors were positioned above each compartment at a 45° angle in order to observe animals in the runways and causeways without disturbing them. Forest and open habitats were modeled with samples of .plant vegetation characteristically found in each area in the fall season. Species of plants utilized in the compartments are listed in Table 15; all species were either dominant or very common in their respective habitats. In plains compartments (Figure 17) ground cover was dominated by an association of lichen and grass; three bracken ferns were also placed in each of the open habitat compartments. Forest associations of red maple, beech, birch and white pine were represented by one short forest compartments (Figure 16). sapling of each in the Six red maple seedlings were placed in each forest compartment and leaf litter was represented by red maple leaves. Stumps and logs were 81 TABLE 15.— Listing of vegetation comprising models of open habitat and forest habitat in enclosures. Habitat Model Open Plant Species Approximate Density in Compartment Deschampsia flexuosa, hair grass 70% of surface area Pteridium aquilinum, bracken fern 1.75 per sq. m.» Cladonia mitis, grey lichen Cladina ranqiferina, lichen 30% of surface area Cladina alpestris, lichen Forest Acer rubrum, red maple; leaves 90% of surface area Acer rubrum, red maple; saplings .43 per sq. m. Betula cerulea, white birch saplings .43 per sq. m. Pinus strobus, white pine seedlings .43 per sp. m. found in both habitats; one small stump and three pieces of logs were placed in each compartment. Each compartment was provided with a nestbox, running wheel, water bottle and wire container of food. Each nestbox (19.1 cm x 19.1 cm x 8 cm) was installed along the compartment wall opposite the partition, 55.3 cm from one corner and 53.3 cm from the other corner. Nestboxes were equipped with a trap door connected to microswitches in order to monitor the amount of time spent in the nestbox. A running wheel with a base of 20.3 cm x 15.2 cm was placed 82 along the same wall/ ,38.1 can from the nestbox and 55.3 cm from the front corner of the compartment. Running wheels were equipped with reed switches in order to measure wheel revolutions and duration of activity. Water was supplied by 100 ml graduated cylinders which were upended and attached to the outside wall so that only the metal drink­ ing tube protruded into the compartment. Food (Purina Mouse Breeder's Chow) was placed in a 1/4 in. wire mesh container which was suspended in one corner of the compartment. Each pen was illuminated by two 400 watt fluore­ scent bulbs attached to crossbeams 198 cm above the pen and parallel to the center partition. The rest of the test room was poorly illuminated in order to emphasize cues within the pens. An artificial light cycle of 14 hours light and 10 hours dark was maintained by means of an automatic timer. A faint illumination of 0.01 footcandles was provided during dark periods to simulate starlight. All time related dependent variables were recorded by a 12 volt Esterline Angus Event Recorder in order to analyze both the distribution of time as well as the total time spent in each compartment. Total time in each com­ partment could be determined by either analyzing Esterline Angus charts or by consulting 120 volt, 24 hour clocks also connected to runway treadle mercury switches (Figure 14). 83 Procedure Prior to each test period, an individual mouse was placed in the central runway and the vertical doors were closed. The individual was allowed to acclimate to the apparatus for 15 minutes. At the end of the acclimation period, the amount of time the animal spent on each end of the runway treadle was measured for a 15 minute period. This variable measured the animal's initial compartment orientation based upon one sensory modality, vision. At the end of this period, the vertical runway doors were pulled up and the animal was free to enter either compartment via the wire causeways. Individuals usually remained in the runway for several minutes before climbing down the causeways; individuals usually climbed down the causeways several times before actually entering a compartment. The initial habitat choice was defined as the first compartment substrate upon which the cinimal placed all four feet and tail. The average time from runway release to compartment choice was 11.4 minutes. Animals were usually released in the pens in the afternoon, between 1600 and 1700 hours (4 to 5 PM) . Individuals were then left in the pens for the following 3 nights and 2 1/2 days. Nestboxes were checked daily (during daylight hours) to verify the animal's position. At the end of the test period, the animal was removed from the enclosure. 84 A large number of animals were placed in each pen prior to any tests in order to distribute scent equally in both compartments. This was done as a precaution against having habitat preference confounded by olfactory stimuli from previous test animals. Variations in behavior due to age and sex effects were avoided by using subadult and adult males of at least 90 days of age as test subjects. Seasonal effects were minimized by using only those mice which were captured in September and October. Data on activity cycles and food and water consump­ tion were also recorded and will be presented in a future paper. Results Of seven dependent variables used to measure behavioral habitat selection, initial cue set choice was based upon the numbers of individuals in each group which selected each cue set; the other six variables were based upon time measures. A summary of the results for each variable is presented in Table 16. Initial choice was defined as the first cue set which was chosen by the animal after its release from the runway. Of 15 individuals captured from forest habitats in the study grid, 14 chose cue sets modeling forest and only 1 chose a cue set modeling open habitat. The differences between the numbers of individuals selecting each cue set mmmmgmmmwmmg! TABLE 16.- Summary of group means of individuals occurring in forest habitat, open habitat and both habitats for seven dependent variables. Time in Compartment -----------------------------------Average 1st 2nd 3rd Nights Night Night Night N x% x% x% x% Habitat Occurrence Habitat Preference N x% Runway Time N x% Initial Choice Time N x% Forest 39 6.8 15 19.5 15 29.5 13 24.3 33.0 17.0 15.9 Open 17 94.2 17 81.7 16 76.5 14 67.0 67.2 64.4 69.5 Both 10 50.0 10 40.2 10 44.5 10 51.5 47.1 49.9 53.2 86 I were highly significant (P << .01) in a chi square test. I In contrast, 5 of those individuals moving between habitats 1 in the study grid chose cue sets modeling forest and 5 I individuals chose cue sets modeling open habitat. I results are clearly not significant. 1 gency test (Table 17) was used to test for differences in These A chi square contin- the proportions of individuals captured in forest and open 1 habitats that chose cue sets modeling these habitats. | proportions of individuals differed significantly (P < .001). j Therefore, significant numbers of individuals from both I forest and open habitats in the study grid chose cue sets I modeling the habitat in which they were captured. 1 neither cue set was chosen by 4 The However, a significant number of those individuals that were trapped in both habitats in the study i grid. j TABLE 17.— Chi square analysis of distributions of individuals captured from forest and open habitats that selected cue sets modeling these habitats. Habitat Occurrence Habitat Model Selection Item Forest Observed 14 Expected 7 Observed 1 Expected 7.9 Open Totals x2 3df P < .001 24.4 Forest 15 Open Totals 1 15 7.9 16 17 9.03 17 32 87 The six time dependent variables were expressed in terms of the percent of time spent in compartments with cue sets modeling open habitats. Percentage data may approximate a normal distribution but the range of values must be from 0% to 100%; in a true normal distribution, the tails of the curve should extend to infinity. For this reason, percentage data will not conform to a normal dis­ tribution unless an arcsin transformation is utilized (Sokal and Rohlf, 1969). Therefore, all percentage data was transformed to arcsin values for statistical tests. Within each group of animals (those captured in forest, open or both habitats), preference was defined as the difference between the group mean of the percent of time spent in compartments modeling open habitats and a theoretical value of 50% (arcsin = 45) which indicated no preference. A 1 sample T test was used to test for pre­ ferences within groups for each time dependent variable. An a priori alpha significance level of .05 was set for eachtest. The results are summarized in Table 18. Group means of those animals captured in forest and those animals cap­ tured from open habitats differed significantly from a mean of 50% at an alpha level of .05 or below for all time dependent variables tested. In both groups, time in the open habitat compartment during the first dark period had the lowest significance level. None of the group means of animals captured in both habitats differed significantly TABLE 18.— Results of a 1 sample T test of within group habitat preferences for 6 time related variables for animals captured in forest, open and both habitats. All data have been transformed to arcsin values. Time in Open Habitat Compartment 2nd Dark Period Arcsin 3rd Dark Period Arcsin Habitat Occurrence Runway Time N Arcsin Forest 15 19.7*** 15 30.3** 13 26.5*** 31.7* 19.5*** 19.1*** Open 17 69.8*** 16 65.8*** 14 56.4** 57.6* 55.8* 60.5* Both 10 34.4 10 44.5 10 46.0 43.5 44.2 47.0 ***p < .001 **p < .01 *p < .05 Average Dark Period N Arcsin 1st Dark Period Arcsin Initial Choice Time N Arcsin 89 from 50% at the alpha level of .05 for any of the dependent variables tested. Therefore, individuals captured in only one habitat preferred models of the same habitat in behavioral selection tests; those individuals captured in both habitats exhibited no such preference. In Hypothesis 3, significant differences were also expected between group means of those animals captured in forest habitats, those captured in open habitats and those captured in both habitats. A one way Analysis of Variance test for unequal sample sizes (Sokal and Rohlf, 1969) was used to test whether any significant differences existed between group means for each time dependent variable. An a priori alpha level of significance of .05 was set for all ANOVA tests. Results are presented for each dependent variable in Analysis of Variance tables included in Tables 19, 21, 23, 25, 27 and 29. Significant differences (P < .001) existed between the group means for each depen­ dent variable tested. However, ANOVA tests cannot indi­ cate the specific sources of variations in group means which produce significant differences. An a posteriori multiple comparisons test, the Tukey HSD test, was used to examine the magnitudes of differences between specific group means (Kirk, 1968). Tukey's test utilizes a predetermined HSD value based upon a "q" value and the square root of the MS error the sample size as shown on the following page. divided by 90 HSD = g Mav MS N error n If the subtracted difference between two group means is greater than the HSD value, then themeans are significantly different at the alpha level specified for the "q" value. In this analysis, HSD values were calculated for alpha levels of .05 and .01. Group mean differences and HSD values are presented for each time dependent variable in Tables 20, 22, 24, 26, 28 and 30. Differences among group means of animals captured in forest and those captured in open habitats were significant (P < .01) for all time dependent measures. Differences among group means of animals captured in forest and those captured in both habitats were significant (P < .01) for measures of preference in the second and third dark periods as well as the average dark period; differences for runway time, initial choice time, and the first dark period were not significant in these two groups. Differences among group means of animals captured in open habitat and those captured in both habitats were significant at P < .01 level for runway time and at the level of P < .05 for initial choice time; dif­ ferences in other variables among these two groups were not significant. These results indicate that there is strong evidence for the acceptance of Hypothesis 3. TABLE 19.— Analysis of variance table for percent of time spent orienting to the open habitat model prior to release from the runway (runway time). df Source of Variation 2 Among groups Within groups 39 SS NS 102814.8 51407.4 18820.1 F Significance Level 106.5 P < .001 482.6 - TABLE 20.— Tukey's test of differences among means for runway time. HSD = q NSerror n Habitat Occurrence x forest = 30,3 xopen = 65‘8 *both " 44,1 **p < .01 H S D (a = .05)7 HSD, A,x = 25.9 (a = .01) • 9 forest — V open 35.5** — Xboth 14.2 21.3* TABLE 21.— Analysis of variance table for percent of time spent orienting to the open habitat model subsequent to runway release and prior to habitat choice (initial choice time). Sources of Variation df SS 2 10056 Among groups Within groups 50283.6 92.65 Significance Level P < .001 542.7 21707.7 40 F MS TABLE 22.— Tukey's test for differences among means for initial choice time. MS HSD - error n Habitat Occurrence ^forest ^ .7 *open “ 69•8 *both = 34-4 **p <.01 *p <.05 4; H S D (ot = .05) " 2 1 ' ^forest — HSD (a = .01) = 27,2 open *both 50.1** 14.7 — 35.4** TABLE 23.— Analysis of variance table for percent of time spent in the open habitat model compartment during the first night (first night compartment time). Sources of Variation df F MS 2 78465.1 39232.6 35 12434.5 355.3 Among groups Within groups SS Significance Level 110.4 P < .001 TABLE 24.— Tukey's test for differences among means for first night compartment time. HSD = q a,v MS error n Habitat Occurrence xforest 5open _ = 57.6 x, .. = 43.5 both **p < .01 *p < .05 .7 H S D . R. 1969. Experimental studies of competitive interaction in a two-species system. I. Microtus and Clethrionomys species in enclosures. Canada J. ZooT: 47:1059-1082. . 1970. Experimental studies of competitive interaction in a two-species system. II. The behaviour of Microtus, Peromyscus and Clethrionomys species. Anim. Behav. 18:411-426. 108 109 Hagen, D. W. 1967; Isolating mechanisms in threespine sticklebacks. J. Fish Res. Bd. Canada 24:16371692. Hall, E. R. and K. R. Kelson. 1959. The Mammals of North America. Ronald, N.Y. 2 Vols. Harris, V. T. 1952. An experimental study of habitat selection by prairie and forest races of the deermouse, Peromyscus maniculatus. Contrib. Lab. Vert. Biol., Univ. Mich. VoTI 56, 53pp. Hasler, A. D. 1966. Underwater guide posts. Wisconsin Press, Madison. Univ. Hatt, R. T., J. Van Tyne, L. C. Stuart, C. H. Pope, and A. B. Grobman. 1948. Island Life: a study of the land vertebrates of the islands of eastern Lake Michigan. Cranbrook Inst, of Sci. Bull. 27, 179pp. Hilden, 0. 1965. Habitat selection in birds. Zoologici Fennici 2:53-75. Annales Jenkins, B. C. 1942. Unpublished Pittman - Robertson quarterly reports. Project 6-R, July 15, 1942. Cusino Wildlife Experiment Station files. Joslin, J. 1971. Visual cues used in orientation by white footed mice, Peromyscus leucopus: a laboratory study. Ph.D. Thesis, Michigan State University. Kelley, R. W. and W. R. Farrand. 1967. The glacial lakes around Michigan. Geol. Survey Mich. Dept. Conserv. Bull. 4, Lansing. Kettlewell, H. B. D. 1959. New aspects of the genetic control of industrial melanism in the Lepidoptera. Nature 183:918-921. __________ . 1965. Insect survival and selection for pattern. Science 148:1290-1296. Kirk, R. E. 1968. Experimental Design: Procedures for the Behavioral Sciences. Wadsworth Publ. Co. , Inc. Belmont, Cal. Klein, J. 1960. Ecological relationships of Peromyscus leucopus noveboracensis and P. maniculatus gracilis in central New York. Ecol. Monogr. 30:387-407. 110 Klopfer, P. H. 1962. Behavioral Aspects of Ecology. Prentice-Hall, Inc., Englewood Cliffs, N.J. ______ 1965. Behavioral aspects of habitat selection. I. Wilson Bull. 77:376-381. 1969. Inc., N.Y. Habitats and Territories. 117p. Basic Books, Kluvyer, H. N. and L. Tinbergen. 1960. Territory and the regulation of density in titmice. Archives Nerrlandaises de Zoologie 10:265-289. Krebs, C. J. 1972. Ecology: The experimental analysis of distribution and abundance. Harper and Row, N.Y. Lack, D. 1933. Habitat selection in birds with special reference to the effects of afforestation on the Breckland avifauna. J. Anim. Ecol. 2:239-262. ________ . 1937. The psychological factor in bird distriEiution. Brit. Birds 31:130-136. MacArthur, R. H. 1965. Patterns of species diversity. Biolog. Revs. 40:510-533. Manville, R. H. 1949. A study of small mammal populations in northern Michigan. Misc. Publ. Mus. Zool. Univ. Mich. 73:1-83. Maturana, H. R., J. Y. Lettvin, W. S. McCullock and W. H. Pitts. 1960. Anatomy and physiology of vision in the frog. J. Gen. Physiol. 43:129-176. Mayr, E. 1947. Ecological factors in speciation. Evolution 1:263-288. Miller, C. A. 1973. Behavioral habitat selection in Peromyscus and Microtus. Paper presented at 53rd Annual Meeting of the American Society of Mammalogists. (with J. F. Cassel) Munn, N. L. 1950. Handbook of Psychological Research on the Rat. Houghton Mifflin Co., Boston. Norris, K. S. 1963. The function of temperature in ecology of the percoid fish Girella nigricans (Ayres). Ecol. Monogr. 33(1):23-62. O'Hara, M. 1974. Effects of developmental stage and prior experience on habitat selection in three species of anuran larvae. M.S. Thesis, Michigan State Univer­ sity. Ill Ozoga, J. J. pers. comm. , and C. J. Phillips. 1964. Mammals of Beaver Island, Michigan. Publ. Mus. Mich. State Univ. Biol. Series 2(6):305-348. , and L. J. Verme. 1968. Small mammals of coni­ fer swamp deeryards in northern Michigan. Paper of the Mich. Acad. Sci., Arts & Letters 53:37-49. Pittendrigh, C. S. 1958. Adaptation, natural selection and behavior. In Behavior and Evolution, ed. by A. Roe and G. G. Simpson: 390-416. Yale Univ. Press, New Haven. Quertemus, C. J. 1972. Prior experience as a factor in habitat selection of the cichled fish Tilapia mossambica. Ph.D. Thesis, Michigan State University, Rahman, H., M. Rahman and J. A. King. 1968. Comparative visual acuity (minimum separable) in five species and subspecies of deermice (Peromyscus) . Physiol. Zool. 41:298-312. Rosenzweig, M. L. and J. Winakur. 1969. Population ecology of desert rodent communities: Habitats and environ­ mental complexity. Ecology 50:558-572. Russell, P. F . , and T. R. Rao. 1942, On relation of mechanical obstruction and shade to ovipositing of Anopheles culicifacies. J. Enp. Zool. 91:303-329. Sale, P. F. 1969. Pertinent stimuli for habitat selection by the juvenile manini, Acanthurus triostegus sandvicensis. Ecology 50 (4) :616-6*23. Sargent, T. D. 1965. The role of experience in the nest building of the zebra finch. Auk 82(1):48-61. Schoonhovern, L. M. 1968. Chemosensory bases of host plant selection. Ann. Rev. Entomol. 13:115-136. Sexton, O. J. 1958. The relationship between habitat pre­ ferences of hatchling Chelydra serpentina and the physical structure of the vegetation. Ecology 39:751-754. Sheppe, W . , Jr. 1961. Systematic and ecological relations of Peromyscus oreas and P. maniculatus. Proc. Amer. Phil. Soc. roS:421-446.” 112 Sokal, R. R. and F. J. Rohlf. 1969- Biometry. The principles and practice of statistics in biological research. W. H. Freeman & Co., San Francisco, 776pp. Sutherland, N. S. 1960. Theories of shape discrimination in octopus. Nature. London. 186:1092-1094. Switzenburg, D. F. 1957. pers. comm. Taylor, D. H. 1972. Extra-optic photoreception and compass orientation in larval and adult salamanders (Ambystoma tigrinum). Anim. Behav. 20(2):233-236. Thoms, C. L. 1974. Effects of prior experience upon nest­ ing and resting site selection by the Mossambique mouth brooder, Tilapia mossambica. Ph.D. Thesis, Michigan State University. Thorpe, W. H. 1939. Further experiments on pre-imaginal conditioning in insects. Proc. Roy. Soc. 127:471-82. __________ . 1945. selection. The evolutionary significance of habitat Jour. Anim. Ecol. 14(2):67-70. __________ . 1963. Learning and Instinct in Animals. Harvard Univ. Press, Cambridge. Tinbergen, L. 1960. The natural control of insects in pine woods. I. Factors influencing the intensity of predation by song birds. Arch. Neerl. Zool. 13:265-343. Turner, E. R. A. 1961. Survival values of different methods of camouflage as shown in a model population. Proc. Zoolog. London. 136:273-284. Uexkiill, J. von. 1921. Umwelt und Innenwelt der Tiere. Springer, Berlin. Van Etter> Robert. 1957. Flat pea plants on Kingston Plains. Survey Report, Sept. 16, 1957. Michigan Department of Natural Resources. Veatch, J. 0., L. R. Schoenmann and F. R. Lesh. 1929. Soil survey of Alger County, Michigan. Bureau of Chemistry and Soils, Series 1929, no. 32. 113 The role of early experience in habi­ Wecker, S. C. 1963. tat selection by the prairie deermouse, Perotnyscus maniculatus bairdii. Ecol. Monogr. 33:367-325J 1970. Effects of early experience on sub­ Wiens , J. A. strate pattern selection in Rana aurora tadpoles. Copeia (3):543-548.