UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The numerical response of great horned owls to the snowshoe hare cycle in the boreal forest Rohner, Christoph 1994

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

Item Metadata

Download

Media
831-ubc_1994-953896.pdf [ 5.59MB ]
Metadata
JSON: 831-1.0088225.json
JSON-LD: 831-1.0088225-ld.json
RDF/XML (Pretty): 831-1.0088225-rdf.xml
RDF/JSON: 831-1.0088225-rdf.json
Turtle: 831-1.0088225-turtle.txt
N-Triples: 831-1.0088225-rdf-ntriples.txt
Original Record: 831-1.0088225-source.json
Full Text
831-1.0088225-fulltext.txt
Citation
831-1.0088225.ris

Full Text

THE NUMERICAL RESPONSE OF GREAT HORNED OWLS TO THE SNOWSHOE HARE CYCLE IN THE BOREAL FOREST by CHRISTOPH ROHNER Dipl. Zool., University of Zurich, 1985 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF ZOOLOGY We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October 1994 © Christoph Rohner, 1994 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. (Signature) Department of ^ ^ ^ ^ ^ The University of British Columbia Vancouver, Canada Date ff / /^/gf DE-6 (2/88) ABSTRACT Great homed owls (Bubo virginianus) are among the most opportunistic avian predators. In the subarctic boreal forest, their diet consists mainly of snowshoe hares (Lepus americanus), which show extreme population cycles with a 8-11 year period. The aim of this thesis was to study the population ecology of great homed owls in a cyclic environment, to investigate the components of the numerical response of this predator to its prey, and to evaluate the evolutionary context of the ecological processes involved. The study was conducted from 1989-92, with some data from 1988 and 1993, at Kluane Lake in the southwestern Yukon, Canada. During the increase phase of the snowshoe hare cycle, 86% of resident owl pairs bred and raised large broods of 2.4-2.6 fledglings per successful nest. Survival of young owls in their first two years of life was high, and two females were observed to breed as yearlings. Densities of territorial owls almost doubled to a maximum of 21-24 pairs/100km2 from 1988-92, but most owls that recruited locally became non-territorial 'floaters', presumably because social behaviour limited the number of territories. Floaters were silent, their ranges overlapped with those of territorial birds, and their density reached 40-50% of the total population. Snowshoe hares began to decline in the winter of 1990/91, and the number of great homed owls recruited in fall dropped from 1.7/pair in 1989-90 to 0.3/pair in 1991. Proximate causes of high pre-dispersal mortality included predation by mammals and parasitism by black flies (Simuliidae) and by Leucocytozoon ziemanni, a blood parasite transmitted by these flies. Post-dispersal mortality and emigration of resident owls also increased as hare densities declined further, floaters being affected before territorial birds. 1 1 Owl densities continued to increase after the hare peak and then declined with a time lag of one year for floaters, and two years for territory holders. Responses to brood size manipulations and food additions suggested that food was not super-abundant during the prey peak. I conclude that territorial behaviour is essential in causing time lags, with some birds monopolizing resources and conserving energy by ceasing reproduction. A review of hfe history variation in northern owls showed that great homed owls are constrained in their phenotypic plasticity to increase reproduction at high prey levels compared to some other species. Based on a comparison of evolution in the genera Bubo and Nyctea, I hypothesize that habitat-specific differences in the availability of prey during reproduction, and also in the mortality on age classes during bottlenecks, favoured the diversity of life histories in northern owls. I l l TABLE OF CONTENTS page Abstract ii Table of Contents iv List of Tables vii List of Figures viii Acknowledgments x Chapter 1: Introduction 1 Predator-prey interactions 1 Study species 2 Aims of this thesis 2 General approach 3 Chapter 2: The numerical response of great horned owls to the snowshoe hare cycle: consequences of non-territorial 'floaters' on demography 5 Abstract 5 Introduction 6 Methods 7 Population model 9 Results 11 Snowshoe hare densities 11 Reproduction 11 Survival 15 Age at breeding and occurrence of non-territorial 'floaters' 15 Movements and emigration 21 Estimating numerical responses 24 Discussion 31 Age of floaters and population dynamics 31 Movements and the threshold model of territoriality 32 Spatial heterogeneity, source and sink populations 34 Relevance to survival studies based on band recoveries 34 Large floating populations: broader implications 36 XV Chapter 3: Survival of great horned owls during their first year of life: black flies, blood parasites, and the paradox of high post-fledging mortality 39 Abstract 39 Introduction 39 Methods 41 Results 43 Nestling survival 43 Post-fledging survival 46 Patterns in juvenile survivorship throughout the first year 46 Causes of mortality 51 Mortalities related to human disturbance 51 Roost site selection 53 Black fly activity 56 Discussion 59 Interactions between mortality factors: does food shortage lead to cascading effects involving disease and predation? 59 Black fly activity and roost site selection 62 Roost site selection: trade-off between predation risk and parasite exposure? 64 Age-specific mortality rates 64 Why is the post-fledging stage a critical one in development? 66 Chapter 4: Brood size manipulations in great horned owls: are predators food limited at the peak of prey cycles? 69 Abstract 69 Introduction 69 Methods 71 Study area and techniques 71 Experimental design 72 Statistical procedures 75 Results 75 Weight changes in fledglings 75 Nocturnal movements of females 80 Diurnal nest attendance of females 83 Discussion 86 Magnitude of effects 86 Food hmitation and parental effort 87 Conclusions 89 Chapter 5: Great horned owls and snowshoe hares: what causes the time lag in the numerical response of predators to cyclic prey? 90 Abstract 90 Introduction 91 Hypotheses and predictions 93 Methods 94 V Results 96 Prediction 1: Population growth during the increase phase 96 Prediction 2: Use of alternative prey during the decline of snowshoe hares 99 Predictions 3a-d: Density-dependence and social behaviour 99 Discussion 102 Population growth and predator response 102 Multiple prey hypothesis 107 Density-dependence and territorial behaviour 108 Individual strategies and the time lag in the predator response 109 Conclusions 111 Chapter 6: Life history variation in northern owls 112 Abstract 112 Introduction 113 Results 114 Extremes in reproductive output 114 Mortality regimes 115 Variation of reproductive investment among northern owls 123 Behavioural, environmental, and phylogenetic correlates 130 Discussion 135 Constraints in phenotypic plasticity 135 Habitat and the diversity of life-histories 137 'Life history profiles' 139 Evolution of northern owls 142 Conclusions 143 Chapter?: Conclusions 145 Floaters and demography 145 Disease and survival 145 Food limitation at the peak of prey cycles? 146 Evolution of reproductive responses to cyclic prey peaks 146 Literature cited 148 V I LIST OF TABLES page Table 2.1 Breeding performance of great homed owls at Kluane Lake, 1989-92. 27 Table 2.2 Survival and emigration of great homed owls at Kluane Lake, Yukon, as determined by radio-telemetry from fall 1989 to fall 1992. 28 Table 3.1 Causes of mortality for juvenile great homed owls older than 35 days post-hatching. 52 Table 4.1 Experimental design for 4 phases of short-term brood size manipulations. 74 Table 5.1 Natural removal experiments and replacements of radio-marked territorial great homed owls at Kluane Lake, Yukon. 105 Table 6.1 Comparison of body size and some life history traits between great homed owls and snowy owls (means of both sexes). 118 Table 6.2 Comparison of reproductive parameters in northern owls in peak years of prey cycles in subarctic and arctic biomes. 124 Table 6.3 Comparison of behavioural and habitat characteristics of northern owls in prey cycles in subarctic and arctic biomes. 133 Table 6.4 Rank correlations between clutch size and behavioural and environmental variables (ranked 1-3 according to intensity). 134 V I 1 LIST OF FIGURES page Figure 2.1 Breeding performance of great homed owls and the snowshoe hare cycle at Kluane Lake, southwestern Yukon. 13 Figure 2.2 Occurence of juvenile great homed owls during two peaks of the snowshoe hare cycle in the Yukon. 17 Figure 2.3 Survivorship of adult owls (territory holders) and young owls (first and second year, floaters) based on radio-telemetry. 19 Figure 2.4 Weekly movements of territory holders and non-territorial •floaters'during 1989-1992. 23 Figure 2.5 Emigration of adult owls (territorial) and young owls (first and second year, floaters) based on radio-telemetry. 26 Figure 2.6 Numerical response of great horned owls (spring densities) to the snowshoe hare cycle. 30 Figure 3.1 Survivorship of nestling great homed owls 1989-91. 45 Figure 3.2 Post-fledging survival of juvenile great homed owls 1989-91. 48 Figure 3.3 Phases of differential mortality in great homed owls during their first year of life. 50 Figure 3.4 Seasonal trends in roost site characteristics (mean+SE) of great homed owls. 55 Figure 3.5 Black fly activity in different habitat positions. 58 Figure 4.1 Weight changes of owlets (deviation from initial weights, average per brood, +SE) when enlarged by one owlet (phase I, 1-10 June 1991). 77 Figure 4.2 Weight change of owlets (deviation from initial weights, average per brood, ±SE) when enlarged by two owlets. 79 Figure 4.3 Nocturnal position of females relative to their broods under treatments varying the level of food stress. 82 Figure 4.4 Daytime nest attendance of female great horned owls (average per nest, +SE, number of nests). 85 Figure 5.1 Annual rates of increase in populations of great homed owls and snowshoe hares. 98 V l l l Figure 5.2 Proportion of snowshoe hares (main prey species) in the diet of great homed owls at Kluane, Yukon, during the last increase phase and the beginning of the decline in the population cycle (means+SE). 101 Figure 5.3 Social behaviour and the limitation of population growth in great homed owls at Kluane, Yukon. 104 Figure 6.1 Frequency distributions of clutch sizes in large owls. 117 Figure 6.2 Mortality regime of great homed owls during varying snowshoe hare densities. 120 Figure 6.3 Clutch sizes in northern owls, as a function of body mass. 126 Figure 6.4 Trade-off between egg volume and clutch size in northern owls (with regression line and 95% confidence limits). 129 Figure 6.5 Asymmetry in the frequency distribution of clutch sizes in northem owls. 132 Figure 6.6 Hypothesized 'life history profiles' for an ancestral Bubo owl in two northern environments with their associated prey. 141 I X ACKNOWLEDGMENTS This study relied on the help of many people, and I would like to thank all of them to have made this project possible. Charley Krebs provided a constant background of experience and support, and had lasting faith in my work despite deviations of my predator focus onto the other side of the fence. Jamie Smith was there in critical moments, was a teacher in the mastery of editing ink, and shared his enthusiasm not only at the shores of Kluane Lake. Christoph ('Gonzo') Schmid helped to start up the fieldwork, and without his skill and friendship, the envisioned direction would not have been possible. Kat Russenberger, Baba Zimmermann, Cilia Kullberg, and Sam Wagniere also devoted a part of their life to helping with strenuous fieldwork on owls. Many thanks go to ('raptor') Frank Doyle, for his help, for his challenges how things can be done differently, and for growing together through a time of enthusiasm and rich experience. Many others helped, particularly with radio-telemetry and when a voice was needed to warn from attacking owl parents - special thanks to Johan Stroman, Troy Wellicome, Paul Heaven, and Brendan Delehanty. I was fortunate to obtain advice from G. Bennett and D. Currie on black flies and blood parasites, and Bruce Hunter's collaboration and enthusiasm lead me into new and exciting approaches. Dave Mossop (Yukon Territorial Government) was a source of experience and inspiration, helped with permits, and allowed me to use his unpublished data. I benefitted from discussions with Stuart Houston and David Andersen - they were both generous in sharing the findings of their research and improved earlier manuscripts with their comments. Dick Cannings provided advice for bone identification during pellet analysis and for field techniques. Tony Sinclair, Ron Ydenberg, Lee Gass, Carl Walters, and Peter Abrams were inspiring throughout the project and provided critical comments on thesis drafts. Erkki Korpimaki, Kathy Martin, and Ian Newton improved the manuscript with their comments. Special thanks to 'the cohort', particularly Dave Hik, Fritz ('Friday night') Mueller, Don Reid, David Ward, and David Westcott - for enlightening discussions on work and life, and for friendship in sharing the highlights and frustrations of field ecologists. I also thank my parents for understanding and encouragement throughout my studies. I was fortunate to be part of the 'Kluane crew', and I was helped in many ways by Irene Wingate, Vilis+Co Nams, Cathy Esser-Doyle, Sabine Schweiger, Mark O'Donoghue, and Carol and Andy Williams. Special thanks to Josie and Frank Sias, Jan and Mike WiUiams, Liz Hofer, Peter Upton, and the 'Breitenmoser connection'. This project was funded by the Natural Sciences and Engineering Council of Canada (grants to C.J. Krebs and J.N.M. Smith), a J.R. Thompson Wildlife Fellowship, a Graduate Fellowship by the University of British Columbia, and a MacLean-Fraser Summer Research Fellowship. CHAPTER 1 Introduction Predator-prey interactions Many animal species show cyclic fluctuations in numbers over time, and predator populations often follow these changes in synchrony or delayed with a time lag of 1-2 years (review in Krebs 1994). The 8-11 year cycle of snowshoe hare populations in boreal Canada and Alaska has been traced back for more than 200 years, and the system has been studied by ecologists for more than 50 years (review in Keith 1963, Krebs et al. 1992, Sinclair et al. 1993). The causes of these oscillations are still unknown, but the most likely explanations involve interactions of food shortage and predation (Keith et al. 1984, Krebs et al. 1992, Hik 1994). Parallel to snowshoe hares, a suite of predator species show similar population cycles. Among those, the regular oscillations of lynx (Lynx canadensis) (Elton and Nicholson 1942, review in Breitenmoser et al. 1993) and of great homed owls (Keith 1963, Adamcik et al. 1978, Houston 1987) are among the best known. In contrast to snowshoe hares, the cause for these cycles in predators is clearly related to food availabihty (reviews for birds of prey in Newton 1976, 1979), and their changes in numbers are therefore called 'numerical responses' to the prey cycle (review in Krebs 1994). Details about these numerical responses of predators are poorly known, but they may be a key element to understand the cyclic patterns of many vertebrate populations in subarctic systems (e.g. Keith et al. 1977, Erlinge et al. 1983, Angelstam 1984, Trostel et al. 1987, Korpimaki 1993). The Kluane Boreal Forest Ecosystem Project, which this study is part of, is investigating the causes and consequences of the snowshoe hare cycle in a collaborative effort involving three Canadian Universities (Krebs et al. 1992, Boutin et al. 1994). Study species Great homed owls (Bubo virginianus) are large, noctumal predators. They are mottled brown in colour, with powerful talons, a large beak, and yellow eyes. They owe their name to the prominent and erectile ear tufts on their head. They weigh 0.9-1.8 kg, and are among the largest North American owls (Earhart and Johnson 1972). Great homed owls are territorial year-round and usually nest in trees. In their diet, they show a wide variety of different prey species, although they prefer large prey such as lagomorphs (Donazar et al. 1989). They are the most widely distributed owl species in the New World, breeding in almost every habitat south of the subarctic tree line in North America to the southern tip of South America (Voous 1989). Aims of this thesis Three goals of this thesis are to: (a) identify the temporal pattern of the numerical response of great homed owls relative to the snowshoe hare cycle, (b) investigate the demographic components of this response, and (c) consider evolutionary consequences for predators that depend on cyclic prey populations. Within this framework, I investigate a number of more specific questions: Do the responses of non-territorial 'floaters' differ from those of territory owners, and does the inclusion of this non-territorial segment change the conclusions for the total population? (Chapter 2.) What are the demographic components of the numerical response in great homed owls? How do varying prey densities affect the reproduction (Chapter 2) and juvenile survival (Chapter 3) of great homed owls? What causes the time lag in the numerical response relative to the hare cycle? Is the decline in great homed owl numbers delayed after the hare peak because the main prey (hares) are still abundant, or because great homed owls shift in their diet and include other prey? (Chapters 4, 5.) General approach Great homed owls are difficult to study in northern environments. They are elusive and hard to locate in coniferous forests, they shift roost sites in a forest of abundant cover and prevent easy collection of pellets for diet study, they cannot be trapped readily, and in this habitat, their nests can be found only with great effort. Very limited road access imposes further logistical problems. Advantages facilitating the study of this species included frequent vocalizations of territorial owls, which are often distinct for individuals, and relatively low height of nest trees in the boreal forest. To overcome some of the limitations imposed by the study system, I used a combination of several approaches. On the broadest level, territorial owls were censused, and nest sites were located by a new technique (Rohner and Doyle 1992a). Broods were then monitored beyond fledging by applying an improved tethering method (Chapter 3). This allowed me to perform short-term brood-size manipulations and food addition experiments with minimal impact on survival of the study animals (Chapter 4). Radio-transmitters with a battery-life of two years made mortality estimates possible, and allowed exciting insights into the secretive life-style of juveniles beyond fledging and independence (Chapters 2, 3). A collaborative effort with D.B. Hunter (Veterinary College of Ontario, University of Guelph) provided detailed analysis of proximate causes of mortality (Hunter and Rohner 1994, Chapter 3). Some aspects relating escape behaviours from parasitism by black flies were further explored by mensurative experiments using Bantam chickens as bird bait (Chapter 3). Finally, features in the life-histories of predators of cyclic prey were evaluated by a comparative approach. How different are great homed owls from other northern owls in similar environments, and what causes some of the differences? These questions, which put the ecological approach used in Chapters 2-5 into a broader evolutionary perspective, are discussed in Chapter 6. CHAPTER 2 The numerical response of great horned owls to the snowshoe hare cycle: consequences of non-territorial 'floaters' on demography ABSTRACT The numerical response of great homed owls (Bubo virginianus) to the 10-year population cycle of snowshoe hares (Lepus americanus) in the boreal forest was examined during 1988-93 in the southwestern Yukon, Canada. Demographic parameters were estimated based on censuses (territorial pairs), nest visits (productivity), and radio-telemetry (survival, emigration, and integration of young birds into the population). Hares rose to peak densities in 1990, and almost all resident owl pairs bred and raised large broods during years of increasing and highest prey abundance. In 1991, the first year of hare decline, all breeding parameters including post-fledging survival were reduced, and recruitment in fall was very low. From 1992 onwards, reproduction was completely suppressed. Survival of young owls in their first two years of life was surprisingly high during the peak of the hare cycle, and a large number of non-territorial 'floaters' were present. These birds were silent, and moved more than territorial owls. Their ranges overlapped broadly with defended territories, and floaters were affected by the hare decline before territory holders. Most ecological studies on birds are based on the territorial and therefore visible fraction of a population. The results of this study show how a large proportion of floaters can delay the detection of population declines in traditional censuses of territorial birds, and can lead to serious underestimates of the impacts of predation. INTRODUCTION Several owl and raptor species in the northern hemisphere show strong numerical responses to population cycles of their prey (Newton 1979, Mikkola 1983, Korpimaki and Norrdahl 1991, Newton 1991b, Korpimaki 1992b). Great horned owls (Bubo virginianus) are large and long-lived predators feeding mainly on lagomorphs. They are territorial year-round, and are widely distributed across North and South America (Voous 1988, Donazar et al. 1989). Occasional irruptions of great homed owls into southern Canada and the northern United States are linked to the decline phase in the 10-year population cycle of snowshoe hares (Lepus americanus) in boreal Canada (Keith and Rusch 1989, Rusch et al. 1972, Mclnvaille and Keith 1974, Adamcik et al. 1978, Houston 1987, Houston and Francis 1995). In this paper, we examine the demography of the numerical response of territorial and non-territorial great homed owls by applying radio-telemetry as a means to directly measure essential demographic parameters at the same field site. The presence of non-territorial 'floaters', which in most cases live a secretive life and form a 'shadow population', is well known for some bird species and assumed for many others (Brown 1964, Watson and Moss 1970, Smith 1978, Newton 1992). Sometimes, such 'surplus' birds live in areas separate from breeding territories, and they may become directly observable when they form social groups (Charles 1972, Birkhead et al. 1986) or they may be detectable in open habitat (Watson 1985, Martin 1989, Hannon and Martin 1995). The evidence for a pool of non-territorial birds in most species, however, is indirect. When territory owners are removed, rapid replacement by birds other than known neighbours commonly occurs (Newton 1992). In at least 26 raptor species, such evidence has been reported in the literature (Newton 1979). New territory holders of unknown origin were also recorded in several owl species (Austing and Holt 1966, Hirons 1985, Franklin 1992). The size of the secretive floater pool has only rarely been measured directly (Smith and Arcese 1989, Matthysen 1989, Newton 1992, but see also Martin 1989, Hannon and Martin 1995). In this paper, we examine how juvenile great homed owls integrate into the breeding population, and we estimate the non-territorial fraction of the population based on direct measures of productivity, survival, emigration, and age at first breeding. Most ecological studies on birds are based on the territorial and therefore visible fraction of a population, but ignoring non-territorial birds may have considerable imphcations. Wilcove and Terborgh (1984) postulated that population declines may affect floaters first, and thus may not be discovered when censusing only the territorial segment of the population. Franklin (1992) demonstrated possible lags in detection time for a population decline in spotted owls (Strix occidentalis) depending on the unknown size of the floater pool. We show here that predator population sizes may have been seriously underestimated in traditional estimates, and our results indicate that even the qualitative pattern of population responses may have to be revised when non-territorial birds are taken into account. METHODS This study is part of the Kluane Boreal Forest Ecosystem Project (Krebs et al. 1992). The population data used here spans the years 1988-93, but most other data are from 1989-92. We worked at Kluane Lake (60" 57'N, 138° 12'W) in the southwestern Yukon, Canada. The study area comprised 350km2 of the Shakwak Trench, a broad glacial valley bounded by alpine areas to the north west and the south east. The valley bottom averages about 900m above sea level and is covered mostly with spruce forest (Picea glauca), shrub thickets (Salix spp.), some aspen forest (Populus tremuloides) grassy meadows with low shrub (Betula glandulosa), old bums, eskers, marshes, small lakes and ponds. Adult snowshoe hares were live-trapped during sessions of 5-6 days in March and April for spring, and in October and November for fall (4-6 grids of 34ha in size). Closed population methods (Otis et al. 1978) were used for calculating numbers (see Krebs et al. 1992 and Hik 1994 for details). Great homed owls were censused when hooting in late winter and their territories were mapped on a lOOkm^ plot within the study area. Nests were located either by monitoring females with radio-transmitters or by triangulating hooting pairs (Rohner and Doyle 1992a). The proportion of breeding pairs was identified in a sample of owls monitored by telemetry. To avoid possible effects of disturbance we started checking nests only when chicks were estimated to be at least than one week old. We measured a total of 116 nestlings during 1989-91, determined their age from feather measurements (Chapter 3), and back-calculated clutch initiation assuming an incubation period of 33 days (Johnsgard 1988). Post-fledging survival was determined according to Chapter 3. Survival estimates are based on individual great homed owls monitored by radio-telemetry. Twenty-one adult territorial owls were captured with mistnets and cage-traps, and 55 owlets were equipped with radio-transmitters before fledging. The radios weighed 50g (<5% of body weight, Kenward 1985), including a shoulder harness of teflon ribbon for attachment. Battery life was 2-2.5 years. The radios were equipped with a position-specific mortality switch, and we monitored the birds at least once per week. Most checks were conducted with hand-held equipment from the Alaska Highway, which follows the valley bottom for the whole length of the study area. In addition, we searched the entire area and its surroundings for radio signals from helicopter or fixed-wing aircraft at least twice per year (in fall after dispersal, and in spring after the onset of breeding). 8 Most telemetry locations for weekly movements were obtained by triangulating roosting owls during daytime with hand-held equipment. Topographical maps were used in the field to estimate the locations and assess the optimal number of bearings needed for reliable fixes. The triangulations were analyzed later with the program "Locate 11" (Nams 1990) for calculating exact locations and distances. Median 95%-error ellipses (Lenth estimator, Saltz and White 1990) were 0.06km2 (quartiles 0.02-0.19km2 n=847) for territory holders, and O.SSkm^ (quartiles 0.14-1.56km2, n=418) for floaters. For survival analysis, the staggered entry design was used (Lee 1980, Pollock et al. 1989). Individuals were identified as "added" (beginning of monitoring), "censored" (end of monitoring) or "dead" for each sampling interval. Therefore, the number of animals at risk and the survival rate are estimated for each time interval. Survival is multiplicative between time intervals and is calculated according to the Kaplan-Meier procedure (Lee 1980, Pollock etal. 1989). To calculate yearly survival, we defined a year from early October, when all juveniles had dispersed from their natal territories, to early October of the following calendar year. Three juveniles monitored in 1988-89 were pooled with the juveniles in 1989-90, assuming that both years were representing similar conditions during the phase of increasing hare densities. Survival estimates of young owls are based on birds that did not disperse beyond the area monitored from the air, which was about five times larger than the intensive study area used for locating nests. Population model The abundance of floaters was estimated with a simple population balance model with discrete time intervals (Walters 1986), in which all of the measured parameters were combined as explained below. The calculations were simplified into five steps according to the seasonal phenology of great homed owls. The exact values of parameters used are given in Tables 2.1-2.2. There was insufficient information for 1988, but because the parameters measured from 6 nests and 4 radio-marked owls were not different from these in 1989, we assumed the more precise values from 1989 as representative for both of these years of early population increase. As described above, owl years begin in early October, when juveniles reach independence and are dispersing. In step (1), the number of floaters F in year / is updated. A cohort of juvenile recruits R bom in the previous summer (year i-1) enter the 'floater pool': F/ = F/-/+R/-7 (1) In the subsequent fall and winter months, a portion of these floaters die (survival rate s^j in year /) or emigrate (emigration rate e^ ,-, note that e^ is called emigration rate for simplicity, but is rather a 'residency rate', or represents 1-emigration): F, = F,.*S;.,*e^,. (2) A number of these birds establish territories in late winter before the beginning of the next breeding season. The number of new territory holders NT,- is estimated from: (a) the actual difference in the number of observed territiories from year i-1 to year i (i.e. the change in numbers of hooting males that were censused), and (b) the number of vacancies that arise from territory holders TH,- that have died or emigrated (survival rate Sji and emigration rate tji): NT,. = TH,- - (TH,..; * sji * ^ Ti) (3) The floater pool is then reduced by the number of birds that have acquired territories: F, = F, - NT, (4) This leaves us with the observed number of territorial adult owls T, and the estimated number of floaters F, in spring of year /, and this is the output that is referred to as 10 numerical response in the results. During the subsequent breeding season, the number of recruits of year / is determined: '^i = ^i*^i*^FLi (5) whereas b^ - is the breeding rate or proportion of resident and territorial pairs that are breeding (hatch eggs), FL^ - is the number of fledged young per reproductive pair (i.e. with a nest in which eggs hatched), and Spu is the survival rate of fledglings before dispersal (post-fledging survival). The loop is then closed by returning to step (1). RESULTS Snowshoe hare densities Snowshoe hare densities recovered from a low in the mid 1980s (Krebs et al. 1992) and increased more than 50-fold in abundance until 1990, when they reached peak values (Fig. 2.1e). Hare density also varied between seasons, increasing from spring to fall because of high reproduction, and dropping over winter because of mortalities. From 1990/91 onwards, reproduction remained below mortality and the hare population declined rapidly according to the long-term pattern of a 10-year cycle (details in Hik 1994). Reproduction The reproductive response by great homed owls was pronounced, and can be described as three phases (Fig. 2.1). During peak densities of hares in 1989 and 1990, a high proportion (86%) of territorial pairs hatched eggs. Clutch sizes and mortality among nestlings in the first week of age are unknown, because we refrained from disturbing at this early stage. Brood sizes were high, however, and nestling mortality was 11 Figure 2.1 Breeding performance of great horned owls and the snowshoe hare cycle at Kluane Lake, southwestern Yukon (sample sizes are given in Table 2.1). A: Laying day in March; B: Breeding rate, or proportion of resident territorial pairs breeding (hatching eggs); C: Number of fledglings per reproductive pair (with hatched eggs); D: Number of juveniles reaching independendence and dispersing in fall (per resident pair); E: Snowshoe hare densities in spring and fall of each year. 12 ^ o \— 2 c CO • D O) c T3 0) 0 JQ « L_ CC Q. 1 10 20 30 /b 50 25 cfl (/} o CO 3 r CC Q. 1— Q. 0} *^ ^ c ^ CO Q. • # - ' C 0) • o CO 0) 1» 2 1 0 1.5 1.0 0.5 3 2 1 clutch initiation proportion breeding fledglings recruits in fall hare density 13 very low (Chapter 3). An average of 2.53 fledglings were produced per reproductive pair (±0.68 S.D., n=30). Post-fledging mortalities were low (Chapter 3), and a high number of 1.7 recruits per resident pair reached independence and dispersed in fall (Fig. 2. Id, see methods for calculation). In the first year of hare decline, 1991, productivity of owls dropped sharply. A higher proportion of pairs did not attempt to breed or failed before hatching (Fig. 2.1b, this trend, however, was not statistically significant for 1989-91). There was a trend towards later clutch initiation from 1989-91, but this result too was not significant statistically. Brood size was lower, and nestling mortalities increased (Chapter 3). This resulted in fewer fledglings produced per reproductive pair compared to 1989-90 (1.75+0.79 S.D., n=27, p<0.005, DF=2, Kruskal-Wallis test. Fig. 2.1c). The most dramatic change in 1991 occurred during post-fledging, when survival fell to 29% of previous levels (Chapter 3). Combining these parameters reveals that only 0.3 offspring per resident pair reached independence in the fall of 1991 compared to 1.7 in 1989-90. Productivity thus fell over 80% from the peak of the snowshoe hare cycle in 1989-90 (Fig. 2. Id). In 1992, no owls produced any young in the study area (Fig. 2.1). There were no signs of nesting attempts, and mates of three monitored pairs did not even roost together, as typically found in reproductive years (Petersen 1979, Rohner and Doyle 1992a). This situation remained unchanged in 1993 (F. Doyle, pers. comm.). To test whether this third phase of almost complete reproductive failure was typical for the cyclic low in snowshoe hares, we used an external set of data. By 1978, David Mossop (Renewable Resources, Yukon Territorial Government) had established routine inspections of raptors and owls that were found injured or dead and reported to conservation officers. This information is based on the area of the entire Yukon and covers two snowshoe hare peaks, thus representing at least some replication over space 14 and time. The same pattern as in our study area was apparent (Fig. 2.2). Although injured or dead adult owls were reported throughout the entire length of the cycle, there were no juveniles during the years of lowest hare densities during 1984-86. Survival Survival of both adult and young owls was high during the peak phase of the cycle (Fig. 2.3). Territorial adults survived at a yearly rate of 95.1% (±6.6% confidence interval) in 1989-90 and 1990-91. Young birds in their first or second year of life had an equally high survival rate during peak hare densities. The values in Fig. 2.3 even exceed those for adult owls, but we assume a shght rounding error because the sample size for young birds was too small to detect survival differences of less than five percent accurately. Hare densities did not recover during summer 1991, and continued to decline in the following years. The survival of adult owls decreased by 13.2% in 1991-92 compared to previous levels (Fig. 2.3). During the same time, the survival of young birds dropped by more than 60%. The sample size of young owls was low in this period, and the difference in survival was not significantly different from adult owls (two-sample tests, Lee 1980). Age at breeding and occurrence of non-territorial 'floaters' The long life spans of radio-transmitters allowed us to examine the integration of fledglings into the breeding population. A minimum of 20 owls were monitored to the end of the first year of their hfe, and a minimum of 9 owls to the end of the second year of their life. Of the cohort hatched in 1988, one of three yearlings, a female, settled in late spring 1989, was actively territorial in fall 1989, and bred successfully in 1990 and 1991. 15 Figure 2.2 Occurrence of juvenile great homed owls during two peaks of the snowshoe hare cycle in the Yukon (56 juveniles, 70 adults, arrows indicate years of highest hare density in Kluane). Based on numbers of injured and dead great homed owls that were reported to Renewable Resources, Yukon Territorial Government (data from D. Mossop, pers. comm.). 16 no. juvenile owls reported - J CD 0) ~NI CD 00 O 00 l\3 00 00 00 00 CD o CO ^ --• --o 1 9 / ^ «) 1 en r ^ • " - o \ 1 y\ 1 1 O CJ1 1 1 • o (0 ' ' * * *«* * * 1 1 1 ' 1 o b - i N) CJI b en b o (a|Bos 6o|) s|Mo Jinpe / anuaAnf OJIBJ Figure 2.3 Survivorship of adult owls (territory holders) and young owls (first and second year, floaters) based on radio-telemetry. A: Probability of survival; B: number of owls monitored. Years begin and end in early October. 18 1.00 0.75 CO > M *^  O § 0.50 !o ca e 0.25 K territory holders floaters ' • • • • 1989 ' 1990 1991 1992 C o E i_ o E C 20 15 10 5 r^ • / r.., V . \ \ B territory holders floaters 1989 1990 1991 1992 19 In 1990, two female siblings settled immediately in the same fall without any of the extended dispersal movements that are typical for juveniles. Both of these owls fledged young in the following spring. Most owls that were monitored after dispersal did not settle within a territory, and did not show any sign of hooting or other territorial defence. In order to test whether these non-territorial 'floaters' would normally be included in a census, a number of owls were monitored within hearing range to record their hooting activity from 3 March to 26 April 1990. Hooting activity was measured as the duration of bouts, each of them considered to be finished when more than 5 minutes elapsed between hoots. Almost all territorial males, and often also females, gave territorial challenges at least for a short time, particularly at dusk and dawn (see also Rohner and Doyle 1992a). In 11 territories specifically monitored for a total of 32.0 hours, all males were recorded giving territorial challenges. Hooting bouts by territory owners lasted 26.7% of the total time (measured mostly at dusk). Of six individual floaters that were monitored for a total of 16.8 hours, none of them was ever heard giving a territorial challenge or any other call. During the same time period, known territorial and non-territorial owls were tested for their responsiveness to playback. Territorial challenges were broadcasted in irregular intervals for a total duration of 20 minutes from a tape-recorder. Twenty-four individually known males were tested in 37 trials, and 28 (75.7%) of the trials resulted in vocal responses. Six individual floaters were each tested in one such trial. Two owls approached the speaker as concluded from telemetry readings, but none of them responded with a vocal signal that would have allowed their detection during a standard census (Fisher's Exact Test, p<0.01, DF=1, n=30). 20 Movements and emigration Juvenile owls started to disperse in September. Median dispersal dates were 20 September 1989,20 September 1990, and 27 September 1991. By the end of the first week in October 1989 and 1990, fewer than 5% (one of 27) remained in their natal territories. In 1991, three of seven juveniles had not dispersed by early October, but these birds never dispersed and in the subsequent winter months they died near where they fledged. Most dispersal movements occurred within the first weeks of leaving natal territories. Some juveniles left the study area within a few days, others stayed or returned after temporary absences. In mid September, median distances from nest sites were 0.6km, 0.7km, and 0.8km for the three years 1989-91. Movements then increased rapidly, and median dispersal distances rose to 24.6km, 18.5km, and 16.0km by mid October. Surveys in the following spring revealed a median natal dispersal of 35km, which is equivalent to about 15 territories in diameter. The movement patterns of territory holders and floaters were substantially different (Fig. 2.4). Territorial owls were very restricted in their movements, and distances between weekly locations very rarely exceeded 3km. Non-territorial owls showed a variety of movement patterns, involving larger dislocations. They did not move randomly, however, and only about 20% of the recorded distances were greater than 10km from one week to another. Typically, a floater would move within an area of about 5-6 times the size of a territory, and then shift to another area over time, sometimes switching between several known areas. Territorial owls almost never overlapped in their space use, but floaters overlapped broadly with defended territories and with each other. Territory holders showed extreme site fidelity, and extra-territorial movements were never observed during the highest hare densities in 1989 and 1990. In September 21 Figure 2.4 Weekly movements of territory holders (filled bars) and non-territorial 'floaters' (empty bars) during 1989-1992. Only movements of resident owls, but not emigrations, were included. The results are based on 17 territory holders (n=637 distances) and 6 floaters (n=170 distances) for 1989-90, 19 territory holders (n=894) and 8 floaters (n=271) for 1990-91, 19 territory holders (n=742) and 3 floaters (n=44) for 1991-92. Years are defined from September to September. 22 03 o O c g O 100 0-3 3-5 5-10 >10 weekly movements (km) 0-3 3-5 5-10 >10 weekly movements (km) 1991-92 UL' 1 0-3 3-5 5-10 >10 weekly movements (km) 1991, two females trespassed into adjacent territories on several nights. Three females left their territories and moved to sites 15km, 28km, and more than 30km away. These birds returned within 2-14 days to their territories. Subsequently, extra-territorial movements occurred at higher frequency. By October 1992, at least 30% (7 of 19) of the monitored adult owls were known to have temporarily left their territories. All these movements concerned residents that returned to their territories, not emigrating birds. Emigration rates showed a similar temporal pattern (Fig. 2.5). None of the monitored owls emigrated during peak densities of snowshoe hares before fall 1990. In winter 1990/91, floaters started to leave the study area. Territory holders did not emigrate before fall 1991. Estimates of emigration rates are conservative and may underestimate real values. Because of possible transmitter failures, we counted lost signals as emigrations only when the movement pattern or locations outside the study area gave further evidence. Thus birds not relocated were not counted as emigrations. Estimating numerical responses The density of non-territorial floaters was estimated based on productivity, survival, and emigration (Tables 2.1-2.2, see methods for details). The results are shown in Fig. 2.6a. Assuming that no floaters were present in spring 1988, the numbers rose quickly from zero to densities similar to territorial owls. The beginning of the hare decline in the winter of 1990/91 resulted in an immediate reduction in population growth due to emigration and lowered production of recruits by territorial pairs. Floater densities reached a peak with a time lag of one year relative to the hare cycle, and then dropped sharply from 1991 onwards, because of increased emigration and mortality, and because no additional juveniles were produced locally that could have compensated for losses in the non-territorial segment of the population. 24 Figure 2.5 Emigration of adult owls (territorial) and young owls (first and second year, floaters) based on radio-telemetry. Presented is the 'residency rate' (1 meaning all owls remain resident, 0 meaning all owls emigrate). Sample sizes as in Fig. 2.3b. 25 c 0 •g w 0 c ' c 'cO E (U c o ••c o Q. O territory holders floaters 1989 1990 1991 1992 year 26 Table 2.1 Breeding performance of great homed owls at Kluane Lake, 1989-92. Parameters used for estimation of floater abundance (see methods for details). Reproductive parameter 1989 1990 1991 1992 Number of owl t^ territories monitored 14 21 27 25 Proportion of pairs producing nestlings b, .86 .86 .78 0 Fledglings per repro-ductive pair, ±S.D.) FL, 2.42±.79 2.61±.61 1.75±.79 Post-fledging survival s^ ,^ (20 weeks, ±S.D.) .800±.211 .795±.177 .232±.151 Fall recruitment rate (off-spring per resident pair) R, 1.66 1.78 .32 0 Table 2.2 Survival and emigration of great homed owls at Kluane Lake, Yukon, as determined by radio-telemetry from fall 1989 to fall 1992. Given are yearly survival rates (Sj-, and S/r,), and yearly 'residency rates' (Cj-, and Cp), for adult owls (territorial) and young owls (first and second year, floaters). Survival rates are (1-mortality), residency rates are (1-emigration). M 00 time period 1989-1990 1990-1991 1991-1992 1989-1992 hare densities peak 1st yr decline 2nd yr decline overall age class adult 1+2 yr adult 1+2 yr adult l+2yr adult 1+2 yr survival ±95% CI .947±.101 1.000 .955±.092 1.000 .819±.258 .353±581 .905±.117 .701±.342 residency ±95% CI 1.000 1.000 .950±.096 .701+.501 .665±.286 .574±.911 .858±.134 .733±.342 n monitored (weekly avg.) 14 11 19 12 13 4 15 9 Figure 2.6 Numerical response of great horned owls (spring densities) to the snowshoe hare cycle. A: Estimated density of non-territorial owls ('floaters'). B: Census of the territorial population (with minimum and maximum estimates), and total population (sum of territorial and non-territorial owls). Arrows indicate the peak year of the snowshoe hare cycle. 29 CM E o o CO c: (U •D o 30 20 10 0 A / / floaters • y f 1 1 1 1 1 1 1 1 1 1 si • / t 1 1 ^ \ \ 1 \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ b-—o 1 1 1 r 88 89 90 91 92 93 CM E o o c 0) •D "i o 80 60 40 20 B total population territory holders 88 89 90 91 92 93 year 30 The number of territorial owls in the study area increased almost linearly from 1988-1992 (Fig. 2.6b). Even when the hare population started to decline in 1990/91, the number of owl territories kept rising until spring 1992. Then, with a time lag of two years relative to the hare cycle, the number of territories dropped in 1993. The numerical response of the total population of great homed owls is given in Fig. 2.6b. Since the territorial segment represented a nearly linear component, the sum of densities or overall pattern closely resembled the floater response with (a) an immediate reduction in population growth as hare densities declined, and (b) with a decline that was delayed by one year relative to the hare cycle. DISCUSSION Age of floaters and population dynamics The age of floaters and their breeding potential are relevant to how natural populations respond to environmental change (Caughley 1977, Lande 1988, Sinclair 1989, Perrins 1991, Newton 1991b and 1991c, Newton 1992). If floaters can breed, but are prevented from doing so by territory holders, they add flexibility to the dynamics of a territorial population. For example, the fast increase of a population of sparrowhawks (Accipiter nisus) recovering from high pesticide levels was possible because of high recruitment of young birds into the breeding segment of the population (Wyllie and Newton 1991). Large owls are known to breed at the end of their first year of life in captivity (Flieg and Meppiel 1972, K. McKeever pers. comm.), but the age at first breeding in natural populations has only been speculated on (Weller 1965, Henny 1972, Adamcik et al. 1978). To our knowledge, our observations of breeding at one year of age are the earliest records of wild great homed owls. All of these birds were females. Earlier onset 31 of breeding in females than males may represent a typical pattern, because both in owls and raptors males are the sole providers of food for the female and the young throughout most of the breeding period, which may be more difficult than laying and incubating eggs (Newton 1979). Our results suggest that only few young birds were recruited into the territorial population at the peak of the snowshoe hare cycle. It is possible that long-distance dispersers (which we were not able to monitor) were more successful than floaters remaining in the study area. It is also possible that transmitter loads had adverse affects on young owls (Gessaman and Nagy 1988, Paton et al. 1991, Vekasy et al. 1994), prevented them from establishing territories, and therefore lead to inflated estimates of the actual floater population. Little is known about the age composition of floater populations, but floaters probably consist primarily of young pre-breeders (Newton 1991c and 1992). For estimating the abundance of non-territorial birds in this study, only owls bom after 1987 were included. This is conservative, because we assumed that no floaters were present early in the increase phase in 1988. It is possible, however, that older non-territorial owls survive to the end of the low phase of the cycle and add to the floater population. We had no possibility of detecting or trapping such birds in our study. Where immigration is used to explain population increases that exceed local production (e.g. Adamcik et al. 1978), the alternative explanation of recruitment from a local floater pool should be considered. Movements and the threshold model of territoriality Great homed owls are nocturnal and difficult to observe, but reports of irmptions into the temperate zone south of the boreal forest occur regularly when snowshoe hares become scarce (Keith and Rusch 1989). Many recoveries of great horned owls banded in 32 Saskatchewan confirm this pattern. The proportion of long-distance recoveries increased during hare declines, and it seems that a cohort of particularly young birds makes directed movements leaving the area affected by the snowshoe hare cycle (Houston and Francis 1995). Territory economics and the threshold model of territoriality predict that birds should defend territories most consistently at intermediate food densities (Davies 1978, Carpenter 1987). Territories should be abandoned when food and intruder pressure increase beyond an upper threshold where the cost of defence exceeds the benefits of exclusive access. A lower threshold is postulated for the case of decreasing food density, where the declining benefit sinks below the cost of territorial defence. There was no indication that great homed owls abandoned territorial behaviour during the peak of the hare cycle, and therefore no evidence for the existence of an upper threshold in the Kluane system. Hooting activity was high, territorial conflicts were frequently observed, and adult owls were never observed to leave their territories at any time of the year. Whether subtle changes in territorial behaviour occurred, such as increased tolerance towards trespassing floaters, is unknown. When hare densities declined, there was evidence for a lower threshold for territorial defence as predicted by theory. The extra-territorial movements first observed in fall 1991 indicate a reduced tendency to defend territories, and may have been exploratory movements. Territorial owls then started emigrating before mortality rates affected them seriously. Floaters also left the area before mortalities occurred. Although this difference is not significant statistically, it is interesting that floaters started to emigrate from the study area before territorial birds. All of the owls that left their territories disappeared from the area and its surroundings, and none remained as floaters. We assumed in our calculations that 33 emigrating owls during the decline were not replaced by immigrants from other areas. We had no means of detecting non-territorial immigrants, which may have arrived or moved through the study area particularly during the decline phase of the snowshoe hare cycle. Spatial heterogeneity, source and sink populations Spatial heterogeneity can affect population parameters, and results from one study site may not necessarily apply to other situations (PuUiam 1988). For the estimation of floater densities, we assumed that natal dispersal would even out. Or in other words, that our study area was neither a source population (more recruits produced than elsewhere) nor a sink population (less recruits produced than elsewhere). It is unclear, whether dispersal distances or the proportion of long-distance dispersers changed over time. There is some evidence that juveniles disperse further at high population densities, when prime habitat may be saturated with territories (Greenwood 1980, Wyllie and Newton 1991). From 1988 to 1991, the proportion of dispersing juveniles that left the study area was 33% (n=3), 55% (n=ll), 63% (n=16), and 75% (n=4). Because of the small sample size any conclusions are premature. Spatial heterogeneity could affect the density estimation of floaters. If survival rates of far dispersing juveniles are lower, the real number of floaters could fall short of the calculated estimate. Variation of these parameters in our population model, however, did not considerably change the pattern of numerical responses shown in Fig. 2.6b. Relevance to survival studies based on band recoveries The analysis of recoveries of banded birds is a useful source of information about mortality in natural populations. Like any estimate of mortality, it is has certain biases (Clobert and Lebreton 1991). One of the most important problems for age-specific 34 survival estimates is the fact that both survival and recovery rates are unknown. Thus, many possible combinations of probabilities of survival and of recovery will fit a given data set. The only solution to this problem is the use of additional, independent field data (Lakhani and Newton 1983). Based on banding recoveries, Adamcik et al. (1978) developed the first life table for boreal populations of great horned owls that took the strongly fluctuating reproductive success into account. They concluded that given almost any assumption about the age at breeding, immigration was necessary to explain the observed increase in the numerical response to the snowshoe hare cycle. At the time of their study, however, only a limited sample size of band recoveries was available, and no statistically significant differences in juvenile survival rates were apparent between hare increases and declines. Compared to our direct survival estimates, their averaged yearly juvenile survival of only 0.45 throughout the cycle is low, and results in a serious underestimation of the potential for increase in great horned owl populations. Houston and Francis (1995) re-analysed the updated recoveries from Saskatchewan great horned owls. Their survival modelling allowed reporting rates to vary over time, and the much improved survival estimates revealed significant differences in juvenile survival during increasing and declining hare densities. The direct measurements of population parameters at our field site do not cover the whole duration of a hare cycle, and further estimates are needed because of small sample sizes, but they provide some independent information that can be used to evaluate unknown parameters. For example, the annual survival values established by Houston and Francis (1995) were 0.56 for the first year, 0.74 for second year, and 0.88 for adults during the high phase. Using a balance equation (Kenny et al. 1970) reveals that for this scenario, a yearly production of 1.80 recruits per female is necessary to prevent the 35 population from declining in the long-term, even if 75% of the yearlings are breeding. Considering brood sizes in Saskatchewan (Table 1 in Houston and Francis 1995), this level of productivity is not reached using any of the data available on the proportion of breeding territory holders (Petersen 1979, Adamcik et al. 1978, this study). Our telemetry data indicate higher survival of 0.91 for adults and 0.70 for young owls during high prey abundance. Further information is needed, but it is unlikely that minor adjustments in absolute levels of survival would change the patterns of differential mortality over different phases of the cycle (Houston and Francis 1995) or age classes (Chapter 6). Large floating populations: broader implications Initially, our main findings of such a high proportion of floaters were a surprise. We found considerable juvenile mortality during the post-fledging period (Chapter 3), but survival during the remaining first two years of life was unexpectedly high. This lead to a large segment of non-territorial floaters in the population. Juvenile survival in birds is generally much lower than that of adults (e.g. Lack 1954, Newton 1979, McCleery and Perrins 1985, SuUivan 1989). The resuhs presented in this study, however, show that this relationship is not constant, but varies greatly according to environmental conditions. In our case, the highest prey densities lead to juvenile survivorship almost identical to adult survival, and to early breeding at least in females. For long-lived predators, models assessing population declines or the capacity for recovery should consider that juvenile survival and age at first reproduction are not constant parameters, and that food availibility can affect these profoundly. The notion of a large fraction of secretive floaters in a predator population is also important for the study of ecological communities. In considering both academic research and applied management questions, many studies have attempted to examine the impact of 36 predators on prey populations (review in Krebs 1994). A common method is to census predator and prey populations, identify the diet of predators, and calculate the predation mortality among prey (e.g. Craighead and Craighead 1956, Keith et al. 1977, Petersen 1979, Angelstam et al. 1984, Trostel et al. 1987, Korpimaki and Norrdahl 1991, Korpimaki 1993). In our case, the territorial great homed owls censused during traditional hooting transects and territory mapping, may have represented only 50-60% of the total population at highest prey densities. Thus predation pressure on prey would have been severely underestimated. Similar problems may occur in other avian predators that are censused by counting nests or other behaviour related to breeding. Underestimation may be particularly severe when prey populations decline and lead to mass emigrations in other areas. Such effects of nomadic predators have been proposed to cause synchronization in population cycles of small mammals (Ydenberg 1987, Korpimaki and Norrdahl 1989, Ims and Steen 1990). Finally, it was unexpected that the population responses of territorial and non-territorial owls would be qualitatively different. The overall population response was only partially reflected in the census of territories. Given the unknown floater pools in most studies of natural populations, many past conclusions need to be viewed with caution. This raises a serious concern for conservation. When, as here, floaters are more affected by decreasing habitat quality than territorial birds, traditional monitoring programmes that are based on censusing territories will not reveal these declines at an early stage (Wilcove and Terborgh 1984). Franklin (1992) estimated that for slowly declining spotted owl populations, declines in territorial owls would not be detected for 15 or more years even when the buffering density of floaters was assumed to be low. Little is known about the size and structure of floater populations (Smith 1978, Smith and Arcese 1989, Matthysen 1989, Newton 1992). Further research is needed to 37 establish how floaters may influence demography of populations that differ in life span, social structure, ecological guild, and habitat. 38 CHAPTER 3 Survival of great horned owls during their first year of life: black flies, blood parasites, and the paradox of high post-fledging mortality ABSTRACT Most bird species have low survival in their first year of life, and the highest losses occur when juveniles become independent and disperse. Young great homed owls (Bubo virginianus), monitored by telemetry in the southwestern Yukon, Canada, survived well during the peak of the population cycle of snowshoe hares (Lepus americanus). Subsequently, juvenile survival collapsed parallel to the decline in hare densities. Disease and predation both increased significantly. We presume that this increased mortality was a cascading result of food shortage. Infestations with Leucocytozoon were unexpectedly high. This blood parasite is transmitted by black flies, and during summer great homed owls shifted their roosting habits to using more exposed sites on the ground where black fly activity was lowest. Mortality rates of juvenile great homed owls peaked before, not after independence. We discuss this early mortality in the context of life history evolution. Extended parental care may be adaptive by enhancing offspring survival after independence, but slow development during post-fledging could have the cost of increased susceptibility to parasites. INTRODUCTION Patterns of survivorship and the ecological dimension of mortality regimes are paramount to understanding lifetime reproductive success, the evolution of life histories, and population dynamics especially in long-lived animal species (e.g. Newton 1979, 39 Newton 1989b, Steams 1992). Such information is not easily available, because long-lived animals are often rare and difficult to study. The large Bubo owls are among the bird species with the most conservative life histories. They occur widely on several continents, but their survival is not well studied. Most of our knowledge on natural populations is based on the long-term banding effort on great homed owls (Bubo virginianus) by C.S. Houston in Saskatchewan (Stewart 1969, Henny 1972, Adamcik and Keith 1978, Houston and Francis 1995). More detailed information can be obtained from monitoring individuals intensively by radio-telemetry, but so far these studies have been limited by small sample sizes (e.g. Dunstan 1970, Petersen 1979). We studied the response of great homed owls to the 10-year population cycle of snowshoe hares in the boreal forest from 1989-92 (Rohner 1994, Krebs et al. 1992). Juvenile survival was very high during the peak of the cycle, but then suddenly dropped in the first year of decline in 1991. Many of the dead owls were highly infected with Leucocytozoon ziemanni (Hunter and Rohner 1994). This genus of blood parasite is prevalent in many bird species, but lethal effects have only been reported in domesticated waterfowl and gallifonnes (Bennett et al. 1993). Little is known about the effects of blood parasites in natural bird populations (Korpimaki et al. 1993); we therefore explored the ecological imphcations of this disease in more detail for the great homed owl. This paper has three objectives: First, we analyse pattems in survivorship from hatching to the end of the first year of life and test the commonly assumed hypothesis that juveniles suffer highest mortalities when they become independent and begin to disperse. Second, we ask whether juvenile mortalities are related to food, predation, or disease. Third, we assess the ecological conditions related to the transmission of Leucocytozoon. This protozoon is transmitted by omithophilic species of black flies (Simuliidae) (Desser and Bennett 1993). After we discovered high loads of Leucocyctozoon infections in 1991, 40 we were unable to study the detailed pathology in 1992, because none of the owls bred (Chapter 2). We conducted mensurative experiments on the abundance and distribution of omithophilic black flies in summer 1992. We hypothesized that (a) black fly numbers sufficiently high for transmission were not an isolated event in 1991, and that (b) great homed owls selected roost sites where the activity of omithophilic black flies was lower than elsewhere. Finally, we discuss how food, disease, and predation might interact to cause mortality, particularly in the context of behavioural decisions to survive when trade-offs exist. We then compare patterns of juvenile mortahty in great homed owls with age-specific survival in other species, and consider why the post-fledging phase is such a sensitive stage. METHODS This study is part of the Kluane Boreal Forest Ecosystem Project (Krebs et al. 1992), and was carried out from 1989-92. We worked at Kluane Lake (60° 57'N, 138° 12'W) in the southwestern Yukon, Canada. The study area comprised SSOkm^ of the Shakwak Trench, a broad glacial valley bounded by alpine areas to the north west and the south east. The valley bottom averages about 900m above sea level and is covered mostly with spruce forest (Picea glauca), shmb thickets (Salix spp.), some aspen forest (Populus tremuloides) grassy meadows with low shrub (Betula glandulosa), old bums, eskers, marshes, small lakes and ponds. We located nests either by monitoring females with radio-transmitters or by triangulating hooting pairs in March and April (Rohner and Doyle 1992a). Because of possible disturbance we started checking nests only when the chicks were estimated older than one week. We measured a total of 116 nestlings during 1989-91, and we calculated 41 their ages from linear growth equations on the fourth and eighth primaries. We calculated growth parameters separate for individual nestlings that were repeatedly measured and found them consistent with previously published results (Petersen and Thompson 1977, Bechard et al. 1985). The nestlings develop their own thermoregulation within 26 days (Turner and McClanahan 1981). After the youngest nestling reached 30 days and oldest members of the brood were close to fledging, we transferred 86 owlets to tethering platforms for further monitoring of diet, growth, and weight changes (Chapters 4-5). A total of 38 broods from 18 different territories were involved (11 in 1989, 14 in 1990, and 13 in 1991). Our tethering platforms followed the suggestions by Petersen and Keir (1976), but we elevated them to a height of about 3.5m above ground to reduce predation risk by mammalian predators. The owlets were checked 2-4 times per week, and they were released after 3-5 weeks at a maximum age of 80 days. We equipped a total of 55 owlets with radio-transmitters, which were attached with a shoulder harness. Fourty-two of these owlets were released from platforms, and an additional 13 fledged from natural nests (4 broods in 1990 and 6 broods in 1991). Radios on the back of owls were hidden in feathers and weighed 50g (<5% of body weight, Kenward 1985). The radios were equipped with a position-specific mortality switch, and we monitored the birds at least once per week. Most checks were conducted with hand-held equipment from the Alaska Highway, which follows the valley bottom for the whole length of the study area. In addition, we searched the area and its surroundings for radio signals from helicopter or fixed-wing aircraft at least twice per year (in fall after dispersal, and in spring after the onset of breeding). We used the staggered entry design for survival analysis (e.g. Lee 1980, Pollock et al. 1989). Individuals were identified as "added" (beginning of monitoring), "censored" (end of monitoring) or "dead" for each sampling interval. Therefore, the number of 42 animals at risk and the survival rate are estimated for each time interval. Survival is multiplicative between time intervals and is calculated according to the Kaplan-Meier or product limit procedure. Five owls injured on platforms, and six owls killed on roads were excluded from survival analysis. Roost sites of owls with radios were located with hand-held antenna, receiver, and binoculars. Perch heights were estimated in meters, and a site was scored as 'exposed' or 'in cover' depending whether dense vegetation was present or absent within a radius of Im of the roosting owl. Most roost sites were examined later in more detail. Omithophilic black flies were captured using the method of Bennett (1960). We exposed cages with live bantam chickens as bait for 20 min, and then sealed them with a collecting cage of fine mesh. Within 30 min after a blood meal, engorged flies had left the chicken and settled on the sides of the cage. They were collected with an aspirator and preserved in 70% ethanol. We sampled at two sites with an identical layout of traps (see Fig. 3.5) on 7 days from 21 July to 5 August 1992. Both sites were near previously-used great homed owl nests, about 5km apart in open spruce forest, which is the predominant habitat type of the area. For statistical analysis, we summed the trapping results across sampling sessions for each trap to avoid pseudo-replication. We define post-fledging according to most ecological studies (but not as in some papers on banding recoveries) as the juvenile period from fledging to the beginning of dispersal (Newton 1979). RESULTS Nestling survival Nestling survival was high in all three years of study (Fig. 3.1). Because of limited sample sizes at early stages, we calculated survival per three weeks from day 15 until 43 Figure 3.1 Survivorship of nestling great homed owls 1989-91. The calculations are based on the staggered entry design (Pollock et al. 1989), and the sample size for each time interval is given below the survival. A total of 116 nestlings originated from 50 broods in 25 different territories (12 broods in 1989, 18 broods in 1990, and 20 broods in 1991). 44 in Cfi +i 15 > 3 W o o 1.00 0.75 0.50 0.25 nestling survival -y"^ / n monitored 100 - 75 - 50 - 25 3 CD « 5' CO V) 3 o 3 CD Q . 5 10 15 20 25 30 35 age (days) 45 early fledging at day 35. Overall survival was 0.954 (95% confidence limits 0.913-0.994) for 3 weeks with an average sample size of 33 nestlings for each day. Figure 3.1 suggests a trend for increasing mortalities towards the end of the nestling period. Early mortalities, however, may be under-represented because of small sample size. Post-fledging survival We defined the post-fledging period as the time between the age of early fledging (35 days) and late dispersal (175 days). The results howed an obvious pattern (Fig. 3.2). During these 20 weeks, total survival was 0.800 and 0.795 for 1989 and 1990, and then fell to 0.232 in 1991 (p<0.05, log-rank test for 1990 vs 1991, approx. X2=6.33, DF=1). The average sample size of monitored owlets for each week was 14, 23, and 14 for 1989, 1990, and 1991 respectively. Patterns in juvenile survivorship throughout the first year If survival rates are equal throughout a series of successive periods, then the number of surviving individuals can be described as an exponential decline over time at a fixed rate (Lee 1980). This null-hypothesis predicts that the hazard function (mortality rates of intervals plotted against time) will be a straight and horizontal line. If the cumulative relative hazard rate is plotted over time, a line with an angle of 45° angle will then connect the origin with the sum of the relative mortalities (=1). The null-hypothesis of constant survival rates is obviously not true for the survivorship of great horned owls during their first year of life (Fig. 3.3). Clearly there are two phases of almost constant, but differing survival rates: mortalities are higher than expected before the beginning of dispersal, and distinctly lower from October throughout the following winter and spring. The difference among weekly hazard rates is significant 46 Figure 3.2 Post-fledging survival of juvenile great homed owls 1989-91. A: Survival curves with 95% confidence intervals (combined for 1989 and 1990). B: sample sizes for each time interval. 47 number monitored survival (+SE) 00 0) (Q CD 1^ CO CD ?r CO o fe; O ^ = = : CO o > -1^ o 1 CD CD CD CD CD CX> CD -»• CD O 0) CQ CD CD CD ^\ CO cn rv3 cn CJl o cn cn IV) o cn p ->. b o CD CD L -CO CO CD CX> O CD Figure 3.3 Phases of differential mortality in great horned owls during their first year of life. Plotted is the cumulative relative hazard rate (above), and sample sizes for each time interval (below). The regression lines are fitted for periods of constant mortality (fledging (F) to early dispersal (D), and early dispersal (D) to end of year). 49 cumulative mortality o 0) (Q <D Q. o o o o CO o o .IS' o o o b I I I ! ' * 1 / • / r 1 ' ' 3 "O 5" (Q 1 1 1 1 o KJ ] J o *» 1 r^ o b> 1 1 o | p —X bo o 1 1 I I 1 1 1 1 3 \ s I 1 1 1 1 1 \ \ 1 O CO C3) CD O O O pejojiuouj u with p<0.001 (Ui8 32=449.5). There is a non-significant trend towards higher survival during the nestling phase compared to post-fledging (visible in Fig. 3.1). Causes of mortality We recorded only five losses during the nestling stage. In 1990, we found a 3 week old owlet dead under the nesting tree. In 1991, a nest with 3 nestlings was preyed upon by a wolverine Gulo gulo, according to the signs present (see also Doyle 1994). In another nest, remains of a nestling banded earlier were found beside a survivor in the nest, possibly revealing a case of siblicide. The causes of mortality for owlets older than 35 days are summarized in Table 3.1. Mortalities were rare on platforms during 1989-90. In 1990, three successive mortalities occurred on the same platform. All of those had symptoms of anaemia, associated with high Leucocytozoon infestations (Hunter and Rohner 1994). In 1991, the number of similar cases tripled (occurrence on 1 of 25 platforms in 1989-90 vs. 7 of 13 platforms in 1991, p<0.001, Fisher's Exact Test). Predations on platforms (by lynx Lynx canadensis, or wolverines) occurred only in 1991. The fate of fledged owlets during later stages before dispersal was more difficult to identify, because usually we found the carcasses with a delay of at least several days, and they had rapidly decayed by that time. The bodies of six owlets were found intact in 1989 and 1990, and we can exclude only predation as a cause of mortality. None of the dead owlets in 1989 and 1990 were touched by predators or scavengers, but this proportion rose sharply in 1991 (p<0.001, Fisher's Exact Test). Mortalities related to human disturbance Summer traffic on the Alaska Highway led to six deaths (Table 3.1). Road kills thus represented 15% of all mortalities of radio-tagged owls during the post-fledging 51 Table 3.1 Causes of mortality for juvenile great homed owls older than 35 days post-hatching. Associated with high levels of parasitism Predation, or signs of scavenging Found intact (disease, starvation, unknown) Traffic mortalities (road kills) Total of mortalities Total of juveniles monitored 1989 0 0 3 1 4 24 1990 3 0 3 3 9 36 1991 10 11 3 2 26 34 52 period. Our sample of territories is strongly biased towards the vicinity of the highway, and because the overwhelming proportion of the boreal forest is without road access, we excluded these mortalities in the survival analysis. An additional source of mortality were deaths related to the tethering method. We lost 2 owlets in 1989, and 3 owlets in 1990 due to peripheral damage that birds suffered at the edge or between the slats of platforms. We lost a total of 5.8%, or on average one individual per 603 exposure days on tethering platforms due to injury. Roost site selection We were surprised to find fledged great horned owls roosting on the ground, often exposed at sites in the sun. Adult owls in winter and early spring always roosted well-hidden and on perches above 4-5m (Rohner and Doyle 1992a). In summer 1991, we quantified some roost site characteristics of juvenile owls. We carefully approached families equipped with radio-transmitters during the day and recorded details of their roost sites (Fig. 3.4a). During June-August, 73% were less than 2m above ground (mean perch height 1.63m, n=26). Fifty percent were on stumps or logs, 35% on the ground, and only 15% in the branches of live trees. We are confident that we detected the initial resting places of owlets without disturbance, because we often found whitewash, feathers, or pellets at the sites, and the perches chosen after occasional displacements tended to be higher above ground (Wilcoxon signed-rank test, z=3.47, n=17, p<0.001). We hypothesized that the roost site selection of great homed owls reflects (a) undeveloped skills of flying, landing, and moving in trees by juveniles, (b) avoidance of black flies, (c) other factors. To attempt to reject (b), we predicted that non-breeding adult owls could show a seasonal trend to a similar summer preference to that found in juveniles. 53 Figure 3.4 Seasonal trends in roost site characteristics (mean+SE) of great homed owls. A: Juvenile owls during the post-fledging stage in 1991. B: Non-breeding, solitarily roosting adult owls in 1992. 54 m CO +1 E. •Q) JZ £? (D Q. 6 k A l l . O juvenile roosting, 1991 H0.8 cover height L H-^ --0.6 8 < (D 0.4 H 0.2 0.0 15Jun 15 Jul 15 Aug 8 r LU W + 1 E • Q ) x : o a. r ^ adult cover ^ \ \ height ^^C 1 I -1 roosting, 1992 :::(7 -•4 : 1 1.0 0.8 0.6 o < 0 0.4 0.2 0.0 15Jun 15 Jul 15 Aug 55 In 1992 no owls bred, and all were roosting solitarily when approached. We located 33 roost sites (Fig. 3.4b). As expected if roosting near the ground was a result of black fly avoidance, perch heights declined significantly from spring to summer (Kruskal-Wallis rank test, approx. X2=10.75, DF=3, p<0.013). Owl also tended to choose more exposed roosts as the season progressed from spring to summer (Kruskal-Wallis, approx. X2=7.178, DF=3, one-tailed p=0.03). Black fly activity We hypothesized that black fly activity was lower at preferred roost sites of great homed owls in mid-summer. We sampled black fly activity across a two transects over a gradient of typical roost sites that were available in the boreal forest of our area (Fig. 3.5a): (1) in the upper canopy level at 9m and 9.5m above ground (canopy heights reached 12-14m), (2) mid-lower canopy level at 3m and 3.5m above ground, (3) on the ground at the base of a tree in cover, (4) on open ground between trees (>4m from willows). In all 8 traps combined, we captured a total of 60 black flies. They included 48 specimens of the Simulium (Eusimulium) aureum complex, 4 Helodon (Distosimulium) pleuralis, 1 Helodon (Parahelodon) decemarticulatus, 1 specimen of the Simulium (Eusimulium) annulum group, 1 Simulium (Eusimulium) cf. pugetense complex, and 1 mutant Simulium (Eusimulium) sp. without signs of blood feeding. The distribution of black fly activity across the sampled gradient is presented in Fig. 3.5b. We used log-transformed values because the data points were not normally distributed. Most black flies were trapped at intermediate canopy height, fewer at the ground in cover, and almost none in the upper canopy and at exposed sites on the ground. The maximum numbers of black flies captured in a single sampling session were 11 for trap (2), 6 for trap (3), 3 for trap (1), and 1 for trap (4). Only one engorged black fly was 56 Figure 3.5 Blackfly activity in different habitat positions. A: Gradient of available roost sites chosen for sampling the activity of omithophilic blackflies. B: Differences in blackfly activity (mean+SE) across a gradient of micro-habitats according to Fig. 5a (7 samples for each location at two sites). 57 00 I—*•* 0) T3 O W o' 3 X blackflies trapped (log scale) p o o r-. I ^ k ' " ^ " ' r ' M r o CO _° ^ o oi o u, b bi bi ""• ^ ^ - i 1 1 — r — \ , en o 3 sampling height (m) o I— en O l encountered on the ground in the open during 14 trapping sessions, whereas trap (2) at mid-canopy level yielded engorged black flies in 11 out of 14 sessions. The effect of trap location across the gradient in Fig. 3.5b was highly significant (ANOVA, F3 8=23.28, p<0.01). DISCUSSION Interactions between mortality factors: does food shortage lead to cascading effects involving disease and predation ? Despite a continuing interest in the evolution and ecology of host-parasite interactions, the effect of blood parasites on body condition, fitness, and population regulation in birds are poorly understood (Loye and Zuk 1991). For more than a decade, there has been an intense interest in the hypothesis (Hamilton and Zuk 1982) that females use male plumage brightness as an honest signal of parasite resistance. Evidence has been found both for and against this hypothesis (review in M0ller 1990). Peirce (1984) suggested that blood parasites reduced body mass in migratory birds, but there is little evidence to support this hypothesis (Ashford 1971, Bennett et al. 1988, Bennett and Bishop 1990). Korpimaki et al. (1993) found a negative correlation between parasite load and reproductive success in female Tengmalm's owls (Aegolius funereus). Blood parasites occur in most species across all avian orders. About 50% of the examined individuals and species of Strigidae were infected with Leucocytozoon (Greiner et al. 1975, Bennett et al. 1982, Bishop and Bennett 1992). For some species, infection rates are even higher. In California and New Mexico, 91% of the investigated spotted owls (Strix occidentalis) were infected by Leucocytozoon (Gutierrez 1989). Korpimaki et al. (1993) found an infection rate of 95% for Tengmalm's owls breeding in boreal Finland. 59 It is well known that in mammals, malarial infections impart a partial immunity to the host, resulting in long-term persistence at low and non-pathogenic levels (e.g. Gamham 1966). Impressions that blood parasites lead to measurable detrimental effects in hosts are based on severe cases reported from domesticated poultry and waterfowl (Bennet et al. 1993). Bennett and Bishop (1990) postulated the null-hypothesis that wild bird populations have co-evolved so successfully, that infections by blood parasites are entirely benign. Parasites may also benefit by keeping the effects on hosts low (e.g. Price 1980, Loye and Zuk 1991). It is remarkable how little evidence for detrimental effects has been found despite intensive effort, and no previous reports of lethal effects on sylvatic birds are known to us. The effect of disease can be masked because of large variation in the sampling technique (M0ller 1990, Bennett and Bishop 1990). Blood smears are often taken from individuals randomly trapped or with only little information on environmental conditions. Focussing on sensitive periods, however, may be the key to understanding host-parasite interactions. For example, the increased stresses of breeding activities such as hormonal change, territory defence, oviposition, and increased foraging are hypothesized to cause the spring relapse phenomenon seen in avian blood parasites (Chemin 1952, Desser et al. 1967). The reproductive output of infected female Tengmalm's owls, for example, was reduced only in years of intermediate prey density but not when food was abundant (Korpimaki et al. 1993). In our case, the ecological factor changing most between years was density of the main prey of great homed owls. Snowshoe hares were at peak levels in 1989 and 1990, and began to decline sharply in winter 1990/91 (Krebs et al. 1992, Hik 1994). At the same time, juvenile owl survival collapsed from very high values in 1989 and 1990 to low survival in 1991 (Fig. 3.2). The causes of mortality were related to parasitemia and 60 predation, and not to food shortage. We offered food to four juveniles that showed symptoms of anaemia on platforms, but the meat was either not accepted or even when force-fed it was regurgitated soon after. The circumstances of predation also provided evidence for interactions between several agents of mortality. For example, a fledgling with anaemic symptoms, was observed to beg during daylight, but was also so lethargic that we could approach it within 3m before it flew to a safer perch. The following week the same bird was found dead, consumed by a predator or scavenger, with only the wings intact. We suggest that food shortage was the root cause of the increase in mortality in 1991, but that cascading effects lead to death by disease and predation as the proximate mortality agents. Beside the physiological interactions described above, there were also cascading effects in the ecological system: food shortage during the decline phase of the snowshoe hare cycle was not limited to great homed owls, but common to most predators in the boreal forest. Therefore the intensity of predation by predators on each other (intraguild predation) increased because of a lack of food. We had no evidence of mammalian predators climbing nest trees or tethering platforms before 1991, the first summer of hare decline. In addition to our evidence of increased predation on great homed owls, we also observed increased predation rates on northern goshawks (Accipiter gentilis), red-tailed hawks (Buteo jamaicensis), and mammalian carnivores (Rohner and Doyle 1992b, O'Donoghue et al. 1994). Finally, our tethering method may have caused stress that contributed to the increased prevalence of blood parasites among juvenile owls. Although the tethering method is widely used in raptor studies, we were interested in assessing detrimental effects and included a sample of 13 owlets that fledged naturally for comparison. The survival of these controls was higher than for tethered owls, but the difference was small 61 (0.638 vs. 0.561 for 20 weeks) and not significant (Log-rank Test, Pollock et al. 1989). The power of this test is low because of the small sample size. Black fly activity and roost site selection It is intriguing to link the seasonal shift to lower and more exposed roost sites with the distribution of black flies. Black flies were most abundant at mid-canopy level, and petered off towards the upper canopy and towards the ground, particularly in the open. Our bird bait was very successful in attracting black flies (11 out 14 trials), whereas few black flies were collected at the exposed sites on the ground (one of 14 trials). The seasonal shift in the roost site selection by great homed owls corresponded with the periods of highest black fly activity during summer (e.g. Bennett 1960, Hunter 1990). Forest owls are known to roost concealed in trees, and aberrant roost sites are highly unusual (Hayward and Garton 1984, Forsman et al. 1984, Belthoff and Ritchison 1990). Ground-roosting has also been reported for great homed owl fledglings in South Dakota by Dunstan (1970). He proposed that great homed owls may prefer the cooler ground level to avoid thermal stress because of high temperatures in summer. In our situation, this explanation is unlikely, because we repeatedly observed owls exposed in sunshine even on hot days. Because black fly activity is influenced by temperature and humidity (e.g. Hunter 1990), great homed owls may respond indirectly to these factors by avoiding black flies. Other alternative hypotheses for ground-roosting may involve the limited locomotory skills of juveniles in branches, or hunting of ground-dwelling insects by juveniles. We never observed any signs of walking or hunting at these sites, and often the presence of feathers, faeces, or pellets indicated that the owls had spent much time at the same spot. Because adult owls without families showed similar shifts in roost sites in 62 1992, we conclude that the roost shifts are not caused by developmental constraints of fledglings. All species of the avian blood parasite Leucocytozoon parasites are transmitted by an omithophilic group of black fly species (Simuliidae), which have modified tarsal claws that enable them to crawl into the plumage of birds (Bennett and Fallis 1960, Sutcliffe 1986, Desser and Bennett 1993). To our knowledge, our sample is the first identification of blood-fed omithophilic black flies for northwestem Canada. The species composition is similar to that found in other regions in northwestem America (Corkum and Currie 1987). The sylvatic species of Simuliidae are known to have a vertically stratified distribution, and many are most abundant at mid-canopy level where the abundance of forest birds may be highest (Bennett 1960, Bennett and Coombs 1975, Greiner et al. 1975). Greiner et al. (1975) compared this vertical distribution with the distribution of infection rates by blood parasites across bird species nesting in different strata. In a continent-wise comparison, there was too much variation for a clear pattern, but at three specific sites in eastern Canada the infection rates were highest in birds nesting at mid-canopy level. In addition to the transmission of blood parasites, black flies may also have direct detrimental effects (see also Fitch et al. 1946). Exposure to a high number of sucking black flies leads to extemal lacerations, and the blood loss may contribute to reduced hematocrit levels and other symptoms of anaemia. This aspect is discussed in more detail in Hunter and Rohner (1994). 63 Roost site selection: trade-off between predation risk and parasite exposure? In a complex environment, animals have to balance conflicting demands in order to enhance body condition, reproduction, and survival. There has been intense interest in trade-offs involving foraging and exposure to predation risk (reviews in Lima and Dill 1990, McNamara and Houston 1990). Other trade-offs may involve exposure to parasitism. Roost site selection by birds is usually seen in the context of finding shelter from thermal stress or from predation risk (e.g. Janes 1985). Little attention has been paid to roost site selection as a behaviour to avoid parasites, or on constraints that apply in the selection of specific roost sites because of their exposure to parasite activity. Exposed sites on the ground may reduce the exposure to blood-sucking flies and the transmission of blood parasites, but they increase predation risk. Juvenile great horned owls suffered high mortalities by mammalian predators, which was an unexpected result that is more easily understood considering high levels of parasite pressure. We suggest that in our case, the trade-off between parasite exposure and predation risk is dependent on prey density as an external variable. It is unknown how roosting birds optimize their behavioural decisions under conflicting demands in changing environmental conditions. Similar trade-offs involving parasites may also occur in other systems. For example, caribou (Rangifer tarandus) in the arctic summer may be restricted to foraging sites where parasitism by biting flies is reduced because of favourable wind conditions (Nixon 1993). Age-specific mortality rates In most bird species, juveniles suffer higher mortality rates in their first year of life compared to later stages (Lack 1954, Newton 1979). This high mortality, especially for 64 passerines, is usually seen as a consequence of an increased risk when young become independent and disperse (Southern 1970, Hirons et al. 1979, McCleery and Perrins 1985, Nilsson and Smith 1985, Sullivan 1989). Great homed owls in our study had the highest mortality during the post-fledging stage. High mortality was not the consequence of independence, since mortality rates decreased before dispersal (Fig. 3.3). Rather, it seems that juveniles have to successfully pass a critical stage in development before they begin to disperse. We support this hypothesis by our observations on juveniles that never dispersed. This occurred only in 1991, when hare densities were declining and post-fledging mortality was very high (Fig. 3.2). Three of seven juveniles stayed in or at the edge of their parental territory, and all of them died in the subsequent fall and winter. A similar case was reported by Miller (1989): Only one of 48 spotted owls did not disperse, and this bird also died on its natal area in the following winter. In young Imperial Eagles (Aquila adalberti), the pre-dispersal period was longer for birds in low nutritional condition as determined by urea levels in the blood (Ferrer 1992). Other studies add to the evidence of high post-fledging mortality in great homed owls. D. Andersen (pers. comm.) monitored 21 great horned owl fledghngs from 11 broods in two years in southeastern Colorado. He found that 60-90% died within the first 6 weeks after fledging. Of 9 fledglings equipped with radio-transmitters and monitored until midwinter in Wisconsin, 2 died in late summer (Petersen 1979). The only mortality among 8 radio-tagged great homed owl fledglings in South Dakota occurred during the dependence period (Dunstan 1970). Spotted owls in Washington, Oregon and California had low first year survival, with a high post-fledging mortality followed by a peak of mortalities after the onset of dispersal (Forsman et al. 1984, Gutierrez et al. 1985, Laymon and Barrett 1985, Miller 65 1989). Eastern screech owls (Otus asio) were vulnerable during the post-fledging period (average weekly mortality 2.27%), but not during dispersal until early winter (0%), and then showed high average weekly mortality rates of 6.53% for the second part of winter (n=22, Belthoff and Ritchison 1989, re-analysed according to Pollock 1989). The most divergent pattern was found in tawny owls (Strix aluco) in Wytham Woods, England (Southern 1970). Almost all owlets survived until dispersal, and high mortality occurred during the following winter. Much of the variation in mortality regimes for different owl species may stem from small sample sizes and ecological differences in the systems. For example, the mostly insectivorous screech owls may face food shortages when forced to capture voles in mid-winter (Belthoff and Ritchison 1989), or avian predation on fledglings may be high when territories overlap with larger owl species (e.g. Mikkola 1983, Duncan 1987, Miller 1989). Detailed information on mortality is often based on radio-telemetry, and concern has been raised that transmitter loads affect the performance and survival of birds (Gessaman and Nagy 1988, Paton et al. 1991). For our study, we have no control data to test this hypothesis directly. First year survival, however, was high during good food years in our study compared with survival information estimated from band recoveries (Houston and Francis 1995, see also Chapters 2, 6), so that potential detrimental effects are restricted to interactions of fitted radios with other factors. Li spotted owls, high post-fledging mortalities were also observed in young birds without transmitters (Forsman et al. 1984, Miller 1989). Why is the post-fledging stage a critical one in development? We focus on the evolutionary background related to life history theory, and point out some physiological characteristics that may influence disease during the post-fledging 66 stage. Almost all of the ecological interest in age-specific mortality has been related to the question of which stages are most critical for population regulation (review in Krebs 1994). From the perspective of life history theory, selection is predicted to increase the proportion of time spent in the safest developmental stages (safe harbour hypothesis, review in Clutton-Brock 1991). For example, both yellow-eyed juncos (Junco phaenotus) and screech owls showed the highest mortality rates (excluding stages in the nest) immediately after fledging (Sulhvan 1989, Belthoff and Ritchison 1989). Because this stage lasts for only a few days, the total number of mortalities is moderate. This may not just be coincidence, but rather the result of adaptation to minimize the time spent in a critical developmental stage. Great homed owls, in contrast, are among the longest-lived birds (Houston and Francis 1995), and prey capture may sometimes be difficult for them because snowshoe hares develop efficient anti-predator strategies (Hik 1994, Rohner and Krebs 1994, see also Chapter 6). Presumably because independence is disproportionately more difficult to achieve, large owls have an extended period of parental care relative to the time spent in the nest. In our study, great homed owls spent 1-2 months in the nest, and were seen begging for another 3-4 months before they dispersed quite synchronously (Fig. 3.3). We presume that strong constraints prevent juveniles from dispersing earlier. Altricial birds seem to delay the development of erythrocyte levels in the blood compared to precocial birds (Hunter et al. 1994). This does not only result in lower potential for physical performance, but also raises susceptibility to anaemia. Nestling bald eagles (Haliaeetus albicilla) had a higher prevalence for Leucocytozoon than fledglings (Hunter et al. 1994). Stage-specific parasitism was also apparent in nestling bald eagles, which were heavily infested by blow flies (Protocalliphora avium) before they were 16 days old, but not much after this age (Bortolotti 1985). 67 We hypothesize that extended parental care may not only increase the immediate success of juveniles, which are enabled to disperse at a time when they are further developed. If trade-offs exist between metabolic turnover and longevity, then slow development will prolong the lifespan, and therefore the reproductive value of offspring (review in Stearns 1992). If extended care is relatively inexpensive for parents, this life history strategy may be optimal for long-lived species, even if it is accompanied by some costs. Costs of slow development of blood chemistry and locomotory skills may consist of higher susceptibility to disease and predation (secondary mortality). Under the average or long-term ecological conditions in which the system evolved, the primary mortality at independence may be reduced. The level of secondary mortality may be increased, but not to the level of primary mortality. Under natural variation in ecological systems, we expect the observed values for mortalities to vary considerably. When we evaluate factors affecting population regulation, we usually ignore the fact that animals may be in the process of adapting to the most severe factors. If organisms are predisposed to a somewhat elevated secondary mortality, we should not be surprised to observe high post-fledging mortalities - during a phase that we consider a safer stage than dispersal. In conclusion, our results suggest that disease affects great horned owl mortality in interaction with ecological factors. Which stage of development is most sensitive, however, may not be determined by external factors but by constraints imposed by the life-history evolution of a long-lived species. 68 CHAPTER 4 Brood size manipulations in great horned owls: are predators food limited at the peak of prey cycles? ABSTRACT The reproduction of diurnal raptors and owls is strongly dependent on food resources. It is unclear whether predators experience super-abundant food during cyclic peaks of prey populations. In order to test this hypothesis, four pairs of great homed owls (Bubo virginianus) with two young were subjected to brood size manipulations at high densities of cyclic snowshoe hare (Lepus americanus) populations in the southwestern Yukon, Canada. Broods older than 35 days were temporarily enlarged by one, and then by two young. No short-term effects were observed when one owlet was added, but the addition of two young resulted in significant weight losses in manipulated broods. Simultaneously, females with enlarged broods moved farther from their nest sites at night, presumably reflecting increased hunting effort. The daytime nest attendance of females was also lower than for controls. Food additions to enlarged broods returned the changes in parental behaviour to control levels. I conclude that the concept of predators experiencing super-abundant food during peaks of prey cycles should be regarded with caution. INTRODUCTION There is much observational evidence that reproduction and abundance of diurnal raptors (Falconiformes) and owls (Strigiformes) is strongly affected by changes in food supply (reviews in Newton 1976, Newton 1979, Mikkola 1983, Voous 1988). Most of 69 these conclusions are derived from comparing successive years with varying prey densities (e.g. Southern 1970, Adamcik et al. 1978, Lundberg 1981, Houston 1987, Pietiainen 1989, Korpimaki and Norrdahl 1991), or from comparing study sites with different prey abundance (e.g. Newton 1986, Newton 1991a). Traditionally, the question whether food limits animal populations has been approached by food supplementation experiments (Boutin 1990). Perhaps because of logistical difficulties, few experiments have been conducted on raptors and owls, but in all of them food additions positively affected reproduction (Newton and Marquiss 1981, Dijkstra et al. 1982, Korpimaki 1989, WeUicome 1993, Wiebe and Bortolotti 1994). Other experimental work has involved brood size manipulations, inspired by David Lack's (1947) ideas on optimal clutch sizes. Lack hypothesized that parents laid as many eggs as they could raise successfully, and he suggested brood size manipulations for testing this hypothesis. Natural populations and their food supphes are not always at stable equilibria (Wiens 1977) as implied for the hypothesis of most productive clutch sizes (Linden and M0ller 1989, Stearns 1992). Population cycles are an interesting situation of environmental variability, because the amphtude of fluctuations is extreme (e.g. Krebs et al. 1992). If predator populations ever experience a superabundant supply of prey, this would be expected during the cyclic peak of prey populations (Lack 1946). How predator populations respond to sudden increases in prey abundance is also relevant to whether predation can cause population cycles (Caughley and Krebs 1983, Trostel et al. 1987). Great homed owls (Bubo virginianus) are large and long-lived predators. They have a small brood size of normally 1-3 young, and cyclic populations of snowshoe hares (Lepus americanus) are their main prey in the boreal nearctic (Keith 1963, Adamcik et al. 1978, Houston 1987, Chapter 5). In this paper, I examine whether breeding great homed 70 owls were food limited during high densities of snowshoe hares in the southwestern Yukon. I used short-term pulse experiments to assess effects on juvenile condition and parental behaviour during the breeding season. First, broods were successively enlarged until a negative effect on the body mass of fledglings was found. Simultaneously, parents of enlarged broods were tested for (a) increased movements at night indicating increased hunting activity, and (b) decreased attendance and defence of broods during daytime. Then, after effects had been found, a food supplementation to enlarged broods was applied to test whether the measured changes were caused by food or other factors. Because of logistical difficulties, the sample sizes were necessarily small. I tried to maintain high reliability by measuring several response variables and by applying replication, reversed treatments, and repeated measurements of individuals. METHODS Study area and techniques The experimental part of this study was carried out in 1991, and observational data were collected in 1989 and 1990. We worked at Kluane Lake (60° 57'N, 138o 12'W) in the southwestern Yukon, Canada. The study area comprised BSOkm^ of the Shakwak Trench, a broad glacial valley bounded by alpine areas to the north west and the south east. The valley bottom averages about 900m above sea level and is mostly covered with spruce forest (Picea glauca), shrub thickets (Salix spp.), some aspen forest (Populus tremuloides) grassy meadows with low shrub (Betula glandulosa), old bums, eskers, marshes, small lakes and ponds. Nests of great horned owls were located in March and April (Rohner and Doyle 1992a). In May and June, owlets were transferred shortly before fledging to elevated 71 tethering platforms (Petersen and Keir 1976), where the parents kept feeding them, and systematically monitored owl famihes for about 5 weeks (details in Chapter 3). The age of owlets during the study period ranged from 35-78 days, when most of them have reached their asymptotic weights (Turner and McClanahan 1981). The owlets were weighed every other day with Pesola spring balances (+ lOg). To prevent the accumulation of adverse effects related to specific positions on the platform (Petersen and Keir 1976), individuals were rotated among positions at every visit. Adult great homed owls were trapped with mist nests prior to the experiments, and equipped with a radio-transmitter weighing 50g (details in Chapter 2). Telemetry locations were obtained by triangulating six females with hand-held equipment every other night during the period of darkest twilight (2300-0300 hrs). The bearings were taken from surveyed stations and plotted in the field in order to immediately assess the quality of locations and adjust the minimum number of bearings needed. The triangulations were later analysed with the program "Locate 11" (Nams 1990) for calculating exact locations and distances. On average, 4.4 bearings were taken for each location, with approximate distances from the owls ranging from 0.3-2km. The median of 95%-error ellipses (Lenth estimator, Saltz and White 1990) was 0.018km2 (quartiles 0.004-0.055km2, n=98). Nest attendance of females during the day was measured by making 3-18 visits (average 9.7) per territory from 1989-91. Only visits at broods with owlets older than approximately 35 days were considered (the average age encountered in 1989-91 was 53, 60, and 55 days respectively). Experimental design Four broods (all with two owlets) were chosen for the short-term pulses of brood size manipulations. These broods are referred to as 'experimental broods' (in contrast to 72 'natural' broods of unchanged brood size). In each trial or phase, owlets from 'donor' broods were added to 'enlarged' broods. The experiments consisted of several subsequent phases of similar design, but increasing degree of manipulation and reversed treatments (see Table 4.1 for a detailed schedule). In phase I (1-10 June), two experimental broods were enlarged by one owlet from two other experimental broods. The reduced broods served as controls together with two natural broods. Among these natural broods, two owlets were exchanged in order to control for possibly negative effects from moving young owls between different territories. Phase n (11-20 June) was similar to phase I, but the experimental broods were now enlarged by two owlets. Brood addition and reduction were reversed in order to avoid cumulative effects between phases. Phase III (23-29 June) was a replication of phase n, with brood addition and reduction reversed among experimental broods. In phase rV (30 June - 6 July), enlarged broods were kept at the same size of four owlets, but food was added to them (ca. 1.5kg per day and brood, consisting of fresh snowshoe hares and ground squirrels Spermophilus parryii). Some changes in the experimental format could not be avoided. During phase n, both broods in the category 'unchanged' were lost on 15 June (predation) and 15-19 June (disease). The missing measurements were replaced by those taken from other natural broods ("K", "L", and "P" in Table 4.1). Because owlets in enlarged broods suffered weight losses during phases n and HI, food was added to owlets of all categories during periods of heaviest weight losses in order to prevent any owlets from starving (lOOg hare meat for each owlet on 18 June, 26 June, and 28 June). Between phases II and in, food was provided ad libitum for enlarged broods to allow them to regain initial conditions for the next trial. 73 Table 4.1 Experimental design for 4 phases of short-term brood size manipulations. Individual owlets are labelled with a letter identifying their brood of origin. Phase I enlarged: +1 Phase IX enlarged: +2 (reversed) Phase III enlarged: +2 (reversed) Phase IV enlarged: +2, +food enlarged J,J+C (1-9 June) G,G+S (2-10 June) C,C+J+X (12-20 June) S,S+C+X (11-19 June) J,J+C+X (24-29 June) G,G+S+X (23-28 June) C,C+J+X (29 June-6 July) J,J+S+X (30 June-6 July) -J controls (donors) C (1-9 June) S (2-10 June) (12-20 June) (11-19 June) (24-29 June) (23-28 June) (30 June-6 July) controls (unchanged) T+F (1-9 June) F+T (2-10 June) F+T (12-14 June) K,K (16-20 June) K,K T+F (11-15 June) (15-19 June) (15-19 June) (24-29 June) (24-29 June) (24-29 June) (24-29 June) (24-29 June) K,K (30 June-6 July) P (30 June-6 July) (24-29 June) In order to reduce disturbances to the owls induced by manipulations and monitoring, the length of the total experimental period was shortened as the experiments progressed. This was achieved through daily monitoring in phases lU-IV, instead of every other day in phases I-II. Statistical procedures For all calculations and statistical tests, broods or territories were considered the experimental unit. Changes in owlet weights were calculated relative to individual weights at the beginning of each phase, were averaged per brood, and then checked for normality with graphical quartile plots. Repeated measures ANOVA's were calculated on absolute brood weights (average of all owlets per brood and day). For all other statistical tests, non-parametric Mann-Whittney U-Tests were applied. One-tailed levels of significance were used for all comparisons, because the tested differences had been predicted beforehand. RESULTS Weight changes in fledglings During a first treatment, the weight development of two broods enlarged by one owlet did not differ from control broods (Fig. 4.1). There was a significant overall increase in weights during this time period (repeated measures ANOVA, F5 6=4.71, p<0.(X)5). In the next phase, broods were enlarged by two. This treatment affected owlet weights (Fig. 4.2a). The enlarged broods lost weight, and by the end of the experiment, their loss in body mass was significantly different from controls (ANOVA for day 10: Fj 6=7.75, p<0.025). An examination of the slopes in weights (average per brood) reveals 75 Figure 4.1 Weight changes of owlets (deviation from initial weights, average per brood, +SE) when enlarged by one owlet (phase 1,1-10 June 1991). Brood sizes: 3 for experimental broods (n=2), 1-2 (average 1.5) for controls (n=4). The two enlarged broods are labelled with a letter according to Table 4.1. 76 average weight change (g) -J ro Q. I " 00 ----• o § 1 • o it o Ol o o o o 1 1 1 d \ ' ' ^ 1 . . 1 . I . Figure 4.2 Weight change of owlets (deviation from initial weights, average per brood, +SE) when enlarged by two owlets. The two enlarged broods are labelled with a letter according to Table 4.1. A: Phase n, 11-20 June 1991. Brood sizes: 4 for experimental broods (n=2), 1-2 (average 1.6) for controls (n=4). B: Reversed treatment. Phase HI, 23-29 June 1991. Brood sizes: 4 for experimental broods (n=2), 1-2 (average 1.13) for controls (n=8). 78 (0 9 I I 1 \ 1 \ 1 <? , « \ / 1 / v/ "H 0) ff 1 0) o _M o c o o • 00 (0 CM O O O o O O o m S 8 S (6) 96UBL|0 }Lj6!9M Q^3^aAB 79 a similar result. The overall slope shows a significant decline (repeated measures ANOVA, F5 6=3.92, p<0.01), but the effect of the brood enlargement on the negative slope is also significant (repeated measures ANOVA, F5 6=2.17, p<0.05). In order to test this limitation in food supply more thoroughly, the same experiment of adding two young was repeated. The treatment was reversed among broods used as 'donors' and 'recipients', and the sample size of controls was increased to enhance the power of statistical tests. The results were almost identical to the previous phase (Fig. 4.2b). In contrast to controls, the enlarged broods had lost body mass by the end of the treatment (ANOVA for day 6: Fj io=4-26, p<0.05). The overall slope for all weights (averages per brood) was not significantly different from 1 (repeated measures ANOVA, F510=1.68), but there was a clear effect of treatment on the slopes (repeated measures ANOVA, F5,io=2.65, p<0.025). Nocturnal movements of females Control females with unchanged brood sizes stayed near their broods (average distances for three females 0.42km (median=0.22km, quartiles 0.13-0.67km, range 0.01-2.08km, n=36). The nocturnal movements of two experimental females during periods of three different treatments are plotted in Figs. 4.3a-b. The distances of the females from their broods are given in Figs. 4.3c-d. A first stage consisted of telemetry locations taken during 'control' conditions, when there was no indication of weight loss in their broods (during brood reductions when one young was removed for enlarging other broods, and during phase I when additions of only one young). All experimental females with radio-transmitters ("C", "J", and "S") stayed close to their broods (median 0.08km, quartiles 0.06-0.22km, range 0.01-0.98km, n=31). 80 Figure 4.3 Nocturnal position of females relative to their broods under treatments varying the level of food stress. Results are shown for control levels (•), enlarged broods (o), and enlarged broods with food addition (^). A, B: Maps with female position for two experimental broods. C, D: Distances of female from nest site (mean±SE) for two experimental broods. 81 , « « ^ E 12.0 re c •D o 8 10 o 1 >> A 1 • , , o • • ^ ^ nest "C" • 0 • A 0 0 ' • — - J 10 2.0 X - coordinates (km) "E a> a c s o 8 10 1 >> B -' ' ' ' 1 nest "J" o o o o ' ' .— , 1 10 2.0 X - coordinates (km) oO^^^°\ .^<.<^«^ E f 1.0 I-w 0) c i 0.5 I °'° ^^"^ %\* e^^^"^  nest "J" ^ i'SP .e<>«6 x^^^V a'P '^^ x 82 A second stage consisted of brood enlargements by two young (phase II and IE). The young of these females lost weight under these conditions. The nocturnal movements of females "C" and "J" increased as broods were enlarged, and the distances to their nest sites were larger than during control conditions (Mann-Whittney, Uio,6=7.0 and Uio,7=2.0 with p<0.01 and p<0.001 respectively). The same difference was apparent in a third experimental female "S" (Ujg 5=21.0, p<0.05). To test whether these increased movements were caused by food shortage or other factors, the enlarged broods "C" and "J" were kept at four young, but provided with surplus prey. Both females responded strongly to this food addition experiment (Fig. 4.3). Their movements ceased, and the nocturnal distances to the nest returned to control levels. This drop in distances from "enlarged" to "enlarged + food" was statistically significant for both female "C" and female "J" (U6,5=27.0 and U7,6=39.0 with p<0.025 and p<0.005 respectively). Diurnal nest attendance of females Females were almost always present in all territories when we visited broods during daytime in 1989 and 1990 (Fig. 4.4). There was some variation in 1991, but females of control nests still attended their broods on 89.7% of visits. In broods enlarged by two young, females were present only in 41.4% of visits. This difference is statistically significant for enlarged broods in 1991 vs. controls of all years (U4 34=129.0, p<0.001), and also for enlarged broods vs. controls within 1991 (1/49=33.5, p<0.05). 83 Figure 4.4 Daytime nest attendance of female great homed owls (average per nest, +SE, number of nests). Results from broods with natural size (controls) and experimental broods enlarged by two owlets. 84 s5 100 <l> o 801 (0 S 601 «s « 40 re i 201 n=11 • controls ° enlarged (b 4 89 90 year 91 85 DISCUSSION Magnitude ofejfects The addition of two young resulted in consistent and significant effects for all four broods chosen for the experiments. Because of the small sample sizes involved, these results should be interpreted with caution. Brood enlargements in kestrels (Falco tinnunculus) caused increased parental effort, reduced growth rate of nestlings, and increased nestling mortality (Dijkstra et al. 1990), although enlarged broods had a higher reproductive value (Daan et al. 1990). None of 10 enlarged broods in savanna hawks (Buteogallus meridionalis) fledged without adverse effects (Mader 1982). In snail kites (Rosthramus sociabilis), which normally have 1-2 young, parents had difficulties raising three young and were unable to raise four young (Beissinger 1990). The last two studies were conducted in tropical environments, but there are also results under more variable conditions. Gard and Bird (1992) manipulated broods of American kestrels (Falco sparverius) in a poor vole year, a good vole year, and in captivity with food ad libitum. They found that kestrels raised larger broods in a good year, but with reduced fledging weights. Prey availability was obviously not as high as under captive conditions, where enlarged broods had high fledging success and were also in good condition. The most comparable results to this study originate from experiments on Tengmalm's owls (Aegolius funereus), a species depending on cyclic vole populations in Fennoscandia. Korpimaki (1989) added food during the pre-laying and laying period and found a positive effect on reproduction even in a peak year of the vole cycle. The owls successfully raised broods enlarged by one young in a year of increasing vole density (although condition beyond fledging was unknown) but not when voles were decreasing (Korpimaki 1988b). Therefore, even in years of highest prey density, there seems to be a 86 bottleneck in food supply in late winter and early spring, but conditions seem more relaxed during the nestling and fledging phase (Korpimaki 1988b). A similar bottleneck during the laying phase may exist for great homed owls in late winter, when snowshoe hare densities are at their seasonal low because of winter mortality (Krebs et al. 1986, Krebs et al. 1992, Hik 1994) and few alternative prey are available (Chapters 5-6). Hare densities peaked in 1990, and started to decline in 1991 (Hik 1994). Brood sizes of great homed owls were also lower in 1991 (Chapter 2), indicating that food was limiting clutch size in late winter despite high hare densities. In contrast to Tengmalm's owls (Korpimaki 1988b, 1989), great horned owls did not sufficiently provision enlarged broods during summer conditions. Although hare densities were still high in 1991 (average 1.1/ha), it should be noted that these brood size manipulations were conducted in the first year after the the peak of the snowshoe hare cycle. Food limitation and parental effort Newton (1979) described three subsequent and overlapping phases of female behaviour typical for breeding raptors and owls: (a) almost continuous brooding or shading before the young have developed their own thermoregulation, (b) presence near the nest for feeding and defending the largely helpless young, (b) attendance with flexible periods of time spent distant from older nestlings or fledged broods. There is much variation in observed attendance patterns during the third phase, and Newton (1979) hypothesized that this variation largely depends on the hunting success of the male. Females seem to leave broods only when the male's efforts are insufficient and require additional hunting. 87 Experiments have largely confirmed Newton's hypothesis. Female flight time tended to parallel manipulated brood sizes in kestrels, and hunting by females in reduced broods ceased almost completely (Dijkstra et al. 1990). Female American kestrels spent more time hunting when broods were enlarged in a poor vole year, but there was no clear pattern in a year of higher vole abundance (Gard and Bird 1990). My results give specific support to Newton's hypothesis: females left their nests during periods when brood enlargements caused young to lose weight, and they remained with enlarged broods when food was added. Parents do not necessarily increase their effort for provisioning enlarged broods, even if sufficient food is available (Ydenberg 1994). Ospreys (Pandion haliaetus), for example, were unresponsive to brood enlargements (Green and Ydenberg 1994). One hypothesis to explain limited parental effort is based on life history-theory. If trade-offs exist between current and future reproduction, individuals may achieve higher fitness by limiting current effort, and brood sizes will fall below the most productive clutch (review in Steams 1992). In kestrels, there is strong experimental evidence that males reach maximum daily energy expenditures and suffer increased mortality when broods are enlarged, although the mechanisms for this trade-off are unknown (Masman et al. 1989, Dijkstra et al. 1990). Female great homed owls increased their effort for enlarged broods, but the response of males is unknown, and it is possible that males were limiting their effort. There may be conflicting interests in the allocation of parental effort between the sexes (see also Beissinger 1990, Ydenberg and Forbes 1991). 88 Conclusions Enlarged broods of great homed owls lose weight, despite high prey abundance and increased effort by females. This suggests that food was not readily accessible. Further information on the hunting effort of males is needed, because they usually provision broods alone. In conclusion, there is no evidence from this study or from reports in the hterature to support the idea that predators experience super-abundant food during peaks of cyclic prey, and this hypothesis requires further testing. 89 CHAPTER 5 Great horned owls and snowshoe hares: what causes the time lag in the numerical response of predators to cyclic prey? ABSTRACT Predator populations often decline with a time lag after the peak of prey cycles. Theoretical models of predator-prey interactions predict that this delay is caused by a higher rate of population growth in prey, which leaves predators with super-abundant food after the peak and buffers their decline. This situation is met when predator populations have a lower innate capacity for increase than their prey or when the increase is inhibited because of territorial behaviour. Here, I refer to this hypothesis as 'single prey hypothesis' (SPH) in contrast to the 'multiple prey hypothesis' (MPH), which predicts that the delayed decline is caused by high availability of other prey species. Results on population growth rates of great homed owls showed that the predictions of SPH were met, although the predicted difference was small when floaters were taken into account or social exclusion from breeding was removed in a population model. In their diet, great homed owls relied to a large degree on the main cyclic prey (snowshoe hares), and thus the results were not in agreement with the MPH. Inverse density-dependent growth rates in the territorial population, density-dependent accumulation of floaters, and replacements of territorial vacancies were consistent with the hypothesis that social behaviour limited the number of owl territories. Reproduction of resident owls was immediately affected by the prey decline, indicating that there was no buffering effect of super-abundant food. Therefore, neither MPH nor SPH were satisfactory explanations, and I propose a mechanism based on individual behaviour to explain delayed numerical responses: territorial predators 90 monopolize a disproportionately large amount of resources for reproduction during the increase and peak of the cycle, and are then buffered against prey declines by adjusting their breeding activities. Non-territorial floaters have lower access to resources and their numbers are affected more immediately by declining prey. INTRODUCTION Population cycles of predators and their prey occur in a variety of taxa and ecological systems (Taylor 1984, Royama 1992). They have attracted much ecological research, because their extreme and parallel population fluctuations suggest that the effect of predators may cause these oscillations. Lotka (1925) and Volterra (1926) independently derived a set of equations, which describe the dynamics between populations of predators and prey resulting in regular oscillations. A series of laboratory studies found evidence for this interaction (review in Krebs 1994), and the early models by Lotka and Volterra have been replaced by models capable of more biological realism (reviews in Taylor 1984, Akgakaya 1992). Both theoretical and field studies have investigated alternative mechanisms for population cycles, in which predation may be included partially or not at all. There is still much controversy about the role of predation, for example in causing population cycles of mammals and birds at northern latitudes (e.g. Keith 1963, Keith et al. 1977, Erlinge et al. 1983, Trostel et al. 1987, Akgakaya 1992, Krebs et al. 1992, Korpimaki 1993, Hanski et al. 1993, Hik 1994). Time lags are an essential condition for oscillations in theoretical models of population growth (Krebs 1994). Depending on the extent of the time lag, a variety of growth patterns is possible, ranging from stable equilibrium, regular oscillations, to 91 diverging or chaotic oscillations (Maynard Smith 1968, Hanski et al. 1993, Ginzburg and Taneyhill 1994). The basic mechanism causing a time lag through predation in theoretical models is the lower rate of increase of predator populations than that of their prey (general review in Krebs 1994, see Keith et al. 1977 for vertebrate predators). The prey population grows faster than the predator population and overshoots an equilibrium density. Meanwhile, predators are not immediately affected when the prey begin to decline, and as a consequence they overshoot their own equilibrium density and then decrease with a time lag. This leads to an extended decline in the prey population, and only when predator densities have dropped low enough can a new cycle begin. In reality, the complexity of predator-prey dynamics exceeds simple two-species interactions. Predator numbers often depend on prey populations, but they cannot be described by a constant conversion factor from the density of main prey, because they show a ftjnctional response to several prey species (HoUing 1959, Fujii et al. 1986). During the decline of the main prey, cyclic predators may kill a high proportion of alternative prey species and may even affect their population dynamics (Hagen 1952, Lack 1954, Keith 1963, Angelstam et al. 1984, Lindstrom et al. 1987, Marcstrom et al 1988, Korpimaki et al. 1990). This implies that the time lag in the predator response is caused by the buffering effect of alternative prey, and not directly by an excess of main prey. If this is true, the existence of a time lag does not prove that predation is causing the oscillation of the main prey species. Given the importance of time lags to explain population cycles in theoretical models, it is surprising how little is known about the mechanisms causing such lags in real predator populations. Time lags in the population decline of vertebrate predators relative to cychc prey have been described for several species (e.g. Elton and Nicholson 1942, 92 Keith et al. 1977, Korpimaki 1993), but they are not consistently found in all situations (e.g. Korpimaki and Norrdahl 1991, Korpimaki 1994). In this paper, I examine causes of the time lag in the numerical response of great homed owls (Bubo virginianus) to the 10-year cycle of snowshoe hares (Lepus americanus) in the boreal forest in the southwestern Yukon. Snowshoe hare populations are cyclic throughout their range in boreal North America, with a period of 8-11 years and an amphtude of 15-200 fold (Keith 1963, Krebs et al. 1992). Great homed owls are large and long-lived predators feeding mainly on lagomorphs, and they are widely distributed across North and South America (Voous 1988, Donazar et al. 1989). They defend long-term territories, and even at northern latitudes they are resident year-round (Chapter 2). The detailed response in reproduction, survival, and emigration to the snowshoe hare cycle is described elsewhere (Chapter 2). Hypotheses and predictions The single prey hypothesis (SPH) predicts that at the beginning of the decline phase, high densities of the main prey species are sufficient to explain the delayed decrease of the predator population. In other words, this hypothesis predicts that the predator population is not food-limited at the peak of the prey cycle. A more detailed prediction can be derived for population growth rates: the SPH can be rejected if the rates of increase between predator and prey populations are not substantially different (prediction 1). Alternatively, the multiple prey hypothesis (MPH) predicts that the declining numbers of main prey are insufficient to maintain stable or increasing predator densities, and that the availability of alternative prey is necessary to explain a time lag in the 93 numerical response of predators. The MPH can be rejected if the proportion of main prey in predator diets does not decline as numbers of main prey begin to drop (prediction 2). The SPH can be further explored. Lower growth rates of predator populations compared to those of their prey may be caused by (i) a lower potential productivity (innate capacity for increase) in the predator population, or (ii) because the maximum density of predators is limited by social behaviour (territoriality in this case). If (ii) is true, the following predictions will be fulfilled: (3a) the observed increase of established territories is lower compared to the potential population growth without social limitation (Davies 1978, Patterson 1980, Newton 1992), (3b) the rate of new territory establishments is inverse density-dependent (territory establishment becomes more difficult with increasing density of existing territories), (3c) the population size of non-territorial 'floaters' is density-dependent (Sinclair 1989), (3d) because territory holders prevent floaters from establishing territories, removed territorial birds will be replaced rapidly (Watson and Moss 1970, Newton 1992). METHODS This study is part of the Kluane Boreal Forest Ecosystem Project (Krebs et al. 1992), and was carried out during 1988-93. We worked at Kluane Lake (60° 57'N, 138° 12'W) in the southwestern Yukon, Canada. The study area comprised 350km2 of the Shakwak Trench, a broad glacial valley bounded by alpine areas to the northwest and the southeast. The valley bottom averages about 900m above sea level and is covered mostly with spruce forest (Picea glauca), shrub thickets (Salix spp.), some aspen forest (Populus tremuloides) grassy meadows with low shrub (Betula glandulosa), old bums, eskers, marshes, small lakes and ponds. 94 Territorial great homed owls were censused in late winter when they were most vocal, their territories were mapped, and most of their nests were found on lOOkm^ within the study area (Rohner and Doyle 1992). The number of non-territorial 'floaters' was estimated based on productivity, recruitment into the territorial population, survival, and emigration (for details see Chapter 2). A total of 21 adult great homed owls and 55 fledghngs were equipped with radio-transmitters and monitored on a weekly basis (see Chapter 2-3 for details). The summer diet of great homed owls was sampled during the breeding season from the beginning of May to mid July. In 1989-91, owlets shortly before fledging were transferred to elevated tethering platforms, where the parents kept feeding them, and we were able to systematically collect pellets for 3-5 weeks (Petersen and Keir 1976). This study is based on the results from 9 platforms for each year from 1989-91 (13 individual territories). Because great homed owls did not breed in 1992, fresh pellets were collected by monitoring six radio-tagged birds on five territories. Pellet analysis was based on minimum counts of prey items as identified by diagnostic bones. Details on the methods of collecting and analysing pellets for summer diets can be found in Rohner and Krebs (1994). Information on winter diets was collected during the months of snow cover, when hibernating and migratory prey were not available to the owls (usually mid October to mid April). Most pellets were collected from roost sites of radio-marked owls, or from roosts found when locating nests in March. Because the sample size was much smaller than during summer, a different method for estimating the proportion in biomass of specific prey was applied. Instead of counting diagnostic bones, the proportion of different prey was estimated for each pellet based on all bones and the volume of fur encountered. The sum of pellets for each prey species was then used to calculate a direct estimate of 95 proportional biomass, assuming that the volume of pellets represented the biomass consumed. A subsample of 528 summer pellets, which was analysed by both methods, revealed that this assumption is valid (C. Rohner, unpubl. data). Sample sizes for winter diets were 18 pellets from 8 territories in 1989-90, 8 pellets from 4 territories in 1990-91, 13 pellets from 6 territories in 1991-92, and 9 pellets from 3 territories in 1992-93. RESULTS Prediction 1: Population growth during the increase phase The territorial owl population at Kluane grew from 11 to 20 pairs/1 OOkm^ during 1988-91 (Fig. 5.1a). This is equivalent to a yearly finite rate of increase of h=l.22. For comparison, the slope from the population growth of hares is overlaid in Fig. 5.1a, plotted from the beginning of the owl census data in 1988. Hare densities in spring were around 0.15/ha during the low phase in the mid 1980s, and then peaked to 1.25-1.61/ha (Krebs et al. 1992, Hik 1994, Chapter 2). For an increase of 4-5 years, this is equivalent to a yearly rate of increase of ^1.5-1.8. An almost identical pattern was found in Rochester, Alberta, during an earlier cycle with peak hare densities in 1971 (analysis of data from Adamcik et al. 1978, Fig. 5.1b). Densities of owl territories were lower, but increased at a similar yearly rate of X=1.26. Hare densities in spring were around 0.24/ha during the low phase, and increased to 5.1/ha at the peak in 1971 at a yearly rate of X=1.8. The growth rates of hare populations were clearly higher than those of territorial owls, but the differences are less pronounced when taking non-territorial birds into account. The growth of the total owl population including floaters increased at a yearly rate of X=1.50 (Fig. 5.1a). An estimate of potential growth in the owl population was calculated based on population parameters in Chapter 2, assuming that young owls, which 96 Figure 5.1 Annual rates of increase in populations of great homed owls (solid line, symbols) and snowshoe hares (broken line, slope overlaid onto owl scale). Years of peak hare densities are indicated by an arrow. A: Kluane, Yukon. B: Rochester, Alberta (data from Adamcik et al. 1978). 97 HARES I 100 O (n o £ o o j2 o CO o OT O E o o _c2 o 50 25 -10 potential owl population ^ total owl population territorial owl population 85 86 87 88 89 90 91 HARES territorial owl population 66 67 68 69 70 71 72 year 98 in reality become floaters, could establish territories and breed as yearlings. These hypothetical owl densities increased at a rate of h=l.66 (Fig. 5.1a). Prediction 2: Use of alternative prey during the decline of snowshoe hares To investigate the summer diet of great homed owls from 1989-92, a total of 946 prey items (322.7 kg in biomass) were identified from pellets. When hare densities were high (>1.1/ha) during the increase, peak, and the first year of decline, owl diets consisted almost exclusively of snowshoe hares (Fig. 5.2). The high proportions of 82.6-86.0% varied insignificantly during 1989-91 (Kruskal Wallis Test, approx. X2=0.91, DF=2, p=0.63, n=27). Hare densities kept declining to 0.43/ha by spring 1992, leading to a drastic change in the summer diet of great homed owls, when snowshoe hares composed only 12.7% of the total biomass. This shift in diet was statistically significant when compared to either all years 1989-91, or only to 1991 (U-Test, U=45, DF=1, p<0.005, n=14). The majority of alternative prey in the diet consisted of voles (mostly Microtus spp.), ground squirrels (Spermophilusparryii), and birds. Winter diets of great homed owls were remarkably different from summer diets, although the results have to be interpreted with caution because only few pellets were found. The proportion of snowshoe hares in winter diets was similar to that in summer diets as long as hare densities were high, but an important difference was observed during the hare decline (Fig. 5.2). In winter, the proportion of hares remained extremely high (83.3-92.3%) despite the rapidly dropping number of hares in the study area. Predictions 3a-d: Density-dependence and social behaviour The number of established owl territories increased throughout 1988-92, but the yearly increase declined towards higher densities of pairs already present (prediction 3b, 99 Figure 5.2 Proportion of snowshoe hares (main prey species) in the diet of great homed owls at Kluane, Yukon, during the last increase phase and the beginning of the decline in the population cycle (means+SE). The year of peak hare densities is indicated by an arrow. 100 100 -CD T3 C CO 0) ^. 03 v O 80 60 40 20 -"•-<> winter 1 0 1 Fig. 5.3a). Although the sample size of only 4 years is small, the negative slope of the regression is significant (y=1.67-0.03x, r2=0.95, p<0.05). The number of floaters increased with density of established owl territories (prediction 3c, Fig. 5.2, y=3.89x-44.84, r2=0.96, p<0.05). The initial floater density in 1988 was assumed to be zero. This may be an overly conservative assumption (Chapter 2). Adding a constant initial number of floaters, however, would affect only the intercept and not the slope of the regression. While monitoring radio-marked great homed owls we observed six vacancies in territories, which served as natural removal experiments (prediction 3d, Table 5.1). Territory holders either died or emigrated, and I recorded whether these vacancies were filled with new birds. In at least five of six vacancies, such replacements occurred. None of these owls were known individuals from the study area. In case 2, it was unclear if the territory holder had been replaced or not. Often it was difficult to observe successful replacements. Checks were made opportunistically, and the dates when new territory holders were confirmed do not reflect the accurate time of the replacement. The estimated intervals are therefore only the upper limits of the real intervals. In case 5, the previous female was observed to return to the occupied territory twice for less than two days. It was unclear whether this female was actively driven off by the replacing female. DISCUSSION Population growth and predator response Predators are usually larger, live longer, and have fewer offspring than their prey (Taylor 1984, Steams 1992). This translates into a lower rate of increase for populations of long-lived species (Caughley and Krebs 1983, Perrins 1991). Our analysis for two 102 Figure 5.3 Social behaviour and the limitation of population growth in great homed owls at Kluane, Yukon. A: Growth rates of the territorial population decline as numbers of owl territories increase in the area (inverse density-dependent growth rate). B: Numbers of non-territorial 'floaters' increase as territories are packed more densely (density-dependent increase). 103 ^ 1.5 sz 1 " 1.0 c .9 c3 §• 0.5 Q. A territories 1 1 y = 1.67-0.03x r'=0.95, p<0.05 .1 . , . . „ , 1 10 15 owl territories / 100 km 20 25 2 J 30 o o "^ 20 C/3 0) CO o 10 -B floaters -J—A-l 10 15 y . / y = 3.89X - 44.84 / r'=0.96, p<0.05 20 25 owl territories / 100 km' 104 Table 5.1 Natural removal experiments and replacements of radio-marked territorial great homed owls at Kluane Lake, Yukon. sex estimated cause vacancy replacement interval to confirmed replacement 1. female 10 Jul 1989 mortality 04 Dec 1990 ca. 4 months 2. female 28Junl991 mortality - ? 3. female 20 Nov 1991 mortahty 12 Mar 1992 < 3.5 months 4. male 25 Jan 1992 mortahty 10 Mar 1992 < 7 weeks 5. female 01 Feb 1992 emigration 12 Mar 1992 < 3 weeks 6. female 26 Feb 1992 mortality 11 Mar 1992 < 6 weeks 105 study sites in boreal Canada resulted in higher growth rates for populations of hares than for those of owls (Fig. 5.1). This result may not be surprising, but it is interesting that the differences in population growth almost disappeared when non-territorial owls were taken into account. The recovery rates of other populations may also be underestimated when floaters are not included. Most studies ignore the presence of floaters, because they are difficult to detect (Chapter 2). Rates of increase may also be underestimated for another reason. Population parameters of long-lived species are often based on measurements from stable populations (Perrins 1991), but if density-dependent mortality occurs (Sinclair 1989), these parameters are likely to underestimate the growth rate of populations with high food availability. The interpretation of population growth, however, does not directly test whether predators are food-limited by their main prey at high densities. Comparing rates of population increase does not provide information on the actual number of animals or on the actual number of prey consumed. For example, predators can vary their energetic requirements depending on their reproductive effort or movement activity (Kasparie 1983, Chapter 6), or they may kill more prey than expected from energetic considerations ('surplus killing', Keith et al. 1977, Jedrzejewska and Jedrzejewski 1989). In conclusion, the single species hypothesis (SPH) cannot be rejected based on the data presented. Prediction 1 was true for territorial owls, and (assuming consumption proportional to their density) they are expected to be below food-limitation by snowshoe hares at highest densities. This was less clear when non-territorial owls were taken into account. Given the data available, comparing population growth rates is insufficient to reach firm conclusions. 106 Multiple prey hypothesis In summer diets, the use of alternative prey increased drastically with a one year delay after the hare peak in 1990. This would indicate that great homed owls were not food-limited by their main prey in 1991, and then depended on other prey as an alternative food source in 1992. Therefore a mix of both hypotheses would explain the continued increase of the territorial owl population from 1990-92. There is some indication, however, that breeding great homed owls were food limited in 1991. Reproduction was reduced in comparison to the peak year (Chapter 2), fledgling survival was much lower than in previous years (Chapter 3), and breeding owls brought insufficient food to their broods when these were increased experimentally (Chapter 4). The winter diet of great homed owls consisted of very high proportions of main prey throughout the study period, independently of hare densities. This result does not fit prediction 2. It is likely that subarctic winter conditions represent a bottleneck, and that the food situation during this time determines the capacity for the number of territories in an area. In conclusion, the multiple prey hypothesis (MPH) is not a sufficient explanation for the time lag in the numerical response of great homed owls. Although alternative prey was being taken in higher proportions during the decline of snowshoe hares, this was not true for the winter months that were probably critical, and the inclusion of alternative prey in owl diets was not sufficient for successful reproduction. Density-dependence and territorial behaviour All predictions (3a-d) derived from the hypothesis that social behaviour limits the maximum density of owls were fulfilled. Some caution is warranted, however, because 107 the study lasted only 4 years, and because removals of territory holders were not manipulative experiments with complete information on all details of the replacements. An alternative explanation for the reduced accumulation of territories at high densities (prediction 3b, Fig. 5.3a) could simply be the fact that hare densities started to decline at that time, and that this reduced prey base reflected reduced potential for the establishment of new territories. Including hare densities into the regression, however, does not improve the fit, and a more comprehensive view of the situation in 1991 and 1992 helps answering the question. As discussed above, there is evidence that breeding owls were food-limited as soon as hares started to decline: reproduction was reduced in 1991, and no owls started breeding in 1992 (Chapter 2). Nevertheless, the territorial population continued to increase from 1991 to 1992. It is difficult to interpret this situation in any other way than by the presence of a high intruder pressure, which forced territory owners to give up space that would be required for successful reproduction. The influence of intruder pressure on territory size has mostly been studied in short-term feeding territories, where high food concentrations and high intruder pressure are usually correlated and very difficult to separate experimentally (Myers et al. 1979, review in Boutin 1990). This is different for long-term territories of great horned owls in the boreal forest. Prey densities decline on a large spatial scale, and floaters have no local food concentrations to turn to. This causes a de-coupling of food density and intruder pressure. Despite reduced food density, high intruder pressure persists and leads to reduced territory size and reproductive output. These results emphasize Carpenter's (1987) postulate that larger scale events must be taken into consideration to understand the territorial behaviour of individuals. 108 Another alternative explanation would predict that delayed maturation in young owls is an adaptive strategy to enhance life-time reproductive success, and that the causes of a large floater pool are physiological and not social (predictions 3a and 3c). In many long-lived species, young birds tend to breed earlier in good years, sometimes even in 'immature' plumage (reviews in Newton 1989a, 1992). Great homed owls are physiologically capable of breeding at the end of their first year of life (Chapter 2), and removed territory owners were replaced in at least 5 of 6 cases (Table 5.1). This does not imply that all floaters were excluded from breeding by social behaviour. Although some studies have shown that non-territorial behaviour or low breeding performance of young birds are not a superior alternative strategy (Smith and Arcese 1989, Newton 1989b, Korpimaki 1988b), more theoretical examination and critical tests of this hypothesis are needed for long-lived species. It is now well established that territorial behaviour limits the density of many raptor and owl populations (Southern 1970, Newton 1976, Village 1983, Hirons 1985, Newton 1992). It is interesting that not only stable populations, but also fluctuating populations linked to prey cycles of extreme amplitude may reach population ceilings without super-abundant food. A similar situation was suggested for Tengmalm's owls (Aegolius funereus) responding to vole cycles in Fennoscandia (Korpimaki 1988a, 1989). Individual strategies and the time lag in the predator response Neither MPH nor SPH are satisfactory explanations for the delayed decline of great homed owls relative to the snowshoe hare cycle. The key to understanding the processes involved seems to lie in the behaviour of individuals. Great homed owls are a socially stmctured predator population. Some individuals monopolize a disproportionate amount of resources, which they defend in large territories. 109 These territories are not individual feeding territories, but larger 'family territories', which supply the prey base necessary for increased reproduction. Breeding pairs may be operating close to food-limitation (Chapter 4), and the decline of the main prey translates into an immediate reduction of reproductive success. The time lag in the population response is not caused by an excess of food, but by a change of individual strategy. By reducing reproduction, a 'family territory' can still provide sufficient food for the owl pair to survive, and emigrations and mortahties do not occur at a substantial rate until a second year of further declining prey. Meanwhile, reproduction and post-fledging survival of juveniles were high during the increase, but most dispersers did not acquire territories (Chapters 2-3). Survival of non-territorial floaters was high, and food seemed to limit only breeding but not survival. Floaters responded to the hare decline with a shorter delay than territory holders (Chapter 2). It is unclear to what degree this difference is caused by territorial exclusion or by a lack of experience in younger birds (see also Newton 1986). The importance of behavioural changes in causing the time lag in the numerical response may also apply to other vertebrate predators. Other avian predators with less pronounced territorial behaviour showed no delay in their numerical response to vole cycles in Finland (Korpimaki and Norrdahl 1991). Goshawks (Accipiter gentilis), which may not have the option of conserving much energy with their fast flying hunting mode, also responded immediately to the decline of snowshoe hares both at Rochester and Kluane (Keith et al. 1977, Doyle and Smith 1994). There should be caution, however, in generalizing the findings of this study. Only one period of the hare cycle was studied, and specific features of systems may affect patterns of predator responses. Adamcik et al. (1978), for example, found that territorial great homed owls in Rochester did not continue to increase after the peak of the hare cycle 110 despite successful reproduction, and they concluded that territoriality was limiting owl numbers. Their study area consisted mostly of farmland (forest <40%), overall owl densities were much lower than in pristine forest at Kluane, and it is Hkely that in their case critical habitat (e.g. for nest sites) and not food was limiting. This has also been found for some other raptor species (Newton 1976, Village 1983). Conclusions What causes the time lag in the numerical response of a predator to its cyclic prey? The approach of contrasting SPH and MPH was a useful tool to bring this question into sharper focus, but also showed clear Umitations. Only in relatively simple systems can we expect proportional responses to external factors, such as changes in food availability. In the case of great homed owls, the time lag appears not to be directly related to external food sources, but also to the social structure in the population and changes in individual behaviour. Individual selection is likely to favour behaviours that provide constantly high food intake for predators, or in other words, that make predators less dependent on fluctuations of prey. Monopolizing resources by defending territories and adjusting energy expenditure to prey densities are fairly common in vertebrate predators (review in Taylor 1984), and it would be interesting to test whether they cause delayed numerical responses in other predator-prey systems. I l l CHAPTER 6 Life history variation in northern owls ABSTRACT Population cycles are common in arctic and subarctic environments, and northern predators experience strongly fluctuating food levels. In this paper, I review the life history variation in northern owls with particular emphasis to reproductive responses to temporarily high prey abundance. The most extreme differences were found in the largest owls of almost equal size. In snowy owls (Nyctea scandiaca), the reproductive response to prey cycles is four times higher than in Bubo owls. The mortality regime of great homed owls (Bubo virginianus) is affected by the snowshoe hare cycle, causing differential mortality on age classes. Differences in mortality regimes may cause habitat-specific differences in the life histories of northern owls. A review of reproductive parameters suggests that owls are generally linked to nocturnal activity, structured habitat with cover, sedentary life-style, and moderate reproductive response to increased prey densities. Only few species deviate strongly from this pattern, and the data suggests the presence of strong constraints. Different hypotheses to explain these deviations are presented, and the evolution of Bubo and Nyctea is discussed in more detail. Investigating the performance of animals under 'peak' conditions (affecting reproduction) and 'bottlenecks' (affecting survival) could be a promising approach to study how environments select for specific life histories. 112 INTRODUCTION Life histories evolve largely in response to the impact of different environments on the survival and fertility of different age classes (Partridge and Harvey 1988). How ecological contexts select for specific combinations of life history parameters is still an issue full of controversy. A theory of density-dependent selection was initially used to compare colonizing with established populations ('r- and K-selection', MacArthur and Wilson 1967, Charlesworth 1980). Other approaches have focussed on optimal life histories under temporal and spatial variability, and emphasized the strategies of extending reproductive life span ('bet-hedging') and of reducing the variance in the mean number of surviving offspring per lifetime (reviews in Stearns 1976, Gillespie 1977, Partridge and Harvey 1988, Steams 1992). Different habitats have often been described based on levels of productivity, disturbance, stability, or unpredictability. Steams (1992) considered 'unpredictable' to be ill defined in respect to its specific effect on life history traits. He pointed out that habitats do not map directly onto life histories but rather cause a specific mortality regime which acts as a selective mechanism (see also Promislow and Harvey 1990). There is a need for specific work on phylogenetically close species, for scaling temporal and spatial features of the corresponding habitats, and for seeking sets of pattems rather than a single generalization (Southwood 1988, Partridge and Harvey 1988, Steams 1992). I studied the population ecology of great homed owls (Bubo virginianus), one of the main predators of cyclic snowshoe hare populations in the nearctic boreal forest (Rohner 1994). Great homed owls are large, long-lived predators, and have a low annual reproductive output - a life history that sharply contrasts the predictions of r- and K-selection for a northem environment with extreme variations in food availability. Most ecological research on great homed owls has emphasized the dietary and demographic 113 responses to the snowshoe hare cycle (reviews in Adamcik et al. 1978, Houston 1987). I obtained similar results to those in previous studies, but the longer I worked on this northern predator-prey system, the more I realized how narrow and obviously constrained the phenotypic plasticity of great homed owls is when compared to other predators of cyclic prey. This paradox of conservative responses to extreme environmental variation led me to a broader review of the life history of northern owls. I chose 'northern' owls, because a feature common to all arctic and subarctic predator-prey systems are population cycles of hares, voles, or lemmings, which occur in periods of 8-11 (hares) and 3-4 years (voles and lemmings) with amphtudes of 15-200 fold (e.g. Keith 1963, Angelstam et al. 1985, Krebs et al. 1992, Stenseth and Ims 1993). In this paper, I compare a group of phylogenetically close species that occupy similar niches but markedly differ in their life histories. I first focus on the extreme difference in reproductive output of two closely-related species of almost identical body size (great homed owl and snowy owl). Then, I present a broader review of reproductive investment in northem owls, how variable investment relates to different environmental and behavioural characteristics, and how evolutionary mechanisms may have caused dichotomies in the adaptations of predators to prey cycles. RESULTS Extremes in reproductive output Snowy owls and the Bubo owls (eagle owl B. bubo in Eurasia and great homed owl B. virginianus in North America) are similar in size and closely related, but show extreme differences in their maximum reproduction. Great homed owls in the southwestern Yukon had adjusted brood sizes in relation to varying snowshoe hare densities of the years before. 114 during, and after a cyclic hare peak (Fig. 6.1a). During times of peak prey abundance, however, the asymmetry in the frequency distribution of brood sizes was extreme. Broods with three young were most common, but no broods of larger size were discovered (unpubl. data, see also Chapter 2). As an almost opposite pattern, the frequency distribution for snowy owls gradually petered off towards maximum clutch sizes, but the lower tail ended abruptly, with 14% clutches of five eggs but no clutches of smaller size (Fig. 6.1b, long-term data for northern Finland from Merikalho in Mikkola 1983). Other life history characteristics covary with this pronounced difference in reproduction (Table 6.1), and allometry (review in Steams 1992) does not account for this difference. Although snowy owls have longer wings and are heavier relative to great homed owls, they have larger clutches, smaller egg weights, a shorter breeding season, and faster nestling growth. Mortality regimes Assuming that reproduction is costly, models of reproductive effort predict that changes in mortality regimes will affect the reproductive investment of an organism (review in Steams 1992). For example, low reproductive output in great homed owls should correspond with high survival, which is not in agreement with reports of crashes in owl numbers following hare peaks (Adamcik et al. 1978, Keith and Rusch 1986). The survival of great homed owls is highest during the peak of the snowshoe hare cycle, and decreases during the low phase (Houston and Francis 1995, Chapters 2-3). Two aspects are of particular interest in this mortality regime (Fig. 6.2): (a) both age classes are affected, (b) adult great horned owls are less affected than juveniles and survive remarkably well through several lean years between peak densities of snowshoe hares. This age difference is significant (ANCOVA, Fj i7=44.67, p<0.001), but statistics 115 Figure 6.1 Frequency distribution of clutch sizes in large owls, (a) Frequency distribution of brood sizes in great homed owls at Kluane, southwestern Yukon, in years before, during, and after a cyclic peak of snowshoe hares, (b) Comparison of clutch sizes in great homed owls and snowy owls (data from Chapter 2, Merikallio in Mikkola 1983). For great homed owls, clutch sizes were not available and brood sizes are given instead (assuming low mortality of young nestlings, see Chapter 3). 116 percent (snowy owl) --— 8 ^ ° 1 ' 1 1 1 1 am m _ in CO CM 00 CO o N W CM XJ T- n in lO o> oo (0 (O in - ^ CO CM spoojq io Aouenbaji o o in ««r o o o CO CM 1 -(|M0 'i| '6) luaojad (Q 117 Table 6.1 Comparison of body size and some hfe history traits between great homed owls and snowy owls (means of both sexes). Wing length (mm) Body mass (g) Brood size Egg weight (g) Breeding period (months) Growth rate, K (logistic equation) Great homed owl 348 1505 2.3 64.4 6 .104 Snowy owl 430 1925 7.8 58.9 3.5 .120 Ratio snowy : g.h. owl 1.24 1.28 3.55 0.91 0.58 1.15 Source 1,2 3,4 5,6 4,2 5,7,2 5,7 Sources: 1) Earhart and Johnson (1970); 2)Portenko(1972); 3) Glutz and Bauer (1980); 4) Voous (1989); 5) this study; 6)Mikkola(1983); 7) Watson (1957). 118 Figure 6.2 Mortality regime of great homed owls during varying snowshoe hare densities. Annual survival rates for adults (filled symbols) and juveniles (open symbols). Data are from three consecutive hare cycles in Saskatchewan (Houston and Francis 1995, circles) and from three years during one cycle in the Yukon (Chapter 2, squares). 119 1.0 0) *•* (0 X . 75 > > ^m 3 0) (0 3 C C (Q 0.8 0.6 0.4 0.2 0.0 — -— cx) s^ adults o \ \ \ o \ \ \ \ o \ \ g N^^  juveniles N \ S \ \ 1 1 High Low Hare density 120 should be regarded with caution because the data points are not entirely independent. How do great horned owls survive through 4-5 years of extremely low hare densities? Four different (but not mutually exclusive) strategies have been proposed: (a) Large-scale movements: Although adult great homed owls are sedentary and long-term territorial (Chapter 2), a fraction of juvenile and adult owls leave and move long distances as hare densities decline, possibly reaching areas of higher prey availability (Houston and Francis 1995, Keith and Rusch 1989, Chapter 2). Presently, it is unclear what proportion of these owls survive or even return. (b) Alternative prey: Great homed owls can increase the proportion of other prey items in their diet when snowshoe hares are scarce (Adamcik et al. 1978). Prey diversity, however, is very limited in the boreal forest during subarctic winter conditions. Even at low hare densities, the winter diet of great homed owls at 61° N in the Yukon consisted almost entirely of snowshoe hares (Chapter 5). (c) Energy conservation: Kasparie (1983) found that captive great homed owls decreased duration and intensity of their activity when fasted. Reduced mobility is predicted by optimality models for perch hunters at low prey densities (Andersson 1981). Great homed owls also lowered their body temperature as a response to food shortage. Based on the regression in Chaplin et al. (1983), the drop in body temperature measured by Kasparie (1983) is equivalent to a reduction of the metabolic rate by about 15%. Besides reducing energy expenditure, owls can survive extended periods of food shortage by losing up to 24% of their initial weight without damage (Ligon 1969, Gessaman 1978, Ceska 1980). (d) Spatial memory: Many bird species show an increase in survival with age (Lack 1954, Newton 1989b). The rate of learning appears to be limited, and when foraging or other essential tasks are difficult, skills improve only slowly over time and 121 with age. Local knowledge may improve foraging efficiency and thus enhance survival at low prey densities. Martin (1986,1990) hypothesized that highly-developed spatial memory may be essential for high hunting success at low light levels, when visual information on obstacles is incomplete in habitats with much cover such as woodlands. In addition, ownership of territories may increase the competitive advantage in successive years. In contrast to great homed owls, virtually no information is available on mortality in snowy owls. Based on environmental conditions, however, differences in the mortality regimes for the two species can be hypothesized: (a) Seasonality: Arctic winters are so extreme and the accessibility of rodents is so much reduced that long-term territoriality is not viable. Snowy owls often leave the breeding grounds before the young are independent (Voous 1989), and although some remain in the arctic in winter, most seem to migrate south and spend the winter months in the temperate zone (Kerlinger and Lein 1988). For a species with high wing loading (ratio of mass to wing area, Johnson 1981), adopting a migratory strategy may be costly and lead to elevated levels of mortality. (b) Spatio-temporal heterogeneity: Little is known about the synchronization of lemming cycles across the Nearctic, but the pattern appears less homogeneous than for the snowshoe hare cycle in the boreal forest (Chitty 1950, Reid 1994). Perhaps as a response to this lack of predictability for a given site, snowy owls are highly nomadic, and breeding pairs concentrate where small mammals reach their cyclic peaks (Voous 1989). Records taken over six years during waterfowl surveys by aircraft across the coastal plain in Alaska showed that breeding sites of snowy owls were not occupied consistently and that the annual variation in breeding pairs was so high that large scale movements were needed to explain local population dynamics (Brackney and King 1991). A nomadic strategy 122 implies losing the benefit of local knowledge. As a consequence, snowy owls may not be able to increase their survival for each year of age as the Bubo owls do. Although there are no data at present, it would be interesting to test whether the ratio of adult to juvenile survival is lower in snowy owls than in great homed owls. Variation of reproductive investment among northern owls How much can reproduction increase under extremely high prey densities in other owl species? For a broader comparison, I compiled data from the arctic, boreal, and boreo-nemoral zone where population cycles of small mammals are predominant (Solheim 1984, Angelstam et al. 1985, Keith 1963). Maximum reproduction over a range of environmental conditions can be measured in several ways. Maximum clutch sizes are easy to find in the ornithological literature, but there are several drawbacks to this variable, such as dependency on sample size, several females laying in the same nest, or reporting errors particularly in earlier records. For analysis, I used the maximum of average clutch sizes that have been sampled in one season at one site as a more robust variable, although sample sizes were small for some species (Table 6.2). Both variables were highly correlated (Spearman r5=0.95, p<0.01, n=ll). Clutch sizes vary considerably, and short-eared owls (Asio flammeus) and northern hawk owls (Sumia ulula) have also been observed to lay more than 10 eggs, with averages of 9.6-10.2 in peak years (Table 6.2). Several patterns become apparent when maximum average clutches are compared to body mass (Fig. 6.3). Clutch size tends to be smaller in larger owl species (rs=-0.52, p<0.10), and the overall level of reproductive output is higher in northern latitudes compared to the temperate zone. At least for some owl species, food availability and not merely geographic location causes this difference 123 Table 6.2 Comparison of reproductive parameters in northern owls in peak years of prey cycles in subarctic and arctic biomes. Species maximum mean clutch size sample size clutch size in maximum year in max. year source bo Snowy owl (Nyctea scandiaca) Short-eared owl (Asio flammeus) Northern hawk owl (Sumia ulula) Pygmy owl (Glaucidium passerinum) Tengmalm's owl (Aegolius funereus) Long-eared owl (Asio otus) Great gray owl (Strix nebulosa) Ural owl (Strix uralensis) Tawny owl (Strix aluco) Eagle owl (Bubo bubo) Great homed owl (Bubo virginianus) 15 14 13 10 10 9 9 8 6 5 4 9.8 9.6 10.2 >7.3> 6.7 6.3 4.6 4.6 4.2 >2.6i 3.1 11 65 5 7 48 7 7 63 70 196 7 1,2 3,4 5 6 7 8 9 10,11 3, 12 5 13 '^mean over several years of high or average prey density Sources: l)Portenko (1972); 2)ParmeleeetaI. (1967); 3) Mikkola (1983); 4) Mikkola and Sulkava (1969); 5) Merikallio cit. in Mikkola (1983); 6) Solheim(1984); 7) Korpimaki (1991); 8) Korpimaki (1992a); 9) Mikkola (1981); 10) Saurola 1989; 11) Pietiainen (1989); 12) Linkola and Myllymaki (1969); 13) Adamcik et al. (1978). Figure 6.3 Clutch sizes in northern owls, as a function of body mass. Lines: regression and 95% confidence limits. Broken line: Regression for the same species in the temperate zone. 125 (D N "w O O NYCTEA AEGOLIUS F. STRIXA. 2 -STRIX U. BUBO V. BUBO B. J L 100 250 500 1000 3000 body mass (g) 126 (Korpimaki 1986). In Fig. 6.3,1 plotted the regression lines and confidence limits including all species, and found that snowy owl, short-eared owl, and northern hawk owl deviate significantly in their clutch sizes from other northern owls. As a general pattern, body mass seems to predict clutch size, but a few species deviate from this pattern and show extreme reproductive responses to high food densities. Considering egg volumes gives further support for this pattern in northern owls. Egg volumes were calculated using Tatum's (1975) index, with data on egg length and egg width based on the review by Glutz and Bauer (1980). The effect of body mass was removed for both egg volume and clutch size, and the residuals are shown in Fig. 6.4. There is a trend for a negative relationship between egg volume and clutch size, suggesting a trade-off in the allocation of available resources to either size or number of offspring (rs=-0.48, p<0.10). The three species with the highest reproductive output (snowy owl, short-eared owl, and hawk owl) had also the lowest egg volumes. Interpretations of individual data points on this graph should not be taken too far, because the sources of measurements were quite heterogeneous. Body weights of owls show considerable seasonal variation (Glutz and Bauer 1980, Ceska 1980). To reduce variance for further comparisons, weights should be taken at the same stage of the breeding cycle, and compared to egg volumes from the same location. In addition to considering the average maximum reproductive rate, the variation in the distribution of clutch sizes is also of interest. Mikkola (1983) provided a subsample of frequency distributions for 7 species, and I added information for great homed owls from my own records (Chapter 2) and from Adamcik et al. (1978). The frequency distributions were standardized by shifting them to a common mode (most frequent clutch size) of zero. The average distribution for all species was then calculated, and the deviations or residuals 127 Figure 6.4 Trade-off between egg volume and clutch size in northern owls (with regression line and 95% confidence limits). Residuals on log-transformed values are used for these analyses to remove the effects of body mass. 128 o E _3 o > O) (D 0.1 -0 •0.1 --1 \ 1 ' \ \ STRIX A. \ Nf STRIX U. \ \ / w • \ \ / \ / * \ BUBO V. V ^.^^^^ STRIX N. N. BUBO B. • ySC. GLAUCIDIUM P . / Nv AEGOLIUS F. 1 1 ASIO 0. ASIO F.' r -SURNIA U. / • \ • \ X NYCTEA \ . \ 1 \ •0.5 0 0.5 clutch size 129 from this common distribution were calculated for each species. In Fig. 6.5, the average residuals for the lower and upper tail of the distribution are plotted. Depending on the variance, a symmetric distribution would have either positive or negative deviation from the overall distribution for both tails. The negative relationship in Fig. 6.5 shows that the frequency distributions tend to be asymmetric, with an extended tail on one side, and a truncation on the other (example in Fig. 6. lb). This indicates that the variation is not random, but subject to some constraint or selective disadvantage. Behavioural, environmental, and phylogenetic correlates In theoretical models, large clutch sizes favours nomadism under cyclic food availability (Andersson 1980). Other behavioural adaptations covarying with clutch size could involve modifications in hunting behaviour, for example increased effort by flight hunting (Masman et al. 1988) or activity extended into daytime. Prey accessibility may be dependent on cover (Janes 1985), therefore resulting in different prey availability for woodlands and more open habitats. I quantified these variables by scoring them from 1 (lowest) to 3 (highest) based on handbooks and the ornithological literature (Table 6.3), and then calculated rank correlations (Table 6.4). Clutch sizes increased with the intensity of nomadism, and they were negatively related to cover (larger clutches in open or semi-open habitats). There was a weaker positive correlation with diurnal activity and a trend for larger clutches in flight hunting owls. The variables other than clutch size were correlated with each other, tending towards a higher degree of nomadism, and perhaps flight hunting and diurnal activity, in open habitats. Statistics with such data, however, are problematic because data points are not independent. Species are related to each other to varying degrees, and this poses a 130 Figure 6.5 Asymmetry in the frequency distribution of clutch sizes in northern owls. The deviations from the overall distribution for all owls are plotted, separated into deviations in the lower tail (clutches smaller than the mode) and in the upper tail (clutches larger than the mode). 131 CO CD o c o CO " > CD • Q c CO CD E 0.2 0 0.2 , , . ._, , , . . BUBO V. / • / / • • STRIX A. / STRIX U. 1 I 1 ASIO F. / BUBO B. / GLAUCIDIUM P. / / • ' • SURNIA U. / -NYCTEA 1 1 -0.025 0 0.025 mean deviation (upper tail) 132 Table 6.3 Comparison of behavioural and habitat characteristics of northern owls in prey cycles in subarctic and arctic biomes. Variables are scores with values 1-3 according to intensity of factor (main sources: Mikkola 1983, Voous 1989, Korpimaki 1992). species habitat cover flight hunting diurnal activity nomadic OJ Eagle owl (Bubo bubo) Great homed owl (Bubo virginianus) Snowy owl (Nyctea scandiaca) Great gray owl (Strix nebulosa) Tawny owl (Strix aluco) Ural owl (Strix uralensis) Pygmy owl (Glaucidium passerinum) Tengmalm's owl (Aegolius funereus) Northern hawk owl (Sumia ulula) Short-eared owl (Asia flammeus) Long-eared owl (Asia otus) 3 3 1 2 3 3 3 3 2 1 1 2 1 2 3 3 1 1 3 2 1 1 3 1 3 3 1 1 1 3 2 1 1 1 2 3 3 3 Table 6.4 Rank correlations between clutch size and behavioural and environmental variables (ranked 1-3 according to intensity), (a) Correlations with effect of body mass removed, (b) correlations with effect of body mass and effect of phylogeny removed. Clutch size Habitat cover Flight hunting Diurnal activity (a) (b) * * Habitat cover Flight hunting Diurnal activity Nomadism Habitat cover Flight hunting Diurnal activity Nomadism -0.85 0.44 0.65* 0.89** -0.27 0.69* 0.27 -0.56 -0.53 -0.91** -0.67* -1.00** 0.45 0.55 0.51 0.67 *p<0.05; **p<0.01 134 problem of pseudo-replication. Felsenstein (1985) suggested that data points should be weighted according to molecular information on relatedness, but at present, samples are incomplete for an analysis at the species level in owls (Sibley and Ahlquist 1990). Alternatively, phylogeny can be subtracted by performing calculations only on residuals of species on the means of a taxonomic unit (Steams 1992). In the owl sample, there was a significant effect of phylogeny on the genus level (ANOVA, Fg ,i=7.36, p<0.05). The results in Table 6.4 are considerably less pronounced after removing the phylogenetic effect, although the same trends remained. This method may be overly conservative when the extent of phylogenetic constraints is unknown (Pagel and Harvey 1988). In addition, the sample size collapsed because several of the owl genera had only one species in northern environments, and the uncertainties in the current classification may distort the results. For example, the genus Nyctea should be pooled with Bubo according to morphological and molecular information, whereas results from DNA hybridization suggested that Strix nebulosa should be placed into a distinct genus (Voous 1989, Sibley and Ahlquist 1990). The problem of phylogenetic effect, however, cannot be ignored, and the ideas presented in this paper need further testing as more information becomes available. DISCUSSION Constraints in phenotypic plasticity All species of northern owls increase their reproduction in response to peaks of cyclic prey (regression lines in Fig. 6.2, review in Korpimaki 1992b). The asymmetry in the frequency distribution of clutch sizes (Fig. 6.1, Fig. 6.5) and the statistical effect of phylogeny on clutch size, also indicate that many northern owls experience strong 135 constraints in their reproductive response to cyclic prey abundance. Most species increase reproduction moderately during prey peaks, and only few respond extremely. What causes these differences in upper contraints on clutch size? Three different hypotheses, which complement rather than exclude each other, are discussed below. (a) Non-adaptive explanations: Reproductive strategies of northern owls are not necessarily adapted to cyclic prey (see Gould and Lewontin 1979). Great homed owls, for example, may not have been exposed long enough to cyclic snowshoe hare populations to have evolved an increased phenotypic plasticity in clutch size. Restricted variation in small brood sizes may account for a slower evolutionary rate (see also Gaston 1992), or gene flow from southern populations may be sufficient to prevent adaptations. Great homed owls do not seem to vary much in their clutch sizes throughout their range (Murray 1976, Johnsgard 1988, Voous 1989). (b) Physiological trade-offs: Because of a possible trade-off between current and future reproduction, long-lived species may limit their energy expenditure for reproduction to a level below their physiological capacity (review in Ydenberg 1994). The maximum working load may be related to basal metabolic rate, and is perhaps a feature of physiological design (Drent and Daan 1980, Daan et al. 1990). This hypothesis may explain the proximate causes for a physiological constraint on maximum brood size. (a) Food limitation: Lack (1946) suggested that avian predators experience periods of super-abundant food during peaks of prey cycles. At least for Tengmalm's owls (Korpimaki 1989) and for great homed owls (Chapters 4-5), this is not necessarily true. The prey base may be overestimated, when their average densities are considered, but not changes in their availability over time. Reproductive success is not dependent on the arithmetic mean of food levels but rather on geometric mean fitness, which takes rare events that lead to starvation into full account (Gillespie 1977, Boyce and Perrins 1987). 136 Habitat and the diversity of life-histories Can these hypotheses be combined to explain differences in the life history variation of northern owls, and what evidence do we have? Some theories in behavioural ecology give insight into the mechanisms that may lead to habitat-related dichotomies in the life history evolution of predatory animals. In his review of provisioning in birds, Ydenberg (1994) points out the implications of rate maximizing versus efficiency maximizing as foraging behaviours, and that efficiency maximizing may be more frequently found than predicted by conventional foraging theory. He does not explicitly deal with changes in prey density, but Andersson (1981) makes specific predictions for foraging predators. When prey density is high, increased foraging effort and rate maximizing can be profitable, whereas at low prey densities, efficiency maximizing is more profitable. Field results agree with these predictions. European kestrels (Falco tinnunculus), for example, use the expensive mode of flight hunting more often when prey is abundant, whereas they preferred perch hunting at low prey abundance (Masman et al. 1988). This basic dichotomy into different strategies depending on environmental condition may affect the evolution of life histories. Prey availibility does not only vary within, but also between habitats. Habitat features affecting life histories have been represented by a number of different approaches, for example as axes of varying disturbance or adversity (reviews in Southwood 1988, Partridge and Harvey 1988). Steams (1992) argued that absolute habitat features are bound to have different effects on different organisms living in the same habitat. Effects of habitat are always interactions with an organism. Differences in prey levels, for example, should be measured from the perspective of an animal and should be defined by a measure such as hunting yield per unit effort (Wijnandts 1984). 137 I suggest that a promising approach could consist in combining 'peaks' (highest prey levels) and 'bottlenecks' (lowest prey levels). Animals optimizing reproductive performance breed during highest prey availability (e.g. Masman 1986), whereas in bottlenecks their main concern is to survive. Particularly seasonal environments are poorly described by an overall measure of variation, because adaptations of animals to reproduce or survive at the extreme ends of such conditions are fundamentally different. As an example illustrated above, rate maximizing would result in maximum reproduction during summer, and efficiency maximizing would improve survival during winter. On theoretical grounds, Abrams (1991) showed that organisms whose foraging goal is to maximize survival are especially unlikely to increase foraging effort in response to increased food availability. Trade-offs between the optimal behaviour during highest and lowest prey abundance can be physiological in nature. For example, the size of organs such as heart, lung, or kidney, or the endocrinal function may have the cost of a high basal metabolism (Daan et al. 1990), and therefore reduce survival during bottlenecks because of reduced efficiency. (Such trade-offs may also account for dichotomies such as reversed size dimorphism in raptorial birds, Hakkarainen and Korpimaki 1991, Ydenberg and Forbes 1991). At the extreme ends of this food availability gradient, an organism could specialize either into a conservative life-style with little reproduction, or into an expensive strategy. An expensive physiological machinery may only be adaptive when exploiting rich habitat patches, but not in bottlenecks. The evolution of a migratory or nomadic strategy to avoid conditions of low food levels may be a direct consequence of such constraints. Different environments or habitats are likely to offer certain opportunities for reproduction, but also impose limitations for survival. From the perspective of an organism, different environments require different alterations in life history traits for a 138 successful colonization. In the case of northern owls, such habitat-specific 'life history profiles' could be based on peak and bottlenecks in prey availability. 'Life history profiles' As an example, simplistic 'life history profiles' for arctic and subarctic environments are portrayed, from the perspective of an ancestral Bubo owl, as a graphic model in Fig. 6.6. Boreal forests (taiga) with hares as the main prey allow the realization of high adult survival during extreme food shortages (X-axis). This was illustrated by the strongly differential mortality for great horned owls (Fig. 6.2). One hypothesis, which may also apply to other owls, has been introduced by Martin (1990) as the 'nocturnal syndrome'. Martin argues that the accumulation of knowledge on the local topography complements the sensory capabilities of owls in structured habitat such as forests and in combination with a sedentary life-style, enables experienced adults to survive through extremely low food levels. This argument is consistent with a strategy of conserving energy at low food levels as discussed before. This X-axis (bottleneck-axis) could be quantified by measuring the performance of birds, for example as the yearly decrease in the effort expended per unit hunting yield of prey, either by comparing age classes or individuals in successive years. On the Y-axis ('peak' axis), this same taiga-hare system seems to offer only limited opportunities for increased reproduction even at high prey densities. Because snowshoe hares live in a structured habitat with abundant cover, they may be able to vary their exposure to predation risk, and therefore reduce their availability for great homed owls (Chapters 4-5, Rohner and Krebs 1994). If owl parents are able to obtain more food during the breeding season, this may only be possible with a considerable increase in 139 Figure 6.6 Hypothesized 'life history profiles' (broken lines) for an ancestral Bubo owl in two northern environments with their associated prey. Differences along the 'peak' axis are caused by different accessibility of prey, whereas differences along the 'bottleneck' axis are caused by habitat-related limitations to an age-specific increase in foraging efficiency (quantified as the yearly decrease of effort necessary per unit hunting yield). Overlaying a fitness isocline (solid line) indicates a dichotomy into separate life history tactics (in this case high reproduction and low survival in the tundra, low reproduction and high survival in the taiga). 140 ® l © o -5 •-UJ s tundra-lemming system 'LIFE HISTORY PROFILES' taiga-hare system line of I equal fitness A effort (unit yield)'^  year"^  'BOTTLENECKS' 141 effort. If there is a trade-off between reproductive effort and adult survival, larger broods will result in lowered fitness, therefore restricting the 'life history profile' along the Y-axis in Fig. 6.6. This axis could be quantified by measuring the hunting yield per unit effort of provisioning males. 'Life history profiles' for a Bubo owl look quite different for the arctic tundra with lemmings as the main prey. The value on the X-axis is small, increments in age-specific survival are not as easily possible when breeding grounds are of only ephemeral use, and stochastic events are likely to predominate when searching for high quality patches (see mortality regimes). The Y-axis, however, seems to offer opportunities for increased reproduction with little or no extra effort during lemming peaks. Little is known about the foraging behaviour of snowy owls, but some evidence suggests that the lack of cover makes prey more accessible in the tundra in summer. Watson (1957) watched a male snowy owl hunting on 10 occasions, in which it never failed to catch a lemming within 5 minutes of continuous hunting. Even when this male was feeding 12-13 young and two females in July, it spent hours sleeping or resting and only about 20% hunting. None of the young lost weight when Watson (1957) removed all lemmings delivered to a nest site, and hunting seemed possible throughout the 24-hour day, even under adverse weather conditions. Evolution of northern owls The evolutionary consequences of habitat-specific 'Ufe history profiles' can be explored by super-imposing a line of equal fitness. Assuming a trade-off between reproductive effort and survival, and given the 'life history profiles' in these habitats, the fitness isocline is discontinuous: the tundra-lemming system can be colonized only by 142 realizing a life history tactic that is distinctly different from the one possible in the taiga-hare system. The vicariant life history tactics in the diurnal snowy owl and the nocturnal Bubo owls are reflected in their parapatric distribution (see Voous 1989 for maps). Great homed owls inhabit almost every habitat from the subarctic tree limit towards the southern end of the South American continent. I have observed several great homed owls in the alpine tundra of the southern and northern Yukon, but no breeding records have been reported outside the northern tree line. It is unknown what prevents the opportunistic Bubo owls from colonizing the arctic. Constrained maximum clutch size may result in an inability to compensate for the severe mortality regime in the arctic, or constraints in deviating from establishing long-term territories and minimizing energy expenditure over winter may be lethal. I agree with others that snowy owls evolved from Bubo owls (review in Voous 1989), and the above life history considerations may provide a selective mechanism. Speciation may indicate that the constraints are strong, perhaps because simultaneous changes of several traits are necessary in order to deviate successfully from the ancestral 'nocturnal syndrome' (Martin 1990). Conclusions This review on life history variation is preliminary, and it would be interesting to test covariations of traits, habitat-related dichotomies, and phylogenetic constraints with extended and refined data. A negative correlation, for example, indicating a trade-off between clutch size and egg volumes, was also found in South African owls (D. Ward, unpubl. data), but not among individuals within some intensively studied owl species in Fennoscandia (Wallin 1988, Hakkarainen 1994). 143 The concept of life history profiles for specific environments may be of limited use, and my graphical model needs formal testing by an analytical model. It makes, however, gaps in our knowledge apparent and can be used to derive some testable predictions. For example, we know very little about mortality regimes and foraging efficiencies of most predators, and particularly how they vary among habitats. A prediction worth testing concerns the interaction of physiology and environment in trade-offs. For example, it would be interesting to test whether snowy owls increase their reproductive effort relative to other Bubo owls, or if the effort is constant and increased reproduction merely reflects a better habitat. Similar tests could be performed within species comparing populations in the temperate and boreal zone (e.g. Korpimaki 1986). 144 CHAPTER 7 Conclusions Floaters and demography Little information is available on densities of non-territorial floaters in natural populations. The very high survival of juvenile great homed owls during peak years of the snowshoe hare cycle led to an estimated proportion of 40-50% floaters in the total population. My results suggest that deterioration in habitat quality affects floaters before territorial birds, and censuses of territorial males may therefore mask the beginning of a population decline. This confirms the suggestion by Wilcove and Terborgh (1984). Further data is needed to evaluate such effects in other species. Floaters are common in bird populations (Newton 1992), but depending on life span and age at maturity, the effect of floaters masking population trends may be negligible for small birds such as passerines. In stable populations, which are not cyclic as in this study, density-dependent mortality (Sinclair 1989) may reduce the size of floater pools to lower levels. Disease and survival Most birds in natural populations are infected by blood parasites, but little is known how these parasites affect their hosts (Bennett et al. 1993). My results on great homed owl mortalities associated with high levels of Leucocytozoon are among the first records of lethal effects in bird populations. Such effects of anaemia may be more common in altricial species with slow juvenile development, such as owls and raptors. My results also suggest, however, that mortalities likely occurred as an interaction with food shortage. Evidence for interactions between food and parasites are also emerging from other studies (e.g. Korpimaki et al. 1993). 145 Haemotophagous black flies lacerate the skin of birds, extract their blood, and can therefore have direct effects on birds beside transmitting blood parasites. The extent of such effects are unknown. The results of this thesis indicate that great homed owls avoided black flies by adjusting their roost site selection while accepting a higher risk of predation. Such trade-offs involving avoidance of parasites provide an exciting direction for future research. Food limitation at the peak of prey cycles? Cyclic peaks have been interpreted as conditions of super-abundant food for predators (Lack 1946). This thesis provides evidence that enlarged broods of great homed owls lost weight, despite increased hunting effort by females when hare densities were high. This suggests food limitation, although for a critical test further information is needed on the hunting effort by males, which are usually the sole providers during breeding. My results on social limitation of the number of breeding territories and the immediate response of reproduction to declining prey abundance also indicate that great homed owls were food limited at high hare densities. Although further evidence is needed to reach firm conclusions, this result may have broader implications. Evidence for super-abundant food would reject the hypothesis of direct effects of predation on peak hare populations, but at present, this evidence is lacking. Evolution of reproductive responses to cyclic prey peaks The reproduction of great homed owls increased as a function of hare density. This confirms previous research in southem parts of the boreal zone (Adamcik et al. 1978, Houston 1987), and I show in more detail how early mortalities affect the reproductive 146 success of great homed owls under varying food conditions. The frequency distribution of brood sizes, however, indicated strong constraints in the phenotypic plasticity of reproductive investment, particularly when compared to a closely-related species such as the snowy owl. A broader comparison among northern owls, which are all exposed to prey cycles, revealed a general pattern of moderate reproductive responses when allometry was taken into account. Only few species strongly deviate from this pattern, and seem to have expanded their upper contraints in reproductive investment. I suggest that this adaptation is accompanied by a habitat shift to open hunting grounds, where prey is more accessible. Mortality regimes seem to correspond with these habitat differences, and they may favour the diversity in observed life histories. Contrasting individual performance during prey 'peaks' and 'bottlenecks' illustrate this pattern for the evolution of the closely-related great homed owl and snowy owl, which are parapatric and seem to have vicariant life histories. While snowy owls have co-evolved with cyclic lemming populations, it is unclear whether great homed owls in the boreal forest are constrained in adapting to cyclic populations of snowshoe hares, or whether their strategy of moderate reproductive investment is optimal in a habitat with abundant cover and agile prey. Despite the difficulties involved, studying predators of cyclic prey is a unique opportunity to leam about the interactions of organisms with their environment. Investigating the behavioural and physiological constraints of a paradoxically conservative predator like the great horned owl may open new and exciting research directions. Without the integration of ecological and evolutionary approaches it is impossible to fully appreciate cyclic predator-prey systems. 147 LITERATURE CITED ABRAMS, P. A. 1991. Life history and the relationship between food availability and foraging effort. Ecology 72:1242-1252. ADAMCIK, R. S., and L. B. KEITH. 1978. Regional movements and mortality of great homed owls in relation to snowshoe hare fluctuations. Can. Field-Nat. 92:228-234. ADAMCIK, R. S., A. W. TODD, and L. B. KEITH. 1978. Demographic and dietary responses of great homed owls during a snowshoe hare cycle. Can. Field-Nat. 92:156-166. AK^AKAYA, H. R. 1992. Population cycles of mammals: evidence for a ratio-dependent predation hypothesis. Ecol. Monogr. 62:119-142. ANDERSSON, M. 1980. Nomadism and nest site tenacity as alternative reproductive tactics in birds. J. Anim. Ecol. 49:175-184. ANDERSSON, M. 1981. On optimal predator search. Theor. Pop. Biol. 19:58-86. ANGELSTAM, P., E. LINDSTROM, and P. WIDEN. 1984. Role of predation in short-term population fluctuations of some birds and mammals in Fennoscandia. Oecologia 62:199-208. ANGELSTAM, P., E. LINDSTROM, and P. WIDEN. 1985. Synchronous short-term population fluctuations of some birds and mammals in Fennoscandia - occurrence and distribution. Holarct. Ecol. 8:285-298. ASHFORD, R. W. 1971. Blood parasites and migratory fat at Lake Chad. Ibis 224:100-101. AUSTING, G. R., and J. B. HOLT. 1966. The world of the Great Homed Owl. Lippincott, New York. BECHARD, M. J., B. W. ZOELLICK, and M. NICKERSON. 1985. Accuracy in determining the age of nestling red-tailed hawks. J. Wildl. Manage. 49:226-228. BORTOLOTTI, G. R. 1985. Frequency of Protocalliphora avium (Diptera: Calliphoridae) infestations on Bald Eagles (Haliaeetus leucocephalus). Can. J. Zool. 63:165-168. BEISSINGER, S. R. 1990. Experimental brood manipulations and the monoparental threshold in snail kites. Am. Nat. 136:20-38. BELTHOFF, J. R., and G. RITCHISON. 1989. Natal dispersal of eastern screech-owls. Condor 91:254-265. 148 BELTHOFF, J. R., and G. RITCHISON. 1990. Roosting behavior of postfledging eastern screech-owls. Auk 107:567-579. BENNETT, G. F. 1960. On some omithophilic blood-sucking Diptera in Algonquin Park, Ontario, Canada. Can. J. Zool. 38:377-389. BENNETT, G. F., and M. BISHOP. 1990. Change in status of haematozooan infections in wild passeriforms sampled in successive years. Proc. zool. Soc, Calcutta 43:9-18. BENNETT, G. F., and R. F. COOMBS. 1975. Omithophilic vectors of avian haematozoa in insular Newfoundland. Can. J. Zool. 53:1241-1246. BENNETT, G. F., and A. M. FALLIS. 1960. Blood parasites of birds in Algonquin Park, Canada, and a discussion of their transmission. Can. J. Zool. 38:261-272. BENNETT, G. F., M. WHITEAWAY, and C. WOODWORTH-LYNAS. 1982. A host-parasite catalogue of the avian haematozoa. Memorial University of Newfoundland Occasional Papers in Biology 5:1-243. BENNETT, G. F., J. R. CAINES, and M. BISHOP. 1988. Influence of blood parasites on the body mass of passeriform birds. J. Wildl. Diseases 24:339-343. BENNETT, G. F., M. A. PEIRCE, and R. W. ASHFORD. 1993. Avian Haemotozoa: mortality and pathogenicity. J. Nat. Hist. 27:993-1001. BIRKHEAD, T. R., S. F. EDEN, K. CARKSON, S. F. GOODBURN, and J. PELLATT. 1986. Social organisation of a population of Magpies Pica pica. Ardea 74:59-68. BISHOP, M., and G. F. BENNETT. 1992. Host-parasite catalogue of the avian haematozoa. Memorial University of Newfoundland Occasional Papers in Biology 15:1-244. BOONSTRA, R., and G. R. SINGLETON. 1993. Population declines in the snowshoe hare and the role of stress. Gen. Comp. Endocrinol. 91:126-143. BOSAKOWSKI, T., R. SPEISER, and D. G. SMITH. 1989. Nesting ecology of forest-dwelling great homed owls, Bubo virginianus, in the eastem deciduous forest biome. Can. Field-Nat. 103:65-69. BOUTIN, S. 1990. Food supplementation experiments with terrestrial vertebrates: patterns, problems, and the future. Can. J. Zool. 68:203-220. BOUTIN, S., C. J. KREBS, A. R. E. SINCLAIR, J. N. M. SMITH, R. BOONSTRA, K. MARTIN, M. R. T. DALE, R. TURKINGTON, M. O'DONOGHUE, F. DOYLE, C. DOYLE, M. BLOWER, E. HOFER, S. SCHWEIGER, V. NAMS, C. ROHNER, A. HUBBS, D. HIK, and D. MURRAY. 1994. Population changes of the vertebrate community during a snowshoe hare cycle in Canada's boreal forest. MS, submitted to Oikos. 149 BOYCE, M. S., and C. M. PERRINS. 1987. Optimizing great tit clutch size in a fluctuating environment. Ecology 68:142-153. BRACKNEY, A. W., and R. J. KING. 1991. Population shifts by snowy owls on the arctic coastal plain of Alaska. Proc. Alaska Bird Conf. 4:69. BREITENMOSER, U., B. G. SLOUGH, and C. BREITENMOSER-WURSTEN. 1993. Predators of cyclic prey: is the Canada lynx victim or profiteer of the snowshoe hare cycle? Oikos 66:551-554. BROWN, J. L. 1964. The evolution of avian diversity in avian territorial systems. Wilson Bull. 76:160-169. BROWN, J. L. 1969. Territorial behavior and population regulation in birds. Wilson Bull. 81:293-329. BRYANT, J. P. 1981. The regulation of snowshoe hare feeding behaviour during winter by plant anti-herbivore chemistry, p. 720-731. In K. Myers and C. D. Maclnnes [eds.], Proceedings of the World Lagomorph Conference. University of Guelph, Guelph, Canada. BULL, E. L., and J. R. DUNCAN. 1993. Great Gray Owl (Strix nebulosa), p. 1-16. In A. Poole and F. Gill [eds.]. The Birds of North America, No. 41. The American Ornithologists' Union, Washington DC. CARPENTER, F. L. 1987. Food abundance and territoriality: to defend or not to defend? Amer. Zool. 27:387-399. CAUGHLEY, G. 1977. Analysis of Vertebrate Populations. Wiley, New York. CAUGHLEY, G., and C. J. KREBS. 1983. Are big mammals simply little mammals writ large? Oecologia 59:7-17. CESKA, V. 1980. Untersuchungen zu Nahrungsverbrauch, Nahrungsnutzung und Energiehaushalt bei Eulen. J. Om. 121:186-199. CHAPLIN, S. B., D. A. DIESEL, and J. A. KASPARIE. 1984. Body temperature regulation in red-tailed hawks and great horned owls: responses to air temperature and food deprivation. Condor 86:175-181. CHARLES, J. K. 1972. Territorial behaviour and the limitation of population size in crows, Corvus corone and C. comix. Ph.D. thesis. University of Aberdeen, Scotland. CHERNIN, E. 1952. The relapse phenomenon in L^MCocjtozoon infection of the domestic duck. Am. J. Hyg. 56:101-118. CHITTY, D. 1950. Canadian arctic wildlife enquiry, 1943-49: with a summary of results since 1933. J. Anim. Ecol. 19:180-193. CLOBERT, J., and J.-D. LEBRETON. 1991. Estimation of demographic parameters in bird populations, p. 75-104. In C. M. Perrins, J.-D. Lebreton and G. J. M. Hirons [eds.]. 150 Bird Population Studies: Relevance to Conservation and Management. Oxford University Press, Oxford. CLUTTON-BROCK, T. H. [ed.]. 1991. The evolution of parental care. Princeton University Press, Princeton, New Jersey. CORKUM, L. D., and D. C. CURRIE. 1987. Distributional patterns of immature Simuliidae (Diptera) in northwestern North America. Freshwater Biology 17:201-211. CRAIGHEAD, J. J., and F. C. CRAIGHEAD. 1956. Hawks, Owls and Wildlife. Stackpole Co., Pennsylvania. CROSSKEY, R. W. 1990. The natural history of blackflies. Wiley, New York. DAAN, S., C. DUKSTRA, and J. M. TINBERGEN. 1990a. Family planning in the Kestrel (Falco tinnunculus): The ultimate control of covariation of laying date and clutch size. Behaviour 114:83-116. DAAN, S., D. MASMAN, and A. GROENEVOLD. 1990b. Avian basal metabolic rates: their association with body composition and energy expenditure in nature. Am. J. Physiol. 259:333-340. DA VIES, N. B. 1978. Ecological questions about territorial behaviour, p. 317-350. In J. R. Krebs and N. B. Davies [eds.]. Behavioural Ecology, an Evolutionary Approach. Blackwell Scientific Publ., Oxford. DAVIES, N.B. 1980. The economics of territorial behaviour in birds. Ardea 68:63-74. DESSER, S. S., and G. F. BENNETT. 1993. The genera Leucocytozoon, Haemoproteus, and Hepatocystis, p. 273-305. In J. P. Kreier [ed.]. Parasitic Protozoa, Second Edition, Vol. 4. Academic Press, London. DESSER, S. S., A. M. FALLIS, and P. C. C. GARNHAM. 1967. Relapses in ducks chronically infected with Leucocytozoon simondi and Parahaemoproteus nettionis. Can. J. Zool. 46:281-285. DUKSTRA, C , L. VUURSTEEN, S. DAAN, and D. MASMAN. 1982. Clutch size and laying date in the Kestrel Falco tinnunculus: Effect of supplementary food. Ibis 124:210-213. DUKSTRA, C , A. BULT, S. BIJLSMA, S. DAAN, T. MEIJER, and M. ZULSTRA. 1990. Brood size manipulations in the kestrel (Falco tinnunculus): Effects on offspring and parent survival. J. Anim. Ecol. 59:269-285. DONAZAR, J. A., F. HIRALDO, M. DELIBES, and R. R. ESTRELLA. 1989. Comparative food habits of the eagle owl Bubo bubo and the great homed owl Bubo virginianus in six palearctic and nearctic biomes. Omis Scand. 20:298-306. DOYLE, F. I., and J. N. M. SMITH. 1994. Population responses of northern goshawks to the 10-year cycle in numbers of snowshoe hares. Condor, in press. 151 DRENT, R. H., and S. DAAN. 1980. The prudent parent: energetic adjustments in avian breeding. Ardea 68:225-252. DUNCAN, J. R. 1987. Movement strategies, mortality, and behavior of radio-marked great gray owls in southeastern Manitoba and northern Minnesota, p. 101-108. /« R. W. Nero, R. J. Clark, R. J. Knapton and R. H. Hamre [eds.]. Biology and conservation of northern forest owls: symposium proceedings. USDA Forest Service, Gen. Techn. Rep. RM-142, 309pp.,. DUNSTAN, T. C. 1970. Post-fledging activities of juvenile great homed owls as determined by radio-telemetry. Ph.D. thesis, Univ. of South Dakota, Vermillion SD. EARHART, C. M., and N. K. JOHNSON. 1970. Size dimorphism and food habits of North American owls. Condor 72:251-264. ELTON, C , and M. NICHOLSON. 1942. The ten-year cycle in numbers of the lynx in Canada. J. Anim. Ecol. 11:215-244. ERLINGE, S., G. GORANSSON, L. HANSSON, G. HOGSTEDT, O. LIBERG, I. N. NILSSON, T. NILSSON, T. VON SCHANTZ, and M. SYLVEN. 1983. Predation as a regulating factor in small rodent populations in southern Sweden. Oikos 40:36-52. FELSENSTEIN, J. 1985. Phylogenies and the comparative method. Am. Nat. 125:1-15. FERRER, M. 1992. Regulation of the period of postfledging dependence in the Spanish Imperial Eagle Aquila adalberti. Ibis 134:128-133. FITCH, H. S., F. SWENSON, and D. F. TILLOTSON. 1946. Behavior and food habits of the Red-tailed Hawk. Condor 48:205-237. FLIEG, G. M., and P. R. MEPPIEL. 1972. An account of trio nesting by yearling snowy owls in captivity. Raptor Research 6:103. FORSMAN, E. D., E. C. MESLOW, and H. M. WIGHT. 1984. Distribution and biology of the spotted owl in Oregon. Wildl. Monogr. 87:1-64. FRANKLIN, A. B. 1992. Population regulation in northern spotted owls: theoretical imphcations for management, p. 815-827. In D. McCuUough and R. Barrett [eds.]. Wildlife 2001: Populations. Elsevier, London. FUJII, K., C. S. HOLLING, and P. M. MACE. 1986. A simple generalized model of attack by predators and parasites. Ecol. Res. 1:141-156. GARD, N. W., and D. M. BIRD. 1990. Breeding behavior of American kestrels raising manipulated brood sizes in years of varying prey abundance. Wilson Bull. 10:605-614. GARD, N. W., and D. M. BIRD. 1992. Nesthng growth and fledging success in manipulated American kestrel broods. Can. J. Zool. 70:2421-2425. 152 GARNHAM, P. C. C. 1966. Malaria parasites and other haemosporidia. Blackwell Scientific Publications, Oxford. GASTON, A. J. 1992. The Ancient Murrelet. A Natural History in the Queen Charlotte Islands. Poyser, London. GESSAMAN, J. A. 1978. Body temperature and heart rate of the snowy owl. Condor 80:243-245. GESSAMAN, J. A., and K. A. NAGY. 1988. Transmitter loads affect the flight speed and metabolism of homing pigeons. Condor 90:662-668. GILLESPIE, J. H. 1977. Natural selection for variances in offspring numbers: a new evolutionary principle. Am. Nat. 111:1010-1014. GINZBURG, L. R., and D. E. TANEYHILL. 1994. Population cycles of forest Lepidoptera: a maternal effect hypothesis. J. Anim. Ecol. 63:79-92. GLUTZ VON BLOTZHEIM, U. N., and K. M. BAUER. 1980. Handbuch der Vogel Mitteleuropas, Vol. 9. Akademische Verlagsgesellschaft, Wiesbaden. GOULD, S. J., and R. C. LEWONTIN. 1979. The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptionist programme. Proc. R. Soc. Lond. B 205:581-598. GREEN, D. J., and R. C. YDENBERG. 1994. Energetics of male ospreys provisioning natural and manipulated broods. Ardea, in press. GREENWOOD, P. J. 1980. Mating systems, philopatry, and dispersal in birds and mammals. Anim. Behaviour 28:1140-1162. GREINER, E. C, G. F. BENNETT, E. M. WHITE, and R. F. COOMBS. 1975. Distribution of the avian hematozoa of North America. Can. J. Zool. 53:1762-1787. GUTIERREZ, R. J. 1989. Hematozoa from the Spotted Owl. J. Wildl. Dis. 25:614-618. GUTIERREZ, R. J., P. J. WARD, A. B. FRANKLIN, W. LAHAYE, and V. MERETSKY. 1985. Dispersal ecology of juvenile spotted owls (Strix occidentalis caurina) in northwestern California. Unpubl. final report, Cooperative Agreement PNW-82-226. Humboldt State Univ., Dept. of WildUfe, Areata CA. HAGEN, Y. 1952. Rovfuglene og viltpleien. Gildendal Norsk forlag, Oslo. HAGEN, Y. 1956. Norwegian studies on the reproduction of birds of prey and owls in relation to micro-rodent population fluctuations. Fauna 22:73-126. HAKKARAINEN, H. 1994. Reproductive effort, body size, and their fitness consequences in Tengmalm's owl (Aegolius funereus) under cyclic food conditions. Ph.D. thesis, Univ. of Turku, Turku. 153 HAKKARAINEN, H., and E. KORPIMAKI. 1991. Reversed sexual size dimorphism in Tengmalm's owl: is small size adaptive? Oikos 61:337-346. HAMILTON, W. D., and M. ZUK. 1982. Heritable true fitness and bright birds. Science 218:384-387. HANNON, S., and K. MARTIN. 1995. Mate fidelity and divorce in ptarmigan: polygyny avoidance on the tundra. In J. Black [ed.]. Ecology of Monogamy: Permanent Partnerships in Birds. Oxford Univ. Press, Oxford, in press. HANSKI, I., P. TURCHIN, E. KORPIMAKI, and H. HENTTONEN. 1993. Population oscillations of boreal rodents: regulation by mustelid predators leads to chaos. Nature 364:232-235. HAYWARD, G. D., and E. O. CARTON. 1984. Roost habitat selection by three small forest owls. Wilson Bull. 96:690-692. HENNY, C. J. 1972. An analysis of the population dynamics of selected avian species with special reference to changes during the modem pesticide era. USDI Wild. Res. Rep. 1, Washington DC, 99pp. HENNY, C. J., W. S. OVERTON, and H. M. WIGHT. 1970. Determining parameters for populations by using structural models. J. Wildl. Manage. 34:690-703. HIK, D. S. 1994. Predation risk and the 10-year snowshoe hare cycle. Ph.D. thesis, Univ. of British Columbia, Vancouver. HIRONS, G. J. M. 1985. The effects of territorial behaviour on the stability and dispersion of Tawny Owl ('i'^ njc a/wcoj populations. J. Zool. Lond. (B) 1:21-48. HIRONS, G., A. HARDY, and P. STANLEY. 1979. Starvation in young tawny owls. Bird Study 26:59-63. HOLLING, C. S. 1959. Some characteristics of simple types of predation and parasitism. Can. Entomol. 91:385-398. HOUSTON, C. S. 1987. Nearly synchronous cycles of the great homed owl and snowshoe hare in Saskatchewan, p. 56-58. In R. W. Nero, R. J. Clark, R. J. Knapton and R. H. Hamre [eds.]. Biology and conservation of northern forest owls: Symposium proceedings. USDA Forest Service, Gen. Techn. Rep. Rm-142, 309pp. HOUSTON, C. S., and C. M. FRANCIS. 1995. Survival of great horned owls in relation to the snowshoe hare cycle. Auk, in press. HUNTER, D. B., and C. ROHNER. 1994. Blood-sucking black flies (Simuliidae) and blood parasites (Leucocytozoon sp.) as causes of mortality in fledgling great homed owls. MS, in prep. HUNTER, D. B., E. DZUS, and K. L. MACHIN. 1994. Hematology of nestling Bald Eagles from Nemeiben Lake, Saskatchewan. Condor, in press. 154 HUNTER, F. F. 1990. Ecological, morphological, and behavioural correlates to cytospecies in the Simulium venustum/verecundum complex (Diptera: Simulidae). Ph.D. thesis. Queen's Univ., Kingston, Ontario. IMS, R. A., and H. STEEN. 1990. Geographical synchrony in microtine population cycles: a theoretical evaluation of the role of nomadic avian predators. Oikos 57:381-387. JANES, S. W. 1985. Habitat selection in raptorial birds, p. 159-188. In M. L. Cody [ed.]. Habitat Selection in Birds. Academic Press, London. JEDRZEJEWSKA, B., and W. JEDRZEJEWSKI. 1988. Seasonal surplus killing as hunting strategy of the weasel Mustela nivalis - test of a hypothesis. Acta Theriol. 34:347-359. JOHNSGARD, P. A. 1988. North American owls. Smithsonian Institution Press, Washington DC. KASPARIE, J. A. 1983. Some physiological and behavioral responses of the great homed owl (Bubo virginianus) to winter conditions. MSc. thesis, Univ. of Missouri-Columbia, Columbia. KEITH, L. B. 1963. Wildlife's ten year cycle. Univ. Wisconsin Press, Madison. KEITH, L. B., J. R. GARY, O. J. RONGSTAD, and M. C. BRTTTINGHAM. 1984. Demography and ecology of a dechning snowshoe hare population. Wildl. Monogr. 90:1-43. KEITH, L. B., and D. H. RUSCH. 1989. Predation's role in the cyclic fluctuations ruffed grouse. Proc. Int. Om. Congr. 19:699-732. KEITH, L. B., A. W. TODD, C. J. BRAND, R. S. ADAMCIK, and D. H. RUSCH. 1977. An analysis of predation during a cyclic fluctuation of snowshoe hares. Proc. Int. Congr. Game Biol. 13:151-175. KENWARD, R. E. 1985. Raptor radio-tracking and telemetry. ICBP techn. Publ. 5:409-420. KERLINGER, P., and M. R. LEIN. 1988. Population ecology of snowy owls during winter on the Great Plains of North America. Condor 90:866-874. KORPIMAKI, E. 1986. Gradients in population flucutations of Tengmalm's owls Aegolius funereus in Europe. Oecologia 69:195-201. KORPIMAKI, E. 1988a. Costs of reproduction and success of manipulated broods under varying food conditions in Tengmalm's owl. J. Anim. Ecol. 57:1027-1039. KORPIMAKI, E. 1988b. Effects of age on breeding performance of Tengmalm's owl Aegolius funereus in western Finland. Omis Scand. 19:21-26. 155 KORPIMAKI, E. 1989. Breeding performance of Tengmalm's Owl Aegolius funereus: effects of supplementary feeding in a peak vole year. Ibis 131:51-56. KORPIMAKI, E. 1992a. Diet composition, prey choice, and breeding success of Long-Eared Owls: effects of multiannual fluctuations in food abundance. Can. J. Zool. 70:2373-2381. KORPIMAKI, E. 1992b. Population dynamics of Fennoscandian owls in relation to wintering conditions and between-year fluctuations of food, p. 1-10. /n C. A. Galbraith, I. R. Taylor and S. Percival [eds.]. The ecology and conservation of European owls. Joint Nature Conservation Committee, Peterborough UK. KORPIMAKI, E. 1993. Regulation of multiannual vole cycles by density-dependent avian and mammalian predation? Oikos 66:359-363. KORPIMAKI, E. 1994. Rapid or delayed tracking of multi-annual vole cycles by avian predators? J. Anim. Ecol. 63:619-628. KORPIMAKI, E., and H. HAKKARAINEN. 1991. Fluctuating food supply affects the clutch size of Tengmalm's owl independent of laying date. Oecologia 85:543-552. KORPIMAKI, E., and K. NORRDAHL. 1989. Predation of Tengmalm's owls: numerical responses, functional responses and dampening impact on population fluctuations of microtines. Oikos 54:154-164. KORPIMAKI, E., and K. NORRDAHL. 1991. Numerical and functional response of Kestrels, Short-eared Owls, and Long-eared Owls to vole densities. Ecology 72:814-826. KORPIMAKI, E., K. HUHTALA, and S. SULKAVA. 1990. Does the year-to-year variation in the diet of eagle and Ural owls support the alternative prey hypothesis? Oikos 58:47-54. KORPIMAKI, E., H. HAKKARAINEN, and G. F. BENNETT. 1993. Blood parasites and reproductive success of Tengmalm's owls: detrimental effects on females but not on males. Funct. Ecol. 7:420-426. KREBS, C. J. 1991. The experimental paradigm and long-term population studies. Ibis 133 suppl. 1:3-8. KREBS, C.J. 1994. Ecology: The Experimental Analysis of Distribution and Abundance, 4th Ed. Harper Collins, New York. KREBS, C. J., B. S. GILBERT, S. BOUTIN, A. R. E. SINCLAIR, and J. N. M. SMITH. 1986. Population biology of snowshoe hares. I. Demography of food supplemented populations in the southern Yukon, 1976-84. J. Anim. Ecol. 55:963-982. KREBS, C. J., R. BOONSTRA, S. BOUTIN, M. DALE, S. HANNON, K. MARTIN, A. R. E. SINCLAIR, J. N. M. SMITH, and R. TURKINGTON. 1992. What drives the Snowshoe Hare cycle in Canada's Yukon?, p. 886-896. In D. McCuUough and R. Barrett [eds.], Wildhfe 2001: Populations. Elsevier, London, UK. 156 LACK, D. 1946. Competition for food by birds of prey. J. Anim. Ecol. 15:123-129. LACK, D. 1947. The significance of clutch size. L Intraspecific variation. Ibis 89:302-352. LACK, D. 1954. The natural regulation of animal numbers. Oxford Univ. Press, London. LAHAYE, W. S., R. J. GUTIRREZ, and D. R. CALL. 1992. Demography of an insular population of spotted owls, p. 803-814. In D. McCuUough and R. Barrett [eds.]. Wildlife 2001: Populations. Elsevier, London. LAKHANI, K. H., and I. NEWTON. 1983. Estimating age-specific bird survival rates from ring recoveries - can it be done? J. Anim. Ecol. 52:83-91. LANDE, R. 1988. Demographic models of the northern spotted owl (Strix occidentalis caurina). Oecologia 75:601 -607. LA YMON, S. A., and R. H. BARRETT. 1985. Analysis of habitat preference and use by dispersing, juvenile spotted owls. Unpubl. final report, PSW-82-0073CA, Univ. of California, Berkeley CA. LEE, E. T. 1980. Statistical methods for survival data analysis. Lifetime Learning Publ., Belmont, California. LIGON, J. D. 1969. Some aspects in temperature relations in small owls. Auk 86:458-472. LIMA, S. L., and L. M. DILL. 1990. Behavioral decisions made under the risk of predation: a review and prospectus. Can. J. Zool. 68:619-640. LINDEN, M., and A. P. M0LLER. 1990. Cost of reproduction and covariation of life history traits in birds. TREE 5:367-371. LINDSTROM, E., P. ANGELSTAM, P. WIDEN, and H. ANDREN. 1987. Do predators synchronize vole and grouse fluctuations? - An experiment. Oikos 48:121-124. LINKOLA, P., and A. MYLLYMAKI. 1969. Der Einfluss der Kleinsaugerfluktuationen auf das Briiten einiger kleinsaugerfressenden Vogel im sudlichen Hame, Mittelfinnland 1952-1966. Omis Fenn. 46:45-78. LOTKA, A. J. 1925. Elements of Physical Biology. Reprinted in 1956 by Dover Publications, New York. LOYE, J. E., and M. ZUK [eds.]. 1991. Bird-Parasite Interactions. Ecology, Evolution, and Behaviour. Oxford University Press, Oxford. LUNDBERG, A. 1981. Population ecology of the Ural Owl Strix uralensis in central Sweden. Omis Scand. 12:111-119. 157 MACARTHUR, R. H., and E. O. WILSON. 1967. Theory of Island Biogeography. Princeton University Press, Princeton. MADER, W. J. 1982. Ecology and breeding habits of the savanna hawk in the Llanos of Venezuela. Condor 84:261-271. MARCSTROM, V., R. E. KENWARD, and E. ENGREN. 1988. The impact of predation on boreal tetraonids during vole cycles: an experimental study. J. Anim. Ecol. 57:859-872. MARTIN, G. 1986. Sensory capacities and the nocturnal habit in owls. Ibis 128:266-277. MARTIN, G. 1990. Birds by Night. Poyser, London. MARTIN, K. 1989. Pairing and adoption of offspring by replacement male willow ptarmigan: behavior, costs, and consequences. Anim. Behav. 37:569-578. MASMAN, D. 1986. The annual cycle of the kestrel Fa/co ftnnMncM/M5. Ph.D. thesis, Univ. of Groningen, Groningen. MASMAN, D., S. DAAN, and C. DIJKSTRA. 1988. Time allocation in the kestrel (Falco tinnunculus), and the principle of energy minimization. J. Anim. Ecol. 57:411-432. MASMAN, D., C. DIJKSTRA, S. DAAN, and A. BULT. 1989. Energetic limitation of avian parental effort: field experiments in the Kestrel (Falco tinnunculus). J. Evol. Biol. 2:435-456. MATTHYSEN, E. 1989. Territorial and non-territorial settling in juvenile Eurasian nuthatches (Sitta europaea) in summer. Auk 106:560-567. MAYNARD SMITH, J. 1968. Mathematical Ideas in Biology. Cambridge University Press, New York. MCCLEERY, R. H., and C. M. PERRINS. 1985. Territory size, reproductive success and population dynamics in the great tit Parus major, p. 353-374. In R. M. Sibley and R. H. Smith [eds.]. Behavioural Ecology: Ecological Consequences of Adaptive Behaviour. Blackwell Scientific Publications, Oxford. MCINVAILLE, W. B., and L. B. KEITH. 1974. Predator-prey relations and breeding biology of the great homed owl and red-tailed hawk in central Alberta. Can. Field-Nat. 88:1-20. MCNAMARA, J. M., and A. I. HOUSTON. 1990. State-dependent ideal free distributions. Evol. Ecology 4:298-311. MIKKOLA, H. 1981. Der Bartkmz Strix nebulosa. Neue Brehm Bucherei 538, Wittenberg-Lutherstadt. 158 MIKKOLA,H. 1983. Owls of Europe. Buteo Books, Vermillion SD. MIKKOLA, H., and S. SULKAVA. 1969. On occurrence and feeding habits of Short-eared Owls in Finland 1964-68. Omis Fenn. 46:188-193. MILLER, G. S. 1989. Dispersal of juvenile Northern Spotted Owls in Western Oregon. MSc. thesis, Oregon State University, Corvallis OR. M0LLER, A. P. 1990. Parasites and sexual selection: current status of the Hamilton-Zuk hypothesis. J. Evol. Biol. 3:319-328. MURRAY, G. A. 1976. Geographic variation in the clutch sizes of seven owl species. Auk 93:602-613. MYERS, J. P., P. G. CONNORS, and F. PITELKA. 1979. Territory size in wintering sanderlings: the effects of prey abundance and intruder density. Auk 96:551-561. NAMS, V. O. 1990. Locate IL Pacer, Truro N.S., Canada. NEWTON, I. 1976. Population limitation in diumal raptors. Can. Field-Nat. 90:274-300. NEWTON,! 1979. Population ecology of raptors. Buteo Books, Vermillion SD. NEWTON, L 1986. The Sparrowhawk. Poyser, Calton U.K. NEWTON, L 1989a. Sparrowhawk, p. 279-296. In I. Newton [ed.]. Lifetime Reproduction in Birds. Academic Press, London. NEWTON, L 1989b. Synthesis, p. 441-469. In I. Newton [ed.]. Lifetime Reproduction in Birds. Academic Press, London. NEWTON,! 1991a. Habitat variation and population regulation in Sparrowhawks. Ibis 133 supp! 1:76-88. NEWTON, I. 1991b. Population limitation in birds of prey: a comparative approach, p. 3-21. In C. M. Perrins, J.-D. Lebreton and G. J. M. Hirons [eds.], Bird Population Studies: Relevance to Conservation and Management. Oxford University Press, Oxford. NEWTON, I. 1991c. The role of recruitment in population regulation. Proc. Int. Om. Congr. 20:1689-1699. NEWTON, I. 1992. Experiments on the limitation of bird numbers by territorial behaviour. Biol. Rev. 67:129-173. NEWTON, ! , and M. MARQUISS. 1981. Effect of additional food On laying dates and clutch sizes of sparrow hawks Accipiter nisus. Omis Scand. 12:224-229. NILSSON, J.-A., and H. G. SMITH. 1985. Early fledging mortality and the timing of juvenile dispersal in the marsh tit Parus palustris. Omis Scand. 16:293-298. 159 NIXON, W. 1993. Habitat selection of Barren Ground Caribou in the Arctic North Slope, Yukon. MSc Thesis, Univ. of Alaska, Fairbanks. O'DONOGHUE, M., E. HOFER, and F. I. DOYLE. 1994. When the going gets tough, the tough get eaten. Natural History, in press. OTIS, D. L., K. P. BURNHAM, G. C. WHITE, and D. R. ANDERSON. 1978. Statistical inference from capture data on closed animal populations. Wildl. Monogr. 62:1-135. PARMELEE, D. F., H. A. STEPHENS, and R. H. SCHMIDT. 1967. The birds of southeastern Victoria Island and adjacent small islands. Bull. Nat. Mus. Canada 222:1-229. PARTRIDGE, L., and P. H. HARVEY. 1988. The ecological context of life history evolution. Science 241:1449-1455. PATON, P. W. C , C. J. ZABEL, D. L. NEAL, G. N. STEGER, N. G. TILGHMAN, and B. R. NOON. 1991. Effects of radio tags on spotted owls. J. Wildl. Manage. 55:617-622. PATTERSON, I. J. 1980. Territorial behaviour and the limitation of population density. Ardea 68:53-62. PEIRCE, M. A. 1984. Weights of birds from Balmoral, Zambia. Bull. Brit. Orn. Union 104:84-85. PERRINS, C. M. 1991. Constraints on the demographic parameters of bird populations, p. 190-206. In C. M. Perrins, J.-D. Lebreton and G. J. M. Hirons [eds.]. Bird Population Studies: Relevance to Conservation and Management. Oxford University Press, Oxford. PETERSEN, L. R. 1979. Ecology of great homed owls and red-tailed hawks in Southeastern Wisconsin. Dept. Nat. Res. Wisconsin, Techn. Bull. 111, 63pp. PETERSEN, L. R., and J. R. KEIR. 1976. Tether platforms - an improved technique for raptor food habits study. Raptor Research 10:21-28. PETERSEN, L. R., and D. R. THOMPSON. 1977. Aging nestling raptors by 4th-primary measurements. J. Wildl. Manage. 41:587-590. PIETIAINEN, H. 1989. Seasonal and individual variation in the production of offspring in the Ural owl Strix uralensis. J. Anim. Ecol. 58:905-920. POLLOCK, K. H., S. R. WINTERSTEIN, C. M. BUNCK, and P. D. CURTIS. 1989. Survival analysis in telemetry studies: the staggered entry design. J. Wildl. Manage. 53:7-15. PORTENKO, L. A. 1972. Die Schnee-Eule. Neue Brehm-Bucherei 454, Wittenberg-Lutherstadt. PRICE, P. W. 1980. Biology of Parasites. Princeton University Press, Princeton, New Jersey. 160 PROMISLOW, D. E. L., and P. H. HARVEY. 1990. Living fast and dying young: a comparative analysis of life history variation among mammads. J. Zool. 220:417-437. PULLIAM, R. 1988. Sources, sinks and population regulation. Am. Nat. 132:652-661. REID, D. 1994. Us and them lemmings. Ph.D. thesis, University of British Columbia, Vancouver. ROHNER, C , and F. I. DOYLE. 1992a. Food-stressed Great Homed Owl kills adult Goshawk: Exceptional observation or community process? J. Raptor Res. 26:261-263. ROHNER, C , and F. I. DOYLE. 1992b. Methods of locating Great Homed Owl nests in the boreal forest. J. Raptor Res. 26:33-35. ROHNER, C , and C. J. KREBS. 1994. Differential predation on snowshoe hares by great homed owls. MS, in prep. ROYAMA, T. 1992. Analytical Population Dynamics. Chapman and Hall, London. RUSCH, D. H., E. C. MESLOW, L. B. KEITH, and P. D. DOERR. 1972. Response of great homed owl populations to changing prey densities. J. Wildl. Manage. 36:282-296. SALTZ, D., and G. C. WHITE. 1990. Comparison of different measures of telemetry error in simulated radio-telemetry locations. J. Wildl. Manage. 54:169-174. SAUROLA, P. 1989. Ural owl, p. 327-345. In I. Newton [ed.]. Lifetime reproduction in birds. Academic Press, London. SIBLEY, C. G., and J. E. AHLQUIST. 1990. Phylogeny and classification of birds: a study in molecular evolution. Yale Univ. Press, New Haven. SINCLAIR, A. R. E. 1989. Population regulation in animals, p. 197-241. In J.M. Cherrett [ed.]. Ecological Concepts. Blackwell Scientific Publications, Oxford. SINCLAIR, A. R. E., J. D. GOSLINE, G. HOLDSWORTH, C. J. KREBS, S. BOUTIN, J. N. M. SMITH, R. BOONSTRA, and M. R. T. DALE. 1993. Can the solar cycle and climate synchronize the snowshoe hare cycle in Czmada? Evidence from tree rings and ice cores. Am. Nat. 141:173-198. SINCLAIR, A. R. E., C. J. KREBS, J. N. M. SMITH, and S. BOUTIN. 1988. Population biology of snowshoe hares. III. Nutrition, plant secondary compounds and food limitation. J. Anim. Ecol. 57:787-806. SMITH, J. N. M., and P. ARCESE. 1989. How fit are floaters? Consequences of alternative territorial behaviors in a nonmigratory sparrow. Am. Nat. 133:830-845. SMITH, S. M. 1978. The "underworld" in a territorial sparrow: adaptive strategy for floaters. Am. Nat. 112:571-582. 161 SOLHEIM, R. 1984. Breeding biology of the pygmy owl Glaucidium passerinum in two biogeographical zones in southeastern Norway. Ann. Zool. Fennici 21:295-300. SOUTHERN, H. N. 1970. The natural control of a population of tawny owls (Strix aluco). J. Zool. 162:197-285. STEARNS, S.C. 1976. Life history tactics: A review of the ideas. Q. Rev. Biol. 51:3-47. STEARNS, S. C. 1992. The Evolution of Life Histories. Oxford University Press, Oxford. STENSETH,N. C , andR. A. IMS. 1993. TheBiology of Lemmings. Academic Press, London. STEPHENS, D. W., and J. R. KREBS. 1986. Foraging theory. Princeton University Press, Princeton New Jersey. STEWART, P. A. 1969. Movements, population fluctuations, and mortality among great homed owls. Wilson Bull. 81:155-161. SULLIVAN, K. A. 1989. Predation and starvation: age-specific mortality in juvenile juncos (Junco phaenotus). J. Anim. Ecol. 58:275-286. SUTCLIFFE, J. F. 1986. Black fly host location: a review. Can. J. Zool. 64:1041-1053. TATUM,LB. 1975. Egg volume. Auk 92:576-580. TAYLOR, R.T. 1984. Predation. Chapman and Hall, New York. TROSTEL, K., A. R. E. SINCLAIR, C. J. WALTERS, and C. J. KREBS. 1987. Can predation cause the 10-year hare cycle? Oecologia 74:185-192. TURNER, J. C , and L. MCCLANAHAN. 1981. Physiogenesis of endothermy and its relation to growth in the great homed owl. Bubo virginianus. Comp. Biochem. Physiol. 68A:167-173. VEKASY, M., J. MARZLUFF, M. MCFADZEN, B. LEHMAN, K. STEENHOF, and M. KOCHERT. 1994. Effects of radio-transmitters on productivity of pairie falcons and golden eagles. J. Raptor Res. 28:66. VILLAGE, A. 1983. The role of nest-site and territorial behaviour in limiting the breeding density of Kestrels. J. Anim. Ecol. 52:635-634. VOLTERRA, V. 1926. Fluctuations in the abundance of species considered mathematically. Nature 118:558-560. VOOUS, K. H. 1988. Owls of the northem hemisphere. Collins Sons and Co., London, U.K. 162 WALLIN, K. 1988. Life history evolution and ecology in the tawny owl (Strix aluco). Ph.D. thesis, Univ. of Goteborg, Goteborg. WALTERS, C. J. 1986. Adaptive management of renewable resources. Macmillan, New York. WATSON, A. 1957. The behaviour, breeding, and food-ecology of the Snowy Owl Nyctea scandiaca. Ibis 99:419-462. WATSON, A. 1985. Social class, socially-induced loss, recruitment and breeding of Red Grouse. Oecologia 67:493-498. WATSON, A., and R. MOSS. 1970. Dominance, spacing behaviour and aggression in relation to population limitation in vertebrates, p. 167-218. In A. Watson [ed.]. Animal populations in relation to their food resources. Blackwell Scientific Publications, Oxford UK. WELLER, M. H. 1965. Bursa regression, gonad cycle and molt of the great-homed owl. Bird Banding 36:102-112. WELLICOME, T. I. 1993. Reproductive performance of burrowing owls (Athene cunicularia): Effects of supplementary food. J. Raptor Res. 27:94-95. WIEBE, K. L., and G. R. BORTOLOTTI. 1994. Food supply and hatching spans of birds: energy constraints or facultative manipulation? Ecology 75:813-823. WEENS, J. A. 1977. On competition and variable environments. Am. Sci. 65:590-597. WIJNANDTS, H. 1984. Ecological energetics of the long-eared owl (Asiootus). Ardea 72:1-92. WILCOVE, D. S., and J. W. TERBORGH. 1984. Patterns of population decline in birds. Am. Birds 38:10-13. WYLLIE, I., and I. NEWTON. 1991. Demography of an increasing population of Sparrowhawks. J. Anim. Ecol. 60:749-766. YDENBERG, R. C. 1987. Nomadic predators and geographical synchrony in microtine population cycles. Oikos 50:270-272. YDENBERG, R. C. 1994. Behavioral ecology of provisioning in birds. Ecoscience 1:1-14. YDENBERG, R. C , and L. S. FORBES. 1991. The survival-reproduction selection equilibrium and reversed size dimorphism in raptors. Oikos 60:115-119. 163 

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.831.1-0088225/manifest

Comment

Related Items