Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Components of hunting mortality in ducks : a management analysis Hochbaum, George Sutton 1980

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1980_A1 H63.pdf [ 4.54MB ]
Metadata
JSON: 831-1.0095027.json
JSON-LD: 831-1.0095027-ld.json
RDF/XML (Pretty): 831-1.0095027-rdf.xml
RDF/JSON: 831-1.0095027-rdf.json
Turtle: 831-1.0095027-turtle.txt
N-Triples: 831-1.0095027-rdf-ntriples.txt
Original Record: 831-1.0095027-source.json
Full Text
831-1.0095027-fulltext.txt
Citation
831-1.0095027.ris

Full Text

COMPONENTS OF HUNTING MORTALITY IN DUCKS: A MANAGEMENT ANALYSIS GEORGE SUTTON HOCHBAUM by B . S c , Colorado State University, 1969 M . F . S c , Yale University, 1971 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY THE FACULTY OF GRADUATE STUDIES We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA Department of Zoology March 1 9 8 0 George Sutton Hochbaum, I 9 8 O In presenting th is thes is in p a r t i a l fu l f i lment of the requirements for an advanced degree at the Un ivers i ty of B r i t i s h Columbia, I agree that the L ibrary sha l l make it f ree ly ava i l ab le for reference and study. I fur ther agree that permission for extensive copying of th is thes is for scho la r ly purposes may be granted by the Head of my Department or by his representat ives . It is understood that copying or pub l ica t ion of th is thes is for f inanc ia l gain sha l l not be allowed without my writ ten permission. Department of The Univers i ty of B r i t i s h Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 Date i i ABSTRACT This study was conducted to evaluate waterfowl harvests on the Delta Marsh, Manitoba, under a variety of options u t i l i z ing a modified predator-prey model. The study involved monitoring of ducks and hunters to examine key components of waterfowl mortality. Seven hypotheses were formulated involving the relationships between duck vulnerability by species and hunter effort , factors determining effective bag s ize, hunter effort , effects of weather on k i l l , and whether or not hunters are selective. Aerial surveys and experimental hunts were conducted and k i l l s tat is t ics gathered on the Pasquia and Delta Marshes in Manitoba to evaluate predictions from the hypotheses. The analysis of f ie ld data revealed that duck vulnerabil ity is negatively proportional to hunting effort and that ducks are most wary when hunter effort is greatest. Hunters were afield in largest numbers during periods of high duck density. Hunters did not shoot selectively and crippling losses approximated one-third of the birds shot. Weather had l i t t l e influence on the k i l l ; bird population size and hunter effort were more important determinants of hunter success. Distribution patterns of ducks were clumped relative to hunting areas, and the probability of mortality for an individual duck was observed to decrease with increasing flock s ize . Handling time per bird downed did not l imit achieved bag size whereas time between encounters was found to be independent of population size and resulted in a non-linear, k i l l -densi ty relationship. Field results were incorporated into the predator-prey model and management schemes involving population s ize , hunter effort , and season length were evaluated. Simulation results revealed that Mallard and Lesser i i i Scaup harvest are not greatly affected by increased hunter effort and that season length and population size strongly influence harvest. Shortened seasons are recommended i f k i l l reductions are desired for Mallard. Lesser Scaup require no special regulations whereas Canvasbacks require short seasons (less than 2 weeks) during times when populations exceed 5000. The predator-prey model is recommended for in-depth analysis of local regulations whereas multi-variate s ta t i s t i ca l models may be more useful in forecasting yields on a regional level . i v TABLE OF CONTENTS PAGE ABSTRACT 1 1 TABLE OF CONTENTS i v LIST OF TABLES v i LIST OF FIGURES v i ' ACKNOWLEDGEMENTS x CHAPTER ONE: INTRODUCTION AND CONCEPTUAL MODEL 1 Interactions and Components 3 Hypotheses ^ CHAPTER TWO: STUDY AREAS, HUNTING REGULATIONS AND METHODS 1 1 The Del ta Marsh • 1 1 The Pasquia Marsh ^ Decoy Tr ia l s ^ Aerial Surveys ^ Spy Blinds and Hunter Effort 1 9 CHAPTER THREE: FIELD RESULTS 2 2 Vulnerab i1i ty 2 2 Hunter Selectivity and Observed K i l l ^ Handli ng Time Relation of Flock Size to Probability of Mortality ^ Density-Kill Relationships 53 Analysis of Hunter Bag Data 5^  Allocation of Effort 59 Crippling Losses CHAPTER FOUR: DISCUSSION OF FIELD RESULTS 65 V TABLE OF CONTENTS cont'd. PAGE CHAPTER FIVE: THE MATHEMATICAL SINGLE HUNTER MODEL 75 Background to the Model 75 Quantifying the Model 7 6 CHAPTER SIX: MANAGEMENT ANALYSIS AND IMPLICATIONS OF FINDINGS 81 Applying the Model to the Delta Marsh 81 Additional Assumptions and Structure 81 Management Options 82 Results and Discussion 83 Management Implications of Findings 9 6 SUMMARY AND CONCLUSIONS 101 BIBLIOGRAPHY 103 v i LIST OF TABLES PAGE Table I Results of 36.8 hours of decoy tr ia l s during the 1973 f ie ld season 2 8 Table II Results of 45.6 hours of decoy tr ia l s during the 1974 f ie ld season 29 Table III Distribution of ducks relative to refuges on the Delta Marsh as measured by aerial surveys during 1973 3 9 Table IV Summary of F ratios generated by regression analysis of vulnerabil ity as related to selected weather variables (df = 32, F = 4.15 for p<.05) **1 Table V Hunter select ivity by species as determined by observation on the Delta Marsh during 1 9 7 3 , 1 9 7 ^ and 1 9 7 7 ^ 3 Table VI Observed k i l l , k i l l per hunter hour, estimated effort and population size for the Delta Marsh in 1 9 7 3 and 1 9 7 ^ . . . ^ Table VII Hunter success on the Delta Marsh expressed as ducks per hunter as determined by hunter bag checks during the period 1973 through 1977 **5 Table VIII Probability of mortality in relation to range from hunters' blinds on the Delta Marsh ( 1 9 7 3 , 1 9 7 4 ) 50 Table IX Calculation of probability of mortality showing adjustment for flock sizes and hunters . . 51 Table X Mean flock sizes of ducks lured to decoys and probability of mortality based on flock size 52 Table XI Estimated k i l l , population size and hunter numbers at Delta for selected weeks during 1975, 1 9 7 6 and 1 9 7 7 6 0 Table XII Crippling losses in relation to range from hunters' blinds on the Delta Marsh ( 1 9 7 3 , 1 9 7 4 , 1 9 7 7 ) 6 Z * Table XIII Possible influences of independent variables on the k i l l of ducks as predicted by three models 69 vi i LIST OF FIGURES PAGE Figure 1 Conceptual model of duck-hunter interactions assumed to represent causal factors determining the k i l l for one hunter. Y=yes, N=no 5 Figure 2 Map of the Delta Marsh 12 Figure 3 Map of the Pasquia Marsh 15 Figure 4 Duck populations on the Delta Marsh during the fa l l s of 1 9 7 3 and 1 9 7 4 as determined by aerial surveys 23 Figure 5 Duck populations on the Delta Marsh during the fa l l s of 1 9 7 5 , 1 9 7 6 and 1 9 7 7 as determined by aerial surveys . . . 2 4 Figure 6 Estimated hunter hours during each week of hunting during 1 9 7 3 and 1 9 7 4 2 5 Figure 7 Estimated numbers of hunters during each week of hunting during 1 9 7 5 , 1 9 7 6 and 1 9 7 7 2 6 Figure 8 Relationship between Mai 1ard vulnerabi1 ity and cumulative hunter effort as determined during decoy tr ia l s at Delta Marsh during 1973 and 1 9 7 4 3 0 Figure 9 Relationship between Lesser Scaup vulnerabil ity and cumulative hunter effort as determined during decoy t r ia l s at Delta Marsh during 1 9 7 3 and 1 9 7 4 31 Figure 10 Relationship between Canvasback vulnerability and cumulative hunter effort as determined during decoy tr ia l s at Delta Marsh during 1 9 7 3 and 1 9 7 4 32 Figure 11 Relationship between Mallard vulnerability and instantaneous hunter effort as determined by decoy tr ia l s at Delta Marsh during 1 9 7 3 and 1 9 7 4 33 Figure 12 Relationship between Lesser Scaup vulnerabil ity and instantaneous hunter effort as determined by decoy tr ia l s at Delta Marsh during 1 9 7 3 and 1 9 7 4 34 Figure 13 Relationship between Canvasback vulnerabil ity and instantaneous hunter effort as determined by decoy t r i a l s at Delta Marsh during 1 9 7 3 and 1 9 7 4 35 Figure 14 Probability of mortality (ratio birds downed to birds fired at) in relation to flock size as determined from spy blinds on the Delta Marsh during 1 9 7 3 and 1 9 7 4 ^ v i i i LIST OF FIGURES cont'd. PAGE Figure 15 Relationship between probability of mortality and flock size as determined from spy blinds on the Delta Marsh during 1973 and 1974. Values adjusted for one hunter 49 Figure 16 Relationship between encounter rates and population size for Mallard as determined by decoy tr ia l s at Delta Marsh during 1973 and 1974 55 Figure 17 Relationship between encounter rates and population size for Lesser Scaup as determined by decoy t r ia l s at Delta Marsh during 1 9 7 3 and 1 9 7 4 56 Figure 18 Relationship between encounter rates and population size for Canvasback as determined by decoy t r i a l s at Delta Marsh during 1973 and 1974 57 Figure 19 Relationship between mean time between encounters and duck abundance during 1973, 1974 and 1 9 7 7 on the Del ta Marsh 58 Figure 20 Relationship between hunter hours and population size on the Delta Marsh during 1973 and 1 9 7 4 62 Figure 21 Relationship between hunter numbers and population size on the Delta Marsh during 1975, 1976 and 1977 6 3 Figure 22 Functional relationship between potential k i l l per hunter hour and duck density 77 Figure 23 Isopleth diagram showing the relationship between regions of Mallard k i l l and population size and season length under normal effort levels observed at Delta 8 5 Figure 24 Isopleth diagram showing the relationship between regions of Mallard k i l l and population size and season length under low effort levels 86 Figure 25 Isopleth diagram showing the relationship between regions of Mallard k i l l and population size and season length under high effort levels 87 Figure 26 Isopleth diagram showing the relationship between regions of Lesser Scaup k i l l and population size and season length under normal effort levels observed at Delta 89 ix LIST OF FIGURES cont'd. PAGE Figure 27 Isopleth diagram showing the relationship between regions of Lesser Scaup k i l l and population size and season length under low effort levels 90 Figure 28 Isopleth diagram showing the relationship between regions of Lesser Scaup k i l l and population size and season length under high effort levels 91 Figure 29 Isopleth diagram showing the relationship between regions of Canvasback k i l l and population size and season length under normal effort levels observed at Delta 93 Figure 30 Isopleth diagram showing the relationship between regions of Canvasback k i l l and population size and season length under low effort levels 94 Figure 31 Isopleth diagram showing the relationship between regions of Canvasback k i l l and population size and season length under high effort levels. 95 X ACKNOWLEDGEMENTS Many people contributed to this project and I am grateful to a l l of those who were involved. I am particularly indebted to my supervisor, Dr. Carl Walters, who saw the program through some bleak moments and to Dr. Conrad Wehrahan whose encouragement during my stay in Vancouver made some of the not-so-pleasant times bearable. I would like to thank Dr. Douglas Stephen of the Canadian Wildl i fe Service for part ia l ly funding the study. Kent Brace and Dr. James Patterson, also of the Canadian Wildl i fe Service, require special mention since they allowed me time to continue the project after I left Vancouver. Dale Caswell, of the Canadian Wildl i fe Service, often gave me valuable s tat i s t ica l and programming advise. I wish to thank Dr. Robert Jones, Dr. Bruce Batt and Peter Ward of the Delta Waterfowl Research Station for financial support and for their advice and encouragement. Monies were also contributed by the Canadian National Sportsman's Show and I am also grateful to this organization. Special thanks go to the Bell Estate for permitting me to use their landings, boats and decoys and to the many guides of St. Ambroise whose advice I frequently required. Louis Ducharme Sr . , Chief Guide of the Bell Estate, deserves special mention for the many thoughts he shared with me on the natural history of ducks and sport hunting in general. I am also indebted to my three assistants, Richard Wishart, Kevin Ward and Valerie Hochbaum, for their endurance and help especially on those mornings in October when the wind was up and the temperature below freezing. I would like to thank my mother, Joan, and my father, Albert, for their never-ending advice, encouragement and financial and moral support. xi Final ly , I am especially indebted to my wife, Valerie , who had to put up with me te l l ing her to "keep paddlin" during her time as my assistant. Her presence in the duck blind made many cold hours pass by quickly and pleasantly and her never-ending interest for the study often times kept it al ive in its infancy. To a l l those members of the duck hunting fraternity at Delta and to the ducks who made the research possible, I thank you. 1 CHAPTER ONE INTRODUCTION AND CONCEPTUAL MODEL Human exploitation of animals is a highly selective, intensive form of predation usually directed at species of economic or recreational value. In this a r t i f i c i a l predator-prey system, man attempts to manipulate game species, their environments and the degree of human pressure toward the maintenance of sustained yie lds . Despite our knowledge of the predation process in animal populations, few attempts have been made to apply this knowledge to man as a predator on the populations he harvests. Holling ( 1 9 5 9 a ; 1 9 6 6 ) in his pioneering work on the functional response of predators to prey densities, developed models to describe the capture rate; Walters et a l . ( 1 9 7 3 ) have applied these equations to ducks. Based on simple f ie ld studies in southwestern Bri t i sh Columbia, Walters.and his colleagues recommended that, i f such models were to be useful in waterfowl management, further research was needed on the basic components (Holling, 1966) of hunting mortality, part icularly in relation to avoidance learning by birds. Several researchers have attempted to forecast long-term waterfowl yields using population dynamics models. Geis et a l . ( 1 9 6 9 ) employed a multiple regression model for estimating allowable harvests of Mallards in the United States. Walters et a l . (1974) incorporated numerous biological relationships into a continental Mallard model but found their forecasts of l i t t l e value because many variables and parameters could not be estimated. More recently, Anderson ( 1 9 7 5 ) and Brown et a l . ( 1 9 7 6 ) have investigated possible Mallard harvests for North America by u t i l i z ing gross habitat measures along with an extension of the Beverton and Holt ( 1 9 5 7 ) 2 stock-recruitment equation. Hochbaum and Caswell ( 1 9 7 8 ) simulated management programs for Mallards on a provincial basis in Western Canada and Cowardin and Johnson ( 1 9 7 9 ) evaluated similar alternatives for local areas. However, none of those models deal with the short-term (within hunting season) human and bird behavioral factors that determine what regulatory actions (such as season length, bag limit) should be employed so as to achieve the allowable long-term harvests. Further, they ignore the multi-species character of waterfowl harvesting; seldom i f ever can regulations be tailored to f i t a l l the bird species that hunters may pursue simultaneously. Throughout North America, ducks are harvested under regulations which assume that a l l species are equally vulnerable to gunning; policy is based primarily on stock size. However, most hunters are aware that the opportunity for taking ducks depends on many factors including species present, location and time of season. Such variations may be due not only to differences in abundance but also to specific avoidance behaviors learned by the birds (Walters et a l . , 1 9 7 3 ) . My objectives In this study were to determine how waterfowl behavior and human factors interact within any hunting season to affect mortality and to incorporate my f ie ld data into a multi-species predator-prey model and thereby explore the management implications of my findings. I intended to meet these objectives through an integrated field-modeling study whereby measurements could be taken of the interactions that occur in a single hunter, multi-prey species situation and the effectiveness of regulations determined through simulation modeling. The study involved three phases, beginning with the development of a conceptual hunting model in which components of mortality were broken down to a series of causal pathways; 3 this formulation led to specific hypotheses and predictions which could be tested in the f i e l d . These predictions led to the f ie ld phase of the study on ducks and hunters at two heavily hunted marshes in Manitoba. Investigations were carried out on the Delta Marsh in south-central Manitoba during the period 1 9 7 3 to 1 9 7 7 . In 1 9 7 4 studies were undertaken at Pasquia Marsh in north-central Manitoba. Both marshes have large populations of ducks during the fa l l and are favoured hunting areas, thus presenting ideal situations for studying bird-hunter relationships. Field studies were followed by the development of a within-season model that would provide a more general analysis of my findings and allow forecasting of the implications of duck hunting regulations on an annual basis. This report outlines a conceptual model and l i s t s hypotheses tested. Study areas and methods are described followed by a presentation of results and a discussion of f ie ld findings. A predator-prey model is then developed and applied to the Delta Marsh yielding a series of management scenarios. The paper is concluded by a discussion of management practices and a summarization of major findings. Interactions and Components An ideal study of ducks and hunters would involve a large scale, controlled experiment where harvests were measured for contrasting combinations of bird density and hunter effort . However, experiments of this scale are not yet considered practical by the agencies that regulate hunting areas. I had to deal with uncontrolled hunting effort, varying with bird abundance in patterns that did not naturally provide appropriate experimental contrasts. The only option in such cases is to attempt to reconstruct the overall relationship from a study of component interactions; k I therefore chose to f i r s t look at how a single hunter acts when pursuing bi rds. My research is based on a conceptual hunter-duck model in which the complexity of events is broken down into a series of causal pathways connecting components (Figure 1). According to this model in i t iat ion of the hunt depends on the number of ducks available, date, weather and various social influences. Once in the marsh, the hunter places his decoys, then hides nearby in his bl ind. When ducks approach the decoys, the hunter must decide whether or not to shoot. If he f i res , a bird may be k i l l ed and brought to bag or, only crippled, the quarry may try to escape. The hunter may then seek to k i l l and bag the cripple . When the hunter holds possession of the number of dead ducks allowed by law, the bag l imit , the hunt terminates. Of course, a hunter may lose interest for lack of birds or extreme cold and the hunt is finished. Within the model the rate of encounters during the hunt is used as a measure of the opportunity to k i l l ducks (Figure 1). Encounter rates reflect vulnerabil ity and may be affected by several variables: foraging and migration behavior, hunter effort , number of birds present, species composition, age structure of flocks, bird distribution in relation to hunters as well as by local weather and by the time of day and season. Several other components influence the outcome of each encounter. The hunter's decision to shoot may be related to species, size of the b ird , and his experience. Selectivity influences the size and composition of the daily bag. Few hunters k i l l ducks in every encounter, success depends on the accuracy of the gunner, range of the ducks, flock size and species. Crippling losses affect the k i l l because birds not retrieved are not included in the bag; handling time spent pursuing cripples is not ut i l ized 5 Figure 1. Conceptual model of duck-hunter interactions assumed to represent causal factors determining the k i l l for one hunter. Y=yes, N=no. 5a IN IT IATION Time of Season Duck Abundance W A I T I N G D I S G R U N T L E D N E N C O U N T E R B i o i i c N Y RETRIEVED M i g r a t o r y Pa t te rns S p e c i e s Present Foraging Behavior N u m b e r of Birds A g e St ructure Hunter Ef for t B i rd D i s t r i b u t i o n A b i o t i c W e a t h e r Time of Day and Season S H O O T S p e c i e s — — S i z e Hunter E x p e r i e n c e N T E R M I N A T I O N B A G C O M P L E T E A N CRIPPLE D i s t a n c e -BIRD(s) D O W N N Hunter E x p e r i e n c e Flock S i z e Spec ies 6 for shooting other ducks which pass by. Some hunters do not pursue cripples lest they lose opportunities to shoot at decoying or passing ducks. A key component that is identified in Figure 1 is the relationship between duck density and the rate of hunter-bird encounters, since this wil l ultimately determine how many birds wil l be fired upon and ki l led during times when the hunter is reactive. Many variables may influence the rate of encounter (Figure 1). Encounter rate per bird present depends on bird movement and reactive behavior to a r t i f i c i a l flocks represented by hunter's decoys. Thus, encounter rates wil l decrease if (1) birds move less or restr ic t movements to areas where decoys are not present, or (2) birds become more wary or less reactive to flocking opportunities. Such changes in bird behavior that decrease vulnerabil ity might be related to foraging requirements, migratory patterns, hunter effort , number and species of birds present, age structure, time and weather. Since pre-history, hunters have used decoys to exploit the flocking behavior of ducks. Waterfowl are gregarious except during the breeding season. Flocking fac i l i tates pair formation during the fa l l and winter (McKinney, 1975), enhances foraging efficiency (Bailey and Batt, 197*0, aids in predator detection (Powell, 197^, Seigfried and Underhill , 1975) and brings birds together to common shelter during wind and storm. Immature birds which have not previously visited the winter grounds join flocks of experienced adults in fa l l migration (Hochbaum, 1955). Hypotheses While Figure 1 identifies the basic components or functional relationships that could influence harvest, i t says l i t t l e about the quantitative nature or relative importance of each. Walters et a l . (1973) suggest for example, that observed k i l l patterns in southern Brit ish Columbia cannot be explained by duck abundance and handling time alone. They believe that avoidance learning has an important influence on hunter success, and that most hunting regulations have l i t t l e impact on the overall harvest. I attempted to identify major uncertainties and interactions suggested by the model and express these as a series of specific hypotheses to be tested in the f i e l d . Hypotheses selected for study were: T. Ducks learn to recognize the danger of hunters' blinds and decoys through experience. The prediction tested for this hypotheses is that vulnerabil ity (likelihood of duck encountering and being k i l l ed by a hunter) wi l l decline as cumulative hunting effort increases through a hunting season. Thus, at the end of the f a l l , birds are expected to be wary and to have modified their behavior so as to avoid either: (a) specific objects such as blinds and decoys, or (b) hunted areas in general. 2. The behaviors associated with wariness of hunters are species-specif ic . The prediction tested here was that avoidance of hunters would change in different ways for different species during the hunting season. 3. The rate of encounter per hunter is a random process related to population size, such that k i l l and encounters increase linearLy with duck abundance. A counter hypothesis is that bird distribution is highly clumped and hunters' search non-randomly so that they see the same number of ducks over a wide range of population sizes. k. Handling time per bird shot or crippled is a large enough component of each hunt to effectively l imit the daily bag except at very low 8 b i r d d e n s i t i e s . A l t e r n a t i v e l y , h a n d l i n g times a r e s m a l l enough so t h a t e i t h e r e n c o u n t e r r a t e s o r management bag l i m i t s a r e the key d e t e r m i n a n t s o f bag per h u n t e r . 5. Hunter e f f o r t i s d e t e r m i n e d by the numbers o f ducks p r e s e n t and time o f the h u n t i n g s e a s o n . I p r e d i c t e d t h a t e f f o r t would be g r e a t e s t e a r l y i n the season when ducks a r e most numerous and would d e c l i n e as l o c a l numbers were d i m i n i s h e d by m i g r a t i o n and by the ad v e n t o f w i n t e r . A c o u n t e r h y p o t h e s i s i s t h a t h u n t i n g p a t t e r n s a r e e s t a b l i s h e d t h r o u g h t r a d i t i o n and w i l l not respond q u i c k l y t o changes i n b i r d d e n s i t y . 6. V u l n e r a b i l i t y i s r e l a t e d t o l o c a l weather c o n d i t i o n s . Thus, I p r e d i c t e d t h a t a v o i d a n c e b e h a v i o r would be i n f l u e n c e d by p r e v a i l i n g w e a t h e r . For example, most h u n t e r s b e l i e v e t h a t duck h u n t i n g i s b e s t when i t i s o v e r c a s t and wi n d y . T h e r e f o r e , h i g h v u l n e r a b i l i t y m ight be r e l a t e d t o weather p a t t e r n s induced by changes i n b a r o m e t r i c p r e s s u r e , t e m p e r a t u r e , a i r movement and r e l a t i v e h u m i d i t y . An a l t e r n a t i v e h y p o t h e s i s i s t h a t k i l l p a t t e r n s a r e not i n f l u e n c e d by l o c a l w e a t h e r . 7. Hu n t e r s s h o o t d i s c r i m i n a t e l y and as a r e s u l t they f o r e g o the o p p o r t u n i t y t o k i l l some s p e c i e s o f d u c k s . A c o u n t e r h y p o t h e s i s i s t h a t h u n t e r s shoot a t a n y t h i n g t h a t f l i e s p a s t , t h e r e b y g e n e r a t i n g c o r r e l a t e d m o r t a l i t y i n a l l s p e c i e s . In a d d i t i o n t o t e s t i n g t h e s e seven h y p o t h e s e s , I hoped t o g a i n a b e t t e r u n d e r s t a n d i n g o f t h e i n f l u e n c e o f age on t h e k i l l and a l s o t o document c r i p p l i n g l o s s e s t h r o u g h d i r e c t o b s e r v a t i o n . S t u d i e s ( e . g . , B e l l r o s e e t a l . , 1961) have shown j u v e n i l e s t o be most v u l n e r a b l e t o gunning and Sowls (1955) 9 found hunters 1 reports of crippling losses to be curiously high on the Delta Marsh. The conceptual model and hypotheses outlined above require some precise definitions for the purpose of f ield study. The following terms are used repeatedly in the text. Avoidance learning: any alteration in the behavior of individual ducks which reduces the likelihood of being ki1 led by a hunter. Crippling loss: a measure of the number of ducks which are hit by gunfire, not immediately k i l l e d , and not retrieved. Birds in this category are often referred to as "sailers". The loss is calculated as the ratio of crippled birds to the number of birds downed by hunters. Decoy t r i a l s : observation periods during which the investigator places himself in an experimental hunting situation using a blind and decoys. The technique is employed to measure the degree of avoidance learning. Birds which respond to the decoys are negatively reinforced by gunfire. Encounter rate: the number of each species observed per hour decoying or passing within 35 m (40 yards) of a blind during a decoy t r i a l . Flare: an avoidance behavior e l i c i t ed when ducks detect a hunter. In f lar ing, the birds turn quickly in the air and fly away from the gunner thus presenting a d i f f i cu l t target. Handl i ng t i me: the amount of time a hunter spends for each encounter in act iv i t ies other than effective search and shooting. This includes time spent reloading the gun and retrieving downed birds. This time is calculated by subtracting mean time between encounters when birds are fired 10 upon (unsuccessfully) from mean time between successful encounters (birds killed) and the next encounter. Hunter effort: an index of the hours hunted per week. The index is calculated by dividing the weekly k i l l (as determined by hunter bag checks) by the number of ducks bagged per hunter hour during spy blind observations. Hunter encounter: duck or ducks passing within k i l l i n g range (35 m) of a hunter, whether shot at or not. K111ing range: the maximum distance at which ducks can be k i l l ed consistently with a 12 gauge shotgun. This distance is 35 m (Bell rose, 1953). I used this s ta t i s t i c in calculations of encounter rates and hunter select i vi ty. Maximum k i l l : the maximum number of ducks k i l l ed per hunter hour. Probability of mortality: the ratio of ducks downed to ducks fired upon. Select i vi ty: the percentage, by species, of ducks fired upon in hunter encounters: (shots at species X) / (encounters with species X) x 100. Spy blind: an observation post, arranged to simulate a hunting situation, from which the investigator observes nearby gunners. Vulnerability coefficient: an index of opportunity to k i l l ducks as measured by the mean weekly encounter rate per hour for a species divided by its estimated population size for that week. The coefficient represents the risk per hour of a bird in the population encountering a hunter and is the rate of effective search (Hoi l ing, 1959a). 11 CHAPTER TWO STUDY AREAS, HUNTING REGULATIONS AND METHODS The Delta Marsh The Delta Marsh (Figure 2), along the south shore of Lake Manitoba, is one of the major waterfowl harvest areas on the Canadian Prair ies . The 20,235 hectares of marshland are separated from the lake by a narrow ridge of sand covered with Poplar (Populus spp.) , Ash (Fraxinus sp . ) , Maple (Acer  negundo) and Willow (Salix spp.). To the south of the wooded ridge l ie vast stands of Reed Grass (Phragmi tes commun i s) broken by large open bays bordered with Bulrush (Scirpus spp.) and Cattail (Typha spp.) . The marsh is joined to the lake by five channe1s a 11owing its water levels to fluctuate with wind tides on the lake. Reed Grass is separated from upland borders by extensive meadows of Whitetop Grass (Schlochloa festucacea). On drier areas, some of which are hayed and grazed, the whitetop is interspersed with Sow Thistle (Sonchus arvens i s), Aster (Aster sp.) , Foxtail (Hordeum jubatum) and Goose foot (Chenopodeum rubrum). Many spring migrants, primarily Mallard (Anas platyrhynchos), Pintail (Anas acuta), Canvasback (Aythya va l i s iner ia ) , Redhead (Aythya ameri cana), Lesser Scaup (Aythya affi nn i s) , Blue-winged Teal (Anas d i scors) , Wigeon (Anas ameri cana), Shoveler (Spatula clypeata) and Gadwall (Anas strepera) stop here in A p r i l . Some remain to nest; others journey on to nesting grounds in western Canada. The large open bays of the Delta Marsh, with their emergent and submergent aquatic plants, continue to serve as a major molting range for dabbling ducks which are numerous there during the fl ightless period (mid-June through mid-August). Diving ducks prefer to molt on several large lakes of west-central Manitoba. Both dabblers and 12 Figure 2. Map of the Delta Marsh. 13 divers gather at Delta, an important staging area from late August through October. Adult Mallard and Pintail are on the wing by early August, their numbers bolstered by young-of-the-year. These two species commonly associate in morning and evening stubble f l ights from marsh to adjacent farmlands after mid-August. Mallard remain common until freeze-up in late October or early November, most Pintail having migrated by mid-September. A build-up of Blue-winged Teal on the sloughs and creeks of the marsh occurs in August, most of these departing by the end of September. Canvasback, mostly young-of-the-year and adult females arrive about the f i r s t of September and depart in mid-October. Redhead, follow a similar pattern. Lesser Scaup are common on the open bays until the end of October, with smaller numbers of Bufflehead (Bucephala albeola) and Green-winged Teal (Anas carol?nens?s) . In 1973, hunting on the eastern portion of the Delta Marsh, site of my studies, did not begin until October 8 . The hunting season at Delta opened on October 7 in 1 9 7 4 . The legal close of shooting was December 1 in 1 9 7 3 and November 30 in 1 9 7 4 . In both years, however, the gunning came to a close when the marsh froze over by November 7 . During both years the daily bag limit was six ducks. This legal k i l l was modified by special regulations for some species. In 1 9 7 3 and 1 9 7 4 only one Canvasback or one Redhead could be taken daily while Mallard k i l l was restricted to three per day in 1 9 7 3 , two per day in 1 9 7 4 . In Manitoba as a whole, two additional Lesser Scaup were allowed as "bonus" ducks after October 8 in 1 9 7 3 , but this extra allowance did not hold the following year. In 1 9 7 5 , 1 9 7 6 and 1 9 7 7 the legal l imit was 6 , 8 and 6 birds per day respectively. During these years hunters were restricted to 3 Mallards daily and either one Canvasback or Redhead. Hunting began October 6 in 1 9 7 5 , 14 September 27 in 1976 and October 8 in 1977 and ended with freeze-up during the second week of November in a l l three years. The Pasquia Marsh The Pasquia Marsh study area of 1,820 hectares is located in northern Manitoba about 32 kilometres southwest of The Pas. Drainage ditches l ink the 65 hectares of open water in the centre of the marsh with a channel leading into The Pas River (Figure 3 ) . The outer borders of Pasquia Marsh are dominated by stands of Willow, Cord Grass (Spartina spp.) and Reed Grass. The heart of the marsh holds islands of Cattail and Hardstem Bulrush (Sc? rpus acutus), with Water Mi l fo i l (Myriophyl1um spp.) the common submergent. Pasquia Marsh is bordered on the north and west by agricultural land and is therefore a favoured staging area for field-feeding Mallard and Pintail in August and September. The marsh is used infrequently by diving ducks which prefer the larger lakes in The Pas region with their dense beds of Sago Pondweed (Potamogeton pecti natus). In 1974, the only year of my study there, the Pasquia Marsh opened to hunting on September 9 with the season closure on November 3 0 . Daily bag limits and restrictions were the same as those on the Delta Marsh during the 1974 hunting season. This area was studied in 1974 to contrast vulnerability with Delta. Al l the birds left the marsh after five days of hunting so a ful l comparison proved impossible. 15 Figure 3- Map of the Pasquia Marsh. PASQUIA L A K E 16 It is not possible to test the hypotheses outlined above by direct manipulation of birds and hunters. However, natural situations involve radical changes in these variables through each hunting season. Thus, appropriate monitoring should provide some of the necessary contrasts, and my f ie ld studies had the fol1 owing objectives: 1. to gather information on vulnerability of ducks according to their experience with hunters, species differences, age and weather; 2. to estimate the numbers and species composition of ducks on the study areas and monitor bird behavior during the f a l l ; and 3. to measure hunter effort and observe hunter behavior (cripple rate, se lect iv i ty , handling time), thereby documenting some of the hunter components of the mortality process at different times during the season. Decoy Tria ls I chose decoy tr ia l s or experimental hunts as a means to evaluate the effects of hunter density, waterfowl species composition and age and weather on vulnerabil i ty . This experimental technique measures the response of ducks to the observer and his decoys in hunting situations. This procedure has been used to monitor hunter-duck interactions by several investigators (Teplov and Kartashev, 1957; Olson, 1965; Stott and Olson, 1972, and Alford and Bolen, 1977). My decoy t r ia l s began one week before hunting each year and continued until freeze-up as part of my routine observations during the 1973 and 1974 hunting seasons. Thirty decoys were used in my t r i a l s . At Delta, there were ten each of Canvasback, Mallard and Lesser Scaup, the most commonly 17 ki l led species. At Pasquia Marsh only Mallard decoys were used, Mallard being the only abundant duck staging there. My observation blinds were always situated in sheltered bays or on the leeward sides of points, sites normally used by feeding and loafing ducks. Al l observations were made on public shooting grounds. When ducks responded to my decoys, shots were fired in order to prevent the birds from becoming habituated to my decoys. Observations were made from sunrise when ducks were easily identified and active, until 0900 CST and from 1500 CST to sunset. From sunrise until sunset, light was sufficient for an accurate identification of birds . Moreover, these times spanned the periods of greatest act iv i ty by ducks and by hunters. Cloud cover and wind speed were measured when each decoy tr ia l was ini t iated. I also recorded barometric pressure and relative humidity each day in order to determine the effects of these quantitative measures of weather on vulnerabil i ty . The latter observations were made at 0600 CST and 1700 CST on most days. During decoy t r i a l s , records were kept of the time of each encounter, of species responding, the number of birds in each flock and approximate minimum range of birds from my bl ind . The ranges were estimated by judging the position of the flock from the decoys, which were placed at known distances from the bl ind. A coefficient of vulnerabil ity for each species was determined weekly during the study from these experimental hunts. This index is calculated by dividing the number of potential k i l l s per hour, (number of the birds less than 35 m) for each species by its population size. Although the k i l l i n g range of a shotgun may be greater than 35 m, I selected this as the vulnerable distance since a duck centered in the pattern up to 35 m can be 18 bagged consistently (Bellrose, 1 9 5 3 ) . For example, 8 0 Mallard may have passed within 35 m of my blind during 20 hours of observations. The rate of encounter per hour then equals 8 0 divided by 20 (4 birds per hour). If there were 1 , 0 0 0 Mallard present in the study area during that period, the vulnerability coefficient would be 4 divided by 1 , 0 0 0 or 0.004. The index represents the risk per hour of a bird in a given population passing within gun range of one hunter, and is species-specific. My model suggests that vulnerabil ity may change due to differences in location, time of season, hunting effort , avoidance learning or flocking behavior. Vulnerability is also related to sex as in Canvasback and Pintai l (Olson, 1 9 & 5 , Alford and Bolen, 1 9 7 7 ) and to age structure of the population (Bellrose et a l . , 1 9 6 1 ) . I did not attempt to measure differences in sexes of birds responding to my decoys. Ratios of juveniles to adults were determined by examination of hunters' bags to evaluate differences in vulnerabil ity related to age. These data were collected weekly on the Delta Marsh and on opening day at Pasquia Marsh in 1 9 7 4 . Aerial Surveys Estimates of population size and species composition on my study areas were accomplished by aerial surveys. On the Delta Marsh these were flown weekly by staff of the Delta Waterfowl Research Station. The f l ights followed east-west section lines at an altitude of 3 0 . 0 metres, beginning at 0 9 0 0 CST after ducks had returned from feeding in stubble f ie lds . Only birds considered to be within 0 . 2 0 kilometres of the observer's side of the aircraft were counted. Fourteen percent of the marsh was covered by each survey. These counts were then extrapolated to provide an estimate for the ent i re marsh. 19 In surveys where species identification was not possible, ducks were recorded simply as dabblers or divers. Estimates were then adjusted for unidentified birds by the following calculations: Y = X • R T = Y + S where: X = ratio of identified ducks by species (S) to total identified dabblers or divers (P) Y = species corrected from unidentified category R = total unidentified dabblers or divers T = adjusted individual species estimate In addition to weekly population surveys, four counts were made in 1973 to determine distribution patterns of ducks on the Delta Marsh relative to hunted areas and refuges where no hunting is permitted (Figure 2). Six transects were flown covering both shooting grounds and refuges. The location of a l l ducks observed was recorded on maps. The percentage of total ducks using refuges was then calculated. These surveys were flown at 0900 CST at an altitude of 100 m. Aerial surveys over the Pasquia Marsh were made on the third day before hunting and on the third and sixth day of the open season. The entire marsh was surveyed, hence a complete count resulted. Spy Blinds and Hunter Effort Since experimental hunts may not be representative of hunters as a whole, I also (1) monitored other hunters from spy blinds, and (2) analyzed 20 overall data on effort and k i l l s collected on the Delta Marsh during 1975 to 1977. I observed and evaluated k i l l s from spy blinds located about 90 m from hunters who were not aware of the nature of my ac t iv i t i e s . My blinds and decoy set-ups were authentic by wildfowling standards and my assistant fired his gun now and again as if hunting. Hunters I watched showed no evidence of being aware they were under observation. Al l my studies were made on public shooting grounds. At peaks, as during opening week and on holidays, three hunting blinds could be observed from my hide. More often, only one party was in view. The entire hunt of each group was followed as often as weather permitted. Calculations involved 179 hunter hours, the product of eight weeks on the marsh during three hunting seasons. The fate of a l l ducks i estimated to be within k i l l i n g range (35.m) of hunters was recorded. Encounters with a l l birds fired upon beyond this range were also recorded. Hunter selectivity was determined from these hunter-bird meetings. For each observation period I recorded species encountered, and time> flock siz< range of birds from guns, number of birds brought to bag and the crippling loss, the latter calculated by dividing the number crippled by the number of birds k i l l ed and crippled. With the assistance of Delta Waterfowl Research Station personnel, a survey of ducks in hunters' bags was conducted on the Delta Marsh during 1973 and 1974. Wings were also collected weekly at private and commercial hunting lodges. Hunter effort was determined each week by dividing the number of ducks in the bag by the number retrieved per hunter hour as observed from my spy blinds. The index represents hunter effort in terms 21 of hunter hours. In 1975, 1976 and 1977 the Manitoba government monitored.hunter effort and duck k i l l on the Delta Marsh, employing aerial surveys and a bag check. Numbers of hunters was estimated during three fl ights a week, counting hunting parties and multiplying by average party size. Flights began at 0800 CST at an altitude of 350 m. Numbers of ducks k i l l ed on the marsh (for each species) were then estimated by multiplying the number of hunters by the average k i l l for each species found in the bag ta l ly during these years. These hunter effort and k i l l figures were used in a l l calculations dealing with the period 1975 through 1977. 22 CHAPTER THREE FIELD RESULTS Duck numbers on the Delta Marsh varied considerably during the fa l l s of 1 9 7 3 and 1 9 7 4 (Figure 4 ) . In 1 9 7 3 , populations approached 1 0 0 , 0 0 0 largely due to an abundance of Canvasbacks. During the hunting season, however, populations followed a declining trend both years (Figure 4 ) . During the period 1 9 7 5 through 1 9 7 7 populations at Delta showed similar trends, peaking at around 4 5 , 0 0 0 in late September then declining during the hunting season (Figure 5 ) • Hunter hours during 1 9 7 3 and 1 9 7 4 and hunter numbers during 1 9 7 5 , 1 9 7 6 and 1 9 7 7 followed the population trends. Hunter effort was greatest early in a l l hunting seasons, declining as duck numbers fe l l with south bound migration (Figures 6 and 7 ) . The pattern of high populations and hunter effort early in the season followed by rapid decline in both variables provided a useful contrast; I was able to make observations over a wide range of waterfowl abundance and hunter effort levels. The close correlation between these variables, however, made it d i f f i cu l t to sort out interactions and effects due to migration behavior, foraging patterns and weather. Vu1nerab i1i ty During the two years of my decoy t r i a l s , 6 , 9 0 4 ducks of 11 species responded to my decoys in experimental hunts. Mallard, Canvasback and Lesser Scaup were the most common ducks on the Delta Marsh during my study. My findings deal mainly with these species. I found Canvasback and Lesser Scaup showing least wariness of hunters and their decoys. Mallards were consistently more wary than either of the 23 Figure h. Duck populations on the Delta Marsh during the fa l l s of 1 9 7 3 and 1 9 7 4 as determined by aerial surveys. 23a 0-1 1 1 1 1 1 1 1 1 1 1 —I 1 S t / ! H U 1 B W S W H W H M B - - N N P i n T X W W U ] DAYS FROM SEPT. 1 2k Figure 5. Duck populations on the Delta Marsh during the fa l l s of 1975, 1976 and 1977 as determined by aerial surveys. N U M B E R D F CL ICKS ( 1 0 0 0 * 5 ) 25 Figure 6. Estimated hunter hours during each week of hunting during 1973 and 1974. 25a 2Mh 04 1 1 1 26 Figure 7 . Estimated numbers of hunters during each week of hunting during 1 9 7 5 , 1 9 7 6 and 1 9 7 7 . N U M B E R D F H U N T E R S m an tsa r** m tsa tsa tsa tsa tsa tsa tS3 tsa tsa * F ? ? — * P 27-diving ducks (Tables I and II). I regressed vulnerabil ity coefficients for Mallard, Canvasback and Lesser Scaup against effort (Log Base of 10 hours hunted per week) to test the hypothesis that vulnerabil ity would decrease as cumulative hunting pressure increased. Although relationships existed (Figures 8, 9 and 10) I rejected the hypothesis since the correlation was positive indicating increased vulnerabil ity with increased cumulative effort . However, vulnerabil ity was s ignif icantly and negatively t.j>p .05) correlated with current (instantaneous) hunting effort for Mallard (R2 =.60, df = 4, y = .00264 - .0007 (X)) (Figure 11) and Lesser Scaup (R2 = .74, df = 4, y = .0142 - .0042 (X)) (Figure 12). There was not a significant relationship (p>.1) for Canvasback (R2 = .03) (Figure 13). These findings suggest that vulnerabil ity is inversely proportional to current hunting effort and I therefore concluded that: (1) the short term (daily) wariness of Mallard and Lesser Scaup is increased through encounters with many hunters, and (2) as duck-hunter meetings decline, so does the degree of vulnerabil i ty . If there is avoidance learning, it is either not retained even through one hunting season, or is masked by other behavioral changes that occur through the fa l l period. In either case, avoidance learning does not protect birds late in the season against the lower levels of hunting that do occur then. Diving ducks are dependent on open waters; Mallard and other dabbling ducks adjust to a much broader pattern of acceptable habitat. Delta Mallards loaf on the shore of Lake Manitoba, an unhunted range. They crossed the marsh morning and evening in their twice daily f l ights to feed on agricultural f ie lds . Under heavy gun pressure, these fl ights are usually above gun range. Canvasback remain on the marsh throughout their stay at Delta; Lesser Scaup frequently trade back and forth between marsh and open Table I. Results of 36.8 hours of decoy tr ia l s during the 1973 f ield season. Total Number Vulnerability Sampling Population Birds 2 Birds Percent Vulnerable Coefficient 6 Period Species Size 1 <35m Responding3 Vulnerable 4 per Hour5 (X10" 4) October 1-6 Mai lard Lesser Scaup Canvasback October 8-13 Mai lard Lesser Scaup Canvasback October 15-19 Mai lard Lesser Scaup Canvasback October 23-26 Mai lard Lesser Scaup Canvasback 17,036 32 51 3,454 90 111 26,189 163 218 4,163 18 57 6,017 64 67 11,442 56 106 3,630 34 78 4,276 82 105 410 34 61 6,851 19 33 6,993 60 64 888 10 10 63 3.56 2.1 81 10.00 28.9 75 18.10 6.9 32 1.22 2.9 96 4.32 7.2 53 3.78 3.3 44 3.40 9-3 78 8.20 19.1 56 3-40 82.9 58 2.44 3.6 94 7.69 11.0 100 1.28 14.4 *from aerial surveys. 2 birds attracted to <35m in decoy t r i a l s . 3 total birds attracted to decoys during decoy tr ia l s -^percent of total birds that were <35m in decoy t r i a l s . 5 birds <35m divided by number of hours of observation. 6number vulnerable per hour divided by population size. Table II. Results of 45.6 hours of decoy tr ia l s during the 1974 f ie ld season. Total Number Vulnerability Sampling Population Birds 2 Birds Percent Vulnerable Coefficient 5 Period Species Size 1 <35m Responding3 Vulnerable 4 per Hour5 (X10" 4) September 30-October 4 Mai 1ard Lesser Scaup Canvasback 32,866 978 10,216 195 50 169 October 7~10 Mai lard Lesser Scaup Canvasback 25,025 7,475 1,679 102 14 79 October 12-17 Mallard Lesser Scaup Canvasback 20,521 1,877 759 49 23 54 October 19-24 Mallard 5,529 - 91 Lesser Scaup 533 43 Canvasback 1,023 47 221 50 200 88 100 85 11.18 3.27 11 .05 3.4 35.2 10.8 254 14 98 40 100 81 9.36 1.28 7.25 3.7 1.7 43.2 75 23 54 65 100 100 5.38 2.53 5.93 2.6 13.5 78.1 137 135 58 66 32 81 8.83 4.17 4.56 15.9 78.2 44.5 1from aerial surveys. 2 birds attracted to <_35m in decoy t r i a l s . 3 total birds attracted to decoys during decoy t r i a l s , ^percent of total birds that were <35m in decoy t r i a l s . 5 birds <35m divided by number of hours of observation. 6number vulnerable per hour divided by population size. 30 Figure 8. Relationship between Mallard vulnerabil ity and cumulative hunter effort as determined during decoy t r ia l s at Delta Marsh during 1973 and 1974. V U L N E R A B I L I T Y C D E T F 1 C 1 E N T C X 10 > S X m ' — — hJ rd m s 31 Figure 9- Relationship between Lesser Scaup vulnerabil ity and cumulative hunter effort as determined during decoy tr ia l s at Delta Marsh during 1973 and 1974. 32 Figure 10. Relationship between Canvasback vulnerability and cumulative hunter effort as determined during decoy t r ia l s at Delta Marsh during 1973 and 1974. 32a C U M U L R T I V E H U N T I N G E F F O R T C L D E B R 5 E 1 0 H D U R 5 ) 33 Figure 11. Relationship between Mallard vulnerability and instantaneous hunter effort as determined by decoy tr ia l s at Delta Marsh during 1973 and 1974. 33a 34 Figure 12. Relationship between Lesser Scaup vulnerability and instantaneous hunter effort as determined by decoy t r ia l s at Delta Marsh during 1973 and 1974. 34a !23t X ! S s s tsi si ta tsa • - • - • * • • — rvi rvi rvi m rn H U N T E R E F F O R T C H H L I R 5 - L O G ' B R 5 E 1 2 ! ) 35 Figure 1 3 . Relationship between Canvasback vulnerability and instantaneous hunter effort as determined by decoy tr ia l s at Delta Marsh during 1 9 7 3 and 1 9 7 4 . H U N T E R E F F O R T C H O U R S ~ L O G B A S E 1 36 lake. None of the diving ducks v i s i t stubble f ie lds . At Pasquia, Mallards loafed on the marsh between their periods of f ie ld feeding, thus being highly vulnerable throughout the day. Under steady gun * pressure, the Mallards vacated the Pasquia Marsh after the f i r s t week of shooting, shifting to isolated lakes, inaccessible to hunters. In addition to the vulnerabil ity coefficient, I also measured vulnerabil ity by the number of ducks that were lured to within 35 m of my blind as a percentage of a l l birds responding. At Pasquia Marsh in 1 9 7 4 , 9 1 . 6 percent ( 3 2 9 of 3 5 9 ) of the Mallard responding during decoy tr ia l s were vulnerable prior to hunting. During the f i r s t week of hunting the percent Mallard vulnerable dropped to 6 7 . 8 ( 7 2 0 of 1 , 0 6 2 ) . There were no birds remaining to hunt after the f i r s t week. At Delta in 1 9 7 4 , the Mallard vulnerable percentage was 8 8 during the week before hunting (Table II). Within the f i r s t week of hunting this vulnerable percentage dropped to 40, evidence of this species avoidance behaviors. Vulnerability increased to 6 5 percent during the second hunting week and remained at that level through the season (Table i l ) . In 1 9 7 3 Mallard vulnerable percentage was 6 3 percent during the week before hunting. This dropped to 32 percent during opening week and then increased to 5 8 percent by the third week of hunting (Table l ) . The large difference observed between Delta and Pasquia may have been due to the fact that some Mallards at Delta may have departed as they did at Pasquia but I was not able to detect the sh i f t . Alternatively, Mallard may not have left Delta because of the lack of unhunted safe areas as occur in the Pasquia area. Canvasback in 1 9 7 4 showed an 85 percent vulnerability during the week before hunting. During the opening week of shooting, this dropped only 37 four points to 81 percent then increased to 1 0 0 percent during the second week of hunting (Table II). In 1 9 7 3 Canvasback were 75 percent vulnerable prior to hunting decreasing to 5 3 percent during the opening week. During the last week of hunting season, Canvasback vulnerability percentage rose to 1 0 0 (Table l) . The percentage vulnerable for Lesser Scaup was 1 0 0 percent prior to hunting in 1 9 7 4 dropping to 32 percent vulnerable by the last week of the season (Table l l ) . In 1 9 7 3 , the percent of the Lesser Scaup vulnerable was highest during the season (Table l ) . These vulnerabil ity findings suggest that Mallard are sensitive to hunters and that avoidance plays a role in their survival when they use heavily shot marshes. The diving ducks, perhaps because they are less adaptable and because their major feeding and resting areas are concentrated on hunted ranges, maintain a high degree of vulnerabil ity throughout the season. I believe avoidance learning is a short-term behavior and that wariness is not retained through the fa l l at the late-season effort levels I observed. I suggest that early avoidance of blinds and hunters is to the noise of shooting (fright syndrome) to which birds later habituate. An alternative explanation is that later in the season birds require more feed (temperatures are lower and energy requirements may be greater) so that the higher vulnerability is a function of increased pre-mi.gratory foraging behavior. Vulnerabi l i t ies , however, are also negatively correlated with bird abundance, vulnerabil ity being highest when numbers were small. This was a unexpected result and may have been due to something besides the correlation between birds and hunters. Possible explanations are ( 1 ) food requirements 38 may be greater later in autumn, moving birds into huntable areas; (2) presence of proportionally more young with high vulnerabil ity later in the season; or (3) migration into the study area of naive birds from unhunted areas farther north. My findings from 1973 aerial surveys indicate a tendency for a l l ducks to move away from hunted areas as the season advances, this shift coming earl ier and in greater numbers for the Mallard and other dabbling ducks. Besides the unshot lakeshore, there are two unhunted refuges on the Delta Marsh (Figure 2 ) . The relative number of ducks on these unshot ranges increased steadily during the hunting season as cumulative hunting pressure increased (Table I I I ) . This suggests an avoidance of hunted zones resulting in clumped distribution centres (Table I I I ) . Age may also be of importance to a duck in establishing and maintaining its safety during the hunting season. During the f i r s t week of the 1974 season at Pasquia, September 9 to 14, Mallard vulnerability was 25.3 x 10~ 4 . At Delta for the opening week, October 7 to 12, 1974, the vulnerabil ity coefficient for Mallard was 3-7 x 1 0 - 4 . The opening day bag at Pasquia showed an adult-juvenile ratio of 1:23 .8 whereas at Delta the opening week bag was 1:1.7. To make a more appropriate comparison of the data, vulnerabil ity coefficients for the two ranges were adjusted for marsh size. This was accomplished by calculating ducks per hectare and dividing this into the rate of encounter per hour for the sampling period. At Pasquia Marsh the adjusted coefficient for Mallard was 46.01 whereas at Delta the adjusted coefficient was 7-56 indicating that birds at Pasquia were six times as vulnerable as Mallards at Delta during the f i r s t week of hunting. I attribute these differences to the larger proportion of immature birds in the more northern population. These limited findings 39 Table III. Distribution of ducks relative to refuges on the Delta Marsh as measured by aerial surveys during 1 9 7 3 -Cumulative Number Birds Total Ducks Hunter Percent Date in Refuges Observed Hours in Refuges October 7 3 1 0 4 , 0 8 0 1 , 6 0 6 8 October 13 6 8 0 4 , 9 7 0 2 , 6 3 5 14 October 18 2 6 0 1 , 1 8 5 2 , 9 3 1 22 October 20 2 2 0 5 9 0 2 , 9 3 1 27 4o support the conclusion of Bell rose et a l . (1961) that populations consisting largely of juveniles are more vulnerable than aggregations of adults. Bellrose et a l . (1961) found a progressive decrease in juvenile to adult vulnerabil ity in male Mallards beginning at 6.5 during the opening week of the season at Havana, I l l inois and declining to 1.5 times as vulnerable toward season's end. In contrast to conventional wisdom,weather had l i t t l e effect on the vulnerabil ity coefficients of Mallard, Lesser Scaup and Canvasback (Table IV). The dependent variable (vulnerability for individual periods of decoy tr ia l s ) was tested in multiple regression analysis using barometric pressure, relative humidity, cloud cover and wind as the independent variables. Each species was tested separately. The one significant correlation was between wind speed and Mallard vulnerabil ity suggesting that when winds are high, Mallard are less wary of hunters (Table IV). Hunter Selectivity and Observed K i l l The performance of 61 hunting parties was observed from spy blinds during 1973, 197** and 1977 on the Delta Marsh and I accumulated observations on 179 hunter hours. Hunters showed l i t t l e se lect ivity and passed up only 39 opportunities to k i l l ducks in 267 encounters (Table V) . Green-winged Teal, Blue-winged Teal and Bufflehead were passed up with greater frequency than the large ducks (X 2 = 20.2 , df = 1, p<.05) . There was variation in the retrieved k i l l per hour and total k i l l s per hour between weeks during 1973 and 1974 (Table VI) . The largest number of birds retrieved per hour was 2.85 and the most ducks downed (k i l l s and cripples) was 4.0 (Table VI) . During most weeks the retrieved k i l l was 41 Table IV. Summary of F ratios generated by regression analysis of vulnerabil ity as related to selected weather variables (df = 32, F = 4.15 for p<.05) . Environmental Variable Barometric Relative Cloud Wind Species Pressure Humidity Cover Speed Mallard 0.94 0.93 4 . 0 8 5-65 Lesser Scaup 1.95 0.04 1.43 0.30 Canvasback 0.11 0.06 0.73 0.22 42 less than one bird per hunter hour suggesting that few hunters f i l l their bags during a single hunt. The overall bag per hunter hour in 1 9 7 3 and 1 9 7 4 was . 91 and . 7 9 respectively. Dividing these values into the bag per hunter for 1 9 7 3 and 1974 gives time per hunt which was 1 . 6 3 hours in 1 9 7 3 and 1 . 7 5 hours in 1 9 7 4 (Table VI). These spy blind observations are supported by the Delta Marsh bag check (Table VI l ) . In the five years of the study, average hunter bags were less than two ducks in a l l years except 1 9 7 6 when i t was 2 . 7 8 . The number of ducks taken by the average hunter was considerably less than the daily bag limit during the study period (Table V I l ) . Thus, bag per hunter is somewhat below management limits suggesting the relationship between k i l l and duck abundance is not l inear. Hand 1ing Time The difference in mean time between encounters when birds were fired upon (unsuccessfully) and mean time between successful encounters, and the next encounter was used as a measure of handling time. The time between encounters fired at takes into account only periods when hunters are reactive to birds. The time between an encounter when a bird is k i l led and the next encounter where the hunter is reactive takes into account the time elapsed when the hunter is not able to react plus the time between encounters. The difference between these two encounter rates is assumed to represent time when a hunter is engaged in nonproductive act iv i t ies such as reloading the gun, pursuing cripples or retrieving downed birds. I accumulated 3 , 7 9 6 minutes of observation on handling time. The time between encounters fired upon was 9 - 3 1 minutes (S 2 = 2 3 - 7 9 ) whereas the time between encounters where birds were ki l led and the next encounter had a 4 3 Table V. Hunter select ivity by species as determined by observation on the Delta Marsh during 1 9 7 3 , 1 9 7 4 , 1 9 7 7 -Total Flocks Flocks Flocks <35 meters <35 meters Percent Species <35 meters Fired At Not Fired At Fired At Blue-winged Teal 3 8 2 7 11 71 . 0 Bufflehead 18 13 5 7 2 . 2 Canvasback 34 30 4 8 8 . 2 Gadwa11 4 3 1 7 5 . 0 Green-winged Teal 2 3 16 7 6 9 . 6 Lesser Scaup 30 2 5 5 8 3 . 3 Mai lard 76 74 2 9 7 . 4 Pintai1 8 8 0 1 0 0 . 0 Redhead 18 16 2 8 8 . 9 Shoveler 9 8 1 8 8 . 9 Wigeon 9 8 1 8 8 . 9 Total 2 6 7 2 2 8 39 8 5 . 4 44 Table VI. Observed k i l l , k i l l per hunter-hour, estimated effort and population size for the Delta Marsh in 1 9 7 3 and 1 9 7 4 . Week of Season 1 2 3 Year 1973 1974 1973 1974 1973 1974 Observed Total K i l l 1 2,088 1,382 8 9 6 808 195 205 Retrieved K i l l / H o u r 2 1.30' . 7 2 . 8 7 .42 .66 2.85 Total K i l l / H o u r 2 3.10 1.10 1.70 .85 1-30 4.00 Effort (Total Hours) 3 1,606 1,919 1 , 029 1,924 296 72 Populat ion Size 27,747 36,605 11,695 25,025 15,574 8,350 • 1 f rom bag t a l l y . 2 f rom spy b l i n d s . C a l c u l a t e d by observed t o t a l b a g / ( r e t r i e v e d k i l l / h o u r ) . 4 5 Table VII. Hunter success on the Delta Marsh expressed as ducks per hunter as determined by hunter bag checks during the period 1 9 7 3 through 1 9 7 7 -Legal Number of Observed Daily Hunters Ducks Ducks/ Hours/ Duck Date Checked Bagged Hunter Hunt Limit 1 9 7 3 9 6 8 1 , 4 3 7 1 .48 1 . 6 3 8 1 9 7 4 5 5 5 7 6 5 1 . 3 8 1 . 7 5 6 1975 9 9 4 1 , 7 5 5 1 . 7 7 6 1 9 7 6 4 1 9 1 , 1 6 5 2 . 7 8 8 1 9 7 7 7 9 7 9 6 9 1 . 2 2 6 46 mean period of 9.29 minutes (S 2 = 17.42). These findings suggest no significant handling time effect since the two refractory periods are essentially the same with no measureable period where a hunter is non-reactive. I suggest this means that hunters do not pursue cripples or retrieve birds or remove time from effective search when there is a poss ibi l i ty of k i l l i n g another b i rd . Therefore it appears that non-linear k i l l patterns do not result from handling time constraints. Relation of Flock Size to Probability of Mortality The probability of ducks being wounded or k i l l ed (ratio birds downed to birds fired upon) when passing a hunter at less than 35 m was 0.21 whereas at greater than 35 m the probability of mortality was 0.11 (Table VIII) . These figures were not adjusted for flock size and they do not take into account the number of hunters present. The results changed considerably when the probability of mortality was related to flock size (Figure 14)'.. Here the chance of a duck being k i l l ed in a flock was calculated by the ratio of birds k i l l ed to birds fired at in each flock category. The data indicated that probability of mortality decreased with increasing flock size. However, a hunter is not less l ikely to hit any bird that he shoots at when that bird is in a large flock. I have adjusted these data to consider mortality per hunter (Figure 15). Mortality rates per hunter were calculated by taking the average number of.shots per hunter for encounters with flocks of a given size. This is divided into the number of shots required to down a bird from a flock of that size yielding the number of encounters required to down one bird (Table IX). This ratio is 'then multiplied times the flock size, giving the number of birds required for one hunter to k i l l one duck (Table IX). The mortality rate is 47 then one divided by the number of birds required by one hunter to k i l l a single duck. The relationship remains similar although the mortality rates are much lower than before adjustment. Flock size is a component of vulnerabil ity because the probability of mortality for the individual declines as the number of its companions becomes larger. Small flocks of 5 or less decoyed most frequently in my decoy t r i a l s . For Lesser Scaup, 76 percent in 1973 and 94 percent in 1974 of small flock encounters were less than 35 m. Seventy-nine percent of the small flocks in 1973 and 87 percent in 1974 of Canvasback and 65 percent in 1973 and 73 percent in 1974 of the small flocks of Mallard came to within less than 35 m of my blinds. From spy blinds I found the average flock size to be less than 3 for a l l species except Lesser Scaup (4.79 in 1974, Table X) . By regressing probability of mortality against flock sizes one through f ive , probabil it ies of mortality for average species-specific flock sizes were determined (Table Xx). The linear relationship between mortality and flock size for birds less than 35 m from hunters' blinds was significant (R2 = .83, df = 3, p<.05, y = .25 - .0355 (X)). Species-specific probabilit ies of mortality for mean flock sizes were calculated using the linear equation (Table X) . There appeared to be an increase in flock sizes for Mallard as the season advanced. Mean flock size per week was regressed against cumulative hunting effort for Mallard to see i f these changes could explain the relationship between vulnerabil ity and hunter effort . The prediction tested was that mean flock sizes would increase over time due to the enlarged benefit for the individual of being in a larger flock. No correlation was found (R2 = .27, p>.05, df = 4) suggesting that ducks do 48 Figure 14. Probability of mortality (ratio birds downed to birds fired at) in relation to flock size as determined from spy blinds on the Delta Marsh during 1 9 7 3 and 1 9 7 4 . 48a Flock Size Figure 1 5 . Relationship between probability of mortality and flock size as determined from spy blinds on the Delta Marsh during 1 9 7 3 and 1 9 7 4 . Values adjusted for one hunter. 49a Flock Size Table VIII. Probability of mortality in relation to range from hunters' blinds on the Delta Marsh (1973, 1974). 1973 1974 Total Bi rds <35 meters 525 166 Birds Downed <35 meters 92 56 Probab i1i ty of Mortali ty .17 .34 Total Bi rds >35 meters 328 309 Bi rds Downed >35 meters 48 25 Probabi1i ty of Mortali ty .15 .08 V/1 o Total 691 148 .21 637 73 .11 Table IX. C a l c u l a t i o n s * of p r o b a b i l i t y of m o r t a l i t y showing adjustment for f lock s i z e s and hunters. P o t e n t i a l Number Number Flock Number B i r d s Birds Size Flocks Downed A v a i l a b l e Average Number Average Number of Shots Average Percent of P o t e n t i a l Shots Downed Number Hunters Taken per Shots Number of Number Present Encounter per of Shots Encounters per Downed per t  per pci uw-.w- r~- — , 3 Taken Encounter Hunter 1 B i r d 2 Encounter One B i r d to Down Number of Bi rds to Down One B i r d 4 Proba-bi 1 i t y M o r t a l i t y for One Hunter 5 1 91 47 91 52.7 213 2.5 .94 4.4 2.3 4.7 4.7 21.3 2 47 38 94 4o.4 117 2.0 1.23 3.1 2.5 2.5 5.0 20.0 3 23 18 69 26.1 51 2.5 .89 2.8 2.2 9.3 10.8 4 17 15 68 22.1 51 2.3 1.21 3.4 3-0 2.8 11.2 8.9 5 7 9 35 25.7 48 2.9 2.40 5.3 6.9 2.2 11.0 9.1 e x a m p l e using Flock s i z e 2. If 1.23 shots were taken per encounter (1) and 3.1 (2) shots were required to down a b i r d U _ = 2 . 5 (3) t h e n u m b e r o f encounters required to k i l l one duck. Flock, s i z e m u l t i p l i e d times number of encounters y i e l d s number of b i r d s (4). P r o b a b i l i t y of M o r t a l i t y is the number one /(4) = (5). Table X. Mean flock sizes of ducks lured to decoys and probability of mortality based on flock size. Spec i es X 1 9 7 3 n S.E. Probabi1i ty of Mortali ty <35 meters X 1 9 7 4 n S.E. Probabi of Mort; <35 mei Blue-winged Teal 1.48 35 . 1 8 . 1 9 1 . 0 0 7 trace .21 Bufflehead 2 . 2 0 41 .24 . 1 6 1 . 8 9 38 . 1 7 . 1 8 Canvasback 2 . 8 3 3 6 6 . 1 6 . 1 5 2 . 7 6 2 0 9 . 2 2 . 1 5 Gadwal1 1 . 9 0 71 . 1 9 . 1 7 1 . 9 5 86 . 1 5 . 1 8 Green-winged Teal 1 . 9 0 30 .40 . 1 7 1 . 3 7 38 .11 . 2 0 Lesser Scaup 2 . 9 0 2 0 9 . 2 0 .14 4 . 7 9 62 1 .47 . 0 8 Mallard 2 . 5 3 3 8 9 . 1 7 . 1 6 2 . 7 1 3 1 3 . 2 5 . 1 5 Pintai1 1 . 5 8 109 .14 . 1 9 1 . 7 5 48 . 1 9 . 1 8 Redhead 2 . 0 0 5 7 .24 . 1 8 1 . 5 8 82 .14 . 1 9 Shoveler 1 .64 28 . 3 3 . 1 9 1.21 71 . 0 6 . 2 0 Wi geon 1 . 7 9 5 3 . 1 5 . 1 8 1.40 35 . 1 3 . 2 0 53 not flock in response to hunting effort or that increasing flock size in Mallard is not sufficient to explain the observed effort-vulnerabil i ty relationship. Density-Kill Relationship In a simple predator-prey system, quarry should be encountered in proportion to their abundance (Hoi l ing , 1959b). Therefore, I regressed encounter rates from decoy tr ia l s against population size to determine i f the k i l l patterns for the three species studied were related to population size and whether or not the rate of encounter is a random process. The linear relationship was not significant for Lesser Scaup or Canvasback (p>.05) (Figures 17 and 1 8 ) . There was a positive trend for Mallard (R2 = .34, df = 4, p>.05) (Figure 1 6 ) . These findings suggest that k i l l should be a non-linear function of density even i f the effects of handling time are ignored, since the encounter rate is apparently related non-1inearly to density: encounters do not result from a random search process. A possible bias in this analysis is the time between encounters caused by my shooting at passing birds to avoid habituation to my decoy tr ia l blinds. As a result of this negative reinforcement, the decoy t r ia l observations may have an a r t i f i c i a l handling time associated with each flock that was shot at. This would result in a saturating (non-linear) encounter pattern, as modelled by the Hoi ling (1959a) disc equation. I further analyzed my spy blind data to see if time between encounters was independent of population size. Mean time between encounters where shots were fired were calculated for periods where population estimates were available during 1973, 1974 and 1377- I then regressed time between encounters with population size and found only a slight negative correlation 54 with a slope near zero (y = 10.867 - .00007 (X) , R 2 = .12, p>.05) (Figure 19). These findings suggest that time between encounters is independent of population size and that there is a relatively constant refractory period after a shot has been f ired . I suggest that the time between encounters is constrained by bird behavior and that there is a refractory period or habituation time which must elapse after gun f i r e . This refractory period, about nine minutes in duration, would result in hunters' bags being below management l imits , giving a non-linear k i l l pattern. This would be especially true i f ducks are active only for short periods during certain times of the day. The slight decrease in time between encounters at higher populations is assumed to be related to a random distribution of birds during early season abundancy. However as populations decline, ducks must have a more contagious pattern in conjunction with a non-random search by hunters. Analysis of Hunter Bag Data To further investigate the k i l l versus population size and effort relationships, a multi-variate analysis of the 1975, 1976 and 1977 census and bag data was performed. Estimates of population size, k i l l and hunter numbers are presented in Table XI. Hunter numbers were adjusted for hours of effective search per hunt. In the analysis I assumed that hunters spent two hours actively hunting ducks since in 1973 and 1974 I found the average hunt to be about that length. A simple model that represents basic constraints and possible non-linearities in the harvesting process would be k = x NB1 E B 2 where k = ki11, N = population size, E = effort , and x, B^  and B^ are f i tted constants. This model says that the k i l l must go to zero i f either the 55 Figure 16. Relationship between encounter rates and population size for Mallard as determined by decoy t r ia l s at Delta Marsh during 1973 and 1974. ENCOUNTER RRTE (B1RD5 VULNERRBLE PER HDUR ) i s w x m m 56 Figure 17. Relationship between encounter rates and population size for Lesser Scaup as determined by decoy t r ia l s at Delta Marsh during 1973 and 1974. ENCDUNTER RRT.E CB1RD5 VULNERRBLE PER HOUR) •a w x tn m mm 57 Figure 1 8 . Relationship between encounter rates and population size for Canvasback as determined by decoy tr ia l s at Delta Marsh during 1973 and 1 9 7 4 . 0 E N C O U N T E R R R T E (BIRD5 V U L N E R R B L E P E R H D U R ) hi m m J ? 2H00 4 • r MB00 n H • z in N PI 7200 t 9E00 + in II ii -< II x X 12000 X 5 8 Figure 19. Relationship between mean time between encounters and duck abundance during 1973, 1974 and 1977 on the Delta Marsh. 58a 15.00T Ul E Id Z • z Ul Z Ul Ul I-Ul to Ul z 1 2 . 0 0 + a . 0 0 + G . 0 0 Y » 1 0 . Q G 7 ~ 7 . B 1 3 E - 0 5 X R 5 Q . = . 1 2 3 DF =6 r -z re ui z 3 . 0 0 0 . 0 0 - f -E3 H— P4 H Q •m H H Ln P D P U L R T 1 D N 5 1 Z E 59 population or hunting effort go to zero. Non-linearity in the k i l l per hunter functional relationship should be reflected by ^ 1 . 0 . Converting the equation to logarithms gives the multivariate linear form: log (k i l l ) = X + B.j (log population size) + B^ (log hunter hours) For Mallard the equation was highly significant (R2 = . 8 3 , df = 7 , p < . 0 5 ) . Thus I suggest that there is a significant relationship between k i l l , Mallard numbers and hunter ef fort . The equation resulting from the analysis was: Z = - . 5 7 5 " . 1 9 (log population size) + 1.1 (log hunter hours). The analysis showed similar results for the two diving ducks studied. For Lesser Scaup the relationship was significant (R2 = . 6 7 , df = 7 , p < . 0 5 , Z = - 1 1 . 9 + .46 (log population size) + 1 . 8 (log hunter hours)). Canvasback k i l l was also correlated with effort and numbers (R2 = . 8 7 , df = 5 , p<.05, Z = 8 . 5 + . 1 9 (log population size) + 1 . 6 (log hunter hours). Thus, a logarithmic model relating k i l l to population size and hunter effort describes k i l l patterns of the three species studied on the Delta Marsh. Furthermore, the B^  values were a l l less than 1 . 0 , suggesting a non-linear, saturating functional response per hunter in agreement with the experimental decoy t r i a l and bag check results. The values were a l l greater than 1 . 0 , suggesting that hunters improve (faci l i tate) one another's chances at finding birds. Allocation of Hunter Effort Duck populations and hunter effort declined as the hunting season progressed during a l l years of the study (Figures 4 , 5 , 6 and 7 ) . During 1 9 7 3 and 1 9 7 4 there was a positive linear correlation between hours hunted 60 Table XI. Estimated k i l l , population size and hunter numbers at Delta for selected weeks during 1975, 1976 and 1977. Mallard Lesser Scaup Canvasback Estimated Est. Pop. Est. Pop. Est. Pop. Number Year Week K i l l 1 S ize 2 K i l l 1 S ize 2 K i l l 1 S ize 2 Hunters 1975 Oct 6-0ct 11 1151 819 376 2145 Oct 13-0ct 18 447 668 507 1691 Oct 2 0 - 0 c t 25 365 4848 259 7408 191 2584 879 Oct 27-Nov 1 493 4 4 5 158 647 Nov 3-Nov 8 163 4656 144 8312 15 768 340 Nov 10-Nov 15 180 94 9 326 1976 Sep 2 7 - 0 c t 2 1265 9185 1244 4286 454 19069 1543 Oct 4 - 0 c t 9 3 3 3 454 126 1149 Oct 11-0ct 16 506 7807 1395 2 1 2 3 5 236 18400 1530 Oct 18-Oct 23 655 8634 1144 12864 171 2 2 3 5 944 Oct 2 5 - 0 c t 3 0 100 3 2 89 1977 Oct 3 - 0 c t 8 7 3 3 15484 6 6 6 7939 216 3167 1879 Oct 10-0ct 15 518 4064 671 5064 148 1 8 3 1342 Oct 17-0ct 22 5 5 3 3754 524 1426 5 8 115 9 1 9 Oct 2 4 - 0 c t 29 327 2592 4 9 9 8084 20 40 536 Oct 31-Nov 5 506 370 12 3 7 2 Nov 7-Nov 12 94 14648 7 3 2 0 8 310 1 Est imated k i l l . Population estimate. 3Data set not included in ana 1ys i s. 61 per week and duck population size for each week (R2 = . 7 4 , df = 4 , Y = 2 0 8 . 3 + . 0 6 5 (X), p<.05) (Figure 2 0 ) . During the 1 9 7 5 , 1 9 7 6 and 1 9 7 7 seasons the relationship between number of hunters and population size was also significant (R2 = . 7 0 , df = 9 , p < - 0 5 , Y = 3 7 4 . 9 + . 0 2 5 (X)) (Figure 2 1 ) . During the two periods of study, 1 9 7 3 and 1974 and 1 9 7 5 through 1 9 7 7 , estimated hunter effort was negatively related to time of the hunting season (Figures 6 and 7 ) . These findings suggest that hunter effort may be related to population size since most hunting occurred when ducks were abundant, and then declined as populations decreased. The negative relationship between effort and time may be related to cold weather since temperatures fa l l as the hunting season advances or simply that hunters lose interest and t i re of hunting as autumn progresses. Crippling Losses The study of hunter behavior is complicated by the unknown loss of crippled birds not retrieved. I found crippling losses to be similar during the f i r s t two years of my study, 3 8 . 6 percent in 1 9 7 3 , 3 3 - 3 percent in 1 9 7 4 . The overall crippling loss was 3 6 . 7 percent. The percent crippled was s ignif icantly higher (p<.05) for birds fired upon when more than 35 m from a hunter; 2 8 . 4 percent for birds less than 3 5 m versus 5 3 . 4 percent for birds greater than 35 m (Table XII) X 2 = 1 3 . 2 1 , df = 1 ) . The crippling loss was 3 5 . 6 percent with the inclusion of 1 9 7 7 data (Table XII). 62 Figure 2 0 . Relationship between hunter hours and population size on the Delta Marsh during 1 9 7 3 and 1 9 7 ^ . 63 Figure 21. Relationship between hunter numbers and population size on the Delta Marsh during 1975, 1976 and 1977-Table XII. Crippling losses in relation to range from hunters' blinds on the Delta Marsh ( 1 9 7 3 , 1 9 7 4 , 1 9 7 7 ) • <35 meters >35 meters Total Year Percent Cri ppled Number Number of Downed Kil led Crippled Birds Percent Crippled Number Number of Downed Kil led Crippled Birds Number Number Kil led Crippled Percent Cr i ppled of Downed Bi rds 1973 1974 1977 1973, 1974 Total 1973, 1974, 1977 Total 65 27 29 .3 21 27 56.3 86 54 38 .6 41 15 26 .8 13 12 48.0 54 27 33.3 23 9 28.1 106 42 28 .4 34 39 53.4 140 81 36.7 163 90 35.6 6 5 CHAPTER FOUR DISCUSSION OF FIELD RESULTS Hoi ling ( 1 9 5 9 b ) pointed out that subsidiary components of predation such as prey characteristics affect the basic functional response of predators to prey density. Avoidance learning by prey was suggested as one means whereby the number of prey consumed could be lowered as prey density rises (Hoi 1 ing, 1966). Avoidance learning is a means whereby ducks improve their survival rate in the presence of hunters. Melzack (1961), working with Mallard ducklings, found that birds readily habituate to model predators. Hochbaum et a l . , ( 1 9 5 4 ) found that field-feeding ducks learned to avoid "scarecrows" which were frequently moved about on fields where the birds were feeding. The effectiveness of these "scarecrows" was in some cases reinforced by gunfire. However, no ducks were k i l l e d . Thorpe ( 1 9 6 3 ) suggests that prey responses to predators are not instinctive, rather prey tend to avoid novel, high intensity or sudden st imuli . There are several explanations of how such learning may occur. Experience depends not only on age but also on the degree of exposure to hunting. Vulnerability was expected to decrease with cumulative experiences with gunners. I assumed at the outset that avoidance learning may be reinforced by gunfire; the more frequently ducks are exposed to gunfire, the more wary they become. This holds for white-tailed deer (Odocoi1 us  vi rg i n ianus) which become less vulnerable as the cumulative hunting pressure increases (Roseberry and Klimstra, 1 9 7 4 ) . However, I found contrary evidence which suggests that vulnerabil ity is related to current hunting effort . The most intensive hunting at Delta occurs early in the season; 66 clucks are most wary and least vulnerable immediately after opening day. As the hunting season progresses, however, my evidence suggests that vulnerability increases. On a fine scale Mallard and Lesser Scaup learn to recognize hunters as dangerous by direct experience with occupied blinds. Thus, I found that individuals and flocks soon related certain parts of the marsh with the sudden, loud, start l ing act ivity of hunters. My study revealed that shortly after the hunting season opened, these species avoided loafing and feeding sites near shooting stands and that birds were least vulnerable during the f irs t week of the season. On a broader scale some ducks learned to avoid hunted areas at Pasquia and Delta by moving to and by remaining in refuges or sites inaccessible to hunters. Therefore, the distribution pattern was non-random reflecting a general avoidance of certain areas. A biotelemetry study of marked birds in Minnesota revealed that waterfowl join in mixed species flocks which avoid hunting areas and restrict their act iv i t ies under heavy gun pressure (Kirby et a l . , 1976). My belief that vulnerability is related to hunter effort is supported by evidence from I l l inois where the total daily bag increased until it reached a point at which more hunters resulted in lowered or stationary daily bags (Bellrose, 1944). Bellrose attributed this to marshes having a carrying capacity for hunters and beyond this level there is so much interference and out-of-range shooting that the k i l l may even decline. Bellrose also found that the fewer the hunters, the greater was the individual success. Studies conducted in Newfoundland on Moose (Alces  a 1ces) (Bergerud et a l . , 1968) revealed that the success of hunters was inversely related to hunter density. 6 7 The analysis of vulnerabil ity may be summarized in relation to three models of k i l l (Table XIII). The f i r s t model assumes a random encounter process (linear model) which I found did not adequately describe k i l l patterns at Delta. A second model (Hoi l ing, 1 9 5 9 a , b) assumes that prey characteristics which reduce chances of capture (for example, avoidance learning on the part of ducks) and that predator characteristics such as "handling time" place constraints on total k i l l . This is the type 2 predation of Hoi 1ing ( 1 9 5 9 b ) . My data relating vulnerabil ity to hunter effort support the use of this model. A more in depth appraisal may be accomplished using the regression coefficients of the logarithmic model (Table XIII). For Mallard was signif icantly less than one (t = 5 - 9 , df = 7 , p<.05) indicating a non-linear relationship between k i l l and population size. This implies that hunters search non-randomly and birds are distributed contagiously or that there is a significant handling time effect. Thus the time lost per encounter is great enough to limit the bag. This would be especially true i f ducks f ly only for short periods each day. The regression coefficient B^ was not significantly different from one (t = . 5 4 , df = J, p>.05) suggesting no interference among hunters. This suggests that my previous finding relating vulnerability to hunter effort may be erroneous and that increased vulnerability through the season is due to some intr ins ic change in bird behavior or else some effect associated with decreasing population s ize . For Lesser Scaup B^  was less than one but not s ignif icantly so (t = 1 . 1 8 , df = 7 , p > . 0 5 ) . This suggests there may be some handling time effect associated with k i l l i n g Lesser Scaup or that hunters search non-randomly for contagiously distributed birds although not s ignif icantly so. The regression coefficient B_ for Lesser Scaup was greater than one 68 although not s ignif icantly so (t = 1.572, df = 7, p>.05). This suggests that hunters may fac i l i ta te one another, perhaps by periodically scaring up bi rds . For Canvasback the regression coefficient was signif icantly less than one (t = 6.k, df = 5, p<.05) suggesting a handling time effect or non-random search associated with clumped distribution of ducks. The coefficient B^ was not different from one (t = 1.3, df = 5, p>.05) suggesting no hunter interference. Overall , the regression coefficients indicate that vulnerability may be a function of intr ins ic behavior such as flocking due to increased pre-migratory food requirements. The results of this analysis indicate that hunters did not interfere with each other. For Lesser Scaup, hunters may fac i l i ta te one another by inducing birds to move about following interferences resulting from various phases of hunting ac t iv i ty . Thus, I conclude that avoidance learning, as observed at Delta, is a short-term behavior associated with a fright syndrome induced by the high levels of hunter effort early in the season. The s tat i s t ica l model of the bag check data support the hypothesis that bird distribution and non-random search patterns by hunters result in a saturating relationship between k i l l and population size. My analysis of the spy blind data suggest mean time between encounters is independent of density due to a behavioral refractory period after each shot f i red . This would contribute to non-linear harvests and result in achieved bags being less than management limits as was the case at Delta. 1 suggest that handling time per bird ki l led does not const rain hunters and has no influence on achieved bag size. 69 Table XIII. Possible influences of independent variables on the k i l l of ducks as predicted by three models. Model Type Hoi 1i ng Total k i l l Variables Linear model prey model logarithmic model Encounter or K i l l Rate Slope constant over time Encounter rate not constant antilog regress ion X Populat ion Size Ki 1 1 proport iona1 to duck abundance Other species present i nf1uence handli ng time B-J< 1 impl ies (1) bird clumping (2) handli ng time effect Hunter Effort Ki 1 1 proport iona1 to hunter Vulnerabi1i ty dependent on effort B2>1 implies hunters faci1i tate one another numbers B 2<1 implies mutual interference among hunters 70 The zone of greatest danger to ducks Is within 3 5 m of hunters' guns. I found considerable variation in the approach of birds to the hunter and his decoys. This was variable by species. Mallard, the most wary variety at Delta, are abundant throughout North America. Canvasback, less wary and more easily k i l l e d , have declined in recent years. I suspect that specific differences in response to hunters may be one important reason for the relative success of the Mallard. Stott and Olson ( 1 9 7 2 ) attribute differences in numbers of three species of sea ducks along the New England coastline to variations in vulnerabil ity to hunters. While most hunters believe that ducks are readily attracted to decoys during windy, wet, cloudy days, I found l i t t l e correlation between quantitative measures of weather and vulnerabi1 i ty. Wind speed affecting the opportunity to k i l l Mallards was one significant relationship. This differs from the conclusions of Koerner et a l . , ( 1 9 7 4 ) who found the k i l l of Canada Geese (Branta canadensis) to be s ignif icantly correlated with barometric pressure and temperature. My findings suggest that fa l l weather has l i t t l e influence on the vulnerabil ity of the three species studied. The varying nature of f l ight between species may have some influence on hunter se lect iv i ty . For instance, i found that hunters were less l ikely to shoot at Green-winged Teal , Blue-winged Teal and Bufflehead than the larger ducks. Boyd ( 1 9 7 1 ) reported similar findings for small ducks in eastern Canada. I believe there is an element of surprise favouring these smaller ducks. They catch a hunter's eye less quickly and their maneuverability allows the smaller birds to swing suddenly to the decoys. Hunters infrequently fired upon ducks making abrupt approaches or coming from behind. 71 The density and quality of alternate food (Leopold, 1 9 3 3 ) and the willingness of the predator to exploit other potential prey are important variables affecting the k i l l of one prey species by a predator. Leopold ( 1 9 3 3 ) referred to alternate foods as "buffer species" and thus the functional response may be lowered when the kinds and palatabi l i ty of "buffers" is increased (Hoi l ing , 1 9 6 6 ) . Therefore, hunter se lect iv i ty has an important effect on size and species composition of the duck k i l l . I found a significant•difference between select ivity for large versus small ducks. This may be the result of choice on the part of hunters or may be the outcome of mechanical (size, f l ight speed and maneuverabi1ity) and behavioral differences among species. 1 have no evidence, however, that small ducks serve as "buffer" species. In my study of hunter se lect iv i ty , hunters shot indiscriminately even when certain species such as Canvasback were protected by restr ict ive regulations. Geis and Crissey ( 1 9 7 3 ) found this to be true in their study of the point system in the United States. There, the legal daily bag depends on numerical values allotted each species and, in some kinds, sex, according to its rar i ty . The k i l l i n g of one Canvasback or Redhead, for example, l imits the hunter to not more than one additional duck in his daily bag, i f he had not k i l l ed other ducks during the hunt. Legally the hunt must terminate i f other birds had been k i l l e d . These authors found that over 9 0 percent of fl ights of Canvasback and Redhead were fired upon. Hunter se lect ivi ty has implications for management. Many states and provinces attempt to reduce the k i l l of certain waterfowl by limiting the number of such species allowed in the daily bag. I found, however, that hunters do not refrain from shooting ducks,.such as Canvasback, theoretically protected by game laws (Hochbaum and Caldwell, 1 9 7 7 ) . Most 72 hunters apparently are unable to identify waterfowl in f l ight to the degree needed for species management (Evard, 1 9 7 0 ) . My findings on the probability of mortality in relationship to flock size support those of Olson (1965) and Boyd ( 1 9 7 1 ) who demonstrated that the probability of an individual being shot decreases with increasing flock size. Powell (1974) suggested that membership in a large flock was an effective anti-predator strategy. Powell's ( 1 9 7 4 ) work with Starlings (Sturnus vulgaris) demonstrates that large flocks detect predators more quickly than do single birds. Siegfried and Underhill ( 1 9 7 5 ) drew a similar conclusion for Laughing Doves (Streptopelia senegalengis) . In observing flocks of ducks I noticed that when one bird f lared, the entire flock would also f lare, although a l l may not have noticed my presence. This behavior is e l ic i ted through social fac i l i ta t ion and has high survival value. In f lar ing, the birds turn quickly and f ly away from potential danger, changing direction and position in the a i r , thus presenting a more d i f f i c u l t target. My overall crippling loss of 3 5 . 6 percent is somewhat higher than rates reported for eastern Canada. Boyd ( 1 9 7 1 ) , observing hunters in the f i e ld , found a crippling loss of 24.8 percent. This rate is similar to the 2 2 . 5 percent cripples reported by Bellrose ( 1 9 5 3 ) who surveyed hunters in the United States. In a more recent study conducted in the United States, Geis and Crissey ( 1 9 7 3 ) reported a loss of only 16.0'percent-. However, Sowls ( 1 9 5 5 ) questioned experienced hunters on the Delta Marsh in the late 1 9 4 0 ' s and found 36 percent crippled ( 3 9 4 birds lost) in a bag of 7 0 5 ducks, closely agreeing with the 3 5 . 6 percent cripple loss I found in my study. In summary, I was able to reject the hypothesis that vulnerabil ity is related to cumulative hunting ef fort . In contrast I found that vulnerability was a function of current hunting ef fort . The results were 73 not def init ive , however, since vulnerability increased as the season progressed and populat ions fel1. I suggest that further research may be required to evaluate the effects of pre-migratory foraging requirements, age, and migration patterns on vulnerabil i ty . This would require marking of individual birds. I found differences in vulnerability by species as was predicted. I conclude that hunters search non-randomly, since k i l l was not a linear function of duck abundance. I found that handling time per bird shot or crippled is not high enough to limit the daily bag and that time between encounters is independent of population size resulting in a saturating k i l l pattern. I demonstrated that hunter effort is correlated with bird abundance but was not able to reject the hypothesis that this may only be related to tradit ion, cold weather or loss of interest. Hunter interviews might reveal more about how hunters allocate effort . Contrary to most hunters' belief , I found that vulnerabi1ity is not related to local weather conditions. In opposition to current species management policy, I found that hunters were not selective even when required to be by law. Several untested assumptions were inherent in the study because I had limited resources, unmarked birds and, in some instances, small sample sizes. In several calculations this required that species be pooled whereas other problems associated with technical aspects such as surveys are necessarily assumed to be constant. For example, in the calculation of the vulnerability coefficient for any week, I assumed that the population size was uniform throughout the week. In addition, the range of birds from my experimental blinds could only be estimated as being less than or greater than 35 m. I assume the error associated with distances is constant since estimates were made by the same observer during both years. The coefficient 74 is also assumed to be independent of population size and only reflects behavioral differences during the study period. The risk of a bird passing a hunter is also thought to be homogeneous throughout the time for which the coefficient was calculated. The results are also conditional on effects of habituation and age ratios. I assumed birds did not habituate to my blinds since we provided negative reinforcement during experimental hunts. Both habituation and age ratio factors could only be incorporated into the vulnerability coefficient i f the population was marked and the status of responding birds known. In the calculation of hunter effort I assume constant effort per bird shot during each survey week. The results are also conditional on the assumption that sample sizes are large enough to assume a homogeneous hunter population. The hunter effort is conditional on the techniques being uniform from year to year and that the calculated values reflect an index of hunter effort . In the calculation of crippling losses and probability of mortality, it was necessary to pool species because of small sample sizes. Therefore, I assume that crippling losses and mortality are species independent. Several problems were associated with estimating waterfowl abundance from the a i r . Replication of counts was not possible so that there is no estimate of sampling precision. It is assumed that a l l birds along the transects are actually seen, so there is no downward counting bias. I believe that the survey is a representative index of species present. In some cases a l l ducks could not be recorded as to species and adjustments were made assuming the rate of unidentified ducks is constant with respect to species. In addition, with unmarked birds the aerial survey gave no indication of the rate of turnover of the population. 75 CHAPTER FIVE MATHEMATICAL MODEL FOR A SINGLE HUNTER Background to the Model It is apparent that birds and hunters interact in a complex fashion through each hunting season. Simulation modeling is one means to capture some of this complexity, in order to address such questions as: If vulnerability decreases with increasing effort , is i t necessary to regulate hunter effort at a l l and, i f so, how can particular target harvests be achieved? What are the harvest implications for any one species with changes in the abundance of other species, given that hunting effort may respond to combined bird densities whereas k i l l i n g birds of other species may compete for the hunter's time and thus tend to protect the one species? The f ield results presented above wil l be essentially misleading and useless in a management context unless they can be tied together in an overall predictive framework. A mu1t?-species model is necessary for several reasons. From the standpoint of the manager, it would be ideal to harvest each species separately, since the yield would be greater than i f a l l are harvested at the same rate (Ricker, 1958). In addition, Ricker pointed out that i f several species with differing characteristics are harvested at one rate, one species may become relatively more abundant whereas others may decline. However, several characteristics of ducks and hunters make yie ld management for individual species unattainable. F i r s t , duck populations in the fa l l consist of many species and hunters shoot indiscriminately. For this reason it is almost impossible to form regulations around one species. Ducks also have species-specific vulnerabil i t ies and behaviour patterns. 76 We may desire a high yield of Mallard in a certain area but i f Canvasback are also present, the k i l l of the latter species wi l l be large because of their greater vulnerabil i ty . Holling (1959b) developed a single species model of the functional response of a predator to prey density which predicts that as prey density increases, the number of prey taken per unit time wil l rise at a decelerating rate to a threshold level (Type 2 predation, Hol l ing, 1959b) (Figure 22). This management analysis examines Holling's model in relation to multi-species waterfowl harvests using f ie ld data I collected on the Delta Marsh. Holling's equation was well suited to my data since it could account for the k i l l rate per hunter as a function of bird abundance. Other responses which Royoma (1970) added to the model were: (1) the allocation of hunting effort between areas and periods of differing prey density and (2) mutual interference of predators as a function of prey density. These other features should be incorporated in a duck harvest model since hunting effort is postulated to be correlated with bird density, and mutual interference may be evidenced in decreasing vulnerability when hunter effort is high. Quantifying the Model The model assumes that, in the absence of regulations, the bag per hunter hour wi11 increase with duck density up to a point at which the k i l l reaches a limit due to handling time (Figure 22). The relationship to describe the k i l l rate per hunter as a function of duck density is derived from Holling's (1959a) disc equation: 77 Figure 22. Functional relationship between potential k i l l per hunter hour and duck density. 77a DENSITY where: 7 8 N c = prey caught N 0 prey density a = encounter rate of predator with prey, times total time per hunt times probability of k i l l per encounter b = handling time per prey encountered, times a This equation is for a predator which consumes one prey type. Now, i f there are n prey species, the mu1ti-species analog for equation ( 1 ) is (Charnov, 1 9 7 3 ) : N = T E X . s I i = 1 T +T. s h (2) where: Now, where: - .th X. = encounter rate of i prey species T g = total time spent searching for prey T, = total time spent hand!ing prey. T u = EX.T h. h i s i h. = handling time for each prey individual 1 of the i t n species so N = T E X . s I i = 1 T +E X.T h. s i s I n E X. I i = 1 ( 3 ) 1+ E X.h. i i 79 but: where: X. = a.N. i i i a. = the vulnerabil ity coefficient for the 1 i t n species, times probability of mortality per encounter for species i times select ivity for species i . t h N. = density of i species. Now, N c I a.N. i i i=1 (h) n 1+ Z a.N.h. i i i i=1 If h. is the handling time for one individual of prey species i then 1/h. wi l l be the maximum rate of k i l l i n g of species i . Let 1/h.=m, the maximum potential k i l l . Then, I a.N. H - -1=1-1!— (5) c n 1+ I a.N./m. i i i i = 1 Now assume that m. and h. are the same for a l l species of ducks, then I I n S a.N. N • ' ' ( 6 ) c n 1+(Z a.N.)/m I I i=l 80 and for the i s p e c i e s N . a i N i (7) i c = n 1+(E a.N.)/m i i i=1 The a. values in this equation are the vulnerability coefficients times probability of mortality times hunter select ivity as described in results (Tables V and X). The vulnerability coefficients (a.) vary over time in relation to hunter effort so the net response cannot be seen by entering a single set of numbers into the equation. The value of m is taken from hunter observations and is the maximum k i l l per hunter hour t h (4.0 from Table V l ) . N. equals the k i l l rate for the i species per hunter hour. 81 CHAPTER SIX MANAGEMENT ANALYSIS AND IMPLICATIONS OF FINDINGS Applying the Model to the Delta Marsh Additional Assumptions and Structure The multi-species disc equation (7) calculates k i l l on an hourly basis for one hunter. I incorporated this function into a simulation model which also includes total hunting effort and population size changes. The simulation then forecasts yields by weeks of the season for Mallard, Canvasback and Lesser Scaup. Inputs are the number of weeks to simulate, maximum k i l l rate, i n i t i a l population sizes, species-specific probabil it ies of mortality and hunter se lect iv i ty . Number of hunter hours is calculated as a function of total population size each week employing the equation: E = B + .065 D (8) where: E = effort at timet D t = total population size at timet B = effort level when population size is zero. The rationale for using this equation comes from my finding that effort was correlated to population size. Init ial efforts can be varied by changing the B value and thus alternative assumptions about hunter effort responses can be used to drive the system. Vulnerability for Mallard and Lesser Scaup is computed dynamically as a function of hunter effort using the linear equations from my f ie ld data analysis: 82 a = .00264 - .0007 (Log Effort) (9) a 2 = .0142 - .0042 (Log Effort) (10) where: a.| = Mallard vulnerabil ity a 2 = Lesser Scaup vulnerabil ity Canvasback vulnerabil ity showed no correlation with measured variables and is assumed constant. The value is entered using the mean vulnerabil ity for the six weeks of hunting during 1973 and 1974 (44.4 x 10 ) . Weekly k i l l s for each species are calculated by multiplying the estimated week's effort by the hourly k i l l for each species as predicted by the disc equation (7). Since no estimates of migrational changeovers are available, I assume a closed system of sub-populations in which the k i l l is removed and the population updated for the next week by the equation: N't+1 = N ' t " K ' t where: Ni = population size in week t for species i K i t = k i l l (including cripples) for species i in week t. Thus the model dynamically considers effort , population size and vulnerability and accounts for harvest losses, interference of hunters and avoidance learning as well as the components of se lect ivity and species-specific probability of mortality. Management Options The model was modified to incorporate my f ie ld data with the objective of providing a general evaluation of alternative harvest strategies and thereby gain insight into the complexities of waterfowl hunting mortality. 83 The intent was not to make precise numeric predictions because population levels cannot be forecasted between weeks. A series of simulations were accomplished for varying season lengths, effort levels and bird abundance in hopes of appraising the effects of the following management options on each of the three species studied: (1) management practices for dealing with changes in potential hunter effort; (2) the influence of abundance on k i l l ; (3) the effect of season length on harvest. Bag l imits , commonly thought to control harvests, were not dealt with in the analysis since I found that few hunters achieve legal daily l imits . On a regional scale, bag limits may affect the numbers of hunters who go a f ie ld , acting as an incentive to hunters as may be the case with squirrels (Christisen, 1971) • However, I had no data on numeric responses of hunters to bag limits on the Delta Marsh and therefore could not incorporate this possible response in my model. Results and Discussion The results of simulated management options demonstrate that each species has its own response to hunter effort and season length even though similar factors are influencing each sub-population at the same time (Figures'23 through 31). These variations in k i l l pattern are due to differences in behavioural (vulnerability) adaptations which are species-specific. Mallard, under various harvest regimes, showed highest k i l l associated with the longest season length and the highest population size (Figures 23, 2k and 25)• Lengthy seasons are needed to obtain intermediate k i l l s at 84 lower population levels. In contrast, shorter seasons require a large number of birds to achieve similar intermediate yields . The management analysis reveals that Mallard respond well to hunting pressure since excessive k i l l s are never achieved under any combination of season length and population size. For example, with the longest season and highest population, only about one-third of the birds are harvested. Under harvest strategies involving short seasons, the k i l l of Mallard remains small even when ducks are abundant (Figure 23). The most sensitive situation is when the population exceeds 15,000 and the season is greater than six weeks in length. With a season constraint of four weeks or less, the k i l l is not strongly influenced by population size. When hunter effort is reduced, the overall response remains similar to that observed under normal hunting conditions (Figure 23). K i l l s differ between low and normal hunter pressure scenarios only in that the harvest under lowered hunter pressure is s l ightly smaller than the k i l l under normal hunter effort at similar densities and season lengths (Figures 23, and 24). The k i l l remains unchanged from normal when hunter effort is increased (Figure 25). In summary, Mallard harvests at Delta are strongly influenced by population size and season length. Mallard survival response is high in the face of large increases in effort with k i l l s remaining relat ively constant. This is due, in part, to a behavioral adaptation whereby individuals increase their wariness in response to higher numbers of hunters. Reductions in the amount of hunter effort wi l l produce only small declines in the overall k i l l . This results from an increase in the vulnerability of Mallard when the number of hunter hours lessens, thereby contributing to a corresponding increase in the k i l l per hunter hour. The 85 Figure 23. Isopleth diagram showing the relationship between regions of Mallard k i l l and population size and season length under normal effort levels observed at Delta. I n i t i a l P o p u l a t i o n S i z e X 1 0 0 Q B98 86 Figure 2h. Isopleth diagram showing the relationship between regions of Mallard k i l l and population size and season length under low effort levels. 87 Figure 25. Isopleth diagram showing the relationship between regions of Mallard k i l l and population size and season length under high effort levels. 88 most appropriate strategy for maintaining reduced k i l l s is to shorten the season since yields wi l l always be small. The season should be lengthened when large harvests are desired. In contrast to the Mallard, which showed largest k i l l s at highest densities, Lesser Scaup demonstrate small k i l l s when numbers are large. Greatest k i l l s occur at intermediate densities (Figures 26, 27 and 28). This is due to the effect of population size on hunter effort and the subsequent depressing effect which hunter effort has on Lesser Scaup vulnerabil i ty . A key, and untested, assumption here is that hunters ignore one another; the modelled effort response is to bird density (equation 8) rather than to realized k i l l per effort . Highest yields are attained in the region of 15,000 birds and a 10 week season under normal hunter effort levels regardless of season length. Adding additional weeks on the season does not influence the k i l l greatly at low or high densities (Figure 26). After 6 weeks, additional lengthening is very important at intermediate population levels (Figure 26) . Under a pattern of reduced effort the k i l l of Lesser Scaup became larger than was observed at normal effort levels (Figure 27). This is due to vulnerabil ity increasing in response to depressed hunter effort levels. In comparison, however, the k i l l was reduced under a pattern of increasing hunter effort (Figure 28). In conclusion, the k i l l of Lesser Scaup under various harvest tactics is strongly influenced by both population size and hunter effort . In contrast to the Mallard, the k i l l does not always increase as the population becomes larger. Season length is of lesser importance than with Mallard and Lesser Scaup are able to respond to large increases in hunter effort to the extent that k i l l s may decline as gun pressure increases. 89 Figure 26. Isopleth diagram showing the relationship between regions of Lesser Scaup k i l l and population size and season length under normal effort levels observed at Delta. 90 Figure 27. Isopleth diagram showing the relationship between regions of Lesser Scaup k i l l and population size and season length under low effort levels . 91 Figure 28. Isopleth diagram showing the relationship between regions of Lesser Scaup k i l l and population size and season length under high effort levels. 92 Special regulations to adjust k i l l are not necessary in the overall management of Lesser Scaup at Delta. In Canvasback, the k i l l is largely dictated by population size, as was the case with Lesser Scaup (Figures 29, 30 and 31) . In comparison to Lesser Scaup, however, the k i l l becomes proportionately larger as the population increases (Figure 29). The k i l l of Canvasback is large even i f seasons are very short (Figure 29, 30 and 31). If the population is small, a long season wi l l produce a similar k i l l to a short season (Figure 29). At high densities, a short season wi l l produce a large k i l l (Figure 29). Reducing hunter effort levels from a 1973 to 1977 baseline does not depress the Canvasback k i l l as much as might be hoped or expected intuit ively (Figure 30). In contrast, increasing the hunter effort wi l l increase the k i l l especially at lower Canvasback densities (Figure 31). In comparing Canvasback with Mallard and Lesser Scaup, it is noteworthy that unlike the latter species, Canvasback are not able to respond in their favor to changes in gunning pressure. In addition, lengthening the season does not greatly influence the k i l l of Canvasback since large portions of the population are k i l l ed even with short seasons. K i l l levels of Canvasback are largely a function of population size. In managing Canvasback, a species which requires special protection because of its small continental population, very short seasons (less than 2 weeks) are required when populations are large. However, when the population is small, (5000 range) season length should not be shortened to protect Canvasback since the k i l l is small irrespective of season length. A shortened season designed to protect Canvasback would simply cause a loss of hunting opportunity for more adaptable species like Mallard and Lesser Scaup. 93 Figure 29. Isopleth diagram showing the relationship between regions of Canvasback k i l l and population size and season length under normal effort levels observed at Delta. 93a I n i t i a l P o p u l a t i o n S i z e X I 0 0 0 3h Figure 30. Isopleth diagram showing the relationship between regions of Canvasback k i l l and population size and season length under low effort levels . 95 Figure 31. Isopleth diagram showing the relationship between regions of Canvasback k i l l and population size and season length under high effort levels. 96 Management Imp 1ications of Findings Ducks, geese, and swans, the Anatidae, comprise a major natural resource of North America. Of the Anatidae, the Anatinae, are the most numerous, their numbers ranging from 35,000,000 to 140,000,000 during the past four decades. Twenty-seven species of ducks share Canada, the United States and Mexico. Their major breeding grounds are in Canada where some range far north in the Arct ic . Most winter in the southern United States and Mexico, although some reach Cuba and South America. Their movements between north and south take place in massive spring and f a l l migrations. There are many stopping or staging areas along migration routes and it is on these sites that much hunting takes place. Some members of the Anatinae are hunted in a l l of Canada, in a l l the United States and in many parts of Mexico. Hunting begins as early as September 1 in Canada followed by successive opening-of-shooting dates and staggered shooting seasons until a l l legal gunning f inal ly ends in February on the most southern wintering grounds. In the early 1970's, the hunting k i l l in North America was estimated at 20,000,000 annually (Bell rose, 1976). Because of the succession of opening dates arranged to meet the oncoming autumn migration, some populations are subject to gunning during five consecutive months. This holds especially for the Mallard and other dabbling ducks as well as the diving ducks including Lesser Scaup. The Mallard is probably the most common duck in the world although populations have declined recently on some breeding ranges of North America. The Canvasback, a once common breeder on prairie marshes has declined in abundance over the past 25 years. The Lesser Scaup remains one of the most plentiful game ducks with an extensive breeding range which includes pra ir ie , parkland, and boreal forest. The total duck population has 97 suffered a decline during the past 25 years perhaps due to increased hunting pressure and declining habitat (Hochbaum and Caswell, 1978). Throughout North America, waterfowl hunting regulations are established by federal governments, with provinces and states defining their own patterns of control within the overall frameworks. The control of harvest in a l l countries is attempted by way of restrictions on the number of birds a hunter may bag in one day (bag limit) and the number of days during which hunting is allowed within a given zone (the open season). This practice has been in operation for some 65 years. In the beginning, bag limits were the same for a l l species within a given country. As some species declined, however, it became the management practice to reduce the bag limit or to disallow shooting altogether for a given kind, as Canvasback and Redhead. As such specific restrictions went into practice, rules were also made allowing larger bags for more plentiful species, as Lesser Scaup. In most instances, management assumed that the hunter could identify a l l species in a l l plumages under f ie ld conditions when v i s i b i l i t y was often hampered by low light and bad weather and by the hunters' requirement to remain concealed. Waterfowl managers generally assume that bag limits and season length can be used to manage the harvest and that regulations involving these variables be based on population parameters such as stock size (Bellrose, 1950; Geis, 1959; Geis, 1963; Geis et a l . , 1969; Nagel and Low, 1971; Rogers et a l . , 1979; and Patterson, 1979)• As a result regulations vary annually, regionally and specif ical ly according to population data available from the breeding grounds each year. These data for the annually varied regulations are gathered by purely empirical methods. These population figures derived from aerial and on-the-ground surveys vary according to the 9 8 experience and training of survey members, as well , upon differential head office interpretation. Assembly of effective restrictions is further complicated by the fact that regulations must be set well in advance of the hunting season due to limitations of administrative procedures. Despite the intensity of annual duck surveys and the complexity of the regulatory problem, the effect of variable regulations on survival has been studied in only the Mallard (Anderson and Burnham, 1 9 7 6 ; Anderson and Burnham, 1 9 7 8 ; and Rogers et a l . , 1 9 7 9 ) . Furthermore, there have been no studies of the effectiveness of regulations or of the complex interactions . between ducks and hunters. The latter must annually memorize new laws covering many species. In the f ie ld the hunter is confronted repeatedly by the requirement to make split-second decisions regarding the legality of each encounter with ducks. As many as 10 or 12 different species may come to his decoys in a given day. My study suggests that complex regulations do not effectively control the k i l l . I found that hunters do not shoot selectively. Although the hunter is under legal restraints to limit his k i l l of Canvasback, for instance, he nevertheless tends to shoot at a l l passing ducks regardless of species. Therefore species-variable bag limits on one shooting area cannot be effective. A species, as Canvasback, is not protected by a one-bird bag limit while larger bag limits are allowed for some other species using the same marsh. In the same vein, regulations aimed at increasing the take of a species, as holds for the Lesser Scaup and some other kinds in the United States, may jeopardize protected species (Hickey, 1 9 5 5 ) . Furthermore, I observed that each species responds differently to hunter effort . If Canvasback are present on a marsh where hunters seek 99 Mallards (with higher bag allowances), the k i l l in Canvasback can be greater than in Mallard. This is because of lesser wariness, hence greater vulnerability of Canvasback in the face of the non-selective shooting which I observed from spy blinds. K i l l in Canvasback is clearly correlated with local population size. I suggest that the only manner by which Canvasback harvest may be controlled is through area closures on a l l ducks when this species is present. In Manitoba this would involve the temporary closure of only a few hunting areas because of the Canvasback's traditional tendency to stage on certain , marshes. My management scenarios also demonstrate some of the not-so-obvious hunter-duck relationships when and where gunners pursue mixed stocks. Mallard k i l l , for instance, remains high in autumn when hunter effort is low. Wariness is not maintained throughout the hunting season. In Lesser Scaup, the k i l l may actually increase under reduced pressure. Both species cope behaviorally with large increases in hunter effort , as applies during the early part of shooting, but evasiveness declines as their encounters with hunters become less frequent as autumn advances. Clearly this does not apply with Canvasback. My analysis of the k i l l data, spy blind observations and decoy tr ia l s indicate that waterfowl harvests are influenced more by bird behavior and abundance than by current regulatory policy. These studies also reveal that hunter behavior is an unknown quantity that demands further investigation. I conclude that because of species-specific differences and because of hunter behavior and its effect on harvest, that regulations be simplif ied, hence more easily understood and enforced. My findings suggest that more 100 thought be given to area closures against a l l duck hunting when a given species in residence deserves protection. F inal ly , I contend that regional harvests can be forecast by using multi-variate equations such as the one developed in this study. A predator-prey simulation model such as the one presented here could be useful for in-depth analysis toward effective regulatory action. The use of models such as the ones discussed wi11 be practicable in the future because management schemes wil l become more refined due to increased numbers of hunters pursuing stable or reduced duck numbers. The effects of more sophisticated management programs wil l not be obvious and may be analyzed using simulation techniques as I have developed prior to policy implementation. 101 SUMMARY AND CONCLUSIONS 1. A conceptual duck-hunter model was developed and several hypotheses and predictions based on causal relationships among model components were formulated and tested under natural f ield conditions in Manitoba. 2. Vulnerabil i ty, an important determinant of k i l l , was found to be a function of instantaneous hunter effort and not to cumulative experiences with hunters. 3. Mallard were found to be most wary, whereas Lesser Scaup and .i . Canvasback proved to be most vulnerable. k. K i l l patterns were not l inearly proportional to population size indicating a non-random encounter process. 5. Handling time was not found to constrain achieved k i l l per hunter. 6. Time between encounters was found to be independent of duck abundance suggesting a refractory period per encounter which may have resulted in the non-linear (saturating) k i l l patterns I observed. 7. Hunter effort was correlated to time of season and duck numbers; however, this finding may be confounded by the fact that hunters simply lose interest as the season passes and winter approaches. 8. No correlation was found between selected weather variables and vulnerabil ity suggesting that weather has l i t t l e influence on the k i l l . 9. Hunters were not selective in their shooting although small ducks were fired upon less frequently than large ducks. 102 10. Northern Manitoba Mallard populations, consisting primarily of juveniles, were 6 times as vulnerable as aggregations mainly of adults at Delta in southern Manitoba. 11. Crippling losses were higher on the Delta Marsh than have been reported in studies conducted elsewhere in North America. 12. Distribution patterns of ducks were non-random with clumping in refuges at Delta during the 1973 hunting season. 13. A multi-species simulation model was developed in order to assess implications of season length, effort and population size on the k i l l of Mallard, Canvasback and Lesser Scaup. 14. Season length manipulations are recommended for reducing k i l l of Mallard whereas Lesser Scaup survival remains l i t t l e changed under increased hunter effort levels thus requiring-no special regulations. Canvasback k i l l s were large under a l l scenarios except those when duck numbers are small. Shortened seasons are recommended as a measure for control 1ing Canvasback harvests. 15. A logarithmic regression model is recommended for forecasting k i l l levels whereas the multi-species simulation model is suggested for evaluation of management options. 103 BIBLIOGRAPHY Alford, J .R. and E .G. Bolen. 1977. Differential responses of male and female Pintai l ducks to decoys. J . Wildl . Manage. 41(4):657-661 . Anderson, D.R. 1975- A population in a Markovian environment. A theory and example. Ecol . 56(6):1281-1296. Anderson, D.R. and K.P. Burnham. 1976. Population ecology of the Mallard: VI. The effect of exploitation on survival . U.S. Fish Wi ld l . Serv. Resour. Publ . 128. 66 p. Anderson, D.R. and K.P. Burnham. 1978. Effect of restr ict ive and l iberal hunting regulations on annual survival rates of the Mallard in North America. Trans. N. Americ. Wi ld l . and Natl . Res. Conf. 43:1 81 -186. Bailey, R.O. and B . D . J . Batt. 1974. Hierarchy of waterfowl feeding with Whi st 1 ing-Swans-. Auk. 91 (3) : 488-493 . Bellrose, F . C . , J r . 1944. Waterfowl hunting in I l l ino i s : its status and problems. Nat. Hist . Survey Biol . Notes No. 17. 33 p. Bellrose, F .C. 1950. Mississippi Flyway: Problems, projects, prospects. Trans. N. Americ. Wi ld l . Conf. 15:123-132. Bellrose, F . C . 1953- A preliminary evaluation of cripple losses in waterfowl. Trans. N. Americ. Wi ld l . Conf. 18:337-360. Bellrose, F .C. 1976. Ducks, geese and swans of North America. Stackpole. Harrisburg, Pa. 543 p. Bellrose, F . C , I.G. Scott, A.S. Hawkins and J . B . Low. 1961. Sex ratios and age ratios in North American ducks. 111. Nat. Hist . Survey Vol. 27, Art i c l e 6. 474 p. i / Bergerud, A . T . , F. Manuel and H. Whalen. 1968. The harvest reduction of-a moose population in Newfoundland. J . Wi ld l . Manage. 32(4):722-728. Beverton, R .J .H. and S . J . Holt. 1957. On the dynamics of exploited fish  populations. Her Majesty's Stationery Office. London. 533 p. Boyd, H. 1971- Observations on duck hunting in eastern Canada in 1968 and 1969. Can. Wi ld l . Serv. Occasional Paper No. 12. 22 p. Brown, G.M., J r . , J . Hammack and M.F. Tillman. 1976. Mallard population dynamics and management mode 1s. J . Wi ld l . Manage. 40(3):542-555. Charnov, E . L . 1973- Optimal foraging: some theoretical explorations. Ph.D. Thesis. U. of Washington. 95 p. Christisen, P.M. 1971. Importance of daily bag in squirrel hunting. Trans. N. Americ. Wi ld l . Conf. 36:322-329. 104 Coward.n, L.M. and D.H. Johnson. 1979- Mathematics and Mallard management. J . Wildl . Manage. 43(1):18-35. Evard, J .O. 1970. Assessing and improving the ab i l i ty of hunters to identify flying waterfowl. J . Wi ld l . Manage. 34(1):114^126. Geis, A.D. 1959- Annual and shooting morta1ity estimates for the Canvasback. J . Wildl'. Manage.' 23(3) :253~261 . Geis, A.D. 1963- Regulations in migratory bird management. Trans. N. Americ. Wi ld l . Conf. 28:164-171. Geis, A.D. and W.F. Crissey. 1973. 1970 test of the point system for regulating duck harvests. Wi ld l . Soc. B u l l . 1(1):1-21. Geis, A . D . , R.K. Martinson and D.R. Anderson. 1969- Establishing hunting regulations and allowable harvest of Mallards in the United States. J . Wi ld l . Manage. 33(4):848-859• fc Hickey, J . J . 1955. Is there a sc ient i f i c basis for flyway management? . Trans. N. Americ. Wi ld l . Conf. 30:126-150. Hoi l ing, C.S. 1959a. Some characteristics of simple types of predation and parasitism. Can. Ent. 91 : 38'5~398. Hoi l ing , C.S. 1959b. The components of predation as revealed by a study of small mammal predation on the European Pine Sawfly. Can. Ent. 91:293"320. Holling, C.S. 1966. The functional response of invertebrate predators to prey density. Mem. Ent. Soc. Canada. No. 48. 1-86. Hochbaum, H.A. 1955. Travels and trad?tions of waterfowl. U. of Minnesota Press. 301 p. Hochbaum, H .A . , S.T. Dillon and J . L. Howard. 1954. An experiment in the | control of waterfowl depredation. Trans. N. Americ. Wi ld l . Conf. v 19:176-185. Hochbaum, G.S. and P . J . Caldwell. 1977. The 1973 k i l l of Canvasback under restrict ive hunting regulations, Delta Marsh, Manitoba. Can. Wi ld l . Serv. Progress Note No. 76. 6 p. Hochbaum, G.S. and F.D. Caswell. 1978. A forecast of long-term trends in breeding Mallard populations on the Canadian prair ies . Can. Wi ld l . Serv. Progress Note No. 90. 8 p. Holsworth, W.N. 1973- Hunting efficiency and white-tailed deer harvest. J . Wi ldl . Manage. 37(3):336-342. Kirby, R . E . , J . H . Riechmann and M.E. Shough. 1976. A preliminary report on Minnesota's innovative 1973 waterfowl season. Wi ld l . Soc. Bul l . 4(2):55-63. 105 Koerner, J .W., T.A. Bookhout and K . E . Bednarik. 1974. Movements of Canada Geese color-marked near southwestern Lake Er ie . J . Wi ld l . Manage. 38(2):275-289. Leopold, A. 1933. Game management. Charles Scribner and Sons, N.Y. 481 p. McKinney, D.F. 1975- The evolution of duck display. P. 331 to 357-G. Baerends, C. Beer and A. Manning (ed) . In Function and Evolution in  Behaviour, Clarendon Press, Oxford. Melzack, P.. 1961. On the survival of Mallard ducks after "habituation" to the hawk-shaped figure. Behaviour. 17:9 - 16. Nagel , J . E . and J . B . Low. 1971. Effect of regulations on waterfowl k i l l in Utah. J . Wi ld l . Manage. 35(3):554-557. Olson, D.P. 1965. Differential vulnerability of male and female Canvasbacks to hunting. Trans. N. Americ. Wi ld l . Conf. 30:121-135. Patterson, J . H . 1979. Can ducks be managed by regulation? Experiences in Canada. Trans. N. Americ. Wi ld l . and Natl . Res . Conf. 44:130-139. Powell, G.V.N. 1974. Experimental analysis of social value of flocking by Starlings (Sturnus vulgaris) in relation to predation and foraging. Animal Beh. 22(2)=501-505. Ricker, W.E. 1958. Maximum sustained yield from fluctuating environments, and mixed stocks. J . Fish. Res. Bd. Canada. 15=991-1006. Rogers, J . P . , J .D. Nichols, F.W. Martin, C.F. Kimball and R.S. Pospahala. 1979. Can ducks be managed by regulation? An examination of harvest and survival rates of ducks in relation to hunting. Trans. N. Americ. Wildl . and Natl . Res. Conf. 44:114-126. Royoma, T. 1970. Evolutionary significance of predator's response to local differences in prey density: A theoretical study. P. 344 to 357-P . J . denBoer and G.R. Gradwell (ed). In Proc. Adv. Study Inst. Dynamics Numbers Popul. Oosterbeek. Siegfried, W.R. and L . G . Underhil l . 1975. Flocking as an anti-predator strategy in doves. Animal Beh. 23:504-508. Sowls, L .K. 1955. Prairie ducks. Stackpole. Harrisburg, Pa. 193 P-Stott, R.S. and D.P. Olson. 1972. Differential vulnerabil ity patterns among three species of sea ducks. J . Wi ld l . Manage. 36(3):775-783. Teplov, V .F . and N.N. Kartashev. 1957. Wildfowl research in Russia. Wildfowl Trust. 9:157-169. Thorpe, W.H. 1963. Learning and instinct in animals. Mathuen. London. 558 p. 106 Walters, C . J . , R. Reed and J . Ward. 1973- Some factors derermining success of duck hunters in southern Bri t i sh Columbia. Can. Wi ld l . Serv. Progress Note No. 36. 6 p. Walters, C . J . , R. Hiborn, E . Oguss, R.M. Peterman and J .M. Stander. 1972*. Development of a simulation model of Mallard duck populations. Can. Wi ld l . Serv. Occasional Paper No. 20. 35 p. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0095027/manifest

Comment

Related Items