Interactions of Humans and Bald Eagles on the Columbia River Estuary Author(S): Kevin Mcgarigal, Robert G

Home , Columbia River Estuary

Interactions of Humans and Bald Eagles on the Columbia River Estuary Author(s): Kevin McGarigal, Robert G. Anthony, Frank B. Isaacs Source: Wildlife Monographs, No. 115, Interactions of Humans and Bald Eagles on the Columbia River Estuary (Apr., 1991), pp. 3-47 Published by: Allen Press Stable URL: http://www.jstor.org/stable/3830569 Accessed: 29/08/2008 08:26

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=acg.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

http://www.jstor.org WILDLIFEMONOGRAPHS (ISSN:0084-0173)

A Publicationof The WildlifeSociety

INTERACTIONSOF HUMANSAND BALDEAGLES ON THE COLUMBIARIVER ESTUARY

by KEVINMcGARIGAL, ROBERT G. ANTHONY, ANDFRANK B. ISAACS

NO. 115 APRIL 1991 FRONTISPIECE.The ColumbiaRiver estuary (Photo by KevinMcGarigal), an adult bald eagle (Photo by FrankB. Isaacs), and boatingactivities on the ColumbiaRiver estuary (Photo by KevinMcGarigal). INTERACTIONSOF HUMANSAND BALD EAGLES ON THE COLUMBIARIVER ESTUARY

KEVINMcGARIGAL1 Oregon Cooperative Wildlife Research Unit, Oregon State University, Corvallis, OR 97331 ROBERTG. ANTHONY U. S. Fish and WildlifeService, Oregon Cooperative WildlifeResearch Unit, Oregon State University, Corvallis, OR 97331 FRANKB. ISAACS Oregon Cooperative Wildlife Research Unit, Oregon State University, Corvallis, OR 97331

Abstract: Human activities have had profound effects on bald eagle (Haliaeetus leucocephalus) populations. Few experimental studies exist that address the effects of human activities on foraging eagles during the breeding season. Information from these types of studies is needed so that resource managers can allow humans and eagles to coexist. We investigated the response of breeding bald eagles to human activities in foraging areas on the Columbia River estuary, Washington-Oregon, during spring and summer, 1985 and 1986. We distinguished between 2 forms of interaction. In the first, a moving human approached a stationary eagle and induced a disturbance. This type of interaction was rare on the Columbia River estuary and accounted for a minor proportion of an eagle's time-energy budget. Only 20% of all moving human activities observed during this study resulted in human-eagle encounters within 500 m, and <6% of all encounters resulted in a visible disturbance to an eagle. In the second form of interaction, an eagle had a choice of alternative foraging sites, some of which may have had human activities occurring nearby. In this situation, the eagle had the freedom to choose a destination, given the pattern of human activities. This type of interaction represented the major form of human-eagle interaction on the Columbia River estuary. To investigate this, we studied 6 pairs of eagles in each of 2 years; each pair was sampled 3 times during the breeding season, corresponding to incubation, nestling, and postfledging stages of the nesting cycle. Each sample consisted of a 3-day control period, during which we monitored "normal" eagle activity patterns, and a 3-day influence period, during which we "disturbed" (i.e., stationary boat with observer) a high-use foraging area. We compared eagle activity patterns within 1,200 m of the experimental disturbance between sampling periods. Eagles typically avoided an area within 400 m of the experimental stationary boat, although avoidance areas ranged between 200-900 m among pairs. In most cases, eagles spent less time and made fewer foraging attempts in the sample area during the influence period. Responses were consistent among nesting stages, although foraging activity increased dramatically and was more concentrated in the high-use areas during the later nesting stages. Although we cannot decisively conclude that human activities were directly responsible for current eagle spatial use patterns, the results of the experimental disturbances confirmed that boating activities have the potential to significantly affect eagle spatial use patterns. Based on these results, we developed a model of human-eagle interactions in foraging areas. We used this model and the results of this study to develop several alternative management strategies involving temporal and spatial restrictions of human activity. We recommend buffer zones 400-800 m wide around high-use foraging areas of bald eagles as the most appropriate management strategy for the Columbia River estuary. WILDL.MONOGR. 115, 1-47

Present address: Department of Forest Science, Oregon State University, Corvallis, OR 97331. 6 WILDLIFE MONOGRAPHS

CONTENTS

INTRODUCTION 6 Response to Stationary Human Activities ...... 24 Acknowledgments 7 Experimental Disturbance: General Effects 24 STUDY AREA 7 Experimental Disturbance: Variation Physical Description 7 Among Pairs 26 Human Population 10 Experimental Disturbance: Variation Eagle Population 10 Among Nesting Stages 29 CONCEPTUAL MODEL AND OBJECTIVES ...... 11 Response to Moving Human Activities 30 METHODS 12 Flush Response Rate and Flush Distance ... 31 Sampling Design 12 Factors Influencing Responses 33 Field Methods 12 Response Behavior 37 Data Analyses 14 DUAL DISTURBANCE THRESHOLD MODEL ...... 37 Human and Eagle Activity Patterns ...... 14 Single Human-Eagle Encounter 38 Experimental Disturbance 15 Moving Human Activities 38 Moving Human Activities 16 Stationary Human Activities 39 Home Range Determination 17 Pair or Population Level 39 RESULTS AND DISCUSSION 18 Research and Management Implications ...... -.....-40 Human Activities 18 Moving Human Activities 40 Temporal Patterns 18 Stationary Human Activities 41 Nesting Stage 18 MANAGEMENTRECOMMENDATIONS 42 Day to Day 18 Spatial Restrictions 42 Time of Day 20 Temporal Restrictions 43 Tide Level 20 Habitat Management 44 Spatial Patterns 20 FUTURE RESEARCH 44 Among Territories 20 LITERATURECITED 44 Within Territories 22

INTRODUCTION al. 1970, Mulhern et al. 1970). These fac- Human activities have had profound ef- tors have continued to impact some pop- fects on wildlife populations. In extreme ulations despite widespread efforts in law cases, species have been exterminated by enforcement and education (Wiemeyer et direct persecution or habitat alteration. al. 1984). Pesticides are unique because More commonly, the effects of human ac- they affect natality rates more often than tivities have been less dramatic and more mortality rates, although the end result is localized. The bald eagle is a good example similar-a reduction in eagle numbers. of a species with many local populations Human activities indirectly affecting that have responded to a variety of human bald eagles present resource managers with activities in a variety of ways. Bald eagles a problem of even greater magnitude and occur over a large area, use a variety of long-term significance. Activities of this habitats (e.g., rivers, lakes, estuaries, etc.), kind fall into 2 general categories: (1) those and are associated with varying degrees of resulting in permanent disruption of the human development (e.g., urban, rural, eagle's environment, and (2) those result- wilderness, etc.). Consequently, local pop- ing in temporary disruption of the eagle's ulations have undergone changes ranging environment. from extermination to increase. Early concerns about permanent dis- Human activities causing direct eagle ruption of bald eagle habitat focused pri- mortality (e.g., electrocution, shooting, marily on the cutting of nest trees (Harlow pesticides, poisoning) are commonly cited 1918, Howell 1941, Broley 1947, Corr 1974, or implied as causes of historicaleagle pop- Weekes 1974). Consequently, most man- ulation declines (Harlow 1918, Hornaday agement efforts to date have centered on 1920, Van Name 1921, Bailey and Wright protection of nest sites. Recent studies have 1931, Howell 1937, Broley 1947, Coon et focused on the destruction of perches and BALD EAGLESAND HUMANS ON THE COLUMBIARIVER-McGarigal et al. 7 winter roosts and the degradation of for- However, most of these studies have po- aging sites and have demonstrated the im- tential biases that may have influenced portance of protecting eagle habitats other their conclusions, because human-eagle than nest sites from permanent distur- interactions occurring away from nest sites bance (Stalmaster 1976, Stalmaster and were not considered. Despite increasing Newman 1979, Hansen et al. 1980, Shapiro awareness among biologists of the impor- et al. 1982, Isaacs and Anthony 1987). Re- tance of foraging areas to eagle survival source managers can use this information and productivity, there is a general lack to identify and protect critical habitats of consideration for the management of from permanent disturbance. these areas for breeding bald eagles. Human activities resulting in temporary In this paper, we investigate human and disruptionof the eagle's environment (e.g., eagle activity patterns and examine how most recreational activities) represent a breeding bald eagles respond to temporary major source of potential disturbance in human activities in foraging areas. Based many eagle populations. Recreational ac- on the results, we develop a model of hu- tivities have increased dramatically in man-eagle interactions that is used to rec- North America during the past few de- ommend both spatial (i.e., buffer zones) cades (U.S. Department of Interior 1982). and temporal (i.e., critical periods) restric- Certain nonconsumptive activities, such as tions of human activities around eagle high- boating, have increased even more dra- use areas during the breeding season. matically (Brockman and Merriam 1973, Acknowledgments.-We thank Geof- Jensen 1973). The unpredictable and tem- frey Dorsey (Corps of Engineers) for his porary nature of these activities makes enthusiastic encouragement and many them extremely difficult to monitor and useful suggestions. We also thank Jim Wat- assess;they present resourcemanagers with son and Monte Garrett for their coopera- a formidable regulatory challenge. Al- tive efforts in the field; they provided in- though such temporary human activities valuable data and site-specific eagle are reportedly very disturbing to winter- information. Special thanks are due Nancy ing and migrating eagles (Stalmasterand McGarigal for many hours of field assis- Newman 1978; Russell 1980; Skagen 1980; tance. We are grateful to Jim Watson, Su- H. Walter and K. L. Garrett, The effect san Skagen, and Monte Garrett for review- of human activity on wintering bald eagles ing an earlier draft of this manuscript and in the Big Bear Valley, California, Unpubl. Gretchen Bracher for assistance in pre- rep. to U.S. For. Serv., 89 pp., 1981; Knight paring figures. This study was conducted and Knight 1984; Smith 1988), virtually under the auspices of the Oregon Coop- nothing is known about how these activi- erative Wildlife Research Unit: Oregon ties, especially those occurring away from Department of Fish and Wildlife, Oregon nest trees, affect breeding bald eagles. State University, U.S. Fish and Wildlife Efforts to manage human activities at Service, and the Wildlife Management In- nest sites rather than at other critical areas stitute cooperating. This is Oregon State within eagle territories stem from the be- University Agricultural Experiment Sta- lief that nest disturbances are the major tion Technical Paper 8911. cause of site desertion and nesting failure. Several studies have associated lowered STUDY AREA productivity and site desertion with nest Physical Description disturbances(Murphy 1965, Retfalvi 1965, Juenemann 1973, Weekes 1974, Grubb Our study was conducted on the Colum- 1976, Anthony and Isaacs 1989); others bia River estuary between Longview, have found little or no evidence of distur- Washington, and the Pacific Ocean, a dis- bance-caused nest failures (Mathisen1968, tance of approximately 98 kilometers (Fig. Grier 1969, Swenson 1975, McEwan and 1). The climate is Pacific Northwest Mar- Hirth 1979, Fraser 1981, Bangset al. 1982). itime and is characterized by mild, wet 00

Baker BE.s

River Cathamet Columbia ^ 'Iff-. L-"i " PACIFIC OCEAN zo o 0 (A Or)

FC FortColumbia YR Young'sRiver AP AldrichPoint RP Rocky Point MC MillCreek CL Clifton DR Deep River TW Twilight MA Maygar

Fig. 1. Study area on the ColumbiaRiver estuary. Circlesrepresent locations of occupiedbald eagle nestingterritories; complete circles represent pairs sampled during this study. BALD EAGLES AND HUMANS ON THE COLUMBIA RIvER-McGarigal et al. 9

Fig. 2. ColumbiaRiver estuary in the vicinityof Astoria,Oregon. Photo taken by U.S. ArmyCorps of Engineers,Portland District. winters (Oct-Jun) and cool, dry summers tween high and low tides during summer. (Jul-Sep); mean annual precipitation and Consequently, at low tide much of the riv- temperature between 1951 and 1980 were er bottom is exposed as tidal mud flats. 176.7 cm and 10.3 C, respectively (records The river is approximately 14 km across at National Weather Service, Astoria, at its widest point near Astoria, Oregon; it Oreg.). becomes progressively narrower upriver. The lower river is characterized by its Numerous man-made (i.e., dredge spoil) expansiveness and the marked cyclic and natural islands create a network of changes in aquatic habitat resulting from channels and sloughs that, when combined pronounced tidal fluctuations(Fig. 2). Wa- with abundant intertidal marshes, create ter depth fluctuates as much as 3.6 m be- a rich and diverse habitat for bald eagles. 10 WILDLIFE MONOGRAPHS

Natural islands are irregularly inundat- tivity, most commercial activities are re- ed by tides and typically consist of a mo- stricted to the main river channel and a saic of shrub- and forest-dominated few side channels. Furthermore, the river- swamps; Sitka spruce (Picea sitchensis), based commerce proceeds at relatively black cottonwood (Populus trichocarpa), steady levels independent of day of the red alder (Alnus rubra), and willow (Salix week and weather. spp.) are major tree species. Man-made During spring and summer, recreational islands support varying vegetative growth, boating activities (mainly associated with ranging from none (i.e., dry sand) to her- fishing) greatly outnumber commercial baceous plants and/or young forests of cot- activities on the river. The influx of rec- tonwood, alder, and willow, depending on reationists increases dramatically as the ea- island age. Marshes occupy the fringe of gle breeding season progresses. In contrast most bays and inlets. to commercial activities, recreational ac- Mainland shores are characterized by tivities have a relatively unrestricted spa- extensive upland forests of young (<80 yrs tial use pattern. With the exception of sev- old), second- and third-growth conifers and eral favorite fishing areas, recreational use isolated patches of old-growth (>200 yrs is widely and irregularly dispersed old) forest. Second- and third-growth for- throughout the river. These activities oc- ests are dominated by Douglas-fir (Pseu- casionally encroach on even the most re- dotsuga menziesii), western hemlock mote areas of the river. Furthermore, lev- (Tsuga heterophylla), and red alder; old- els of these activities are subject to widely growth forests also contain Sitka spruce fluctuating temporal changes in response and western redcedar (Thuja plicata). to factors such as tide level, weather, and day of the week. Human Population Eagle Population Much of the Oregon shore, and to a much lesser extent the Washington shore, is de- Year-round, resident, adult bald eagles veloped. Shores in the vicinity of Young's occupied 23 territories (13 in Wash., 10 in Bay on the Oregon side and Baker Bay on Oreg.) on the Columbia River estuary (Fig. the Washington side (Fig. 1) are occupied 1). Between 1978 and 1990, breeding suc- by high-density industrial, commercial, cess and productivity of this population and residential development; the remain- were below state and regional averages and ing developed shore consists of scattered well below levels required for delisting by residential and agricultural development. the Pacific States Bald Eagle Recovery Plan A railroad runs immediately adjacent to (U.S. Fish and Wildlife Service 1986). the Oregon shoreline, and primary and Breeding success averaged 38.5% (range secondary paved roads parallel the Wash- 23-55%), and productivity averaged 0.56 ington shoreline throughout much of the young/occupied site (range 0.32-0.77) study area. during 1984-87 (see Postupalsky 1974 and The Columbia River estuary is a major Isaacs et al. 1983 for terminology and passageway for commercial vessels mov- methods, respectively). High levels of DDE ing to and from Portland, Vancouver, and PCB's in breeding eagles on the Co- Longview, and other cities located upriv- lumbia River estuary (Robert G. Anthony, er. Many side channels and sloughs are unpubl. data) suggest that contaminants used for storing log rafts, which are even- may be 1 cause of the reduced productiv- tually transported by tugboat to Astoria or ity. There is considerable evidence that Longview for milling or export. The river environmental contaminants have caused is periodically open to commercial fishing. reductions in the productivity of other bald During gill-net seasons, fishing boats oc- eagle populations (Stickel et al. 1966; cupy nearly all open water areas. Except Krantz et al. 1970; Postupalsky 1971; Wie- for gill netting and some log storage ac- meyer et al. 1972, 1984; Sprunt et al. 1973). BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 11

Nest sites generally were associatedwith surmisedthat the irregularspatial pattern standsof old-growthtrees and all nests were and widely fluctuatingtemporal pattern within 1.6 km of open water. Mean dis- of recreationalboating activities were po- tance between active nests was 7.1 km tentiallydisturbing to eagles. (range 2.1-22.1 km, SE = 0.7, n = 41). We noted that human-eagle interac- Foraging areas consisted of shorelines and tionsinvolving moving boats were rare and interchannel tidal mud flats. Diets were accountedfor a smallproportion of an ea- extremely variable among eagle pairs and gle's time budget.Stationary vessels (such varied on a seasonal basis; waterfowl con- as anchoredfishing boats), on the other sumption by some eagle pairs increased hand, disturbedthe eagles' environment during the winter period (Watson et al. for long periodsof time and, therefore, 1991). Overall, however, fish composed representeda greater potentialfor eagle 90% (based on frequency) of the eagle's disturbance.Interactions involving sta- year-round diet (Watson et al. 1991). tionaryboats were not easily assessed, how- The Columbia River estuary also is a ever, becausestationary activities did not major wintering area for nonresident ea- resultin visibleeagle disturbances,such as gles. Eagle numbers increased in Novem- flushingan eagle from a perch. Instead, ber and December, decreased in January, we hypothesizedthat stationaryboating and peaked in February or March. Major activitieswere resultingin changesin spa- population fluctuations coincided with tial and temporaluse patternsof eagles, southward and northward migrations. even though these changeswere not im- During our 3-year study, it was common mediatelyevident. to observe as many as 100 eagles on the Thesepreliminary observations formed estuary in 1 day. the basis for 2 conceptualmodels. These modelsestablished the frameworkfor our CONCEPTUALMODEL AND researchand are best understoodby de- OBJECTIVES scribing2 contrastingscenarios. In the first scenario(hereafter referred to as Active We formulated a conceptual model af- Displacement),the foragingarea consists ter observing human and eagle activity of a relativelynarrow river stretch where patterns and human-eagle interactions eagleshunt from shorelinetrees or gravel throughout the study area during April- bars and where boaterscome into close September 1984. Our observations were contactwith eagles. In this situation,hu- extensive but nonsystematic and subjec- mansactively approach or passby eagles, tive. In addition, we intensively monitored andthe eaglesreact (generally by flushing) 4 breeding pairs of eagles during July- if humanscome too close.Active Displace- September 1984 (see Home Range Deter- ment is typical of winteringpopulations mination) to detail their activity patterns. on riversin Washingtonwhere eagles are We rarely observed obvious human dis- quitemobile and rely heavilyon spawned- turbances of eagles, even though human outsalmon as a foodsource (Stalmaster and use of the study area was high. We never Newman 1978, Knightand Knight1984). observed nest disturbances and rarely ob- For this scenario,the researchapproach served human activities near active nests. has been to study eagle responseto ap- Most nest sites were relatively inaccessible proachingboats or pedestrians. to humans, which led us to suspect that In the contrastingscenario (hereafter re- nest disturbances probably did not repre- ferredto as PassiveDisplacement), the for- sent an important form of human-eagle agingarea consists of a largebody of water, interaction. On the other hand, we ob- suchas a largeriver, lake, estuary, or coast- served high levels of human activities in al area, where both humanand eagle ac- eagle foraging areas and noted that rec- tivitiesare not restrictedto a narrowcor- reational boating activities greatly out- ridor. In this situation,eagles hunt from numbered all other human activities. We shorelinetrees, pilings, or tidal mud flats, 12 WILDLIFE MONOGRAPHS and boats remain stationary most of the sponding to 3 major stages of the nesting time. Human activities influence the ea- cycle: incubation (23 Mar-13 May), nest- gle's environment and cause eagles to ling (10 May-13 Jul), and postfledging (3 change their foraging locations and be- Jul-27 Aug). Each sample consisted of 2 haviors. Passive Displacement is typical of sampling periods: (1) a control period dur- breeding populationswhere eagles are sed- ing which we monitored "normal" eagle entary on fixed home ranges. activity patterns for 3 days prior to ex- Our preliminary observationsindicated perimental disturbance, followed by (2) an that Passive Displacement represented the influence period during which we "dis- majorform of human-eagle interaction on turbed" a high-use foraging area for 3 days. the Columbia River estuary. We also no- The foraging area that received the most ticed that each pair of eagles concentrated eagle use during the control period and their foraging activity in only a few lo- also afforded the observer good visibility cations within their home range and that of the surrounding area was selected for high-use foraging areas usually were as- each experiment. The disturbance consist- sociated with shallow water that was rel- ed of the senior author in a stationary boat atively distant from boating activity cen- positioned in the center of a high-use for- ters. We hypothesized that, under existing aging area. Three different open, outboard conditions, stationary boating activities in motorboats (3.7-5.5 m) were used during high-use foraging areas represented the the study. The center of the high-use area most significant potential for human-eagle was defined subjectively as the geometric interaction on the Columbia River estuary. center of eagle use in a particular foraging To evaluate this, we developed the follow- area during the control period. We posi- ing study objectives: tioned the stationary boat as close to the 1. To describe the and geometric center as possible (always with- temporal spatial in 100 The location of the patterns of humans and breed- m). stationary activity boat was considered the center of the ing bald eagles in foraging areas. high- 2. To determine the relative abundance use area in all analyses. With a few ex- observation of stationary and ac- ceptions, daily periods were 8 moving boating hours at first and tivities in foraging areas. long, began light, in- 3. To determine how cluded morning low tides because this co- stationary boating incided with activity in high-use foraging areas in- peak eagle foraging activity. fluenced daily activity patterns of breeding bald eagles. Field Methods 4. To determine how and at what distance the control we observed eagles responded to moving boats and During period, from a motorboat or land-based ve- what factors influenced their response. eagles hicle, depending on accessibility. We tried to maintain a minimum distance of 500 m METHODS between the observer and eagles. During Sampling Design the influence period, we observed eagles from a motorboat positioned centrally in Sampling was designed to accomplish a high-use foraging area. Consequently, objective 3; data for other objectives were observations during the influence period collected coincidentally. We studied 6 pairs were restricted to a smaller area than dur- of breeding bald eagles each year during ing the control period. Early in the breed- spring and summer, 1985 and 1986; three ing season, when 1 eagle of a pair re- of the pairswere studied during both years. mained at the nest site while the other Pairs were widely distributed throughout foraged, we focused observations on the the study area (Fig. 1) to represent a range foraging eagle. Later in the breeding sea- of conditions. We sampled each pair 3 son, during late nestling and postfledging times during the breeding season, corre- stages, when both adults foraged away from BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 13 the nest site simultaneously, we attempted observation by general activity (e.g., to monitor both adults. As a result of our perching, directional flight, soaring flight, efforts to maintain continuous and detailed foraging attempt, etc.), specific activity observations on eagles throughout their (e.g., prey pursuit, hunting, resting, han- large home ranges, we were not able to dling prey, feeding self, etc.), habitat type document all human activities occurring (e.g., open water, mud flat, old-growth co- within the home ranges. Consequently, we nifer, etc.), perch substrate (e.g., tree spe- recorded complete observationson the for- cies, piling, driftwood, ground, etc.), tide aging areas and noted human activities level (1 = all tidal flatsexposed, water depth outside of these areas whenever possible. <0.3 m; 2 = tidal flats partially exposed, We plotted all human and eagle move- water depth 0.3-0.9 m; 3 = tidal flats ments on enlarged large-scale (1:12,000) slightly exposed, water depth 0.9-1.5 m; black and white photographsthat allowed and 4 = tidal flats covered, water depth us to identify individual trees, pilings, and > 1.5 m), and weatherconditions (i.e., cloud other fixed objects and record eagle perch cover, precipitation, and wind speed). We sites accurately to within a few meters. also recorded the time of day, time since Human and eagle locations in open water last known feeding (in hours), duration of were variously difficult to plot; all loca- activity (in seconds), distance from the tions, however, were accurate to within 20 center of the high-use foraging area, and m. Consequently, distance calculations height above water. Height above water were accurate to within 20 m. Human ac- was estimated visually in the field by com- tivities around permanently occupied ar- parisonto objectswith known height (mea- eas such as homes, highways, and marinas sured with a clinometer) and was accurate were not recorded for 2 reasons. First, hu- to within 20 m, depending on the birds man activities associated with these areas height above water. Distances were mea- were difficult to observe and were too sured to the nearest meter from plotted abundant and frequent to record accu- map locations using a digitizer. Based on rately. Second, these activities were reg- horizontal distance and height above ular in occurrence (often nearly continu- water, we converted all distance mea- ous) and, therefore, did not bias sures to straight-line distance for analyses. comparisons of eagle activity patterns be- We noted hunting tactic (direct pre- tween control and influence periods. We dation of live prey, scavenge, pirate), out- differentiated between research- and non- come (successful, unsuccessful), prey spe- research-related human activities; unless cies, and prey size for foraging attempts otherwise noted, research activities were whenever possible. Multiple dives at the not included in the analyses. same prey were considered 1 attempt. We We categorized human activities by type defined foraging success as the percent of (pedestrian, boat, automobile, aircraft, foraging attempts with known outcome re- train, industrial noise, gunshot, blasting, sulting in prey capture. chainsaw or logging noise), motion (sta- To assess human-eagle interactions, we tionary, moving), relative noise level (soft, noted each time a human activity occurred moderate, loud), and relative speed (slow, within 500 m of an eagle. We chose 500 moderate, fast), and recorded the time of m because we had previously never wit- day, duration of activity (in seconds), and nessed a human-induced eagle distur- distance from the center of the high-use bance at distances >500 m. A human ac- foraging area. Noise level and speed were tivity occurring within 500 m of an eagle categorized subjectively. Distances were was considered a single encounter; a single measured to the nearest meter from plot- human activity could be involved in more ted map locations using a digitizer. than 1 encounter. We defined "encounter We recorded time and activity budgets rate" as the percent of all human activities for eagles. Each time an eagle changed approaching within 500 m of an eagle, locations or behaviors, we categorized the where each individual human activity was 14 WILDLIFE MONOGRAPHS considered 1 observation regardless of the to all boating activity minutes, including number of eagles encountered. We de- both stationary and moving BAM's. Be- fined "disturbance" as any human pres- cause we could not effectively observe and ence or activity that caused a change in document all human activities, the data on the behaviorof nearbyeagles (Fraser1984). human activity patterns represent conser- We defined "disturbancerate" as the per- vative measuresof real human activity lev- cent of all encounters resulting in a visible els. eagle disturbance. Disturbances were fur- To describe eagle activity patterns, we ther separated by whether they caused an used eagle activity minutes (EAM's)with- agitation or flush response. We defined in 400 m of high-use foraging areas and "agitation response"as a visibly disturbed number of foraging attempts within 1,200 behavioral state that did not result in a m of high-use foraging areas as indices of flush. Keen attentiveness and frequent ag- foraging effort. We limited EAM'sto with- gressive vocalizations characterized this in 400 m to exclude nonforaging activities condition when directed at stationary hu- from the measure. Two nest sites were lo- man activities; flightiness (e.g., wing-flick- cated approximately 500 m from the for- ing, flight-ready posture) and rapid head aging areas under study, making it difficult movements were characteristic responses in those cases to differentiate foraging ac- to approachinghumans. We defined "flush tivity from other activities. We limited for- response"as any human-inducedflight that aging attempts to within 1,200 m because was preceded by agitated behavior. When this was the extent of our sampling area no agitation behavior was observed prior during both sampling periods (see Exper- to flight, human-induced flights were dif- imental Disturbance). Trends for total ea- ficult to distinguish from natural flights. gle activity time (total time) and foraging Therefore, we confined the analysesto ver- activity time (search-capture time, see Ex- ified flush responses. perimental Disturbance) were identical For each encounter, we recorded eagle because most eagle activity within 400 m response behavior (no response, agitation of high-use foraging areas was associated response, flush response), minimum dis- with the search and capture of prey. All tance between human and eagle, mini- measures were standardizedon a per hour mum distance between human and nest, of observation (hr obs.) basis to facilitate minimum distance between eagle and nest, comparisons. duration of behavioral response (in sec- To assessthe relationshipbetween boat- onds), and flight distance following a flush. ing activity and eagle foraging activity lev- Following flush responses, we noted els on a daily basis, we regressed 3 boating whether the eagle left the immediate area, activity parameters (total, moving, and flew to a nearby perch, or returned to the stationary BAM's) on 2 eagle foraging ac- original perch. Again, we measured all dis- tivity parameters(search-capture time and tances from plotted map locations using a number of foraging attempts) both within digitizer and converted them to straight- 400 and 1,200 m of high-use foraging areas line distance. in 12 separate simple linear regressions (n = 96 days). We included only control pe- Data Analyses riod days and used log or reciprocal trans- formationsof some variablesto better meet Human and Eagle Activity Patterns.- regression assumptions. To describe human activity patterns, we To assess the relationship between time report summary statistics in units of boat- of day and eagle foraging activity, we re- ing activity minutes (BAM's), unless oth- gressed hour on number of foraging at- erwise noted. A stationaryBAM represents tempts per hour of observationwithin 1,200 1 boat occupying a fixed location for 1 m of high-use foraging areas pooled across minute. A moving BAM represents 1 boat all pairs. The data were grouped on the in motion for 1 minute. Total BAM'srefers basis of 2 factors: (1) eagle pairs subjected BALD EAGLES AND HUMANS ON THE COLUMBIA RIvER-McGarigal et al. 15 to low levels of boating activity (<40 total BAM/hr obs. within home range; 4 pairs) versus pairs subjected to high levels of boating activity (>81 total BAM/hr obs. within home range; 5 pairs), including only control period foraging observations and (2) control period observations versus influence period observations. The division between pairs subjected to high versus low human activity was done after completion of the field work to reflect obvious differences among territories in human activity levels. To assess whether the relationship between time of day and eagle foraging differed between groups, we treated each grouping factor as an indicator variable in separate multiple regressions. For each grouping factor, this was analogous to test- ing whether the slopes of the regressions for each group were different. Experimental Disturbance.-To assess eagle response to the experimental stationary boat, we defined a sample area surrounding each high-use area as the visible area <1,200 m from the observer during Fig.3. Samplearea (bold line) and influencezones (concentric the influence period (Fig. 3). This bands)around 1 high-useforaging area forthe Deep Riverbald repre- eagle pairon the ColumbiaRiver estuary. sented the largest area that we could con- sistently and effectively observe during both sampling periods. In addition, we de- gaged in prey capture; it did not in- fined a series of 12 concentric, 100-m-wide clude flight times or time spent engaged influence zones around the experimental in feeding or other activities from a stationary boat (Fig. 3) to aid in comparing perch. The subjectivity involved in de- eagle activity between sampling periods; termining an eagle's behavioral status influence zones were defined arbitrarily (i.e., actively hunting or not) limited the within the data base and were not used to accuracy of this measure. delineate observations in the field. The fol- 3. Perch visits: Number of perch visits to lowing 5 variables were used to measure each influence zone. A single perch visit eagle activity and were standardized to an was defined as an entry into an influ- 8-hour day: ence zone associated with 1 or more fixed locations (i.e., perched eagle or 1. Total time: Duration of time spent in foraging attempt). Flights between each influence zone. This measure in- perches in the same zone that did not cluded all behavioral activities associ- enter the next closer zone were includ- ated with a fixed location (i.e., perched ed in the same visit. Flights entering eagle or foraging attempt); flight times the next closer zone terminated the visit between perch locations were not in- to that zone. Number of perch visits was cluded. more appropriate than number of perch 2. Search-capture time: Duration of time locations as an index to foraging area spent in foraging-related activities in use because the latter was subject to the each influence zone. This measure in- spurious effects of multiple, short perch- cluded time spent actively searching for to-perch movements during a single prey from a perch and time spent en- "visit" to a foraging area. 16 WILDLIFE MONOGRAPHS

4. Perch-flight visits: Number of perch that reflected the obvious changes in eagle visits plus flight-only visits to each in- use of influence zones between sampling fluence zone. A single flight-only visit periods apparent in graphical displays of was defined as an entry into an influ- the data. We defined "responsedistance" ence zone not associated with a fixed as the distance between the experimental location. Only the closest influence zone stationaryboat and the outside edge of the approached in a flight received a visit. closest influence zone used as expected or This measure was designed to assess more than expected during the influence sensitivity of eagle flight movements to period. the experimental stationary boat. Moving Human Activities.-We cate- 5. Foraging attempts: Number of forag- gorized all human-eagle encounters into ing attempts in each influence zone. This 100-m intervals, based on minimum dis- measure included successful and un- tance between human and eagle, and used successful attempts involving actual Pearson's chi-square statistic (2-dimen- strikes (for live prey), active pursuits sional models) to test for independence of (for pirate attempts), and scavenging response behavior (i.e., no response vs. attempts. Sorties not associated with an flush) and encounter distance (Fienberg actual strike and flights over carcasses 1982). Because of sample size limitations, were not included. Consequently, this we included encounters involving both re- measure underestimated activity asso- search and nonresearch human activities ciated directly with the capture of prey. and all eagles, including breeding adults, nonbreeding adults, and immatures. Based We compared eagle activity in the sam- on these results (i.e., the distance at which ple area between the control and influence response behavior changed significantly), periods. No statistical procedure was com- we reclassified encounters into 2 distance pletely effective in quantifying this rela- intervals (<200 m and >200 m) and used tionship, largely because some influence the maximum likelihood ratio statistic (G2; zones were not used by eagles and because 3-dimensional models) to examine the in- some data were not normally distributed. dividual effects of several factors (Table Consequently, we developed a nonstatis- 1) on response behavior while controlling tical procedure for quantitatively com- for encounter distance (Fienberg 1982). paring eagle activity between sampling For each factor, we established categories periods. Expected use of each influence that were biologically meaningful and zone was based on eagle activity during combined categories when necessary to the control period. A zone was used less ensure adequate expected cell values than expected if use during the influence (Fienberg 1982). The procedure involved period was <50% of control-period use. a hierarchical partitioning of the 3-dimen- Likewise, a zone was used more than ex- sional log-linear model into its individual- pected if use during the influence period effect and interactive-effect components. was >50% of control-period use. We chose The difference in G2 between the model 50% subjectively for 2 reasons. First, 50% of no 3-way interaction (includes all 3 assured us that "significant" differences in 2-way interactions) and the model without eagle use of influence zones between sam- the factor-response interaction (includes pling periods (i.e., used more or less than distance-response and distance-factor in- expected during the influence period) rep- teractions) is a statistical measure of the resented real and biologically meaningful effect of the factor on response behavior differences in eagle use patterns. Lower while controlling for encounter distance percentages allowed smaller differences in (Fienberg 1982). Interpretation of factors eagle use to become "significant," but did with significant effects was based on an so at the cost of lowered confidence that analysis of the standardized residuals for measured differences reflected real biolog- the model without the factor-response in- ical responses and not sampling error. Sec- teraction. For flush responses, we used ond, 50% provided a quantitative measure nonparametric procedures (Mann-Whit- BALD EAGLESAND HUMANSON THE COLUMBIARIVER-McGarigal et al. 17

Table 1. Factors evaluatedfor theireffect on bald eagle flush response rate and flush distance for human-eagleencounters involvingmoving human activities on the ColumbiaRiver estuary, 1985-86.

Factor Levels Additional notes Eagle characteristics Perch height <1 m 1-10 m >10 m Distance to nest <400 m 400-800 m > 800 m Activity on nest Included incubating, brooding, feeding young. foraging or feeding resting or other Included miscellaneous activities associated with resting (e.g., preening). Age adult Based on plumage characteristics. subadult Breeding status active Terminology defined by Postupalsky (1974). inactive or failure Residence status resident adult all others Hunger condition fed within 2 hrs Did not include eagles with unknown condition status. not fed Nesting stage incubation Adult on eggs. nestling Young in the nest. postfledging Young out of the nest. Human characteristics Approaching speed slow Based on relative speeds. moderate fast Noise intensity soft Based on relative noise. loud Other characteristics Time of day <0600 hrs 0600-0800 hrs 0800-1000 hrs -1000 hrs Precipitation none rainy Steady or intermittent, light, or heavy rain. Sky cover clear <25% cloud cover. cloudy 25-95% cloud cover. overcast or fog >95% cloud cover.

ney U test [MW] for factors with 2 groups, of observation. We scheduled observation Kruskal-Wallis test [KW] for factors with bouts to correspond with peak activity pe- >2 groups) to examine the individual ef- riods (i.e., early morning and late after- fects of several factors (Table 1) on flush noon). We observed eagles from a 5.8-m distance. outboard motorboat or land-based vehicle, depending on site accessibility. We plotted Home Range Determination locations of perches and foraging attempts on enlarged 1:12,000 aerial photographs We determined home ranges for 9 pairs and transposed locations onto 1:24,000 U.S. of eagles. Each pair was monitored during Geological Survey topographic maps to 8-hour observation bouts at least 4-6 times/ determine Universal Transverse Mercator month over a 1-year period during July coordinates. 1984-April 1986 for a total of 2,405 hours We analyzed the locational data using 18 WILDLIFE MONOGRAPHS

A FREQUENCY 2 or more movements (i.e., to and from the site). Almost all stationary boating activities were recreational in nature. Off- Pedestrian Moving 5% . Boat 67% road vehicles (e.g., beach combing for fire- Auto 84% wood) and pedestrians (e.g., beach walk- 5%- Stationary Other 6% 33% ing) accounted for 5% each by frequency and 2% and 7% by time, respectively, but % Total % Boat these activities were largely confined to a single site. These results confirmed our preliminary B TIME observations and supported the Passive model as the Moving Displacement conceptual Pedestrian 13% most appropriate basis for the study de- 7% Boat sign. We are not aware of other attempts Auto 2% 89% Stationary 87% to distinguish between the effects of sta- Other 2?A tionary and moving activities on bald ea- % Total % Boat gles. Investigations designed to measure disturbance responses usually have been Fig.4. Relativeoccurrence of majorhuman activity types and formulated on the basis of the Active Dis- movingversus stationaryboats by (A) frequency(n = 2,967) and (B) time (n = 138,645 min)on the ColumbiaRiver estuary, placement model and often without rec- 1985-86. ognition of the differences among alternative conceptual models. However, most of the previous work has been done on relatively small rivers where stationary ac- the MCPAAL software program (Stuwe tivities are relatively less common. and Blowhowiak 1986); locations for male and female of each were combined pair Patterns for analyses. We used the harmonic mean Temporal and to de- method (Dixon Chapman 1980) Nesting activity levels of home and Stage.-Human termine the sizes ranges high- varied with nesting stage and were greatest was defined as the use areas. Home range when eagle foraging activity was highest. area that included all activities of a pair Boating activity increased >6-fold from con- of resident eagles; the 95% utilization incubation (mid-Mar to mid-May) to post- tour was chosen as the most realistic esti- fledging (Jul-Aug); eagle foraging activity mator of home range size. High-use areas increased >4-fold for the same period (Fig. were defined as specific areas within home 5A). Increased eagle foraging activity re- ranges that received disproportionately flected the increasing nutritional needs of high eagle use; the 50% utilization contour growing young. Increased foraging effort was chosen as the estimator for high-use also may have reflected a decrease in fish areas. availability during the postfledging period (Watson et al. 1991). It also is possible that RESULTS AND DISCUSSION the seasonal increase in eagle foraging ef- Human Activities fort was the result of a human-induced decrease in foraging efficiency. Knight and Stationary boating activities represented Knight (1986) found that feeding efficien- the greatest potential source of disturbance cy declined as the possibility of human to breeding bald eagles on the Columbia encounter increased, largely as a result of River estuary, as they accounted for 87% more time spent scanning for intruders. of all boating activities by time and 33% Day to Day.-There was no strong ev- by frequency (Fig. 4). The high frequency idence that eagles modified their foraging of moving boats reflects the fact that each activity level in relation to daily fluctua- stationary boat was usually associated with tions in human activity (Fig. 5B). Eagle BALD EAGLESAND HUMANSON THE COLUMBIARIVER-McGarigal et al. 19

A -2.5 0 25 -2.0 > (cnI) -1.5 m 15, 0.6 3 0 -1.0 3 10. cr I 0.3 O C0 05 5 LU a- 0 t I L o to O 0 INCU- NEST- POST- w 1: BATION UNG FLEDGING aC NESTINGSTAGE w TIMEOF DAY 0. (hrs) 30 B -200 0 1.2 z ui 25- < 0.9 20- 15- -100 0.6 10 - -50-I 0.3 5TH 0 M T W TH F S S (low) 2 3 4 (high) DAYOF THEWEEK TIDE LEVEL Fig. 5. Temporalpatterns of adult,breeding bald eagles and boatingactivities by (A)nesting stage, (B) day of week, (C) time of day, and (D) tide level on the ColumbiaRiver estuary, 1985-86. Eagle activityfor nesting stage and day of week represent total eagle activityminutes within400 m of high-use foragingareas duringthe control period(n = 10,910 min)per hour of observation(EAM/hr obs.). Eagle activityfor time of day represents numberof foragingattempts within1,200 m of high-use foragingareas duringthe controlperiod (n = 144) per hour of observation.Eagle activityfor tide level represents numberof foragingattempts observed throughoutthe study area (n = 701) per hourof observation.Boating activity for nestingstage and day of week representstotal boatingactivity minutes observed throughoutthe study area duringcontrol and influenceperiods (n = 123,574 min) per hour of observation(BAM/hr obs.). Boating activityfor time of day represents frequencyof boating activitiesobserved throughoutthe study area duringcontrol and influenceperiods (n = 2,498 boats) per hourof observation. search-capture time and number of for- Boating activity increased >3-fold from aging attempts within 1,200 m of high-use weekdays to weekend, but eagle foraging foraging areas, and number of foraging activity was not strongly affected by this attempts within 400 m of foraging areas temporal factor (Fig. 5B). Eagle search- exhibited mild negative correlations with capture time within 400 m of high-use ar- moving boating activity minutes within eas was, however, slightly greater on week- 1,200 and 400 m of foraging areas, re- ends (weekday x = 77 EAM/day, SE = 13, spectively (r > -0.208, n = 96 days, P < n = 83 days; weekend x = 120 EAM/day, 0.042, 3 separate simple linear regres- SE = 26, n = 24 days; Mann-Whitney U sions). All other possible correlations be- test, MW = 1,245, P = 0.061). The eagle's tween eagle activity parameters and boat- persistent, or perhaps even increased, for- ing activity parameters within either 400 aging activity on the weekend may have or 1,200 m of high-use foraging areas were been the result of decreased foraging ef- inconsistent and insignificant (r < ?0.167, ficiency, their unwillingnessto reduce food n = 96 days, P > 0.108, 9 separate simple supplies to their young, or their unwill- linear regressions). ingness to abandon the nest territory for Similarly, there was little evidence that areas of lower human activity. We noted eagle foraging activity was affected by the that eagles that failed in their nesting at- weekend influx of humans on the estuary. tempt were quite mobile and often spent 20 WILDLIFE MONOGRAPHS considerable time outside their established subjected to relatively low boating activity "breeding" territories,suggesting that un- levels (difference in slopes, F = 5.78, 1 df, successful or nonbreeding pairs may be P = 0.027; Fig. 6A), indicating a possible more easily displaced by periodic influxes human-induced change in eagle temporal of human activities. activity patterns. In addition, eagles con- Bald eagles wintering on the Skagit Riv- centrated their foraging in the early morn- er in Washington were easily displaced by ing during the control period more than temporally high levels of human activity during the influence period (difference in (M. V. Stalmaster, Stalmaster and Associ- slopes, F = 13.93, 1 df, P = 0.002; Fig. ates, Milton, Wash., pers. commun.). He 6B). Knight et al. (In Press)noted a similar observed a significant decrease in the num- shift in foraging activity of wintering ea- ber of feeding eagles and the amount of gles on the Toutle River in Washington feeding activity on weekends in associa- from early morning to late afternoon in tion with increased boating activity. Like- response to anglers. The delay in foraging wise, Smith (1988) observed significantly activity during the influence period may fewer summering and migrating eagles on have been due to later low tides in addition Jordon Lake and Falls Lake in North Car- to the experimental disturbance. Regard- olina on weekends when human activity less of the reasonfor the observed temporal levels were higher than on weekdays. Un- patterns, it is apparent that early morning like wintering or postbreeding eagles, human activities are potentially the most breeding eagles on the Columbia River es- disruptive to eagle foraging activity on the tuary were either unaffected by the week- ColumbiaRiver estuary.Skagen (1980) also end influx of human activities or they in- concluded that human activities during the creased their foraging effort to compensate early morning hours were most disturbing for human-caused reductions in foraging to foraging eagles wintering on the Skagit efficiency. River. Time of Day. -Different daily activity Tide Level.-Eagle foraging patterns patterns of humans and eagles resulted in were strongly influenced by tidal fluctu- fewer potential interactions. Boating ac- ations, which altered the exposure of tidal tivity increased steadily from sunrise mud flats and shorelines. Eagles foraged (0430-0610 hrs) to 0900 hours and there- much more than expected at low tides and after remained relatively constant. Eagle less than expected at high tides (Fig. 5D) foraging activity, on the other hand, was (Watson et al. 1991). Other investigators highest from 0400 to 0600 hours and de- have suggested a similar relation between creased steadily thereafter (Fig. 5C). Sim- tide and foraging activity (Broley 1947, ilar trends in eagle foraging activity have Hancock 1964, Ofelt 1975, Todd 1979), been noted for other populations(Servheen but we are unaware of other attempts to 1975, Skagen 1980, Stalmaster 1980, Har- quantify this relationship. More fish car- mata 1984), although some populationsalso casses on the mud flats and less boating foraged extensively in the evening activity in the shallowsboth may have con- (Servheen 1975, Steenhof 1976). Early tributed to the increased foraging at low morning foraging is most likely related to tide. Most boating activity was restricted food stress, although it may be an effort to deep water during low tide. Unfortu- by eagles to minimize their interactionwith nately, we were unable to separate the ef- humans. Stalmaster (pers. commun.) par- fects of increased prey availability from tially attributes higher morning counts of the effects of reduced human activities on wintering bald eagles on the Skagit River tidal mud flats. to less boating activity early in the day. On the Columbia River estuary, eagle Spatial Patterns pairs subjected to relatively high boating activity levels concentrated their foraging Among Territories.-Boating activity in the early morning more so than pairs levels varied considerably among eagle BALD EAGLESAND HUMANSON THE COLUMBIARIVER-McGarigal et al. 21

-.0(- LOW BOAT ACTIVITY - r -0.943, P < 0.000 Y = 0.174 0.016X - HIGH BOAT ACTIVITY Y = 0.271 - 0.027X

O ccn o,- I I I I I I I O

I uW 0.5-

0.2-O r = -0.554, P = 0.0770

0 TIME OF DAY (hrs) <0500 0700 0900 1100 1300 >1400

TIME OF DAY (hrs) Fig. 6. Foragingattempts within1,200 m of high-useforaging areas of adultbreeding bald eagles per hourof observation(hr obs.) in relationto time of day for (A)eagle pairssubjected to high(>81) versus low (<40) boatingactivity minutes per hourof observationand (B) control periodversus influenceperiod observations on the ColumbiaRiver estuary, 1985-86. Foraging activityfor part A includescontrol period observations only. territories (range 0.4-296.5 BAM/hr obs.), areas. Few people used this portion of the largely depending on proximity to favored river for temporary activities, but there fishing areas. Four territories accounted were considerable habitat alterations and for 87% of the total BAM's recorded and permanent centers of human activity (i.e., were responsible for most of the temporal highway traffic, houses). In contrast, we changes in human activities discussed pre- recorded the highest human activity in a viously. We recorded the lowest human territory near Rocky Point, Washington activity in an atypical territory on lower (Fig. 1). The primary foraging area con- Young's River, a tributary of the Columbia sisted of an undeveloped inlet with exten- River (Fig. 1). The primary foraging area sive tidal mud flats near a major fishing adjoined the nest site and consisted of a area. Consequently, although there was narrow river channel that dissected exten- virtually no permanent disturbance of the sive agricultural (i.e., predominantly dairy area, there was considerable temporary pasture) and low-density, rural-residential boating activity. These sites illustrate the 22 WILDLIFE MONOGRAPHS

boating activity levels were much lower during this time of day, their relative distribution among territories remained nearly the same. Again, there was no strong relationship between eagle activity and boating activity, although there was a weak negative correlation between stationary boating activity and search-capture time within 400 m of high-use foraging areas (r = -0.567, n = 9 pairs, P = 0.112). These data suggest that within the observed range of boating activity levels, overall long-term (i.e., breeding season) eagle use of high-use foraging areas was largely independent of overall boating activity in those areas, although there was a noticeable trend for high-use areas with greater boating activity to be used less by eagles. Apparently, any human-induced changes in eagle foraging activity patterns were obscured at these coarse spatial and temporal scales. Our previous results indicate that eagles may be responding at Fig. 7. Breedingseason home range (HarmonicMean 95% more complex temporal scales, for exam- utilizationcontour) and high-useareas (HarmonicMean 50% ple, by concentrating foraging activity in utilizationcontours) for the Deep Riverbald eagle pairon the ColumbiaRiver estuary, 1985-86. the early morning and at low tide. Perhaps, such subtle temporal shifts in activity patterns in response to increasing human ac- disparity of conditions within the study tivity levels precedes any large-scale spa- area and the distinction between tempo- tial responses. Unfortunately, these data do rary human activities in foraging areas and not allow us to predict at what human other forms of permanent human activi- activity levels permanent displacement ties. from these areas might occur or whether We found little evidence that pairs with eagles already were being displaced to higher boating activity in their high-use these areas from other preferred foraging foraging areas were using these areas less areas. than pairs with lower boating activity. To- Within Territories.-Within territo- tal eagle activity time, search-capture time, ries, eagles and humans differed in their and number of foraging attempts were not spatial use of the habitat. Most eagle pairs correlated with moving, stationary, or total concentrated their activity in a few loca- boating activity minutes within either 400 tions (i.e., high-use areas), which included or 1,200 m of high-use foraging areas (r < the nest site and key foraging areas (Fig. ?0.415, n = 9 pairs, P > 0.267, 18 separate 7). High-use areas based on harmonic mean simple linear regressions). Because eagles 50% utilization contours composed an av- foraged most intensely in the early morn- erage of only 6% (1.3 km2) of the average ing, foraging activity may have been un- home range size (range 0.1-4.2 km2, SE = affected by human activities occurring lat- 0.47, n = 9 pairs). All eagle pairs dem- er in the day. To evaluate this, we tested onstrated this highly concentrated spatial for independence of eagle activity and use pattern. Within high-use foraging ar- boating activity within 400 and 1,200 m eas, eagles usually concentrated their ac- of high-use foraging areas during early tivity around 1 or more hunting perches morning hours only (<0900 hrs). Although that offered good visibility of the adjacent BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 23

50 - 100 -

B MOVINGBOATS 40 M EAGLEACTIVITY 80 - E STATIONARYBOATS * BOATACTIVITY - 30- z U- 60- 0 1r w 20- QI 40-

10 20-

I1 i;iima - ill.l-1 - - 0 -1 1 1 6-7 8 1 0-1 I 2-3 I 4-5 ! 6-7 I 8-9 I 10-11 I nf 1-2 3-4 5-6 7-8 9-10 11-12 <400 400-800 800-1,200 >1,200 DISTANCE FROM HUA (x100 m) DISTANCEFROM HUA (m) Fig. 8. Distributionof adult,breeding bald eagle foragingac- = n= and Fig.9. Distributionof boatingactivity (n 69,229 min)during tivity(search-capture time, 12,138 min) boatingactivity the control in relationto distance from the centers of (n = 15,807 min)during the controlperiod in relationto distance period fromthe centers of the areas on the the high-useforaging areas (HUA)of adult,breeding bald ea- high-useforaging (HUA) on the ColumbiaRiver 1985-86, and the relative ColumbiaRiver estuary, 1985-86. gles estuary, contributionof stationaryversus movingboats to this pattern. foraging area. Consequently, eagle foraging activity within high-use areas gen- fish carcasses, which are a large compo- erally decreased abruptly with increasing nent of the eagle's diet, are more abundant distance from the high-use center or fa- on exposed mud flats (Watson et al. 1991). vorite perch site (Fig. 8). It also is widely acknowledged that human Conversely, boating activities were less activities sometimes displace eagles to ar- common in high-use foraging areas than eas of lower human activity. In Minnesota elsewhere. Less than 12% of all boating and Maryland, nests are farther from per- activity minutes occurred within 800 m of manent human activity centers than ran- high-use foraging areas (Fig. 9). This pat- dom points (Andrew and Mosher 1982, tern was mostly governed by the distri- Fraser et al. 1985). Wintering eagles on bution of stationary boating activities. Less the Nooksack River in Washington dis- than 7% of all stationary BAM's occurred proportionately used areas of low human within 800 m of high-use foraging areas, activity (Stalmaster and Newman 1978). largely because most stationary boating ac- These authors and Servheen (1975) also tivities were associated with deep-water noted significantly higher eagle use of the areas, whereas foraging areas were asso- riverside with least human activity on the ciated with shallow water or exposed mud Nooksack and Skagit rivers, respectively. flats. In contrast, almost 50% of all moving Corr (1974) speculated that shoreline sec- boating activities occurred within 800 m tions near eagle nests in southeast Alaska of high-use foraging areas, largely because that were heavily used by commercial and boating channels were often close (200- recreational crafts received less eagle use. 800 m) to these foraging areas. Smith (1988) noted a spatial partitioning These data indicate that eagles and hu- between humansand eagles on JordanLake mans spatially partitioned the foraging and Falls Lake, North Carolina, and re- habitat. It is unclear, however, whether lated it to varying water depths. He sug- human activities displaced eagles to a few gested that shallow water created unde- concentrated, shallow-water foraging ar- sirable conditions for boating-related eas or whether eagles actually preferred recreationalactivities and also concentrat- these areas and the spatial partitioning of ed fish, perhaps increasing eagle foraging the habitat was coincidental. On the Co- success. These spatial use patternsare sim- lumbia River estuary, observations suggest ilar to the one we observed on the Colum- that fish are more available and perhaps bia River estuary. It is plausible, therefore, more vulnerable in shallow water and that that both foraging preferences and hu- 24 WILDLIFE MONOGRAPHS man-induced displacement operated si- remainder of this section, we analyze how multaneously to restrict spatial use pat- eagle activity patterns were modified in terns of foraging eagles on the Columbia response to the experimental, stationary River estuary. boating activity. Disturbance: General Response to Stationary Experimental Ef- fects.-Eagle activity in high-use foraging Human Activities areaswas modified substantiallyduring the Most eagles did not approach stationary influence period in response to the exper- human activities at close distances and were imental stationary boat. On the average, almost never visibly disturbed when they eagles avoided an area within 300-400 m did so. Boating activities accounted for 91% of the experimental stationary boat (Figs. of all stationary human activities by time 10-12). During the control period, total (n = 118,206 min). Of 822 individual sta- time spent in the foraging area decreased tionary boating activities, where each steadily from zero to 400 m (Fig. 10A). boating activity was considered a single This was largely due to a pattern of con- observation regardless of the number of centrated use around key foraging perch- eagles encountered, only 15% resulted in es. In at least 4 areas, however, eagle use an encounter (i.e., <500 m) with eagles, was more dispersed, and no single perch and only 0.12% resulted in a visible dis- dominated the spatial use pattern. Beyond turbance to an eagle. When each encoun- 400 m, eagle activity typically was largely ter within 500 m was considered separate- unaffected by the experimental distur- ly (i.e., each human activity could be bance. The large expenditure of time be- involved in several encounters with dif- tween 400-600 m was mainly a result of ferent eagles), including those involving nesting activity; nest trees were located research-related activities, we observed approximately 500 m from the center of disturbances in only 2.3% of the cases (n 2 foraging areas under study. Total time = 1,461) involving stationary boats. Based spent actively hunting clearly depicted an on these data, stationary boats did not ap- avoidance of the foraging area during the pear to be very disturbing to eagles. In this influence period (Fig. 10B). type of interaction (Passive Displacement Number of perch visits did not exhibit model), however, the eagles largely con- the strong bimodal pattern evident with trolled the encounters because they could total time and, to a lesser extent, search- modify their spatial use patterns in re- capture time (Fig. 11A). Further, the pat- sponse to the stationary boats. The ques- tern of rapidly decreasing use within 400 tion was then: Do eagles modify their spa- m was less dramatic with perch visits than tial use patterns in response to stationary both total time and search-capture time boats and thereby result in such low en- (Fig. 11A). This largely reflects the fact counter and disturbance rates? that eagles often perched in single loca- The limited number of disturbances in- tions for long periods. Also, eagle foraging volving stationary human activities that we effort often was very casual in nature; ea- observed were predominantly agitation re- gles frequently perched in foraging areas sponses and were almost always in re- for long periods without actively scanning sponse to the research boat during the in- for prey. Hence, perch visits was less influence period. In most cases eagles flew fluenced by temporal considerations and to the disturbed area, perched 100-200 m provided a clearer means of separating away, displayed agitation behaviors, and spatial use patterns from spatio-temporal subsequently flushed and left the area or use patterns than total time and search- moved to a perch at a "comfortable" dis- capture time. tance. We discuss the implications of these Eagle flight visits to the high-use for- behavioral responses from a theoretical aging areas were not modified to the same perspective in a subsequent section (see degree as perching activities during the Dual Disturbance Threshold Model). In the influence period. Eagle perch and flight- BALD EAGLESAND HUMANSON THE COLUMBIARIVER-McGarigal et al. 25

80 1.8- A PERCHVISITS

O] CONTROL 1.4- * INFLUENCE

1.0--

0.2-

0 c: a. U) 30- co 2-8 - B PERCH-FLIGHTVIS z 5 2.4- 25-

20-

15 -

10. -

DISTANCEFROM HUA (x100 m) DISTANCEFROM HUA (x100 m)

Fig. 10. Total eagle activity time (A) and search-capture time Fig. 11. Number of perch visits (A) and number of perch-flight = (B) during control (n = 102 days) and influence (n = 100 days) visits (B) during control (n = 102 days) and influence (n 100 periods in relation to distance from the centers of the high-use days) periods in relation to distance from the centers of the foraging areas (HUA) on the Columbia River estuary, 1985- high-use foraging areas (HUA) on the Columbia River estuary, 86, for all bald eagle pairs combined. 1985-86, for all bald eagle pairs combined. only visits were greatly reduced within number of foraging attempts within 400 200-300 m of the experimental stationary m of the disturbance was dramatically less boat (Fig. liB). Even within 200 m, the during the influence period (Fig. 12). perch-flight response was much less pro- Hence, foraging activity was greatly re- nounced compared to the other eagle ac- duced within 400 m of the disturbance, tivity measures. During the influence pe- although eagles did occasionally forage in riod, eagles often flew close to the observer the disturbed area, apparently when there as they moved between other undisturbed was higher probability of success. foraging areas. Also, eagles often flew di- Eagles did not simply redistribute their rectly past the observer and perched 300- use of the sample areas during the influ- 500 m away. After this "inspection," some ence period; that is, they did not increase eagles vacated the area, but other eagles their use of the outer portions of the sample remained at a "comfortable" distance and areas to compensate for their reduced use made foraging flights into the disturbed of the centers. In fact, eagle use beyond area. These forays sometimes resulted in 400 m remained virtually the same during approaches to within 100 m and foraging both sampling periods (Figs. 10-12). Ad- attempts often were successful. Foraging ditional observations outside the sample success within 400 m of the experimental areas indicated that eagles sometimes stationary boat during the influence period shifted their activities to other previously was slightly higher (78%, n = 18) than undocumented foraging areas within their during the control period (66%, n = 82), territory. We documented this phenome- although this difference was not significant non for a single pair during 2 sample pe- (x2 = 0.96, 1 df, P = 0.326). However, the riods. These observations indicated that a 26 WILDLIFE MONOGRAPHS

0.36

mI CONTROL

< 0.30- 1 | INFLUENCE D

a. 0.24 1 a. t 0.18-1

z 012

0-1 12 2-3 34 45 5-6 6-7 7-8 9 9-10 10'1 11-12

DISTANCEFROM HUA (x100 m) Fig. 12. Numberof foragingattempts during control (n = 102 days) and influence(n = 100 days) periodsin relationto distance fromthe centers of the high-useforaging areas (HUA)on the ColumbiaRiver estuary, 1985-86, for all baldeagle pairscombined. single, stationary boat had the potential to Bear Valley, California, Unpubl. rep. to affect a pair's home range use patterns be- U.S. For. Serv., 89pp., 1981) noted that yond the obvious proximal effects. wintering eagles in Big Bear Valley, Cal- The experiments conclusively demon- ifornia, temporarily shifted use away from strated that temporary changes in eagle temporary human activities such as wa- spatial use patterns were the direct result terfowl hunting. Similarly, Smith (1988) of the human activity. The experimental noted that summering and migrating ea- stationary boat not only affected general gles on Jordan Lake and Falls Lake, North eagle use of the foraging areas, but also Carolina, shifted spatial use patterns in re- reduced the number of foraging attempts sponse to the weekend influx of people. and probably reduced overall foraging ef- These studies, encompassing summering ficiency. We are not aware of similar ex- and wintering populations throughout perimental studies on territorial, breeding much of the species' range, combined with bald eagles, but observations on the for- our experimental findings, confirm that aging behavior of nonbreeding eagles sup- human activities can elicit dramatic port our findings. Skagen (1980) noted that changes in spatial use patterns of bald ea- winter feeding activity on the Skagit River gles. in Washington was significantly reduced Experimental Disturbance: Variation for periods of up to 30 minutes following Among Pairs.-Based on the number of a human activity. Similarly, Stalmaster and perch visits, estimates of response distance Newman (1978) noted that winter feeding for most pairs (5 of 9) were between 400 behavior on the Nooksack River in Wash- and 600 m (range 200-900 m; Table 2). ington was disrupted by human presence. Deep River and Aldrich Point pairs de- Humans temporarily displaced eagles to viated noticeably from that distance. The marginal habitat and confined the popu- Clifton pair also exhibited a marked de- lation to a smaller area, and disturbed birds viation, however the data for this pair were avoided the same feeding area for long too scarce to interpret reliably. The Deep periods following the disturbance. H. Wal- River pair was by far the most tolerant of ter and K. L. Garrett (The effect of human human activity. They avoided an area activity on wintering bald eagles in the Big within 100-200 m of the experimental sta- BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 27

Table2. Response of breedingbald eagle pairsto an experimentalstationary boat in high-useforaging areas on the Columbia Riverestuary, 1985-86. Minussigns representinfluence zones used >50% less than expected duringthe influenceperiod; plus signs representinfluence zones used >50% more than expected duringthe influenceperiod; zeros representinfluence zones used as expected, based on controlperiod use; blanksrepresent influence zones not used duringeither sampling period. Pairs are arrangedin orderof least to greatest response distance,where response distance is definedas the contiguousarea around the experimentaldisturbance used less than expected. For3 pairsstudied during both years, response distances are given by year.

Influence zone (x 100 m)a Eagle pair 1 2 3 4 5 6 7 8 9 Deep River - 0 0 - 0 0 + 0 - Clifton - 0 - 0 - - - 0 + Young's River 1986 - - 0 + 0 0 + + + Mill Creek 1985 - - - + + - 0- Mill Creek 1986 - - + + - + Fort Columbia 1986 - - - 0 - + + + 0 Rocky Point - - - - + + - - 0 Maygar - - - - 0 0 0 Young's River 1985 - - - - 0 0 0 - 0 Fort Columbia 1985 - - - - - 0 0 + - Twilight - - - - - 0 0 Aldrich Point ------+ All pairs combined - - - 0 0 0 0 0 +

a Influence zones represent 100-m-wide concentric bands around the location of the experimental stationary boat in the center of the high-use foraging areas. tionary boat (Table 2). In contrast, the Al- mans, particularly the number and out- drich Point eagle pair was the least toler- come of previous interactions (Fraser 1984, ant. They avoided an area within 800-900 Knight and Skagen 1987). Although re- m of the experimental stationary boat (Ta- searchers generally acknowledge that ble 2). Differences between these pairs learning plays a major role in avian be- probably was related to the characteristics havior, we do not clearly understand how of their high-use foraging areas, particu- learning influences an eagle's tolerance of larly the size of the foraging areas and the humans. Eagles may either become sen- availability of alternative perch sites. For sitized or habituated to human activities. example, the Deep River foraging sites in- For example, Stalmaster and Newman cluded extensive tidal mud flats (i.e., pre- (1978) noted that adults flushed at greater ferred foraging habitat) with several al- distances from humans than immatures, ternative hunting perches (Fig. 13A). Perch suggesting that conditioning was taking trees were well distributed along the shore, place. However, others have not detected and pilings and large driftwood were scat- differences between age classes in flush re- tered throughout the tidal mud flats. Con- sponse rate or flush distance (Russell 1980, sequently, the eagles were able to shift their Knight and Knight 1984, this study), and, use to alternative perch sites and still hunt in a recent study, Stalmaster (pers. com- the same tidal mud flats. In contrast, the mun.) noted that immatures were more Aldrich Point foraging sites were com- easily disturbed than adults on the Skagit prised of smaller, localized tidal mud flats River. Fraser et al. (1985) found that in- with few or no alternative perches. Eagle cubating eagles flew at greater distances use was concentrated on single, isolated with each repeated disturbance. On the trees on shrub-dominated islands (Fig. other hand, Stalmaster and Newman (1978) 13B). Hence, the eagles exhibited an "all saw no change in flush distance of winter or nothing" response; they either used the migrants with repeated disturbances of area or they did not. perched eagles. These confusing observa- Variation among pairs also may have tions may reflect differences in method- reflected previous experiences with hu- ologies and populations studied. For ex- 28 WILDLIFE MONOGRAPHS

Fig. 13. Foragingareas of Deep River(top) and AldrichPoint (bottom) bald eagle pairs on the ColumbiaRiver estuary. Deep Riverhigh-use areas were located on and adjacentto the prominentpoints along the shoreline.Aldrich Point high-use areas were located on and adjacentto the isolated tree perches on the shrub-dominatedislands. BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 29 ample, Fraser et al. (1985) found that Washington were more easily disturbed by incubating birds were sensitized by re- approaches that occurred on the river peated disturbances; the same birds were channel than from adjacent farm mead- disturbed on subsequent occasions and, ows where human activity was common. therefore, were capable of learning and Similarly, Russell (1980) found that win- showing different responses to the same tering eagles on the Sauk and Suiattle riv- intruder. In contrast, the other studies cit- ers in Washington were more tolerant of ed involved winter transient populations. boating activities on river stretches with In these studies, repeated disturbances of higher levels of human activity. the same individuals were probably rare, Differential exposure to persecution and, therefore, the birds were not likely to (Fraser 1984, Knight 1984) an unequal food become sensitized. availability among territories also may have If pairs were becoming habituated or influenced differences in response patterns sensitized to humans in our study area, among pairs. Skagen (1980) and Knight then we would expect to see a positive and Knight (1984) associated food scarcity (sensitized) or negative (habituated) cor- with flush responses of wintering eagles on relation between site-specific human ac- the Skagit and Nooksack rivers. We were tivity levels and response distance. In other unable to quantify site-specific prey levels words, we might expect pairs with higher and, therefore, cannot adequately address boating activity levels to be either more this possibility. Apparently no single factor sensitive (greater response distances) or effectively explains variation among pairs. more tolerant (lower response distances) Hence, it is likely that these and other un- of human activity. However, response dis- identified factors interact in a complex tance was neither positively nor negatively manner to affect response patterns. correlated with boating activity levels (i.e., Experimental Disturbance: Variation stationary, moving, and total BAM's) with- Among Nesting Stages.-Although the in entire territories (r < ?0.425, n = 9 average response distance remained ap- pairs, P > 0.261, 3 separate simple linear proximately the same (300-400 m) among regressions), within 1,200 m of high-use the 3 nesting stages, the difference in eagle foraging areas (r < ?0.358, n = 9 pairs, use between control and influence periods P > 0.344, 3 separate simple linear re- was least during the incubation stage and gressions), or within 400 m of high-use became increasingly pronounced during foraging areas (r < ?0.316, n = 9 pairs, the subsequent 2 stages (Fig. 14). In other P > 0.407, 3 separate simple linear re- words, the magnitude of difference be- gressions). Hence, we had no indication tween control-period use and influence- that previous exposure to human distur- period use within 400 m of the high-use bance influenced response distance. foraging areas increased from incubation Several additional factors may have in- to postfledging stages (Fig. 14). This was fluenced the observed variation in re- due to increased intensity of foraging ac- sponse patterns among pairs. Pairs may tivity during the latter nesting stages (Fig. have reacted differently to the location of 5A). Eagle use also became more concen- the experimental disturbance within their trated in the high-use foraging areas; that home ranges. That is, eagles may have be- is, the dramatic increase in activity oc- come differentially habituated or sensi- curred in the centers of the high-use areas tized to human activities in different areas (Fig. 15). This concentrated use resulted within their home ranges. Stalmaster and from a preference for 1 or more foraging Newman (1978) suggested that tolerance perches in the center of most foraging ar- to human activity is related to the location eas. Increasing foraging demands during of the disturbance; that is, eagles are more the nestling and postfledging stages re- tolerant of human activities where they sulted in more frequent visits to those cen- expect to see them. They showed that win- tral perches. Therefore, the ecological con- tering eagles on the Nooksack River in sequences of disturbing high-use foraging 30 WILDLIFE MONOGRAPHS

1.5

:_ INCUBATION M NESTLING < 100 - U POSTFLEDGING

. 80-- 09

0-100 100200 200300 300400 0-100 100-200 200-300 300-400

DISTANCEFROM HUA (m) C cc, Fig. 15. Total bald eagle activitytime by nesting stage (in- 0. cubation, nestling, postfledging)during the control period in relationto distance fromthe centers of the high-useforaging :,) areas (HUA)on the ColumbiaRiver estuary, 1985-86, for all eagle pairs combined.

Response to Moving 3.0. C POSTFLEDGING Human Activities 2.5. Most humans did not approach eagles 2.0- at close distances and rarely disturbed ea- when did. activities ac- 1.5--1 gles they Boating counted for 79% of all moving human ac- 1.0 tivities by time (n = 20,439 min). Pedestriansand automobilesaccounted for 0.5- 9-kse11~ 8 an additional 12%and 7%,respectively, of I L-"-IV- --a it ^ *- :- l 0v I0 1aruw2r lw;6rF 89 F^ P these human activities. Of 1,676 individual 0-1 I 2-3 I 4-5 6-7 I 8-9 10-11 I 1-2 3-4 5-6 7-8 9-10 11-12 moving boating activities, where each was considered a DISTANCEFROM HUA (x100 m) boating activity single observation regardless of the number of Fig. 14. Numberof perch visits duringcontrol and influence periodsin relationto distancefrom the centers of the high-use eagles encountered, only 21% resulted in foragingareas (HUA)for (A) incubation,(B) nestling,and (C) contact <500 m with eagles (i.e., encoun- postfledgingstages on the ColumbiaRiver estuary, 1985-86, ter rate; and 0.78%resulted for all baldeagle pairscombined. Note differencesin scales on Fig. 16A), only the Y-axes. in a visible disturbance to an eagle. When each human-eagle encounter within 500 m was considered separately (i.e., each hu- areas (with respect to time-energy budgets man activity could be involved in several and productivity) are potentially much encounters with different eagles), includ- greater during the nestling and postfledging those involving research-relatedactiv- ing stages, particularlybecause human ac- ities, we observeddisturbances in only 6.4% tivity levels are greatest during this period. of the cases (n = 765) involving moving Of course, extensive disturbances in for- boats (i.e., disturbance rate; Fig. 16B). We aging areas early in the nesting cycle could noted very different patterns of encounter cause adults to abandon early nesting ef- rates and disturbance rates for other hu- forts altogether, although this occurrence man activities (Fig. 16). With the excep- is unlikely with current human activity tion of train activities, high levels of these patterns (Fig. 5A). other moving activities could result in sig- BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 31

A ENCOUNTER RATES (32 of 54)

1 o nf 10 z 0 0 30- CLLL B DISTURBANCE RATES

1f1 nf 19N 25-

20- (4 of 22)

15-

10-

(0 of 55) 0 AUTO- PEDESTRIANAIRCRAFT BOAT TRAIN MOBILE HUMAN ACTIVITYTYPE

Fig. 16. (A) Proportionof individualmoving human activitiesapproaching within 500 m of bald eagles (where each human activitywas considered 1 observation)and (B) the proportionof separate human-eagleencounters within 500 m resultingin visibleeagle disturbances(where each movinghuman activity could be involvedin > 1 encounter)on the ColumbiaRiver estuary, 1985-86. Disturbancerates (B) includeencounters involving research activities. nificant numbers of eagle disturbances. At limit the remainder of this section to flush current levels, however, they are of little responsesonly. Eighty percent of the flush consequence to the eagle population. In- responsesinvolved a boating activity. Only teractions involving moving boating activ- a few flush responses were caused by pe- ities, because of their greater numbers, are destrians (4), low-flying aircraft (4), and of much greater consequence to the eagle off-road automobiles (2). Small sample siz- population. es prohibited us from analyzing each hu- Disturbances involving moving human man activity separately. Therefore, we activities consisted mostly of flush re- limited the subsequent analyses to boating sponses (50 of 61 visible disturbances); the activity; where there were instructive dif- remainder were agitation responses. We ferences, we also analyzed all human ac- discuss agitation responses from a theoret- tivities collectively. ical perspective in a subsequent section (see Flush Response Rate and Flush Dis- Dual Disturbance Threshold Model). We tance.-Flush responserate decreasedwith 32 WILDLIFE MONOGRAPHS

300- A ENCOUNTER!S count for the fact that an eagle that flushed at - 1 particular distance probably would 250 have flushed at lesser distances as well had co B NO RESPONSE i 200- * FLUSH it encountered the human under similar z conditions. It is perhaps more instructive O 150- to define response rates on the basis of z cumulative the 100 - response frequency (i.e., total number of eagles flushing at or be- 50 - yond a particulardistance), which assumes that all flushed eagles also would have flushed had they encountered the activity 100" B FLUSH RATE at all closer distances;this may or may not be true. This was the approach used by 80- E CUMULATIVEFLUSH RATE Stalmasterand Newman and E INCREMENTALFLUSH RATE (1978) Knight and this Z 60 Knight (1984). Using approach, 71.4% of all eagles (n = 56) flushed when LU a. 40 boats approached within 100 m (Fig. 17B). This approach generates maximum esti- 20 mates of response rates. The true response rates are probably bounded by these 2 estimates. Regardlessof approach, even high 0-100 100-200 200-300 300-400 400-500 responserates at these distances are of little ENCOUNTERDISTANCE (m) consequence, because eagles rarely encountered boats and other human activi- Fig.17. (A)Numberof separate human-eagle encounters and flush responses (whereeach movinghuman activity could be ties at close distances (<200 m; Fig. 17A). involved in >1 encounter) and (B) correspondingflush re- Our response rates were much lower sponse rates in relationto encounterdistance for movingboating activitieson the ColumbiaRiver estuary, 1985-86. Incre- than those observed for other populations. mentalflush rate = no. flushes - no. encounters;cumulative Almost 43% of wintering eagles on the flush rate is based on the sum of all flushes occurringat a Skagit River flushed when boats, particulardistance and pedes- greater. trians, or land-based vehicles approached within 500 m (Skagen 1980), compared to increasing encounter distance (Fig. 17B). the 5.7% flush response rate we observed Flush response rates were much higher for all human activities combined (n = than expected at 0-200 m and less than 874). Stalmaster(pers. commun.) reported expected beyond 200 m (Table 3). The similar high response rates on the Skagit standardized residuals in Table 3 indicate River (26-51% for perched birds, 86-88% that response behavior was fairly consis- for birds on the ground). Differences may tent within these distance categories; that reflect differences between wintering and is, the significance of the difference be- breeding birds. Territorial,resident eagles tween observed response rate and expect- may be more secure in their surroundings ed response rate, measured by the mag- than nonresident wintering eagles. How- nitude of the standardized residuals, was ever, 80% of the encounters reported in similar between 0-100 and 100-200 m, our study involved nonbreeding adults and and likewise was similar for the zones be- immatures. These populations may have yond 200 m. Nevertheless, few eagles were been exposed to different levels and con- flushed by moving boats, even when boats ditions of disturbance, and, consequently, approached at close distances (Fig. 17A). developed different tolerances to humans Only 5.3% of all eagles (n = 756) flushed (Fraser 1981). For example, Knight and when boats approached within 500 m; even Knight (1984) noted that wintering eagles when boats approached within 100 m, only were more tolerant of boating activity on 27.3% of the eagles (n = 22) flushed (Fig. the Skagit River where human activity was 17B). However, these statistics do not ac- relatively high than on the Nooksack River BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 33

Table3. Response of baldeagles to movingboating activities and all movinghuman activities combined on the ColumbiaRiver estuary, 1985-86. Each observationrepresents a single human-eagle encounterwithin 500 m, includingboth research- and nonresearch-relatedhuman activities and both breedingand nonbreedingbald eagles.

Responsebehavior No response Flush response No. of Standardized No. of Standardized Encounterdistance x2 P observations residualsa observations residuals Total Boating activities 55.14 <0.000 0-100 m 16 -1.06 6 +4.48 22 101-200 m 107 -1.13 19 +4.78 126 201-300 m 254 +0.43 7 -1.83 261 301-400 m 225 +0.42 6 -1.78 231 401-500 m 114 +0.39 2 -1.67 116 Total 716 40 756 All human activities 41.93 <0.000 0-100 m 30 -0.97 8 +3.95 38 101-200 m 162 -0.94 23 +3.82 185 201-300 m 267 +0.36 10 -1.47 277 301-400 m 235 +0.52 6 -2.10 241 401-500 m 130 +0.41 3 -1.67 133 Total 824 50 874

a (observed- expected) V/expected. where human activity was relatively low. were obtained under varying conditions The most likely explanation for these dif- and represent the range in responses by ferences, however, is that the populations individuals. The range in mean population were subjected to different human en- flushdistances was somewhat less. Further, counter rates at close distances. On the flush distances obtained under conditions Columbia River, a much higher propor- comparable to this study are surprisingly tion of the human-eagle encounters oc- similar. Mean flush distance for wintering curred beyond 200 m (Fig. 17A). Low re- adult bald eagles responding to pedestrian sponse rates beyond 200 m decreased the disturbances on the Nooksack River in overall response rate. Clearly, response Washington was 196 m (Stalmaster and rates are only meaningful if evaluated rel- Newman 1978). Similarly, mean flush dis- ative to both encounter distance and the tances in response to boating disturbances distribution of encounters among distanc- for wintering eagles perched in trees on es; response rates alone may lead to erro- the Skagit and Nooksack rivers in Wash- neous conclusions. ington were 168 and 150, respectively Flush distances ranged from 50 to 468 (Knight and Knight 1984), and mean flush m and averaged 197 m (SE = 15, n = 40); distance in response to boating distur- 95% of the eagles that flushed did so at bances for summering and migrating ea- distances <350 m (Fig. 18). Flush dis- gles on Jordan Lake and Falls Lake in tances for all human activities combined North Carolina was 137 m (Smith 1988). were similar, although mean flush distance The similarity in these distances suggests for aircraft disturbances was notably less that there may be a general tolerance than other human activities (x = 87 m, SE threshold for foraging eagles. = 23, n = 4). Factors Influencing Responses.-No Reported flush distances range from 25 differences in flush response (whether ea- to 990 m (Vian 1971, Steenhof 1976, Nye gles flew or not) or flush distance were and Suring 1978, Stalmasterand Newman attributable to nesting stage, cloud cover, 1978, Skagen 1980, Knight and Knight eagle hunger condition, eagle age, breed- 1984, Fraser et al. 1985). These estimates ing status, or residence status (P > 0.170). 34 WILDLIFE MONOGRAPHS

100 -0- CUMULATIVERESPONSE *- INCREMENTALRESPONSE 95% 80

i- z 60 LLI 0 cQ w a- 40

0 I 100-150 1 200-250 I 300-350 50-100 150-200 250-300 350-400 450-500

FLUSHDISTANCE (m) Fig. 18. Percentof baldeagles flushed(n = 40) by movingboats in relationto distancebetween eagle and boat (flushdistance) on the ColumbiaRiver estuary, 1985-86. Incrementalresponse equals the percent of total flush responses occurringat a particularencounter distance; cumulative response equals the percentof totalflush responses occurringat a particularencounter distance or less.

Seven factors, however, influenced flush trees (G2 = 34.97, 2 df, P < 0.000; Fig. response, 3 of which also influenced flush 19A). Eagles on the ground also flew at distance (Table 4). greater distances than eagles perched Eagle perch height had a particularly higher off the ground (Table 4). Other in- strong influence on response patterns. Ea- vestigators have noted similar patterns (Jo- gles perched on or close to the ground nen 1973; Stalmaster 1976; Skagen 1980; flushed in response to approaching boats Knight et al., In Press; Stalmaster, pers. more often than eagles perched higher in commun.). Knight and Knight (1984) re-

Table 4. Mean flush distances for factors (Table 1) with significanteffects on bald eagle flush distance for movingboating activities(n = 40) and all humanactivities combined (n = 50) on the ColumbiaRiver estuary, 1985-86.

All human activities Boating activities Flush distance Flush distance KW or MWa KW or MW Factor and level n Mean SE (P value) n Mean SE (P value) Eagle perch height <1 m 17 251 24 10.34 14 249 23 7.73 1-10 m 17 146 21 (0.006) 13 152 23 (0.021) ?-10 m 16 180 24 13 184 29 Eagle activity Foraging or feeding 33 167 15 175 24 175 19 136 Resting or other 17 242 26 (0.030) 16 229 25 (0.118) Time of day -0600 hrs 11 153 31 5.17 9 160 35 7.63 0600-0800 hrs 18 224 21 (0.160) 15 236 22 (0.054) 0800-1000 hrs 10 189 32 7 149 20 >1000 hrs 11 185 35 9 205 39 a KW = Kruskall-Wallis analysis of variance for factors with >2 levels; MW = Mann-Whitney U Statistic for factors with 2 levels. BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 35

40- A EAGLE HEIGHT 30- (2of 7) B DISTANCE FROM NEST 35- s11 m 25- I_400 m 30- m1-10m inbom w 25- 20- 0 0 20- 15- 15- uL 10- 10- 0 5- 5- I1 ( of 276) (0 of1) - 0- 0 u-z;j 0-200 200-500

of UJ 40- C BOAT SPEED 45- (7 16) D BOATNOISE w 35- - SLOW 40- - SOFT CO 30- MODERATE 35- LOUD w =] FAST 30- 25- |1(I5 of 27) z 25- 0 20 0. 20- u0 15- 15- 10- 10- of 64) (30f66) 0D 5- 5- (12 s 5 0 (3 of 101) LL 0 0- 0-200 200-500 0-200 200-500 ENCOUNTER DISTANCE (m) ENCOUNTER DISTANCE (m)

1-11 30-g (622) E PRECIPITATION w H 25- m NO PRECIPITATION M RAINY w 20-

w 15

0-200 200-500 ENCOUNTER DISTANCE (m) - Fig. 19. Percentof baldeagles flushed(flush response rate = no. flushes no. human-eagleencounters) by movingboats in relationto distance between eagle and boat (encounterdistance) and (A) eagle perch height, (B) distance between eagle and nest (breedingadults only),(C) relativeboat speed, (D) relativeboat noise, and (E)precipitation on the ColumbiaRiver estuary, 1985-86. corded nearly identical flush distances for as their major food source, because these wintering eagles on the ground and eagles must spend considerabletime on the perched in trees on the Skagit and Nook- ground feeding (Knight et al., In Press.). sack rivers. These consistent findings con- Eagles on or near their nests tolerated firm that eagles perched on or close to the approaching humans more than did eagles ground are more easily disturbed by hu- away from their nests. Specifically, eagles mans than eagles perched higher in trees. perched -800 m from their nests flew in The implications of this are most serious response to a boat more often than eagles for eagles dependent on large fish carcasses perched <400 m from their nests (G2 = 36 WILDLIFE MONOGRAPHS

20- A EAGLEACTIVITY.

0-- M ON NEST w ~15- M HUNTINGOR FEEDING *l RESTINGOR OTHER w o 10- Cl) w

I 5- (12 of C/) 278) 7 261Dof U- 0 (0 of 20) (Oof 12)26)

20- B TIMEOF DAY 0 < 0600 HRS uJ 0601-0800 HRS 15- El < 0801 HRS-1000 HRS LU >1000 HRS C) z o 10- cn03 7 of 79)

l 5 5- U- (4of 295) I 0- 0-200 200-500 ENCOUNTERDISTANCE (m) Fig. 20. Percentof baldeagles flushed(flush response rate = no. flushes - no. human-eagleencounters) by movinghumans (allhuman activities combined) in relationto distance between eagle and human(encounter distance) and (A)eagle activityand (B) time of day on the ColumbiaRiver estuary, 1985-86.

14.02, 2 df, P < 0.000; Fig. 19B), although the nest (e.g., incubating, brooding) were flush distances were similar (MW = 12, 1 less disturbed by approaching humans (all df, P = 0.078). The significance of this human activities combined) than foraging relationship was largely determined by the or feeding eagles and resting eagles (G2 = higher than expected response rate of ea- 9.27,2 df, P = 0.010; Fig. 20A), and resting gles >800 m from their nests and lower eagles flushed at greater distances than for- than expected response rate of eagles < 400 aging or feeding eagles (Table 4). These m from their nests for human-eagle en- patterns were evident with boating activ- counters occurring beyond 200 m. The high ities alone, but the relationships were not response rates for encounters within 200 significant (G2 = 3.33, 2 df, P = 0.210; m (Fig. 19B) did not contribute signifi- Table 4). cantly to the overall relationship between Flush responsewas affected by the speed response behavior and distance from nest and noise intensity of approaching boats. because the number of encounters within Eagles flushed more often than expected 200 m was small compared to the number when boats approachedslowly or were loud of encounters beyond 200 m. Eagles on than when boats approached rapidly or BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 37 were quiet (speed, G2 = 18.16, 2 df, P < Response Behavior. -Although most 0.000, Fig. 19C; noise, G2 = 4.84, 1 df, P eagles continued to use the same foraging = 0.028, Fig. 19D), although flush dis- area following a disturbance, several ea- tances were similar (speed, KW = 1.56, 2 gles were displaced from the area com- df, P = 0.459; noise, MW = 172, 1 df, P pletely. Twenty-six percent of flushed ea- = 0.501). Boat speed and noise were par- gles (n = 50) flew to unknown destinations ticularly influential for human-eagle en- away from the immediate foraging area; counters within 200 m. Stalmaster (pers. the remainder (74%) flew an average of commun.) also observed slow moving boats 469 m (range 0-1,596 m, SE = 69, n = 37) to be more disruptive of eagle feeding ac- to the next perch. Two eagles (4%) re- tivity than fast moving boats on the Skagit turned to the original perch following dis- River. Fraser (1981) observed a similar re- turbances by low-flying aircraft. In con- lationship for pedestrian disturbances of trast, wintering eagles on the Nooksack incubating and brooding eagles. He sug- River flew between 50 and 500 m and gested that slower nest approaches allowed rarely left the river system (Stalmaster and eagles more time to "decide" to fly. We Newman 1978). These authors demon- also noted that eagles were largely unaf- strated that visual buffers effectively re- fected by fast moving land-based vehicles, duced flush distance. The openness of the but became increasingly agitated as ve- Columbia River estuary in comparison to hicles slowed to a stop nearby. H. Walter smaller rivers may account for the differ- and K. L. Garrett (The effect of human ences in avoidance flight patterns. On the activity on wintering bald eagles in the Big Columbia River, longer avoidance flights Bear Valley, California, Unpubl. rep. to may be necessary for eagles to visually U.S. For. Serv., 89pp., 1981) noted that loud buffer themselves from the source of dis- vehicle noises between 50-200 m flushed turbance or to find tree perches; suitable wintering eagles in Big Bear Valley. On perches are widely scattered as a result of the Columbia River, we observed the same intensive tree harvests. relationship for boats. Time of day also influenced flush re- DUALDISTURBANCE sponse. Eagles flushed in response to all THRESHOLDMODEL human activities combined more often before 0800 hours than after 1000 hours (G2 We developed a conceptual model to = 9.80,3 df, P = 0.021; Fig. 20B), although help explain how eagles respond to tem- the same relationship was not apparent for porary human activities in foraging areas boating activities alone (G2 = 2.95, 3 df, P based on the results of our study. We dis- = 0.420). Time of day had a mild effect cuss this model at the individual human- on flush distance for boating activities eagle encounter level and extended it to alone, but inconsistent flush distances pre- the pair and population levels. This model vented us from interpreting the relation- is useful for evaluating disturbances in ship (Table 4). These results support our populations subject to any level of human earlier conclusion that human activities disturbance and is particularly helpful during early morning were potentially when temporary disturbances of foraging more disturbing to foraging eagles. Al- areas are of concern or when considering though human activity has been reported flush distance as a measure of disturbance to disrupt eagle foraging activity in early reaction (Hediger 1968:40-41). More im- morning (Skagen 1980), we are unaware portantly, it provides a conceptual basis of other quantitative evaluations of this for establishing buffer zones that effec- relationship. Eagles also flew in response tively satisfy management objectives. We to boats more often on rainy days than on emphasize that this model provides a con- clear days (G2 = 3.69, 1 df, P = 0.055; Fig. ceptual framework for developing effec- 19E), but flush distances were similar (MW tive research and management strategies = 107, 1 df, P = 0.174). that reflect the predominant types of hu- 38 WILDLIFE MONOGRAPHS

sponse by an eagle. AD is always greater A than or to FD and is much ui equal generally CO) z 0 greater. a. FLUSH NCE the existence of AD is intu- ZONE Although ZONE itive, it is not an parameter to estimate Jc:3w easy IDCC in the field. On the Columbia River, in w0 FLUSHDISTANCE to OVERTIME \ addition the agitation behaviors preced- ing human-induced flushes, we observed 11 agitation responses to moving human FLUSHDISTANCE AGilTATION DISTANCE activities that did not result in a flush from

EAGLE HUMAN the perch; AD ranged from 80 to 486 m (x = 231, SE = In these cases, DISTANCEBETWEEN HUMAN AND EAG 43). eagles were visibly disturbed by the approaching Fig. 21. Hypotheticalrelationship between tt to elicita flush response and the distance betwveen human and human, but the human apparently did not bald eagle for a single human-eagle encountter (Dual Distur- approach close enough to elicit a flush re- bance ThresholdModel). sponse. Because of our inability to effectively assess this behavioral state, these ob- man-eagle interactions in the area under servations underestimate the true AD's. consideration. The model itse]If does not Fraser (1981) observed a related response provide specific management rrecommen- pattern during intentional nest distur- dations;it provides the framewc)rk for gen- bances in Minnesota. During nest ap- erating recommendations. proaches, he noted disturbance vocalizations 190 and 50 m prior to flushing on 2 Single Human-Eagle Encounter occasions. He reasoned that the birds had become psychologically disturbed well in Moving Human Activities.- -Under any advance of the time at which they actually specific set of conditions, an e agle has at took flight. Similarly, Greig-Smith (1981) least 2 distinguishable disturba]nece thresh- also distinguished between 2 disturbance olds (i.e., dual disturbancethres ;holds). For thresholds in barred ground doves (Geo- any human-eagle encounter, a relation- pelia striata). We are unaware of other ship exists between the hum;an-to-eagle attempts to distinguish between these 2 distance and the eagle's behavior. As the disturbance thresholds, although we sus- human-to-eagle distance decreeases (i.e., as pect that observations similar to ours are a human approaches an eagle) , eagle be- more common than published reports in- havior changes markedly at 2 [)oints (Fig. dicate. 21): (1) the distance at which an eagle ex- Our observations also suggest that at dis- periences human-induced physiiological or tances between AD and FD, the human- psychologicalchanges (e.g., increased heart to-eagle distance and the time required to rate, diversion of attention) and (2) the elicit overt escape behavior are related (Fig. distance at which an eagle exhi[bits escape 21). If a human approaches an eagle and behavior (i.e., flushes).The fornner may be remains at some distance between the 2 considered the agitation distance (AD) and thresholds, the eagle's disturbance level in- is recognized as the distance alt which an creases over time until eventually the eagle approaching human first elicit:s a change flushes. In effect, the eagle's FD gradually in an eagle's physiological or ps>ychological increases as the disturbance remains in ef- state. At distances greater than this, the fect. Naturally, the closer the human is to eagle may be aware of the huiman pres- the eagle's FD, the less time required to ence, but experiences no agitatic)n. The lat- elicit a flush response. Although the shape ter may be considered the flus;h distance of this relationship is unknown, the cur- (FD) and is the distance at whiich an ap- vilinear pattern depicted in Figure 21 il- proaching human first elicits :a flush re- lustrates the concept. BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 39

On the Columbia River, we quantified approaches AD for prolonged stationary this "time-lag" response on 5 occasions. In activities, and, as a result, FD has limited these cases, the human approached a utility for interactionsinvolving prolonged perched eagle at distances of 87-324 m (x stationary activities. = 219, SE = 54) and remained at that On the Columbia River, we quantified distance for 2-25 minutes (x = 7, SE = 4) areas of high avoidance around the ex- before the eagle flushed. Prior to flushing, perimental stationaryboat and smaller ar- eagles exhibited agitated behaviors as de- eas of complete avoidance. We noted that scribed previously. Fraser (1981) inter- eagles sometimes (n = 36) approached mittently approached nests in an attempt within their AD for brief periods, even to flush incubating or brooding eagles. He though they were noticeably agitated. speculated that the intermittent approach These approaches were generally associ- allowed birds more time to "decide" to fly ated with successful foraging attempts. and that this partially explained the rela- Foraging success within 400 m of the sta- tively high flush distances observed. This tionary boat was greater during the influ- observation supports our suggestion that ence period than during the control period given enough time, human activities lo- (see Experimental Disturbance: General cated between AD and FD will elicit a Effects), suggesting that eagles only ap- flush responseand effectively increase FD. proached within their AD when the prob- Stationary Human Activities. -The ability of a successful prey capture was model also applies to interactionsinvolving very high. stationary human activities. The existence of AD is based on the assumption that an Pair or Population Level eagle will not subject itself to disturbing conditions for prolonged periods; it is rec- We have defined the relationship be- ognized as the general shift in eagle activ- tween human-to-eagle distance and eagle ity away from a stationaryhuman activity. behavior for a single human-eagle en- The existence of FD is based on the as- counter. Each human-eagle encounter has sumption that an eagle will never volun- a unique relationship (Fig. 21) and is in- tarily approach a human closer than its fluenced by a multitude of factors. Thus, FD; it is recognized as the area of complete the relationship for a single human-eagle avoidance around a stationary human ac- encounter has limited management utility. tivity. Although flush distance implies an However, if we determine the relationship actively approaching human, we use the for a number of encounters over a range term here for consistency. of meaningful conditions (i.e., many hu- The time-lag response for encounters man-eagle encounters), involving either between AD and FD also is applicable for the same pair or a number of different stationary activities. Eagles sometimes ap- eagles, we can develop a family of response proach stationary humans at distances less curves (Fig. 22A) pertaining to a specific than their AD for specific purposes (e.g., pair or populationof interest. Because each foraging attempts), but generally leave im- encounter establishesa single AD and FD, mediately after the tasks are completed. we can produce frequency distributions Perhaps, eagles can be attracted to a dis- for the disturbance thresholds (Fig. 22B). turbed area as long as their foraging drive These distributions can be used to deter- is greater than their agitation level. Once mine disturbance thresholds for pairs or in the disturbed area, however, an eagle's populations, based on average threshold agitation level builds until it surpassesthe distances, or any specified cumulative per- foraging drive and eventually the bird cent (e.g., see Fig. 18). Because recovery leaves. Thus, the disturbed area may re- or management plans usually involve pop- ceive limited use, but effective use of the ulations,this information has obvious man- area is largely precluded. In effect, FD agement utility. Further, this information 40 WILDLIFE MONOGRAPHS

approach of a human. This procedure is ,) objective and can be approached experi- 0 mentally or observationally. In an experimental encoun- w approach, human-eagle ters are strictly controlled; eagles are flushed under regulated conditions. This

LL approach is labor efficient, allows for control over conditions of the human-eagle encounters,and allows for strong statistical inferences. However, in this approach, eagles are subjected to highly regulated and C B somewhat artificial disturbances, which may or may not represent the conditions fu --FD Z --- AD of natural disturbances. Investigations of flush distance often have taken an experimental approach. In most cases, eagles were approached by the same researchers in the same manner. Circumstantial evi- EAGLE HUMAN dence suggests that individual eagles learn DISTANCEBETWEEN HUMAN AND EAGLE to recognize specific humans or vehicles and become sensitized to them. Fig. 22. Hypotheticalrelationship between the time required may (Fra- to elicita flush and the distancebetween humanand ser to thebehavior response and needs of the individ- 1981, Harmata 1984). Knight and bald eagle for several human-eagle encounters (A) and the Temple (1986) found that behavioral frequencyualsccupying distributions theof the territory. disturbancethresholds (B). FD = flush distance; AD = agitationdistance. changes in red-winged blackbirds (Age- laius phoeniceus) resulted from learning to recognize a particular individual; it is can be used to develop site-specific terri- likely that bald eagles have the same ca- tory management plans that are sensitive pability. This is typically not a concern tout behaviorthe and ofexpeindivid- when studying large wintering eagle pop- uals occupying the territory. ulations,because the same eagles are rarely encountered more than a few times. For Research and Management breeding populations and small, localized Implications wintering populations the bias may be significant if this factor is not considered in Moving Human Activities.-Estimat- the sampling design. ing AD involves determining the distance In an observational approach such as at which an eagle becomes agitated at the ours, human-eagle encounters are ob- approach of a human. This requires re- served under naturally occurring condi- mote sensing of individual eagle's physi- tions involving naturally occurring human ological reactions (e.g., increased heart rate) activities. This approach usually is very and direct observation of behavioral time consuming, results in low sample siz- changes (e.g., diverted attention). Such in- es, does not allow for control over condi- formation can be gathered through exper- tions of the human-eagle encounters, and imentation or observation. Either ap- allows for fewer and weaker statistical in- proach is highly subjective and extremely ferences. However, in this approach, hu- difficult without the aid of expensive and man-eagle encounters are natural and complicated equipment. We are not aware therefore represent the likely range of con- of any attempts to quantify AD of eagles ditions under which humans and eagles for interactions involving moving human will interact. Furthermore, this approach activities. allows the investigator to estimate the nat- Estimating FD involves determining the urally occurring flush response rate and, distance at which an eagle flushes at the thereby, better gauge the relative impor- BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 41 tance of interactions resulting in flush re- bias and interpretationalerror and, there- sponses compared to other forms of hu- fore, should be interpreted cautiously. For man-eagle interaction. example, in a recent study of wintering Several authors have advocated the use eagles on the SkagitRiver, Stalmaster(pers. of flush distances and flush response rates commun.) noted that the most sensitive to establish buffer zones around important eagles flushed very early in the day and eagle-use areas (Stalmaster and Newman left the river system. Consequently, any 1978, Knight and Knight 1984, Fraser et sampling scheme that failed to detect these al. 1985). Our results indicate that esti- sensitive birds would artificially deflate mates of flush distances and flush response flushresponse rates and flushdistances. Our rates can lead to very different buffer zone results suggest that flush distances should recommendations.On the Columbia River be interpreted in conjunction with flush estuary, based on an average of incremen- response rates; flush distances alone may tal and cumulative flush response rates, lead to buffer zone recommendations that buffer zones of roughly 100 m would be do not meet the management objectives. sufficient to protect 50% of the breeding Stationary Human Activities.-Esti- eagle population from human-induced mating AD involves determining the shift flushes; buffer zones of roughly 300 m (measured in distance) in eagle activity would be needed to protect 95% of the away from a stationary human activity. population (Fig. 17, Table 5). Based on This procedure is relatively straightfor- flush distances, however, buffer zones of ward and objective, but extremely time 200 and 350 m would be needed to protect consuming; it does not depend on the ob- 50% and 95%, respectively, of the popu- server'sability to determine when an eagle lation from human-induced flushes (Fig. becomes agitated. The most practical ap- 18, Table 5). The 2-fold difference be- proach is experimental, in which each of tween approaches in buffer zone widths several samples consists of 2 sampling pe- required to protect 50% of the population riods: (1) a control period during which results because the 2 estimates represent normal eagle activity patterns in the for- different information. Flush response rate aging area are monitored, and (2) an in- is defined as the percent of total human- fluence period during which the foraging eagle encounters resulting in a flush re- area is disturbed. The disturbance consists sponse. Flush response rates are therefore of a stationary human activity positioned based on the entire eagle population;every centrally in the foraging area. Agitation eagle encountered is considered, regard- distance is determined by comparing eagle less of whether it flushes or not. Flush dis- activity between the 2 sampling periods; tance is defined as the distance at which the shift (measured in distance) in eagle an eagle flushes at the approach of a hu- activity away from the human activity man. Only those eagles that flush are con- during the influence period is equal to or sidered in generating a mean flush dis- greater than AD. Consequently, the AD tance (or some specified cumulative estimate is not likely to underestimate the percent flush distance). The choice of flush human disturbance impacts. The repre- distances or flush responserates as the basis sentativeness of the disturbance and ha- for making buffer zone recommendations, bituation or sensitization to the researcher therefore, depends on whether the desire should be considered in the study design. is to consider each eagle in the population The former concern was dealt with in this equally or to emphasize the portion of the study by observing naturallyoccurring hu- population most sensitive to human dis- man activities during the firstyear of study turbance(i.e., those that flush).Buffer zones prior to any experimental disturbance usually are suggested on the basis of flush work. The later concern was partiallydealt distances. with in this study by using 3 different boats Both flush response rates and flush dis- with which to conduct experimental distances are subject to potential sampling turbances. 42 WILDLIFE MONOGRAPHS

MANAGEMENT Threshold Model and the empirical results RECOMMENDATIONS of this study, we developed alternative buffer zone strategies (Table 5). The dif- Eagles, on the average, avoided an area ference in buffer zone widths between within 300-400 m of the experimental sta- strategies based on moving versus station- tionary boat. In effect, a single, stationary ary human activities is especially note- boat displaced eagles from 28 to 50 ha of worthy and emphasizes the importance of available foraging habitat. High-use areas designing studies that reflect the true na- for pairs of breeding bald eagles encom- ture of human-eagle interactions occur- passed an average of 1.3 km2 (169 ha), ring in the population. based on harmonic mean 50% utilization The choice of an appropriatebuffer zone contours (see Spatial Patterns-Within strategy depends on management objec- Territories).Under these conditions, a few tives. In most cases, the objectives will re- strategically located boats would be suffi- quire humans and eagles to coexist with cient to effectively disturb a pair's entire minimal interference. This necessitates a high-use foraging area. Because eagles on thorough understandingof the spatial and the Columbia River estuary relied on rel- temporal activity patterns of both humans atively few (2-6) high-use foraging areas and eagles and the types and exact nature to provide the majority of their foraging of human-eagle interactions occurring in needs, human activities in these areascould foraging areas. The choice of strategiesfor prohibit eagles from meeting their ener- establishing buffer width depends on at getic demands, and ultimately could affect least 4 considerations:(1) the predominant productivity and impede recovery of this type of human activity (e.g., boats, pedes- population. We recommend management trians, aircraft, etc.), (2) whether the pre- of foraging areas, including spatial and dominant form of human activity is mov- temporal restrictions on human activity ing or stationary, (3) the degree of eagle and measures to maintain or improve for- protection sought (i.e., protection from ag- aging habitat quality. itation or flushing), and (4) the desired management level (i.e., management to Spatial Restrictions ensure protection of a certain percentage of the population). Designation of buffer zones is a widely On the Columbia River estuary, station- used management strategy for limiting hu- ary boating activities greatly outnumbered man disturbance at nest sites, communal all other types of human activities in eagle roosts,and concentrated winter feeding ar- foraging areas. Further, eagles rarely en- eas (Mathisenet al. 1977, Stalmaster1980, countered moving boats and other human Howard and Postovit 1987). Human ac- activities at distances that resulted in flush tivities around important resource areas responses. Therefore, stationary boats rep- are perhaps best managed using this con- resented the dominant source of distur- cept (Howard and Postovit 1987, Knight bance to eagles, and estimates of flush re- and Skagen 1987). The major difficulty in sponse rates and flush distance based on establishing buffer zones is determining moving human activities are not recom- zone size; any measure of disturbance re- mended measures for establishing buffer action may be used to estimate zone width. zone width. Because flush distance is in- For example, agitation and flush distances appropriate for prolonged stationary ac- for moving and stationary human activi- tivities (see Dual Disturbance Threshold ties often will differ greatly and therefore Model), we suggest the use of agitation lead to very different buffer zone widths. distance from experiments involving sta- We suggest the use of buffer zones to tionary boats to establish buffer zone width. protect high-use foraging areas of breed- We recommend 400-m-wide buffer ing bald eagles on the Columbia River es- zones around high-use foraging areas as tuary. Based on the Dual Disturbance the most appropriate and practical man- BALD EAGLES AND HUMANSON THE COLUMBIARIVER-McGarigal et al. 43

Table 5. Alternativebuffer-zone widths based on the predominantform of humanactivity, management objective, and management level (50%or 95% of the populationprotected) for high-useforaging areas of breedingbald eagles on the Columbia Riverestuary.

Predomi- Management nant form objective Managementlevel of human activity Protectionfrom 50% 95% Comments Moving Agitationa ? ? Not quantified. Moving Flushingb 100c 300 Of minor importance relative to stationary activities on the Co- lumbia River estuary; considers disturbed and undisturbed eagles. Moving Flushingd 200 350 Of minor importance relative to stationary activities on the Co- lumbia River estuary; considers only disturbed eagles. Stationary Flushinge 200 400 Not appropriate on the Columbia River estuary because flush distance equals agitation distance for prolonged stationary activities (see Fig. 21). Stationary Agitationf 400 800 Most appropriate strategy on the Columbia River estuary.

a Basedon agitationdistance. b Basedon flushresponse rates. c Minimumbuffer-zone width (m). d Basedon flushdistances. e Basedon area of complete avoidancearound stationary human activities(equals flush distance for stationaryactivities; see Dual Disturbance ThresholdModel). f Basedon area of generalavoidance around stationary human activities (equals agitation distance for stationaryactivities; see Dual Disturbance ThresholdModel). agement strategy (Table 5). Buffer zones On the Columbia River estuary, human of this width would protect >95% of all activities were highest when eagle forag- eagles from being flushed by moving boats ing efforts were greatest, both seasonally and at least 50% of all resident adults from and daily. Human and eagle activity levels the effects of stationary boats in or near increased from incubation to postfledging their high-use foraging areas. Buffer zones stages (Mar-Aug; Figs. 5A, 14). Addition- should be expanded to 800 m on a site- ally, eagle use was more concentrated in specific basis to protect sensitive pairs or the foraging areas during the later nesting sensitive foraging areas. stages (Fig. 15). As a result, human dis- turbancesin foragingareas potentially have Temporal Restrictions more ecological consequence during the nestling and postfledging stages. We sug- Temporal restrictions of human activity gest that buffer zones around high-use for- generally are used in conjunction with aging areas would be most beneficial if buffer zones as part of management strat- used during the nestling and postfledging egies. Restrictions on temporary human stages (May-Aug). This recommendation activities need only be in effect during does not negate the importance of mini- times when the birds are using the critical mizing nest disturbancesduring the early resource (Knight and Skagen 1987). Typ- nesting stages, because, in many raptor ically, temporary human activities (e.g., species, early disturbances of the nest site most recreational activities) are prohibited may lead to desertion or abandonment of within designated buffer zones during crit- eggs, whereas desertion is much less likely ical periods of the year (e.g., breeding sea- after the young hatch (Fyfe and Olendorff son). We suggest using temporal restric- 1976, Grier and Fyfe 1987). This recom- tions in combination with buffer zones to mendation also does not ignore the poten- allow eagles undisturbed use of their for- tial significance of foraging area distur- aging areas during important periods, bances during the early nesting stages, but, while also lessening the restrictions on hu- given current human activity patterns on man activity. the Columbia River estuary, extensive for- 44 WILDLIFE MONOGRAPHS aging area disturbancesare not likely dur- along the shoreline, isolated trees on the ing this period. shorelines of natural, shrub-dominated is- Our results also indicate that eagle feed- lands, and pilings on tidal mud flats. We ing activity was greatest during early suggest that these foraging perches be pro- morning hours (<1000 hrs, Fig. 5C) and tected and that previously suggested buff- that human activities during this period er zones be established around these sites. were more disturbing than activities oc- later in the Al- curring day (Fig. 20B). FUTURE RESEARCH though human activity levels were lowest during this portion of the day, consider- Fraser et al. (1985) noted the futility of able potential for interaction still exists. conducting observational studies to deter- We suggest that buffer zones would be mine the effect of human activities on bald most effective during early morning eagle parameterssuch as productivity. He (< 1000 hrs) and that restrictionswould not suggested that experiments in which a sub- be needed after 1000 hours each day. A stantial number of eagles are disrupted to restriction of this type may be difficult to the point of nest failure by a variety of enforce over a large area; however, it may human activities be conducted to address be practical for critical sites. this question. Studies of this nature are highly undesirable for threatened or en- Habitat Management dangered populations; yet, these populations need immediate research and man- Variation among pairs in responses to agement consideration. For these the experimental stationaryboat (Table 2) populations, we suggest the need for non- may have been related to differences in disruptive experiments, such as ours, de- habitat characteristicsof the high-use for- signed to relate bald eagle behavior (not aging areas. We suspect that the size of productivity) directly to human activities the foraging areas and the availability of without jeopardizing site occupancy or alternative perch sites influenced response nesting success. We further suggest that patterns. If this is true, then small, con- these experiments be formulated on the centrated foraging areas with few or no basis of a thorough assessment of the nat- alternative perch sites may be most vul- urally occurring human-eagle interactions nerable to the adverse effects of human and be designed to produce information activities. Providing alternative foraging useful to the resource manager. perches, either by retaining shoreline trees Results of our study demonstrate the or by providing artificial perches such as complexity of human-eagle interactions. pilings, may increase the likelihood of si- Disturbance is not always in the form of multaneous use of foraging areas by hu- a human-induced flush from a perch. In- mans and eagles, and thereby minimize direct disturbances,not easily investigated the ecological consequences of human ac- with an observational study, potentially tivities in high-use foraging areas. Adding have significant ecological consequences. pilings or similar structures to large tidal Future investigations on this topic should mud flatscurrently void of elevated perch- consideralternative conceptual models and es also may prove to be an effective means develop a sampling strategy that accu- of increasing foraging use of habitat cur- rately reflects the predominant form of rently not used, although this needs to be human-eagle interaction occurring in the investigated further. target area. Eagles concentrated their foraging activity in a few areas. These high-use areas were typically centered around 1 or more LITERATURECITED key foraging perches. These perches usually provided eagles with a commanding ANDREW,J. M., ANDJ. A. MOSHER. 1982. Bald eagle nest site selection and nesting habitat in Mary- view of the surrounding foraging habitat land. J. Wildl. Manage. 46:382-390. and included trees on prominent points ANTHONY, R. G., AND F. B. ISAACS. 1989. Char- BALD EAGLES AND HUMANS ON THE COLUMBIA RIVER-McGarigal et al. 45

acteristics of bald eagle nest sites in Oregon. J. southern Gulf Islands, British Columbia. Wilson Wildl. Manage. 53:148-159. Bull. 76:117-120. BAILEY,A. M., AND E. G. WRIGHT. 1931. Birds of HANSEN, A. J., M. V. STALMASTER, AND J. R. NEW- southern Louisiana. Wilson Bull. 43:190-219. MAN. 1980. Habitat characteristics, function, BANGS,E. E., T. N. BAILEY,AND V. D. BERNS. 1982. and destruction of bald eagle communal roosts Ecology of nesting bald eagles on the Kenai Na- in western Washington. Pages 221-229 in R. L. tional Wildlife Refuge, Alaska. Pages 47-54 in Knight, G. T. Allen, M. V. Stalmaster, and C. W. W. N. Ladd and P. F. Schempf, eds. Proc. sym- Servheen, eds. Proc. Washington bald eagle symposium and workshop on raptor management posium. The Nat. Conserv., Seattle, Wash. and biology in Alaska and western Canada. U.S. HARLOW, R. C. 1918. Notes on the breeding birds Fish and Wildl. Serv., Anchorage, Alas. FWS/ of Pennsylvania and New Jersey. Auk 35:18-29. AK/proc-82. HARMATA, A. R. 1984. Bald eagles of the San Luis BROCKMAN,C. F., AND L. C. MERRIAM,JR. 1973. Valley, Colorado: their winter ecology and spring Recreational use of wild lands. Second ed. Mc- migration. Ph.D. Thesis, Montana State Univ., Graw-Hill Book Co., New York, N.Y. 329pp. Bozeman. 222pp. BROLEY, C. L. 1947. Migration and nesting of Flor- HEDIGER, H. 1968. The psychology and behavior ida bald eagles. Wilson Bull. 59:3-20. of animals in zoos and circuses. Dover Publ., Inc., COON, N. C., L. N. LOCKE,E. CROMARTIE,AND W. New York, N.Y. 166pp. L. REICHEL. 1970. Causes of bald eagle mor- HORNADAY,W. T. 1920. Alaska can save the Amer- tality, 1960-1965. J. Wildl. Dis. 6:72-76. ican eagle. Nat. Hist. 20:117-119. CORR, P. 0. 1974. Bald eagle (Haliaeetus leuco- HOWARD,R., AND B. C. POSTOVIT. 1987. Impacts cephalus) nesting related to forestry in south- and mitigation techniques. Pages 183-213 in B. eastern Alaska. M.S. Thesis, Univ. Alaska, Fair- A. G. Pendleton, B. A. Millsap, K. W. Cline, and banks. 144pp. D. M. Bird, eds. Raptor management techniques DIXON, K. R., ANDJ. A. CHAPMAN. 1980. Harmonic manual. Natl. Wildl. Fed., Washington, D.C., mean measure of animal activity areas. Ecology Sci. Tech. Ser. 10. 61:1040-1044. HOWELL, J. C. 1937. The nesting bald eagles of FIENBERG,S. E. 1982. The analysis of cross-classi- southeastern Florida. Auk 54:296-299. fied categorical data. The MIT Press, Cambridge, . 1941. Comparison of 1935 and 1940 pop- Mass. 198pp. ulations of nesting bald eagles in east-central FRASER, J. D. 1981. The breeding biology and status Florida. Auk 58:402-403. of the bald eagle on the Chippewa National For- ISAACS,F. B., AND R. G. ANTHONY. 1987. Abun- est. Ph.D. Thesis, Univ. Minnesota, St. Paul. dance, foraging, and roosting of bald eagles win- 235pp. tering in the Harney Basin, Oregon. Northwest 1984. The impact of human activities on Sci. 61:114-121. bald eagle populations-a review. Pages 68-84 , AND R. J. ANDERSON. 1983. Dis- in J. M. Gerrard and T. M. Ingram, eds. The tribution and productivity of nesting bald eagles bald eagle in Canada. White Horse Plains Publ., in Oregon, 1978-1982. Murrelet 64:33-38. Headingly, Manitoba, and The Eagle Found., JENSEN, C. R. 1973. Outdoor recreation in America. Apple River, Ill. Burgess Publ. Co., Minneapolis, Minn. 284pp. , L. D. FRENZEL,AND J. E. MATHISEN. 1985. JONEN,J. R. 1973. The winter ecology of the bald The impact of human activities on breeding bald eagle in west-central Illinois. M.S. Thesis, West- eagles in northcentral Minnesota. J. Wildl. Man- ern Illinois Univ., Macomb. 84pp. age. 49:585-592. JUENEMANN, B. G. 1973. Habitat evaluation of se- FYFE, R. W., AND R. R. OLENDORFF. 1976. Mini- lected bald eagle nests on the Chippewa National mizing the dangers of nesting studies on raptors Forest. M.S. Thesis, Univ. Minnesota, St. Paul. and other sensitive species. Can. Wildl. Serv. Oc- 143pp. cas. Pap. 23. 17pp. KNIGHT, R. L. 1984. Responses of nesting ravens GREIG-SMITH, P. W. 1981. Responses to disturbance to people in areas of different human densities. in relation to flock size in foraging groups of Condor 86:345-346. barred ground doves (Geopelia striata). Ibis 123: , D. P. ANDERSON,AND N. V. MARR. 1991. 103-106. Responses of an avian scavenging guild to an- GRIER,J. W. 1969. Bald eagle behavior and pro- glers. Biol. Conserv. 56:In Press. ductivity responses to climbing nests. J. Wildl. , AND S. K. KNIGHT. 1984. Responses of win- Manage. 41:438-443. tering bald eagles to boating activity. J. Wildl. , AND R. W. FYFE. 1987. Preventing re- Manage. 48:999-1004. search and management disturbance. Pages 173- , AND . 1986. Vigilance patterns of 182 in B. A. G. Pendleton, B. A. Millsap, K. W. bald eagles feeding in groups. Auk 103:263-272. Cline, and D. M. Bird, eds. Raptor management , AND S. K. SKAGEN. 1987. Effects of rec- techniques manual. Natl. Wildl. Fed., Washing- reational disturbance on birds of prey: a review. ton, D.C., Sci. Tech. Ser. 10. Pages 1-5 in R. L. Glinski, B. G. Pendleton, M. GRUBB,T. G. 1976. A survey and analysis of bald B. Moss, M. N. LeFranc, Jr., B. A. Millsap, and eagle nesting in western Washington. M.S. The- S. W. Hoffman, eds. Proc. southwest raptor man- sis, Univ. Washington, Seattle. 87pp. agement symposium and workshop, Natl. Wildl. HANCOCK, D. 1964. Bald eagles wintering in the Fed., Washington, D.C., Sci. Tech. Ser. 11. 46 WILDLIFE MONOGRAPHS

, AND S. A. TEMPLE. 1986. Methodological Knight, G. T. Allen, M. V. Stalmaster, and C. W. problems in studies of avian nest defence. Anim. Servheen, eds. Proc. Washington bald eagle sym- Behav. 34:561-566. posium. The Nat. Conserv., Seattle, Wash. KRANTZ,W. C., B. M. MULHERN,G. E. BAGLEY, A. SMITH,T. J. 1988. The effect of human activities SPRUNT,IV, F. J. LIGAS,AND W. B. ROBERTSON, on the distribution and abundance of the Jordan JR. 1970. Organochlorine and heavy metal res- Lake-Falls Lake bald eagles. M.S. Thesis, Vir- idues in bald eagle eggs. Pestic. Monit. J. 3:136- ginia Polytechnic Inst. and State Univ., Blacks- 140. burg. 100pp. MATHISEN, J. E. 1968. Effects of human disturbance SPRUNT, A., IV, W. B. ROBERTSON, JR., S. on nesting bald eagles. J. Wildl. Manage. 32:1-6. POSTUPALSKY,R. J. HENSEL,C. E. KNODER,AND , D. J. SORENSON,L. D. FRENZEL,AND T. C. F. J. LIGAS. 1973. Comparative productivity of DUNSTAN. 1977. A management strategy for six bald eagle populations. Trans. North Am. bald eagles. Trans. North Am. Wildl. Nat. Re- Wildl. Nat. Resour. Conf. 38:96-106. sour. Conf. 42:86-92. STALMASTER,M. V. 1976. Winter ecology and ef- MCEWAN,L. C., AND D. H. HIRTH. 1979. Southern fects of human activity on bald eagles in the bald eagle productivity and nest site selection. J. Nooksack River Valley, Washington. M.S. Thesis, Wildl. Manage. 43:585-594. Western Washington State Coll., Bellingham. MULHERN,B. M., W. L. REICHEL,L. N. LOCKE,T. 100pp. G. LAMONT, A. A. BELILSE, E. CROMARTIE, G. 1980. Management strategies for wintering E. BAGLEY, AND R. M. PROUTY. 1970. Organo- bald eagles in the Pacific Northwest. Pages 49- chlorine residues and autopsy data from bald 67 in R. L. Knight, G. T. Allen, M. V. Stalmaster, eagles 1966-68. Pestic. Monit. J. 4:141-144. and C. W. Servheen, eds. Proc. Washington bald MURPHY,J. R. 1965. Nest site selection of the bald eagle symposium. The Nat. Conserv., Seattle, eagle in Yellowstone National Park. Proc. Utah Wash. Acad. Sci., Arts, and Lett. 42:261-264. , AND J. R. NEWMAN. 1978. Behavioral re- NYE, P. E., AND L. H. SURING. 1978. Observations sponses of wintering bald eagles to human activ- concerning a wintering population of bald eagles ity. J. Wild. Manage. 42:506-513. on an area in southeastern New York. N.Y. Fish , AND . 1979. Perch-site preferences and Game J. 25:91-107. of wintering bald eagles in northwest Washing- OFELT, C. H. 1975. Food habits of nesting bald ton. J. Wildl. Manage. 43:221-224. eagles in southeast Alaska. Condor 72:358-361. STEENHOF,K. 1976. The ecology of wintering bald POSTUPALSKY,S. 1971. Toxic chemicals and de- eagles in southeastern South Dakota. M.S. Thesis, clining bald eagles and cormorants in Ontario. Univ. Missouri, Columbia. 146pp. Can. Wildl. Serv., Pestic. Sect. Rep. 20. 45pp. STICKEL,L. F., N. J. CHURA, P. A. STEWART,C. M. . 1974. Raptor reproductive success: some MENZIE,R. M. PROUTY,AND W. L. REICHEL. problems with methods, criteria, and terminol- 1966. Bald eagle pesticide relations. Trans. North ogy. Pages 21-31 in F. N. Hamerstrom, Jr., B. Am. Wildl. Nat. Resour. Conf. 31:190-200. E. Harrell, and R. R. Ohlendorff, eds. Manage- STUWE,M., AND C. E. BLOWHOWIAK.1986. Micro ment of raptors. Proc. conference on raptor con- computer program for the analysis of animal lo- servation techniques, Raptor Res. Rep. 2, Raptor cations. Conserv. and Res. Cent. Natl. Zool. Park, Res. Found., Inc., Vermillion, S.D. Smithsonian Inst., Washington, D.C. 18pp. RETFALVI,L. I. 1965. Breeding behavior and feed- SWENSON, J. E. 1975. Ecology of the bald eagle and ing habits of the bald eagle (Haliaeetus leuco- osprey in Yellowstone National Park. M.S. Thesis, cephalus 1.) on San Juan Island, Washington. M.R. Montana State Univ., Bozeman. 146pp. Thesis, Univ. British Columbia, Vancouver. TODD, C. S. 1979. The ecology of the bald eagle 193pp. in Maine. M.S. Thesis, Univ. Maine, Orono. 91pp. RUSSELL,D. 1980. Occurrence and human distur- U.S. DEPARTMENTOF INTERIOR. 1982. 1980 na- bance sensitivity of wintering bald eagles on the tional survey of fishing, hunting and wildlife as- Sauk and Suiattle rivers, Washington. Pages 165- sociated recreation. U.S. Gov. Printing Off., 174 in R. L. Knight, G. T. Allen, M. V. Stalmaster, Washington, D.C. 152pp. and C. W. Servheen, eds. Proc. Washington bald U.S. FISH AND WILDLIFESERVICE. 1986. Recovery eagle symposium. The Nat. Conserv., Seattle, plan for the Pacific bald eagle. U.S. Fish and Wash. Wildl. Serv., Portland, Oreg. 160pp. SERVHEEN,C. W. 1975. Ecology of the wintering VAN NAME, W. G. 1921. Threatened extinction of bald eagles on the Skagit River, Washington. M.S. the bald eagle. Ecology 2:76-78. Thesis, Univ. Washington, Seattle. 96pp. VIAN, W. E. 1971. The wintering bald eagle (Hal- SHAPIRO,A. E., F. MONTALBANO,III, ANDD. MAGER. iaeetus leucocephalus) on the Platte River in 1982. Implications of construction of a flood southcentral Nebraska. M.S.E. Thesis, Kearney control project upon bald eagle nesting activity. State Coll., Kearney, Nebr. 60pp. Wilson Bull. 94:555-563. WATSON,J. W., R. G. ANTHONY,AND M. GARRETT. SKAGEN,S. K. 1980. Behavioral response of win- In Press. Foraging ecology of bald eagles in the tering bald eagles to human activity on the Skagit Columbia River estuary. J. Wildl. Manage. River, Washington. Pages 231-241 in R. L. WEEKES,F. M. 1974. A survey of bald eagle nesting BALD EAGLES AND HUMANS ON THE COLUMBIA RIvER-McGarigal et al. 47

attempts in southern Ontario, 1969-1973. Can. J. E. MATHISEN, F. C. ROBARDS, AND S. Field-Nat. 88:415-419. POSTUPALSKY. 1972. Residues of organochlo- WIEMEYER,S. N., T. G. LAMONT,C. M. BUNCK, C. rine pesticides, polychlorinated biphenyls, and R. SINDELAR,F. J. CRAMLICK,J. D. FRASER,AND mercury in bald eagle eggs and changes in shell M. A. BYRD. 1984. Organochlorine pesticide, thickness, 1969 and 1970. Pestic. Monit. J. 6:50- polychlorobiphenyl, and mercury residues in bald 55. eagle eggs-1969-79-and their relationships to shell thinning and reproduction. Arch. Environ. Contam. Toxicol. 13:529-549. Received 18 May 1989. , B. M. MULHERN,F. J. LIGAS,R. J. HENSEN, Accepted 11 October 1990.