U M I INFORMATION TO USERS. continuing from left to right in equal sections with small overlaps. Each

Transcription

1 INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedtbrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. U M I University Microfilms International A Bell & Howell Information Company 300 North Zeeb Road. Ann Arbor. MI USA 313/ :

2

3 Order Number Competition and predation as processes affecting community patterns of geese Gawlik, Dale Edward, Ph.D. Texas AkM University, 1994 U M I 300 N. Zeeb Rd. Ann AIbor, MI48106

4

5 COMPETITION AND PREDATION AS PROCESSES AFFECTING COMMUNITY PATTERNS OF GEESE A Dissertation by DALE EDWARD GAWLIK Submitted to the Office of Graduate Studies of Texas A&M University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY August 1994 Major Subject: Wildlife and Fisheries Sciences

6 COMPETITION AND PREDATION AS PROCESSES AFFECTING COMMUNITY PATTERNS OF GEESE A Dissertation by DALE EDWARD GAWLIK Submitted to Texas A&M University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Approved as to style and content by: (Chair of Committee) ~~ (Member) Milton W. Weller (Member) NovafSilVY (Member) flk~ Robert D. Brown (Head of Department) August 1994 Major Subject: Wildlife and Fisheries Sciences

7 iii ABSTRACT Competition and Predation as Processes Affecting Community Patterns of Geese. (August 1994) Dale Edward Gawlik, B.S., University of Wisconsin Stevens Point; M.S., Winthrop College Chair of Advisory Committee: Dr. R. Douglas Slack I investigated patterns of behavior, association, and abundance for geese at three organizational scales (Le., population, flock, and individual) to determine the relative effects of interspecific competition, intraspecific competition, and predation. I examined patterns of population density, habitat use, niche breadth, niche overlap, flock structure, foraging behavior, and absolute energetic expenditure for Snow Geese (Chen caenl1escens), Greater White-fronted Geese (Anser albifrons), and Canada Geese (Branta canadensis). This study was conducted during November 1991 to February 1992 and October 1992 to February 1993, in southeastern Texas. Snow Geese overlapped considerably with white-fronted and Canada geese along resource axes of time, habitat, and food, and used a broader range of these resources than either of the latter two species. White-fronted and Canada geese overlapped less on time and habitat use and more on diet. These two species also showed more restrictive patterns of temporal abundance and habitat use than did Snow Geese. Collectively, these patterns are consistent with those of a community structured to minimize competition for food by exhibiting niche complimentarity. Regression analyses indicated food (waste rice grain) most strongly influenced population abundance of white-fronted geese, whereas snow and Canada geese were affected by

8 IV hunting pressure and food (green vegetation and waste rice grains) indirectly through interactions. At the flock level, all three species exhibited patterns strongly indicative of a response to predation, with lesser effects from food and heterospecifics. At the individual level, foraging behaviors were primarily affected by food type and abundance, with lesser effects from predation risk. The close link between Greater White-fronted Geese and waste-rice grain suggests that current declines in rice agriculture and the subsequent conversion of rice fields to pastureland wi11likely impact populations of white-fronted geese more severely than those of either snow or Canada geese.

9 v ACKNOWLEDGEMENTS Funding for this project was provided by the Southeast Texas Wildlife Foundation and the Texas Agricultural Experiment Station. Further support was provided by the Office of Graduate Studies at Texas A&M University in the form of a Tom Slick Senior Graduate Fellowship. The U. S. Fish and Wildlife Service at Attwater's Prairie Chicken National Wildlife Refuge provided housing and logistical support. I especially thank my committee chair, R. Douglas Slack, for his guidance and support throughout my tenure at Texas A&M University. His encouragement and personal challenge to think beyond the problem at hand shaped this study and directed my interest into exciting areas of ecology I had not previously discovered. I also thank Nova Silvy for extended discussions on many topics, but in particular, for reminding me of the difference between new ideas and better ideas. I am grateful to John Eltinge for many hours of consultation and the fresh insight he provided regarding the fundamental commonality between ecological and statistical problems. I acknowledge Kenneth Risenhoover for the use of his lab facilities and the encouragement to critically evaluate my own ideas as well as those of others. I thank all members of my graduate committee for their time and expertise in reviewing drafts of this dissertation and preceding proposal. I also thank numerous friends that provided their insightful ideas, comments, and encouragement throughout this study. Especially, Donna Renfrow, Felipe Chavez-Ramirez, Thomas Lewis, Douglas Harpole, and James Thomas. I am indebted to many landowners, farmers, ranchers, wildlife professionals, and hunting guides, in my study area whom provided access to their property and shared their knowledge of the rice-prairie region. In particular, William Hobaugh, David

10 vi Wintermann, Larry Sebesta, Jenny Hoskins, Michael Morrow, William Lehrer, Jessie Kalina, Harry Lobpries, Bill Krenek, and Bill Blair. Their perspectives allowed me to better perceive the interdependence of people and wildlife, and to identify the breadth of wildlife management issues in this region. I also thank David Wintermann for his support beyond the scope of this study, and for his model as a true conservationist. This study would not have occurred without the continuing support and encouragement of my parents, Louis and Carol Gawlik, and introduction to the wildlife profession by my grandmother, Rosemary Rossier. Their influence extends far beyond my professional self. Also, I acknowledge the thoughtfulness and efforts of Drs. Frederick and Marguerite Baumgartner, who transformed my passion into the beginning of a career. Drs. Frances and the late Fredrick Hamerstrom instilled in me a philosophical and scientific foundation that continues to shape my view of ecological issues. There has been no greater influence on my professional development, and for that I am grateful. I especially thank my wife, Robin, who supported me in every way possible with unwavering confidence. Her understanding, challenges, and insight made this study as enjoyable as it was productive. Finally, I thank my daughter, Glynnis, who is too young to appreciate the satisfaction and pride she provided me in the final stages of this endeavor.

11 vii CHAPTER TABLE OF CONTENTS Page I INTRODUCTION 1 Interspecific competition... 4 Intraspecific competition... 6 Predation II STUDY AREA AND METHODS... 9 III POPULATION-LEVEL PATTERNS Introduction Methods Results Discussion IV FLOCK-LEVEL PATTERNS Introduction Methods Results Discussion V INDIVIDUAL-LEVEL PATTERNS Introduction Methods Results Discussion VI CONCLUSIONS AND MANAGEMENT IMPLICATIONS Specifying model terms Comparisons among levels of study Species differences and management implications " 59 LITERATURE CITED VITA... 68

12 viii Table LIST OF TABLES Page 1 Predicted effects of competition and predation on community patterns of geese Multiple regression model of Snow Goose population density relative to abundance of rice grains, green vegetation, white-fronted goose density, Canada Goose density, hunting pressure, and interactions between rice grains and white-fronted goose density, rice grains and Canada Goose density, hunting pressure and white-fronted goose density, hunting pressure and Canada Goose density, and between white-fronted goose density and Canada Goose density (df = 99, &2 = 0040, :e <.001) Multiple regression model of Greater White-fronted Goose population density relative to abundance of rice grains, green vegetation, Snow Goose density, Canada Goose density, and hunting pressure (df = 99, &2 = 0.20, :e <.001) Multiple regression model of Canada Goose population density (log) relative to abundance of rice grains (log), green vegetation (log), Snow Goose density, white-fronted goose density, hunting pressure, and interactions between Snow goose and hunting pressure, Snow Goose and white-fronted goose, and between green vegetation (log) and white-fronted goose (df = 99, R2 = 0.44, :e <.001) " Frequency and species-composition of all goose flocks and a subset of flocks used in flock-level regression analyses for rice stubble and non-rice stubble fields in southeastern Texas ~requency.of habitats used by goose flocks dominated (> 75 %) by a SIngle species Multiple regression model of Snow Goose cohort size for flocks in rice stubble fields relative to abundance of rice grains, green vegetation, white-fronted goose cohort size, Canada Goose cohort size, distance from cover, and interactions between distance from cover and white-fronted goose cohort size,..'! and between rice grains and Canada Goose cohort size (df = 47, R~ = 0.77, :e <.001). Also, multiple regression model of Snow Goose cohort size for flocks in non-rice stubble fields relative to abundance of rice grains, green vegetation, white-fronted goose cohort size, Canada Goose cohort size, and distance from cover (df = 34, &2 = 0.49, :e =.001)

13 ix Table LIST OF TABLES (Continued) Page 8 Multiple regression model of Greater White-fronted Goose cohort size (log) for flocks in rice stubble fields relative to abundance of rice grains, green vegetation, Snow Goose cohort size, Canada Goose cohort size, and distance from cover (df = 83, &2 = 0.31, <.001). Also, multiple regression model of Greater White-fronted Goose cohort size for flocks in non-rice stubble fields relative to abundance of green vegetation, Snow Goose cohort size, Canada Goose cohort size, distance from cover, and interactions between Canada Goose cohort size and distance from cover, Canada Goose cohort size and Snow Goose cohort size, Canada Goose cohort size and green vegetation, Snow Goose cohort size and green vegetation, Snow Goose cohort size and distance from cover (df = 68, &2 = 0.64, :e <.001) Multiple regression model of Canada Goose cohort size for flocks in nonrice stubble fields relative to abundance of green vegetation, Snow Goose cohort size, white-fronted goose cohort size, distance from cover, and interactions between white-fronted goose cohort size and distance from cover, white-fronted cohort size and Snow Goose cohort size, and between white-fronted goose cohort size and green vegetation (df = 44, &2 = 0.75, <.001) Energetic expenditures of behaviors expressed as multiples of the basal metabolic rate Esophageal and proventriculi contents (% dry weight excluding feathers and masticated green vegetation) of Snow Geese, Greater White-fronted Geese, and Canada Geese in rice stubble and non-rice stubble fields in southeastern Texas Diet overlap and diet breadth (n) for Snow Geese, Greater White-fronted Geese, and Canada Geese collected in rice stubble and non-rice stubble fields in southeastern Texas Estimated energy expenditure (kj/h) for Snow Geese, Greater White-fronted Geese, and Canada Geese foraging in rice stubble fields and non-rice stubble fields in southeastern Texas Non-significant multiple regression models of absolute energy expenditure (kj/h) for Snow Geese and Greater White-fronted Geese relative to abundance of rice grains, green vegetation, conspecifics, heterospecifics, and distance from cover for birds in rice stubble fields, and for Snow Geese, Greater White-fronted Geese, and Canada Geese relative to abundance of green vegetation, con specifics, heterospecifics, and distance from cover for birds in non-rice stubble fields

14 x Table LIST OF TABLES (Continued) Page 15 Percent time in various activities for Snow Geese, Greater White-fronted Geese, and Canada Geese in rice stubble fields and non-rice stubble fields in southeastern Texas Frequency of aggressive interactions with focal Snow Geese, Greater White-fronted Geese, and Canada Geese in southeastern Texas Multiple regression model of percent time spent on feeding locomotion for Snow Geese in rice stubble fields relative to abundance of rice grains, green vegetation, Snow Goose cohort size, white-fronted goose cohort size, Canada Goose cohort size, and distance from cover (df = 36,.&2 = 0.40, =.051) Multiple regression model of percent time spent on non-feeding locomotion for Snow Geese in rice stubble fields relative to abundance of rice grains, green vegetation, Snow Goose cohort size, white-fronted goose cohort size, Canada Goose cohort size, distance from cover, and interactions between Canada Goose cohort size and white-fronted goose cohort size, and between Canada Goose cohort size and distance from cover (df = 36,.&2 = 0.63, <.001) Multiple regression model of percent time spent on aggression for Snow Geese in rice stubble fields relative to abundance of rice grains, green vegetation, Snow Goose cohort size, white-fronted goose cohort size, Canada Goose cohort size, and distance from cover (df = 36,.&2 = 0.33, =.049) Multiple regression model of percent time spent on comfort activities for Snow Geese in rice stubble fields relative to abundance of rice grains, green vegetation, Snow Goose cohort size, white-fronted goose cohort size, Canada Goose cohort size, distance from cover, and the interaction between rice grain and Canada Goose cohort size (df = 36,.&2 = 0.78, <.001) Non-significant multiple regression models of percent time spent on various behaviors relative to abundance of rice grains, green vegetation, Snow Goose cohort size, white-fronted goose cohort size, Canada Goose cohort size, and distance from cover for Snow Geese in rice stubble fields (df = 36), and relative to green vegetation, Snow Goose cohort size, white-fronted goose cohort size, Canada Goose cohort size, and distance from cover for Snow Geese in non-rice stubble fields (df = 27)... 47

15 Xl Table LIST OF TABLES (Continued) Page 22 Multiple regression model of percent time spent feeding stationary for Greater White-fronted Geese in rice stubble fields relative to abundance of rice grains, green vegetation, white-fronted goose cohort size, Snow Goose cohort size, Canada Goose cohort size, and distance from cover (df = 66, &2 = 0.27,.e =.003) Multiple regression model of percent time spent feeding locomotion for Greater White-fronted Geese in rice stubble fields relative to abundance of rice grains, green vegetation, white-fronted goose cohort size, Snow Goose cohort size, Canada Goose cohort size, and distance from cover (df = 66, &2 = 0.29,.e =.002) Multiple regression model of percent time spent on aggressive behaviors for Greater White-fronted Geese in rice stubble fields relative to abundance of rice grains, green vegetation, white-fronted goose cohort size, Snow Goose cohort size, Canada Goose cohort size, distance from cover, and the interaction between rice grains and green vegetation (df = 66, &2 = 0.51, <.001) Multiple regression model of percent time spent on comfort activities relative to abundance of rice grains, green vegetation, white-fronted goose cohort size, Snow Goose cohort size, Canada Goose cohort size, and distance from cover for Greater White-fronted Geese in rice stubble fields (df = 66, &2 = 0.78.e <.001), and relative to abundance of green vegetation, white-fronted goose cohort size, Snow Goose cohort size, Canada Goose cohort size, distance from cover, and the interaction between white-fronted goose cohort size and distance from cover for Greater White-fronted Geese in non-rice stubble fields (df = 54, &2 = 0.44.e <.001) Non-significant multiple regression models of percent time spent on various behaviors relative to abundance of rice grains, green vegetation, white-fronted goose cohort size, Snow Goose cohort size, Canada Goose cohort size, and distance from cover for Greater White-fronted Geese in rice stubble fields (df = 66), and relative to green vegetation, white-fronted goose cohort size, Snow Goose cohort size, Canada Goose cohort size, and distance from cover for Greater White-fronted Geese in non-rice stubble fields (df = 54) Non-significant multiple regression models of percent time spent on various behaviors for Canada Geese in non-rice stubble fields relative to abundance of green vegetation, Canada Goose cohort size, white-fronted goose cohort size, Snow Goose cohort size, and distance from cover (df = 27)... 53

16 xii Figure LIST OF FIGURES Page 1. Weekly density (birds/ha) of (A) Snow Geese, (B) Greater White-fronted Geese, and (C) Canada Geese in southeastern Texas Weekly (A) abundance of rice grains, (B) abundance of green stems, and (C) hunting pressure in southeastern Texas

17 CHAPTER I INTRODUCTION Many species of arctic-nesting geese in North America and Europe have undergone a dramatic increase in population size (Cooch et al. 1989, Madsen 1991, Owen and Black 1991, Warren and Sutherland 1992), and have expanded their ranges to include non-traditional feeding sites (Bateman et al. 1988, Hobaugh et al. 1989, Perco 1991, Prokosch 1991) since at least the 1960's. Several goose populations have increased to the point where they are exhibiting density-dependent characteristics (Cooch et al. 1991, Ebbinge 1991, Owen and Black 1991). Recent evidence indicates that populations of Lesser Snow Geese (Chen caerulescens caerulescens) and Barnacle Geese (Branta leucopsis) now are limited by feeding constraints on the arctic breeding grounds ( Cooch et ai. 1991, Kalchreuter 1991, Owen and Black 1991). Owen and Black (1991) examined long-term population trends of geese in Europe and North America and hypothesized that historically, populations of most arcticbreeding geese were limited by food on the wintering grounds. However, with the advent of modem agricultural practices, winter food is no longer limiting (Unlimitedresource hypothesis; cf. Ebbinge 1991). Two corollaries emerge from this hypothesis: historically, the winter goose community was structured by processes resulting from resource limitations, and the community is currently structured by processes other than resource limitations. Format and style follow Ecology.

18 2 Selected terms used in this paper are defined as follows. Community: a group of individuals of several species that co-occur in time and space (Wiens 1989g). Density: number of individuals per unit area (Ralph 1981). Density dependency: annual variation in fecundity and survivorship is inversely related to population size (Cooch et al. 1991). Competition: an interaction between two or more individuals that, as a consequence either of exploitation of a shared resource or interference related to that resource, has a negative effect on fitness-related characteristics of at least one of the individuals (Wiens 1989b). Flock: greater than two individual geese in periodic visual contact, with every individual being within 50 m of another. Cohort size: number of conspecifics in a flock. If the unlimited-resource hypothesis is valid, patterns of goose abundance, distribution, and species co-occurrence were affected by multiple processes. When multiple processes are important, additional insight may be gained by conducting a study in a framework of several competing hypotheses (Wiens 1989g). Examining multiple hypotheses avoids the frequent problem of researchers interpreting data in the context of a single hypothesis and developing conclusions that substantiate their hypothesis, without considering other equally viable alternatives (Wiens 1989g, Q). This suggests that both resource-limited processes (e.g., competition) and resourceunlimited proc!:sses (e.g., predation, parasitism, and disturbance) should be considered simultaneously. The unlimited-resource hypothesis addresses population-level patterns; however, the proposed explanation for these patterns is based on individual-level processes (i.e., effects of food on individuals). These individual-level processes may result in predicted patterns at the population level; however, behavioral responses by individuals also may produce unexpected patterns at higher levels (Wiens 1989g).

19 3 Thus, determining the scale which provides the "correct" pattern may be dependent on the process of interest. Indeed the very notion that a system or process has a "correct" or inherent biological scale has generated considerable controversy (Carlile et al. 1989). Developing theory on this question now suggest that observed patterns are at least partly a function of the scale at which they are viewed (Wiens 1989~, Levin 1992), therefore it is particularly informative to conduct an investigation in a hierarchial framework which includes analyses at several levels (O'Neill et al. 1986, Ives et al. 1993). In this study, I selected three scales of organization (e.g., population, flock, individual) to examine patterns of goose species abundance, behavior, and cooccurrence. Each level of organization reflects the sampling unit of the response variables. Population-level variables are measured with a spatial grain of several kilometers (e.g., density of conspecifics, regional resource abundance), whereas flock variables are measured with a spatial grain of several meters (e.g., flock size, and spatial arrangement of flocks relative to environmental features within fields). Individual-level variables are measured with the same spatial grain as flocks because geese can respond to their environment at that scale (e.g., Black et al. 1992). However, I consider these variables to describe events at the individual level because they relate directly to behaviors of individual geese. The broad objective of my study is to test the unlimited-resource hypothesis by assessing the effects of interspecific competition, intraspecific competition, and predation, on community patterns of geese at three hierarchial levels. The general approach is to examine patterns of species co-occurrence and abundance at the population and flock levels, habitat use at the flock level, and to estimate the relative energetic costs of foraging as well as behavioral differences at the individual level.

20 4 At each level, three specific statistical null hypotheses can be tested: no effect from interspecific competition, no effect from intraspecific corppetition, and no effect from predation. These hypotheses lead to predictions regarding the relationship between variables selected for study (Table 1). Because this study is not experimental, however, consistency with predictions does not ensure that causative processes can be identified. Observed patterns consistent with predictions from a process merely support the possibility of causation. Further, results are interpreted only with respect to the three processes identified a priori. Other ad hoc explanations are undoubtedly possible but cannot be addressed here. With these cautionary remarks in mind, the a priori predictions between patterns and processes should be viewed as a framework for interpreting observational data, that allows for stronger inference than a posteriori explanations alone. Interspecific comootition Interspecific competition has been the most widely proposed explanation of community patterns; however, the results have often proven ambiguous because of the vagueness of what competition was assumed to represent (Wiens 1989h). The widespread focus on competition in the face of ambiguities has led to a more critical evaluation of competition as a dominant process (e.g., Connor and Simberloff 1979), and the development of criteria for documenting the occurrence of competition (Wiens 1989l!). Thus, to determine the effects of interspecific competition, it is necessary to: document patterns of species co-occurrence that are consistent with predictions based

21 5 TABLE 1. Predicted effects of competition and predation on community patterns of geese. Cell values indicate the predicted correlafion betw~en the variables selected for study as positive (+). negative (-). or absent (0). Pattern Process Study level Independent Dependent Interspecific Intraspecific Predation variable variable competition competition Population Food abundance Number conspecifics Predation pressure Number conspecifics 0 0 Number heterospecifics Number conspecifics 0 0 Flock Food abundance Number conspecifics Predation risk Number conspecifics Number heterospecifics Number conspecifics 0 + Individual Food abundance E expenditure 0 Predation risk E expenditure Number heterospecifics E expenditure + 0 Number conspecifics E expenditure on a competition hypothesis, show that resources are shared, and that a negative effect on a fitness-related characteristic for at least one species results from either direct sharing of resources or interference. At the population and flock levels, I look for the effects of interspecific competition by quantifying patterns of abundance for conspecifics and heterospecifics. At the flock level, I examine patterns of habitat use. At the individual level, esophageal contents are examined to determine resource overlap and niche breadth. Also, I estimated energetic costs of foraging (e.g., Black et al. 1992) and behavioral changes relative to the presence of heterospecifics.

22 6 Intraspecific competition Temporal or spatial changes in resource levels may produce corresponding changes in species' abundances or niche patterns independent of any effects from interactions with other species (Wiens ). In such cases, interspecific competition is less important, but intraspecific competition for resources may be intense and become an important process in determining the community pattern (Abrams 1980). To document intraspecific competition, it is necessary to show that: a population responds in a density-dependent fashion to changing resource levels independent of interspecific effects, and individuals experience a reduction in some fitness related component. In my study, I assess the effects of intraspecific competition by examining whether the density of con specifics at the population and flock levels is dependent on resource levels and heterospecifics. At the individual level, I determine if energetic expenditure and behavior are related to levels of resource abundance and conspecifics. Predation Traditionally, predation has been viewed as a process of direct mortality that affects abundance and population dynamics of prey species (Brown 1992). Although rates of predation depend on a wide variety of factors, density of a predator population can be an important determinant of prey population density (Wiens 1989b). For many species of geese, the greatest cause of direct mortality is hunting (Boyd 1957, Owen 1980, Francis et al. 1992). Thus, if predation has an effect on a goose population, it can be expected to correspond to the density of the dominant

23 7 predator population, which in the case of this study is the human-hunter population. As an evolutionary force, hunting may act through natural selection as do other forms of predation. Although speculative, it is possible that geese which possess traits making them vulnerable to hunting do not reproduce and perpetuate those traits as do geese which are less vulnerable to hunting mortality. Anecdotal evidence suggests even in as short a time period as 30 years, geese have become less vulnerable to certain hunting techniques. For example, in the 1960's when geese first started using the rice-prairie region of Texas in large numbers, hunters could regularly lure them into shooting range using sheets of scattered newspaper as decoys (Justin Hurst, hunting guide, personal communication). Since that time geese have become less vulnerable to such practices, and hunters now must resort to cloth or plastic decoys that more closely resemble feeding geese. Hunting also resembles other forms of predation from a more proximate perspective. For example, personal observation indicates that when Bald Eagles (Haliaeetus leucocephalus) attack geese, it disrupts the entire flock as birds change flight path or take flight from the ground. Likewise, hunters shooting at geese cause them to quickly change flight path and birds feeding nearby also may take flight. Thus, as with other forms of predation, hunting causes considerable disturbance to individuals not directly removed from a population (Madsen 1988, Ebbinge 1991). Here, I use density of hunters to examine effects of predation pressure on density of geese at the population level. Although previous studies have examined the response of geese to presence and absence of hunters (Hobaugh 1982, Ely 1992), I am not aware of any studies that have looked at abundance patterns of geese in response to different levels of hunting pressure. Non-lethal effects of predation can strongly influence behavior of individuals and

24 8 their patterns of species co-occurrence (Barnard and Thompson 1985, Brown 1992). Correspondingly, the predictability of foraging models improved when predation risk was incorporated (e.g., Waite and Grubb 1987, Poysa 1991, Lima 1992). Thus, at individual and flock levels, risk associated with a specific foraging site may be as, or more important, than predator population density. Indeed, predation risk for social birds inhabiting open environments is inversely related to a flock's distance from cover (Lima 1993). In my study, I view distance to cover as a measure of individual predation risk at the flock and individual levels.

25 9 CHAPTER II STUDY AREA AND METHODS This study was conducted from November 1991 to February 1992 and October 1992 to February 1993 southwest of Houston, Texas. This area, known as the riceprairie region of Texas, lies inland from the coastal marshes, and extends from Port Lavaca, Texas, eastward to the Louisiana border (Hobaugh et al. 1989). Major land uses are grazing and cultivated croplands, with domestic rice (Oryza sativa) being the primary crop (United States Department of Commerce 1989). Rice is cultivated most commonly on a 3-year rotation such that during the winter, fields are classified as plowed soil, rice stubble, or annual plants. Plowed soil represents fields in preparation for planting; rice stubble represents harvested rice; and annual plants represent rice fields fallowed for at least one year. The goose community occurring in the study area consists of Lesser Snow Geese, Greater White-fronted Geese (Anser albifrons), Canada Geese (Branta canadensis), and Ross' Geese (Chen rossii) (Hobaugh 1982, Haskins 1993). Ross' Geese, however, comprise less than 7% of the total goose population (Harpole et al. unpublished data). For this reason, and because of the difficulty in discriminating visually at a distance between Ross' and Snow Geese, the two species are treated as one. Four groups of fields, two each in Colorado and Wharton counties, were selected as sample areas based on landowner permission to access contiguous blocks of property. Each sample area was approximately 2,000 ha and contained the three common fields (Le., plowed soil, rice stubble, and annual plants). Total hectares of

26 10 each field type in each sample area was based on aerial photos and calculated using the loris! geographic information system (Eastman 1992). To minimize observer disturbance to geese, a survey route was established such that fields could be viewed with a spotting scope and binoculars from a vehicle. To minimize conflicts with large numbers of hunters on weekends, data were collected only on weekdays. Surveys were conducted on three to four days per week except during days of moderate or heavy precipitation. Surveys were conducted each week of the study excluding one week in December To reduce biases in selecting daily-sample areas, the sequence of visitation to each sample area was chosen randomly. Each area was traversed from the pre-established route starting in early morning and usually ending by early afternoon. The direction of travel was reversed on each subsequent visit to reduce visual and temporal biases (Gawlik and Bildstein 1990).

27 11 CHAPTER III POPULATION-LEVEL PATTERNS Introduction Owen and Black (1991) examined long-term population trends of geese in Europe and North America and hypothesized that historically, populations of most arcticbreeding geese were limited by food on the wintering grounds. However, with the advent of modem agricultural practices, winter food is no longer limiting (Unlimitedresource hypothesis; cf. Ebbinge 1991). Thus, geese were released from their historic population constraint. Two corollaries emerge from this hypothesis: historically, the winter goose community was structured by processes resulting from resource limitations, and the community is currently structured by processes not resulting from resource limitations. Wiens (1989b) suggested both competition and predation may playa role in structuring communities; however, conditions under which either process is likely to be important are dependent on a wide range of environmental factors. At the population level, I examine patterns of temporal niche width and niche overlap for evidence of interspecific competition (Shoener 1974, DuBowy 1988, Wiens ). Also, I examine temporal patterns of goose abundance relative to food abundance, predation pressure, and density of heterospecifics to assess the effects of interspecific competition, intraspecific competition, and predation. Variables are measured on a spatial scale of several kilometers. My null hypotheses are there is no effect on the population densities of geese from each of the three

28 12 processes under consideration. Methods Individual Snow Geese exhibit site fidelity to the broad region of the upper Texas coast among years (Jennings 1990); however, there is no evidence of regional site fidelity within years (Robertson 1991). Daily movements of individual Snow Geese occur at a scale of the entire upper Texas coast, and therefore probability of resighting individuals is directly related to proportion of population sampled. Because I sampled a relatively small portion of the estimated 800,000 Snow Geese on the upper Texas coast (estimates from Haskins 1993), I assumed data collected during daily goose surveys to be independent across days, and all data within a survey were pooled. All goose flocks seen along survey routes were examined to determine species composition and relative abundance. Thus, each survey provided a single datum on total number of each species in the sample area at a given time. To calculate a measure of goose density, total number of each species were divided by the total area (ha) of searched fields. To quantify how geese co-occur on a seasonal basis, measures of niche breadth were calculated using Levins' standardized measure with weeks as resource categories (Krebs 1989). Levins' standardized index ranges from 0 to 1 with small values indicating smaller temporal breadth and large values indicating greater temporal breadth. Also, I calculated the degree to which species overlap in time using the percentage overlap measure (Krebs 1989). Percentage overlap ranges from 0 to 100 with large values indicating a high degree of overlap. Although there is considerable controversy over which measure of niche overlap is best, Abrams (1980) recommends percentage overlap as the best overall

29 13 measure, partly because it is insensitive to how investigators perceive and categorize resources. Resource levels were monitored by sampling food immediately following each daily-goose survey, at three 20-m radius plots chosen randomly and stratified by field type (one plot per field type). Each plot was sub sampled at the center, and four subplots at 90 0 intervals, 20 m from the center. The direction of the first subplot relative to the center was random and each subsequent subplot was 90 0 from the one preceding it. At each subplot, the frequency of food items (i.e., rice grains and leaf-bearing stems of green vegetation) that occurred within a 28 x 28 cm quadrat were recorded. Subsamples within a plot were not independent; therefore to avoid pseudoreplication (Hurlbert 1984), subsamples were pooled to obtain a mean abundance of food items per quadrat. Because relatively few samples were taken per week, daily samples were averaged within weeks and the subsequent weekly-mean value was used as a measure of resource abundance. A predation index was calculated as the mean number of hunting parties/day/station that harvested at least one goose. This number was obtained from records of two commercial goose-cleaning stations, one each located in Colorado and Wharton counties. To account for a possible residual effect from heavy hunting on weekends, the predation index was averaged across days within weekly periods. To test simultaneously for an effect from each hypothesized process, I constructed full and reduced multiple regression models (Freund and Wilson 1993) for each goose species with goose density as the dependent variable. When visual inspection of residual plots indicated variances were distributed non-normally, data were transformed using natural logs (Freund and Wilson 1993). Regression analyses were conducted with REG and GLM procedures (SAS Institute Inc. 1988), using a critical

30 14 level of.05. Full models contained a term for abundance of rice grains, green vegetation, heterospecifics (one each), hunter pressure, and interactions deemed possible based on literature and personal observation. A reduced model was subsequently constructed that contained the four main effects and any significant interactions. I viewed a significant coefficient as evidence that the corresponding process had an effect after accounting for the effects of all other variables in the model (e.g., Crowell and Pimm 1976, Hallett and Pimm 1979, Rosenzweig et al. 1984). Because interactions were not specified a priori during the study design, subsequent analyses should be considered exploratory and,e-values may be liberal. Throughout the study, I conducted 100 goose surveys with corresponding measures of food and hunting pressure. White-fronted geese were the first to arrive in October followed by Snow Geese and Canada Geese, respectively (Fig. 1). Population densities of geese fluctuated in a species-specific manner throughout the winter season. White-fronted Geese were most abundant early in the season, and then decreased slowly until the end of the season (Niche breadth = 0.482). Canada Geese exhibited a complementary pattern whereby they gradually increased in numbers to their highest levels at the end of the season (Niche breadth = 0.488). Thus, white-fronted and Canada geese occurred in the study site for approximately the same length of time, but each species was most abundant when the other was least abundant (Overlap = 43%). In contrast, mean densities of Snow Geese were similar but highly variable throughout the season. Also, Snow Geese occurred in the study site for a longer period of time (Niche breadth = 0.676) than white-fronted

31 <Ll (f) 2.0 <Ll <Ll ( A ~ 0 c (f) <Ll (f) 5 <Ll <Ll 4 (.9 "0 <Ll 3 +J C 0 I... 2 " Ī B <Ll -'-' ~ o C ~ 2.0 <Ll (f) ~ 1.5 (.9 o 1.0 "0 o C o 0.5 U 0.0 -'--..,----""---..., , Oct 19 Nov 17 Dec 14 Jan 11 Feb Date Fig. 1. Weekly density (birds/ha) of (A) Snow Geese, (B) Greater White-fronted Geese, and (C) Canada Geese in southeastern Texas. Bars indicate the mean and 1 SE.

32 16 and Canada geese. Correspondingly, Snow Geese exhibited a higher degree of temporal overlap with both white-fronted geese (67%) and Canada Geese (57%). Resource levels varied throughout winter as did hunting pressure (Fig. 2). Rice grains were most abundant following harvest in October, and subsequently decreased until December or January when no longer present. In contrast, green vegetation increased in abundance throughout the season. Hunting pressure was nearly constant until February when it decreased. Averaged over the entire season, Snow Geese were the most abundant species, with a mean density of 1.5 birds/ha. A significant negative interaction between abundance of white-fronted geese and rice indicated the relationship between Snow Geese and white-fronted geese was positive when rice was scarce and negative when rice was abundant (Table 2). The positive interaction between rice and Canada Geese indicated the positive relationship between Snow and Canada Geese was strongest when rice was abundant. Also, the positive interaction between hunting pressure and white-fronted goose density suggested the positive relationship between snow and white-fronted geese was strongest when hunting pressure was low. Similarly, the negative interaction between Canada goose density and hunting pressure indicated the positive relationship between Snow and Canada Geese was strongest when hunting pressure was low. A negative interaction between whitefronted geese and Canada Geese indicated the positive relationship between Snow Geese and white-fronted geese was strongest when Canada Geese were abundant. Also, this interaction indicated the relationship between Snow Geese and Canada Geese was positive when white-fronted geese were scarce and negative when whitefronted geese were abundant.

33 17 N 50 ~ ~ A E 45 u 40 ~ 35 ~ 30 (f) 25 c o 20 I I!.l u 10 0::: 5 o ~~~~~~~~~~~~------~--~ N ~~------~ ~ 140 ' 120 OJ ~ 100 E 80..., I!.l 60 (f) C I!.l I!.l I (!) o ~~~~~~~~~~~~~~~~~~ 16 --r _ ~ 14 - C \J 12 - (f) "'" Q.. :::l o I I... I!.l..., C :::l I o I 22 Oct Jlq,-----,II 19 Nov 17 Dec 14 Jan 11 Feb Date Fig. 2. Weekly (A) abundance of rice grains, (B) abundance of green stems, and (C) hunting pressure in southeastern Texas. Bars indicate the mean and 1 SE.

34 18 TABLE 2. Multiple regression model of Snow Goose population density relative to abundance of rice grains, green vegetation, white-fronted goose density, Canada Goose density, hunting pressure, and interactions between rice grains and white-fronted goose density, rice grains and Canada Goose density, hunting pressure and white-fronted goose density, hunting pressure and Canada Goose density, and between white-fronted goose density and Canada Goose density (df = 99, R2 = 0.40, f <.001). Parameter Estimate SE P Rice grain Green vegetation White-fronted goose Canada Goose Hunting pressure Rice grain white-fronted goose Rice grain Canada Goose <.001 Hunting pressure white-fronted goose Hunting pressure Canada Goose <.001 White-fronted goose Canada Goose White-fronted geese were the second most abundant species in the sample areas, with a mean density of 0.8 birds/ha. Density of white-fronted geese was positively related to both rice grains and density of Snow Geese (Table 3). Canada Geese averaged 0.3 birds/ha. Density of Canada geese was negatively related to rice grains, green vegetation, and density of white-fronted geese (Table 4). The interaction between green vegetation and white-fronted geese indicated the negative relationship between Canada Geese and green vegetation was weaker when white-fronted geese were abundant. Also, the negative relationship between Canada and white-fronted geese was greater when green vegetation was scarce. The interaction between snow and white-fronted geese indicated the negative relationship between Canada and white-fronted geese was greater when Snow Geese were

35 19 TABLE 3. Multiple regression model of Greater White-fronted Goose population density relative to abundance of rice grains, green vegetation, Snow Goose density, Canada Goose density, and hunting pressure (df = 99, R2 = 0.20, <.001). Parameter Estimate SE P Rice grain Green vegetation Snow Goose Canada Goose Hunting pressure TABLE 4. Multiple regression model of Canada Goose population density (log) relative to abundance of rice grains (log), green vegetation (log), Snow Goose density, white-fronted goose density, hunting pressure, and interactions between Snow goose and hunting pressure, Snow Goose and white-fronted goose, and between green vegetation (log) and white-fronted goose (df = 99, R2 = 0.44, P <.001). Parameter Estimate SE P Rice grain (log) <.001 Green vegewtion (log) Snow Goose <.001 White-fronted goose Hunting pressure Snow Goose hunting pressure <.001 Snow Goose white-fronted goose Dl5.029 White-fronted goose green vegetation (log)

36 20 abundant. Density of Canada geese was positively related to density of Snow Geese. The negative interactions between Snow Geese and hunting pressure, and Snow Geese and white-fronted geese, indicated the positive relationship between Canada Geese and Snow Geese was strongest when hunting pressure and density of white-fronted geese were low. Discussion Taken alone, the positive relationships of Snow Goose densities and white-fronted goose densities to heterospecifics and food, suggests that both species may be responding to similar processes. One explanation for this pattern is that geese are opportunistic species that track fluctuating resources and may not be indicative of any type of competitive interaction (Wiens 1989a). In other words, Snow Geese and white-fronted geese may both be exploiting abundant waste rice grain following harvest (Hobaugh 1984). For white-fronted geese, the strong positive relationship with rice abundance in the absence of a negative interspecific effect is consistent with the intraspecific competition hypothesis because it suggests these birds enter and leave the region in response to rice abundance. Considering main effects in light of significant interactions, however, provides more insight into the way multiple factors interact. For Snow Geese, both rice grains and hunting pressure affected goose density through interactions with heterospecifics. Rice grains and hunting pressure had the opposite effect on the relationship between Snow Geese and white-fronted geese, and Snow Geese and Canada Geese. Canada Geese were most closely associated with Snow Geese when

37 21 rice was abundant, whereas white-fronted geese were negatively associated with Snow Geese when rice was abundant. The negative relationship between snow and white-fronted geese when rice was abundant is consistent with a competitive interaction. Both species consume rice grains (Alisauskas et ai. 1988, Ely 1992) and the expansion of these two species into their current winter range coincided closely with the developing rice agriculture industry (Hobaugh et ai. 1989, Robertson and Slack 1994). Nevertheless, the lack of a strong negative interspecific interaction suggests that interspecific competition between snow and white-fronted geese and snow and Canada geese does not have an overwhelming effect on Snow Goose populations. Likewise, the lack of a positive main effect from either rice or green vegetation suggests that intraspecific competition may not be a strong process affecting Snow Goose populations. The negative relationship between Canada Geese and white-fronted goose population density was particularly strong when green vegetation was scarce and Snow Geese were abundant. Also, the negative relationship between Canada Goose density and green vegetation was strongest when white-fronted geese were scarce. This interaction makes direct interpretation of the negative effect from green vegetation difficult because it is possible that Canada Geese were responding more strongly to white-fronted geese than green vegetation. Overlap indices suggest that Canada Geese reduce temporal overlap with white-fronted geese by increasing in density later in the season when white-fronted geese are scarce. These patterns are most consistent with the interspecific-competition hypothesis which predicts as food increases in abundance, two competitors should show a greater degree of separation, in this case temporal. Superficially, the negative relationship between Canada Geese and rice appears to confound this hypothesis because is appears population density is