migration and collision avoidance of eiders and …

Printed on recycled paper.

MIGRATION AND COLLISION AVOIDANCE OF EIDERS AND OTHER BIRDS AT NORTHSTAR ISLAND, ALASKA, 2001–2004: FINAL REPORT

Prepared for

BP Exploration (Alaska) Inc.P.O. Box 196612

Anchorage, AK 99519–6612

Prepared by

Robert H. DayAlexander K. Prichard

John R. Rose

ABR, Inc.—Environmental Research & ServicesP.O. Box 80410

Fairbanks, AK 99708–0410

July 2005

ABR Final Report i Migration at Northstar Island

EXECUTIVE SUMMARY

• During migration, 450,000+ Common, King,Spectacled, and Steller's eiders pass throughthe Beaufort and Chukchi seas during theirpassage between breeding and winteringgrounds. Because they fly rapidly and low overthe water while migrating and because theymay become attracted to lights, eiders may behighly susceptible to collision withhuman-made structures.

• The Business Rationale for this study istwofold. First, BP Exploration (Alaska) Inc.(BP), as a global company has a statedaspiration of doing no damage to theenvironment. Second, BP was mandated by itsconstruction permit to conduct a study of eidermigration and collision potential at NorthstarIsland. In addition to the U.S. Army Corps ofEngineers (USACE) permit stipulations, theU.S. Fish and Wildlife Service stipulatedseveral “reasonable and prudent” measures thatshould be taken by BP to minimize the “take”of Spectacled and Steller’s eiders at NorthstarIsland. BP met the stipulations and is fundingthis study to investigate movements ofmigrating eiders and other birds in the vicinityof Northstar Island and their responses to theseanti-collision lights.

• The objectives of this study were (1) to useornithological radar and visual sampling tomonitor the migration and behavior of birdsmigrating past Northstar Island; (2) todetermine whether migrating birds detectNorthstar Island and, if so, what their responseis and the distance at which the responseoccurs; (3) to determine whether the flightbehavior of migrating birds changes duringperiods of lowered visibility, therebyincreasing the chances of collision withNorthstar Island or its structures; and (4) todetermine the effects of anti-collision lightingon the movements and responses of birdsmigrating near Northstar Island. This is thefinal report for this study.

• We collected data on the movements, behavior,and flight altitudes of birds during 80 daystotal during 24 August–12 September 2001,11–29 September 2002, 27 August–18 September

2003, and 5–22 October 2004. We sampledwith ornithological radar and visual equipmentfor ~7–11 h/day. We alternated the use of theanti-collision lights by turning them on for thefirst half of each hour, then turning them off forthe second half; we then reversed the sequencein the next hour. This way, we were able to usethe anti-collision lights as a controlledexperiment to determine their effects onmigratory movements.

• We examined migration in the context of sixfactors that have been shown to affect birdmigration, collision rates, or both. These sixfactors included period (presence or absence ofsea ice), time of day, precipitation level,visibility, wind direction, and wind strength. Ineffect, we statistically removed the effects ofthese factors (which were important sources ofvariation) so that we could determine theeffects of the anti-collision lights on migratorymovements. We examined the effects of thesesix factors, plus lights, on seven aspects ofmigration that might indicate the effectivenessof the lights at reducing the probability ofcollision: movement rates, flight velocity,flight direction, flight behavior, island-passingsuccess, island-passing distance, and spatialdistribution. We also needed to ensure that thelights did not disrupt the birds’ migration.

• Over the 20 days of sampling in 2001, the 19days in 2002, the 23 days in 2003, and the 18days in 2004, we recorded 928 radar targetsthat we called “eiders” (690 in 2001, 105 in2002, 25 in 2003, and 108 in 2004) and 1,485radar targets that we called “non-eiders” (778in 2001, 251 in 2002, 420 in 2003, and 36 in2004), based on flight characteristics or visualidentification. Hence, we had a total of 238“eider” radar targets in 2002–2004 combined(a period with no sea ice near the island), so themajority of radar data on “eiders” are from2001 (a period with sea ice near the island).

• Birds identified visually as eiders generallyflew with a high mean velocity (~45 mi/h) andhad a high percentage of straight-line(directional) flight. In contrast, “non-eiders”tended to fly considerably more slowly than“eiders” did (~28 mi/h) and tended to exhibit a

Migration at Northstar Island ii ABR Final Report

much lower tendency for straight-line flightthan “eiders” did. The maximalmisidentification rate of radar targets as“eiders” was 55%, although we believe that itwas substantially lower than that, in that manyof the misidentified “eider” targets occurred onone night; excluding that night, the maximalmisidentification rate was 38%.

• The anti-collision lights at Northstar Islandcaused “eiders” to respond in some ways butnot in others. These lights significantlyreduced velocity at night, resulted in a spatialredistribution of the “eiders” away from theisland when sea ice was present, resulted ingreater course change (vertex) distances (butonly when visibility was good or duringtailwinds), and caused a shift farther offshorein two distance zones, resulting in a netincrease in passing distance from what itwould have been had they not changed course.They also were involved in severallights*factor interactions, especially thoseinvolving lights*winds. Overall, the lightscaused avoidance of the island, but the spatialavoidance that they caused was not dramatic.

• The anti-collision lights at Northstar Islandcaused “non-eiders” to respond in fewsignificant and meaningful ways that we wereable to detect. The lights caused velocities todrop in calm winds and tailwinds, theydecreased the proportion of non-directionalbehavior during poor visibility and duringcrosswinds and tailwinds, and they increasedthe island-passing distance in crosswinds(when birds might hit the island more easily).Hence, some of these results were difficult tointerpret in the context of collision avoidance.These lights also caused an attraction towardthe island in some cases. Hence, these lightsdid not help at increasing collision avoidanceof “non-eiders.”

• Both “eiders” and “non-eiders” exhibitedpulsed, irregular periods of movement andwere generally similar in nightly movementrates. Movement rates of “eiders” were higherwhen sea ice was present (2001), higher duringperiods without precipitation, and higherduring tailwinds and crosswinds. Movement

rates of “non-eiders” were higher when sea icewas present, higher during the day, higherduring periods without precipitation, higherduring calm winds, and higher during weakcrosswinds and tailwinds. These pulses ofmovement appear to be related to movementsof weather systems through the area thatresulted in alternating periods of favorable andunfavorable migration conditions, especiallyfor winds.

• “Eider” velocities (groundspeeds) averaged~48 mi/h overall and were higher when ice waspresent, higher with tailwinds, and higher withstrong tailwinds; importantly, they also weresignificantly lower at night when the lightswere on, indicating that the lights made themslow significantly. Velocities of “non-eiders”were higher when ice was present, higher withtailwinds, higher with strong tailwinds and(surprisingly) strong headwinds, higher withthe lights on during the day (but no effect oflights at night), and higher with the lights onduring calm winds and tailwinds.

• Both “eiders” and “non-eiders” exhibited abimodal distribution of flight direction, beingtoward both the northwest and the southeast;however, “eiders” exhibited a much strongerdirectionality than non-eiders did. Flightdirections for “eiders” averaged 294° (i.e., verystrongly toward the northwest) and weresignificantly affected only by period (299°when ice was present and 281° when ice wasabsent) and winds, with mean directions duringcrosswinds and tailwinds being toward thenorthwest, during headwinds being toward thesouthwest, and during calm conditions beingtoward the northeast. The directionality of“non-eiders” was so inconsistent overall thatwe were unable to conduct statistical tests onfactors affecting flight direction; qualitatively,they were heading toward the south duringcalm conditions, toward the west duringcrosswinds and headwinds, and toward thenorthwest during tailwinds.

• “Eiders” exhibited little variation in flightbehavior as they passed the island, with ~95%of all flocks flying in a straight-line(directional) manner, ~4% flying erratically,

ABR Final Report iii Migration at Northstar Island

and ~1% circling. The proportion ofnondirectional behavior was higher when icewas present, higher with tailwinds, higher withweak winds, and higher near the Full Moonwhen it was not visible. “Non-eiders”exhibited only ~76% straight-line (directional)behavior, ~11% exhibited erratic behavior, and~13% exhibited circling behavior. Theproportion of nondirectional behavior washigher when ice was present, higher at night,higher with good visibility, higher withtailwinds, higher with strong tailwinds, higherwith the lights off when visibility was poor,higher with the lights off in crosswinds andtailwinds, higher when the moon was full, andsignificantly higher when the moon was notvisible; however, the moon phase*moonvisibility interaction was not significant. Themoon phase/moon visibility relationshipsuggested that migrating birds of both groupsmay use the Full Moon for orientation duringmigration, possibly becoming disoriented attimes by large sources of lights when the FullMoon is not visible.

• A gas-flaring event on the evening on20 September 2002 released >3,844 Mcf(>3,844 × 106 cf) of gas and resulted in theattraction and, in some cases, near-collision, ofprimarily Long-tailed Ducks and GlaucousGulls. We cannot, however, separateconclusively the effects of the increased lightfrom the effects of the increased soundassociated with this event.

• “Eiders” exhibited almost no variation inisland-passing success (>99% successful),indicating that the island’s infrastructure andthe anti-collision lights did not affect or disruptthe birds’ migratory movements. “Non-eiders”also exhibited almost no variation inisland-passing success (~95% successful),indicating that the island’s infrastructure andthe anti-collision lights did not disrupt thebirds’ migratory movements. The only factoraffecting passing success of “non-eiders” wasperiod, in that success was higher when icewas present.

• “Eiders” varied in island-passing distance,with the overall mean being ~1,431 m overall.

(The radar could see all birds to ~2,770 mnorth and south of the radar site, orconsiderably farther than this mean distance.)Passing distances were larger when ice waspresent and larger during all winds other thancalm winds, suggesting that these birds gavethe island a wider berth when the wind wasblowing. “Non-eiders” passed the island withan overall mean distance of ~1,132 m. Passingdistances were larger when ice was absent,larger when visibility was poor, larger withcalm winds, larger with strong headwinds, andlarger with the lights on in crosswinds. Theseresults suggest that these birds gave the islanda wider berth when visibility was poor andwhen lights were on during crosswinds (whenbirds would be pushed toward the island asthey passed it).

• “Eiders” exhibited nonrandom spatialmovements through the study area, both withthe anti-collision lights off and with them on.A Before–After analysis indicated that turningon the lights caused a net decrease in themovement density of “eiders” near the islandand a net increase far from the island (i.e.,avoidance), with the net change becomingpositive beyond ~1,550 m; unfortunately, thispattern was seen only when ice was present.However, the avoidance response was not large(mean ~56 m), and the lights explained only~4% of the total variation, suggesting thatother factors had a greater effect on the spatialdistribution of these birds than the lights did.“Non-eiders” also exhibited nonrandommovements in the study area, both with thelights off and with them on. A Before–Afteranalysis indicated that turning on theanti-collision lights caused a weak attraction of“non-eiders” toward the island in all years,although the relationship was marginallysignificant when ice was present and notsignificant when ice was absent.

• High-resolution spatial analyses of individualflightlines of “eiders” indicated that they sawthe island and exhibited several behaviors thatcould be interpreted as natural anti-collisionbehavior; however, the anti-collision lightingsystem increased this anti-collision behavioronly in some cases, and never dramatically.

Migration at Northstar Island iv ABR Final Report

“Eiders” exhibited more course changes(vertices) as they approached the island andgreater angular changes/vertex as theyapproached the island; however, theanti-collision lights did not significantly affectthese responses. In contrast, for those “eiders”that originally were going to pass near theisland, the lights caused a shift of some birdsfarther offshore between 500 and 1,500 m anda net increase in island-passing distance.Hence, these birds exhibited some responsesthat indicated natural anti-collision behavior,and the lights enhanced those responses in afew cases.

• During the 80 days of sampling, we visuallyrecorded 1,075 flocks of birds totaling 17,040birds.

• Eiders flew at a mean altitude of ~6 m agl/asl(i.e., at altitudes low enough to collide withNorthstar Island and its structures) butexhibited no high-resolution responses to theisland. Mean flight altitudes of non-eiderspecies varied among species-groups, from~9 m for loons to ~19 m for gulls and averaged~11 m across all visually identified non-eiderspecies; again, these were altitudes low enoughfor birds to collide with Northstar Island andits structures. Island-passing behavior of“non-eiders” varied among species-groups,with loons exhibiting little response but allother groups exhibiting changes in flightdirection, flight altitude, or both; behavioralflaring (extreme anti-collision behavior) waseven recorded in a few instances during agas-flaring event.

• We evaluated patterns of downing andmortality of eiders at Northstar Island andnearby Endicott Island for 36 eiders and 13Long-tailed Ducks that were downed on 17total nights over the 4 fall migration periods.Patterns of downing and mortality suggest thatthe downing and mortality of eiders isassociated primarily with a Full Moon, awaxing moon, and a weakly changingbarometer. Patterns of downing and mortalityof Long-tailed Ducks suggest that the downingand mortality of these ducks is associatedprimarily with a waxing moon, a rising

barometer, and a high probability of fog. Thesepatterns suggest a possible interaction betweenthe visibility of the Moon and downing. Wecaution, however, that sample sizes are small,especially for those of nights on which birdswere downed.

• These anti-collision lights caused avoidance ofNorthstar Island by eiders, but the responsewas inconsistent and not dramatic. Flight speedwas lower at night when the lights are on, andthere was a net spatial avoidance of the islandcaused by turning on the lights. Further, thehigh-resolution analyses clearly showed anatural anti-collision response to the island, butthat response was increased both modestly andinconsistently by the anti-collision lights.Finally, these lights clearly did not causenon-eider species to avoid the island andactually appeared to cause attraction at times.

• The gas-flaring events of 2002 suggest thatflaring should be avoided (when and ifpossible) at night and during periods of limitedvisibility, to minimize the attraction of birds tothe island.

• We suggest that more birds are around theisland when ice is present: we recorded 690“eider” flocks in 1 year when ice was near theisland but only 238 over 3 years combinedwhen the ice edge was far offshore. Althoughsome of the decrease probably was related todecreased ability to sample with the radarunder such conditions, we believe that mosteider flocks cross the open Beaufort Seafarther offshore under such open-waterconditions. During the 1970s, when sea icewas heavier than it is now, Northstar Islandwas located at the inner edge of the eidermigration corridor crossing the Beaufort Sea.Hence, we suggest that fewer eiders will passnear Northstar Island as sea ice retreats as aresult of climate change in this region.

ABR Final Report v Migration at Northstar Island

TABLE OF CONTENTS

EXECUTIVE SUMMARY ............................................................................................................................ iINTRODUCTION ......................................................................................................................................... 1

BUSINESS RATIONALE.......................................................................................................................... 2OBJECTIVES ............................................................................................................................................. 3

STUDY AREA .............................................................................................................................................. 3METHODS.................................................................................................................................................... 6

DATA COLLECTION ............................................................................................................................... 6RADAR .................................................................................................................................................. 14VISUAL ................................................................................................................................................. 17

DATA ANALYSIS................................................................................................................................... 18RADAR .................................................................................................................................................. 20VISUAL ................................................................................................................................................. 25DOWNING AND MORTALITY .......................................................................................................... 25

LIMITATIONS OF THE DATA AND ANALYSES .............................................................................. 26RESULTS.................................................................................................................................................... 27

RADAR..................................................................................................................................................... 27TARGET IDENTIFICATION ............................................................................................................... 28MOVEMENT RATES ........................................................................................................................... 30FLIGHT VELOCITY............................................................................................................................. 42FLIGHT DIRECTION ........................................................................................................................... 45FLIGHT BEHAVIOR ............................................................................................................................ 47ISLAND-PASSING SUCCESS ............................................................................................................. 56ISLAND-PASSING DISTANCE........................................................................................................... 59SPATIAL DISTRIBUTION................................................................................................................... 63HIGH-RESOLUTION VARIATIONS IN SPATIAL DISTRIBUTION............................................... 66

VISUAL .................................................................................................................................................... 80FLOCK SIZE ......................................................................................................................................... 80FLIGHT ALTITUDE ............................................................................................................................. 80FLIGHT BEHAVIOR ............................................................................................................................ 89ISLAND-PASSING SUCCESS ............................................................................................................. 89ISLAND-PASSING BEHAVIOR.......................................................................................................... 89EFFECTS OF DISTANCE TO ISLAND ON RESPONSES .............................................................. 108

DOWNING AND MORTALITY........................................................................................................... 108DISCUSSION............................................................................................................................................ 112

“EIDERS” ............................................................................................................................................... 112MOVEMENT RATES ......................................................................................................................... 112FLIGHT VELOCITY........................................................................................................................... 113FLIGHT DIRECTION ......................................................................................................................... 113FLIGHT BEHAVIOR .......................................................................................................................... 114ISLAND-PASSING SUCCESS ........................................................................................................... 115ISLAND-PASSING DISTANCE......................................................................................................... 115SPATIAL DISTRIBUTION................................................................................................................. 115HIGH-RESOLUTION VARIATIONS IN SPATIAL DISTRIBUTION............................................. 116VISUAL FLIGHT CHARACTERISTICS........................................................................................... 116DOWNING AND MORTALITY ........................................................................................................ 116

“NON-EIDERS” ..................................................................................................................................... 117MOVEMENT RATES ......................................................................................................................... 117

Migration at Northstar Island vi ABR Final Report

FLIGHT VELOCITY........................................................................................................................... 118FLIGHT DIRECTION ......................................................................................................................... 118FLIGHT BEHAVIOR .......................................................................................................................... 118ISLAND-PASSING SUCCESS ........................................................................................................... 119ISLAND-PASSING DISTANCE......................................................................................................... 119SPATIAL DISTRIBUTION................................................................................................................. 119VISUAL FLIGHT CHARACTERISTICS........................................................................................... 119DOWNING AND MORTALITY ........................................................................................................ 120

EFFECTS OF LIGHTS........................................................................................................................... 120“EIDERS” ............................................................................................................................................ 120“NON-EIDER” SPECIES .................................................................................................................... 123CONCLUSIONS AND RECOMMENDATIONS............................................................................... 123

LITERATURE CITED.............................................................................................................................. 125

LIST OF FIGURES

Figure 1. Location of Northstar Island, northern Alaska, and radar sampling site. ................................. 4

Figure 2. Aerial photo of Northstar Island, northern Alaska; view is from the northwest, looking toward the southeast. ................................................................................................................ 5

Figure 3. Strobe light used in anti-collision lighting system at Northstar Island, northern Alaska, 2001–2003. ............................................................................................................................... 5

Figure 4. Radar screen at Barrow, northern Alaska, in August 1997, showing successive echoes plotted from flocks of migrating eiders. ................................................................................. 16

Figure 5. Radar screen at Northstar Island, northern Alaska, in September 2001, showing echoes of numerous icebergs around the island...................................................................................... 16

Figure 6. Examples of straight-line, erratic, and circling flightlines on ornithological radar near Northstar Island, northern Alaska........................................................................................... 17

Figure 7. Movement rates of “eider” targets on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, by date. ........................................................................................... 30

Figure 8. Movement rates of “non-eider” targets on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, by date. ............................................................................. 31

Figure 9. Frequencies of flight directions of “eider” targets on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004 combined, by 10× categories. ................................ 45

Figure 10. Frequencies of flight directions of “non-eider” targets on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004 combined, by 10× categories. ................ 45

Figure 11. Percentage of “eider” targets on ornithological radar exhibiting non-directional flight behavior near Northstar Island, northern Alaska, fall 2001–2004 combined, by moon phase and moon visibility. ...................................................................................................... 54

Figure 12. Percentage of “non-eider” targets on ornithological radar exhibiting non-directional flight behavior near Northstar Island, northern Alaska, fall 2001–2004 combined, by moon phase and moon visibility. ...................................................................................................... 55

Figure 13. Statistical power to detect a specified change in mean passing distance of “eiders” on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004 combined. ...... 63

Figure 14. Movement density of “eiders” on ornithological radar near Northstar Island, northern Alaska, during a period with ice present, by anti-collision lighting setting. .......................... 64

ABR Final Report vii Migration at Northstar Island

Figure 15. Movement density of “eiders” on ornithological radar near Northstar Island, northern Alaska, during periods with ice absent, by anti-collision lighting setting.............................. 65

Figure 16. Net difference in movement density of “eiders” on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, between periods and anti-collision lighting settings. ................................................................................................................................... 67

Figure 17. Regression of net difference between anti-collision lighting settings in movement density of “eiders” near Northstar Island, by period with ice present and with ice absent, by distance of the grid cell from the island.................................................................................. 68

Figure 18. Movement density of “non-eiders” on ornithological radar near Northstar Island, northern Alaska, during a period with ice present, by anti-collision lighting setting. .......................... 69

Figure 19. Movement density of “non-eiders” on ornithological radar near Northstar Island, northern Alaska, during periods with ice absent, by anti-collision lighting setting.............................. 70

Figure 20. Net difference in movement density of “non-eiders” on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, between periods and anti-collision lighting settings. ................................................................................................................................... 71

Figure 21. Regression of net difference between anti-collision lighting settings in movement density of “non-eiders” near Northstar Island, by period with ice present and with ice absent, by distance of the grid cell from the island............................................................................. 72

Figure 22. Mean numbers of “eider” trackline vertices and mean angle changed for all vertices near Northstar Island, fall 2001–2004 combined, by distance from the island and anti-collision lightning setting. ..................................................................................................................... 73

Figure 23. Mean numbers of “eider” trackline vertices ≥5×/km of line within that distance zone for all lines near Northstar Island, northern Alaska, fall 2001–2004 combined, by distance from the island and anti-collision lighting setting. ................................................................. 74

Figure 24. Number of original and actual “eider” tracklines passing within each distance zone near Northstar Island, northern Alaska, fall 2001–2004 combined, by distance from the island and anti-collision lighting setting. ......................................................................................... 78

Figure 25. Mean changes in nearest distances of “eider” flightlines near Northstar Island, northern Alaska, fall 2001–2004 combined, by distance from the island and anti-collision lighting setting...................................................................................................................................... 79

Figure 26. Eider migration corridor in the Alaska Beaufort Sea in the 1970s, as determined by Divoky. ................................................................................................................................. 124

LIST OF TABLES

Table 1. Dates and times of sampling and sample sizes during radar and visual sampling for bird movements near Northstar Island, northern Alaska, fall 2001 ................................................. 7

Table 2. Dates and times of sampling and sample sizes during radar and visual sampling for bird movements near Northstar Island, northern Alaska, fall 2002 ................................................. 9

Table 3. Dates and times of sampling and sample sizes during radar and visual sampling for bird movements near Northstar Island, northern Alaska, fall 2003 ............................................... 11

Table 4. Dates and times of sampling and sample sizes during radar and visual sampling for bird movements near Northstar Island, northern Alaska, fall 2004 ............................................... 13

Migration at Northstar Island viii ABR Final Report

Table 5. Number and percentage of radar targets seen near Northstar Island, fall 2001–2004, that were eiderlike and non-eiderlike in flight characteristics on radar, mean velocity, and percentage exhibiting straight-line flight, by species-group................................................... 29

Table 6. Movement rates (targets/h) of “eiders” and “non-eiders” migrating near Northstar Island, northern Alaska, fall 2001–2004, by species-group, period, factor, anti-collision lighting setting, and period................................................................................................................... 32

Table 7. Significance of factors affecting movement rates, velocity, flight behavior, passing success, and passing distance of “eiders” and “non-eiders” migrating near Northstar Island, northern Alaska, fall 2001–2004................................................................................. 33

Table 8. Sum of Akaike Weights for the model parameters in candidate models for each response variable ................................................................................................................................... 36

Table 9. Model-weighted parameter estimates for factors affecting movement rates, velocity, flight behavior, passing distance, and passing success of “eiders” and “non-eiders” migrating near Northstar Island, northern Alaska, fall 2001–2004 ........................................ 37

Table 10. Mean flight velocity of birds seen on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, by species-group/period, factor, and anti-collision lighting setting...................................................................................................................................... 43

Table 11. Flight directions of radar targets seen near Northstar Island, northern Alaska, fall 2001–2004, by species-group, factor, anti-collision lighting setting, and period................... 46

Table 12. Significance of factors affecting flight directions of “eiders” migrating near Northstar Island, northern Alaska, fall 2001–2004................................................................................. 47

Table 13. Flight directions of radar targets seen near Northstar Island, northern Alaska, fall 2001–2004 combined, by species-group, factor, and anti-collision lighting setting .............. 48

Table 14. Frequencies and percentages of general flight behaviors of birds seen on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, by species-group/period, factor, and anti-collision lighting setting ................................................................................ 49

Table 15. Frequencies and percentages of island-passing success of birds seen on ornithological radar near Northstar Island, northern Alaska, fall 2001–2003, by species-group/period, factor, and anti-collision lighting setting ................................................................................ 57

Table 16. Island-passing distances of “eiders” seen on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, by species-group/period, factor, and anti-collision lighting setting ........................................................................................................................ 60

Table 17. Characteristics of records of “eider” and “non-eider” targets on ornithological radar that passed over Northstar Island, northern Alaska, fall 2001–2004, by species-group ............... 62

Table 18. Significance of factors affecting the distance of vertices ≥5°, the mean angular change for vertices ≥5°, and the difference between original and actual nearest-passing distances of "eiders" migrating near Northstar Island, fall 2001–2004. ..................................................... 75

Table 19. Sum of Akaike Weights for the model parameters in candidate models for each response variable. .................................................................................................................................. 76

Table 20. Model-weighted parameter estimates for factors affecting the distance of vectors ≥5°, the mean angular change for vertices ≥5°, and the difference between original and actual passing distances of “eiders” migrating near Northstar Island, fall 2001–2004 ................................. 76

ABR Final Report ix Migration at Northstar Island

Table 21. Mean flock size of birds seen visually near Northstar Island, northern Alaska, fall 2001–2004, by species-group, factor, and anti-collision lighting setting............................... 81

Table 22. Mean flight altitude of birds seen visually near Northstar Island, northern Alaska, fall 2001–2003, by species-group, factor, and anticollision lighting setting ................................ 85

Table 23. Frequencies and percentages of general flight behaviors of birds seen visually near Northstar Island, northern Alaska, fall 2001–2004, by species-group, factor, and anti-collision lighting setting. ................................................................................................. 90

Table 24. Frequencies and percentages of passing success of birds seen visually near Northstar Island, northern Alaska, fall 2001–2004, by species-group, factor, and anti-collision lighting setting ........................................................................................................................ 97

Table 25. Island-passing behavior of birds seen visually near Northstar Island, northern Alaska, fall 2001–2003, by species-group, factor, and anti-collision lighting setting ...................... 101

Table 26. Frequencies and percentages of passing success and behavioral responses of birds passing Northstar Island, northern Alaska, fall 2001–2003, by closest distance to island..................................................................................................................................... 109

Table 27. Environmental and other characteristics associated with the downing of Common and King eiders and Long-tailed Ducks at Northstar Island, 2001–2004 ................................... 110

Table 28. Patterns of downing of Common and King eiders and Long-tailed Ducks at Northstar Island, 2001–2004, by night and by individual bird............................................................. 111

Table 29. Summary of effects of anti-collision lights on migratory movements and behavior of “eiders” and “non-eiders,” 2001–2004 ................................................................................. 121

LIST OF APPENDICES

Appendix 1. Statistical review of the Draft Report and the response to that review. ........................ 129

Appendix 2. Movement density of “eiders” on ornithological radar near Northstar Island, northern Alaska, fall 2002, by anti-collision lighting setting. ..................................... 132



Appendix 5. Movement density of “non-eiders” on ornithological radar near Northstar Island, northern Alaska, fall 2002, by anti-collision lighting setting. ..................................... 135



Appendix 8. Numbers of flocks and numbers of birds recorded during visual sampling near Northstar Island, northern Alaska, fall 2001–2004. ..................................................... 138

Appendix 9. Memorandum about avian vision at night. ................................................................... 139

Migration at Northstar Island x ABR Final Report

ACKNOWLEDGMENTS

This study was funded by BP Exploration (Alaska) Inc.; however, the conclusions are ours and do notnecessarily represent those of BP Exploration. We thank Dr. Bill Streever of BP Exploration (Alaska) Inc.,and Philip D. Martin and Theodore Swem of the U.S. Fish and Wildlife Service for discussions about thisstudy and for help with designing some of the analyses. We thank Dave Trudgen, Allison Erickson, andWilson Cullor of Oasis Environmental, Inc., for logistical help with this study. We thank NorthstarConstruction Engineers/Island Engineers Jeff Huey, Earl Beverly, and David McGuire; Northstar HSEAdvisors Pam Pope and Julie Arin; and Northstar Environmental Specialists Joe Serra and Hunter Ervinfor on-island help. We particularly thank the folks in Operations on Northstar Island for help with the lightsand radios. Dr. Edward Murphy of the University of Alaska—Fairbanks provided statistical advice. AtABR, Lauren Attanas, Jill Keen, Michael Knoche, Tim Obritschkewitsch. Julie Parrett, John Shook, AliceStickney, and Mark Vekasy helped with the data collection; Doris Armijo helped with logistics; JonathanPlissner and Richard Blaha helped with the Oriana figures; Allison Zusi-Cobb, Dorte Dissing, and TaiGraham made the GIS figures; Tom DeLong provided fiscal management, and Pam Odom produced thereport. This report has been improved by the review of Brian Cooper of ABR, Inc., and Bill Streever of BPExploration (Alaska), Inc.

Introduction

ABR Final Report 1 Migration at Northstar Island

INTRODUCTION

During migration, large numbers of Common(Somateria mollissima) and King (S. spectabilis)eiders pass through the Beaufort and Chukchi seasduring their passage between breeding andwintering grounds (Frontispiece; Bailey 1948,Thompson and Person 1963, Johnson 1971,Richardson and Johnson 1981, Johnson andRichardson 1982, Woodby and Divoky 1982,Divoky 1984a; Suydam et al. 1997, 2000b;Quakenbush and Suydam 2004). In addition, small(but unknown) numbers of Spectacled (S. fischeri)and Steller’s (Polysticta stelleri) eiders, both ofwhich currently are protected by the EndangeredSpecies Act, move through the central and westernBeaufort Sea (USFWS 1996, 2002; TERA 1999,Suydam et al. 2000b, Quakenbush et al. 2002);probably less than 1% of the world populations ofeach species migrate across the Beaufort Sea(Petersen et al. 1999, Larned et al. 2005; M.Petersen, USGS—BRD, Anchorage, AK, in litt.).Common and King eider populations in theBeaufort Sea also are declining (Suydam et al.2000a), so any additional sources of mortality forany of the four species of eiders is of concern.

Eiders are thought to be susceptible tocollision with human-made structures because theyfly low over the water while migrating (Thompsonand Person 1963, Johnson and Richardson 1982;Day et al. 2001, 2004b), fly rapidly whilemigrating (Day et al. 2004b), and become attractedto bright lights at night (Dick and Donaldson 1978;John J. Burns, Alaska Department of Fish andGame [retired], Fairbanks, AK, pers. comm.;Patricia Kaminsky, University of Alaska/SewardMarine Station, Seward, AK, pers. comm; Lori T.Quakenbush, Alaska Department of Fish andGame, Fairbanks, AK, pers. comm.; John L. Sease,National Marine Mammal Laboratory, Seattle, WA,pers. comm.). They also have been reported simplyflying into the ground during periods of poorvisibility (Mallory et al. 2001), suggesting limitedability to avoid collision in poor visibility.Waterfowl in general commonly collide withman-made structures (e.g., Anderson 1978, Dickand Donaldson 1978), and many species ofmigrating birds are attracted to lights, especiallyduring periods of inclement weather at night (e.g.,Kramer 1948, Cochran and Graber 1958, Telfer et

al. 1987, Jehl 1993), sometimes colliding withstructures at the same time. Numerous otherinstances of birds coming aboard ocean-goingboats at night have been recorded, even when theweather was not stormy (Day, pers. obs.; JohnBurns, pers. comm.), indicating that light attractionis not an isolated phenomenon. Lights clearlyexacerbate attraction and collision problems duringperiods of misty or rainy weather, and eiders, inparticular, have been found to be attracted to lightsduring such weather (Dick and Donaldson 1978;John Burns, pers. comm.; Patricia Kaminsky, pers.comm.; Lori Quakenbush, pers. comm.; JohnSease, pers. comm.; James Short, ARCO Alaska,Inc., Anchorage, AK, pers. comm.). For example,migrating shorebirds have been attracted to, andcollided with, lighted boats on the North Pacificduring storms (Day, pers. obs.), and both eiders andother seaducks collided with the Prudhoe BaySaltwater Treatment Plant during its first fallseason of operation in the early 1980s (JamesShort, pers. comm.), presumably because of lightattraction.

Ornithological radar has been an importanttool in ornithological research for five decades(Eastwood 1967) because it can overcome some ofthe limitations of visual observation techniqueswhen birds travel during periods of low light orrestricted visibility (e.g., in fog, at night). Radarhas been used successfully in studies ofnight-flying geese, cranes, and waterfowl (Cooperand Mabee 1990, Cooper et al. 1991; Dirksen et al.1997; Day et al. 1998, 2000a, 2000b, 2001, 2004a;Tulp et al. 1999) and on endangered, nocturnalseabirds and bats (Day and Cooper 1995, Hamer etal. 1995; Cooper and Day 1998, 2003; Cooper etal. 2001, Day et al. 2003a, 2003b). Radar also canbe used to collect data over large areas that cannotbe covered by a single visual observer (e.g.,Richardson and Johnson 1981, Johnson andRichardson 1982, Hamer et al. 1995) and to helpvisual observers detect and locate flying birds thatotherwise would be missed by visual observers(e.g., Kerlinger and Gauthreaux 1984, 1985;Cooper and Ritchie 1995). This is not to say thatornithological radar always detects all birds in anarea, for it also is a sampling tool with its own setof biases and limitations. Radar and visualtechniques, however, provide different types of

Introduction

Migration at Northstar Island 2 ABR Final Report

data that are complementary in studies of birdmigration and movements.

BP Exploration (Alaska) Inc. (hereafter, BP),has constructed an offshore island in the BeaufortSea as part of the development of the NorthstarOilfield. Northstar Island lies ~4 mi (~6.5 km)north of Egg Island and ~8 mi (~13 km) northwestof Prudhoe Bay’s West Dock. Because eidersappear to have a broad-front migration in this areaand do not concentrate near the coast (Johnson andRichardson 1982), substantial numbers of themmay pass close to this new structure duringmigration, thus putting themselves in jeopardy ofcolliding with the island or structures on it. At thistime, however, knowledge of eider migration (andthe migration of other birds) and its variability inthe immediate vicinity of Northstar Island isextremely limited; the results of the first threeyears of this study (Day et al. 2004a) represent thefirst in-depth examination of eider migration in thislocation.

BUSINESS RATIONALE

The Business Rationale for this study istwofold. First, BP as a global company has a statedaspiration of doing no damage to the environment(see website www.bp.com/environ_social/environment/index.asp). As a way of achieving thisgoal, BP has instituted a Biodiversity Action Planfor minimizing impacts of their operations on theNorth Slope (BP 2003). Second, BP was mandatedto conduct a study of eider migration and collisionpotential at Northstar Island and, in effect, todevelop an anti-collision lighting system to helpmigrating birds avoid colliding with the island.

As part of its construction permit, BP wasrequired to conduct a study of eider migration andcollision at Northstar Island. U.S. Army Corps ofEngineers (USACE) Permit #N-950372 (dated3 May 1999) included two stipulations involvingmigrating eiders. Special Condition #1 stipulatedthat BP could not take any endangered species andthat BP must implement “reasonable and prudentmeasures” stipulated in the Biological Opinion (seebelow) to avoid taking any endangered species atNorthstar Island. Special Condition #11 stipulatedthat BP must conduct a post-constructionmonitoring study to evaluate whether the islandresulted in the injury or mortality of birds,

especially Spectacled Eiders. That SpecialCondition also stipulated that BP should conductthis study for a minimum of 5 years and that thesampling protocol should be worked out betweenBP, the USACE, and the U.S. Fish and WildlifeService (USFWS).

In addition to the USACE Permit stipulations,the USFWS published a Final Biological Opinion(dated 10 March 1999) for the Northstar Project.That Biological Opinion stipulated several“reasonable and prudent” measures that should betaken by BP to minimize the incidental “take” ofSpectacled and Steller’s eiders at Northstar Island.These measures were: (1) structures should bepainted and/or lighted (to increase overall visibilityto birds), lighting should be directed inward towardthe island (to reduce their attractiveness to birds),and crane booms should be lowered when not inuse (to reduce the chances of collision); and(2) oilspill mitigation plans that minimized effectson eiders should be developed and implemented.Hence, one major source of incidental take wasconsidered to involve the collision of migratingeiders with island-associated structures duringperiods of darkness or inclement weather.

These concerns about the collision-causedmortality of eiders at Northstar Island were notentirely unfounded, although the actual mortalityso far has been small. For example, 2 CommonEiders, 1 Long-tailed Duck, and 2 unidentifiedwaterfowl were killed at Northstar Island in fall2000; 8 Common, 4 King, and 1 unidentified eiderand 8 Long-tailed Ducks (Clangula hyemalis) werekilled or injured there in fall 2001; no eiders and 3Long-tailed Ducks were killed there in fall 2002; 4Common Eiders and 2 Long-tailed Ducks werekilled there in fall 2003; and 3 Common Eiders and1 unidentified eider were killed there in fall 2004.This total of 20 dead eiders for the 4 yearscombined represent ~0.004% of the estimated450,000+ eiders migrating through the AlaskaBeaufort Sea each fall (Suydam et al. 2000a,2000b). Both the USFWS and BP, however, areconcerned about the possibility of a kill of a flockof eiders on an occasional night that has acombination of weather factors that lead tocollisions. In other words, the desire is to keepeider mortality rates very small and, preferably, atzero.

Study Area


In response to the permit stipulations and theBiological Opinion, BP met the stipulations andpainted the buildings on the island a light brown,directed shielded lights inward (toward the interiorof the island), and installed a set of anti-collisionlights on the perimeter of the island. BP also isfunding this study to investigate movements ofmigrating eiders and their responses to these birdanti-collision lights at Northstar Island. The centralquestions of this study are: (1) are migrating eidersand other birds that are passing near NorthstarIsland altering any aspect of their behavior (i.e.,flight velocity, flight direction, flight behavior,island-passing success or distance, spatialdistribution) that influences their chances ofcolliding with the island (i.e., are they beingattracted to or repelled by the island, or are they notchanging behavior as they approach the island?),and (2) does the anti-collision lighting system onthe island attract or repel migrating eiders andother birds (i.e., do they significantly alter anyaspect of behavior in response to the lights?)?

OBJECTIVES

Given the Business Rationale and the permit stipulations, the objectives of this study were:

• to use ornithological radar and visual sampling to monitor the migration and behavior of birds migrating past Northstar Island;

• to determine whether migrating birds detect Northstar Island and, if so, what their response is and the distance at which the response occurs;

• to determine whether the flight behavior of migrating birds changes during periods of lowered visibility, thereby affecting the chances of collision with Northstar Island or its structures; and

• to determine the effects of anti-collision lighting on the movements and responses of birds migrating near Northstar Island.

To complete these objectives, we conducted aradar and visual study of bird migration nearNorthstar Island, Alaska, in 2001–2004. Wedescribe both movement patterns and responses ofbirds to the island and its lights; although theemphasis of the study was on eiders, we also

present information on other species. This reportsummarizes the results of all four years of thestudies and is presented as a Final Report of ouractivities and findings.1

STUDY AREA

Northstar Island is a small, human-madegravel island that lies in the Beaufort Sea ~4 mi(~6.5 km) north of Egg Island, in the Jones Islands(Figs. 1 and 2). Northstar also lies ~6 mi (~10 km)north of the coastline of mainland Alaska and islocated northwest of Prudhoe Bay and almost duewest of the Midway Islands and Cross Island. Onthe nearby mainland are an assortment of facilitiesfor the North Slope oilfields. Northstar Islandpumps ~70,000 bbl of crude oil/day, or ~1% of USdaily production.

Northstar Island is ~500 ft (~150 m) long inan east–west direction, ~700 ft (~210 m) wide in anorth–south direction, and ~25 ft (~8 m) high at theseawall; it is 8.77 ac (3.55 ha) in area. The islanditself consists of a sheetpile wall, made of steelplates pounded vertically into the seabed andgravel, that encloses the bulk of the island. Thatwall, in turn, is surrounded by gravel covered withsteel-reinforced concrete “beach armor” thatextends the perimeter of the island outward. Thepipe rack and two electrical-control buildings riseabove the sheetpile wall on the eastern side of theisland, the pipe rack runs above the northern sideof the island, the highest structure on the island(the Processing Facility) rises to ~125 ft (~40 m)above sea level along the northwestern part of theisland, and the smaller Permanent Living Quarters,which is ~50 ft (~15 m) high, is located in thesouthwestern part of the island. A drilling rig~120 ft (~35 m) high is present on the eastern sideof the island. During August–September 2001, twolarge construction cranes also were present on theisland; that number dropped to one crane in2002–2004.

The anti-collision lighting system consists ofa series of 14 white strobe lights mounted on mastsalong the perimeter of the sheetpile wall, with 4 onthe eastern side, 3 on the northern side, 5 on thewestern side, and 2 on the southern side of theisland (Fig. 3). These lights are mounted ~45 ft

1. BP Exploration (Alaska) Inc., has agreed to support the publication of this research.

Study Area

Migration at N

orthstar Island4

AB

R F

inal Report

Figure 1. Location of Northstar Island, northern Alaska, and radar sampling site.

Sa

ga

va

nir

kto

kR

iver

Kup

aru

kR

iver

B e a u f o r t S e a

W

Prudhoe

Bay

Deadhorse

Midway Islands Cross Island

Endicott Development

OliktokPoint

Egg Island

Stump Island

Pingok Island

Long Island

Colv

ille

Riv

er

J o n e s I s l a n d s

151°0'0"W

151°0'0"W

149°0'0"W

149°0'0"W

147°0'0"W

147°0'0"W

70

°0'0

"N

70

°0'0

"N

70

°30

'0"N

70

°30

'0"N

71

°0'0

"N71°0'0"N

410 0 10 20 30

Miles

10 0 10 20 30

Kilometers

Point Belcher

Barrow

Cape Halkett

Colvil le River

Enlarged Area

Location Map

O I L F I E L D S

Radar Sampling

Site

ProcessingFacility

PermanentLiving

Quarters Northstar Island

0 100 200 300 Feet

ABR file: Study_Area_03_153.mxd; 3 March 2004

Study Area


Figure 2. Aerial photo of Northstar Island, northern Alaska; view is from the northwest, looking toward the southeast. Photo courtesy of BP Exploration (Alaska) Inc.

Figure 3. Strobe light (arrow) used in anti-collision lighting system at Northstar Island, northern Alaska, 2001–2003. Photo by R. H. Day.

Methods


(~14 m) above the ocean’s surface. They are set tofire at the rate of 40 flashes/min, and those withinone side of the island fire asynchronously, bothwithin that group and from those on other sides ofthe island. These Honeywell Flashguard 2000Bstrobe lights emit white light (i.e., all wavelengths)from a daytime strobe light (20,000 candela) and anighttime flashing light (2,000 candela); aphotocell controls the switching between the twomodes. This type of light is described at theHoneywell website (www.airportsystems.honeywell.com/ oblighting/products/fg2000b) as amedium-intensity obstacle light for marking tallstructures for aviation safety (FAA type L-865 orL-866).

A variety of birds, including various eiderspecies, nests on nearby islands (Schamel 1977,Gavin 1977) and the nearby mainland (Bergman etal. 1977) and migrates through the surroundingarea (Flock 1973, Schamel 1978, Richardson andJohnson 1981, Johnson and Richardson 1982;Divoky 1984a, 1984b; Fischer and Larned 2004).Additional information on birds in this region ispresented in Troy (1985), Johnson and Herter(1989), and Hohenberger et al. (1994).

METHODS

We collected data on the movements,behavior, and flight altitudes of migrating eidersand other birds during 20 nights from late Augustto mid-September 2001 (Table 1), on 19 nightsfrom mid- to late September 2002 (Table 2), on 23nights from late August to mid-September 2003(Table 3), and on 18 nights from early tomid-October 2004 (Table 4). We sampled for~7–11 h/day with both ornithological radar andvisual equipment (both 10× binoculars andnight-vision equipment with a 5× eyepiece). Anight’s sampling usually began at ~2100 (in 2001and 2002) or ~1800 (in 2003 and 2004) andfinished at 0500, with a break usually around0000–0100. For the purposes of recording data, acalendar day began at 0700 and ended at 0659 thefollowing morning; that way, an evening and thatnight were classified during analyses as occurringon the same day.

DATA COLLECTION

When possible, we collected radar and visualdata concurrently, so that the radar operator couldhelp the visual observer locate birds foridentification and data collection and so the visualobserver could provide information to the radaroperator on the identity of individual targets.Although we always attempted to sampleconcurrently, heavy rain and/or fog prevented usfrom collecting data during 2 radar samplingsessions and 15 visual sampling sessions in 2001(Table 1). Rain/snow/sleet/fog, heavy sea clutterfrom high winds, and/or the presence of PolarBears (Ursus arctos) prevented us from collectingdata during 137 radar sampling sessions and 17visual sampling sessions in 2002 (Table 2). Heavysea clutter from high winds (a persistent problem in2003), rain/snow/sleet, and the lack of boat accessto the island prevented us from collecting dataduring ~190 radar sampling sessions and no visualsampling sessions in 2003 (Table 3). High windsprevented installation of the radar antenna forseveral days and a crew change-out prevented usfrom collecting data during 201 radar samplingsessions and 19 visual sampling sessions in 2004(Table 4).

We attempted to collect data during 25-minsessions for both sampling methods, then used5-min breaks between sessions to collect weatherdata and to allow observers to switch samplingmodes. Actual lengths of sampling sessions were10–25 min for radar data and 5–25 min for visualdata, with nearly all sessions of both types being 25minutes long. In 2001, we collected 114.1 h of dataon 187 radar sampling sessions and 108.8 h of dataon 262 visual sampling sessions. In 2002, wecollected 55.1 h of data on 133 radar samplingsessions and 110.9 h of data on 267 visualsampling sessions. In 2003, we collected 51.8 h ofdata on 126 radar sampling sessions and 171.1 h ofdata on 414 visual sampling sessions. In 2004, wecollected 77.2 h of data on 187 radar samplingsessions and 134.2 h of data on 351 visualsampling sessions.

We recorded the following data at thebeginning of each sampling session:

Methods


Table 1. Dates and times of sampling and sample sizes (n sampling sessions) during radar and visual sampling for bird movements near Northstar Island, northern Alaska, fall 2001. Data for a particular date were collected between 0700 in the morning and 0659 the next morning.

Time of sampling (n)

Date Radar Visual Comments 22–23 AU – (0) – (0) R. Day, J. Rose, and M. Knoche arrive; set up and

test radar. 24 AU 2230–0200,

0300–0500 (11) 2230–0200,

0300–0500 (11) Foggy; winds moderate; bird lights inoperative.

25 AU 2100–0200, 0300–0500 (14)

2100–0200, 0300–0500 (14)

Foggy; winds moderate; bird lights inoperative; some sea clutter.

26 AU 2130–0130, 0230–0500 (13)

2130–0130, 0230–0500 (13)

R. Day departs; foggy/drizzly; winds light–moderate; sea clutter reduced; bird lights operative for first time.

27 AU 2130–0500 (15) 2130–0500 (15) Scattered fog/drizzle, stopping and clearing in middle of night; winds light; seas calm.

28 AU 2100–0500 (16) 2100–0500 (16) Heavy fog; high winds with heavy sea clutter to 2 km.

29 AU 2130–0500 (15) 2130–0030, 0400–0500 (8)

Fog so heavy that no visual sampling 0030–0400; winds light.

30 AU 2100–0500 (16) 2100–0000, 0400–0500 (8)

Fog so heavy that no visual sampling 0000–0400; winds light–moderate.

31 AU 2130–0200, 0300–0500 (13)

2130–0200, 0300–0500 (13)

J. Rose and M. Knoche depart, R. Day and J. Keen arrive; Concrete Island Drilling Structure moved out today; foggy; much scattered ice; winds/sea clutter light; could control lights for only eastern side of island 2135–0200; lost 15 min of radar sampling time (0315–0330) due to loss of electrical power.

1 SE 2100–0130, 0230–0500 (14)

2100–0130, 0230–0500 (14)

Scattered ice; occasional fog; winds light.

2 SE 2100–0200, 0300–0500 (14)

2100–0200, 0300–0500 (14)

Much ice; winds light; some circling of island by shorebirds in middle of night.

3 SE 2100–0200, 0300–0500 (14)

2100–0200, 0300–0500 (14)

Foggy, but visual observers could see floodlights of crew boats out to 1.3 km; winds high; heavy sea clutter to 2.1 km; visit from Allison Erickson.

4 SE 2100–0200, 0300–0500 (14)

2100–0200, 0300–0500 (14)

Heavy sea clutter to 2.3 km early, then disappearing as winds/seas calm; foggy.

5 SE 2100–0200, 0300–0500 (14)

2100–0200, 0300–0500 (14)

Winds light; at 2200–0200, portable floodlights set up on eastern shoreline for repairing beach armor; extensive circling of island by waterfowl and shorebirds.

6 SE 2100–0200, 0300–0500 (14)

2100–0200, 0300–0500 (14)

R. Day departs, A. Stickney arrives; winds light; great visibility; turned off lights for helicopter visit at 2240; photographer visits at 2100–2200.

7 SE 2100–0000, 0100–0200,

0300–0500 (12)a

2100–0200, 0300–0500 (14)

Lost 1 h radar sampling time to rain; winds light; turned off lights for crew boat during arrival at 2325; several floodlights on eastern shore make visual sampling difficult.

Methods


Table 1. Continued.


Date Radar Visual Comments 8 SE 2100–0200,

0300–0500 (14) 2100–0200,

0300–0500 (14) Much ice to north; some sea clutter early, disappearing by ~0000 as winds/seas decrease.

9 SE 2100–0200, 0300–0500 (14)

2100–0200, 0300–0500 (14)

Many large ice floes near island; cloud cover decreased during night; winds 10–15 mi/h, but no sea clutter.

10 SE 2100–0200, 0300–0500 (14)

2100–0200, 0300–0500 (14)

Winds light; no sea clutter; broken cloud cover.

11 SE 2100–0200, 0300–0500 (14)

2100–0200, 0300–0500 (14)

Winds light; light–moderate sea clutter; lost 5 min of radar and visual sampling at 2235–2240.

12 SE 1800–2300 (10) 1800–2300 (10) Heavy fog; winds light; much ice. 13–15 SE – (0) – (0) A. Stickney and J. Keen depart.

a Lost sampling time because of rain or snow.

Methods




Date Radar Visual Comments 9–10 SE – (0) – (0) R. Day, A. Stickney, and J. Shook arrive; set

up and test radar. 11 SE 2200–0100,

0200–0500 (12)a 2030–0100,

0200–0500 (11) No ice; winds moderate, but sea clutter out to 1.8 km at times; scattered showers.

12 SE 2000–0100, 0200–0400 (14)

2000–0100, 0200–0500 (16)

Fog and rain in latter half of night; sea clutter builds out to 2.8 km, resulting in cancellation of final two radar sampling sessions.

13 SE 2030–0100, 0200–0530 (16)

2030–0100, 0200–0530 (16)

Sea clutter out to 1.6 km at times.

14 SE – (0) 2000–0030, 0200–0500 (15)

High winds and heavy sea clutter to edge of radar screen all night, resulting in cancellation of all radar sampling; fog in latter half of night.

15 SE 2000–2045, 2200–0100,

0200–0500 (14)

2000–0100, 0200–0500 (16)

Polar Bear attempts to land on island, so radar sampling cancelled 2045–2200; sea clutter out to 500–800 m; fog all night.

16 SE 2100–2130, 2300–0000,

0100–0200 (5)

2100–0000, 0100–0200 (8)

A. Stickney departs, J. Rose arrives; winds moderate, but sea clutter out to >1.4 km at times, so re-shim radar antenna; fog in first half of night; Polar Bears attempt to land on island repeatedly, so all sampling finally cancelled at 0200.

17 SE 2000–0000, 0100–0430 (15)a

2000–0000, 0100–0500 (16)

Rain off and on all night, becoming so heavy that radar sampling cancelled at 0430.

18 SE – (0)a 2000–2230 (5)a Heavy rain/sleet/snow, high winds, and heavy sea clutter all night, resulting in cancellation of all radar sampling and limitation of visual sampling.

19 SE 0400–0500 (2) 2000–0100, 0200–0500 (16)

J. Shook departs, R. Day arrives; sea clutter to edge of screen until 0400; heavy gas flaring 2330–0500; mixed rain/sleet/snow all night, but not so severe that it hinders sampling.

20 SE 1830–2230, 2300–0000,

0100–0500 (19)

1830–0000, 0100–0500 (19)

Visit from Pam Pope; heavy gas flaring 1845–0300, resulting in attraction of birds to island; scattered snow flurries.

21 SE 2000–0000, 0100–0500 (16)

2000–0000, 0100–0500 (16)

Sea clutter low–moderate at times, but still able to sample.

22 SE – (0) 2030–0000, 0130–0500 (14)

R. Day departs, J. Parrett arrives; high winds and heavy sea clutter to edge of screen all night, resulting in cancellation of radar sampling.

Methods


Table 2. Continued.


Date Radar Visual Comments 23 SE – (0) 1830–2200,

2300–0300, 0400–0500 (17)

High winds and heavy sea clutter to edge of screen all night, resulting in cancellation of radar sampling; seas so rough that crew boat is unable to operate; some fog late in morning.

24 SE – (0) 2030–0130, 0300–0500 (14)

High winds and heavy sea clutter to edge of screen all night, resulting in cancellation of radar sampling.

25 SE 2000–0000, 0100–0500 (16)

2100–0000, 0100–0500 (14)

Sea clutter builds out to 2.2 km at 0230, then dropping; fog most of night.

26 SE 0300–0500 (4) 2030–0030, 0130–0500 (15)

High winds and heavy sea clutter to edge of screen until 0300, resulting in cancellation of most radar sampling.

27 SE – (0)a 2000–2300, 2330–0130

0230–0500 (15)

Rain, high winds, and heavy sea clutter to edge of screen all night, resulting in cancellation of radar sampling; Polar Bear attempts to land on island 1930–2000, resulting in cancellation of first sampling period.

28 SE – (0) 2000–0000, 0100–0500 (16)

High winds and heavy sea clutter to edge of screen all night, resulting in cancellation of radar sampling; fog all night.

29 SE – (0) 2000–0000 (8) Heavy sea clutter, so no radar sampling; tear down/pack radar and other equipment 0000–0500.

30 SE – (0) – (0) J. Rose and J. Parrett depart.

a Lost sampling time because of rain or snow.

Methods




Date Radar Visual Comments 26 AU – (0) – (0) R. Day and M. Vekasy arrive; set up and test

radar 27 AU – (0) 2100–2200,

2345–0100, 0200–0500 (11)

Radar/visual training; windy with high seas and sea clutter too extensive for radar sampling all night.

28 AU 0230–0500 (5) 2030–0100 0200–0500 (15)

Foggy; windy with high seas and sea clutter too extensive for radar sampling until later in the night.

29 AU – (0) 2030–0100, 0200–0500 (15)

Mostly clear with excellent Noctron viewing conditions; windy with high seas and sea clutter too extensive for radar sampling all night; small gas-flaring event at ~2310.

30 AU 2200–0100, 0200–0500 (12)

2030–0100, 0300–0500 (13)

Foggy; high seas still causing extensive sea clutter early in evening; flaring events 2210–2250 and 0337–0400.

31 AU – (0) 2030–0100, 0200–0500 (15)

Light fog and mist; windy with high seas and sea clutter too extensive for radar sampling all night; heavy equipment and floodlights all night at southern end of beach.

1 SE – (0) – (0) No sampling possible—transportation to island cancelled.

2 SE 2000–0000, 0100–0500 (16)

2000–0000, 0100–0500 (16)

R. Day leaves, J. Rose arrives; mostly clear with light or calm winds; sea clutter not too severe most of the night.

3 SE – (0) 1830–0000, 0100–0600 (21)

Windy with high seas and sea clutter too extensive for radar sampling all night; rain showers; heavy equipment and floodlights all night near Bird Hut; gas-flaring event at 0200.

4 SE 0130–0430 (6) 1800–0000, 0100–0600 (22)

Windy with high seas and sea clutter too extensive for radar sampling much of night; rain/snow clutter off and on later that night; heavy equipment and floodlights all night near Bird Hut; gas-flaring event 0246–0255.

5 SE 1800–0000 (12) 1800–0000, 0100–0600 (22)

Clear; good radar sampling conditions early in evening, but seas building until sea clutter too severe for radar sampling later that night; heavy equipment and floodlights all night near Bird Hut.

6 SE 1815–0000, 0100–0600 (22)

1800–0000, 0100–0600 (22)

Partly cloudy and light winds; excellent radar conditions; heavy equipment and floodlights all night near Bird Hut.

7 SE 1815–2330, 0100–0530 (20)

1815–0000, 0100–0600 (22)

Winds light; sea clutter low but building throughout the night; heavy equipment and floodlights all night near Bird Hut.

Methods


Table 3. Continued.


Date Radar Visual Comments 8 SE – (0) 1815–0000,

0100–0600 (22) Sky clear; windy with high seas and sea clutter too extensive for radar sampling all night; heavy equipment and floodlights all night near Bird Hut.

9 SE – (0) 1930–0000, 0100–0600 (19)

J. Rose leaves, J. Parrett arrives; windy with high seas and sea clutter too extensive for radar sampling all night; heavy equipment and floodlights all night near Bird Hut.

10 SE – (0) 1800–0000, 0100–0600 (22)

Windy with high seas and sea clutter too extensive for radar sampling all night; snow flurries; heavy equipment and floodlights all night near Bird Hut.

11 SE – (0) 1800–0000, 0100–0600 (22)

Windy with high seas and sea clutter too extensive for radar sampling all night; snow flurries; excellent Noctron sampling conditions; heavy equipment and floodlights gone.

12 SE 0300–0330, 0400–0500 (3)

1800–0000, 0100–0300,

0330–0600 (21)

Windy with high seas and sea clutter too extensive for radar sampling most of night; snow clutter off and on.

13 SE 1830–0000, 0100–0600 (21)

1900–0000, 0100–0600 (20)

Some sea clutter, but workable; snow clutter off and on later in evening.

14 SE 1815–2200, 2230–2300 (9)

1815–0000, 0100–0600 (22)

Good weather early in evening, changing to rain and heavy sea clutter by middle of night.

15 SE – (0) 1800–0000, 0100–0600 (22)

Fog and snow flurries; severe storm with seas washing over beach up to Bird Hut; all radar sampling cancelled because of safety concerns.

16 SE – (0) 1815–0000, 0100–0600 (22)

Severe storm continues; radar sampling impossible because of extensive sea clutter; snow flurries.

17 SE – (0) 1800–0000, 0100–0600 (22)

Severe storm continues; radar sampling impossible because of extensive sea clutter; visibility decreases.

18 SE – (0) 1800–2100 (6) Severe storm continues; radar sampling impossible because of extensive sea clutter; snow flurries; pack up radar laboratory for shipping.

19 SE – (0) – (0) J. Parrett and M. Vekasy depart.

Methods




Date Radar Visual Comments 4 OC – (0) – (0) R. Day and A. Stickney arrive. 5 OC – (0) 1550–1720 (4) Set up part of radar; too windy to move radar

antenna to roof of Electrical IO Building, so unable to complete setup and test radar; overcast with rain.

6 OC – (0) 0700–0930, 1000–1200, 1300–1500,

1530–1730 0200–0500 (18)

High winds; overcast; rain and snow alternating.

7 OC – (0) 0700–0900, 0945–1100, 1115–1200, 1345–1500, 1515–

1730 (18)

High winds; overcast; snow all day.

8 OC – (0) 0700–0900, 0930–1000, 1830–2030

(10)

High winds; overcast; late at night, snow too heavy to continue visual observations. R. Day leaves, T. Obritschkewitsch arrives.

9 OC – (0) 0700–0900, 0930–1130, 1330–1730, 1830–2030, 2100–2300, 0000–0230, 0300–0500 (33)

High winds; overcast; snow flurries.

10 OC – (0) 0700–1030, 1045–1130, 1330–1410, 1515–1730, 1930–2130, 2200–2300, 0030–0230, 0300–

0500 (30)

Install antenna and test radar; however, winds so high that sea clutter too extensive for radar sampling; overcast; snow flurries.

11 OC – (0) 0700–1200, 1330–1410, 1450–1730, 1830–2030, 2100–2300, 0000–0230, 0315–0500 (35)

Winds so high that sea clutter too extensive for radar sampling; overcast; snow flurries; generator shutdown; gas flaring off and on tonight.

12 OC – (0) 0700–1200, 1330–1730, 1830–2030, 2110–2300, 0030–0230, 0300–0500

(34)

Winds so high that sea clutter too extensive for radar sampling; overcast; snow flurries.

13 OC 1830–2300, 0000–0500 (19)

0700–1200, 1330–1700 (16)

High winds, but sea clutter patchy; overcast; snow flurries.

14 OC 1840–2300, 0030–0500 (17)

1930–2300, 0030–0500 (16)

Moderate winds; overcast with rain early in evening.

15 OC 1845–2300, 0000–0430 (18)

1845–2300, 0000–0500 (18)

Moderate winds; partly cloudy to clear; no precipitation.

16 OC 1830–2300, 0000–0430 (18)

1800–2300, 0000–0430 (18)

Light winds; cloud cover decreasing throughout night; no precipitation.

17 OC 1900–2300, 0030–0530 (19)

– (0) A. Stickney leaves; winds light; partly cloudy; no precipitation.

Methods


• ordinal wind direction (10 categories)— calm, north, northeast, east, southeast, south, southwest, west, northwest, variable/erratic;

• light condition (6 categories)—daylight with or without precipitation, crepuscular (twilight) with or without precipitation, darkness with or without precipitation;

• precipitation (11 categories)—none, fog, drizzle, light rain, heavy rain, scattered showers, sleet, snow, hail, mixed rain/sleet/snow, snow flurries;

• minimal visibility (2 categories)—poor (<500 m), good (≥500 m);

• moon visible to a ground-based observer (2 categories)—no, yes; and

• anti-collision lighting setting (2 categories)—off, on.

We used the anti-collision lighting system as acontrolled experiment. We collected data with thelights off for one half-hour sampling session, thenturned them on for the next session. In the nexthour, we reversed the sequence in which the lightswere turned off and on. Although such a samplingprocedure provides similar sampling intensity ineach light-setting category, the number of birdsflying on the radar screen varied among samplingsessions and was beyond our control. Hence, thenumber of bird targets actually seen in each

light-setting category was not identical, but it wassimilar, between categories.

RADARWe monitored movements of migrating birds

with an ornithological radar. This FurunoFCR-1411 surveillance radar is a standard X-bandradar transmitting at 9,410 MHz with a peak poweroutput of 10 kW. A similar surveillance radar isdescribed in Cooper et al. (1991). The range of thisradar was set at 1.5 nautical miles (~2.8 km), thepulse setting was 0.60 µsec, and the plottingfunction was set to plot target locations every15 sec. We mounted the radar’s antenna on the roofof a building near the northeastern corner of theisland, ~50 ft (~15 m) above the beach (Fig. 1). In2001–2003, the radar laboratory (i.e., where theradar display was located and where we collectedthe data) was a small shed with plexiglas windowshanging off the sheetpile wall, directly below theelectrical building where the antenna was mounted;in 2004, we sampled from the Electrical IOBuilding itself.

The radar scanned a 360° arc around the radarlaboratory and was used to obtain information onflight paths, movement rates, behavior, andgroundspeeds of birds. This radar can be operatedat a variety of ranges and pulse lengths. The longerpulse length used in this study helped increase thedetectability of echoes. (An echo is a picture of atarget on the radar display screen; a target is one ormore birds displayed as a single echo on the radar

Table 4. Continued.


Date Radar Visual Comments 18 OC 1930–2300,

0000–0530 (18) 2000–2300,

0000–0330 (13) L. Attanas arrives; moderate winds; partly cloudy; fog early in evening, but otherwise no precipitation.

19 OC 1830–2300, 0000–0530 (20)

1830–2300, 0000–0530 (20)

Light winds; partly cloudy; no precipitation.

20 OC 1820–2300, 0000–0500 (20)

1820–2300, 0000–0500 (20)

Light to moderate winds; no precipitation; partly cloudy; gas flaring at ~0000; radar conditions excellent.

21 OC 1830–2300, 0000–0530 (20)

1830–2300, 0000–0530 (20)

Light winds; no precipitation; partly cloudy; radar conditions excellent.

22 OC 1830–2300, 0000–0430 (18)

1830–2300, 0000–0430 (18)

Light to moderate winds; partly cloudy, clearing in evening; no precipitation.

23 OC – (0) – (0) T. Obritschkewitsch and L. Attanas leave.

Methods


display screen. Flocks of birds typically appear as asingle echo because the individual birds are flyingtoo closely to one another for the radar to be able todifferentiate them.) This radar has a digital colordisplay with several scientifically useful features,including color-coded echoes (to differentiate thestrength of return signals), on-screen plotting of asequence of echoes (to depict flight paths), andTrue North correction for the display screen (todetermine flight directions easily). The plottingfunction recorded the location of a target every15 sec; because time intervals are fixed,groundspeed is directly proportional to the distancebetween consecutive echoes (Fig. 4) and can bemeasured with a hand-held scale.

Whenever energy is reflected from theground, surrounding vegetation, waves on the sea,or other objects that surround the radar unit, a“ground-clutter” or “sea-clutter” echo appears onthe display screen (Figs. 4 and 5). Because clutterechoes can obscure bird echoes, we attempted tominimize their occurrence by elevating the forwardedge of the antenna, using a ground-clutterreduction screen mounted to the bottom of theantenna face (described in Cooper et al. 1991), andusing some of the island’s buildings as a radarfence (see Eastwood 1967). We also were able todecrease sea clutter to some extent by turning onthe “sea clutter” (STC) function to its minimalsetting; however, in 2002 and 2003, sea clutteroften was so severe that we were unable toeliminate it with the STC adjustment. We did notturn up the sea clutter filter enough to eliminate seaclutter completely, for doing so also eliminatedbird-caused echoes. At the minimal STC setting wewere using, we easily could see flocks of eidersflying farther out, but they often were lost in thesea clutter next to the island during periods ofheavy sea clutter. That clutter was not eliminated,but it was reduced enough to allow us to collectdata on migrating eiders more closely to the islandthan we otherwise would have been able to do.

Maximal distances of detection of birds byradar depends on the body size, flight profile, andnumber of and distance between flying birds;atmospheric conditions that may cause attenuationof the radar signal; pulse length of the radar signal;and, to some extent, the location and extent ofclutter. For example, single, small passerines aredetectable to ~1 km, whereas large flocks of

high-flying waterfowl are detectable to several km(Cooper et al. 1991; Day and Cooper, pers. obs.).At Northstar Island, we saw some targets trackingacross the entire radar screen from the east(~3,670 m at this range-setting). Day et al. (2001)also saw eiders to the edge of the radar screen atBarrow with this radar.

We collected radar data during 721 samplingsessions during the 80 days of sampling over the 4years (Tables 1–4). The emphasis of the radarsampling was on quantifying the number of flocksof birds flying within a specified distance of theisland (the area covered by the radar screen),describing aspects of the behavior of those flocksof birds, and mapping their movements intwo-dimensional space. The sampling unit was aradar echo (target) on the display screen (i.e., aflock of birds, regardless of its size). Fromwatching these targets cross the screen, we saw noevidence (e.g., several following one another) thatthese flocks were not acting independently; onewould expect such independence of flocks,especially of those migrating at night, when theycould not see each other.

We collected the following data on each target echo seen on the radar display screen:

• time of day;

• observation number (a unique number assigned to that observation during that sampling session);

• target type (2 categories)—”eiderlike,” “non-eiderlike” (see below);

• cardinal transect crossed (4 categories)— north, east, south, west (the four compass bearings that are used to tell in which general direction from the radar sampling site the target occurred, with the emphasis on the northern and southern transects);

• flight direction (to the nearest 1° True; if direction of a target changed, we used the overall directional vector, from the beginning to the end of the line; some targets, especially circling ones, had no directional vector);

• flight velocity (to the nearest 5 mi/h [8 km/h]);

Methods


Figure 4. Radar screen at Barrow, northern Alaska, in August 1997, showing successive echoes plotted from flocks of migrating eiders (long arrows). Photo by R. H. Day.

Figure 5. Radar screen at Northstar Island, northern Alaska, in September 2001, showing echoes of numerous icebergs around the island. Photo by R. H. Day.

Methods


• general flight behavior (3 categories)— straight-line (highly directional, with long stretches of straight-line movements), erratic (may or may not be directional; often moved so erratically, especially toward the end of the line examined, that we could not predict with confidence where the target was going), circling (rarely directional and showing evidence of circling the island; Fig. 6);

• island-passing success (3 categories)—no, yes, probably yes (based on extrapolation of flightlines), depending on whether the target crossed an imaginary north–south line passing through the eastern side of the island; and

• island-passing distance (meters)— distance from the radar to the target (including extrapolated lines, if needed) as it crossed an imaginary north–south line passing through the eastern side of the island.

When possible, we determined visually thespecies and number of birds represented by theradar echo. We also traced each echo’s trackline(i.e., flightline) across the screen onto clear plasticacetate sheets. Each flightline was given a uniqueidentification number that cross-referenced it backto the individual line of data in the radar data set.

An “eiderlike” target on radar was flying withfairly specific characteristics. Eiders tend to fly intight flocks that may exhibit lateral and vertical

motion. On the radar display screen, their echoesgenerally are large and rapidly moving and oftenplot every 15 sec as a double echo. “Non-eiderlike”targets were represented by all other flightcharacteristics and ranged between small and largesizes, slow and fast speeds, and variable directions.Similarly, at Barrow, 95% of all targets visuallyconfirmed as eiders were flying 40–60 mi/h(64–96 km/h), only 1.4% were flying <40 mi/h(<64 km/h), and the targets had mostly directionalflight behavior (Day et al. 2004b).

The term “straight-line, directional flight”needs further explanation in the context of eidermigration. As indicated above, eiders may migratein long, undulating strings of birds that exhibiterratic lateral motion. Although these flocksexhibit somewhat variable (i.e., possibly erratic)flight behavior at a small scale, they almost alwaysare strongly directional at a larger scale (e.g., seeRichardson and Johnson 1981). The scale at whichwe were sampling was intermediate between thesetwo scales, so we sometimes could see smallundulations in radar flightlines, but the flightlinesalmost always were strongly directional overall. Itis for this sense of overall directional movementthat we use the term “straight-line.”

VISUALDuring the 80 days of sampling over the 4

years, we collected visual data during 1,294sampling sessions (Tables 1–4). The emphasis ofthe visual sampling was on identifying birds,quantifying flight altitudes, and describing detailed

Figure 6. Examples of straight-line, erratic, and circling flightlines on ornithological radar near Northstar Island, northern Alaska.

1,000 0 1,000 2,000meters 5

Straight-line Erratic Circling

Northstar

Island$

Northstar

Island$

Northstar

Island$

Methods


behavior of birds as they approached the island.The sampling unit was a flock of birds, regardlessof size.

We collected the following data on each bird or flock of birds seen:

• identification (to lowest practical taxon);

• flock size;

• ordinal flight direction (9 categories)— north, northeast, east, southeast, south, southwest, west, northwest, variable/erratic;

• minimal flight altitude (estimated to the nearest 1 m when flying ≤25 m agl/asl, in 5-m increments from 26 to 50 m agl/asl, in 10-m increments from 51 to 100 m agl/asl, and in 25-m increments above 100 m agl/asl), with the emphasis on altitude as the birds crossed an imaginary north–south line running through the eastern side of the island;

• island-passing behavior (4 categories)— change in flight altitude, change in flight direction, change in both, change in neither as the birds passed the island and crossed an imaginary north–south line running through the eastern end of the island;

• change in flight altitude (9 categories)—no change, decreases of 1–5 m, 6–10 m, 11–20 m, and >20 m, increases of 1–5 m, 6–10 m, 11–20 m, and >20 m as the birds passed the island and crossed an imaginary north–south line running through the eastern end of the island;

• distance from the island at which behavior changed (7 categories)—no change, 1–50 m, 51–100 m, 101–250 m, 251–500 m, 501–1,000 m, >1,000 m; and

• closest distance from the island the flock approached (7 categories)—0 m, 1–50 m, 51–100 m, 101–250 m, 251–500 m, 501–1,000 m, >1,000 m.

We also collected data on time of day, generalflight behavior, and island-passing success thesame as we did for radar data. If a visually detectedflock also was detected with the radar, we recorded

the radar observation number on the visual datasheet so that a cross-reference was available.

Migrating eiders sometimes travel in long,undulating strings that exhibit both lateral andvertical motion. Although there is lateral motion,the flocks clearly exhibit a strongly directionalcomponent of flight. As described above for radarsampling, that strongly directional movement ofeider flocks is what we are referring to with theterm “straight-line flight.” The vertical undulationsare irregular but generally do not vary by morethan 10 m (Day et al. 2001, 2004b). Hence, weconsidered a “significant” change in flightdirection or flight altitude as one that is moreextreme than these background undulations, basedon subjective judgment.

DATA ANALYSIS

For both radar and visual data, we pooledindividual species into species-groups to increasesample sizes for analyses. We used the categories(1) eiders (= Steller’s Eider + King Eider +Common Eider + unidentified eider); (2) loons(= Red-throated Loon [Gavia stellata] + PacificLoon [G. pacifica] + Yellow-billed Loon [G.adamsii] + unidentified loon); (3) other ducks(= American Wigeon [Anas americana] + NorthernPintail [A. acuta] + Long-tailed Duck [Clangulahyemalis]); (4) unidentified ducks (= unidentifiedduck + unidentified waterfowl); (5) shorebirds(= American Golden-Plover [Pluvialis dominica] +Semipalmated Sandpiper [Calidris pusilla] +unidentified phalarope [Phalaropus spp.] +unidentified sandpiper + unidentified shorebird);(6) gulls (= Parasitic Jaeger [Stercorariusparasiticus] + unidentified jaeger + Herring Gull[Larus argentatus] + Glaucous Gull [L.hyperboreus] + Black-legged Kittiwake [Rissatridactyla] + unidentified gull); (7) alcids(= Common Murre [Uria aalge] + unidentifiedmurre + Black Guillemot [Cepphus grylle] +Horned Puffin [Fratercula corniculata]); (8)unidentified eiderlike (radar) target (= targets thatwere not identified visually but that had target type“eiderlike”); (9) unidentified non-eiderlike (radar)target (= targets that were not identified visuallybut that had target type “non-eider”); (10) “eiders”(= targets that were identified visually as eiders +eiderlike targets that were identified visually as

Methods


unidentified ducks in 2001 and 2003 + unidentifiedeiderlike targets); and (11) “non-eiders” (= targetsthat were identified visually as non-eiders +non-eiderlike targets that were identified visuallyas unidentified ducks in 2003 + unidentifiednon-eiderlike targets). Hence, categories 1–7included both radar and visual categories and weresubsets of categories 10–11, whereas categories8–11 included only radar categories (and categories8–9 also were subsets of categories 10–11). In theradar section of the report, we first present data onall taxa to see whether they potentially could bemisidentified as eiders, then concentrate on twogroups: “eiders” and “non-eiders.” Although flightcharacteristics of geese (especially Brant) on radarare similar to those of eiders (Day et al. 2001,2004b), we saw only one flock of geese during thisstudy.

In data summaries and analyses, we examinedthe data in terms of six environmental factors thathave been shown to affect movements and/orcollision rates of migrating birds. We examined theeffects of period by categorizing each year aseither (sea) “ice present” (2001; the only year inwhich pack ice was present near the island) or “iceabsent” (2002–2004; when the edge of the pack icewas far offshore). We examined the effects of timeof day after pooling all light-condition samplesfrom daytime and crepuscular periods into“daytime” and retaining all samples from nighttimeperiods as “nighttime.” We examined the effects ofprecipitation level after pooling thelight-condition categories into “no precipitation”and “precipitation” (includes fog). We examinedthe effects of session visibility after pooling thevisibility categories into “good visibility” (≥500 m)and “poor visibility” (<500 m). We examined theeffects of wind direction on movements of birds,because it also has been shown to affect birdmigration and/or collision rates. We used“theoretical directions” for all analyses exceptflight velocity, for which we used “relativedirections.” Theoretical directions were based onthe idea that all eiders and most other birds wouldbe leaving the Beaufort Sea via Barrow, so weassumed that the primary flight direction would betoward the northwest. (The direction fromNorthstar Island to the base of Barrow Spit is~290°, or west–northwest.) Hence, we assumedthat winds blowing from the west, northwest, or

north would represent a headwind, those blowingfrom the east, southeast, or south would represent atailwind, those from the northeast and southwestwould represent a crosswind, and no winds andlight/variable winds would represent calmconditions. “Relative directions” were based on theactual direction of the wind relative to the actualflight direction of the bird and were used only inanalyses of flight velocity, because that factor hadimplications for collision avoidance. We examinedthe effects of wind strength on movements ofbirds, especially in the context of a winddirection*wind strength interaction, because thisfactor has been found to be important to migratingeiders at Barrow (Day et al. 2004b); wecategorized the strength as “weak” (≤10 mi/h[≤16 km/h]) or “strong” (>10 mi/h [>16 km/h]).Finally, we also examined the effects of theanti-collision lights, to determine whether they hadany effect on the response variables.

For some analyses, we also used the moonphase, as indicated by the fraction of the moon’sdisk illuminated on each date. The information ondisk illumination is available at the website(aa.usno.navy.mil/data/docs/MoonFraction.html). Wearbitrarily categorized the moon as being “full” ifthe disk was ≥75% illuminated and as “not full” ifit was <75% illuminated. Based on these criteria,the moon is full ~10–11 days per month.

We used the software SPSS and MicrosoftExcel for all analyses and data summaries. Allstatistical tests were 2-tailed, and the level ofsignificance (α) for all tests was 0.05, unless wedid a Bonferroni adjustment for multiple inference(see below). In some cases, overall sample sizeswere too small, depending on the stratificationused, to conduct statistical tests. In those cases, wesimply summarized and discussed the data.

We used a Bonferroni adjustment for somesets of statistical tests to achieve an overall α of0.05 across all tests combined. This adjustment isnecessary to achieve an overall Type I (incorrectlyconcluding that a difference exists when it actuallydoes not) error rate of 0.05 across a family ofconceptually related hypotheses and is mosteffective when the number of tests is small (Bealand Khamis 1991). Using a Bonferroni adjustment,however, also increases the probability ofcommitting a Type II error (incorrectly concludingthat a difference does not exist when it actually

Methods


does)—you simply may not find any significanteffects because α in each test is so small. In thisstudy, we used these adjustments sparingly toachieve an overall α of 0.05 in a few analyses inwhich we were unable to analyze multiple factorssimultaneously (e.g., a series of single tests ratherthan an ANOVA), while recognizing that, by doingso, we might be inflating the probability ofconcluding that the anti-collision lights weresignificantly affecting movements of these birdsacross the entire set of analyses for thatspecies-group. In reality, however, the effects oflights rarely were found to be significant (seeResults), so we, we conclude that our applicationof Bonferroni adjustments was judicious andforged a reasonable compromise betweenstatistical power and Type I errors.

RADARWe first tabulated counts of numbers of

targets recorded during each sampling session.These counts then were converted to estimates ofmovement rates (targets/h) for each session, basedon the number of minutes that we actually sampledin that session. Heavy precipitation sometimesprevented sampling for the entire session.

We examined the effects of the fiveenvironmental factors (see above) and theanti-collision lighting system on several possibleresponse variables that might indicate theeffectiveness of the anti-collisions lights at keepingbirds away from the island and, hence, at reducingthe probability of collision. We also needed toensure that the anti-collision lights did not disruptthe birds’ migration by causing an excessivelynegative response. In effect, the environmentalfactors were important factors that removed someof the variation in the data, so that we coulddetermine the true effect of the anti-collision lightson movements.

We examined movement rates to see whetheroverall migration rates changed when theanti-collision lights were on; for example,increased movement rates when the lights were onwould suggest attraction of migrating birds towardthe island, whereas decreased movement rateswould suggest the opposite. We examined flightvelocity to see whether the groundspeed of thetargets changed when the anti-collision lights wereon; for example, slower velocities would give birds

additional time to react and avoid colliding withstructures on the island. We examined flightdirection to see whether it changed when theanti-collision lights were on; for example, if flightdirections became reversed when the lights wereon, it would suggest that there was excessiverepulsion and disruption of the birds’ migratoryorientation. We examined flight behavior to seewhether behaviors of the targets changed when theanti-collision lights were on; for example, if targetsexhibited a higher proportion of non-directionalbehavior when the lights were on, it would suggestdisruption of migratory directionality. Weexamined island-passing success to see whether thebirds’ ability to migrate past the island wasaffected when the anti-collision lights were on; forexample, if passing success declined dramaticallywhen the lights were on, it would suggest that theyapparently were too effective at repulsion and thatthe lighting setup needed to be modified. Weexamined island-passing distances to see whethertargets flew at different distances when theanti-collision lights were on; for example, ifpassing distances decreased dramatically when thelights were on, it would suggest that the lights wereattracting birds and, thus, increasing the probabilityof collision. Finally, we examined spatialdistribution to see whether that changed when theanti-collision lights were on; for example, if targetsmoved in discrete corridors far from the islandwhen the lights were off but spread out evenly inspace when the lights were on, that change mightincrease the probability of collision.

For three of the response variables (movementrates, flight velocity, and island-passing distance),we calculated a series of multifactor ANOVAmodels containing all combinations of the maineffects (including no variables) period, time of day,precipitation level or session visibility (dependingon the response variable), wind direction, windspeed (included as both a main effect and aninteraction between wind direction and windspeed), and anti-collision lighting setting, plus theinteraction terms lights*time of day,lights*precipitation level (or session visibility),and lights*wind direction. In these models, weincluded interaction terms only in those models inwhich both factors were also included as maineffects. Wind speed was only included if winddirection was already in the model. For each

Methods


response variable, there were 114 models each for“eiders” and “non-eiders”. This random-blockdesign may not have as much statistical power asthat for a paired-sample random-block design, butsample sizes resulting from the paired-sampleblocking were too small to achieve a substantialincrease in statistical power (Appendix 1). Hence,we conducted analyses with the random-blockdesign, rather than the paired-sample design.

For both “eiders” and “non-eiders,” wecompared competing models with aKullback–Liebler information–theoretic approach(Burnham and Anderson 1998). This approachacknowledges uncertainty in determining the bestcombination of possible explanatory variables bycalculating the relative strength of all models in aset of candidate models based on AkaikeInformation Criteria (AIC) and allowsmodel-selection uncertainty to be incorporated inthe parameter estimates and their standard errors(SE). We calculated Akaike Information Criteriaadjusted for small sample sizes (AICc) with theformula for least-squares models and used theAkaike Weights to estimate the relative probabilitythat each model was the best model in the set(Anderson et al. 2000). We considered the potential“best model” to be any of those models in whichthe Δ AICc was ≤2.0 units from the model with thelowest AICc. For each variable, we calculated thesum of Akaike Weights for all models containingthat variable to estimate the probability that a givenvariable is in the best model. We then calculatedthe parameter estimates and SEs for each modeland calculated model-weighted parameterestimates and SEs by weighting the average byAkaike Weight (relative strength of the model) tocalculate unconditional parameter estimates andunconditional SEs (Burnham and Anderson 1998,Anderson et al. 2000). In all of these models, thenull hypothesis was that the model factors(environmental factors, anti-collision light setting)had no effect on the response variable (movementrate, velocity, passing distance) for thatspecies-group (“eiders,” “non-eiders”).

For two of the response variables (flightbehavior, island-passing success), we calculated aseries of multifactor logistic-regression modelscontaining the factors period, time of day, sessionvisibility, wind direction, wind speed (included asboth a main effect and an interaction between wind

direction and wind speed), and anti-collisionlighting setting, plus the interaction termslights*time of day, lights*session visibility, andlights*wind direction. Again, we includedinteraction terms only in those models in whichboth factors were included as main effects. Forflight behavior, we also included the variablesmoon visible (yes or no), moon full (yes or no), andthe interaction term for these two variables in allmodels except the model with no variables, basedon results from 2001. For “eiders,” we did notinclude the interaction term lights*wind directionin the behavior analyses because of insufficientsample sizes in some combinations of lights andwind. We calculated the AICc value based on thelog-likelihood of each model (Burnham andAnderson 1998), then compared models based onAkaike Weights and calculated model-weightedparameter estimates and SEs, as described above.We also did not run any models for passing successof “eiders” because only 8 of 928 “eider” targetsfailed to pass the island during the four years; inthat case, we conducted Chi-square tests for row ×column independence separately for each factor. Inall of these models, the null hypothesis was that themodel factors had no effect on the responsevariable (flight behavior, island-passing success)for that species-group.

We used the estimated movement rates foreach sampling session to calculate the mean ± 1standard error (SE) movement rate and the range inmovement rates of “eiders” and “non-eiders” byday. We then calculated mean, SE, and range inmovement rates by factor (time of day,precipitation level, session visibility, winddirection, wind speed) and anti-collision lightingsetting (lights off, lights on). We examined theeffects of the anti-collision lights on movementrates by testing the above factors with allcombinations of multifactor ANOVA modelscontaining the above main effects. Precipitationlevel and session visibility were correlated, andprecipitation level appeared to be a better predictorof movement rates than was session visibility(based on RSS and R² in exploratory analyses), sowe included it, rather than visibility, in thesemultifactor ANOVAs. Prior to conductingstatistical analyses, we added 0.167 to avoidcomputing the logarithm of zero (followingMosteller and Tukey 1977), then ln-transformed

Methods


the movement data to normalize them. The largenumber of zeros was potentially a problem, butMonte Carlo simulations (not presented here)indicated that the significance levels of t-tests didnot differ greatly from expected; and, the t-test isrobust to moderate deviations from the assumptionof normality (Zar 1984).

We calculated the mean, SE, and range inflight velocity of each species-group by factor andanti-collision lighting setting. We then examinedthe effects of the anti-collision lights on flightvelocities by testing the above factors (usingvisibility rather than precipitation) for eachspecies-group, as described above.

We calculated mean, circular SD (S’), andlength of directional vector (r) of flight directionsof each species-group by factor and anti-collisionlighting setting. We then calculated a series ofsingle-factor models containing the main effectsperiod, time of day, precipitation level, visibility,wind direction, wind speed, and lighting setting.Following Zar (1984:446–450), we used amultisample Watson–Williams test for differencesin mean vectors. Because no statistical tests allowthe use of ANOVA analyses for circular statistics,we had to examine the data as a series of separateanalyses for each factor, then use a Bonferroniadjustment for multiple inference to achieve anoverall α of 0.05 across all tests by dividing 0.05by the number of tests (7 in this case) as our levelof significance for each factor (see Beal andKhamis 1991). Before conducting the analyses, wetested whether we could pool data between periodsby conducting a Watson–Williams test of overalldirections between periods; if period was notsignificant, we pooled them before the aboveanalyses. The directionality of “non-eider” flightwas too low to conduct Watson-Williams tests.

To examine general behavioral responses tothe island, we first summarized the flight-behaviordata into the three behavioral categories by factorand anti-collision lighting setting. Because two ofthe three behaviors (erratic and circling) were rare,we pooled them as non-directional behavior, thentested for differences between proportions ofstraight-line (i.e., directional) and non-directionalflight behavior with a series of multifactorlogistic-regression models. As described above, wealso included in all models whether the moon wasor was not visible (either being obscured by clouds

or by being below the horizon) and whether it wasor was not full (see above for definitions). In thesemodels, we used session visibility, rather thanprecipitation level, based on exploratory models.The interaction terms wind direction*wind strengthand lights*wind direction were not included for“eiders” in any models because the sample sizeswere small to achieve accurate parameterestimates.

To examine island-passing success, wesummarized the success data into two categories(unsuccessful vs. successful) by factor andanti-collision lighting setting. We pooled definitelysuccessful and probably successful into“successful,” then tested for differences betweenproportions of unsuccessful and successful passingevents with a series of multifactor logistic-regression models. In these models, we again usedsession visibility instead of precipitation level.Passing success of “eiders” was essentially 100%,so only statistical analyses of factors affecting thepassing success of “eiders” was conducted, aChi-square test for row × column independencethat examined the effects of lights on passingsuccess.

We calculated actual island-passing distancesby first converting the actual measurements(distance from the radar) to distances from theisland by subtracting 40 m (the distance from theradar site to the northern shoreline) from thepassing distance if the target crossed north of theisland and 170 m (the distance from the radar siteto the southern shoreline) to the distance if thetarget crossed south of the island, and classifyingtargets crossing within these distances as havingcrossed over the island (i.e., crossing distance of0 m); unsuccessful birds were excluded from thisanalysis. We then used these corrected distances tocalculate mean and SE passing distances by factorand collision lighting setting, then examined theeffects of the anti-collision lights on passingdistances by testing the factors, as above; in thesemodels, we again used session visibility, ratherthan precipitation level. Prior to analyses, weexcluded all passing distances >4,000 m (i.e., faroff the radar screen) and used a square-roottransformation of the distance to normalize thedata.

Because we did not detect a significant effectof lights on the passing distances of eiders (see

Methods


Results), we conducted a series of powercalculations that incorporated the actual variabilityin the data. We used the actual “lights off”passing-distance data in a series of Monte Carlosimulations by taking random samples (withreplacement) of the lights-off data, adding variousdistances (e.g., 100 m, 150 m, 200 m) to thesamples (as if there was a mean shift of thatdistance caused by turning on the lights), thentesting the shifted data against the unshifted datafor 2,000 iterations of a simple t-test. This methodassumes that there is no change in the shape of thestatistical distribution of the distances; a plot of thestatistical distribution of the actual data setsuggests that this assumption is correct. Becausewe were just doing t-tests and not including othersources of variation (e.g., wind, precipitation) inthis power analysis, the resulting power estimatesare likely underestimates. We also calculated theprobability of detecting a spatial shift of certainmean distances (100 m, 150 m, 200 m) at differentsample sizes, given the variability in the passingdata, with similar Monte Carlo techniques.

To examine the effects of the anti-collisionlighting on spatial variations in movement, wedigitized the flightlines of all bird flocks seen onradar, then used the GIS software ArcGIS v.9 andplotted the data by species-group and anti-collisionlighting setting. During this work, we not onlydigitized the actual flightline tracing, but alsodigitized extrapolated flightlines before and/orafter the actual flightline if they would occur on theeastern half of the radar screen (the direction mostbirds were coming from). This extrapolation wasdone so that we would have equal samplingintensity, and, hence, could calculate overallmovement intensity, across the entire eastern halfof the screen. We overlaid these flightlines over agrid of 250-m × 250-m cells, then excluded partialcells, leaving us 261 complete grid cells foranalysis. We then counted the number of flightlinescrossing each cell. Because of the GIS projection,grid cells differed slightly in area, so we calculatedthe number of targets/km² for each grid cell. Wethen used the total length of sampling effort (whichdiffered in total length between “lights off” and“lights on” settings) to calculate the movementdensity (targets/km²/h) for each grid cell. Weplotted movement rates for each grid cell, thensubtracted the rates for “lights on” from rates in the

same cells for “lights off,” similar to aBefore–After test (Wiens and Parker 1995). Wethen conducted regression analyses of thedifference(After – Before) in movement rates in each gridcell versus distance of the midpoint of each cellfrom the island. An ANCOVA analysis indicatedthat the slopes of the lines did differ betweenperiods for “eiders” and that the intercepts weresignificantly different between periods for“non-eiders” (see Results).

We also used individual flightlines to conducta series of high-resolution analyses of theresponses of individual flocks to the island and tothe anti-collision lights. We selected only observed“eider” targets that were not circling, then used thesubset of those lines that crossed an imaginarynorth–south line passing through the eastern side ofthe island (i.e., that line that was used for passingsuccess and passing distance). (In the earlieranalyses of passing distances and spatialdistribution, we used both observed andextrapolated lines to ensure that we had equalcoverage of the eastern half of the radar screen.)This stratification resulted in a maximum of 622lines that could be used for the high-resolutionanalyses; if further subsets were used, the numberof flightlines available for analysis was evensmaller. For each flightline, we calculated theclosest distance to the island the target would havebeen if the target had stayed on its original courseand what the actual closest distance to the islandwas (calculated as the minimum of the actualclosest distance or the closest distance of theextrapolated last flight bearing). We then identifiedevery change in direction along the 622 lines andmeasured the angular change at every vertex.There were 836 vertices where “eider” flockschanged direction ≥5° (this requirement was usedto ensure that small changes in direction caused bytracing/digitizing errors were excluded).Subsequent analyses were conducted with eitherindividual vertices or individual lines as the sampleunit.

We conducted three sets of vertex analyses.These analyses examined course changes, whichwere based on changes of ≥5° in flight direction.We marked the location of each vertex andmeasured the angle of change (degrees) at eachvertex. We then used the GIS software to generatea series of concentric distance zones around the

Methods


island, split the flightlines into different segmentsfor each concentric circle, categorized each linesegment into a distance zone, and calculated thenumber of vertices ≥5°/km of line for each line ineach distance zone. Although we originally hadwanted to use 250-m distance zones, sample sizesin the innermost zone (0–249 m) were too small,even for lights off and on combined, forstatistically valid analyses; hence, we had to uselarger 500-m distance zones in analyses involvingdistance. We then calculated the mean number ofvertices/km of line for each zone, plotted themeans for each anti-collision lighting setting, andtested within each zone whether there was asignificant difference in the mean number ofvertices/km of line. We used an overall α of 0.05and bootstrap resampling to estimate the variabilitywithin each zone and to test for differences withineach distance zone and among zones. For thisanalysis, we used Bonferroni adjustments tomaintain an overall α of 0.05 across the five tests,as described above. For each combination ofdistance zone and lights, we took a random samplewith replacement of the line segments equal to theoriginal sample size, then calculated the number ofvertices/km of line for each new sample. Werepeated this bootstrap process 3,000 times, thencompared the 3,000 resulting values with the actualvalues observed. If the mean value of one categorywas greater than the mean value of the othercategory for more than 97.5% of 3,000 resampledvalues, we concluded that there was a significantdifference at α of 0.05. The null hypothesis wasthat, for each distance zone, the mean number ofvertices/line did not differ between anti-collisionlighting settings. We also compared the number ofvertices/km of line between each of the closest twodistance categories and all other distancecategories with a bootstrap analysis of 3,000repetitions. Again, we used Bonferroni adjustmentsto maintain an overall α of 0.05. This bootstrapapproach, rather than traditional parametricstatistics, was used in these tests because it usedthe total number of vertices over all lines and,therefore, was not affected by the large variation inthe length of individual line segments. Finally, wealso did a similar bootstrap analysis with 3,000repetitions on those targets whose originalflightlines would have passed <500 m from theisland, versus those targets that would have passed

≥500 m from the island, to see whether the numberof vertices/km of line differed between thosetargets that originally were going to pass near theisland and those that were not and to see the effectsof the lights on those patterns. In these tests, we didnot need Bonferroni adjustments to maintain anoverall α of 0.05 because there was only one test inthis family of tests.

We tested for factors affecting the distancefrom the island of vertices ≥5° with a series ofmultifactor ANOVAs containing the factors time ofday, session visibility, wind direction, wind speed(included as a main effect and a winddirection*wind speed interaction), and lights, andthe interaction terms lights*time of day,lights*session visibility, and lights*wind directionfollowing the Burnham and Andersonmodel-selection procedures described above. Thenull hypothesis was that none of the factorssignificantly affected vertex distances.

We examined whether the mean angularchange at each vertex differed by distance andbetween anti-collision lighting settings. Wecalculated the mean angular change for just thosevertices having a course change ≥5°, plotted themeans for each anti-collision lighting setting bydistance zone, and tested for factors affecting themean angular change with a series of multifactorANCOVAs containing the factors time of day,session visibility, wind direction, wind speed(included as a main effect and a winddirection*wind speed interaction), and lights, theinteraction terms lights*time of day, lights*sessionvisibility, and lights*wind direction, and thecovariate distance to island, following theBurnham and Anderson model-selectionprocedures described above. The null hypothesiswas that none of the factors significantly affectedthe angular change.

We conducted two sets of line analyses onthose lines having vertices ≥5°. These analysesexamined the overall characteristics of flightlinesand nearest distances that the flightlinesapproached from the island. We determined theclosest distance to the island of the originalflightline had the birds not changed course, thendetermined the actual closest distance (theminimum of the closest distance of the actual lineor the closest distance of the extrapolation of thefinal bearing). We measured the original passing

Methods


distance and the actual passing distance, thensubtracted the actual from the original to get the netchange caused by the course change; thus, apositive net change indicated that the coursechange caused the target to pass farther from theisland, whereas a negative change caused the targetto pass closer to the island. We then plottedfrequencies of original versus actual passingdistances by 500-m distance zones andanti-collision lighting setting and plotted mean netchanges in passing distances (actual – original) by500-m zones and anti-collision lighting setting; wethen compared mean changes by zone with atwo-factor ANOVA. The null hypothesis was thatmean changes in passing distance did not differbetween lighting settings and between distancezones.

We also tested for factors affecting the netchange with a series of multifactor ANOVAscontaining the factors time of day, sessionvisibility, wind direction, wind speed (included as amain effect and a wind direction*wind speedinteraction), lights, and the interaction termslights*time of day, lights*session visibility, andlights*wind direction, following the Burnham andAnderson model-selection procedures describedabove. Only non-straight flightlines (lines showingchanges in bearing of >10°/km and/or having ≥1vertices of ≥5°) were included in this analysis. Thenull hypothesis was that none of the factorssignificantly affected the net change in passingdistance.

VISUALAlthough we collected a large amount of

visual data, the number of samples of eachspecies-group was fairly small, especially when thedata were stratified by the various factors.Consequently, we simply summarized anddescribed the data qualitatively; we caution thatthese tentative conclusions may change when wehave large enough samples for statistical testing.

We calculated mean, SE, and range in flocksizes of each species-group by environmentalfactor (time of day, precipitation level, sessionvisibility, “theoretical” wind direction, windstrength) and anti-collision lighting setting (lightsoff, lights on). We also calculated mean, SE, andrange in flight altitudes of each species-group by

visibility category and anti-collision lightingsetting.

We summarized the data on general flightbehavior of each species-group into the threebehavioral categories by factor and anti-collisionlighting setting. We also summarized the data onisland-passing success of each species-group intotwo categories (unsuccessful, successful [=successful/probably successful]) and the data onisland-passing behavior into five groups (nochange direction/no change altitude, changedirection/no change altitude, no changedirection/change altitude, change direction/changealtitude, flare) by factor and anti-collision lightingsetting. We then stratified the success andisland-passing behavior data into two categories(“near” [≤500 m from the island], “far” [>500 mfrom the island]) of closest distance from theisland, then compared success and responsesbetween the two distance categories.

DOWNING AND MORTALITYWe examined patterns of downing and

mortality of eiders and Long-tailed Ducks byexamining environmental and other factorsassociated with these events. In the analyses, weused data on 12 Common and King eiders and 8Long-tailed Ducks from Northstar Island in fall2001, 16 Common and King eiders from nearbyEndicott Island in fall 2001, 3 Long-tailed Ducksfrom Northstar Island in fall 2002, 4 CommonEiders and 2 Long-tailed Ducks from NorthstarIsland in fall 2003, and 3 Common Eiders and 1unidentified eider from Northstar Island in fall2004. All birds except one were found dead.Although two Common Eiders were found dead atNorthstar Island on 2 August 2000, we excludedthose data from analyses because the entire island’sstructures and lights were not present until 2001and because no weather data were locally availablefrom Northstar Island in that year.

When possible, we used the automaticallycollected weather data from Northstar Island in fall(1 August–30 November) 2001–2004 in analyses.The weather station at Northstar was inoperableduring all except four days during the period2 September–25 November 2001; hence, duringthat period, we used weather data from DeadhorseAirport, whenever possible. We categorized eachday’s moon phase (full, not full) and categorized

Methods


each day’s lunar trend (waxing, waning) from dataat the U.S. Naval Observatory, as described above.We determined mean daily wind speeds from dataat Northstar Island, whenever possible; however,they had to be determined from Deadhorse for theabove period in fall 2001 and 2004. We determinedall mean daily wind directions from the DeadhorseAirport because of concerns that the tall structureson Northstar Island might bias those data. Wedetermined the mean daily barometric pressurefrom data at Northstar Island, whenever possible;however, it had to be determined from Deadhorsefor the above period in fall 2001 and 2004. Withthose data, we categorized each daily barometricchange as falling or rising (daily change >0.5 mb)or as steady (daily change ≤0.5 mb); we alsocategorized each daily change as weak (dailychange <5.0 mb) or strong (daily change ≥5.0 mb).We categorized each day’s probability of havingfog by comparing the mean daily air temperatureand the mean daily dew point. We categorized aday as probably having fog if the mean airtemperature was ≤1°C above the mean dew point.Because temperatures and dew points almostcertainly differed between Northstar Island andDeadhorse, these probabilities could only becomputed with Northstar data; hence, thesecategorizations are missing for the above period infall 2001 and 2004.

We summarized the data for eiders separatelyfrom those for Long-tailed Ducks and examinedthe data in two ways: with nights as the samplingunit and with individual birds as the sampling unit.First, we summarized environmental conditions fornights on which birds of either group weredowned. We also used the daily weather summariesthat we had compiled to determine what the overalltrends were across the three years. Second, wesummarized environmental patterns for individualbirds that were downed. Hence, the first analysisemphasized nightly patterns of downing andignored the number of birds that were downed,whereas the second one emphasized data fromnights on which multiple birds were downed.

LIMITATIONS OF THE DATA AND ANALYSES

There are three possible limitations of the dataand analyses, none of which, in our opinion,compromises the validity or conclusions of this

study. First, the data were not collected during anentire fall-migration period (July–November) inany year. We believe, however, that this datacollection over part of four fall-migration periodsshould provide a representative sample of thebehavior and responses of migrating birds. Hence,collecting data over a period less than the entire fallmigration should not result in a biased estimate ofthe migratory characteristics of these birds.

The second possible limitation of the data isthat we can detect birds only within a certaindistance from the island with radar (~2.8 km to thenorth and south and ~3.7 km to the east). Thislimitation is caused by contrasting pressuresbetween (a) a desire to sample over as large an areaas possible and (b) an inability to collect data overlarger and larger areas immediately adjacent to theisland as the sampling range increases because theinner zone of ground and sea clutter increases withincreasing range setting (i.e., we cannot collectdata in the zone where highly detailed informationon collision avoidance is needed the most, and thesize of that unsurveyable zone increases outwardfrom the island as the radar’s range increases). Werecognize the possibility that responses or changesin the behavior of migrating birds may haveoccurred off the screen (i.e., at distances greaterthan we were able to sample), but we also stressthat collecting high-resolution data near the islandwas more important than looking for behavioralresponses several kilometers from the island.

The final possible limitation is that, to ourknowledge, no mortality of either eiders or anyother birds occurred during any of our sampling forthis study. Therefore, the exact combination(s) ofenvironmental conditions that lead to mortalitymay not have occurred during our sampling. Onthe other hand, we did have several nights withprecipitation and/or heavy fog, which are factorsoften associated with bird collisions (seeIntroduction). Hence, we believe that we were ableto collect data during at least some of thoseconditions that often result in collision-causedmortality (see Results).

Results


RESULTS

In 2001, weather was good overall (Table 1).We had no major storms but did have substantialwinds on two nights. We also had littleprecipitation other than fog, so we lost littleradar-sampling time because of heavyprecipitation. In addition, the island wassurrounded by sea ice, so there were no largewaves on the ocean. In contrast, weather in2002–2004 was much worse, with high winds andassociated high seas resulting from the fact that theedge of the pack ice was up to 300 mi offshore anddecreased the ability of the radar to sample (Tables2–4). In addition, heavy precipitation occurredmore frequently in 2002–2004 than in 2001.Consequently, we lost a substantial amount ofradar-sampling time in 2002–2004.

In 2001, we conducted this study during aperiod of rapidly changing light conditions. Duringthe first night of sampling (24–25 August), the sunset at 2221 and rose the next morning at 0536 h;evening twilight ended at 2354, and morningtwilight began at 0405, for a total of 4 h:11 min ofcomplete darkness. (All sunrise, sunset, andevening and morning twilight times are taken fromastronomical tables generated at the websitesunrisesunset.com.) During the last night ofsampling (12–13 September), the sun set at 2047and rose the next morning at 0656; eveningtwilight ended at 2154 and morning twilight beganat 0551, for a total of 7 h:57 min of completedarkness. The lunar First Quarter occurred on25 August 2001, the Full Moon occurred on2 September, and the Third Quarter occurred on10 September.

In 2002, we conducted the sampling ~20 dayslater than what we had done in 2001. During thefirst night of sampling (11–12 September), the sunset at 2053 and rose the next morning at 0651;evening twilight ended at 2201 and morningtwilight began at 0544, for a total of 7 h:43 min ofcomplete darkness. During the last night ofsampling (29–30 September), the sun set at 1928and rose the next morning at 0803; eveningtwilight ended at 2030 and morning twilight beganat 0701, for a total of 10 h:51 min of completedarkness. The lunar First Quarter occurred on13 September, the Full Moon occurred on21 September, and the Third Quarter occurred on29 September.

In 2003, we conducted the sampling withextensive temporal overlap with what we had donein 2001. During the first night of sampling(27–28 August), the sun set at 2208 and rose thenext morning at 0456; evening twilight ended at2336 and morning twilight began at 0421, for atotal of 4 h:45 min of complete darkness. Duringthe last night of sampling (18–19 September), thesun set at 2021 and rose the next morning at 0718;evening twilight ended at 2125 and morningtwilight began at 0614, for a total of 8 h:49 min ofcomplete darkness. The lunar New Moon occurredon 27 August, the First Quarter occurred on3 September, the Full Moon occurred on10 September, and the Third Quarter occurred on18 September.

In 2004, we conducted the sampling inOctober, which is when the most birds are downedat the island; hence, there was no temporal overlapwith any of the other data sets. During the firstnight of sampling (5–6 October), the sun set at1858 and rose the next morning at 0830; eveningtwilight ended at 1959 and morning twilight beganat 0728, for a total of 11 h:29 min of completedarkness. During the last night of sampling(22–23 October), the sun set at 1737 and rose thenext morning at 0943; evening twilight ended at1844 and morning twilight began at 0835, for atotal of 13 h:51 min of complete darkness. Thelunar Third Quarter occurred on 6 October, theNew Moon occurred on 13 October, the FirstQuarter occurred on 20 October, and the Full Moonoccurred on 27 October.

RADAR

Over the 20 days of sampling in 2001, werecorded 690 radar targets (356 with the lights offand 334 with the lights on) that we called “eiders”and recorded 778 radar targets (375 with the lightsoff and 403 with the lights on) that we called“non-eiders,” based on flight characteristics orvisual identification. Hence, for bothspecies-groups, the number of flocks recorded wassimilar between the two anti-collision lightingsettings. In addition, we visually identified 21flocks of birds as eiders.

Over the 19 days of sampling in 2002, werecorded 105 radar targets (51 with the lights offand 54 with the lights on) that we called “eiders”and recorded 251 radar targets (123 with the lights

Results


off and 128 with the lights on) that we called“non-eiders,” based on flight characteristics orvisual identification. Hence, for bothspecies-groups, the number of flocks recorded wassimilar between the two anti-collision lightingsettings. In addition, we visually identified sixflocks of birds as eiders.

Over the 23 days of sampling in 2003, werecorded 25 radar targets (11 with the lights off and14 with the lights on) that we called “eiders” andrecorded 422 radar targets (229 with the lights offand 193 with the lights on) that we called“non-eiders,” based on flight characteristics orvisual identification. Hence, for bothspecies-groups, the number of flocks recorded wassimilar between the two anti-collision lightingsettings. In addition, we visually identified 43flocks of birds as eiders.

Over the 18 days of sampling in 2004, werecorded 108 radar targets (52 with the lights offand 56 with the lights on) that we called “eiders”and recorded 34 radar targets (17 with the lights offand 17 with the lights on) that we called“non-eiders,” based on flight characteristics orvisual identification. Hence, for bothspecies-groups, the number of flocks recorded wassimilar between the two anti-collision lightingsettings. In addition, we visually identified 39flocks of birds as eiders. We recorded only 238“eider” radar targets in 2002–2004 combined,however, so most radar data on “eiders” are from2001.

TARGET IDENTIFICATIONBirds identified visually as eiders tended to

fly with distinctive characteristics (Table 5). Theyflew with a high mean velocity (~45 mi/h[~72 km/h]), and most exhibited straight-line(directional) flight behavior; the other birdsexhibited erratic flight, whereas none exhibitedcircling flight. Overall, eiders generally were fastand directional in flight behavior. Thesecharacteristics differed slightly from what we haveseen at Barrow (Day et al. 2001, 2004b), whereeider flight paths tend to be more erratic at a finescale. Some unidentified ducks were eiderlike inboth velocity and flight behavior; hence, theirtargets seen in 2001 and 2003 probably wereeiders, and we pooled them with eiders in

subsequent radar analyses. Most of the targets thatwe eventually classified as “eiders,” however, wereunidentified eiderlike targets.

“Eiders” included targets that were identifiedvisually as eiders, targets that were identifiedvisually as unidentified ducks in 2001 and 2003,and unidentified targets that were eiderlike incharacteristics. This category excluded targets thatwere eiderlike in characteristics but were identifiedvisually as being of other species. These eiderlike,but actually non-eider, targets flew with speedsslightly higher than those of visually identifiedeiders and were highly directional in flightbehavior. About 47% of all radar targets seen in2001, ~29% of all seen in 2002, ~5% of all seen in2003, and ~76% of all seen in 2004 were classifiedas “eiders,” for a total of ~38% of all targets seen inthe 4 years combined.

In contrast to “eiders,” the various“non-eider” targets (with the exception of someLong-tailed Ducks) had flight characteristics thatexhibited little similarity to those of eiders (Table5). “Non-eiders” tended to fly considerably moreslowly than “eiders” did (mean ~28 mi/h) andtended to exhibit a much lower tendency forstraight-line flight behavior than “eiders” did. Onlyloons and Long-tailed Ducks exhibitedcharacteristics somewhat similar to those of“eiders,” but both flew more slowly and withgenerally less directionality than “eiders” did.About 53% of all radar targets seen in 2001, ~71%of all seen in 2002, ~95% of all seen in 2003, and~24% of all seen in 2004 were classified as“non-eiders,” for a total of ~62% of all targets seenin the 4 years combined.

During the radar sampling, we visuallyrecorded 26 radar targets of eiders and 6unidentified duck targets that were flying with“eiderlike” characteristics and recorded 3 radartargets of eiders that were not flying with“eiderlike” characteristics. We also recorded 36radar targets that were flying like eiders butactually were of other species (loons, Long-tailedDucks, and unidentified waterfowl). Hence, thepotential misidentification rate was (3 + 36 =39)/(32 + 39 = 71), or 55%. We consider thisestimate to be an overestimate, however, becausemost of the misidentification occurred on the night

Results

AB

R F

inal Report

29M

igration at Northstar Island

Table 5. Number and percentage of radar targets seen near Northstar Island, northern Alaska, fall 2001–2004, that were eiderlike and non-eiderlike in flight characteristics on radar, mean velocity (mi/h), and percentage exhibiting straight-line flight, by species-group. In all cases, n is the number of flocks.

Number (percentage)

Species-group Eiderlike Non-eiderlike Total (n) Mean

velocity (n) Percent straight-

line flight (n)

King Eider 1 (100.0) 0 (0) 1 50.0 (1) 100.0 (1) Unidentified eider 25 (89.3) 3 (10.7) 28 44.6 (26) 85.7 (28) Total eiders 26 (89.7) 3 (10.3) 29 44.8 (27) 86.2 (29) Unidentified duck 6 (100.0) 0 (0) 6 43.0 (5) 100.0 (6) Unidentified eiderlike targets 893 (100.0) 0 (0) 893 48.0 (887) 95.5 (893) "EIDERS" 925 (99.7) 3 (0.3) 928 47.9 (919) 95.3 (928) Pacific Loon 6 (17.6) 28 (82.4) 34 29.5 (32) 91.2 (34) Unidentified loon 9 (15.3) 50 (84.7) 59 28.1 (57) 94.9 (59) Total loons 15 (16.1) 78 (83.9) 93 28.6 (89) 93.5 (93) Northern Pintail 0 (0) 2 (100.0) 2 15.0 (2) 100.0 (2) Long-tailed Duck 20 (28.6) 50 (71.4) 70 31.9 (67) 47.1 (70) Unidentified duck 0 (0) 16 (100.0) 16 29.3 (23) 93.8 (16) Total other ducks 20 (22.7) 68 (77.3) 88 30.9 (92) 56.8 (88) Unidentified waterfowl 1 (16.7) 6 (83.3) 7 34.3 (7) 100.0 (7) Unidentified shorebird 0 (0) 1 (100.0) 1 30.0 (1) 0 (1) Glaucous Gull 0 (0) 50 (100.0) 50 19.9 (48) 68.0 (50) Black-legged Kittiwake 0 (0) 9 (100.0) 9 20.6 (9) 88.9 (9) Unidentified gull 1 (50.0) 1 (50.0) 2 30.0 (2) 100.0 (2) Total gulls 1 (0.6) 60 (98.4) 61 20.3 (59) 72.1 (61) Black Guillemot 0 (0) 1 (100.0) 1 20.0 (1) 100.0 (1) Unidentified bird 0 (0) 1 (100.0) 1 30.0 (1) 100.0 (1) Unidentified non-eiderlike targets 0 (0) 1,233 (100.0) 1,233 28.2 (1,154) 76.5 (1,233) "NON-EIDERS" 37 (2.5) 1,448 (97.5) 1,485 28.0 (1,397) 76.3 (1,485)

Results


of 20 September 2002, when a gas-flaring eventresulted in the attraction of 19 flocks ofLong-tailed Ducks to the island and appeared toincrease the velocity of these birds. The meanvelocity of Long-tailed Duck targets that appearedto be eiders that night was 44.3 mi/h (n = 15), orconsiderably higher than the overall averagevelocity for this species (Table 5). Hence, webelieve that this attraction and circling caused themisidentification rate to be inflated. Excluding thatnight, the maximal misidentification rate was (3 +17 = 20)/(32 + 20 = 52), or 38%.

MOVEMENT RATESIn 2001, both “eiders” and “non-eiders”

exhibited pulsed, irregular periods of movement(Fig. 7). We classified all of these data as occurringin the “ice present” period because ice was presentaround the island the entire time. Early in 2001(24 August–2 September), “eiders” exhibited littlemovement, with none being recorded on 5 of the10 nights (Fig. 7). Later in 2001 (3–12 September),however, movement rates were moderate–high,with “eiders” being recorded on all nights.“Non-eiders” exhibited a movement pattern similarto that seen for “eiders,” with low–moderatemovements early in 2001 and moderate–high

movement rates late in 2001 (Fig. 8,).“Non-eiders” were not recorded on 4 of 10 nightsin the early period, but they were recorded on allnights in the late period.

In 2002, both “eiders” and “non-eiders” againexhibited pulsed, irregular periods of movement(Figs. 7 and 8). We classified all of these data asoccurring in the “ice absent” period because theedge of the pack ice was ~75 mi (~120 km)offshore. “Eiders” exhibited low movement ratesoverall, with no birds being recorded during 5 of 11nights of sampling (Fig. 7). “Non-eiders” exhibiteda movement pattern somewhat similar to that seenfor “eiders,” with low–moderate movement ratesoverall and with no birds being recorded on onenight of sampling (Fig. 8).

In 2003, both “eiders” and “non-eiders” againexhibited pulsed, irregular periods of movement(Figs. 7 and 8). We classified all of these data asoccurring in the “ice absent” period because theedge of the pack ice was ~150 mi (~240 km)offshore. “Eiders” exhibited low movement ratesoverall, with no birds being recorded during 5 of10 nights of sampling (Fig. 7). “Non-eiders”exhibited a movement pattern somewhat similar tothat seen for “eiders,” with low–moderate

Figure 7. Movement rates (targets/h) of “eider” targets on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, by date.

0

5

10

15

20

25

30

35

40

24

AU

26

AU

28

AU

30

AU

01

SE

03

SE

05

SE

07

SE

09

SE

11

SE

13

SE

15

SE

17

SE

19

SE

21

SE

23

SE

25

SE

13

OC

15

OC

17

OC

19

OC

21

OC

DATE

ME

AN

MO

VE

ME

NT

RA

TE

(T

AR

GE

TS

/H)

2001 2002 2003 2004

Results


movement rates overall and with no birds beingrecorded on two nights of sampling (Fig. 8).

In 2004, both “eiders” and “non-eiders” againexhibited pulsed, irregular periods of movement(Figs. 7 and 8). We classified all of these data asoccurring in the “ice absent” period because theedge of the pack ice was ~300 mi (~480 km)offshore. “Eiders” exhibited low movement ratesoverall, with no birds being recorded during 1 of10 nights of sampling; movement rates increasedduring our last night of sampling, as migrationconditions improved (Fig. 7). “Non-eiders”exhibited a movement pattern somewhat similar tothat seen for “eiders,” with low movement ratesoverall and with no birds being recorded on fournights of sampling (Fig. 8).

Overall nightly movement rates were highlysimilar between the two species-groups across allyears (Pearson product-moment correlation;r = 0.486; df = 49; P < 0.001), indicatingsubstantial concordance in overall movements.Within years, nightly movement rates were highlycorrelated between the two species-groups in 2001(r = 0.532, df = 18, P = 0.016), 2003 (r = 0.733,df = 8, P = 0.016), and 2004 (r = 0.840, df = 8,P = 0.002), but they were not in 2002 (r = 0.512,df = 9, P = 0.107).

In 2001–2004, mean movement rates of“eiders” qualitatively exhibited dramatic variationsby time of day, precipitation level, sessionvisibility, wind direction, and wind strength butwere not substantially different betweenanti-collision lighting settings (Table 6). Thesequalitative patterns were supported by thestatistical analyses, in that the best-approximatingmodel describing movement rates of “eiders”included the parameters period (Σwi = 1.000),precipitation level (Σwi = 1.000), and winddirection (Σwi = 0.995), all of which were includedin all models in the best-model set (i.e., thosemodels whose AICc values were within 2.0 unitsof the model with the smallest AICc value [Tables7–9]). This model had an Akaike Weight of 0.303,which was substantially higher than that of thesecond-best model (0.195). Movement rates of“eiders” were significantly higher when sea icewas present (2001) than when it was absent(2002–2004), significantly higher withoutprecipitation, and significantly higher duringtailwinds and crosswinds than headwinds but notdifferent among calm winds, crosswinds, andtailwinds, but were not different between calmwinds and headwinds (based on multiplecomparisons in the best model). Movement ratesdid not differ significantly by time of day, wind

Figure 8. Movement rates (targets/h) of “non-eider” targets on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, by date.

0

5

10

15

20

25

30

35

40

24AU

26AU

28AU

30AU

01SE

03SE

05SE

07SE

09SE

11SE

13SE

15SE

17SE

19SE

21SE

23SE

25SE

13OC

15OC

17OC

19OC

21OC

DATE

ME

AN

MO

VE

ME

NT

RA

TE

(T

AR

GE

TS

/H)

2001 2002 2003 2004

Results

Migration at N

orthstar Island32

AB

R F

inal Report

Table 6. Movement rates (targets/h) of “eiders” and “non-eiders” migrating near Northstar Island, northern Alaska, fall 2001–2004, by species-group, period, factor, anti-collision lighting setting, and period. Data are presented as mean ± SE and n sampling sessions.

Period

Ice present Ice absent Total

Species-group Factor Attribute Mean ± SE n Mean ± SE n Mean ± SE n

"Eiders" Time of day Daytime 4.0 ± 0.9 84 1.0 ± 0.2 85 2.5 ± 0.5 169 Nighttime 6.8 ± 1.2 191 1.3 ± 0.2 361 3.2 ± 0.4 552 Precipitation level No precipitation 13.3 ± 1.9 114 1.3 ± 0.2 344 4.3 ± 0.5 458 Precipitation 0.8 ± 0.1 161 1.2 ± 0.3 102 0.9 ± 0.2 263 Session visibility Good 11.6 ± 1.6 133 1.2 ± 0.2 409 3.8 ± 0.5 542 Poor 0.7 ± 0.1 142 1.8 ± 0.8 37 0.9 ± 0.2 179 Wind (theoretical) Calm 3.2 ± 0.9 17 0.5 ± 0.3 23 1.7 ± 0.5 40 Crosswind 10.7 ± 2.5 37 1.3 ± 0.4 113 3.7 ± 0.8 150 Headwind 0.6 ± 0.2 75 1.4 ± 0.4 95 1.0 ± 0.2 170 Tailwind 7.8 ± 1.4 146 1.3 ± 0.2 215 3.9 ± 0.5 361 Wind strength Weak 4.1 ± 1.1 186 1.2 ± 0.2 327 3.3 ± 0.5 513 Strong 6.9 ± 1.2 89 1.3 ± 0.2 119 2.5 ± 0.5 208 Lights Off 5.7 ± 1.2 147 1.1 ± 0.2 230 2.9 ± 0.5 377 On 6.3 ± 1.2 128 1.4 ± 0.2 216 3.2 ± 0.5 344 Total Total 6.0 ± 1.0 275 1.3 ± 0.4 446 3.1 ± 0.4 721 "Non-eiders" Time of day Daytime 3.4 ± 0.5 84 8.7 ± 0.9 85 8.0 ± 0.6 169 Nighttime 8.1 ± 0.8 191 2.2 ± 0.3 361 4.2 ± 0.4 552 Precipitation level No precipitation 12.2 ± 1.1 114 4.2 ± 0.4 344 6.2 ± 0.4 458 Precipitation 2.8 ± 0.5 161 2.5 ± 0.5 102 2.7 ± 0.4 263 Session visibility Good 11.4 ± 1.1 133 4.0 ± 0.3 409 5.8 ± 0.4 542 Poor 2.3 ± 0.4 142 1.8 ± 0.8 37 2.2 ± 0.3 179 Wind (theoretical) Calm 6.1 ± 1.2 17 12.1 ± 1.9 23 9.5 ± 1.3 40 Crosswind 8.9 ± 1.2 37 3.9 ± 0.6 113 3.7 ± 0.8 150 Headwind 6.3 ± 0.8 75 6.9 ± 0.9 95 6.6 ± 0.6 170 Tailwind 6.4 ± 1.0 146 1.6 ± 0.3 215 3.5 ± 0.5 361 Wind strength Weak 8.6 ± 0.8 186 4.9 ± 0.4 327 6.2 ± 0.4 513 Strong 2.7 ± 0.5 89 0.9 ± 0.3 119 2.5 ± 0.5 208 Lights Off 6.0 ± 0.8 147 3.8 ± 0.4 230 4.7 ± 0.4 377 On 7.4 ± 1.0 128 3.8 ± 0.5 216 5.2 ± 0.5 344 Total Total 6.7 ± 0.6 275 3.8 ± 0.3 446 4.9 ± 0.3 721

Results

AB

R F

inal Report

33M


Table 7. Significance of factors affecting movement rates (targets/h), velocity, flight behavior, passing success, and passing distance of “eiders” and “non-eiders” migrating near Northstar Island, northern Alaska, fall 2001–2004. All models examined the effects of the factors period, time of day, precipitation level or session visibility, wind direction (relative for velocity; theoretical for all others), wind strength, and lights on the response variable. For each response variable and species-group, these models have a AICc of 2. No models were calculated for passing success of “eiders” because success was so uniformly high.

Response variable Species-group Model RSSa nb Kc AICcd ΔAICce wif

Movement rate "Eiders" Period, precipitation, wind direction 1,773.0 721 7 662.90 0.00 0.303 Period, precipitation, wind direction, lights 1,770.1 721 8 663.78 0.88 0.195 Period, time of day, precipitation, wind

direction 1,772.5 721 8 664.76 1.86 0.119

"Non-eiders"

Period, time of day, precipitation, wind direction, wind strength, wind direction*wind strength

2,023.5 721 11 766.40 0.00 0.542

Velocity

"Eiders"

Period, time of day, wind direction, wind strength, lights, wind direction*wind strength, lights*time of day

41,028.4 912 12 3,495.76 0.00 0.373

Period, time of day, visibility, wind direction, wind strength, lights, wind direction*wind strength, lights*time of day

40,976.9 912 13 3,496.68 0.91 0.236

"Non-eiders"

Period, time of day, wind direction, wind strength, lights, wind direction*wind strength, lights*time of day, lights*wind direction

107,199.1 1200 15 5,421.25 0.00 0.361

Period, time of day, visibility, wind direction, wind strength, lights, wind direction*wind strength, lights*time of day, lights*wind direction

107,081.1 1200 16 5,421.98 0.73 0.250

Passing distance "Eiders" Period, visibility, wind direction 112,446.1 903 7 4,370.65 0.00 0.210 Period, wind direction 112,757.3 903 6 4,371.12 0.46 0.166 Period, visibility, wind direction, lights 112,439.7 903 8 4,372.64 1.98 0.078 "Non-eiders"

Period, visibility, wind direction, wind strength, lights, wind direction*wind strength, lights*wind direction

179,903.6 1344 14 6,609.58 0.00 0.221

Period, visibility, wind direction, wind strength, lights, wind direction*wind strength, lights*velocity, lights*wind direction

179,712.6 1344 15 6,610.19 0.62 0.162

Period, time of day, visibility, wind direction, wind strength, lights, wind direction*wind strength, lights*wind direction

179,847.7 1344 15 6,611.20 1.63 0.098

Results

Migration at N

orthstar Island34

AB

R F

inal Report

Table 7. Continued.

Response variable Species-group Model –2LLg nb Kc AICcd ΔAICce wif

Flight behavior

"Eiders"

Period, wind direction, wind strength, moon visibility, moon phase, moon visibility*moon phase

309.4 928 9 327.58 0.00 0.210

Period, time of day, wind direction, wind strength, lights, lights*time of day, moon visibility, moon phase, moon visibility*moon phase

304.1 928 12 328.46 0.88 0.136

Period, visibility, wind direction, wind strength, moon visibility, moon phase, moon visibility*moon phase

308.7 928 10 328.98 1.40 0.104

Period, time of day, wind direction, wind strength, moon visibility, moon phase, moon visibility*moon phase

308.8 928 10 329.04 1.46 0.101

Period, time of day, visibility, wind direction, wind strength, lights, lights*time of day, moon visibility, moon phase, moon visibility*moon phase

302.9 928 13 329.33 1.75 0.088

Period, wind direction, wind strength, lights, moon visibility, moon phase, moon visibility*moon phase

309.3 928 10 329.54 1.96 0.079

"Non-eiders"

Period, time of day, visibility, wind direction, wind strength, lights, wind direction*wind strength, lights*visibility, lights*wind direction, moon visibility, moon phase, moon visibility*moon phase

1,330.2 1485 18 1,366.68 0.00 0.478

Period, time of day, visibility, wind direction, wind strength, lights, wind direction*wind strength, lights*time of day, lights*visibility, lights*wind direction, moon visibility, moon phase, moon visibility*moon phase

1,329.5 1485 19 1,367.99 1.31 0.249

Results

AB

R F

inal Report

35M


Table 7. Continued.

Response variable Species-group Model –2LLg nb Kc AICcd ΔAICce wif

Passing success "Non-eiders" Period, time of day, lights 581.3 1483 4 589.28 0.00 0.091 Period, time of day 583.9 1483 3 589.91 0.62 0.067 Period, time of day, visibility, lights,

lights*visibility 578.7 1483 6 590.72 1.43 0.045

Period, lights 584.9 1483 3 590.95 1.66 0.040 Period, time of day, visibility, lights, 581.0 1483 5 591.03 1.74 0.038 Period, time of day, wind direction, wind

strength, lights 575.0 1483 8 591.09 1.80 0.037

Period, time of day, wind direction, wind strength, lights, lights*wind direction

568.9 1483 11 591.11 1.83 0.037

Period 587.2 1483 2 591.20 1.91 0.035 Period, time of day, lights, lights*time of day 581.2 1483 5 591.21 1.93 0.035

a Residual Sum of Squares. b Sample size. c Number of estimatable parameters in the approximating model. d Akaike’s Information Criterion corrected for small sample size. e Difference in value between AICc of the current model and that of the best approximating model (AICcmin). f Akaike Weight—probability that the current model (i) is the best approximating model among those considered. g –2 log-likelihood.

Results


Table 8. Sum of Akaike Weights for the model parameters in candidate models for each response variable. Not all parameters were used in all models for a response variable.

Response variable

Species-group Model parameter Movement

rates Velocity Flight

behavior a Passing success

Passing distance

"Eiders" Period 1.000 0.747 0.987 – 0.878

Time of day 0.361 0.999 0.527 – 0.294

Precipitation level 1.000 – – – –

Session visibility – 0.475 0.408 – 0.592

Wind direction 0.995 1.000 0.990 – 1.000

Wind strength b 0.077 1.000 0.949 0.085

Lights 0.538 0.995 0.511 – 0.364

Lights*time of day 0.110 0.993 0.297 – 0.037

Lights*precipitation 0.142 – – – –

Lights*visibility – 0.125 0.083 – 0.066

Lights*wind direction 0.034 0.061 – – 0.042 "Non-eiders" Period 1.000 1.000 1.000 0.936 1.000

Time of day 0.994 0.820 1.000 0.716 0.390

Precipitation level 1.000 – – – –

Session visibility – 0.502 1.000 0.423 1.000

Wind direction 1.000 1.000 1.000 0.495 1.000

Wind strength b 1.000 0.998 0.971 0.374 0.775

Lights 0.455 1.000 0.976 0.739 0.891

Lights*time of day 0.139 0.726 0.358 0.154 0.100

Lights*precipitation 0.151 – – – –

Lights*visibility – 0.165 0.811 0.181 0.367 Lights*wind direction 0.033 1.000 0.890 0.203 0.811

a No sum of Akaike weights was calculated for moon phase*moon visibility because this factor was present in all models. b This factor includes both wind strength as a main effect and a wind direction*wind strength interaction.

Results

AB

R F

inal Report

37M


Table 9. Model-weighted parameter estimates for factors affecting movement rates, velocity, flight behavior, passing distance, and passing success of “eiders” and “non-eiders” migrating near Northstar Island, northern Alaska, fall 2001–2004. No models of passing success for “eiders” were tested because success was uniformly high.

Species-group

"Eiders" "Non-eiders"

Response variable Model parameter Estimate SE P Estimate SE P

Movement rate Intercept –1.764 0.155 <0.001 –1.508 0.191 <0.001 Period 1.280 0.131 <0.001 1.195 0.142 <0.001 Lights off –0.111 0.148 0.454 0.069 0.176 0.693 Daytime 0.121 0.184 0.511 0.545 0.166 0.001 No precipitation 1.170 0.139 <0.001 1.285 0.155 <0.001 Calm –0.173 0.271 0.522 1.322 0.302 <0.001 Crosswind –0.005 0.159 0.975 0.590 0.193 0.002 Headwind –0.572 0.153 <0.001 0.813 0.192 <0.001 Wind strong –0.045 0.169 0.791 –0.966 0.181 <0.001 Strong/crosswind –0.165 0.526 0.754 –0.387 0.563 0.492 Strong/headwind –0.306 0.400 0.444 0.358 0.429 0.404 Lights off/daytime –0.379 0.279 0.173 –0.202 0.300 0.502 Lights off/no precipitation 0.007 0.245 0.978 –0.215 0.263 0.415 Lights off/calm –0.224 0.535 0.676 0.016 0.573 0.977 Lights off/crosswind 0.010 0.308 0.974 0.182 0.331 0.583 Lights off/headwind 0.202 0.297 0.497 0.326 0.318 0.306

Results

Migration at N

orthstar Island38

AB

R F

inal Report

Table 9. Continued.

Species-group



Velocity Intercept 46.765 0.949 <0.001 25.808 1.277 <0.001 Period 1.075 0.526 0.041 8.615 0.640 <0.001 Lights off 0.971 0.653 0.137 –3.321 1.271 0.009 Daytime 0.466 0.785 0.553 1.893 0.968 0.051 Visibility good 0.939 0.948 0.322 1.239 1.074 0.249 Calm –1.387 1.310 0.290 –1.229 1.447 0.396 Crosswind –1.786 0.641 0.005 –1.544 1.119 0.168 Headwind –2.352 0.808 0.004 –4.829 0.985 <0.001 Wind strong 3.705 0.634 <0.001 5.371 1.483 <0.001 Strong/crosswind –3.074 1.803 0.088 –6.255 2.488 0.012 Strong/headwind –5.125 1.563 0.001 –2.287 1.977 0.247 Lights off/daytime –4.441 1.168 <0.001 –2.901 1.177 0.014 Lights off/good visibility 0.213 1.680 0.899 –1.339 1.741 0.442 Lights off/calm wind –1.137 2.558 0.657 –4.416 1.965 0.025 Lights off/crosswind 0.704 1.151 0.541 4.649 1.562 0.003 Lights off/headwind 0.526 1.362 0.699 4.357 1.382 0.002

Results

AB

R F

inal Report

39M


Table 9. Continued.

Species-group



Passing distance Intercept 38.748 1.705 <0.001 37.285 1.605 <0.001 Period –2.259 0.925 0.015 –4.765 0.725 <0.001 Lights off 0.126 1.203 0.917 0.042 1.651 0.980 Daytime 0.045 1.032 0.965 0.506 0.888 0.569 Visibility good –2.187 1.485 0.141 –4.586 1.329 <0.001 Calm –10.313 2.113 <0.001 –1.499 1.668 0.369 Crosswind 0.877 0.938 0.350 4.116 1.398 0.003 Headwind 2.283 1.554 0.142 –0.687 1.147 0.549 Wind strong 1.004 1.024 0.327 0.957 1.482 0.518 Strong/crosswind 0.753 3.308 0.820 –1.302 3.644 0.721 Strong/headwind –2.430 8.133 0.765 4.674 2.472 0.059 Lights off/daytime –1.136 1.934 0.557 0.496 1.465 0.735 Lights off/visibility good –1.154 2.723 0.672 2.263 1.975 0.252 Lights off/calm wind –0.804 4.202 0.848 –2.097 2.264 0.354 Lights off/crosswind –2.086 1.817 0.251 –4.290 1.755 0.014 Lights off/headwind –2.917 2.794 0.297 1.340 1.558 0.390

Results

Migration at N

orthstar Island40

AB

R F

inal Report

Table 9. Continued.

Species-group



Flight behavior Intercept –6.274 1.093 <0.001 –2.434 0.591 <0.001 Period 1.643 0.579 0.005 0.751 0.159 <0.001 Moon full 1.915 0.529 <0.001 0.971 0.213 <0.001 Moon visible –0.283 0.560 0.613 –1.160 0.332 <0.001 Moon full/moon visible –2.194 0.851 0.010 0.207 0.394 0.600 Lights off 0.513 0.588 0.383 0.894 0.584 0.126 Daytime 0.813 0.665 0.221 –1.414 0.234 <0.001 Visibility good 0.627 0.670 0.349 1.773 0.474 <0.001 Calm wind –1.788 0.808 0.027 –1.619 0.614 0.008 Crosswind –0.019 0.471 0.969 –2.295 0.939 0.014 Headwind –1.802 0.812 0.027 –1.637 0.763 0.032 Wind weak 1.586 0.641 0.013 –0.822 0.309 0.008 Weak/crosswind – – – 1.049 0.876 0.231 Weak/headwind – – – 2.171 0.703 0.002 Lights off/daytime –1.695 0.797 0.033 –0.355 0.370 0.337 Lights off/visibility good –0.846 1.018 0.406 –1.118 0.504 0.027 Lights off/calm wind – – – –0.085 0.809 0.916 Lights off/crosswind – – – 1.048 0.413 0.011 Lights off/headwind – – – –0.330 0.316 0.295

Results

AB

R F

inal Report

41M


Table 9. Continued.

Species-group



Passing success Intercept – – – 2.756 0.733 <0.001 Period – – – 0.779 0.293 0.008 Lights off – – – –0.835 0.740 0.259 Daytime – – – –0.502 0.308 0.103 Visibility good – – – –0.274 0.882 0.756 Calm – – – –0.271 0.517 0.600 Crosswind – – – 0.326 0.471 0.490 Headwind – – – 0.021 0.518 0.968 Wind weak – – – 0.765 0.349 0.028 Weak/crosswind – – – – – – Weak/headwind – – – – – – Lights off/daytime – – – 0.013 0.538 0.981 Lights off/visibility good – – – 1.455 1.116 0.192 Lights off/calm wind – – – 0.607 0.746 0.416 Lights off/crosswind – – – –0.176 0.741 0.813 Lights off/headwind – – – 1.320 0.616 0.032

Results


strength, lights, or any interaction term, includinglights.

In 2001–2004, mean movement rates of“non-eiders” exhibited dramatic variations byperiod, time of day, precipitation level, winddirection, and wind strength but were notsubstantially different between anti-collisionlighting settings (Table 6). Statistical results for“non-eiders” were similar to those for “eiders,” inthat the best-approximating model describingmovement rates included the parameters period(Σwi = 1.000), time of day (Σwi = 0.994),precipitation level (Σwi = 1.000), wind direction(Σwi = 1.000), and a wind direction*wind speedinteraction (Σwi = 1.000). The overall best modelincluded these five variables and had an AkaikeWeight of 0.542 (Tables 7–9); no other model hadan AICc value within 2 units of this model,indicating that it had a high probability of being thebest model in the candidate set. Movement rates of“non-eiders” were significantly higher when icewas present (2001) than when it was absent(2002–2004), significantly higher during the day,significantly higher without precipitation,significantly higher during calm winds thancrosswinds and tailwinds, significantly higherduring headwinds than tailwinds, but not differentbetween calm winds and headwinds, crosswindsand headwinds, or crosswinds and tailwinds, andsignificantly higher for weak crosswinds andtailwinds than for strong crosswinds and tailwinds(based on multiple comparisons in the best model).Movement rates did not differ significantly bylights or any other interaction term.

FLIGHT VELOCITY“Eiders” exhibited little variation in flight

velocities across most visibility categories andother characteristics, averaging ~45–48 mi/h(72–77 km/h) for most factors (Table 10) and47.9 mi/h (77.1 km/h) overall. Relative winddirection was the main factor qualitativelyaffecting velocities and was expected as animportant factor because we measuredgroundspeed, not airspeed, of targets. Thebest-approximating model describing velocity of“eiders” included the parameters period (Σwi =0.747), relative wind direction (Σwi = 1.000), winddirection*wind strength (Σwi = 1.000), andlights*time of day (Σwi = 0.993; Tables 7–9); both

time of day and lights had to be included in themodel because the interaction term was significant.This model had an Akaike Weight of 0.373 (Table7), which was much higher than that of thesecond-best model (0.236). Velocities weresignificantly higher when ice was present,significantly higher with tailwinds than with anyother wind direction, and significantly higher withstrong tailwinds than with weak ones but did notdiffer significantly by wind strength for any otherwind type; they also were significantly higherduring the daytime when the lights were on thanwhen they were off but were significantly higher inthe nighttime when the lights were off than whenthey were on (based on multiple comparisons in thebest model). Velocity did not differ significantly byvisibility.

“Non-eiders” averaged flight velocities of~24–31 mi/h (38–50 km/h) for most factors (Table10) and 28.0 mi/h (45.1 km/h) overall. Velocitiesqualitatively were higher with ice present andduring tailwinds and/or crosswinds. In contrast tothe pattern seen for “eiders,” velocities of“non-eiders” always were higher with the lights onthan with them off. The best-approximating modeldescribing velocity of “non-eiders” included theparameters period (Σwi = 1.000), time of day (Σwi= 0.820), relative wind direction (Σwi = 1.000),wind direction*wind strength (Σwi = 0.998), lights(Σwi = 1.000), lights*time of day (Σwi = 0.726),and lights*wind direction (Σwi = 1.000; Tables7–9). This model had an Akaike Weight of 0.361,which was substantially higher than that of thesecond-best model (0.250). Velocities weresignificantly higher when ice was present,significantly higher with a tailwind than with aheadwind or calm winds, significantly higher witha crosswind than calm winds, significantly higherwith a strong tailwind than a weak headwind or aweak tailwind, significantly higher with a strongheadwind than a weak headwind, significantlyhigher with the lights on than off during the day butnot significantly different by anti-collision lightsetting during the night, and significantly higherwith lights on than off during calm winds andtailwinds (based on multiple comparisons in thebest model); they did not differ by light settingbetween crosswinds and headwinds. Velocity didnot differ significantly by visibility or theinteraction lights*visibility.

Results

AB

R F

inal Report

43M


Table 10. Mean flight velocity (mi/h) of birds seen on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, by species-group/period, factor, and anti-collision lighting setting. Data are presented as mean ± SE (n flocks).

Lights

Off On Total Species-group/ period Factor Attribute Mean ± SE (n) Mean ± SE (n) Mean ± SE (n)

"Eiders" Time of day Daytime 45.4 ± 0.8 (58) 49.5 ± 0.8 (82) 47.8 ± 0.6 (140) (Ice present) Nighttime 49.0 ± 0.4 (297) 47.3 ± 0.4 (251) 48.3 ± 0.3 (548) Precipitation level No precipitation 48.6 ± 0.4 (333) 48.0 ± 0.4 (300) 48.3 ± 0.3 (630) Precipitation 46.0 ± 1.1 (25) 46.8 ± 1.0 (33) 46.5 ± 0.8 (58) Session visibility Good 48.6 ± 0.4 (330) 47.8 ± 0.4 (310) 48.2 ± 0.3 (640) Poor 46.0 ± 1.1 (25) 48.5 ± 1.2 (23) 47.2 ± 0.8 (48) Wind (relative) Calm 44.8 ± 1.2 (21) 49.4 ± 1.9 (9) 46.2 ± 1.1 (30) Crosswind 46.9 ± 0.7 (65) 46.6 ± 0.7 (73) 46.8 ± 0.5 (138) Headwind 46.4 ± 1.1 (37) 47.1 ± 1.1 (45) 46.8 ± 0.8 (82) Tailwind 49.6 ± 0.4 (229) 48.5 ± 0.4 (203) 49.1 ± 0.3 (432) Wind strength Weak 47.7 ± 0.4 (286) 47.2 ± 0.4 (252) 47.5 ± 0.3 (538) Strong 51.5 ± 0.8 (69) 49.9 ± 0.7 (81) 50.7 ± 0.5 (150) Total Total 48.5 ± 0.3 (355) 47.9 ± 0.2 (333) 48.2 ± 0.2 (688) "Eiders" Time of day Daytime 38.1 ± 2.7 (13) 40.5 ± 1.9 (19) 39.5 ± 1.5 (32) (Ice absent) Nighttime 47.7 ± 0.9 (97) 48.9 ± 0.8 (102) 48.3 ± 0.6 (199) Precipitation level No precipitation 46.2 ± 1.0 (78) 47.7 ± 0.9 (105) 47.0 ± 0.7 (183) Precipitation 47.7 ± 1.8 (32) 47.2 ± 1.4 (16) 47.5 ± 1.3 (48) Session visibility Good 46.3 ± 0.9 (100) 47.7 ± 0.9 (105) 47.0 ± 0.6 (205) Poor 50.0 ± 3.7 (10) 47.2 ± 1.4 (16) 48.3 ± 1.6 (26) Wind (relative) Calm 57.5 ± 2.5 (2) 37.5 ± 7.5 (2) 47.5 ± 6.6 (4) Crosswind 46.5 ± 1.3 (31) 43.2 ± 1.1 (19) 45.2 ± 0.9 (50) Headwind 42.1 ± 2.2 (21) 40.9 ± 1.8 (23) 41.5 ± 1.4 (44) Tailwind 48.1 ± 1.3 (55) 51.0 ± 0.8 (77) 49.8 ± 0.7 (132) Wind strength Weak 46.3 ± 0.9 (83) 46.0 ± 0.8 (81) 46.1 ± 0.6 (164) Strong 47.6 ± 2.5 (27) 50.9 ± 1.5 (40) 49.6 ± 1.4 (67) Total Total 46.6 ± 0.9 (110) 47.6 ± 0.8 (121) 47.1 ± 0.6 (231)

Results

Migration at N

orthstar Island44

AB

R F

inal Report

Table 10. Continued.

Lights

Off On Total Species-group/ period Factor Attribute Mean ± SE (n) Mean ± SE (n) Mean ± SE (n) "Non-eiders" Time of day Daytime 36.2 ± 1.2 (62) 35.7 ± 1.2 (53) 36.0 ± 0.8 (115) (Ice present) Nighttime 30.0 ± 0.5 (303) 30.9 ± 0.5 (322) 30.5 ± 0.3 (625) Precipitation level No precipitation 31.6 ± 0.6 (269) 32.0 ± 0.5 (288) 31.8 ± 0.4 (557) Precipitation 29.5 ± 0.8 (96) 30.1 ± 0.7 (87) 29.8 ± 0.6 (183) Session visibility Good 31.2 ± 0.6 (289) 31.8 ± 0.5 (318) 31.5 ± 0.4 (607) Poor 30.6 ± 0.9 (76) 30.1 ± 0.8 (57) 30.4 ± 0.6 (133) Wind (relative) Calm 31.2 ± 1.1 (26) 29.4 ± 1.1 (18) 30.5 ± 0.8 (44) Crosswind 34.1 ± 0.9 (78) 30.8 ± 0.8 (67) 32.6 ± 0.6 (145) Headwind 32.1 ± 0.9 (83) 30.7 ± 0.7 (107) 31.3 ± 0.6 (190) Tailwind 34.9 ± 1.0 (89) 35.1 ± 0.8 (125) 35.0 ± 0.6 (214) Wind strength Weak 30.3 ± 0.5 (314) 31.1 ± 0.5 (330) 30.7 ± 0.3 (644) Strong 35.7 ± 1.2 (51) 34.8 ± 1.3 (45) 35.3 ± 0.9 (96) Total Total 31.1 ± 0.5 (365) 31.5 ± 0.4 (375) 31.3 ± 0.3 (740) "Non-eiders" Time of day Daytime 19.4 ± 0.8 (171) 26.2 ± 0.8 (189) 23.0 ± 0.6 (360) (Ice absent) Nighttime 23.9 ± 0.9 (174) 28.8 ± 0.9 (123) 25.9 ± 0.6 (297) Precipitation level No precipitation 20.9 ± 0.7 (272) 27.1 ± 0.6 (293) 24.1 ± 0.5 (565) Precipitation 24.7 ± 1.1 (73) 30.3 ± 1.2 (19) 25.8 ± 0.9 (92) Session visibility Good 21.8 ± 0.6 (332) 27.2 ± 0.6 (302) 24.4 ± 0.5 (634) Poor 18.8 ± 2.5 (13) 30.0 ± 1.7 (10) 23.7 ± 2.0 (23) Wind (relative) Calm 14.3 ± 1.0 (59) 28.2 ± 1.8 (47) 20.5 ± 1.2 (106) Crosswind 25.9 ± 1.4 (55) 28.8 ± 1.4 (55) 27.4 ± 1.0 (110) Headwind 22.0 ± 1.0 (126) 23.3 ± 0.9 (108) 22.6 ± 0.7 (234) Tailwind 20.7 ± 1.3 (83) 28.7 ± 1.2 (74) 24.5 ± 0.9 (157) Wind strength Weak 21.0 ± 0.6 (320) 27.6 ± 0.6 (295) 24.2 ± 0.5 (615) Strong 30.2 ± 1.7 (25) 21.5 ± 3.4 (17) 26.7 ± 1.8 (42) Total Total 21.7 ± 0.6 (345) 27.3 ± 0.6 (312) 24.3 ± 0.4 (657)

Results


FLIGHT DIRECTIONBoth “eiders” and “non-eiders” exhibited a

pronounced pattern of flight direction across allyears (Figs. 9–10). “Eiders” exhibited a bimodalpattern with strong directionality. Altogether,~53% of all “eider” targets (n = 928) flew in thedirection 290–319°, ~68% flew in the direction270–329°, and the highest percentage (~27% of all

targets) flew in the direction 300–309°. In contrast,only ~14% flew in the direction 080–139°.“Non-eiders” also exhibited a bimodal pattern, butwith much less directionality than “eiders”exhibited. Altogether, ~22% of all “non-eider”targets (n = 1,275) flew in the direction 290–319°,~34% flew in the direction 270–329°, and thehighest percentage (~9% of all targets) flew in thedirection 290–299°. In contrast, ~24% flew in thedirection 080–139°.

The overall mean flight direction for “eiders”differed significantly between periods, being 299°when ice was present and 281° when ice wasabsent; it also was affected significantly by winddirection, being significantly different when thewinds were calm, were headwinds, or werecrosswinds or tailwinds (Fig. 9, Tables 11 and 12).To a great extent, these overall differences in meandirections were caused by among-year differencesin proportions of “eiders” seen during various winddirections, which significantly affect flightdirections (Table 12), rather than being caused bythe birds actually flying in different overalldirections. Indeed, headwinds, which cause asouthwesterly vector in these birds, occurred

Figure 9. Frequencies of flight directions (° True) of “eider” targets on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004 combined, by 10° categories. Mean flight directions differed significantly between periods when ice was present and when it was absent.

Eiders-Ice Present

400 400

400

400

300 300

300

300

200 200

200

200

100 100

100

100

0

90

180

270

Ice Present (n=684)

100 200100

100

300

200

200300

300

0°

270°

180°

90°

Eiders-Ice Absent

70 70

70

70

60 60

60

60

50 50

50

50

40 40

40

40

30 30

30

30

20 20

20

20

10 10

10

10

0

90

180

270

Ice Absent (n=237)

10 20 30 40

0°

270°

180°

90°50 60

20

40

30

50

60

10

Figure 10. Frequencies of flight directions (° True) of “non-eider” targets on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004 combined, by 10° categories. Mean flight direction was not significantly directional.

Non-eiders

200 200

200

200

150 150

150

150

100 100

100

100

50 50

50

50

0

90

180

270

(n=1,275)

50

0°

270°

180°

90°

150

150100

100

Results

Migration at N

orthstar Island46

AB

R F

inal Report

Table 11. Flight directions of radar targets seen near Northstar Island, northern Alaska, fall 2001–2004, by species-group, factor, anti-collision lighting setting, and period. Data are presented as mean (° True), circular SD, directional vector r, and n flocks.

Period

Ice present Ice absent

Species-group Factor Attribute Mean SD r n Mean SD r n

"Eiders" Time of day Daytime 296 54 0.6406 141 283 65 0.5302 35 Nighttime 299 62 0.5545 543 281 51 0.6689 202 Precipitation level No precipitation 298 59 0.5877 628 278 52 0.6614 188 Precipitation 301 78 0.3978 56 295 56 0.6191 49 Session visibility Good 298 59 0.5837 635 279 55 0.6267 210 Poor 305 75 0.4249 49 297 32 0.8548 27 Wind (theoretical) Calm 62 92 0.2781 30 275 15 0.9683 5 Crosswind 296 74 0.4332 164 292 46 0.7257 62 Headwind 241 73 0.4398 22 262 49 0.6923 53 Tailwind 299 50 0.6831 468 285 58 0.5986 117 Wind strength Weak 299 64 0.5409 534 278 51 0.6679 169 Strong 297 50 0.6834 150 291 57 0.6115 68 Lights Off 300 57 0.6094 353 279 60 0.5748 113 On 297 64 0.5327 331 283 47 0.7163 124 Total Total 299 61 0.5721 684 281 53 0.6484 237 "Non-eiders" Time of day Daytime 290 99 0.2224 111 268 73 0.4430 363 Nighttime 96 110 0.1575 508 274 115 0.1327 293 Precipitation level No precipitation 328 148 0.0354 449 271 89 0.3008 569 Precipitation 102 79 0.3898 170 261 85 0.3301 87 Session visibility Good 337 153 0.0284 484 267 89 0.2962 629 Poor 102 71 0.4674 135 297 62 0.5542 27 Wind (theoretical) Calm 98 54 0.6405 53 239 86 0.3274 114 Crosswind 327 153 0.0285 129 263 106 0.1785 175 Headwind 113 107 0.1739 143 282 68 0.4976 221 Tailwind 9 146 0.0390 294 265 103 0.2013 146 Wind strength Weak 89 117 0.1235 524 272 88 0.3083 612 Strong 257 126 0.0902 95 231 86 0.3219 44 Lights Off 84 110 0.1575 286 274 91 0.2819 346 On 115 146 0.0382 333 265 85 0.3312 310 Total Total 90 125 0.0910 619 270 88 0.3041 656

Results


during ~3% of the “eider” targets when ice waspresent but during ~22% of the “eider” targetswhen ice was absent; that difference aloneprobably accounts for most of the among-yeardifferences in overall flight vectors. Consequently,we concluded that the differences were statisticallysignificant but biologically meaningless, so wepooled the data across all years (Fig. 9, Table 13).The overall mean flight direction across all yearswas 294°, or slightly south of northwest. Meanflight directions during crosswinds and tailwindswere similar and toward the northwest, duringheadwinds were significantly different from theothers and toward the southwest, and during calmconditions were significantly different from theothers and toward the northeast. Flight directionswere essentially identical between anti-collisionlight settings.

The overall mean flight direction of“non-eiders” varied between periods, being 090°when ice was present and 270° when ice wasabsent (Fig. 10, Table 11). Because of the highvariability in flight direction, the overall weightedvector length rw was too small for us to be able totest for differences among any factors; hence, wepooled the data across all years and concluded thatflight direction was non-directional and, thus, thatit was inappropriate to test for differences in flightdirections (Fig. 10, Table 13). The overall meanflight direction was 269°, or almost exactly due

west. Qualitatively, they were headed toward thewest during the daytime but toward the east atnight, toward the west during no precipitation buttoward the southeast during precipitation, andtoward the west during good visibility but towardthe east during poor visibility. They were headedtoward the south during calm conditions, towardthe west during crosswinds and headwinds, andtoward the northwest during tailwinds. Finally,they were headed toward the west during weakwinds and toward the southwest during strongwinds and were headed toward the west–northwestwith the lights off and west–southwest with thelights on. Because the directionality of all factorswas so low (Table 13), however, all of thesepatterns should be interpreted with caution.

FLIGHT BEHAVIOR“Eiders” exhibited little variation in flight

behavior: ~95% of all targets flew in a straight-line(directional) manner, ~4% flew erratically, and~1% flew by circling (Table 14). Thebest-approximating model describing behavior of“eiders” included the parameters period (Σwi =0.987), wind direction (Σwi = 0.990), wind strength(Σwi = 0.949; the wind direction*wind strengthinteraction was unable to be included in candidatemodels because the model was unable to generateparameter estimates, so this analysis included windstrength only as a main effect), and a moon

Table 12. Significance of factors affecting flight directions of “eiders” migrating near Northstar Island, northern Alaska, fall 2001–2004. Results are from multisample Watson–Williams tests for differences in mean angles. Because the weighted directionality (rw) of “non-eiders” was so low, we were unable to compute any test statistics for that species-group; hence, we conclude that that species-group exhibited no real differences for any of the factors. We used Bonferroni adjustments for multiple inference, so the level of significance (α) for individual tests here is 0.007.

Factor F df P Conclusion

Period 14.954 1, 919 <0.001 ice present ≠ ice absent Time of day 0.001 1, 919 0.978 Precipitation level 0.337 1, 919 0.562 Session visibility 1.181 1, 919 0.277 Wind direction (theoretical) 12.582 3, 917 <0.001

calm ≠ headwind ≠ crosswind, tailwind; crosswind = tailwind

Wind strength 0.185 1, 919 0.667 Lights 0.404 1, 919 0.525

a Calm = no wind or light/variable winds; crosswind = wind from northeast or southwest; headwind = wind from west, northwest, or north; tailwind = wind from east, southeast, or south.

Results


Table 13. Flight directions of radar targets seen near Northstar Island, northern Alaska, fall 2001–2004 combined, by species-group, factor, and anti-collision lighting setting. Data are presented as mean (° True), circular SD, directional vector r, and n flocks.

Direction (° True)

Species-group Factor Attribute Mean SD r n

"Eiders" Time of day Daytime 294 56 0.6165 176 Nighttime 294 60 0.5793 745 Precipitation level No precipitation 293 58 0.5977 816 Precipitation 297 67 0.5004 105 Session visibility Good 293 59 0.5878 845 Poor 301 60 0.5761 76 Wind (theoretical) Calm 031 113 0.1428 35 Crosswind 295 66 0.5130 226 Headwind 258 57 0.6114 75 Tailwind 297 52 0.6633 585 Wind strength Weak 293 61 0.5636 703 Strong 295 52 0.6603 218 Lights Off 295 58 0.5940 466 On 292 60 0.5790 455 Total Total 294 59 0.5864 921 "Non-eiders" Time of day Daytime 271 79 0.3882 474 Nighttime 098 140 0.0514 801 Precipitation level No precipitation 275 107 0.1771 1018 Precipitation 116 110 0.1583 257 Session visibility Good 271 107 0.1721 1113 Poor 097 89 0.3010 162 Wind (theoretical) Calm 176 112 0.1464 167 Crosswind 269 121 0.1086 304 Headwind 279 97 0.2353 364 Tailwind 288 134 0.0656 440 Wind strength Weak 274 121 0.1093 1136 Strong 241 110 0.1596 139 Lights Off 283 127 0.0853 632 On 261 113 0.1428 643 Total Total 269 120 0.1123 1275

Results

AB

R F

inal Report

49M


Table 14. Frequencies and percentages of general flight behaviors of birds seen on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, by species-group/period, factor, and anti-collision lighting setting. Data are presented as number (%) and n (flocks).

Flight behavior

Species-group/ Straight-line Erratic Circling

period Factor Attribute Lights Number (%) Number (%) Number (%) n

"Eiders" Time of day Daytime Off 55 (93.2) 4 (6.8) 0 (0) 59

(Ice present) On 72 (82.0) 10 (12.2) 0 (0) 82 Total 127 (90.1) 14 (9.9) 0 (0) 141 Nighttime Off 279 (93.9) 15 (5.1) 3 (1.0) 297 On 244 (96.8) 5 (2.0) 3 (1.2) 252 Total 523 (95.3) 20 (3.6) 6 (1.1) 549 Precipitation level No precipitation Off 310 (93.9) 18 (5.5) 2 (0.6) 330 On 288 (96.0) 12 (4.0) 0 (0) 300 Total 598 (94.9) 30 (4.8) 2 (0.3) 630 Precipitation Off 24 (92.3) 1 (3.8) 1 (3.8) 26 On 28 (82.4) 3 (8.8) 3 (8.8) 34 Total 52 (86.7) 4 (6.7) 4 (6.7) 60 Session visibility Good Off 310 (93.9) 18 (5.5) 2 (0.6) 330 On 293 (94.5) 14 (4.5) 3 (1.0) 310 Total 603 (94.2) 32 (5.0) 5 (0.8) 640 Poor Off 24 (92.3) 1 (3.8) 1 (3.8) 26 On 23 (95.8) 1 (4.2) 0 (0) 24 Total 47 (94.0) 2 (4.0) 1 (2.0) 50 Wind (theoretical) Calm Off 19 (90.5) 2 (9.5) 0 (0) 21 On 9 (100.0) 0 (0) 0 (0) 9 Total 28 (93.3) 2 (6.7) 0 (0) 30 Crosswind Off 63 (92.6) 5 (7.4) 0 (0) 68 On 93 (96.9) 3 (3.1) 0 (0) 96 Total 156 (95.1) 8 (4.9) 0 (0) 164 Headwind Off 6 (100.0) 0 (0) 0 (0) 6 On 14 (87.5) 2 (12.5) 0 (0) 16 Total 20 (90.9) 2 (9.1) 0 (0) 22 Tailwind Off 246 (94.3) 12 (4.6) 3 (1.1) 261 On 200 (93.9) 10 (4.7) 3 (1.4) 213 Total 446 (94.1) 22 (4.6) 6 (1.3) 474 Wind strength Weak Off 265 (92.3) 19 (6.6) 3 (1.0) 287 On 238 (94.1) 12 (4.7) 3 (1.2) 253 Total 503 (93.1) 31 (5.7) 6 (1.1) 540 Strong Off 69 (100.0) 0 (0) 0 (0) 69 On 78 (96.3) 3 (3.7) 0 (0) 81 Total 147 (98.0) 3 (2.0) 0 (0) 150 Total Total Off 334 (93.8) 19 (5.3) 3 (0.8) 356 On 316 (94.6) 15 (4.5) 3 (0.9) 334 Total 650 (94.2) 34 (4.9) 6 (0.9) 690

Results

Migration at N

orthstar Island50

AB

R F

inal Report

Table 14. Continued. Flight behavior


Species-group/year Factor Attribute Lights Number (%) Number (%) Number (%) n

"Eiders" Time of day Daytime Off 14 (100.0) 0 (0) 0 (0) 14

(Ice absent) On 21 (100.0) 0 (0) 0 (0) 21 Total 35 (100.0) 0 (0) 0 (0) 35 Nighttime Off 98 (98.0) 2 (2.0) 0 (0) 100 On 101 (98.1) 1 (1.0) 1 (1.0) 103 Total 199 (98.0) 3 (1.5) 1 (0.5) 203 Precipitation level No precipitation Off 81 (100.0) 0 (0) 0 (0) 81 On 106 (99.1) 0 (0) 1 (0.9) 107 Total 187 (99.5) 0 (0) 1 (0.5) 188 Precipitation Off 31 (93.9) 2 (6.1) 0 (0) 33 On 16 (94.1) 1 (5.9) 0 (0) 17 Total 47 (94.0) 3 (6.0) 0 (0) 50 Session visibility Good Off 103 (100.0) 0 (0) 0 (0) 103 On 106 (99.1) 0 (0) 1 (0.9) 107 Total 209 (99.5) 0 (0) 1 (0.5) 210 Poor Off 9 (81.8) 2 (18.2) 0 (0) 11 On 16 (94.1) 1 (5.9) 0 (0) 28 Total 25 (89.3) 3 (10.7) 0 (0) 28 Wind (theoretical) Calm Off 2 (100.0) 0 (0) 0 (0) 2 On 3 (100.0) 0 (0) 0 (0) 3 Total 5 (100.0) 0 (0) 0 (0) 5 Crosswind Off 26 (92.9) 2 (7.1) 0 (0) 28 On 34 (97.1) 1 (2.9) 0 (0) 35 Total 60 (95.2) 3 (4.8) 0 (0) 63 Headwind Off 32 (100.0) 0 (0) 0 (0) 32 On 21 (100.0) 0 (0) 0 (0) 21 Total 53 (100.0) 0 (0) 0 (0) 53 Tailwind Off 52 (100.0) 0 (0) 0 (0) 52 On 64 (98.5) 0 (0) 1 (1.5) 65 Total 116 (99.1) 0 (0) 1 (0.9) 117 Wind strength Weak Off 84 (97.7) 2 (2.3) 0 (0) 86 On 82 (97.6) 1 (1.2) 1 (1.2) 84 Total 166 (97.6) 3 (1.8) 1 (0.6) 170 Strong Off 28 (100.0) 0 (0) 0 (0) 28 On 40 (100.0) 0 (0) 0 (0) 40 Total 68 (100.0) 0 (0) 0 (0) 68 Total Total Off 112 (98.2) 2 (1.8) 0 (0) 114 On 122 (98.4) 1 (0.8) 1 (0.8) 124 Total 234 (98.3) 3 (1.3) 1 (0.4) 238

Results

AB

R F

inal Report

51M





"Eiders" Wind (theoretical) Calm Off 2 (100.0) 0 (0) 0 (0) 2

(Ice absent) On 3 (100.0) 0 (0) 0 (0) 3 (continued) Total 5 (100.0) 0 (0) 0 (0) 5

Crosswind Off 26 (92.9) 2 (7.1) 0 (0) 28 On 34 (97.1) 1 (2.9) 0 (0) 35 Total 60 (95.2) 3 (4.8) 0 (0) 63 Headwind Off 32 (100.0) 0 (0) 0 (0) 32 On 21 (100.0) 0 (0) 0 (0) 21 Total 53 (100.0) 0 (0) 0 (0) 53 Tailwind Off 52 (100.0) 0 (0) 0 (0) 52 On 64 (98.5) 0 (0) 1 (1.5) 65 Total 116 (99.1) 0 (0) 1 (0.9) 117 Wind strength Weak Off 84 (97.7) 2 (2.3) 0 (0) 86 On 82 (97.6) 1 (1.2) 1 (1.2) 84 Total 166 (97.6) 3 (1.8) 1 (0.6) 170 Strong Off 28 (100.0) 0 (0) 0 (0) 28 On 40 (100.0) 0 (0) 0 (0) 40 Total 68 (100.0) 0 (0) 0 (0) 68 Total Total Off 112 (98.2) 2 (1.8) 0 (0) 114 On 122 (98.4) 1 (0.8) 1 (0.8) 124 Total 234 (98.3) 3 (1.3) 1 (0.4) 238

Results

Migration at N

orthstar Island52

AB

R F

inal Report




"Non-eiders" Time of day Daytime Off 59 (89.4) 2 (3.0) 5 (7.6) 66 (Ice present) On 43 (81.1) 8 (15.1) 2 (3.8) 53

Total 102 (85.7) 10 (8.4) 7 (5.9) 119 Nighttime Off 192 (62.1) 41 (13.3) 76 (24.6) 309 On 243 (69.4) 42 (12.0) 65 (18.6) 350 Total 435 (66.0) 83 (12.6) 141 (21.4) 659 Precipitation level No precipitation Off 180 (65.5) 28 (10.2) 67 (24.4) 275 On 209 (68.8) 40 (13.2) 55 (18.1) 304 Total 389 (67.2) 68 (11.7) 122 (21.1) 579 Precipitation Off 71 (71.0) 15 (15.0) 14 (14.0) 100 On 77 (77.8) 10 (10.1) 12 (12.1) 99 Total 148 (74.4) 25 (12.6) 26 (13.1) 199 Session visibility Good Off 189 (64.1) 34 (11.5) 72 (24.4) 295 On 224 (66.9) 46 (13.7) 65 (19.4) 335 Total 413 (65.6) 80 (12.7) 137 (21.7) 630 Poor Off 62 (77.5) 9 (11.3) 9 (11.3) 80 On 62 (91.2) 4 (5.9) 2 (2.9) 68 Total 124 (83.8) 13 (8.8) 11 (7.4) 148 Wind (theoretical) Calm Off 28 (96.6) 1 (3.4) 0 (0) 29 On 22 (91.7) 2 (8.3) 0 (0) 24 Total 50 (94.3) 3 (5.7) 0 (0) 53 Crosswind Off 50 (79.4) 11 (17.5) 2 (3.2) 63 On 65 (89.0) 7 (9.6) 1 (1.4) 73 Total 115 (84.6) 18 (13.2) 3 (2.2) 136 Headwind Off 63 (57.3) 11 (10.0) 36 (32.7) 110 On 57 (64.0) 14 (15.7) 18 (20.2) 89 Total 120 (60.3) 25 (12.6) 54 (27.1) 199 Tailwind Off 110 (63.6) 20 (11.6) 43 (24.9) 173 On 142 (65.4) 27 (12.4) 48 (22.1) 217 Total 252 (64.6) 47 (12.1) 91 (23.3) 390 Wind strength Weak Off 207 (64.1) 36 (11.1) 80 (24.8) 323 On 251 (70.1) 41 (11.5) 66 (18.4) 358 Total 458 (67.3) 77 (11.3) 146 (21.4) 681 Strong Off 44 (84.6) 7 (13.5) 1 (1.9) 52 On 35 (77.8) 9 (20.0) 1 (2.2) 45 Total 79 (81.4) 16 (16.5) 2 (2.1) 97 Total Total Off 251 (66.9) 43 (11.5) 81 (21.6) 375 On 286 (71.0) 50 (12.4) 67 (16.6) 403 Total 537 (69.0) 93 (12.0) 148 (19.0) 778

Results

AB

R F

inal Report

53M





"Non-eiders" Time of day Daytime Off 163 (92.6) 12 (6.8) 1 (0.6) 176

(Ice absent) On 180 (90.0) 12 (6.0) 8 (4.0) 200 Total 343 (91.2) 24 (6.4) 9 (2.4) 376 Nighttime Off 149 (77.2) 25 (13.0) 19 (9.8) 193 On 104 (75.4) 18 (13.0) 16 (11.6) 138 Total 253 (76.4) 43 (13.0) 35 (10.6) 331 Precipitation level No precipitation Off 252 (87.8) 29 (10.1) 6 (2.1) 287 On 265 (84.7) 28 (8.9) 20 (6.4) 313 Total 517 (86.2) 57 (9.5) 26 (4.3) 600 Precipitation Off 60 (73.2) 8 (9.8) 14 (17.1) 82 On 19 (76.0) 2 (8.0) 4 (16.0) 25 Total 79 (73.8) 10 (9.3) 18 (16.8) 107 Session visibility Good Off 297 (84.6) 34 (9.7) 20 (5.7) 351 On 276 (84.1) 29 (8.8) 23 (7.0) 328 Total 573 (84.4) 63 (9.3) 43 (6.3) 679 Poor Off 15 (83.3) 3 (16.7) 0 (0) 18 On 8 (80.0) 1 (10.0) 1 (10.0) 10 Total 23 (82.1) 4 (14.3) 1 (3.6) 28 Wind (theoretical) Calm Off 61 (93.8) 4 (6.2) 0 (0) 65 On 48 (98.0) 0 (0) 1 (2.0) 49 Total 109 (95.6) 4 (3.5) 1 (0.9) 114 Crosswind Off 71 (80.7) 13 (14.8) 4 (4.5) 88 On 86 (93.5) 5 (5.4) 1 (1.1) 92 Total 157 (87.2) 18 (10.0) 5 (2.8) 180 Headwind Off 117 (80.7) 12 (8.3) 16 (11.0) 145 On 86 (71.1) 15 (12.4) 20 (16.5) 121 Total 203 (76.3) 27 (10.2) 36 (13.5) 266 Tailwind Off 63 (88.7) 8 (11.3) 0 (0) 71 On 64 (84.2) 10 (13.2) 2 (2.6) 76 Total 127 (86.4) 18 (12.2) 2 (1.4) 147 Wind strength Weak Off 293 (85.9) 28 (8.2) 20 (5.9) 341 On 269 (83.8) 29 (9.0) 23 (7.2) 321 Total 562 (84.9) 57 (8.6) 43 (6.5) 662 Strong Off 19 (67.9) 9 (32.1) 0 (0) 28 On 15 (88.2) 1 (5.9) 1 (5.9) 17 Total 34 (75.6) 10 (22.2) 1 (2.2) 45 Total Total Off 312 (84.6) 37 (10.0) 20 (5.4) 369 On 284 (84.0) 30 (8.9) 24 (7.1) 338 Total 596 (84.3) 67 (9.5) 44 (6.2) 707

Results


phase*moon visibility interaction (included in allcandidate models; Fig. 11, Tables 7–9) This modelhad an Akaike Weight of 0.210, which wassomewhat higher than that of the next-best model(0.136). The proportion of “eiders” exhibitingnon-directional flight was significantly higherwhen ice was present, significantly higher withtailwinds than headwinds or calm winds, andsignificantly higher with weak winds than strongones (Table 14). As expected, the moonphase*moon visibility interaction was highlysignificant: when the moon was full, the proportionof non-directional behavior was significantlyhigher when the moon was not visible than when itwas visible, but there was no difference due tomoon visibility when the moon was not full(Fig. 11). Although one comparison for thelights*time of day interaction had a significantparameter estimate (Table 9), the additionalexplanatory power was not sufficient to justifyadding the three additional parameters to themodel; hence, we concluded that this factor wasnot significant.

“Non-eiders” exhibited much less behavioralconsistency than “eiders” did: ~76% flew in astraight-line, directional manner, ~11% flewerratically, and ~13% flew by circling (Table 14).

In 2001, the highest percentage of non-directionalbehavior was concentrated during the period2–8 September. These non-directional individualsconsisted of a diversity of visually identifiedspecies, including Pacific Loons, Northern Pintails,unidentified ducks, American Golden-Plovers,unidentified shorebirds (both medium-sized, suchas Pectoral Sandpipers [Calidris melanotos], andsmall peeps), and Glaucous Gulls. The NorthernPintails, American Golden-Plovers, shorebirds,and Glaucous Gulls were the primary species seencircling the island at several times during thisperiod; in addition, we saw Long-tailed Duckscircling the island but could not tie those visualobservations to individual radar targets. In 2002,“non-eiders” exhibited a high percentage ofnon-directional behavior during the period16–26 September. These non-directionalindividuals consisted of Long-tailed Ducks andGlaucous Gulls, which were attracted to the islandduring a gas-flaring event (see below). In 2003,“non-eiders” exhibited a high percentage ofnon-directional behavior during the period5–14 September. These non-directional individualsconsisted of Long-tailed Ducks and unidentifiedducks and smaller numbers of Pacific Loons,unidentified loons, Glaucous Gulls, and

Figure 11. Percentage of “eider” targets on ornithological radar exhibiting non-directional flight behavior near Northstar Island, northern Alaska, fall 2001–2004 combined, by moon phase and moon visibility. Numbers above columns are total numbers of samples in each category. The moon phase*moon visibility interaction was significant.

0

5

10

15

NOT FULL FULL

MOON PHASE

% N

ON

-DIR

EC

TIO

NA

L F

LIG

HT

MOON NOT VISIBLE MOON VISIBLE

236299

165

120

Results


Black-legged Kittiwakes. In 2004, we recorded sofew “non-eiders” that we were unable to detect atemporal pattern of non-directional behavior.

The best-approximating model describing thebehavior of “non-eiders” included the parametersperiod (Σwi = 1.000), time of day (Σwi = 1.000),session visibility (Σwi = 1.000), wind direction(Σwi = 1.000), wind direction*wind strength (Σwi= 0.971), lights (Σwi = 0.976), a lights*visibilityinteraction (Σwi = 0.811), a lights*wind directioninteraction (Σwi = 0.890), and a moon phase*moonvisibility interaction (included in all candidatemodels, so Σ wi = 1.000; Fig. 12, Tables 7–9). Thismodel had an Akaike weight of 0.478, which wassubstantially higher than that of the second-bestmodel (0.249). The proportion of non-directionalflight was significantly higher when ice waspresent, significantly higher at night, significantlyhigher with good visibility, significantly higherwith tailwinds than calm winds, crosswinds, orheadwinds, significantly higher with strongtailwinds than weak tailwinds, significantly higherwith the lights off than on when visibility was poorbut not different when visibility was good,significantly higher with the lights off than on incrosswinds and tailwinds but not different between

light settings for calm winds and headwinds,significantly higher when the moon was full, andsignificantly higher when the moon was not visible(however, the moon phase*moon visibilityinteraction was not significant; Fig. 12, Table 14).

Effects of gas flaringOn the evening of 19 September 2002, an

electrical problem occurred in the ProcessingFacility at ~2330 in the evening. This problemresulted in gas flaring because the ProcessingFacility was unable to reinject the gas. While thegas flaring was occurring, one could see severalhundred meters offshore with the naked eye. Thisgas flaring continued the rest of the night, stilloccurring when we left at 0545. Altogether, thisgas flaring resulted in the burning of ~1,900 Mcf(1,900 × 106 cf) of gas during the period when wewere present on the island (A. Fiedler, BPExploration, in litt.).

On the evening of 20 September (the nightbefore the Full Moon), the gas flaring had stoppedwhen we arrived on island at 1800. Around 1830,however, gas flaring began again and wasconsiderably larger than what had occurred theprevious night, in that we were able to see>1,000 m from the island with the naked eye;

Figure 12. Percentage of “non-eider” targets on ornithological radar exhibiting non-directional flight behavior near Northstar Island, northern Alaska, fall 2001–2004 combined, by moon phase and moon visibility. Numbers above columns are total numbers of samples in each category. The moon phase*moon visibility interaction was significant.

0

10

20

30

40

NOT FULL FULL

MOON PHASE

% N

ON

-DIR

EC

TIO

NA

L F

LIG

HT

MOON NOT VISIBLE MOON VISIBLE

353

152

830

116

Results


consequently, we used binoculars, rather than thenight-vision scope, to sample. This gas flaringcontinued most of the night, ending at 0300, andreleased >3,844 Mcf of gas (A. Fiedler, in litt.). Itattracted several hundred birds, especiallyLong-tailed Ducks and Glaucous Gulls, to theisland.

During this evening, none (0%) of the 34 total“eider” targets exhibited non-directional flight. Wesaw them only during the gas flaring, recordingnone during periods when no gas flaring wasoccurring. Nevertheless, it appeared that the gasflaring had no effect on the flight behavior of“eiders.” (Note: The gas flaring did affect the speedand behavior of Long-tailed Ducks, some of whichappeared to be “eiderlike” on radar but wereexcluded from the “eider” category because visualidentification confirmed that they were“non-eiders.”)

During this evening, 40 (44.9%) of the 89total “non-eider” targets exhibited non-directionalflight. We saw 5 during non-flaring periods and 84during periods when gas was being flared. Duringthe non-flaring period, none (0%) of the 5 targetsexhibited non-directional behavior; in contrast,during the gas flaring, 40 (47.6%) of the targetsexhibited non-directional behavior. Surprisingly,this difference was not statistically significant(χ² = 2.614; df = 1; P = 0.106). We believe,however, that this lack of significance resultedfrom low statistical power caused by the smallsample size during the non-flaring period, ratherthan being caused by no difference between thetwo proportions. Hence, we suspect that the gasflaring caused a change in the behavior of“non-eiders.”

During this evening, we recorded many birdsvisually: loons (1 Pacific Loon, 5 unidentifiedloons in 2 groups), eiders (15 unidentified eiders in1 group), other ducks (867 Long-tailed Ducks in 31groups), gulls (31 Glaucous Gulls in 14 groups),and an unidentified passerine. The eiders passedfar (i.e., >500 m) from the island and exhibited noresponse as they passed. The Pacific Loon passednear (i.e., ≤500 m from) the island, and theunidentified loons passed far from the island; noneof the loons exhibited any response to the island asthey passed. The Long-tailed Ducks becameattracted to the island, heading toward the islandfrom distances >2,000 m away; they circled

repeatedly, disappearing slowly over the course ofseveral hours. Of the 30 groups of Long-tailedDucks for which we had behavioral data, 12 (40%)exhibited a behavioral response, with 8 (66.7%)exhibiting directional responses and 4 (33.3%)exhibiting behavioral flaring (i.e., extreme changesin direction and altitude done to avoid collision).(Note that this behavioral flaring is different fromgas flaring.) All four of these groups exhibitingbehavioral flaring did so to avoid hitting theelectrical buildings located at the northeasterncorner of the island, and all passed <50 m fromthose buildings; at least two passed within 5–10 mof the buildings. The Glaucous Gulls also becameattracted to the island, flying in from >1,000 maway and circling the island repeatedly over thecourse of several hours. Overall, 2 (14.3%) of 14groups for which we had behavioral data exhibitedbehavioral responses, with those responsesconsisting of changing direction; we saw nobehavioral-flaring, even though 13 of the 14groups passed within 500 m of the island.

ISLAND-PASSING SUCCESS“Eiders” exhibited almost no variation in

island-passing success: >99% of all targetssuccessfully passed the island (i.e., only 8 of 928flocks did not successfully pass the island; Table15). Because nearly all “eiders” passedsuccessfully, we could not conduct any multifactoranalyses of factors affecting passing successbecause not all models would converge toward asolution; instead, we used a Likelihood-ratioChi-square test to test the effects of lights onpassing success. Five failures occurred with thelights off and three occurred with the lights on(χ² = 0.454; df = 1; P = 0.501). Hence, as would beexpected, lights did not significantly affect theprobability that “eiders” successfully passed theisland.

“Non-eiders” also had a high overall rate ofpassing success, although it was lower than that for“eiders”: ~95% of all targets successfully passedthe island (Table 15). The best-approximatingmodel describing passing success of “non-eiders”included period (Σwi = 0.936), time of day(Σwi = 0.716), and lights (Σwi = 0.739; Tables7–9). Although the lights*wind directioninteraction had one significant model-weightedparameter estimate, the sum of Akaike weights for

Results

AB

R F

inal Report

57M


Table 15. Frequencies and percentages of island-passing success of birds seen on ornithological radar near Northstar Island, northern Alaska, fall 2001–2003, by species-group/period, factor, and anti-collision lighting setting. Data are presented as number (%) and n (flocks for Total columns).

Lights

Species-group/ Off On Total

period Factor Attribute Unsuccessful Successful Unsuccessful Successful Unsuccessful Successful n

"Eiders" Time of day Daytime 1 (1.7) 58 (98.3) 0 (0) 82 (100.0) 1 (0.7) 140 (99.3) 141 (Ice present) Nighttime 1 (0.3) 296 (99.7) 1 (0.4) 251 (99.6) 2 (0.4) 547 (99.6) 549 Precipitation level No precipitation 2 (0.6) 328 (99.4) 1 (0.3) 299 (99.7) 3 (0.5) 627 (99.5) 630 Precipitation 0 (0) 26 (100.0) 0 (0) 34 (100.0) 0 (0) 60 (100.0) 60 Session visibility Good 2 (0.6) 328 (99.4) 1 (0.3) 309 (99.7) 3 (0.5) 637 (99.5) 640 Poor 0 (0) 26 (100.0) 0 (0) 24 (100.0) 0 (0) 50 (100.0) 50 Wind (theoretical) Calm 1 (4.8) 20 (95.2) 0 (0) 9 (100.0) 1 (3.3) 29 (96.7) 30 Crosswind 0 (0) 68 (100.0) 1 (1.0) 95 (99.0) 1 (0.6) 163 (99.4) 164 Headwind 0 (0) 6 (100.0) 0 (0) 16 (100.0) 0 (0) 22 (100.0) 22 Tailwind 1 (0.4) 260 (99.6) 0 (0) 213 (100.0) 1 (0.2) 473 (99.8) 474 Wind strength Weak 2 (0.7) 285 (99.3) 0 (0) 253 (100.0) 2 (0.4) 538 (99.6) 540 Strong 0 (0) 69 (100.0) 1 (1.2) 80 (98.8) 1 (0.7) 149 (99.3) 150 Total Total 2 (0.6) 354 (99.4) 1 (0.3) 333 (99.7) 3 (0.4) 687 (99.6) 690 "Eiders" Time of day Daytime 0 (0) 14 (100.0) 0 (0) 21 (100.0) 0 (0) 35 (100.0) 35 (Ice absent) Nighttime 3 (3.0) 97 (97.0) 2 (1.9) 101 (98.1) 5 (2.5) 198 (97.5) 203 Precipitation level No precipitation 2 (2.5) 79 (97.5) 2 (1.9) 105 (98.1) 4 (2.1) 184 (97.9) 188 Precipitation 1 (3.0) 32 (97.0) 0 (0) 17 (100.0) 1 (2.0) 49 (98.0) 50 Session visibility Good 2 (1.9) 101 (98.1) 2 (1.9) 105 (98.1) 4 (1.9) 206 (98.1) 210 Poor 1 (9.1) 10 (90.9) 0 (0) 17 (100.0) 1 (3.6) 27 (96.4) 28 Wind (theoretical) Calm 0 (0) 2 (100.0) 0 (0) 3 (100.0) 0 (0) 5 (100.0) 5 Crosswind 1 (3.6) 27 (96.4) 0 (0) 35 (100.0) 1 (1.6) 62 (98.4) 63 Headwind 0 (0) 32 (100.0) 0 (0) 21 (100.0) 0 (0) 53 (100.0) 53 Tailwind 2 (3.8) 50 (96.2) 2 (3.1) 63 (96.9) 4 (3.4) 113 (96.6) 117 Wind strength Weak 2 (2.3) 84 (97.7) 1 (1.2) 83 (98.8) 3 (1.8) 167 (98.2) 170 Strong 1 (3.6) 27 (96.4) 1 (2.5) 39 (97.5) 2 (2.9) 66 (97.1) 68 Total Total 3 (2.6) 111 (97.4) 2 (1.6) 122 (98.4) 5 (2.1) 233 (97.9) 238

Results

Migration at N

orthstar Island58

AB

R F

inal Report


Lights

Species-group/ Off On Total

period Factor Attribute Unsuccessful Successful Unsuccessful Successful Unsuccessful Successful n

"Non-eiders" Time of day Daytime 3 (4.5) 63 (95.5) 0 (0) 53 (100.0) 3 (2.5) 116 (97.5) 119 (Ice present) Nighttime 13 (4.2) 296 (95.8) 9 (2.6) 341 (97.4) 22 (3.3) 637 (96.7) 659 Precipitation level No precipitation 11 (4.0) 264 (96.0) 7 (2.3) 297 (97.7) 18 (3.1) 561 (96.9) 579 Precipitation 5 (5.0) 95 (95.0) 2 (2.0) 97 (98.0) 7 (3.5) 192 (96.5) 199 Session visibility Good 12 (4.1) 283 (95.9) 8 (2.4) 327 (97.6) 20 (3.2) 610 (96.8) 630 Poor 4 (5.0) 76 (95.0) 1 (1.5) 67 (98.5) 5 (3.4) 143 (96.6) 148 Wind (theoretical) Calm 0 (0) 29 (100.0) 1 (4.2) 23 (95.8) 1 (1.9) 52 (98.1) 53 Crosswind 4 (6.3) 59 (93.7) 1 (1.4) 72 (98.6) 5 (3.7) 131 (96.3) 136 Headwind 5 (4.5) 105 (95.5) 2 (2.2) 87 (97.8) 7 (3.5) 192 (96.5) 199 Tailwind 7 (4.0) 166 (96.0) 5 (2.3) 212 (97.7) 12 (3.1) 378 (96.9) 390 Wind strength Weak 15 (4.6) 308 (95.4) 7 (2.0) 351 (98.0) 22 (3.2) 659 (96.8) 681 Strong 1 (1.9) 51 (98.1) 2 (4.4) 43 (95.6) 3 (3.1) 94 (96.9) 97 Total Total 16 (4.3) 359 (95.7) 9 (2.2) 394 (97.8) 25 (3.2) 753 (96.8) 778 "Non-eiders" Time of day Daytime 18 (10.3) 157 (89.7) 17 (8.5) 183 (91.5) 35 (9.3) 340 (90.7) 375 (Ice absent) Nighttime 11 (5.7) 182 (94.3) 5 (3.6) 132 (96.4) 16 (4.8) 314 (95.2) 330 Precipitation level No precipitation 22 (7.7) 264 (92.3) 21 (6.7) 292 (93.3) 43 (7.2) 556 (92.8) 599 Precipitation 7 (8.5) 75 (91.5) 1 (4.2) 23 (95.8) 8 (7.5) 98 (92.5) 106 Session visibility Good 26 (7.4) 324 (92.6) 22 (6.7) 306 (93.3) 48 (7.1) 630 (92.9) 678 Poor 3 (16.7) 15 (83.3) 0 (0) 9 (100.0) 3 (11.1) 24 (88.9) 27 Wind (theoretical) Calm 8 (12.3) 57 (87.7) 4 (8.2) 45 (91.8) 12 (10.5) 102 (89.5) 114 Crosswind 6 (6.8) 82 (93.2) 3 (3.3) 89 (96.7) 9 (5.0) 171 (95.0) 180 Headwind 5 (3.5) 139 (96.5) 11 (9.1) 110 (90.9) 16 (6.0) 249 (94.0) 265 Tailwind 10 (14.1) 61 (85.9) 4 (5.3) 71 (94.7) 14 (9.6) 132 (90.4) 146 Wind strength Weak 22 (6.5) 318 (93.5) 20 (6.3) 300 (93.8) 42 (6.4) 618 (93.6) 660 Strong 7 (25.0) 21 (75.0) 2 (11.8) 15 (88.2) 9 (20.0) 36 (80.0) 45 Total Total 29 (7.9) 339 (92.1) 22 (6.5) 315 (93.5) 51 (7.2) 654 (92.8) 705

Results


this interaction was very low (0.203), and thisinteraction was not present in most of the bestmodels; hence, we believe that the one significantmodel-weighted parameter estimate was spurious.There also was a significant model-weightedparameter estimate for wind strength; this variablewas not in the best model because we onlyconsidered wind strength in models alreadycontaining wind type, and wind type was not asignificant variable. The best model had an Akaikeweight of just 0.091, which was only slightlyhigher than that of the second-best model (0.067).There also were nine competing best models, againsuggesting that the explanatory power of allmodels was poor. Passing success was significantlyhigher when ice was present; although lights andtime of day both were in the best model, neithermodel-averaged parameter estimate wassignificantly different from zero (P > 0.05).Success also did not differ significantly byvisibility, wind direction, or any interaction terms.

ISLAND-PASSING DISTANCE“Eiders” exhibited substantial variation in

island-passing distances, with means being1,403 m with the lights off, 1,459 m with the lightson, and ~1,431 m overall (Table 16). (The radarcould see all birds to ~2,770 m north and south ofthe radar site, so this mean is well within thatmaximal distance.) The best-approximating modeldescribing the passing distance of “eiders”included the parameters period (Σwi = 0.878),session visibility (Σwi = 0.592), and wind direction(Σwi = 1.000; Tables 7–9). This model had anAkaike weight of 0.210, which was only slightlyhigher than that of the second-best model (0.166).Passing distance was significantly larger when icewas absent and significantly smaller with calmwinds than with crosswinds, headwinds, andtailwinds; it was not different between crosswinds,headwinds, and tailwinds (Table 16). Passingdistance did not differ by time of day, visibility (theparameter-weighted estimate for visibility was notsignificantly different from zero, indicating thatpassing distance was not significantly different byvisibility category, even though this factor was inthe best model), lights, or any interaction.

We conducted power analyses to examinewhether we had a sufficiently large sample size todetect a difference in the passing distance of

“eiders,” given the variability in the data set(Figure 13). As might be expected, powerincreased substantially with increasing samplesizes. With our sample sizes in the “lights off” and“lights on” categories (457 and 446, respectively)we should have been able to detect a change (i.e.,an effect size) of 150 m 85% of the time. We alsocalculated the sample sizes needed to detect aspecified change (effect size) in mean passingdistance (Figure 13). These calculations indicatethat we need 1,250 samples in each light-settingcategory (i.e., N = 2,500) to be able to detect a100-m change with 90% probability, given thevariability in the data; if we wanted to detect a56-m change with 90% probability, it wouldrequire sample sizes larger than these. On the otherhand, we can say with 90% probability that, if therewas an increase in mean passing distance causedby the anti-collision lights, it was less than 170 m.

“Non-eiders” also exhibited considerablevariation in island-passing distances, with meansbeing ~1,150 m with the lights off, ~1,114 m withthe lights on, and ~1,132 m overall (Table 16). Thebest-approximating model describing the passingdistance of “non-eiders” included the parametersperiod (Σwi = 1.000), session visibility(Σwi = 1.000), wind direction (Σwi = 1.000), winddirection*wind strength (Σwi = 0.775), lights(Σwi = 0.891), and the lights*wind directioninteraction (Σwi = 0.811; Tables 7–9). This modelhad a Akaike Weight of 0.221, which was onlyslightly better than that of the second-best model(0.162). The mean passing distance of “non-eiders”was significantly higher when ice was absent,significantly higher in poor visibility, significantlyhigher with calm winds than headwinds,significantly higher with strong headwinds thanweak headwinds, and significantly higher with thelights on than off in crosswinds but not differentbetween lighting settings for calm winds,headwinds, and tailwinds (Table 16). There was nosignificant difference in passing distance by timeof day, lights*time of day, or lights*sessionvisibility.

Targets that passed over the islandOver the four years, we recorded five “eider”

targets passing over (i.e., the calculated passingdistance was zero) Northstar Island (Table 17). Ofthese, all were recorded at night, during periods

Results

Migration at N

orthstar Island60

AB

R F

inal Report

Table 16. Island-passing distances (m) of “eiders” seen on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, by species-group/period, factor, and anti-collision lighting setting. Data are presented as mean ± SE (n flocks) distance from the shoreline of Northstar Island at which birds crossed a north–south line; passing distances >4,000 m have been excluded.

Species-group/ period

Factor

Attribute

Lights off

Lights on

Total

"Eiders" Time of day Daytime 1,260.1 ± 87.2 (53) 1,451.8 ± 90.6 (70) 1,369.2 ± 64.1 (123) (Ice present) Nighttime 1,334.9 ± 53.0 (216) 1,491.5 ± 61.1 (189) 1,408.0 ± 40.3 (405) Precipitation level No precipitation 1,317.8 ± 46.9 (244) 1,451.8 ± 52.0 (228) 1,382.5 ± 35.0 (472) Precipitation 1,342.8 ± 188.7 (25) 1,694.0 ± 181.5 (31) 1,537.2 ± 132.0 (56) Session visibility Good 1,317.8 ± 46.9 (244) 1,448.6 ± 50.6 (237) 1,382.3 ± 34.6 (481) Poor 1,342.8 ± 188.7 (25) 1,826.6 ± 237.3 (22) 1,569.3 ± 152.3 (47) Wind (theoretical) Calm 694.6 ± 127.3 (20) 669.3 ± 227.8 (8) 687.3 ± 109.5 (28) Crosswind 1,346.2 ± 89.7 (59) 1,582.8 ± 87.2 (75) 1,478.6 ± 63.4 (134) Headwind 1,121.3 ± 138.8 (6) 1,517.1 ± 203.4 (16) 1,409.1 ± 155.6 (22) Tailwind 1,386.3 ± 56.7 (184) 1,469.9 ± 66.1 (160) 1,425.1 ± 43.2 (344) Wind strength Weak 1,278.7 ± 49.7 (212) 1,474.0 ± 58.2 (192) 1,371.5 ± 38.3 (404) Strong 1,474.3 ± 111.1 (57) 1,500.1 ± 104.3 (67) 1,488.2 ± 75.7 (124) Total Total 1,320.1 ± 45.9 (269) 1,480.8 ± 50.8 (259) 1,398.9 ± 34.3 (528) "Eiders" Time of day Daytime 1,517.9 ± 239.7 (14) 1,349.5 ± 160.2 (20) 1,418.8 ± 135.0 (34) (ice absent) Nighttime 1,654.1 ± 83.8 (90) 1,554.6 ± 88.3 (88) 1,604.9 ± 60.8 (178) Precipitation level No precipitation 1,691.0 ± 91.4 (73) 1,580.9 ± 85.6 (94) 1,629.0 ± 62.5 (167) Precipitation 1,505.7 ± 154.9 (31) 1,085.1 ± 130.8 (14) 1,374.9 ± 117.1 (45) Session visibility Good 1,620.9 ± 82.4 (94) 1,580.9 ± 85.6 (94) 1,600.9 ± 59.3 (188) Poor 1,775.5 ± 287.5 (10) 1,085.1 ± 130.8 (14) 1,372.8 ± 155.4 (24) Wind (theoretical) Calm 1,480.0 ± 50.0 (2) 1,209.0 ± 467.0 (3) 1,317.4 ± 264.7 (5) Crosswind 1,705.8 ± 164.5 (26) 1,510.3 ± 178.6 (28) 1,604.4 ± 121.5 (54) Headwind 1,655.6 ± 164.0 (32) 1,807.0 ± 212.9 (17) 1,708.1 ± 129.2 (49) Tailwind 1,587.0 ± 108.8 (44) 1,452.7 ± 92.8 (60) 1,509.5 ± 70.5 (104) Wind strength Weak 1,673.8 ± 89.4 (80) 1,480.1 ± 100.9 (71) 1,563.6 ± 67.1 (151) Strong 1,629.1 ± 172.3 (24) 1,586.8 ± 120.5 (37) 1,603.4 ± 98.8 (61) Total Total 1,635.8 ± 79.1 (104) 1,516.6 ± 77.9 (108) 1,575.1 ± 55.5 (212)

Results

AB

R F

inal Report

61M



Species-group/ year

Factor

Attribute

Lights off

Lights on

Total

"Non-eiders" Time of day Daytime 1,238.9 ± 104.2 (58) 1,305.1 ± 118.2 (49) 1,269.2 ± 77.9 (107) (Ice present) Nighttime 964.5 ± 44.1 (267) 1,023.5 ± 42.9 (295) 995.5 ± 30.8 (562) Precipitation level No precipitation 980.9 ± 48.6 (232) 1,024.8 ± 48.7 (254) 1,003.8 ± 34.4 (486) Precipitation 1,094.9 ± 76.6 (93) 1,173.0 ± 72.4 (90) 1,133.3 ± 52.7 (183) Session visibility Good 987.7 ± 47.3 (251) 1,022.2 ± 45.1 (281) 1,005.9 ± 32.6 (532) Poor 1,110.8 ± 82.6 (74) 1,248.1 ± 92.0 (63) 1,168.5 ± 61.6 (137) Wind (theoretical) Calm 739.6 ± 143.2 (27) 841.5 ± 142.0 (22) 785.3 ± 100.7 (49) Crosswind 1,059.7 ± 90.2 (55) 1,330.2 ± 110.3 (61) 1,201.9 ± 72.9 (116) Headwind 1,068.3 ± 75.8 (101) 1,050.8 ± 71.5 (85) 1,060.3 ± 52.4 (186) Tailwind 1,008.7 ± 62.7 (142) 1,005.1 ± 56.8 (176) 1,006.7 ± 42.0 (318) Wind strength Weak 990.1 ± 43.1 (275) 1,055.4 ± 42.8 (304) 1,024.4 ± 30.4 (579) Strong 1,142.2 ± 122.4 (50) 1,125.7 ± 131.1 (40) 1,134.9 ± 89.0 (90) Total Total 1,013.5 ± 41.1 (325) 1,063.6 ± 40.7 (344) 1,039.2 ± 28.9 (669) "Non-eiders" Time of day Daytime 1,222.2 ± 61.5 (141) 1,137.4 ± 57.0 (160) 1,177.1 ± 41.8 (301) (Ice absent) Nighttime 1,407.5 ± 65.6 (168) 1,313.4 ± 74.4 (119) 1,368.5 ± 49.2 (287) Precipitation level No precipitation 1,363.6 ± 52.0 (238) 1,191.4 ± 46.0 (261) 1,273.5 ± 34.7 (499) Precipitation 1,186.4 ± 94.1 (71) 1,518.1 ± 236.9 (18) 1,253.5 ± 89.5 (89) Session visibility Good 1,306.1 ± 46.9 (294) 1,205.7 ± 46.2 (270) 1,258.0 ± 33.0 (564) Poor 1,653.1 ± 185.0 (15) 1,416.4 ± 312.0 (9) 1,564.3 ± 162.2 (24) Wind (theoretical) Calm 1,226.1 ± 137.3 (51) 1,219.9 ± 92.1 (39) 1,223.5 ± 87.0 (90) Crosswind 1,244.1 ± 95.7 (75) 1,396.4 ± 100.4 (70) 1,317.6 ± 69.3 (145) Headwind 1,327.9 ± 61.4 (129) 1,056.4 ± 72.5 (104) 1,206.7 ± 47.7 (233) Tailwind 1,511.9 ± 108.9 (54) 1,258.9 ± 96.3 (66) 1,372.8 ± 72.8 (120) Wind strength Weak 1,310.9 ± 47.2 (291) 1,207.8 ± 47.3 (265) 1,261.7 ± 33.5 (556) Strong 1,517.5 ± 171.8 (18) 1,301.1 ± 181.4 (14) 1,422.8 ± 124.6 (32) Total Total 1,322.9 ± 45.6 (309) 1,212.5 ± 45.8 (279) 1,270.5 ± 32.4 (588)

Results


Table 17. Characteristics of records of “eider” and “non-eider” targets on ornithological radar that passed over Northstar Island, northern Alaska, fall 2001–2004, by species-group.

Species-group

Characteristic "Eiders" (n = 5) "Non-eiders" (n = 21)

Year 2001 (80.0%); 2004 (20.0%) 2001 (61.9%); 2002 (14.3%); 2003 (23.8%)

Ice status ice present (80.0%); ice absent (20.0%)

ice present (61.9%); ice absent (38.1%)

Time of day night (100.0%) day (28.6%); night (71.4%) Precipitation level no precipitation (100.0%) no precipitation (95.2%);

precipitation (4.8%) Session visibility good (100.0%) good (100.0%) Wind direction tailwind (100.0%) calm (4.8%); crosswind (19.0%);

headwind (28.6%); tailwind (47.6%) Wind strength strong (20.0%): weak (80.0%) strong (19.0%); weak (81.0%) Lights off (100.0%) off (52.4%); on (47.6%) Mean velocity (mi/h) 48.8 ± SE 1.2 (n = 4) 31.6 ± 2.1 (n = 20) Flight behavior straight-line (100.0%) straight-line (100.0%) Passing success (%) successful (100.0%) successful (100.0%) Mean cloud cover (%) 30.0 ± SE 10.6 (n = 4) 60.3 ± 8.2 (n = 20) Ceiling height (m) 1,001–2,500 (100.0%) 1–500 (9.5%); 501–1,000 (9.5%);

1,001–2,500 (71.4%); clear sky (9.5%)

Moon visible? yes (80.0%); no (20.0%) yes (52.4%); no (47.6%) Illumination of lunar disk not full (100.0%) not full (47.6%); full (52.4%)

Results


without precipitation, during good visibility, duringperiods of tailwinds, when the anti-collision lightswere off, during moderate ceiling heights, andwhen the moon was not full. Most were recorded in2001 (and, hence, when ice was present), duringweak winds, during light cloud cover, and when themoon was visible. In addition, they flew atabout-average velocities and with straight-lineflight, and all were successful at passing the island.

Over the 4 years, we recorded 21 “non-eider”targets passing over Northstar Island (Table 17). Ofthese, all were recorded during good visibility.Most were recorded in 2001 (and, hence, when icewas present), at night, during periods withoutprecipitation, during tailwinds (but also occurringwith all other wind directions), and during highcloud cover (but occurring during all levels ofcover). In addition, they flew at above-averagevelocities and with straight-line flight, and all weresuccessful at passing the island. They wererecorded in almost identical proportions byanti-collision light setting, by when the moon wasvisible or not, and by lunar phase.

SPATIAL DISTRIBUTION“Eiders” exhibited nonrandom movements

through the study area, both with the anti-collisionlights off and with them on and both with icepresent and ice absent (Figs. 14–15, Appendices2–4, with the data for 2001 being presented inFig. 14). (In these figures, we are interested only inthe eastern part of the radar screen, which is wheremost birds were originating and where we couldget a complete survey with the radar. Hence, thewestern half of the radar screen is left blank.)When ice was present, there was a largeconcentration of radar targets northeast of theisland that was moving toward the northwest and asmaller band south of the island that was movingtoward the northwest; considerably lower numbersof targets flew near the island and out near the edgeof the radar screen (Fig. 14). When ice was absent,there also was a concentration of radar targetsnortheast of the island, although it was not asnarrow and concentrated as it was when ice waspresent; it also appeared to be more concentratedwhen the lights were on than when they were off(Fig. 15).

Figure 13. Statistical power to detect a specified change in mean passing distance of “eiders” on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004 combined. The two numbers for a line refer to sample sizes in “lights off” and “lights on” categories; the set with sample sizes of 457 and 466 refers to sample sizes for this study.

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

50 100 150 200 250 300

EFFECT SIZE (M)

PO

WE

R

100, 100

200, 200

300, 300

457, 466

500, 500

750, 750

1000, 1000

1250, 1250

Results


Figure 14. Movement density (targets/km²/h) of “eiders” on ornithological radar near Northstar Island, northern Alaska, during a period with ice present (2001), by anti-collision lighting setting. Grid cell size is 250 m × 250 m.

Lights Off

4ABR file: Eiders_Ice_05_153.apr; 13 April 2005

"Eiders"

Northstar

Island

Northstar

Island

500 0 500 1,000 1,500 2,000 Meters

Ice Present

Movement Rates(targets/km

2/h)

0 – 1.0

1.0 – 3.0

3.0 – 5.0

5.0 – 7.0

7.0 – 9.0

9.0 – 11.0

> 11.0

Lights On

Results


Figure 15. Movement density (targets/km²/h) of “eiders” on ornithological radar near Northstar Island, northern Alaska, during periods with ice absent (2002–2004), by anti-collision lighting setting. Grid cell size is 250 m × 250 m. Data are presented by year in Appendices 2–4.

Lights Off

4ABR file: Eiders_NoIce_05_153.apr; 13 April 2005

"Eiders"

Northstar

Island

Northstar

Island

500 0 500 1,000 1,500 2,000 Meters

Ice Absent


2/h)

0

0.0 – 0.5

0.5 – 1.0

1.0 – 1.5

1.5 – 2.0

2.0 – 2.5

>2.5

Lights On

Results


The Before–After analysis for “eiders”suggested that turning on the anti-collision lightsgenerally resulted in net decreases in movementdensity near the island, whereas it generallyresulted in net increases far from the island,especially north of the island and especially whenice was present (Fig. 16). When ice was present,turning on the lights resulted in a significantlypositive slope to a line fitted to the points (Fig. 17),indicating an avoidance of the island and asignificant movement of “eiders” away from theisland when the lights were on (Y = –0.645 +0.0004X; R² = 0.041; P = 0.001). The regressionline crossed a Before–After difference of zero at adistance of ~1,550 m from the island, suggesting atleast some avoidance of the island out to thatdistance. When ice was absent, however, turningon the anti-collision lights resulted in a veryweakly (but not significant) positive slope to a linefitted to the points (Fig. 17), indicating no effect ofthe lights on spatial distribution (Y = 0.182 +0.000002 X; R² = 0.000006; P = 0.969).

“Non-eiders” also exhibited nonrandommovements through the study area, both with theanti-collision lights off and with them on and bothwith ice present and ice absent (Figs. 18–19,Appendices 5–7, with the data for 2001 beingpresented in Fig. 18). When ice was present, thelarge band of concentrated movement near theisland represented large numbers of circling birdsseen on some nights; movement densities droppedoff with increasing distance from the island, and itactually appeared that having the lights onexacerbated the problem. When ice was absent,there was a fairly similar patter of concentrationnear the island, especially when the lights were on,although it was not as pronounced as theconcentrated zone seen when ice was present;when the lights were off, it almost appeared thatthere was a concentration of movement toward thenorthwest on the seaward side of the island, similarto that generally seen for “eiders.”

The Before–After analysis for “non-eiders”suggested that turning on the anti-collision lightsgenerally resulted in net increases in movementdensity near the island, whereas it generallyresulted in net decreases far from the island,especially when ice was present (Fig. 20). Whenice was present, turning on the lights resulted in anon-significantly negative slope to a line fitted to

the points (Fig. 21), suggesting a weak attractiontoward the island and a movement of “non-eiders”toward the island when the lights were on(Y = 1.836 – 0.0003X; R² = 0.014; P = 0.054).When ice was absent, turning on the lights alsoresulted in a non-significantly negative slope to aline fitted to the points (Fig. 21), suggesting a weakattraction toward the island and a movement of“non-eiders” toward the island when the lightswere on (Y = 0.175 – 0.0001X; R² = 0.006;P = 0.205).

HIGH-RESOLUTION VARIATIONS IN SPATIAL DISTRIBUTION

“Eiders” exhibited variable numbers of coursechanges (vertices) ≥5° at all distances from theisland. When the data are grouped by distance fromthe island, however, it is clear that there are fewervertices/km of line with increasing distance(Fig. 22), suggesting that the “eiders” are seeingthe island and responding to it naturally by turningmore often as they approach it (presumably toavoid collision). Mean numbers of vertices/km forthe innermost two distance zones (0–499 m and500–999 m) were significantly different from thosefor the outermost two zones (1,500–1,999 m and>1,999 m; P = 0.003 for 0–499 vs. 1,500–1,999;P < 0.001 for other comparisons). Within theinnermost distance zone, the mean number ofvertices/km appeared to be higher with theanti-collision lights on than with them off;however, the 95% CIs from the bootstrappedestimates overlapped, indicating that there was nota significant effect of the lights on the meannumber of vertices/km (P = 0.192). In fact, lightshad no significant effect on the mean number ofvertices/km in any distance category. Hence,“eiders” were responding to the island by having agreater number of course changes (vertices) as theyapproached it, but the anti-collision lights were notsignificantly increasing that responsiveness in anydistance category.

We also conducted a bootstrap analysis thatcompared the effects of the anti-collision lights onthe mean number of vertices/km between thosetargets whose original flightlines were going topass <500 m from the island and those thatoriginally were going to pass ≥500 m from theisland (Fig. 23). There was no difference in theeffect of anti-collision light settings on the mean

Results


Figure 16. Net difference in movement density (targets/km²/h) of “eiders” on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, between periods and anti-collision lighting settings. Data are calculated as density (lights on) – density (lights off).

Ice Present

Ice Absent

4ABR file: Diff_Eiders_Ice_05-153.mxd; 13 April 2005

"Eiders"

Northstar

Island

Northstar

Island

Difference BetweenLights On and

Lights Off

500 0 500 1,000 1,500 2,000 Meters


2/h)

-7.0 – -2.0

-2.0 – -1.0

-1.0 – -0.5

-0.5 – 0.5

0.5 – 1.0

1.0 – 2.0

2.0 – 7.0

Results


Figure 17. Regression of net difference between anti-collision lighting settings in movement density (targets/km²/h) of “eiders” on ornithological radar near Northstar Island, northern Alaska, by period with ice present (top) and with ice absent (bottom), by distance of the grid cell from the island. The slope of the upper line is significantly different from zero, but the slope of the bottom line is not.

ICE PRESENT

-8

-6

-4

-2

0

2

4

6

0 500 1,000 1,500 2,000 2,500 3,000 3,500

DIF

FE

RE

NC

E (

TA

RG

ET

S/K

M2/H

)

ICE ABSENT

-3

-2

-1

0

1

2

3

4

0 500 1,000 1,500 2,000 2,500 3,000 3,500

DISTANCE FROM ISLAND (M)

DIF

FE

RE

NC

E (

TA

RG

ET

S/K

M2

/H)

DIF

FE

RE

NC

E (

TA

RG

ET

S/K

M2

/H)

Results


Figure 18. Movement density (targets/km²/h) of “non-eiders” on ornithological radar near Northstar Island, northern Alaska, during a period with ice present (2001), by anti-collision lighting setting. Grid cell size is 250 m × 250 m.

Lights Off

4ABR file: NonEiders_Ice_05_153.apr; 13 April 2005

"Non-Eiders"

Northstar

Island

Northstar

Island

500 0 500 1,000 1,500 2,000 Meters

Ice Present


2/h)

0 – 3.0

3.0 – 5.0

5.0 – 7.0

7.0 – 9.0

9.0 – 11.0

11.0 – 13.0

> 13.0

Lights On

Results


Figure 19. Movement density (targets/km²/h) of “non-eiders” on ornithological radar near Northstar Island, northern Alaska, during periods with ice absent (2002–2004), by anti-collision lighting setting. Grid cell size is 250 m × 250 m. Data are presented by year in Appendices 5–7.

Lights Off

4ABR file: NonEiders_NoIce_05_153.apr; 13 April 2005

"Non-Eiders"

Northstar

Island

Northstar

Island

500 0 500 1,000 1,500 2,000 Meters

Ice Absent


2/h)

0 – 2.0

2.0 – 3.0

3.0 – 4.0

4.0 – 5.0

5.0 – 6.0

6.0 – 7.0

> 7.0

Lights On

Results


Figure 20. Net difference in movement density (targets/km²/h) of “non-eiders” on ornithological radar near Northstar Island, northern Alaska, fall 2001–2004, between periods and anti-collision lighting settings. Data are calculated as density (lights on) – density (lights off).

Ice Present

Ice Absent

4ABR file: Diff_NonEiders_Ice_05-153.mxd; 13 April 2005

"Non-Eiders"

Northstar

Island

Northstar

Island

Difference BetweenLights On and

Lights Off

500 0 500 1,000 1,500 2,000 Meters


2/h)

-7.0 – -2.0

-2.0 – -1.0

-1.0 – -0.5

-0.5 – 0.5

0.5 – 1.0

1.0 – 2.0

2.0 – 7.0

Results


Figure 21. Regression of net difference between anti-collision lighting settings in movement density (targets/km²/h) of “non-eiders” on ornithological radar near Northstar Island, northern Alaska, by period with ice present (top) and with ice absent (bottom), by distance of the grid cell from the island. The slopes of these lines are not significantly different from zero.

ICE PRESENT

-8

-6

-4

-2

0

2

4

6

8

0 500 1,000 1,500 2,000 2,500 3,000 3,500

DIF

FE

RE

NC

E (

TA

RG

ET

S/K

M2

/H)

ICE ABSENT

-4

-3

-2

-1

0

1

2

3

4

0 500 1,000 1,500 2,000 2,500 3,000 3,500


DIF

FE

RE

NC

E (

TA

RG

ET

S/K

M2

/H)

DIF

FE

RE

NC

E (

TA

RG

ET

S/K

M2

/H)

Results


Figure 22. Mean numbers of “eider” trackline vertices (course changes) ≥5°/km of line within that distance zone for all lines (top) and mean angle changed for all vertices ≥5° (bottom) near Northstar Island, northern Alaska, fall 2001–2004 combined, by distance from the island and anti-collision lighting setting. Vertical lines represent 95% confidence intervals.

0.0

0.4

0.8

1.2

1.6

2.0

0-499 500-999 1,000-1,499 1,500-1,999 >1,999

VE

RT

ICE

S/K

M O

F L

INE

LIGHTS OFF LIGHTS ON

5

10

15

20

25

0-499 500-999 1,000-1,499 1,500-1,999 >1,999


AN

GU

LA

R C

HA

NG

E (

°)

Results


number of vertices/km for targets originallyheading <500 m from the island (P = 0.873) or fortargets originally heading ≥500 m from the island(P = 0.052, but α was 0.0125 after Bonferroniadjustments). There also was no difference in themean number of vertices/km between targetsoriginally heading <500 m from the island andthose originally heading ≥500 m from the islandwith the lights off (P = 0.085), but there weresignificantly more vertices/km for targetsoriginally heading <500 m from the island thanthose originally heading ≥500 m from the islandwith the lights on (P < 0.001). Hence, the meannumber of vertices/km did not differ between lightsettings at either nearest-distance category andbetween distance categories when the lights wereoff. This result implies that birds originally headingtoward the island changed course more often thanbirds not heading toward it but that the number ofcourse changes occurred independently of lightsetting.

A third way of examining possible effects ofthe anti-collision lights is to determine their effecton the distance from the island when course

changes occurred (i.e., vertex distances). Thebest-approximating model describing the vertexdistance of “eiders” included the parameterssession visibility (Σwi = 0.998), wind direction(Σwi = 1.000), lights (Σwi = 1.000), and theinteractions lights*visibility (Σwi = 0.997) andlights*wind direction (Σwi = 0.861; Tables 18–20).This model had an Akaike Weight of 0.288 (Table18), which was just slightly higher than that of thesecond-best model (0.268). Vertex distances weresignificantly smaller with poor visibility, weresignificantly smaller with calm winds thancrosswinds and tailwinds but not headwinds, weresignificantly larger when lights were on than offwhen visibility was good but were significantlylarger when lights were off than on when visibilitywas poor, and were significantly larger with thelights off than on during tailwinds but not differentby light settings for calm winds, crosswinds, andheadwinds (Table 20). Vertex distances did notdiffer significantly by time of day or winddirection*wind strength.

“Eiders” also exhibited variably-sized angularchanges for vertices ≥5° at all distances from the

Figure 23. Mean numbers of “eider” trackline vertices (course changes) ≥5°/km of line within that distance zone for all lines near Northstar Island, northern Alaska, fall 2001–2004 combined, by distance from the island and anti-collision lighting setting. Vertical lines represent 95% confidence intervals.

0.4

0.6

0.8

1.0

1.2

1.4

0-499 500-4,000

ORIGINAL DISTANCE FROM ISLAND (M)

VE

RT

ICE

S/K

M O

F L

INE


Results

AB

R F

inal Report

75M


Table 18. Significance of factors affecting the distance of vertices (course changes) >5°, the mean angular change for vertices >5°, and the difference between original (without course changes) and actual nearest-passing distances of "eiders" migrating near Northstar Island, northern Alaska, fall 2001–2004. All models examined the effects of the factors year, time of day, session visibility, wind (theoretical), and lights on the response variable. For each response variable and species-group, these models have a have a ΔAICc of ≤2.

Response variable Model RSSa nb Kc AICcd Δie wi

f

Vertex distance Visibility, wind direction, lights, lights*visibility,

lights*wind direction 234,540,721 836 11 10,509.5 1.00 0.288

Time of day, visibility, wind direction, lights, lights* visibility, lights*wind direction 234,003,239 836 12 10,509.7 0.93 0.268

Time of day, visibility, wind direction, lights, lights*time of day, lights* visibility, lights*wind direction 233,631,885 836 13 10,510.4 0.65 0.185

Angular change Intercept 254 836 2 –992.4 1.00 0.294 Distance 254 836 3 –990.6 0.41 0.121 Actual – original distance Time of day, visibility, wind direction, wind strength,

lights, wind direction*wind strength, lights*wind direction 21,363,658 351 14 3,896.0 1.00 0.211

Visibility, wind direction, wind strength, lights, wind direction*wind strength, lights*wind direction 21,531,067 351 13 3,896.6 0.75 0.158

Time of day, visibility, wind direction, wind strength, wind direction*wind strength 21,999,940 351 10 3,897.7 0.43 0.090

a Residual Sum of Squares. b Sample size. c Number of estimatable parameters in the approximating model. d Akaike’s Information Criterion corrected for small sample size. e Difference in value between AICc of the current model and that of the best approximating model (AICcmin). f Akaike weight—probability that the current model (i) is the best approximating model among those considered.

Results


Table 19. Sum of Akaike Weights for the model parameters in candidate models for each response variable.

Table 20. Model-weighted parameter estimates for factors affecting the distance of vectors (course changes) >5°, the mean angular change for vertices >5°, and the difference between original and actual passing distances of “eiders” migrating near Northstar Island, northern Alaska, fall 2001–2004.

Response variable

Model parameter Vertex distance Angular change Actual – original distance

Distance from island – 0.706 –

Time of day 0.635 0.245 0.679

Session visibility 0.998 0.223 0.866

Wind direction 1.000 0.255 0.934

Wind strength 0.133 0.148 0.967

Lights 1.000 0.418 0.856

Lights*time of day 0.257 0.122 0.164

Lights*visibility 0.997 0.038 0.194 Lights*wind direction 0.861 0.186 0.711

Response variable Model parameter Estimate SE P

Vertex distance Intercept 1,013.583 99.362 <0.001 Daytime –48.131 62.322 0.440 Visibility good 333.420 92.671 <0.001 Calm –146.311 177.842 0.411 Crosswind 195.807 73.293 0.008 Headwind 56.228 90.203 0.533 Wind strong 82.833 59.330 0.163 Strong/crosswind 82.349 278.162 0.767 Strong/headwind –66.381 391.224 0.865 Lights off 470.380 143.974 0.001 Lights off/daytime –106.434 94.590 0.260 Lights off/visibility good –526.106 132.176 <0.001 Lights off/calm wind –413.447 185.019 0.025 Lights off/crosswind –222.261 89.271 0.013 Lights off/headwind –185.296 150.785 0.219

Results


island, with what appeared to be an increase inmean angular change as they approached the island(Fig. 22, Tables 18–20). The best-approximatingmodel describing angular changes of “eiders” hadno independent variables, indicating that none ofthese factors had particularly strong explanatorypower (Tables 18 and 19). This model had anAkaike Weight of 0.294 (Table 18), or substantiallyhigher than that of the second-best model (0.121).Hence, when course changes did occur, none of the

factors had a significant effect on the mean angularchange.

Another issue that needs to be examined isthat of where the birds would have passed withrespect to the island versus where they actually didpass because of course changes. In the innermostdistance zone (i.e., 0–499 m), slightly fewerflightlines actually passed the island than therewould have been had course changes not occurred,suggesting possibly very weak avoidance at thisdistance (Fig. 24); however, this pattern did not


Response variable Model parameter Estimate SE P

Angular change Intercept 2.628 0.057 <0.001 Distance from islanda –0.017 0.035 0.630 Daytime 0.033 0.052 0.522 Visibility good –0.020 0.069 0.775 Calm 0.197 0.171 0.248 Crosswind –0.013 0.065 0.843 Headwind –0.064 0.111 0.562 Wind strong –0.008 0.060 0.891 Strong/crosswind 0.145 0.288 0.615 Strong/headwind 1.026 0.406 0.011 Lights off –0.055 0.056 0.321 Lights off/daytime –0.019 0.038 0.625 Lights off/visibility good –0.015 0.130 0.906 Lights off/calm –0.227 0.185 0.220 Lights off/crosswind 0.068 0.091 0.454 Lights off/headwind 0.423 0.157 0.007

Actual – original distance Intercept 123.549 72.988 0.091 Daytime 68.097 41.457 0.100 Visibility good –131.112 60.616 0.031 Calm wind –155.565 127.747 0.223 Crosswind –85.275 63.378 0.178 Headwind –23.730 73.049 0.745 Wind strong –37.792 39.400 0.337 Strong/crosswind –653.441 187.899 0.001 Strong/headwind 3.733 197.384 0.985 Lights off –77.298 57.618 0.180 Lights off/daytime –37.934 61.521 0.538 Lights off/visibility good 24.398 107.868 0.821 Lights off/calm wind 164.222 122.550 0.180 Lights off/crosswind 159.914 60.452 0.008 Lights off/headwind –4.236 104.905 0.968

a Although distances were in meters, we converted these values to km so that numbers other than zero would be present here.

Results


differ between lighting settings, indicating that thelights were not having a significant effect on thischange. At a distance of 500–999 m, there weresubstantially more flightlines than originallyexpected with the lights off but slightly fewer withthem on. At a distance of 1,000–1,499 m, therewere slightly fewer flightlines than expected with

the lights off but substantially more than expectedwith the lights on. At 1,500–1,999 m, numberswere essentially identical, whereas numbers atdistances >1,999 m suggested that substantialnumbers of flightlines were changing course toapproach the island more closely. Theanti-collision lights had little effect on these overall

Figure 24. Number of original (before any course changes) and actual “eider” tracklines passing within each distance zone near Northstar Island, northern Alaska, fall 2001–2004 combined, by distance from the island and anti-collision lighting setting.

LIGHTS ON

0

20

40

60

80

100

120

0-499 500-999 1,000-1,499 1,500-1,999 >1,999

CLOSEST DISTANCE (M)

NU

MB

ER

OF

TA

RG

ET

SLIGHTS OFF

0

20

40

60

80

100

120

0-499 500-999 1,000-1,499 1,500-1,999 >1,999

NU

MB

ER

OF

TA

RG

ET

SORIGINAL ACTUAL

Results


patterns, with the exception that the mostflightlines passed within 500–999 m of the islandwhen the lights were off but that most passedwithin 1,000–1,499 m of the island when the lightswere on, indicating a shift of birds farther awayfrom the island when the lights were on. Thispattern matches the general net spatial shift awayfrom the island when the lights are on (Fig. 17);also see “Passing Distance,” above).

Another way to examine these data is tosubtract the original passing distance from theactual passing distance and to examine theresulting spatial patterns (Fig. 25). In this figure, apositive net change represents a net movementaway from the island after a course change,whereas a negative number represents a netmovement toward the island after a course change.The mean net change in the innermost distancezone was away from the island (by an average of~155 m), whereas, at greater distances, it was about

zero (500–999 m distance zone) or negative(distances beyond that), suggesting no effect orslight net movement toward the island at greaterdistances. Although there was a net movementaway from the island in the innermost distancezone, the lights had no significant effect on thisresponse at any distance (P ≥ 0.180 for all). Oneword of caution should be said about these results,however. Because a target originally heading nearthe island could change course only a smalldistance to cross the island, it also could changecourse to pass much farther away, potentiallybiasing these results. For example, a target thatoriginally would have passed 50 m from the islandcould have made net course changes of between–50 m (i.e., it changed course to cross over theisland) and +2,770 m (i.e., it changed course tocross at the edge of the radar screen). Hence,especially near the island, the potential changes arenot symmetrical around zero. Nevertheless,

Figure 25. Mean changes in nearest distances of “eider” flightlines near Northstar Island, northern Alaska, fall 2001–2004 combined, by distance from the island and anti-collision lighting setting.

-150

-100

-50

0

50

100

150

200

250

0-499 500-999 1,000-1,499 1,500-1,999 >1,999

DISTANCE OF ORIGINAL HEADING FROM ISLAND (M)

PA

SS

ING

CH

AN

GE

(A

CT

UA

L –

OR

IGIN

AL

) .


Results


negative net changes still are possible for targetsnear the island. Because the statistical distributionof net changes in this innermost zone indicated thatnearly all net changes were either zero or positive,with only a few targets heading closer to the island,we conclude that avoidance truly is occurring butthat the anti-collision lights have no significanteffect on that net change.

We also examined factors affecting this netchange. The best-approximating model describingthe net change in passing distances (actual –original distance) of “eiders” included theparameters time of day (Σwi = 0.679), visibility(Σwi = 0.866), wind direction (Σwi = 0.934), winddirection*wind strength (Σwi = 0.967), lights (Σwi= 0.856), and the interaction lights*wind direction(Σwi = 0.711); wind direction, wind strength, andlights had to be in the model because theinteraction terms were significant (Tables 18–20).This model had an Akaike Weight of 0.211, or onlyslightly better than that of the second-best model(0.158). The net change in passing distance wassignificantly larger during poor visibility,significantly larger with weak crosswinds thanstrong crosswinds but did not differ by windstrength for other wind directions, significantlylarger for strong tailwinds than strong crosswindsbut did not differ among wind directions duringweak winds, and significantly larger when thelights were on than off in tailwinds but did notdiffer by light setting for other wind directions(Table 20). Although time of day was in the bestmodel, the parameter estimate was not significant(Table 20). Hence, the anti-collision lights had aninconsistent effect on net passing distance, beingsignificant in only one wind-direction category.

VISUAL

During the 80 days of visual sampling, werecorded 1,075 flocks (1 bird or ≥2 birds flyingtogether) totaling 17,040 birds (Appendix 8). Werecorded these as either individual species orspecies-groups, depending on how well we couldsee them. The species-group seen most often wasgulls, with 261 flocks totaling 649 birds; loonswere second in frequency, with 215 flocks totaling421 birds. Unidentified ducks were third infrequency, with 187 flocks totaling 7,779 birds.Other ducks were fourth in frequency, with 184

flocks totaling 4,260 birds. Eiders were fifth infrequency, with 109 flocks totaling 2,558 birds.Shorebirds were sixth in frequency, with 69 flockstotaling 207 birds. Alcids consisted of 14 flockstotaling 22 birds. Finally, we saw 1,144unidentified birds, other birds, cormorants, andswans/geese in 36 flocks, most at a great distance;these data are not presented here. Becausesample-sizes for all species-groups are too small inmost cases for statistical analyses, we simply willdiscuss overall patterns.

FLOCK SIZEFlock sizes of eiders ranged between 1 and

160 birds, with an overall mean of 23.4 birds/flock(Table 21). Flock sizes appeared to be larger withthe anti-collision lights off than with them on,during the daytime (although there were only sixnighttime samples), during no precipitation, duringgood visibility (although sample sizes for poorvisibility were small), during tailwinds andcrosswinds (sample sizes for calm winds weresmall), and with weak winds. In all cases, however,the SEs were so large that the differences wouldnot have been statistically significant.

Flock sizes of loons, shorebirds, gulls, andalcids averaged 1–3 birds; because of the smallflock sizes, there was not enough variation to seeany dramatic patterns with respect to visibilitycategory and lights (Table 21). The one suggestionof a pattern was that flock sizes of loons andshorebirds were larger with tailwinds, crosswinds,and/or calm winds than with headwinds, similar tothe pattern seen for eiders. Other ducks averaged23.2 birds/flock, and unidentified ducks averaged41.6 birds/flock. Both other ducks and unidentifiedducks averaged much larger flock sizes in thedaytime than at night, with no precipitation thanwith precipitation, and with good visibility thanwith poor visibility; they also averaged the largestflock sizes with tailwinds and strong winds.

FLIGHT ALTITUDEEiders flew at a mean flight altitude of ~6 m

agl/asl, with a range of 1–50 m agl/asl (Table 22).Mean altitudes were slightly higher with theanti-collision lights off than with them on, althoughthe difference was only ~1 m; the only dramaticpattern with reasonable sample sizes was that flightaltitudes were highest during tailwinds. Again,

Results

AB

R F

inal Report

81M


Table 21. Mean flock size of birds seen visually near Northstar Island, northern Alaska, fall 2001–2004, by species-group, factor, and anti-collision lighting setting. Data are presented as mean ± SE (n flocks) and range. Because of small sample sizes, cormorants, swans/geese, other birds, and unidentified birds are not analyzed.

Lights

Off On Total

Species-group Factor Attribute Mean ± SE (n) Range Mean ± SE (n) Range Mean ± SE (n) Range

Eiders Time of day Daytime 24.8 ± 4.6 (68) 1–160 22.5 ± 5.2 (35) 1–130 24.0 ± 3.5 (103) 1–160 Nighttime 18.0 ± 16.0 (3) 1–50 10.0 ± 5.1 (3) 3–20 14.0 ± 7.7 (6) 1–50 Precipitation level No precipitation 25.8 ± 6.4 (44) 1–160 23.0 ± 6.2 (27) 1–130 24.7 ± 4.6 (71) 1–160 Precipitation 22.4 ± 5.5 (27) 1–100 17.9 ± 7.4 (11) 1–75 21.1 ± 4.4 (38) 1–100 Session visibility Good 25.1 ± 4.5 (69) 1–160 20.6 ± 5.0 (35) 1–130 23.6 ± 3.5 (104) 1–160 Poor 4.5 ± 1.5 (2) 3–6 32.7 ± 21.7 (3) 3–75 21.4 ± 13.8 (5) 3–75 Wind (theoretical) Calm 76.5 ± 73.5 (2) 3–150 75.0 ± – (1) 75–75 76.0 ± 42.4 (3) 3–150 Crosswind 32.6 ± 8.0 (21) 1–150 14.3 ± 12.0 (4) 1–50 29.7 ± 7.0 (25) 1–150 Headwind 11.1 ± 7.3 (10) 1–75 4.5 ± 2.0 (13) 1–27 7.4 ± 3.3 (23) 1–75 Tailwind 20.8 ± 5.7 (38) 1–160 31.4 ± 7.7 (20) 3–130 24.4 ± 4.6 (58) 1–160 Wind strength Weak 35.2 ± 11.8 (19) 1–150 16.9 ± 5.3 (18) 1–75 26.3 ± 6.7 (37) 1–150 Strong 20.6 ± 4.2 (52) 1–160 25.7 ± 8.0 (20) 1–130 22.0 ± 3.7 (72) 1–160 Total Total 24.5 ± 4.4 (71) 1–160 21.5 ± 4.9 (38) 1–130 23.4 ± 3.3 (109) 1–160 Loons Time of day Daytime 2.5 ± 0.4 (103) 1–30 1.5 ± 0.1 (112) 1–7 2.0 ± 0.2 (215) 1–30 Nighttime – ± – (0) – – ± – (0) – – ± – (0) – Precipitation level No precipitation 2.5 ± 0.5 (102) 1–30 1.5 ± 0.1 (90) 1–7 2.0 ± 0.2 (192) 1–30 Precipitation 2.0 ± – (1) 2–2 1.3 ± 0.3 (22) 1–7 1.3 ± 0.3 (23) 1–7 Session visibility Good 2.5 ± 0.4 (103) 1–30 1.5 ± 0.1 (111) 1–7 2.0 ± 0.2 (214) 1–30 Poor – ± – (0) – 1.0 ± – (1) 1–1 1.0 ± – (1) 1–1 Wind (theoretical) Calm 1.1 ± 0.1 (8) 1–2 1.6 ± 0.4 (17) 1–7 1.5 ± 0.2 (25) 1–7 Crosswind 1.1 ± 0.1 (23) 1–2 1.3 ± 0.3 (19) 1–7 1.2 ± 0.1 (42) 1–7 Headwind 1.8 ± 0.2 (50) 1–6 1.6 ± 0.2 (55) 1–7 1.7 ± 0.1 (105) 1–7 Tailwind 5.9 ± 1.9 (22) 1–30 1.2 ± 0.1 (21) 1–2 3.6 ± 1.0 (43) 1–30 Wind strength Weak 1.5 ± 0.2 (41) 1–6 1.7 ± 0.2 (76) 1–7 1.6 ± 0.1 (117) 1–7 Strong 3.1 ± 0.7 (62) 1–30 1.1 ± 0.1 (36) 1–2 2.4 ± 0.5 (98) 1–30 Total Total 2.5 ± 0.4 (103) 1–30 1.5 ± 0.1 (112) 1–7 2.0 ± 0.2 (215) 1–30

Results

Migration at N

orthstar Island82

AB

R F

inal Report


Lights

Off On Total


Other ducks Time of day Daytime 26.6 ± 6.3 (65) 1–350 26.7 ± 5.3 (72) 1–220 26.7 ± 4.1 (137) 1–350 Nighttime 12.3 ± 3.6 (25) 1–75 13.5 ± 4.2 (22) 1–80 12.9 ± 2.7 (47) 1–80 Precipitation level No precipitation 27.1 ± 6.0 (69) 1–350 24.5 ± 4.4 (75) 1–200 25.8 ± 3.6 (144) 1–350 Precipitation 7.9 ± 2.1 (21) 1–35 20.1 ± 11.7 (19) 1–220 13.7 ± 5.7 (40) 1–220 Session visibility Good 24.1 ± 5.0 (84) 1–350 22.8 ± 4.1 (93) 1–220 23.4 ± 3.2 (177) 1–350 Poor 2.5 ± 0.7 (6) 1–5 105 ± – (1) 105–105 17.1 ±14.7 (7) 1–105 Wind (theoretical) Calm 20.2 ± 16.3 (5) 1–85 9.5 ± 5.2 (4) 3–25 15.4 ± 9.0 (9) 1–85 Crosswind 10.9 ± 4.8 (9) 1–40 20.9 ± 8.8 (27) 1–220 18.4 ± 6.7 (36) 1–220 Headwind 11.2 ± 2.4 (40) 1–75 17.6 ± 4.2 (45) 1–125 14.6 ± 2.5 (85) 1–125 Tailwind 38.7 ± 10.7 (36) 1–350 45.8 ± 13.1 (18) 1–200 41.1 ± 8.3 (54) 1–350 Wind strength Weak 15.9 ± 3.4 (36) 1–85 23.5 ± 5.3 (55) 1–220 20.5 ± 3.5 (91) 1–220 Strong 27.1 ± 7.4 (54) 1–350 23.8 ± 6.8 (39) 1–200 25.7 ± 5.2 (93) 1–350 Total Total 22.7 ± 4.7 (90) 1–350 23.6 ± 4.2 (94) 1–220 23.2 ± 3.1 (184) 1–350 Unid. ducks Time of day Daytime 66.7 ± 10.0 (58) 1–260 38.7 ± 6.3 (97) 1–370 49.8 ± 5.5 (155) 1–370 Nighttime 5.0 ± 1.3 (21) 1–30 4.3 ± 2.1 (11) 1–24 4.8 ± 1.1 (32) 1–30 Precipitation level No precipitation 53.5 ± 8.4 (74) 1–260 39.7 ± 6.6 (92) 1–370 45.8 ± 5.2 (166) 1–370 Precipitation 3.8 ± 1.2 (5) 1–7 9.3 ± 3.3 (16) 1–50 8.0 ± 2.6 (21) 1–50 Session visibility Good 51.5 ± 8.1 (77) 1–260 35.3 ± 5.8 (107) 1–370 42.1 ± 4.8 (184) 1–370 Poor 4.0 ± 3.0 (2) 1–7 24.0 ± – (1) 24–24 10.7 ± 6.9 (3) 1–24 Wind (theoretical) Calm 1.0 ± – (1) 1–1 8.0 ± 5.5 (5) 1–30 6.8 ± 4.7 (6) 1–30 Crosswind 34.3 ± 17.1 (12) 1–200 7.7 ± 4.2 (12) 1–50 21.0 ± 9.0 (24) 1–200 Headwind 17.3 ± 12.3 (20) 1–250 27.2 ± 12.4 (36) 1–370 23.7 ± 9.1 (56) 1–370 Tailwind 69.9 ± 10.9 (46) 1–260 48.9 ± 7.3 (55) 1–300 58.5 ± 6.4 (101) 1–300 Wind strength Weak 19.8 ± 12.4 (20) 1–250 30.5 ± 12.7 (35) 1–370 26.6 ± 9.2 (55) 1–370 Strong 60.7 ± 9.4 (59) 1–260 37.5 ± 6.0 (73) 1–300 47.8 ± 5.4 (132) 1–300 Total Total 50.3 ± 7.9 (79) 1–260 35.2 ± 5.7 (108) 1–370 41.6 ± 4.7 (187) 1–370

Results

AB

R F

inal Report

83M



Lights

Off On Total


Shorebirds Time of day Daytime 1.0 ± 0 (2) 1–1 3.7 ± 2.2 (3) 1–8 2.6 ± 1.3 (5) 1–8 Nighttime 2.9 ± 1.0 (25) 1–25 3.1 ± 1.1 (39) 1–30 3.0 ± 0.7 (64) 1–30 Precipitation level No precipitation 3.3 ± 1.6 (15) 1–25 2.9 ± 0.9 (26) 1–25 3.0 ± 0.8 (26) 1–25 Precipitation 2.1 ± 0.5 (12) 1–7 3.6 ± 1.8 (16) 1–30 2.9 ± 1.1 (28) 1–30 Session visibility Good 2.7 ± 1.0 (25) 1–25 3.1 ± 0.9 (42) 1–30 3.0 ± 0.7 (67) 1–30 Poor 4.0 ± 3.0 (2) 1–7 – ± – (0) – 4.0 ± 3.0 (2) 1–7 Wind (theoretical) Calm – ± – (0) – – ± – (0) – – ± – (0) – Crosswind 4.4 ± 3.0 (8) 1–25 5.2 ± 2.1 (11) 1–25 4.8 ± 1.7 (19) 1–25 Headwind 2.0 ± 0.4 (10) 1–4 1.5 ± 0.3 (10) 1–4 1.8 ± 0.2 (20) 1–4 Tailwind 2.2 ± 0.7 (9) 1–7 2.9 ± 1.4 (21) 1–30 2.7 ± 1.0 (30) 1–30 Wind strength Weak 3.8 ± 1.7 (14) 1–25 3.2 ± 1.1 (22) 1–25 3.4 ± 0.9 (36) 1–25 Strong 1.7 ± 0.5 (13) 1–7 3.1 ± 1.5 (20) 1–30 2.5 ± 0.9 (33) 1–30 Total Total 2.8 ± 0.9 (27) 1–25 3.1 ± 0.9 (42) 1–30 3.0 ± 0.6 (69) 1–30 Gulls Time of day Daytime 2.0 ± 0.3 (127) 1–30 2.9 ± 0.4 (94) 1–27 2.4 ± 0.2 (221) 1–30 Nighttime 3.9 ± 1.7 (25) 1–35 1.7 ± 0.4 (15) 1–7 3.1 ± 1.1 (40) 1–35 Precipitation level No precipitation 2.7 ± 0.6 (94) 1–35 2.6 ± 0.4 (80) 1–18 2.6 ± 0.4 (174) 1–35 Precipitation 1.7 ± 0.2 (58) 1–6 3.2 ± 0.9 (29) 1–27 2.2 ± 0.3 (87) 1–27 Session visibility Good 2.4 ± 0.4 (136) 1–35 2.8 ± 0.4 (104) 1–27 2.6 ± 0.3 (240) 1–35 Poor 1.3 ± 0.2 (16) 1–4 1.2 ± 0.2 (5) 1–2 1.3 ± 0.2 (20) 1–4 Wind (theoretical) Calm 1.6 ± 0.5 (8) 1–5 – ± – (0) – 1.6 ± 0.5 (8) 1–5 Crosswind 1.6 ± 0.2 (29) 1–5 3.0 ± 0.7 (29) 1–14 2.3 ± 0.4 (58) 1–14 Headwind 2.4 ± 0.5 (64) 1–30 2.5 ± 0.4 (57) 1–18 2.5 ± 0.3 (121) 1–30 Tailwind 2.7 ± 0.9 (51) 1–35 2.8 ± 1.1 (23) 1–27 1.5 ± 0.7 (74) 1–35 Wind strength Weak 2.0 ± 0.5 (60) 1–30 2.7 ± 0.4 (73) 1–18 2.4 ± 0.3 (133) 1–30 Strong 2.5 ± 0.5 (92) 1–35 2.9 ± 0.8 (36) 1–27 2.6 ± 0.4 (128) 1–35 Total Total 2.3 ± 0.4 (152) 1–35 2.7 ± 0.4 (109) 1–27 2.5 ± 0.3 (261) 1–35

Results

Migration at N

orthstar Island84

AB

R F

inal Report


Lights

Off On Total


Alcids Time of day Daytime 1.5 ± 0.3 (4) 1–2 1.3 ± 0.3 (6) 1–3 1.4 ± 0.2 (10) 1–3 Nighttime 1.0 ± – (1) 1–1 2.3 ± 0.9 (3) 1–4 2.0 ± 0.7 (4) 1–4 Precipitation level No precipitation 1.3 ± 0.3 (3) 1–2 1.9 ± 0.5 (7) 1–4 1.7 ± 0.3 (10) 1–4 Precipitation 1.5 ± 0.5 (2) 1–2 1.0 ± 0 (2) 1–1 1.3 ± 0.3 (4) 1–2 Session visibility Good 1.3 ± 0.3 (4) 1–2 1.9 ± 0.5 (7) 1–4 1.6 ± 0.3 (11) 1–4 Poor 2.0 ± – (1) 2–2 1.0 ± 0 (2) 1–1 1.3 ± 0.3 (3) 1–2 Wind (theoretical) Calm 1.0 ± – (1) 1–1 1.0 ± – (1) 1–1 1.0 ± 0 (2) 1–1 Crosswind 1.0 ± – (1) 1–1 1.0 ± – (1) 1–1 1.0 ± 0 (2) 1–1 Headwind 1.5 ± 0.5 (2) 1–2 1.0 ± – (1) 1–1 1.3 ± 0.3 (3) 1–2 Tailwind 2.0 ± – (1) 2–2 2.0 ± 0.5 (6) 1–4 2.0 ± 0.4 (7) 1–4 Wind strength Weak 1.0 ± 0 (2) 1–1 1.7 ± 0.5 (6) 1–4 1.5 ± 0.4 (8) 1–4 Strong 1.7 ± 0.3 (3) 1–2 1.7 ± 0.7 (3) 1–3 1.7 ± 0.3 (6) 1–3 Total Total 1.4 ± 0.2 (5) 1–2 1.7 ± 0.4 (9) 1–4 1.6 ± 0.3 (14) 1–4

Results

AB

R F

inal Report

85M


Table 22. Mean flight altitude (m agl or asl) of birds seen visually near Northstar Island, northern Alaska, fall 2001–2003, by species-group, factor, and anti-collision lighting setting. Data are presented as mean ± SE (n flocks) and range.

Lights

Off On Total


Eiders Time of day Daytime 5.9 ± 1.0 (68) 1–50 4.9 ± 1.1 (35) 1–30 5.6 ± 0.8 (103) 1–50 Nighttime 16.7 ± 7.3 (3) 5–30 11.0 ± 4.9 (3) 3–20 13.8 ± 4.1 (6) 3–30 Precipitation level No precipitation 7.9 ± 1.5 (44) 1–50 5.0 ± 1.3 (27) 1–30 6.8 ± 1.1 (71) 1–50 Precipitation 3.9 ± 0.7 (27) 1–15 6.2 ± 2.2 (11) 1–20 4.6 ± 0.8 (38) 1–20 Session visibility Good 6.4 ± 1.0 (69) 1–50 5.1 ± 1.1 (35) 1–30 6.0 ± 0.8 (104) 1–50 Poor 6.0 ± 1.0 (2) 5–7 8.7 ± 5.7 (3) 3–20 7.6 ± 3.2 (5) 3–20 Wind (theoretical) Calm 1.5 ± 0.5 (2) 1–2 3.0 ± – (1) 3–3 2.0 ± 0.6 (3) 1–3 Crosswind 6.7 ± 2.4 (21) 1–50 2.3 ± 0.9 (4) 1–5 6.0 ± 2.1 (25) 1–50 Headwind 1.5 ± 0.2 (10) 1–3 4.8 ± 2.4 (13) 1–30 3.4 ± 1.4 (23) 1–30 Tailwind 7.8 ± 1.3 (38) 1–30 6.4 ± 1.4 (20) 1–20 7.3 ± 1.0 (58) 1–30 Wind strength Weak 7.5 ± 2.2 (19) 1–30 5.6 ± 1.8 (18) 1–30 6.6 ± 1.4 (37) 1–30 Strong 6.0 ± 1.1 (52) 1–50 5.1 ± 1.4 (20) 1–20 5.7 ± 0.9 (72) 1–50 Total Total 6.4 ± 1.0 (71) 1–50 5.3 ± 1.1 (38) 1–30 6.0 ± 0.8 (109) 1–50 Loons Time of day Daytime 9.8 ± 1.6 (103) 1–100 7.9 ± 1.4 (112) 1–100 8.8 ± 1.0 (215) 1–100 Nighttime – ± – (0) – – ± – (0) – – ± – (0) – Precipitation level No precipitation 9.8 ± 1.6 (102) 1–100 9.0 ± 1.7 (90) 1–100 9.4 ± 1.2 (192) 1–100 Precipitation 3.0 ± – (1) 3–3 3.4 ± 1.0 (22) 1–20 3.4 ± 0.9 (23) 1–20 Session visibility Good 9.8 ± 1.6 (103) 1–100 7.9 ± 1.4 (111) 1–100 8.8 ± 1.1 (214) 1–100 Poor – ± – (0) – 5.0 ± – (1) 5–5 5.0 ± – (1) 5–5 Wind (theoretical) Calm 2.4 ± 1.1 (8) 1–10 5.8 ± 2.3 (17) 1–30 4.7 ± 1.6 (25) 1–30 Crosswind 4.9 ± 1.3 (23) 1–20 5.9 ± 2.9 (19) 1–55 5.4 ± 1.5 (42) 1–55 Headwind 7.3 ± 1.5 (50) 1–45 4.3 ± 0.8 (55) 1–30 5.7 ± 0.8 (105) 1–45 Tailwind 23.1 ± 5.5 (22) 1–100 20.7 ± 5.7 (21) 1–100 22.0 ± 3.9 (43) 1–100 Wind strength Weak 8.0 ± 2.7 (41) 1–100 5.3 ± 1.0 (76) 1–55 6.3 ± 1.2 (117) 1–100 Strong 11.0 ± 1.9 (62) 1–70 13.3 ± 3.7 (36) 1–100 11.8 ± 1.8 (98) 1–100 Total Total 9.8 ± 1.6 (103) 1–100 7.9 ± 1.4 (112) 1–100 8.8 ± 1.0 (215) 1–100

Results

Migration at N

orthstar Island86

AB

R F

inal Report


Lights

Off On Total


Other ducks Time of day Daytime 3.2 ± 0.8 (65) 1–45 3.1 ± 0.6 (72) 1–30 3.0 ± 0.5 (137) 1–45 Nighttime 16.5 ± 3.1 (25) 1–55 18.2 ± 2.7 (22) 3–40 17.3 ± 2.1 (47) 1–55 Precipitation level No precipitation 5.9 ± 1.3 (69) 1–55 6.3 ± 1.1 (75) 1–40 6.1 ± 0.8 (144) 1–55 Precipitation 10.1 ± 2.8 (21) 1–50 8.1 ± 2.4 (19) 1–30 9.1 ± 1.8 (40) 1–50 Session visibility Good 6.7 ± 1.3 (84) 1–55 6.6 ± 1.0 (93) 1–40 6.7 ± 0.8 (177) 1–55 Poor 9.8 ± 4.5 (6) 2–30 5.0 ± – (1) 5–5 9.1 ± 3.8 (7) 2–30 Wind (theoretical) Calm 1.2 ± 0.2 (5) 1–2 1.0 ± 0 (4) 1–1 1.1 ± 0.1 (9) 1–2 Crosswind 1.9 ± 0.5 (9) 1–5 5.7 ± 1.8 (27) 1–40 4.8 ± 1.4 (36) 1–40 Headwind 5.9 ± 1.5 (40) 1–50 6.4 ± 1.3 (45) 1–40 6.1 ± 1.0 (85) 1–50 Tailwind 10.1 ± 2.4 (36) 1–55 9.9 ± 3.1 (18) 1–40 10.0 ± 1.9 (54) 1–55 Wind strength Weak 7.1 ± 1.9 (36) 1–55 6.1 ± 1.1 (55) 1–40 6.5 ± 1.0 (91) 1–55 Strong 6.7 ± 1.6 (54) 1–50 7.3 ± 1.8 (39) 1–40 7.0 ± 1.2 (93) 1–50 Total Total 6.9 ± 1.2 (90) 1–55 6.6 ± 1.0 (94) 1–40 6.8 ± 0.8 (184) 1–55 Unid. ducks Time of day Daytime 6.1 ± 0.9 (58) 1–30 6.9 ± 1.0 (97) 1–60 6.6 ± 0.7 (155) 1–60 Nighttime 13.9 ± 2.8 (21) 1–50 20.4 ± 4.1 (10) 1–40 16.0 ± 2.3 (31) 1–50 Precipitation level No precipitation 8.2 ± 1.1 (74) 1–50 8.1 ± 1.1 (92) 1–60 8.2 ± 0.8 (166) 1–60 Precipitation 7.6 ± 4.5 (5) 1–25 7.8 ± 3.1 (16) 1–50 7.7 ± 2.5 (21) 1–50 Session visibility Good 8.2 ± 1.1 (77) 1–50 8.1 ± 1.1 (107) 1–60 8.2 ± 0.8 (184) 1–60 Poor 4.0 ± 3.0 (2) 1–7 3.0 ± – (1) 3–3 3.7 ± 1.8 (3) 1–7 Wind (theoretical) Calm 1.0 ± – (1) 1–1 1.2 ± 0.2 (5) 1–2 1.2 ± 0.2 (6) 1–2 Crosswind 6.7 ± 1.9 (12) 1–20 7.3 ± 2.3 (12) 1–25 7.0 ± 1.4 (24) 1–25 Headwind 3.7 ± 1.0 (20) 1–20 4.7 ± 1.5 (36) 1–50 4.4 ± 1.0 (56) 1–50 Tailwind 10.6 ± 1.5 (46) 1–50 11.1 ± 1.7 (55) 1–60 10.9 ± 1.2 (101) 1–60 Wind strength Weak 8.0 ± 2.7 (20) 1–50 6.9 ± 1.9 (35) 1–50 7.3 ± 1.5 (55) 1–50 Strong 8.2 ± 1.1 (59) 1–30 8.6 ± 1.3 (73) 1–60 8.4 ± 0.9 (132) 1–60 Total Total 8.1 ± 1.0 (79) 1–50 8.1 ± 1.1 (108) 1–60 8.1 ± 0.8 (187) 1–60

Results

AB

R F

inal Report

87M



Lights

Off On Total


Shorebirds Time of day Daytime 1.0 ± 0 (2) 1–1 1.7 ± 0.7 (3) 1–3 1.4 ± 0.4 (5) 1–3 Nighttime 11.9 ± 2.8 (25) 1–50 14.4 ± 2.2 (39) 1–65 13.5 ± 1.7 (64) 1–65 Precipitation level No precipitation 15.3 ± 4.3 (15) 1–50 15.7 ± 3.1 (26) 1–65 15.6 ± 2.5 (41) 1–65 Precipitation 5.8 ± 1.6 (12) 1–20 9.9 ± 2.1 (16) 1–25 8.2 ± 1.4 (28) 1–25 Session visibility Good 11.8 ± 2.8 (25) 1–50 13.5 ± 2.1 (42) 1–65 12.9 ± 1.7 (67) 1–65 Poor 2.0 ± 1.0 (2) 1–3 – ± – (0) – 2.0 ± 1.0 (2) 1–3 Wind (theoretical) Calm – ± – (0) – – ± – (0) – – ± – (0) – Crosswind 8.0 ± 2.8 (8) 1–20 14.3 ± 3.3 (11) 1–35 11.6 ± 2.3 (19) 1–35 Headwind 6.7 ± 1.9 (10) 3–20 8.3 ± 3.0 (10) 1–30 7.5 ± 1.7 (20) 1–30 Tailwind 18.8 ± 6.8 (9) 1–50 15.6 ± 3.5 (21) 1–65 16.6 ± 3.1 (30) 1–65 Wind strength Weak 16.0 ± 3.7 (14) 1–50 20.7 ± 3.2 (22) 1–65 18.9 ± 2.4 (36) 1–65 Strong 5.8 ± 3.3 (13) 1–45 5.7 ± 1.2 (20) 1–20 5.7 ± 1.4 (33) 1–45 Total Total 11.1 ± 2.6 (27) 1–50 13.5 ± 2.1 (42) 1–65 12.6 ± 1.6 (69) 1–65 Gulls Time of day Daytime 21.7 ± 2.9 (127) 1–350 17.5 ± 1.6 (94) 1–80 19.9 ± 1.8 (221) 1–350 Nighttime 17.0 ± 2.6 (25) 1–50 13.6 ± 2.9 (15) 1–35 15.7 ± 2.0 (40) 1–50 Precipitation level No precipitation 23.5 ± 3.9 (94) 1–350 17.7 ± 1.8 (80) 1–80 20.9 ± 2.2 (174) 1–350 Precipitation 16.6 ± 1.9 (58) 1–50 14.8 ± 2.1 (29) 1–40 16.0 ± 1.4 (87) 1–50 Session visibility Good 21.8 ± 2.8 (136) 1–350 17.0 ± 1.5 (104) 1–80 19.7 ± 1.7 (240) 1–350 Poor 13.4 ± 2.6 (16) 1–40 17.0 ± 4.4 (5) 5–30 14.3 ± 2.2 (21) 1–40 Wind (theoretical) Calm 19.4 ± 0.6 (8) 15–20 – ± – (0) – 19.4 ± 0.6 (8) 15–20 Crosswind 25.3 ± 12.3 (29) 1–350 17.7 ± 2.9 (29) 1–65 21.4 ± 6.2 (58) 1–350 Headwind 19.5 ± 1.8 (64) 1–50 16.4 ± 2.0 (57) 1–80 18.1 ± 1.3 (121) 1–80 Tailwind 20.4 ± 1.9 (51) 1–50 17.4 ± 3.0 (23) 1–45 19.5 ± 1.6 (74) 1–50 Wind strength Weak 25.3 ± 5.9 (60) 1–350 19.0 ± 1.8 (73) 1–80 21.8 ± 2.8 (133) 1–350 Strong 18.1 ± 1.5 (92) 1–50 12.8 ± 2.3 (36) 1–45 16.6 ± 1.3 (128) 1–50 Total Total 20.9 ± 2.5 (152) 1–350 17.0 ± 1.4 (109) 1–80 19.2 ± 1.6 (261) 1–350

Results

Migration at N

orthstar Island88

AB

R F

inal Report


Lights

Off On Total


Alcids Time of day Daytime 3.0 ± 0.4 (4) 2–4 1.5 ± 0.3 (6) 1–3 2.1 ± 0.3 (10) 1–4 Nighttime 2.0 ± – (1) 2–2 7.0 ± 3.0 (3) 1–10 5.8 ± 2.5 (4) 1–10 Precipitation level No precipitation 3.3 ± 0.3 (3) 3–4 4.0 ± 1.6 (7) 1–10 3.8 ± 1.1 (10) 1–10 Precipitation 2.0 ± 0 (2) 2–2 1.0 ± 0 (2) 1–1 1.5 ± 0.3 (4) 1–2 Session visibility Good 3.0 ± 0.4 (4) 2–4 4.0 ± 1.6 (7) 1–10 3.6 ± 1.0 (11) 1–10 Poor 2.0 ± – (1) 2–2 1.0 ± 0 (2) 1–1 1.3 ± 0.3 (3) 1–2 Wind (theoretical) Calm 3.0 ± – (1) 3–3 1.0 ± – (1) 1–1 2.0 ± 1.0 (2) 1–3 Crosswind 2.0 ± – (1) 2–2 2.0 ± – (1) 2–2 2.0 ± 0 (2) 2–2 Headwind 3.5 ± 0.5 (2) 3–4 1.0 ± – (1) 1–1 2.7 ± 0.9 (3) 1–4 Tailwind 2.0 ± – (1) 2–2 4.3 ± 1.8 (6) 1–10 4.0 ± 1.6 (7) 1–10 Wind strength Weak 2.5 ± 0.5 (2) 2–3 4.2 ± 1.9 (6) 1–10 3.8 ± 1.4 (8) 1–10 Strong 3.0 ± 0.6 (3) 2–4 1.7 ± 0.7 (3) 1–3 2.3 ± 0.5 (6) 1–4 Total Total 2.8 ± 0.4 (5) 2–4 3.3 ± 1.3 (9) 1–10 3.1 ± 0.8 (14) 1–10

Results


however, the variation was small and SEs generallywere so large that the differences would not havebeen statistically significant.

Loons averaged flight altitudes of ~9 magl/asl, with a range of 1–100 m agl/asl (Table 22).Similar to the pattern seen for eiders and gulls,flight altitudes of loons were higher with theanti-collision lights off than with them on and werehighest with tailwinds. Other ducks averaged ~7 magl/asl, unidentified ducks averaged ~8 m agl/asl,shorebirds averaged ~13 m agl/asl, gulls averaged~19 m agl/asl, and alcids averaged ~3 m agl/asl.For nearly all of these species-groups, flightaltitudes were higher with the anti-collision lightson than with them off. Other ducks, unidentifiedducks, shorebirds, and alcids also tended to fly atthe highest altitudes with tailwinds. Other ducks,unidentified ducks, shorebirds, and alcids alsotended to fly higher at night than during the day.

FLIGHT BEHAVIOREiders exhibited little variation in general

flight behavior, regardless of the visibility categoryand anti-collision lighting setting (Table 23).Nearly all birds flew with straight-line (directional)behavior; however, three flocks were flyingerratically or circling with the lights off, whereasnone were flying this way with the lights on. Therealso was a tendency for a lower frequency ofstraight-line behavior with headwinds, althoughthe difference was not great.

Loons exhibited straight-line behavior in~96% of all flocks, with the anti-collision lightsapparently having no effect on behavior (Table 23).Other ducks exhibited a high percentage of circlingbehavior that primarily was associated with agas-flaring event in September 2002 (see “Radar:Behavior,” above); the anti-collision lights had noeffect on behavior. They also exhibited a lowerpercentage of straight-line flight during headwindsthan with winds of other types. Unidentified ducksexhibited primarily straight-line behavior, with theanti-collision lights having no effect on behavior.Shorebirds exhibited a high percentage ofnon-directional behaviors, especially circlingbehavior; however, there was little differencebetween lighting settings in the percentage ofnon-directional behaviors. Both gulls and alcidsexhibited moderate–high percentages of

straight-line behavior, with the anti-collision lightshaving little effect on behavior.

ISLAND-PASSING SUCCESSEiders exhibited little variation in

island-passing success, regardless of the factor andanti-collision lighting setting (Table 24).Approximately 92% of all eider flocks for whichwe knew the success status were successful inpassing the island. Success rates were higher withthe anti-collision lights off than with them on, butthe difference was slight. In addition, success ratestended to be lower with headwinds than with otherwind directions.

All other taxa exhibited moderate–highoverall passing success (70–94%; Table 24). Therewas a suggestion of slightly lower success with theanti-collision lights off than with them on for otherducks, unidentified ducks, gulls, and alcids,although differences were not dramatic. Few otherpatterns were evident, although passing success forother ducks, unidentified ducks, shorebirds, andgulls was lower with headwinds than with otherwind directions. In contrast, passing success forloons was highest with headwinds.

ISLAND-PASSING BEHAVIOREiders exhibited little variation in

high-resolution island-passing behavior, regardlessof the factor and anti-collision lighting setting(Table 25). About 10% of the flocks passed theisland with a detectable change in behavior. Fiveflocks exhibited a “significant” change in flightdirection, three exhibited a “significant” change inflight altitude, and two exhibited a “significant”change in both (see definition of "significant" inthis context in Methods). No eider flocks exhibitedbehavioral flaring. Of those flocks exhibiting achange in altitude as they passed the island, threedecreased altitude by 1–5 m, one decreased altitudeby 11–20 m, and one increased altitude by 1–5 m.

Similar to the pattern seen for eiders, loonsexhibited little variation in island-passing behavior,regardless of the factor and anti-collision lightingsetting (Table 25). In contrast, all otherspecies-groups exhibited substantial reaction to theisland, with shorebirds, other ducks, and gullshaving the highest response rate but onlyshorebirds exhibiting a response rate >50% of thetime. Reactions included changes in flight

Results

Migration at N

orthstar Island90

AB

R F

inal Report

Table 23. Frequencies and percentages of general flight behaviors of birds seen visually near Northstar Island, northern Alaska, fall 2001–2004, by species-group, factor, and anti-collision lighting setting. Data are presented as number (%) and n (flocks).

Flight behavior

Species-group Factor Attribute Lights Straight-line Erratic Circling n

Eiders Time of day Daytime Off 62 (92.5) 4 (6.0) 1 (1.5) 67 On 35 (100.0) 0 (0) 0 (0) 35 Total 97 (95.1) 4 (3.9) 1 (1.0) 102 Nighttime Off 1 (50.0) 1 (50.0) – (–) 2 On 3 (100.0) 0 (0) 0 (0) 3 Total 4 (80.0) 1 (20.0) 0 (0) 5 Precipitation level No precipitation Off 39 (92.9) 2 (4.8) 1 (2.4) 42 On 27 (100.0) 0 (0) 0 (0) 27 Total 66 (95.6) 2 (2.9) 1 (1.4) 69 Precipitation Off 24 (88.9) 3 (11.1) 0 (0) 27 On 11 (100.0) 0 (0) 0 (0) 11 Total 35 (92.1) 3 (7.9) 0 (0) 38 Session visibility Good Off 61 (91.0) 5 (7.5) 1 (1.5) 67 On 35 (100.0) 0 (0) 0 (0) 35 Total 96 (94.1) 5 (4.9) 1 (1.0) 102 Poor Off 2 (100.0) 0 (0) 0 (0) 2 On 3 (100.0) 0 (0) 0 (0) 3 Total 5 (100.0) 0 (0) 0 (0) 5 Wind (theoretical) Calm Off 2 (100.0) 0 (0) 0 (0) 2 On 1 (100.0) 0 (0) 0 (0) 1 Total 3 (100.0) 0 (0) 0 (0) 3 Crosswind Off 20 (95.2) 1 (4.8) 0 (0) 21 On 4 (100.0) 0 (0) 0 (0) 4 Total 24 (96.0) 1 (4.0) 0 (0) 25 Headwind Off 8 (80.0) 1 (10.0) 1 (10.0) 10 On 13 (100.0) 0 (0) 0 (0) 13 Total 21 (91.3) 1 (4.3) 1 (4.3) 23 Tailwind Off 33 (91.7) 3 (8.3) 0 (0) 36 On 20 (100.0) 0 (0) 0 (0) 20 Total 53 (94.6) 3 (5.4) 0 (0) 56 Wind strength Weak Off 16 (94.1) 0 (0) 1 (5.9) 17 On 18 (100.0) 0 (0) 0 (0) 18 Total 34 (97.1) 0 (0) 1 (2.9) 35 Strong Off 47 (90.4) 5 (9.6) 0 (0) 52 On 20 (100.0) 0 (0) 0 (0) 20 Total 67 (93.1) 5 (6.9) 0 (0) 72 Total Total Off 63 (91.3) 5 (7.2) 1 (1.4) 69 On 38 (100.0) 0 (0) 0 (0) 38 Total 101 (94.4) 5 (4.7) 1 (0.9) 107

Results

AB

R F

inal Report

91M



Flight behavior


Loons Time of day Daytime Off 98 (97.0) 2 (2.0) 1 (1.0) 101 On 106 (94.6) 5 (4.5) 1 (0.9) 112 Total 204 (95.8) 7 (3.3) 2 (0.9) 213 Nighttime Off – (–) – (–) – (–) 0 On – (–) – (–) – (–) 0 Total – (–) – (–) – (–) 0 Precipitation level No precipitation Off 97 (97.0) 2 (2.0) 1 (1.0) 100 On 85 (94.4) 4 (4.4) 1 (1.1) 90 Total 182 (95.7) 6 (3.2) 2 (1.1) 190 Precipitation Off 1 (100.0) 0 (0) 0 (0) 1 On 21 (95.5) 1 (4.5) 0 (0) 22 Total 22 (95.7) 1 (4.3) 0 (0) 23 Session visibility Good Off 98 (97.0) 2 (2.0) 1 (1.0) 101 On 105 (94.6) 5 (4.5) 1 (0.9) 111 Total 203 (95.7) 7 (3.3) 2 (0.9) 212 Poor Off – (–) – (–) – (–) 0 On 1 (100.0) 0 (0) 0 (0) 1 Total 1 (100.0) 0 (0) 0 (0) 1 Wind (theoretical) Calm Off 8 (100.0) 0 (0) 0 (0) 8 On 17 (100.0) 0 (0) 0 (0) 17 Total 25 (100.0) 0 (0) 0 (0) 25 Crosswind Off 20 (90.9) 2 (9.1) 0 (0) 22 On 18 (94.7) 1 (5.3) 0 (0) 19 Total 38 (92.7) 3 (7.3) 0 (0) 41 Headwind Off 48 (98.0) 0 (0) 1 (2.0) 49 On 51 (92.7) 3 (5.5) 1 (0) 55 Total 99 (95.2) 3 (2.9) 2 (1.9) 104 Tailwind Off 22 (100.0) 0 (0) 0 (0) 22 On 20 (95.2) 1 (4.8) 0 (0) 21 Total 42 (97.7) 1 (2.3) 0 (0) 43 Wind strength Weak Off 39 (97.5) 1 (2.5) 0 (0) 40 On 73 (96.1) 2 (2.6) 1 (1.3) 76 Total 112 (96.6) 3 (2.6) 1 (0.9) 116 Strong Off 59 (96.7) 1 (1.6) 1 (1.6) 61 On 33 (91.7) 3 (8.3) 0 (0) 36 Total 92 (94.8) 4 (4.1) 1 (1.0) 97 Total Total Off 98 (97.0) 2 (2.0) 1 (1.0) 101 On 106 (94.6) 5 (4.5) 1 (0.9) 112 Total 204 (95.8) 7 (3.3) 2 (0.9) 213

Results

Migration at N

orthstar Island92

AB

R F

inal Report


Flight behavior


Other ducks Time of day Daytime Off 63 (96.9) 2 (3.1) 0 (0) 65 On 62 (87.3) 5 (7.0) 4 (5.6) 71 Total 125 (91.9) 7 (5.1) 4 (2.9) 136 Nighttime Off 8 (32.0) 3 (12.0) 14 (56.0) 25 On 12 (54.5) 0 (0) 10 (45.5) 22 Total 20 (42.5) 3 (6.4) 24 (51.1) 47 Precipitation level No precipitation Off 60 (87.0) 2 (2.9) 7 (10.1) 69 On 58 (77.3) 4 (5.3) 13 (17.3) 75 Total 118 (81.9) 6 (4.2) 20 (13.9) 144 Precipitation Off 11 (52.4) 3 (14.3) 7 (33.3) 21 On 16 (88.9) 1 (5.6) 1 (5.6) 18 Total 25 (64.1) 4 (10.3) 8 (20.5) 39 Session visibility Good Off 69 (82.1) 4 (4.8) 11 (13.1) 84 On 73 (79.3) 5 (5.4) 14 (15.2) 92 Total 142 (80.7) 9 (5.1) 25 (14.2) 176 Poor Off 2 (33.3) 1 (16.7) 3 (50.0) 6 On 1 (100.0) 0 (0) 0 (0) 1 Total 3 (42.9) 1 (14.3) 3 (42.9) 7 Wind (theoretical) Calm Off 5 (100.0) 0 (0) 0 (0) 5 On 4 (100.0) 0 (0) 0 (0) 4 Total 9 (100.0) 0 (0) 0 (0) 9 Crosswind Off 9 (100.0) 0 (0) 0 (0) 9 On 25 (96.2) 0 (0) 1 (3.8) 26 Total 34 (97.1) 0 (0) 1 (2.9) 35 Headwind Off 24 (60.0) 4 (10.0) 12 (30.0) 40 On 28 (62.2) 5 (11.1) 12 (26.7) 45 Total 52 (61.2) 9 (10.6) 24 (28.2) 85 Tailwind Off 33 (91.7) 1 (2.8) 2 (5.6) 36 On 17 (94.4) 0 (0) 1 (5.6) 18 Total 50 (92.6) 1 (1.9) 3 (5.6) 54 Wind strength Weak Off 22 (61.1) 3 (8.3) 11 (30.6) 36 On 38 (69.1) 4 (7.3) 13 (23.6) 55 Total 60 (65.9) 7 (7.7) 24 (26.4) 91 Strong Off 49 (90.7) 2 (3.7) 3 (5.6) 54 On 36 (94.7) 1 (2.6) 1 (2.6) 39 Total 85 (92.4) 3 (3.3) 4 (4.3) 92 Total Total Off 71 (78.9) 5 (5.6) 14 (15.6) 90 On 74 (79.6) 5 (5.4) 14 (15.2) 93 Total 145 (79.2) 10 (5.5) 28 (15.3) 183

Results

AB

R F

inal Report

93M



Flight behavior


Unid. ducks Time of day Daytime Off 57 (100.0) 0 (0) 0 (0) 57 On 92 (96.8) 3 (3.2) 0 (0) 95 Total 149 (98.0) 3 (2.0) 0 (0) 152 Nighttime Off 19 (90.5) 1 (4.8) 1 (4.8) 21 On 10 (90.9) 0 (0) 1 (9.1) 11 Total 29 (90.6) 1 (3.1) 2 (6.3) 32 Precipitation level No precipitation Off 71 (97.3) 1 (1.4) 1 (1.4) 73 On 87 (96.7) 2 (2.2) 1 (1.1) 90 Total 158 (96.9) 3 (1.9) 2 (1.2) 162 Precipitation Off 5 (100.0) 0 (0) 0 (0) 5 On 15 (93.8) 1 (6.3) 0 (0) 16 Total 20 (95.2) 1 (4.8) 0 (0) 21 Session visibility Good Off 74 (97.4) 1 (1.3) 1 (1.3) 76 On 101 (96.2) 3 (2.9) 1 (1.0) 105 Total 175 (96.7) 4 (2.2) 2 (1.1) 181 Poor Off 2 (100.0) 0 (0) 0 (0) 2 On 1 (100.0) 0 (0) 0 (0) 1 Total 3 (100.0) 0 (0) 0 (0) 3 Wind (theoretical) Calm Off 1 (100.0) 0 (0) 0 (0) 1 On 5 (100.0) 0 (0) 0 (0) 5 Total 6 (100.0) 0 (0) 0 (0) 6 Crosswind Off 11 (100.0) 0 (0) 0 (0) 11 On 12 (100.0) 0 (0) 0 (0) 11 Total 23 (100.0) 0 (0) 0 (0) 23 Headwind Off 20 (100.0) 0 (0) 0 (0) 20 On 32 (94.1) 2 (5.9) 0 (0) 34 Total 52 (96.3) 2 (3.7) 0 (0) 54 Tailwind Off 44 (95.6) 1 (2.2) 1 (2.2) 46 On 53 (96.4) 1 (1.8) 1 (1.8) 55 Total 97 (96.0) 2 (2.0) 2 (2.0) 101 Wind strength Weak Off 18 (90.0) 1 (5.0) 1 (5.0) 20 On 31 (93.9) 1 (3.0) 1 (3.0) 33 Total 49 (92.5) 2 (3.8) 2 (3.8) 53 Strong Off 58 (100.0) 0 (0) 0 (0) 58 On 71 (97.3) 2 (2.7) 0 (0) 73 Total 129 (98.5) 2 (1.5) 0 (0) 131 Total Total Off 76 (97.4) 1 (1.3) 1 (1.3) 78 On 102 (96.2) 3 (2.8) 1 (0.9) 106 Total 178 (96.7) 4 (2.2) 2 (1.1) 187

Results

Migration at N

orthstar Island94

AB

R F

inal Report


Flight behavior


Shorebirds Time of day Daytime Off 2 (100.0) 0 (0) 0 (0) 2 On 2 (66.7) 1 (33.3) 0 (0) 3 Total 4 (80.0) 1 (20.0) 0 (0) 5 Nighttime Off 12 (48.0) 7 (28.0) 6 (24.0) 25 On 17 (43.6) 11 (28.2) 11 (28.2) 39 Total 29 (45.3) 18 (28.1) 17 (26.6) 64 Precipitation level No precipitation Off 6 (40.0) 4 (26.7) 5 (33.3) 15 On 9 (34.6) 7 (26.9) 10 (38.5) 26 Total 15 (36.6) 11 (26.8) 15 (36.6) 41 Precipitation Off 8 (66.7) 3 (25.0) 1 (8.3) 12 On 10 (62.5) 5 (31.3) 1 (6.3) 16 Total 18 (64.3) 8 (28.6) 2 (7.1) 28 Session visibility Good Off 13 (52.0) 6 (24.0) 6 (24.0) 25 On 19 (45.2) 12 (28.6) 11 (26.2) 42 Total 32 (47.8) 18 (26.9) 17 (25.3) 67 Poor Off 1 (50.0) 1 (50.0) 0 (0) 2 On – (–) – (–) – (–) 0 Total 1 (50.0) 1 (50.0) 0 (0) 2 Wind (theoretical) Calm Off – (–) – (–) – (–) 0 On – (–) – (–) – (–) 0 Total – (–) – (–) – (–) 0 Crosswind Off 5 (62.5) 2 (25.0) 1 (12.5) 8 On 6 (54.5) 4 (36.4) 1 (9.1) 11 Total 11 (57.8) 6 (31.6) 2 (10.5) 19 Headwind Off 5 (50.0) 3 (30.0) 2 (20.0) 10 On 5 (50.0) 2 (20.0) 3 (30.0) 10 Total 10 (50.0) 5 (25.0) 5 (25.0) 20 Tailwind Off 4 (44.4) 2 (22.2) 3 (33.3) 9 On 8 (38.1) 6 (28.6) 7 (33.3) 21 Total 12 (40.0) 8 (26.7) 10 (33.3) 30 Wind strength Weak Off 6 (42.9) 2 (14.3) 6 (42.9) 14 On 10 (45.5) 3 (13.6) 9 (40.9) 22 Total 16 (44.4) 5 (13.9) 15 (41.7) 36 Strong Off 8 (61.5) 5 (38.5) 0 (0) 13 On 9 (45.0) 9 (45.0) 2 (10.0) 20 Total 17 (51.5) 14 (42.4) 2 (6.1) 33 Total Total Off 14 (51.9) 7 (25.9) 6 (22.2) 27 On 19 (45.2) 12 (28.6) 11 (26.2) 42 Total 33 (69.0) 19 (27.5) 17 (24.6) 69

Results

AB

R F

inal Report

95M



Flight behavior


Gulls Time of day Daytime Off 104 (82.5) 14 (11.1) 8 (6.3) 126 On 66 (70.2) 19 (20.2) 9 (9.6) 94 Total 170 (77.3) 33 (15.0) 17 (7.7) 220 Nighttime Off 13 (54.2) 1 (4.2) 10 (41.7) 24 On 8 (53.3) 2 (13.3) 5 (33.3) 15 Total 21 (53.8) 3 (20.5) 15 (38.5) 39 Precipitation level No precipitation Off 78 (83.9) 5 (5.4) 10 (10.8) 94 On 54 (67.5) 18 (22.5) 8 (10.0) 80 Total 132 (76.3) 23 (13.3) 18 (10.4) 173 Precipitation Off 39 (68.4) 10 (17.5) 8 (14.0) 57 On 20 (69.0) 3 (10.3) 6 (20.7) 29 Total 59 (68.6) 13 (15.1) 14 (16.3) 86 Session visibility Good Off 108 (80.6) 12 (9.0) 14 (10.4) 134 On 71 (68.3) 21 (20.2) 12 (11.5) 104 Total 179 (75.2) 33 (13.9) 26 (10.9) 238 Poor Off 9 (56.2) 3 (18.8) 4 (25) 16 On 3 (60.0) 0 (0) 2 (40.0) 5 Total 12 (57.1) 3 (14.3) 6 (28.6) 21 Wind (theoretical) Calm Off 8 (100.0) 0 (0) 0 (0) 8 On – (–) – (–) – (–) 0 Total 8 (100.0) 0 (0) 0 (0) 8 Crosswind Off 17 (58.6) 7 (24.1) 5 (17.2) 29 On 19 (65.5) 5 (17.2) 5 (17.2) 29 Total 36 (62.1) 12 (20.7) 10 (17.2) 58 Headwind Off 50 (79.4) 7 (11.1) 6 (9.5) 63 On 37 (64.9) 14 (24.6) 6 (10.5) 57 Total 87 (72.5) 21 (17.5) 12 (10.0) 120 Tailwind Off 42 (84.0) 1 (2.0) 7 (13.0) 50 On 18 (78.3) 2 (8.7) 3 (14.3) 23 Total 60 (82.2) 3 (4.1) 10 (13.7) 73 Wind strength Weak Off 42 (71.2) 6 (10.2) 11 (18.6) 59 On 54 (74.0) 9 (12.3) 10 (13.7) 73 Total 96 (72.7) 15 (11.4) 21 (15.9) 132 Strong Off 75 (82.4) 9 (9.9) 7 (7.7) 91 On 20 (55.6) 12 (33.3) 4 (11.1) 36 Total 95 (74.8) 21 (16.5) 11 (8.7) 127 Total Total Off 117 (78.0) 15 (10.0) 18 (12.0) 150 On 74 (67.9) 21 (19.3) 14 (12.8) 109 Total 191 (73.7) 36 (13.9) 32 (12.4) 259

Results

Migration at N

orthstar Island96

AB

R F

inal Report


Flight behavior


Alcids Time of day Daytime Off 4 (100.0) 0 (0) 0 (0) 4 On 6 (100.0) 0 (0) 0 (0) 6 Total 10 (100.0) 0 (0) 0 (0) 10 Nighttime Off 0 (0) 1 (100.0) 0 (0) 1 On 1 (33.3) 2 (66.7) 0 (0) 3 Total 1 (25.0) 3 (75.0) 0 (0) 4 Precipitation level No precipitation Off 3 (100.0) 0 (0) 0 (0) 3 On 5 (71.4) 2 (28.6) 0 (0) 7 Total 8 (80.0) 2 (50.0) 0 (0) 10 Precipitation Off 1 (50.0) 1 (50.0) 0 (0) 2 On 2 (100.0) 0 (0) 0 (0) 2 Total 3 (75.0) 1 (25.0) 0 (0) 4 Session visibility Good Off 3 (75.0) 1 (25.0) 0 (0) 4 On 5 (71.4) 2 (28.6) 0 (0) 7 Total 8 (72.7) 3 (27.3) 0 (0) 11 Poor Off 1 (100.0) 0 (0) 0 (0) 1 On 2 (100.0) 0 (0) 0 (0) 2 Total 3 (100.0) 0 (0) 0 (0) 3 Wind (theoretical) Calm Off 1 (100.0) 0 (0) 0 (0) 1 On 1 (100.0) 0 (0) 0 (0) 1 Total 2 (100.0) 0 (0) 0 (0) 2 Crosswind Off 0 (0) 1 (100.0) 0 (0) 1 On 1 (100.0) 0 (0) 0 (0) 1 Total 1 (50.0) 1 (50.0) 0 (0) 2 Headwind Off 2 (100.0) 0 (0) 0 (0) 2 On 1 (100.0) 0 (0) 0 (0) 1 Total 3 (100.0) 0 (0) 0 (0) 3 Tailwind Off 1 (100.0) 0 (0) 0 (0) 1 On 4 (66.7) 2 (33.3) 0 (0) 6 Total 5 (71.4) 2 (28.6) 0 (0) 7 Wind strength Weak Off 1 (50.0) 1 (50.0) 0 (0) 2 On 4 (66.7) 2 (33.3) 0 (0) 6 Total 5 (62.5) 3 (37.5) 0 (0) 8 Strong Off 3 (100.0) 0 (0) 0 (0) 3 On 3 (100.0) 0 (0) 0 (0) 3 Total 6 (100.0) 0 (0) 0 (0) 6 Total Total Off 4 (80.0) 1 (20.0) 0 (0) 5 On 7 (77.8) 2 (22.2) 0 (0) 9 Total 11 (78.6) 3 (21.4) 0 (0) 14

Results

AB

R F

inal Report

97M


Table 24. Frequencies and percentages of passing success of birds seen visually near Northstar Island, northern Alaska, fall 2001–2004, by species-group, factor, and anti-collision lighting setting. Data are presented as number (%) and n (flocks for Total columns).

Lights

Off On Total

Species-group Factor Attribute Unsuccessful Successful Unsuccessful Successful Unsuccessful Successful n

Eiders Time of day Daytime 6 (9.1) 60 (90.1) 2 (5.7) 33 (94.3) 8 (7.9) 93 (92.1) 101 Nighttime 0 (0) 3 (100.0) 1 (33.3) 2 (66.7) 1 (16.7) 5 (83.3) 6 Precipitation level No precipitation 4 (9.1) 40 (90.9) 1 (3.7) 26 (96.3) 5 (7.0) 66 (93.0) 71 Precipitation 2 (8.0) 23 (92.0) 2 (18.2) 9 (81.8) 4 (11.1) 32 (88.9) 36 Session visibility Good 6 (9.0) 61 (91.0) 2 (5.7) 33 (94.3) 8 (7.8) 94 (92.2) 102 Poor 0 (0) 2 (100.0) 1 (33.3) 2 (66.7) 1 (20.0) 4 (80.0) 5 Wind (theoretical) Calm 0 (0) 2 (100.0) 0 (0) 1 (100.0) 0 (0) 3 (100.0) 3 Crosswind 1 (5.0) 19 (95.0) 1 (25.0) 3 (75.0) 2 (8.3) 22 (91.7) 24 Headwind 4 (40.0) 6 (60.0) 1 (7.7) 12 (92.3) 5 (21.7) 18 (78.3) 23 Tailwind 1 (2.7) 36 (97.3) 1 (5.0) 19 (95.0) 2 (3.5) 55 (96.5) 57 Wind strength Weak 1 (5.3) 18 (94.7) 1 (5.6) 17 (94.4) 2 (5.4) 35 (94.6) 37 Strong 5 (10.0) 45 (90.0) 2 (10.0) 18 (90.0) 7 (10.0) 63 (90.0) 70 Total Total 6 (8.7) 63 (91.3) 3 (7.9) 35 (92.1) 9 (8.4) 98 (91.6) 107 Loons Time of day Daytime 17 (18.9) 73 (81.1) 20 (18.9) 86 (81.1) 37 (18.9) 159 (81.1) 196 Nighttime – (–) – (–) – (–) – (–) – (–) – (–) 0 Precipitation level No precipitation 17 (19.1) 72 (80.9) 18 (21.2) 67 (78.8) 35 (20.1) 139 (79.9) 174 Precipitation 0 (0) 1 (100.0) 2 (9.5) 19 (90.5) 2 (9.1) 20 (90.9) 22 Session visibility Good 17 (18.9) 73 (81.1) 20 (19.0) 85 (81.0) 37 (19.2) 158 (81.0) 195 Poor – (–) – (–) 0 (0) 1 (100.0) 0 (0) 1 (100.0) 1 Wind (theoretical) Calm 1 (20.0) 4 (80.0) 8 (47.1) 9 (52.9) 9 (40.9) 13 (59.1) 22 Crosswind 5 (22.7) 17 (77.3) 3 (18.8) 13 (81.1) 8 (21.1) 30 (78.9) 38 Headwind 8 (17.8) 37 (82.2) 6 (10.9) 49 (89.1) 14 (14.0) 86 (86.0) 100 Tailwind 3 (16.7) 15 (83.3) 3 (16.7) 15 (83.3) 6 (16.7) 30 (83.3) 36 Wind strength Weak 5 (15.6) 27 (84.4) 16 (22.2) 56 (77.8) 21 (20.2) 83 (79.8) 104 Strong 12 (20.7) 46 (79.3) 4 (11.8) 30 (88.2) 16 (17.4) 76 (82.6) 92 Total Total 17 (18.9) 73 (81.1) 20 (18.9) 86 (81.1) 37 (18.9) 159 (81.1) 196

Results

Migration at N

orthstar Island98

AB

R F

inal Report


Lights

Off On Total


Other ducks Time of day Daytime 8 (12.9) 54 (87.1) 9 (13.4) 58 (86.6) 17 (13.2) 112 (86.8) 129 Nighttime 1 (4.5) 21 (95.5) 3 (14.3) 18 (85.7) 4 (9.3) 39 (90.7) 43 Precipitation level No precipitation 7 (10.8) 58 (89.2) 10 (14.5) 59 (85.5) 17 (12.7) 117 (87.3) 134 Precipitation 2 (10.5) 17 (89.5) 2 (10.5) 17 (89.5) 4 (10.5) 34 (89.5) 38 Session visibility Good 9 (11.5) 69 (88.5) 12 (13.8) 75 (86.2) 12 (12.7) 144 (87.3) 165 Poor 0 (0) 6 (100.0) 0 (0) 1 (100.0) 0 (0) 7 (100.0) 7 Wind (theoretical) Calm 2 (50.0) 2 (50.0) 0 (0) 3 (100.0) 2 (28.6) 5 (71.4) 7 Crosswind 0 (0) 9 (100.0) 1 (3.8) 25 (96.2) 1 (2.9) 34 (97.1) 35 Headwind 7 (18.9) 30 (81.1) 11 (26.2) 31 (73.8) 18 (22.8) 61 (77.2) 79 Tailwind 0 (0) 34 (100.0) 0 (0) 17 (100.0) 0 (0) 51 (100.0) 51 Wind strength Weak 4 (12.5) 28 (87.5) 10 (20.0) 40 (80.0) 14 (17.1) 68 (82.9) 82 Strong 5 (9.6) 47 (90.4) 2 (5.3) 36 (94.7) 7 (7.8) 83 (92.2) 90 Total Total 9 (10.7) 75 (89.3) 12 (13.6) 76 (86.4) 21 (12.2) 151 (87.8) 172 Unid. ducks Time of day Daytime 2 (3.8) 51 (96.2) 7 (7.4) 87 (92.6) 9 (6.1) 138 (93.9) 147 Nighttime 1 (5.3) 18 (94.7) 0 (0) 8 (100.0) 1 (3.7) 26 (96.3) 27 Precipitation level No precipitation 3 (4.4) 65 (95.6) 5 (5.7) 83 (94.3) 8 (5.1) 148 (94.9) 156 Precipitation 0 (0) 4 (100.0) 2 (14.3) 12 (85.7) 2 (11.1) 16 (88.9) 18 Session visibility Good 3 (4.2) 68 (95.8) 7 (6.9) 94 (93.1) 10 (5.8) 162 (94.2) 172 Poor 0 (0) 1 (100.0) 0 (0) 1 (100.0) 0 (0) 2 (100.0) 2 Wind (theoretical) Calm – (–) – (–) 0 (0) 5 (100.0) 0 (0) 5 (100.0) 5 Crosswind 1 (9.1) 10 (90.9) 0 (0) 9 (100.0) 1 (5.0) 19 (95.0) 20 Headwind 1 (5.9) 16 (94.1) 6 (17.6) 28 (82.4) 7 (13.7) 44 (86.3) 51 Tailwind 1 (2.3) 42 (97.7) 1 (1.9) 53 (98.1) 2 (2.0) 96 (98.0) 98 Wind strength Weak 1 (6.3) 15 (93.8) 3 (8.8) 31 (91.2) 4 (8.0) 46 (92.0) 50 Strong 2 (3.6) 54 (96.0) 4 (5.9) 64 (94.1) 6 (4.8) 118 (95.2) 124 Total Total 3 (4.2) 69 (95.8) 7 (6.9) 95 (93.1) 10 (5.7) 164 (94.3) 174

Results

AB

R F

inal Report

99M



Lights

Off On Total


Shorebirds Time of day Daytime 1 (50.0) 1 (50.0) 2 (66.7) 1 (33.3) 3 (60.0) 2 (40.0) 5 Nighttime 6 (24.0) 19 (76.0) 6 (17.6) 28 (82.4) 12 (20.3) 47 (79.7) 59 Precipitation level No precipitation 3 (20.0) 12 (80.0) 4 (18.2) 18 (81.8) 7 (18.9) 30 (81.1) 37 Precipitation 4 (33.3) 8 (66.7) 4 (26.7) 11 (73.3) 8 (29.6) 19 (70.4) 27 Session visibility Good 6 (24.0) 19 (76.0) 8 (21.6) 29 (78.4) 14 (22.6) 48 (77.4) 48 Poor 1 (50.0) 1 (50.0) – (–) – (–) 1 (50.0) 1 (50.0) 2 Wind (theoretical) Calm – (–) – (–) – (–) – (–) – (–) – (–) 0 Crosswind 2 (25.0) 6 (75.0) 1 (10.0) 9 (90.0) 3 (16.7) 15 (83.3) 18 Headwind 3 (30.0) 7 (70.0) 3 (37.5) 5 (62.5) 6 (33.3) 12 (66.7) 18 Tailwind 2 (22.2) 7 (77.8) 4 (21.1) 15 (78.9) 6 (21.4) 22 (78.6) 28 Wind strength Weak 1 (7.1) 13 (92.9) 2 (11.1) 16 (88.9) 3 (9.4) 29 (90.6) 32 Strong 6 (46.2) 7 (53.8) 6 (31.6) 13 (68.4) 12 (37.5) 20 (62.5) 32 Total Total 7 (25.9) 20 (74.1) 8 (21.6) 29 (78.4) 15 (23.4) 49 (76.6) 64 Gulls Time of day Daytime 13 (10.9) 106 (89.1) 13 (16.5) 66 (83.5) 26 (13.1) 172 (86.9) 198 Nighttime 0 (0) 20 (100.0) 3 (25.0) 9 (75.0) 3 (9.4) 29 (90.6) 32 Precipitation level No precipitation 10 (11.4) 78 (86.6) 11 (16.7) 55 (83.3) 21 (13.6) 133 (86.4) 154 Precipitation 3 (5.9) 48 (94.1) 5 (20.0) 20 (80.0) 8 (10.5) 68 (89.5) 76 Session visibility Good 12 (9.4) 115 (90.6) 15 (17.2) 72 (82.8) 27 (12.6) 187 (87.4) 214 Poor 1 (8.3) 11 (91.7) 1 (25.0) 3 (75.0) 2 (12.5) 14 (87.5) 16 Wind (theoretical) Calm 0 (0) 7 (100.0) – (–) – (–) 0 (0) 7 (100.0) 7 Crosswind 5 (18.5) 22 (81.5) 2 (9.5) 19 (90.5) 7 (14.6) 41 (85.4) 48 Headwind 7 (11.9) 52 (88.1) 11 (21.6) 40 (78.4) 18 (16.4) 92 (83.6) 110 Tailwind 1 (2.2) 45 (97.8) 3 (15.8) 16 (84.2) 4 (6.2) 61 (93.8) 65 Wind strength Weak 6 (10.7) 50 (89.3) 5 (7.8) 59 (92.2) 11 (9.2) 109 (90.8) 120 Strong 7 (8.4) 76 (91.6) 11 (40.7) 16 (59.3) 18 (16.4) 92 (83.6) 110 Total Total 13 (9.4) 126 (90.6) 16 (17.6) 75 (82.4) 29 (12.6) 201 (87.4) 230

Results

Migration at N

orthstar Island100

AB

R F

inal Report


Lights

Off On Total


Alcids Time of day Daytime 1 (33.3) 2 (66.7) 1 (16.7) 5 (83.3) 2 (22.2) 7 (77.8) 9 Nighttime 0 (0) 1 (100.0) 2 (66.7) 1 (33.3) 2 (50.0) 2 (50.0) 4 Precipitation level No precipitation 0 (0) 2 (100.0) 2 (28.6) 5 (71.4) 2 (22.2) 7 (77.8) 9 Precipitation 1 (50.0) 1 (50.0) 1 (50.0) 1 (50.0) 2 (50.0) 2 (50.0) 4 Session visibility Good 0 (0) 3 (100.0) 2 (28.6) 5 (71.4) 2 (20.0) 8 (80.0) 10 Poor 1 (100.0) 0 (0) 1 (50.0) 1 (50.0) 2 (66.7) 1 (33.3) 3 Wind (theoretical) Calm 0 (0) 1 (100.0) 0 (0) 1 (100.0) 0 (0) 2 (100.0) 2 Crosswind 0 (0) 1 (100.0) 0 (0) 1 (100.0) 0 (0) 2 (100.0) 2 Headwind 0 (0) 1 (100.0) 0 (0) 1 (100.0) 0 (0) 2 (100.0) 2 Tailwind 1 (100.0) 0 (0) 3 (50.0) 3 (50.0) 4 (57.1) 3 (42.9) 7 Wind strength Weak 0 (0) 2 (100.0) 2 (33.3) 4 (66.7) 2 (25.0) 6 (75.0) 8 Strong 1 (50.0) 1 (50.0) 1 (33.3) 2 (66.7) 2 (40.0) 3 (60.0) 5 Total Total 1 (25.0) 3 (75.0) 3 (33.3) 6 (66.7) 4 (30.8) 9 (69.2) 13

Results

AB

R F

inal Report

101M


Table 25. Island-passing behavior of birds seen visually near Northstar Island, northern Alaska, fall 2001–2003, by species-group, factor, and anti-collision lighting setting. Data are presented as number (%) and n (flocks).

Behavior

Species-group Factor Attribute Lights

No Δ direction/ no Δ altitude

Δ direction/ no Δ altitude

No Δ direction/ Δ altitude

Δ direction/ Δ altitude Flare n

Eiders Time of day Daytime Off 54 (90.0) 3 (5.0) 2 (3.3) 1 (1.7) 0 (0) 60 On 33 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 33 Total 87 (93.5) 3 (3.2) 2 (2.2) 1 (1.1) 0 (0) 93 Nighttime Off 1 (33.3) 1 (33.3) 1 (33.3) 0 (0) 0 (0) 3 On 0 (0) 1 (50.0) 0 (0) 1 (50.0) 0 (0) 2 Total 1 (20.0) 2 (40.0) 1 (20.0) 1 (20.0) 0 (0) 5 Precipitation level No precipitation Off 37 (92.5) 2 (5.0) 1 (2.5) 0 (0) 0 (0) 40 On 26 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 26 Total 63 (95.5) 2 (3.0) 1 (1.5) 0 (0) 0 (0) 66 Precipitation Off 18 (78.3) 2 (8.7) 2 (8.7) 1 (4.3) 0 (0) 23 On 7 (77.8) 1 (11.1) 0 (0) 1 (11.1) 0 (0) 9 Total 25 (78.1) 3 (9.4) 2 (6.3) 2 (6.3) 0 (0) 32 Session visibility Good Off 53 (86.9) 4 (6.6) 3 (4.9) 1 (1.6) 0 (0) 61 On 32 (97.0) 1 (3.0) 0 (0) 0 (0) 0 (0) 33 Total 85 (90.4) 5 (5.3) 3 (3.2) 1 (1.1) 0 (0) 94 Poor Off 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 On 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Total 3 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 3 Wind (theoretical) Calm Off 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 On 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Total 3 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 3 Crosswind Off 15 (83.3) 1 (5.6) 1 (5.6) 1 (5.6) 0 (0) 18 On 3 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 3 Total 18 (85.7) 1 (4.8) 1 (4.8) 1 (4.8) 0 (0) 21 Headwind Off 8 (88.9) 0 (0) 1 (11.1) 0 (0) 0 (0) 9 On 12 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 12 Total 20 (95.2) 0 (0) 1 (4.8) 0 (0) 0 (0) 21 Tailwind Off 30 (88.2) 3 (8.8) 1 (2.9) 0 (0) 0 (0) 34 On 17 (89.5) 1 (5.3) 0 (0) 1 (5.3) 0 (0) 19 Total 47 (88.7) 4 (7.5) 1 (1.9) 1 (1.9) 0 (0) 53 Wind strength Weak Off 16 (94.1) 1 (5.9) 0 (0) 0 (0) 0 (0) 17 On 17 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 17 Total 33 (97.1) 1 (2.9) 0 (0) 0 (0) 0 (0) 34 Strong Off 39 (84.8) 3 (6.5) 3 (6.5) 1 (2.2) 0 (0) 46 On 16 (88.9) 1 (5.6) 0 (0) 1 (5.6) 0 (0) 18 Total 55 (85.9) 4 (6.3) 3 (4.7) 2 (3.1) 0 (0) 64 Total Total Off 55 (87.3) 4 (6.3) 3 (4.8) 1 (1.6) 0 (0) 63 On 33 (94.3) 1 (2.9) 0 (0) 1 (2.9) 0 (0) 35 Total 88 (89.8) 5 (5.1) 3 (3.1) 2 (2.0) 0 (0) 98

Results

Migration at N

orthstar Island102

AB

R F

inal Report

Table 25. Continued. Behavior






Loons Time of day Daytime Off 90 (96.8) 2 (2.1) 1 (1.1) 0 (0) 0 (0) 93 On 102 (95.3) 2 (1.9) 3 (2.8) 0 (0) 0 (0) 107 Total 192 (96.0) 4 (2.0) 4 (2.0) 0 (0) 0 (0) 200 Nighttime Off – (–) – (–) – (–) – (–) – (–) 0 On – (–) – (–) – (–) – (–) – (–) 0 Total – (–) – (–) – (–) – (–) – (–) 0 Precipitation level No precipitation Off 89 (96.7) 2 (2.2) 1 (1.1) 0 (0) 0 (0) 92 On 84 (96.6) 1 (1.1) 2 (2.3) 0 (0) 0 (0) 87 Total 173 (96.7) 3 (1.7) 3 (1.7) 0 (0) 0 (0) 179 Precipitation Off 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 On 18 (90.0) 1 (5.0) 1 (5.0) 0 (0) 0 (0) 20 Total 19 (90.5) 1 (4.8) 1 (4.8) 0 (0) 0 (0) 21 Session visibility Good Off 90 (96.8) 2 (2.2) 1 (1.1) 0 (0) 0 (0) 93 On 101 (95.3) 2 (1.9) 3 (2.8) 0 (0) 0 (0) 106 Total 191 (96.0) 4 (2.0) 4 (4.1) 0 (0) 0 (0) 199 Poor Off – (–) – (–) – (–) – (–) – (–) 0 On 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Total 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Wind (theoretical) Calm Off 8 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 8 On 17 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 17 Total 25 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 25 Crosswind Off 19 (95.0) 1 (5.0) 0 (0) 0 (0) 0 (0) 20 On 17 (94.4) 0 (0) 1 (5.6) 0 (0) 0 (0) 18 Total 36 (94.7) 1 (2.6) 1 (2.6) 0 (0) 0 (0) 38 Headwind Off 41 (95.3) 1 (2.3) 1 (0) 0 (0) 0 (0) 43 On 48 (92.3) 2 (3.8) 2 (3.8) 0 (0) 0 (0) 52 Total 89 (93.7) 3 (3.2) 3 (3.2) 0 (0) 0 (0) 95 Tailwind Off 22 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 22 On 20 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 20 Total 42 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 42 Wind strength Weak Off 40 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 40 On 73 (98.6) 1 (1.4) 0 (0) 0 (0) 0 (0) 74 Total 113 (99.1) 1 (0.9) 0 (0) 0 (0) 0 (0) 114 Strong Off 50 (94.3) 2 (3.8) 1 (1.9) 0 (0) 0 (0) 53 On 29 (87.9) 1 (3.0) 3 (9.1) 0 (0) 0 (0) 33 Total 79 (91.9) 3 (3.5) 4 (4.7) 0 (0) 0 (0) 86 Total Total Off 90 (96.8) 2 (2.2) 1 (1.1) 0 (0) 0 (0) 93 On 102 (95.3) 2 (1.9) 3 (2.8) 0 (0) 0 (0) 107 Total 192 (96.0) 4 (2.0) 4 (2.0) 0 (0) 0 (0) 200

Results

AB

R F

inal Report

103M








Other ducks Time of day Daytime Off 60 (96.8) 1 (1.6) 1 (1.6) 0 (0) 0 (0) 62 On 57 (95.0) 3 (5.0) 0 (0) 0 (0) 0 (0) 60 Total 117 (95.9) 4 (3.3) 1 (0.8) 0 (0) 0 (0) 122 Nighttime Off 13 (54.2) 8 (33.3) 2 (8.3) 0 (0) 1 (4.2) 24 On 13 (61.9) 3 (14.3) 1 (4.8) 0 (0) 4 (19.0) 21 Total 26 (57.8) 11 (24.4) 3 (6.8) 0 (0) 5 (11.1) 45 Precipitation level No precipitation Off 61 (92.4) 4 (6.1) 1 (1.5) 0 (0) 0 (0) 66 On 54 (85.7) 5 (7.9) 0 (0) 0 (0) 4 (6.3) 63 Total 115 (89.2) 9 (7.0) 1 (0.8) 0 (0) 4 (3.1) 129 Precipitation Off 12 (60.0) 5 (25.0) 2 (10.0) 0 (0) 1 (5.0) 20 On 16 (88.9) 1 (5.6) 1 (5.6) 0 (0) 0 (0) 18 Total 28 (73.7) 6 (15.8) 3 (7.7) 0 (0) 1 (2.6) 38 Session visibility Good Off 70 (87.5) 9 (11.2) 1 (1.3) 0 (0) 0 (0) 80 On 69 (86.3) 6 (7.5) 1 (1.3) 0 (0) 4 (5.0) 80 Total 139 (86.9) 15 (9.4) 2 (1.3) 0 (0) 4 (2.5) 160 Poor Off 3 (50.0) 0 (0) 2 (33.3) 0 (0) 1 (16.7) 6 On 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Total 4 (57.1) 0 (0) 2 (28.6) 0 (0) 1 (14.3) 7 Wind (theoretical) Calm Off 5 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 5 On 4 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 4 Total 9 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 9 Crosswind Off 8 (88.9) 0 (0) 1 (11.1) 0 (0) 0 (0) 9 On 24 (96.0) 1 (4.0) 0 (0) 0 (0) 0 (0) 25 Total 32 (94.1) 1 (2.9) 1 (2.9) 0 (0) 0 (0) 34 Headwind Off 30 (81.1) 5 (13.5) 2 (5.4) 0 (0) 0 (0) 37 On 26 (72.2) 5 (13.9) 1 (2.8) 0 (0) 4 (11.1) 36 Total 56 (76.7) 10 (13.7) 3 (4.1) 0 (0) 4 (5.5) 73 Tailwind Off 30 (85.7) 4 (11.4) 0 (0) 0 (0) 1 (2.9) 35 On 16 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 16 Total 46 (90.2) 4 (7.8) 0 (0) 0 (0) 1 (2.0) 51 Wind strength Weak Off 30 (83.3) 5 (13.9) 0 (0) 0 (0) 1 (2.8) 36 On 37 (80.4) 5 (10.9) 0 (0) 0 (0) 4 (8.7) 46 Total 67 (81.7) 10 (12.2) 0 (0) 0 (0) 5 (6.1) 82 Strong Off 43 (86.0) 4 (8.0) 3 (6.0) 0 (0) 0 (0) 50 On 33 (94.3) 1 (2.9) 1 (2.9) 0 (0) 0 (0) 35 Total 76 (89.4) 5 (5.9) 4 (4.7) 0 (0) 0 (0) 85 Total Total Off 73 (84.9) 9 (10.5) 3 (3.5) 0 (0) 1 (1.2) 86 On 70 (86.4) 6 (7.4) 1 (1.2) 0 (0) 4 (4.9) 81 Total 143 (85.6) 15 (9.0) 4 (2.4) 0 (0) 5 (3.0) 167

Results

Migration at N

orthstar Island104

AB

R F

inal Report







Unid. ducks Time of day Daytime Off 56 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 56 On 84 (98.8) 0 (0) 1 (1.2) 0 (0) 0 (0) 85 Total 140 (99.3) 0 (0) 1 (0.7) 0 (0) 0 (0) 141 Nighttime Off 19 (90.5) 2 (9.5) 0 (0) 0 (0) 0 (0) 21 On 8 (80.0) 2 (20.0) 0 (0) 0 (0) 0 (0) 10 Total 27 (87.1) 4 (12.9) 0 (0) 0 (0) 0 (0) 31 Precipitation level No precipitation Off 72 (98.6) 1 (1.4) 0 (0) 0 (0) 0 (0) 73 On 81 (96.4) 2 (2.4) 1 (1.2) 0 (0) 0 (0) 84 Total 153 (97.5) 3 (1.9) 1 (0.6) 0 (0) 0 (0) 157 Precipitation Off 3 (75.0) 1 (25.0) 0 (0) 0 (0) 0 (0) 4 On 11 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 11 Total 14 (93.3) 1 (6.7) 0 (0) 0 (0) 0 (0) 15 Session visibility Good Off 74 (97.4) 2 (2.6) 0 (0) 0 (0) 0 (0) 76 On 91 (96.8) 2 (2.1) 1 (1.1) 0 (0) 0 (0) 94 Total 165 (97.1) 4 (2.4) 1 (0.6) 0 (0) 0 (0) 170 Poor Off 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 On 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Total 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 Wind (theoretical) Calm Off 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 On 5 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 5 Total 6 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 6 Crosswind Off 11 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 11 On 10 (90.9) 1 (9.1) 0 (0) 0 (0) 0 (0) 11 Total 21 (95.5) 1 (4.5) 0 (0) 0 (0) 0 (0) 22 Headwind Off 19 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 19 On 25 (96.2) 0 (0) 1 (3.8) 0 (0) 0 (0) 26 Total 44 (97.8) 0 (0) 1 (2.2) 0 (0) 0 (0) 45 Tailwind Off 44 (95.7) 2 (4.3) 0 (0) 0 (0) 0 (0) 46 On 52 (98.1) 1 (1.9) 0 (0) 0 (0) 0 (0) 53 Total 96 (97.0) 3 (3.0) 0 (0) 0 (0) 0 (0) 99 Wind strength Weak Off 18 (90.0) 2 (10.0) 0 (0) 0 (0) 0 (0) 20 On 29 (96.7) 1 (3.3) 0 (0) 0 (0) 0 (0) 30 Total 47 (94.0) 3 (6.0) 0 (0) 0 (0) 0 (0) 50 Strong Off 57 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 57 On 63 (96.9) 1 (1.5) 1 (1.5) 0 (0) 0 (0) 65 Total 120 (98.4) 1 (0.8) 1 (0.8) 0 (0) 0 (0) 122 Total Total Off 75 (97.4) 2 (2.6) 0 (0) 0 (0) 0 (0) 77 On 92 (96.8) 2 (2.1) 1 (1.1) 0 (0) 0 (0) 95 Total 167 (97.1) 4 (2.4) 1 (0.6) 0 (0) 0 (0) 172

Results

AB

R F

inal Report

105M








Shorebirds Time of day Daytime Off 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 On 2 (66.7) 0 (0) 1 (33.3) 0 (0) 0 (0) 3 Total 3 (75.0) 0 (0) 1 (25.0) 0 (0) 0 (0) 4 Nighttime Off 11 (47.8) 6 (26.1) 3 (13.0) 1 (4.3) 2 (8.7) 23 On 16 (45.7) 9 (25.7) 3 (8.6) 7 (20.0) 0 (0) 35 Total 27 (46.6) 15 (25.9) 6 (10.3) 8 (13.8) 2 (3.4) 58 Precipitation level No precipitation Off 4 (30.8) 4 (30.8) 2 (15.4) 1 (7.7) 2 (15.4) 13 On 12 (50.0) 6 (25.0) 1 (4.2) 5 (20.8) 0 (0) 24 Total 16 (43.2) 10 (27.0) 3 (8.2) 6 (16.2) 2 (5.4) 37 Precipitation Off 8 (72.7) 2 (18.2) 1 (9.1) 0 (0) 0 (0) 11 On 6 (42.9) 3 (21.4) 3 (21.4) 2 (14.3) 0 (0) 14 Total 14 (56.0) 5 (20.0) 4 (16.0) 2 (8.0) 0 (0) 25 Session visibility Good Off 10 (45.5) 6 (27.3) 3 (13.6) 1 (4.5) 2 (9.1) 22 On 18 (47.4) 9 (23.7) 4 (10.5) 7 (18.4) 0 (0) 38 Total 28 (46.7) 15 (25.0) 7 (11.7 ) 8 (13.3) 2 (3.3) 60 Poor Off 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 On – (–) – (–) – (–) – (–) – (–) 0 Total 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 Wind (theoretical) Calm Off – (–) – (–) – (–) – (–) – (–) 0 On – (–) – (–) – (–) – (–) – (–) 0 Total – (–) – (–) – (–) – (–) – (–) 0 Crosswind Off 4 (66.7) 1 (16.7) 1 (16.7) 0 (0) 0 (0) 6 On 5 (45.5) 3 (27.3) 2 (18.2) 1 (9.1) 0 (0) 11 Total 9 (52.9) 4 (23.5) 3 (17.6) 1 (5.9) 0 (0) 17 Headwind Off 4 (44.4) 3 (33.3) 1 (11.1) 0 (0) 1 (11.1) 9 On 4 (50.0) 1 (12.5) 1 (12.5) 2 (25.0) 0 (0) 8 Total 8 (47.1) 4 (23.5) 2 (11.8) 2 (11.8) 1 (5.9) 17 Tailwind Off 4 (44.4) 2 (22.2) 1 (11.1) 1 (11.1) 1 (11.1) 9 On 9 (47.4) 5 (26.3) 1 (5.3) 4 (21.1) 0 (0) 19 Total 13 (46.4) 7 (25.0) 2 (7.1) 5 (17.9) 1 (3.6) 28 Wind strength Weak Off 3 (23.1) 5 (38.5) 3 (23.1) 0 (0) 2 (15.4) 13 On 9 (40.9) 7 (31.8) 3 (13.6) 3 (13.6) 0 (0) 22 Total 12 (34.3) 12 (34.3) 6 (17.1) 3 (8.6) 2 (5.7) 35 Strong Off 9 (81.8) 1 (9.1) 0 (0) 1 (9.1) 0 (0) 11 On 9 (56.3) 2 (12.5) 1 (6.3) 4 (25.0) 0 (0) 16 Total 18 (66.7) 3 (11.1) 1 (3.7) 5 (18.5) 0 (0) 27 Total Total Off 12 (50.0) 6 (25.0) 3 (12.5) 1 (4.2) 2 (8.3) 24 On 18 (47.4) 9 (23.7) 4 (10.5) 7 (18.4) 0 (0) 38 Total 30 (48.4) 15 (24.2) 7 (11.3) 8 (12.9) 2 (3.2) 62

Results

Migration at N

orthstar Island106

AB

R F

inal Report







Gulls Time of day Daytime Off 106 (89.1) 6 (5.0) 7 (5.9) 0 (0) 0 (0) 119 On 75 (92.6) 3 (3.8) 2 (2.5) 1 (1.2) 0 (0) 81 Total 181 (90.5) 8 (4.5) 9 (4.5) 1 (0.5) 0 (0) 200 Nighttime Off 19 (79.2) 4 (16.7) 0 (0) 1 (4.2) 0 (0) 24 On 10 (71.4) 0 (0) 4 (28.6) 0 (0) 0 (0) 14 Total 29 (76.3) 4 (10.5) 4 (10.5) 1 (2.6) 0 (0) 38 Precipitation level No precipitation Off 76 (85.4) 8 (9.0) 5 (4.5) 1 (1.1) 0 (0) 89 On 62 (91.2) 3 (4.4) 3 (4.4) 0 (0) 0 (0) 68 Total 138 (87.9) 11 (7.0) 7 (4.5) 1 (0.6) 0 (0) 157 Precipitation Off 49 (90.7) 2 (3.7) 3 (5.6) 0 (0) 0 (0) 54 On 23 (85.2) 0 (0) 3 (11.1) 1 (3.7) 0 (0) 27 Total 72 (88.9) 2 (2.5) 6 (7.4) 1 (1.2) 0 (0) 81 Session visibility Good Off 114 (89.1) 9 (7.0) 4 (3.1) 1 (0.8) 0 (0) 128 On 80 (88.9) 3 (3.3) 6 (6.7) 1 (1.1) 0 (0) 90 Total 194 (89.0) 12 (5.5) 10 (4.6) 2 (0.9) 0 (0) 218 Poor Off 11 (73.3) 1 (6.7) 3 (20.0) 0 (0) 0 (0) 15 On 5 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 5 Total 16 (80.0) 1 (5.0) 3 (15.0) 0 (0) 0 (0) 20 Wind (theoretical) Calm Off 7 (87.5) 1 (12.5) 0 (0) 0 (0) 0 (0) 8 On – (–) – (–) – (–) – (–) – (–) 0 Total 7 (87.5) 1 (12.5) 0 (0) 0 (0) 0 (0) 8 Crosswind Off 21.0 (80.8) 3 (11.5) 2 (7.7) 0 (0) 0 (0) 26 On 22 (88.0) 0 (0) 2 (8.0) 1 (4.0) 0 (0) 25 Total 43 (84.3) 3 (5.9) 4 (7.8) 1 (2.0) 0 (0) 51 Headwind Off 51 (86.4) 2 (3.4) 5 (8.5) 1 (1.7) 0 (0) 59 On 45 (93.8) 2 (4.2) 1 (2.1) 0 (0) 0 (0) 48 Total 96 (89.7) 4 (3.7) 6 (5.6) 1 (0.9) 0 (0) 107 Tailwind Off 46 (92.0) 4 (8.0) 0 (0) 0 (0) 0 (0) 50 On 18 (81.8) 1 (4.5) 3 (13.6) 0 (0) 0 (0) 22 Total 64 (88.9) 5 (6.9) 3 (4.2) 0 (0) 0 (0) 72 Wind strength Weak Off 51 (86.4) 6 (10.2) 1 (1.7) 1 (1.7) 0 (0) 59 On 62 (93.9) 1 (1.5) 3 (4.5) 0 (0) 0 (0) 66 Total 113 (90.4) 7 (5.6) 4 (3.2) 1 (0.8) 0 (0) 125 Strong Off 74 (88.1) 4 (4.8) 6 (7.1) 0 (0) 0 (0) 84 On 23 (79.3) 2 (6.9) 3 (10.3) 1 (3.4) 0 (0) 29 Total 97 (85.8) 6 (5.3) 9 (8.0) 1 (0.9) 0 (0) 113 Total Total Off 125 (87.4) 10 (7.0) 7 (4.9) 1 (0.7) 0 (0) 143 On 85 (89.5) 2 (3.2) 6 (6.3) 1 (1.1) 0 (0) 95 Total 210 (88.2) 13 (5.5) 13 (5.5) 2 (0.8) 0 (0) 238

Results

AB

R F

inal Report

107M








Alcids Time of day Daytime Off 2 (66.7) 1 (33.3) 0 (0) 0 (0) 0 (0) 3 On 5 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 5 Total 7 (87.5) 1 (12.5) 0 (0) 0 (0) 0 (0) 8 Nighttime Off 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 On 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 Total 3 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 3 Precipitation level No precipitation Off 2 (66.7) 1 (33.3) 0 (0) 0 (0) 0 (0) 3 On 6 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 6 Total 8 (88.9) 1 (11.1) 0 (0) 0 (0) 0 (0) 9 Precipitation Off 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 On 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Total 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 Session visibility Good Off 3 (75.0) 1 (25.0) 0 (0) 0 (0) 0 (0) 4 On 6 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 6 Total 9 (90.0) 1 (10.0) 0 (0) 0 (0) 0 (0) 10 Poor Off – (–) – (–) – (–) – (–) – (–) 0 On 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Total 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Wind (theoretical) Calm Off 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 On 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Total 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 Crosswind Off 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 On 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Total 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 Headwind Off 1 (50.0) 1 (50.0) 0 (0) 0 (0) 0 (0) 2 On 1 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 1 Total 2 (66.7) 1 (33.3) 0 (0) 0 (0) 0 (0) 3 Tailwind Off – (–) – (–) – (–) – (–) – (–) 0 On 4 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 4 Total 4 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 4 Wind strength Weak Off 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 On 5 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 5 Total 7 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 7 Strong Off 1 (50.0) 1 (50.0) 0 (0) 0 (0) 0 (0) 2 On 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2 Total 3 (75.0) 1 (25.0) 0 (0) 0 (0) 0 (0) 4 Total Total Off 3 (75.0) 1 (25.0) 0 (0) 0 (0) 0 (0) 4 On 7 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 7 Total 10 (90.9) 1 (9.1) 0 (0) 0 (0) 0 (0) 11

Results


direction, flight altitude, both, or behavioralflaring; changes in direction were most common,however. Only other ducks and shorebirds wereseen behavioral flaring, which is the most dramaticresponse of birds to possible collision; thisbehavioral flaring was seen with both theanti-collision lights off and on. For all non-eiderspecies-groups, the percentage exhibitingresponses tended to be similar betweenanti-collision light settings. For shorebirds andgulls, the percentage exhibiting responses wasmuch higher at night than during the daytime,suggesting that there was an exacerbating effect ofthe island.

EFFECTS OF DISTANCE TO ISLAND ON RESPONSES

Because it was possible that responses to theisland might differ with distance from the island,we also stratified the data as “near” (i.e., the closestdistance the flock of birds was seen was ≤500 mfrom the island) and “far” (i.e., closest distance>500 m from the island) and examineddistance-caused effects on passing success andbehavior. For eiders, passing success was slightlyhigher in near flocks, whereas behavioralresponses were slightly more frequent in nearflocks (Table 26). In both cases, however,differences were slight. Similar to the pattern seenfor eiders, all non-eider taxa except alcids (whichhad small sample sizes) also had a higher passingsuccess in near flocks and a higher frequency ofbehavioral responses in near flocks.

DOWNING AND MORTALITY

Eiders and/or Long-tailed Ducks were eitherdowned alive or killed at Northstar Island on 17nights in the falls of 2001–2004 (Table 27). Inaddition to these data, to enhance ourunderstanding of patterns of downing and mortalityof these birds, we have included with these resultsdata from one night in which 16 eiders weredowned at nearby Endicott Island on 27 October2001. Both eiders and Long-tailed Ducks werekilled on two nights, only eiders were downed onnine nights, and only Long-tailed Ducks weredowned on seven nights.

Altogether, we have information on 36 eiders(20 from Northstar Island and 16 from EndicottIsland), with data on one additional eider from

Northstar Island being unavailable; all weredowned in fall 2001, 2003, or 2004. Altogether, wehave data on 13 Long-tailed Ducks, all fromNorthstar Island; 8 were recorded in fall 2001, 3were recorded in fall 2002, and 2 were recorded infall 2003. Because the automated weather station atNorthstar Island was out of service during most ofthe fall of 2001, we were unable to determine theprobability of fog on all nights in 2001 on whichbirds were downed (Table 27).

In 2001, downings occurred in two clusters,one between 26 September and 4 October and onebetween 24 and 27 October (Table 27). In 2002,downings occurred in one cluster of10–11 October. In 2003, downings occurred onfour widely separated dates. In 2004, downingsoccurred in two clusters separated by a month. Inall cases, we assumed that the birds that werefound on a particular date were downed on theprevious night.

When the patterns of downing are examinedwith nights as the sampling unit, eiders exhibitedtwo main patterns, one related to the lunar cycleand one related to the barometric trend (Table 28).In terms of lunar patterns, ~64% of the nights onwhich eiders were downed occurred during a FullMoon, and ~91% occurred during a waxing moon.In terms of barometric patterns, ~73% of the nightson which eiders were downed occurred during aweakly changing barometer. The overall windspeed was 10.2 mi/h (~16 km/h) on nights onwhich downings occurred (lower than averagenightly conditions), and the overall wind directionwas 022° (nearly identical to average nightlyconditions); the very limited information on theprobability of fog suggested that eiders weredowned during periods of fog. In contrast, over the4 years of this study, a Full Moon has occurredonly during ~36% of all nights, and waxing andwaning moons each have occurred on ~50% of allnights. (A Full Moon occurs on ~33% of nights,based on our definition, but the lunar cycle variesmonthly and annually, so we had an extra FullMoon during the four-month period in one year,making the percentage slightly higher overall.) Thebarometric patterns also differed in terms ofnightly conditions, in that the barometer wasweakly changing on only ~57% of the nights; someof this higher percentage occurred because a steady

Results

AB

R F

inal Report

109M


Table 26. Frequencies and percentages of passing success and behavioral responses of birds passing Northstar Island, northern Alaska, fall 2001–2003, by closest distance to island. “Near” refers to the closest visually determined distance <500 m from the island, and “far” refers to 500 m from the island. Data are presented as number (%) and n (flocks).

Behavior

Species-group

Distance

Successful

(number, %)

n

No Δ direction/ no

Δ altitude

Δ direction/ no Δ

altitude


Δ direction/ Δ altitude

Flare

n

Eiders Near 51 (92.7) 55 42 (84.0) 5 (10.0) 1 (2.0) 2 (4.0) 0 (0) 50 Far 47 (90.4) 52 46 (95.8) 0 (0) 2 (4.2) 0 (0) 0 (0) 48 Loons Near 50 (89.3) 56 54 (93.1) 3 (5.2) 1 (1.7) 0 (0) 0 (0) 58 Far 109 (77.9) 140 138 (97.2) 1 (0.7) 3 (2.1) 0 (0) 0 (0) 142 Other ducks Near 88 (91.7) 96 73 (75.3) 15 (15.5) 4 (4.1) 0 (0) 5 (5.2) 97 Far 63 (82.9) 76 69 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 69 Unid. ducks Near 37 (97.4) 38 38 (90.5) 4 (9.5) 0 (0) 0 (0) 0 (0) 42 Far 127 (93.4) 136 129 (99.2) 0 (0) 1 (0.8) 0 (0) 0 (0) 130 Shorebirds Near 47 (77.0) 61 27 (45.8) 15 (25.4) 7 (11.9) 8 (13.6) 2 (3.4) 59 Far 2 (66.7) 3 3 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 3 Gulls Near 159 (87.8) 181 166 (86.0) 13 (6.7) 12 (6.2) 2 (1.0) 0 (0) 193 Far 41 (85.4) 48 42 (97.7) 0 (0) 1 (2.3) 0 (0) 0 (0) 43 Alcids Near 8 (66.7) 12 8 (88.9) 1 (11.1) 0 (0) 0 (0) 0 (0) 9 Far 1 (100.0) 1 2 (100.0) 0 (0) 0 (0) 0 (0) 0 (0) 2

Results

Migration at N

orthstar Island110

AB

R F

inal Report

Table 27. Environmental and other characteristics associated with the downing of Common and King eiders and Long-tailed Ducks at Northstar Island, 2001–2004; data on downed eiders from nearby Endicott Island in fall (27 October) 2001 also are included. Wind speed is in mi/h.

Date Number of birds a Previous Moon Mean wind Barometric Fog

found COEI KIEI LTDU Total night Phase Trend Speed Direction Trend Rate probable?

09/26/01 0 0 2 2 9/25/01 not full waxing 13.7 028° rising strongly – 09/27/01 0 1 0 1 9/26/01 full waxing 5.6 028° rising strongly – 09/28/01 6 1 2 9 9/27/01 full waxing 10.5 007° rising weakly – 09/29/01 0 1 1 2 9/28/01 full waxing 9.8 003° rising weakly – 10/01/01 0 1 0 1 9/30/01 full waxing 22.6 005° falling weakly – 10/02/01 0 0 2 2 10/1/01 full waxing 17.2 004° rising weakly – 10/04/01 1 0 0 1 10/3/01 full waning 4.1 010° rising weakly – 10/24/01 1 0 0 1 10/23/01 not full waxing 7.7 004° rising strongly – 10/27/01 15 1 0 16 10/26/01 full waxing 7.8 024° falling weakly – 11/08/01 0 0 1 1 11/7/01 not full waning 12.9 021° falling weakly – 10/10/02 0 0 1 1 10/9/02 not full waxing 17.5 028° rising strongly yes 10/11/02 0 0 2 2 10/10/02 not full waxing 10.8 024° rising strongly yes 10/16/03 0 0 1 1 10/15/03 not full waning 20.8 001° falling strongly yes 10/26/03 2 0 0 2 10/25/03 not full waxing 8.0 023° falling weakly yes 10/28/03 0 0 1 1 10/27/03 not full waxing 22.2 025° rising strongly yes 11/26/03 2 0 0 2 11/25/03 not full waxing 17.1 024° rising strongly – 09/23/04 0 1 b 0 1 09/22/04 not full waxing 11.2 104° falling weakly – 10/23/04 3 0 0 3 10/22/04 full waxing 8.3 191° falling weakly yes

a COEI = Common Eider; KIEI = King Eider; LTDU = Long-tailed Duck. b Actually an unidentified eider but probably of this species.

Results

AB

R F

inal Report

111M


Table 28. Patterns of downing of Common and King eiders and Long-tailed Ducks at Northstar Island, 2001–2004, by night and by individual bird. For comparison, patterns across all nights of data also are included.

Moon Mean wind Barometric

Summary/taxon Phase Trend Speed (mi/h) Direction Trend Rate of change Fog probable?

NIGHTS Eiders full (63.6%) waxing (90.9%) 10.2 022° rising (54.5%) weakly (72.7%) no (0%) a (n = 11) not full (36.4%) waning (9.1%) falling (45.5%) strongly (27.3%) yes (100.0%) a Long-tailed Ducks full (33.3%) waxing (77.8%) 15.0 016° rising (77.8%) weakly (55.6%) no (0%) b (n = 9) not full (66.7%) waning (22.2%) falling (22.2%) strongly (44.4%) yes (100.0) b Nightly conditions full (35.8%) waxing (50.5%) 13.7 c 016° d rising (45.6%) weakly (57.1%) no (59.3) e (n = 488) not full (64.2%) waning (49.5%) falling (49.7%) strongly (42.9%) yes (40.7) e steady (4.6%) INDIVIDUALS Eiders full (83.3%) waxing (97.2%) 9.3 021° rising (36.1%) weakly (88.9%) no (0%) f (n = 36) not full (16.7%) waning (2.8%) falling (63.9%) strongly (11.1%) yes (100.0%) f Long-tailed Ducks full (38.5%) waxing (84.6%) 14.4 016° rising (84.6%) weakly (46.2%) no (0%) b (n = 13) not full (61.5%) waning (15.4%) falling (15.4%) strongly (53.8%) yes (100.0%) b

a n = 1. b n = 4. c n = 286. d n = 397. e n = 216. Data are available only from August through October 2001–2003. f n = 3.

Discussion


barometer also was defined as having a weaknightly change.

When the patterns of downing are examinedwith nights as the sampling unit, Long-tailedDucks exhibited two main patterns, one related tothe lunar cycle and one related to the barometrictrend (Table 28). In terms of lunar patterns, ~78%of the nights on which these birds were downedoccurred during a waxing moon; the proportion ofFull Moons, however, was similar to what waspresent on average nightly conditions. In terms ofbarometric patterns, ~78% of the nights on whichthese birds were downed occurred during a risingbarometer; in contrast, the percentage occurringduring a weakly changing barometer was similar toaverage nightly conditions. Overall wind speedwas 15.0 mi/h (~24 km/h) on nights on which thesebirds were downed, and overall wind direction was016° (both similar to or identical to average nightlyconditions). The four nights that we have been ableto examine probably had fog; in contrast, fogprobably occurs only ~41% of the time duringAugust–October and may occur less frequentlythan that, depending on when the ocean freezes inOctober.

When the patterns of downing are examinedwith individual birds as the sampling unit, eidersagain exhibited two main patterns, one related tothe lunar cycle and one related to the barometrictrend (Table 28). In terms of lunar patterns, ~83%of the eiders were downed during a Full Moon, and~97% were downed during a waxing moon (bothmuch higher than the percentage of nights onwhich a Full Moon and a waxing moon occurred).In terms of barometric patterns, ~89% of the birdswere downed during a weakly changing barometer(much higher than the percentage of nights onwhich the barometer was weakly changing),whereas the percentage of birds that were downedduring a rising barometer was slightly lower thanthe percentage of nights on which the barometerwas rising. The overall wind speed was 9.3 mi/h(~15 km/h) during downings (lower than averagenightly conditions), and the overall wind directionwas 021° (similar to average nightly conditions).Again, the few eiders for which we had dataprobably were downed during foggy conditions.

When the patterns of downing are examinedwith individual birds as the sampling unit,Long-tailed Ducks again exhibited two main

patterns, one related to the lunar cycle and onerelated to the barometric trend (Table 28). In termsof lunar patterns, ~85% of the birds were downedduring a waxing moon (much higher than thepercentage of nights on which a waxing moonoccurred), whereas the percentage that wasdowned during a Full Moon was similar to thepercentage of nights on which a Full Moonoccurred. In terms of barometric patterns, ~85% ofthese birds were downed during a rising barometer(much higher than the percentage of nights onwhich the barometer was rising); in contrast, thepercentage downed during a weakly changingbarometer was similar to the percentage of nightson which the barometer was weakly changing.Overall wind speed was 14.4 mi/h duringdownings (similar to average nightly conditions),and overall wind direction was 016° (identical toaverage nightly conditions). Again, all birds aboutwhich we have the information were downedduring nights on which fog probably occurred.

DISCUSSION

“EIDERS”

Over the 4 years, we collected data on 928radar targets that we believe were flocks of eidersof various species. If all of these targets actuallywere eiders, and if all targets had an average flocksize similar to that for flocks that we were able tolocate visually (23.4 birds/flock), we would havecollected data on ~21,750 migrating eiders in thevicinity of Northstar Island during the four years ofthe study. On the other hand, if we assume anidentification accuracy rate of =62% (assumed tobe higher than the calculated rate of 45%), wewould have collected data on ~13,500 migratingeiders. For reference, ~450,000+ eiders migratepast Barrow each fall (Suydam et al. 2000a,2000b).

MOVEMENT RATES“Eiders” exhibited a highly pulsed pattern of

migration timing at Northstar Island in all fouryears. Movement rates were low–moderate in lateAugust–early September 2001 and in 2002, 2003,and 2004, whereas they were moderate–highduring early to mid-September 2001. Movementrates varied dramatically among nights, indicatingpulses of migratory movement, presumably

Discussion


reflecting pulses of weather favorable orunfavorable for migration. This pulsed movementpattern of eider migration also has been seen atBarrow (Thompson and Person 1963, Suydam etal. 1997; Day et al. 2001, 2004b) and along theBeaufort Sea coast of Alaska near present-dayNorthstar Island (Johnson and Richardson 1982).

Movement rates of “eiders” at NorthstarIsland were affected significantly by period (i.e.,the presence or absence of sea ice), precipitationlevel, and wind direction. In contrast, time of day,wind strength, and anti-collision lights (see below)had no effect on movement rates. Movement rateswere highest when ice was present and were muchlower at other times, were higher during periods ofno precipitation, and were highest during tailwindsand crosswinds and lowest during calm winds andheadwinds. At Barrow, periods of heavy fog (i.e.,precipitation and poor visibility) also depressmovement rates (Day et al. 2004b). Migrating birdsin general avoid areas of precipitation (Alerstam1990, Richardson 1990), although one wouldassume that the effect would be smaller onwaterbirds than on terrestrial passerines and that itwould be smaller in fog than in heavy rain. AtBarrow, tailwinds also resulted in highermovement rates and headwinds resulted in lowermovement rates (Thompson and Person 1963, Dayet al. 2004b); wind direction*wind strength alsoaffected movement rates (Day et al. 2004b),although we did not see a significant effect in thisstudy. It is clear that a substantial number of eidersmigrated past Northstar Island at night. Similar tothe results of this study, the research at Barrowfound no difference in migration rates by time ofday (Day et al. 2004b). Alerstam et al. (1974)found that ~20% of the total spring migration ofCommon Eiders in the Baltic Sea occurs at night,indicating that nocturnal migration of eiders mayoccur if migration conditions are good.

FLIGHT VELOCITY“Eiders” averaged overall flight velocities of

~48 mi/h at Northstar Island. Factors thatsignificantly affected flight velocity were period,relative wind direction, wind direction*windstrength, and a lights*time of day interaction. Mostencouraging, the time of day*lights interactionindicated that “eiders” slowed significantly when

the lights were on at night, indicating that theywere seeing the lights and responding to them.

“Eiders” flew ~48 mi/h (~77 km/h) when icewas present but ~47 mi/h (~76 km/h) when ice wasabsent, flew ~49 mi/h (~79 km/h) with a tailwind,~45 mi/h (~72 km/h) with a headwind, and~46 mi/h (~74 km/h) with a crosswind or calmwinds. During spring migration in the Baltic Sea,most Common Eiders fly at groundspeeds of45–70 mi/h (70–110 km/h; Alerstam et al. 1974).At Barrow, radar targets visually identified aseiders and “eiderlike” targets (similar to our“eiders” category) averaged ~52 mi/h (~84 km/h)overall, slightly higher than what we recordedhere; velocities were higher with good visibility,higher with crosswinds and tailwinds thanheadwinds, higher with weak headwinds thanstrong ones, and higher with strong tailwinds andcrosswinds than weak ones (Day et al. 2004b).

FLIGHT DIRECTION“Eiders” showed a strong and consistent mean

flight direction of 299° at Northstar Island whenice was present and 281° when ice was absent;however, these apparent differences appeared to becaused more by among-period differences in winddirections (especially headwinds, which causedbirds to fly more toward the southwest) than bydifferences in where the birds were going. Inaddition, these differences may have beenstatistically significant, but they were notbiologically meaningful. Thus, the overall flightdirection across all years combined was 294°. Theoverall distribution of flight directions wasbimodal, with a large westerly/northwesterlycomponent and a small easterly/southeasterlycomponent. There are several possible reasons forthis unanticipated easterly component. First, thiseasterly movement seems to occur primarily withcalmer wind conditions, so perhaps some of theeasterly heading targets were eiders that simplywere making local movements prior to the largewestward migration push. Another possibleexplanation is that the imperfect targetidentification suggests that some of these easterlyheading targets may have been of otherspecies-groups. Third, there are polynyas (e.g.,Bathurst Polynya) in the eastern Beaufort Seawhere eiders can overwinter (Richardson andJohnson 1981; World Wildlife Fund Canada poster

Discussion


at www.wwwf.ca/NewsAndFacts/Newsroom/Supplemental/Migratory), so perhaps some ofthese easterly heading targets were heading there tooverwinter. Fourth, and similar to the results foundhere, a small percentage of the eiders migratingalong the Beaufort Sea coast in summer is headingeast, rather than west (Johnson and Richardson1982), recent telemetry studies on eiders fromPrudhoe Bay indicate brief periods of eastwardmovement (Troy 2003), and ~5% of the radartargets examined by Flock (1973: 268) on twonights in late August 1969 were heading indirections other than northwest. Finally, it ispossible that some of these easterly heading birdswere undergoing reverse migration, although thisphenomenon has not been documented previouslyfor waterbirds (Alerstam 1990).

The only factor that had a significant effect onflight direction was wind direction. “Eiders” flewtoward the northwest with crosswinds andtailwinds, toward the southwest with headwinds,and toward the northeast with calm conditions. Theonly one of these results that is truly surprising isthe mean eastward vector during calm conditions.These calm conditions occurred primarily duringthree consecutive days of 27–29 August in 2001,so perhaps the early date and the good weatherconditions resulted in local movements that had anoverall vector to the east. There also was highvariability in flight directions during this calmweather, and 10 of the 29 “eider” targets seenduring this calm weather were heading in anorthwesterly direction.

The overall direction from Northstar Island tothe base of Barrow Spit (Fig. 1) is ~290°, as is theoverall direction of the nearby coastline fromFoggy Island Bay to east of Oliktok Point (i.e.,before it starts angling southwestward toward theColville River Delta). Thus, eiders passingNorthstar Island are flying almost exactly thedirection they need to be going to get to Barrow. Atthis time, however, we cannot determine whetherthese birds are flying a straight-line course towardBarrow or they simply are following the nearestcoastline; however, coastlines in general have aleading-line effect that results in paralleling flight(Alerstam 1990). Some of the deflection of“eiders” around Northstar Island may account forthe small difference between the geographicdirection these birds need to be heading and the

direction they actually are heading, or it maysimply reflect collision-avoidance behavior: inSweden, migrating Common Eiders clearly see andchange course to avoid hitting offshore windturbines (Anonymous 2001). In a general sense,however, the overall vector of these birds towardthe northwest reflects the fact that the only waythat birds migrating in a westerly direction over theocean can leave the Beaufort Sea is by flying in adirection of ~290°.

FLIGHT BEHAVIOR“Eiders” exhibited little variation in flight

behavior, with ~95% flying in a straight-line(directional) manner and ~5% flying in an erraticor circling (non-directional) manner; the <1% ofbirds exhibiting circling behavior represented only6 targets, all seen in 2001. Although there was littlevariation in the proportion of non-directional flightbehavior, several factors appear to be important inaffecting it: period, wind direction, wind strength,and a moon phase*moon visibility interaction. Theproportion exhibiting non-directional flight wassignificantly higher when ice was present, higherwith tailwinds and crosswinds, higher during weakwinds, and (when there was a Full Moon) higherwhen the moon was not visible than when it wasvisible. These last results suggest that eiders mayuse the Full Moon for help in orientation duringmigration, resulting in confusion in flight behavior(non-directional flight) and/or attraction to largesources of light if the Full Moon is not visible.

Another facet of this issue is of the effects ofgas flaring on the behavior of migrating eiders,which we were able to examine in September2002. Surprisingly, we saw no effect of that gasflaring on the proportion of non-directionalbehavior of “eiders,” even though the Full Moonwas not visible at the time and the gas flaringresulted in the attraction of large number of otherbirds to the island (see below). Hence, gas flaringmay not be a source of concern for eiders, althoughit would be good to observe additional gas-flaringevents when these birds are moving to be certain ofthat conclusion.

Little research has been done on the effects ofthe moon on attraction to lights and subsequentcollision-related mortality, other than in the contextthat most large kills of migrating birds in easternNorth America in the fall occur during cloudy,

Discussion


inclement weather that blocks the ability of birds tosee the moon and lowers flight altitudes. Verheijen(1980, 1981a, 1981b) suggested a lunar-drivenperiodicity to bird kills at tall structures, with theoverall mortality pattern centered only on the NewMoon and essentially no mortality occurringaround the Full Moon. For example, such a patternhas been seen for nocturnal procellariiformseabirds in several locations (Telfer et al. 1987,Day and Cooper 1995 and references therein). Incontrast to Verheijen’s conclusions, however,Crawford (1981a) used a stronger data set andfound a bimodal pattern centered on both the NewMoon and the Full Moon. Qualitative studies alsosuggest that the highest light-attraction andcollision rates of migrating passerines at theWashington Monument occur when the moon isnot visible, either being obscured or below thehorizon (Overing 1936, 1938). Further, migratingbirds are most strongly attracted to light caused bygas flaring at offshore oil platforms in the NorthSea during cloudy and hazy nights (Sage 1979,Hope Jones 1980, Wallis 1981, Wiese et al. 2001),leaving as soon as dawn arrives (Alerstam 1990),also suggesting that lunar visibility is important tomany night-migrating birds and that its absencemay result in attraction to other large sources oflight. Hence, for at least some bird species, theabsence of moonlight during migration may resultin attraction to other sources of bright light.

ISLAND-PASSING SUCCESSEssentially all “eiders” successfully passed

the island, and none of the factors significantlyaffected passing success. Although we found thatlights had no effect, probably because of theextremely high rate of success, the very high rateof passing success itself suggests that statisticalvalidation of this point probably is not evennecessary.

ISLAND-PASSING DISTANCE“Eiders” passed the island at the substantial

mean distance of ~1,400 m with ice present,~1,575 m with ice absent, and ~1,430 m overall.The only factors that significantly affected passingdistance were period and wind direction, in that“eiders” passed the island at a greater distancewhen ice was present and at a greater distanceduring all wind directions other than calm winds.

Hence, these results suggest that the birds gave theisland a wider berth when ice was absent or whenthe wind was blowing.

The few targets that passed over the island didso in a variety of environmental conditions. All didso during good visibility, suggesting that poorvisibility was not causing the flyovers. The secondfactor of particular interest is wind direction, inthat all of the flyovers occurred during tailwinds.Hence, tailwinds may cause enough difficulties inmaneuvering that some birds simply are pushedover the island. Finally, all flyovers occurred withthe lights off at night.

Little research has been conducted on thedistance at which migrating birds approachstructures under various visibility and windconditions. The primary attraction occurs when thesky is overcast (Crawford 1981b), usually withlow, heavy fog and bright lights (Kramer 1948);however, these conditions are not the same as whenvisibility is poor. A study of the behavior of divingducks around offshore wind turbines in theNetherlands, however, suggested that birds gavethe windfield a wider berth during moonless nightsthan during nights when moonlight made thewindfield easier to see (Dirksen et al. 1997),suggesting that a poor ability to see at night clearlyincreases behavioral caution in at least some divingducks. (On the other hand, these were resident,rather than migratory, birds that knew where thewindfield was, so their avoidance response mayhave been stronger than that exhibited by migratingbirds first encountering such structures.) Thus,although we know of little applicable informationin the literature, it appears that migrating eiderspass Northstar Island at a greater distance duringperiods of ice and wind conditions when collisionrisk would be highest.

SPATIAL DISTRIBUTION“Eiders” exhibited nonrandom movement

through the study area, both with the anti-collisionlights off and with them on. The greatestconcentration passed on the seaward (northern)side of the island, with a smaller number movinglandward (south) of the island. Such a consistentpattern of spatial use also is seen at Barrow, whereeiders migrating from across the Beaufort Seaattempt to cross Barrow Spit and enter the ChukchiSea only in a narrow zone at its base (Thompson

Discussion


and Person 1963; Suydam et al. 1997, 2000a; Dayet al. 1998, 2001, 2004b).

The Before–After analysis for “eiders”suggested that turning on the anti-collision lightsresulted in net decreases in movement density nearthe island and net increases in movement densityfar from the island, especially toward the northeastof the island; the regression line crossed aBefore–After difference of zero at a distance of~1,550 m from the island, suggesting that some“eiders” avoided the island out to that distance.This avoidance effect, however, was seen onlywhen ice was present. We caution, however, thatthe R² for this regression was only 0.05, suggestingthat there was much additional variation that wasunexplained by turning on the anti-collision lights;however, given the effects of several of theenvironmental factors on movements that we haveseen above, we are not surprised that theexplanatory power was so low. However, thenature of this analysis makes it impossible toincorporate those other factors into this analysis.

HIGH-RESOLUTION VARIATIONS IN SPATIAL DISTRIBUTION

“Eiders” clearly exhibited responses to theisland, and those responses generally increased asthey approached the island. For example, theyexhibited significantly greater number of vertices(course changes)/km of line near the island than farfrom it, they exhibited significantly greater angularchanges near the island than far from it, and, forthose that would have passed closest to the island,they exhibited course changes that resulted in a netincrease in passing distance. All of these behaviorsare concordant with an interpretation that thesebirds are aware the island is there and respond toavoid colliding with it, with the responsesincreasing with increasing proximity.

Although such a pattern of response andnatural anti-collision avoidance behavior wasapparent, a pronounced and consistent response tothe anti-collision lighting system was lacking. Thelights had no effect on the distance at which coursechanges occurred and on mean angular changeswhen course changes actually occurred. The lightshad an inconsistent effect on the vertex distance,being significantly larger when the lights were onin good visibility but larger when the lights wereoff in poor visibility; however, it was encouraging

that the mean vertex distances were significantlylarger during tailwinds and that the net change inpassing distance indicated that birds moved awayfrom the island during tailwinds, which is when thebirds are traveling the fastest over the ground (i.e.,groundspeed is highest) and, hence, have less timeto avoid collision. There was a shift of birds awayfrom the island when the lights were on, but aneffect was not seen at the closest distances. Thereclearly was natural avoidance of the island,however.

VISUAL FLIGHT CHARACTERISTICSVisual data on eiders at Northstar Island are

limited and were presented here in a qualitativefashion. These birds flew at a mean altitude of~6 m agl/asl, with a suggestion that they flew ~1 mhigher with the lights off than with them on and athigher altitudes with tailwinds. They exhibited novariation in general flight behavior (essentially allflew in a straight-line manner), island-passingsuccess (nearly all passed successfully), orisland-passing behavior (only ~10% exhibited adetectable change in flight direction and/oraltitude), regardless of visibility category andanti-collision lighting setting. Finally,island-passing success and island-passingbehavioral responses were a little, but notdramatically, higher near the island. Hence, eidersflew at altitudes low enough to collide withNorthstar Island and its structures but exhibitedfew visually detectable responses to it. We caution,however, that these birds may have adjusted flightdirections or other behaviors far from the island,resulting in responses that we could not detectvisually.

The only information available fromelsewhere is on flight-altitude data from Barrow(Day et al. 2004b). There, the mean flight altitudewas ~12 m, or higher than at Northstar Island;however, birds there were approaching the base ofBarrow Spit, where they are hunted, so they mayhave been rising in elevation either to cross land orto avoid hunters. Birds at Barrow flew higher withtailwinds and crosswinds (Day et al. 2004b),similar to what we saw here.

DOWNING AND MORTALITYThe limited data so far on downing of eiders

at Northstar Island and Endicott Island in

Discussion


2001–2004 suggest a strong lunar component and afairly strong barometric component but no effect ofwind speed or wind direction. The lunarcomponent indicated a strong tendency fordowning during a Full Moon (as we have definedit) and, especially, during a waxing moon; bothtendencies were strong, both for nights on whicheiders were downed and for individual birds. Thebarometric component suggests a strong tendencyfor downing during a weakly changing barometer.The limited data so far suggest that fog was presentduring the mortality events for which we have data(see below).

Although Verheijen (1980, 1981a, 1981b)concluded that most birds are killed during nightsnear a New Moon and very few are killed near aFull Moon, Crawford (1981) used a systematicallycollected data set to show that there are twoclusters of mortality, one near the New Moon and asecond one near the Full Moon. Thus, our datasuggest a downing pattern for eiders closer toCrawford’s than to Verheijen’s and are moststrongly associated with a Full Moon.

Why might eiders become downed atNorthstar Island and Endicott Island primarily neara Full Moon and a waxing moon? The previouslypresented data on flight behavior of eiders indicatea moon phase*moon visibility interaction thatshows a higher frequency of non-directional flightbehavior around a Full Moon if it is not visible thanif it is visible. Hence, we suggest that a componentof the mortality is related to visibility during a FullMoon, especially in October. Based on acomparison of air temperature and the dew point in2001–2003 at Northstar Island, there is a largeincrease in the probable occurrence of fog fromAugust (~18% of the days) to September (~23%)and October (~92%); once the ocean freezescompletely, however (probably sometime inOctober or early November), the occurrence of fogprobably decreases quickly. Hence, the higherprobability of fog, coupled with the occurrence of aFull Moon, may help explain the observed lunarperiodicity of eider downings so far.

A waxing moon also is strongly involved inthe mortality of eiders. Examination of the timingof moonrise at Prudhoe Bay during the four fallperiods suggest that, in general, a waxing FullMoon may not be visible for several days(especially in the evening), whereas a waning Full

Moon always is up. An absence of a visible FullMoon, especially early in the evening, probablycauses as much confusion in eiders as a Full Moonthat is up but obscured by clouds or fog. Thispatterns is not an exact fit, however, for a waxingFull Moon sometimes is visible during theafternoon and evening in some months and years.

The barometric pattern is more difficult toexplain than is the lunar one. The evidencesuggests a strong association with a weaklychanging barometer. Because a rapidly changingbarometer would be associated with the passage ofstrong weather fronts with which many bird killsare associated, it appears that most eider mortalityis not associated with the passage of fronts orstorms.

“NON-EIDERS”

Over the two years, we collected data on1,485 radar targets that represented a diverse groupof species, including loons, other waterfowl,shorebirds, gulls, and alcids. With a mean flocksize of 15.0 birds/flock for visually identified birdflocks, we collected data on ~22,300 birds, andprobably more.

MOVEMENT RATES“Non-eiders” exhibited a highly pulsed

pattern of migration at Northstar Island. Movementrates were low–moderate in late August–earlySeptember 2001 and in 2002–2004, whereas theywere moderate–high during early tomid-September 2001. Movement rates varieddramatically among nights, indicating pulses ofmigratory movement. All of these characteristicswere highly similar to those seen for “eiders” andwere corroborated by the significant correlation innightly movement rates between the two groups in2001, 2003, 2004, and overall.

Movement rates of “non-eiders” weresignificantly affected by period, time of day,precipitation, wind direction, and winddirection*wind speed. Movement rates werehighest when ice was present, were higher duringperiods without precipitation, were higher duringcalm winds and headwinds, and were higher forweak crosswinds and tailwinds than for strongones. In general, the migration volume of birds inboth spring and fall is highest with tailwinds andlowest with headwinds and decreases during

Discussion


periods of low visibility and fog (Richardson1978). Further, because of birds’ preference fortailwinds, migration is pulsed because easterlymoving weather patterns of alternating high- andlow-pressure systems result in alternating periodsof favorable and unfavorable winds (Richardson1978). Hence, the results from this study are atodds with a majority of the scientific literature onbird migration and wind direction. This study,however, examined a variety of species, some ofwhich were migrating eastward (i.e., into whattheoretically were tailwinds), rather than westward,in substantial numbers, possibly explaining theunusual results. The low directionality alsoreflected the fact that 24% of these flocks wereheading into what theoretically were tailwinds.

FLIGHT VELOCITY“Non-eiders” exhibited flight velocities that

averaged ~28 mi/h, or only ~58% of thoseexhibited by “eiders;” however, there was someoverlap between the two groups in velocities attimes, especially for loons and Long-tailed Ducks.Velocities were higher when ice was present,highest with tailwinds and crosswinds, higher witha strong tailwind than a weak headwind or a weaktailwind, higher with a strong headwind than aweak headwind, higher with the lights on than offduring the day, and significantly higher with lightson than off during calm winds and tailwinds. Noliterature is available on the effects of period ortime of day on velocities, but at least part of thedifference in this study probably was caused byvariation in species-composition.

FLIGHT DIRECTION“Non-eiders” exhibited low overall flight

directionality, making it impossible for us toconduct statistical tests, so all patterns arequalitative at best. These birds exhibited a bimodalpattern of flight to the northwest and southeast,generally similar to that of “eiders;” however,overall directionality was weak, and birds wererecorded flying nearly every direction. The meanvector was toward the west during the daytime, noprecipitation, and good visibility, but toward theeast at night and during good visibility and towardthe southeast during precipitation. The mean vectorwas toward the northwest during tailwinds, towardthe southwest during crosswinds and headwinds,

and toward the south during calm conditions.Finally, they were headed toward the west duringweak winds and toward the southwest duringstrong winds and were headed toward thewest–northwest with the lights off andwest–southwest with the lights on. These patterns,however, all are qualitative and should beconsidered with caution because, in ourexperience, mean flight-direction vectors may bemisleading if there is not a strong directionalitycomponent in the data. Some of these differencesmay have been caused by different species-groupsmoving during different weather conditions (e.g.,see Flock 1973: 269) or possibly by some speciesmigrating from Siberia to Alaska and theneastward along the coast before turning southwardover North America (e.g., Alerstam andGudmundsson 1999a, 1999b).

FLIGHT BEHAVIORIn contrast to the pattern seen for “eiders,”

only ~76% of all “non-eiders” exhibitedstraight-line (directional) flight behavior; the other23% of targets exhibited non-directional behavior,consisting of 11% erratic behavior and 13%circling behavior. The proportion ofnon-directional flight behavior was higher whenice was present, higher at night, higher with goodvisibility, higher with tailwinds, higher with strongtailwinds, higher with the lights off than on whenvisibility was poor, and higher with the lights offthan on in crosswinds and tailwinds. In otherwords, these birds responded to essentially everyenvironmental factor that was examined. Some ofthese patterns, however, suggest that “non-eiders”became strongly attracted to the island and circledit or flew erratically around it at night and in somewind conditions with the lights off. Surprisingly,the proportion of attraction and circling behavioralso was higher during good visibility than poorvisibility; we are unable to explain this result in thecontext of migratory behavior in birds, and it maybe a statistical artifact.

In addition to the main effects, “non-eiders”also exhibited a lunar effect on flight behavior,similar to that seen in “eiders.” They exhibited asignificantly higher proportion of non-directionalbehavior when the moon was full and when it wasnot visible. These results are similar to those for“eiders” and suggest that many of these non-eider

Discussion


species may use the Full Moon for help inorientation during migration, resulting in confusioncause by and/or attraction to large sources of lightsif the Full Moon is not visible.

The gas-flaring event of 20 September 2002indicated that gas flaring may cause attraction of“non-eiders” to the island: none of the targetscircled the island when it was not flaring, whereas~48% circled the island when the gas flaring wasoccurring. Long-tailed Ducks and Glaucous Gullsappeared to be the species most affected, althoughit is possible that others not migrating in numberson that night also may be affected by such events.Because the gas-flaring event caused an increase inthe amount of light emitted by the island, we inferthat it was the light that was the attractant for thesebirds; however, the level of ambient sound alsoincreased dramatically that night, so we are unableto separate conclusively the effects of the increasedlight from the effects of the increased soundassociated with this event. During this attraction,some flocks of Long-tailed Ducks passed so closeto colliding with buildings that they had to takeextreme action in the form of what is calledbehavioral flaring (i.e., extreme changes in flightdirection and flight altitude) to avoid collision.

ISLAND-PASSING SUCCESS“Non-eiders” had a high rate of passing

success, but it was slightly lower than that seen for“eiders.” Only one factor, period, was found toaffect passing success, probably because overallpassing success was so high that one would notexpect many of the factors to have a significanteffect.

ISLAND-PASSING DISTANCE“Non-eiders” exhibited great variation in

passing distances, being ~1,040 m when ice waspresent, ~1,270 m when ice was absent, and~1,130 m overall; hence, passing distances wereconsiderably smaller than those for “eiders” as awhole. To some extent, this smaller mean distanceis related to the more extensive attraction to theisland and subsequent circling behavior exhibitedby several of these non-eider species. Passingdistances were greater when ice was absent, greaterduring poor visibility, higher with winds other thanheadwinds, greater with strong headwinds, andgreater with the lights off than on in crosswinds.

The greater distances during poor visibility aregood, in that it means that these birds are giving theisland a wider berth during conditions that makethe island difficult to see. The greater distanceswith the lights off during crosswinds is difficult tointerpret, however.

Targets that passed over the island did so in avariety of environmental conditions. All did soduring good visibility, suggesting that poorvisibility was not causing the flyovers andreflecting the fact that these birds passed the islandat smaller average distances when visibility wasgood than when it was poor. The second factor ofparticular interest is wind direction, in that a higherpercentage of the flyovers occurred duringtailwinds (~48%) than occurred across all sessions(~40%). Hence, tailwinds may cause enoughdifficulties in maneuvering that some birds simplyare pushed over the island. Finally, essentiallyequal percentages of flyovers occurred with thelights off and on.

SPATIAL DISTRIBUTION“Non-eiders” exhibited nonrandom

movements, both with the anti-collision lights offand with them on. The extensive attraction tooverall lights on the island and subsequent circlingbehavior are clearly visible on the plots for whenice was present but are less evident for when icewas absent. In addition, there was a zone of higheruse passing toward the northwest on the seawardside of the island, similar to, but weaker than, thatseen for eiders. There was a clear pattern ofattraction and concentration around the island thatwas driven by several species-groups, especiallyducks, shorebirds, and gulls. The regressionanalyses indicated attraction caused by turning onthe anti-collision lights during both periods,although the effect was strongest when ice waspresent and non-significant when ice was absent.

VISUAL FLIGHT CHARACTERISTICSVisual data on these other species-groups

were presented here in a qualitative fashion. Meanflight altitudes varied among species-groups, from~7 m for other ducks to ~19 m for and gulls andaveraged ~11 m across all visually identified flocksof non-eider species; gulls and shorebirds had highmean altitudes, so these birds mostly were lookingdown onto the island’s lights. For all

Discussion


species-groups, proportions of non-directional(erratic or circling) general flight behavior did notdiffer between anti-collision lighting setting.Island-passing success was high in mostspecies-groups, but success of some groups wasslightly lower when the anti-collision lights wereoff than on. Island-passing behavior varieddramatically among species-groups, with loonsexhibiting little response but all other groupsexhibiting moderate to frequent changes in flightdirection, flight altitude, or both; behavioral flaring(extreme anti-collision behavior) was evenrecorded in a few instances when Long-tailedDucks nearly hit the electrical buildings at thenortheastern corner of the island during agas-flaring event and when shorebirds werecircling extensively. All species-groups tended toexhibit higher passing success and higher responsefrequencies when near the island. Hence, all ofthese species flew at altitudes low enough for themto be able to collide with Northstar Island orstructures on it, and most exhibited dramaticbehavioral responses to the island, but many of theresponses did not suggest major problems causedby the anti-collision lights or by proximity to theisland.

Data on flight altitudes of non-eider species innorthern Alaska are limited, but some areavailable. At Barrow, flight altitudes were highlyvariable among species-groups (Day et al., unpubl.data). Loons averaged 58 m agl (n = 71 flocks),other ducks averaged 19 m agl (n = 15 flocks),shorebirds averaged 19 m agl (n = 19 flocks), andgulls averaged 20 m agl (n = 14 flocks). Along theBeaufort Sea coast, migrating Long-tailed Ducksprimarily flew low over the water, with 46% of allbirds flying 2 m agl and the rest being seen ataltitudes up to 500 m agl (Johnson and Richardson1982).

DOWNING AND MORTALITYThe limited data set so far on downing of

Long-tailed Ducks at Northstar Island in2001–2004 suggests strong lunar and barometriccomponents, whereas wind speed and direction hadno effect on downing. The lunar componentindicated a strong tendency for downing during awaxing moon (both for nights and individuals assampling units); in contrast, lunar phases occurredwith a frequency similar to those for the four years

combined. The barometric component indicated astrong affinity for a rising barometer (for bothnights and individuals) and no effect of the rate ofbarometric change. For those nights for which wehave data, fog was probable on all of the nights onwhich birds were downed.

Although Verheijen (1980, 1981a, 1981b)concluded that most birds collide with structuresduring nights near a New Moon and Crawford(1981) concluded that there were clusters ofmortality near both the New Moon and the FullMoon, our data suggest a pattern for Long-tailedDucks similar to neither lunar pattern. In effect,these ducks were downed with lunar phases similarto the proportions in which they occurred duringthe four falls. On the other hand, neither of theseauthors examined the association between downingand lunar trend. We suspect that the associationwith a waxing moon may be similar to thatdiscussed for eiders.

The association between barometric patternsand the downing of Long-tailed Ducks isconfusing: there was a strong association with arising barometer but no relationship with the rate ofbarometric change. A rising barometer generallyreflects improving weather conditions andnortherly winds and often is accompanied byhigher migration volume in the fall (Richardson1978); however, such patterns are important forbirds that are migrating in a north–south direction,rather than the east–west one found in the BeaufortSea. Because a rising barometer generally signalsbetter weather (i.e., fewer, rather than more, birdsshould be killed), perhaps a rising barometer in theBeaufort Sea in fall is accompanied by cooler airtemperatures that, in turn, would increase thechances of fog over the ocean. This is the onlypossible scenario we can develop for such anassociation with higher downing rates inLong-tailed Ducks at this time.

EFFECTS OF LIGHTS

“EIDERS”The anti-collision lights at Northstar Island

caused “eiders” to respond in some ways thatsuggested that they were enhancing collisionavoidance (Table 29). These anti-collision lightssignificantly reduced velocity at night, resulted in aspatial redistribution of the “eiders” away from the

Discussion


Table 29. Summary of effects of anti-collision lights on migratory movements and behavior of “eiders” and “non-eiders,” 2001–2004. For those analyses that showed significant effects of lights on a response variable, the nature of that response is described; ns = not significant statistically.

Species-group Response variable "Eiders" "Non-eiders"

Movement rates ns ns

Velocity higher in daytime when lights on; higher at night when lights off

higher in daytime when lights on; higher when lights on in calm winds

and tailwinds

Flight direction ns ns

Proportion non-directional flight behavior

ns higher with lights off when visibility poor; higher when lights off in

crosswinds and tailwinds

Island-passing success ns ns

Island-passing distance ns higher when lights on in crosswinds

Spatial changes net spatial avoidance out to ~1,550 m when ice present; ns when

ice absent

ns (weak attraction when lights on)

Course changes by actual

distance categories (vertices/km)

ns –

Course changes (vertices/km)

by near vs. far original distance categories

ns –

Vertex distance inconsistent effect—larger when

lights on when visibility good but larger when lights off when visibility poor; larger when lights off during

tailwinds

–

Angular change ns –

Original vs. actual passing

distance shift of some birds farther offshore

when lights on, but no effect at closest distances

–

Net change in passing

distance movement away from island when

lights on in tailwinds –

Discussion


island, were involved with a shift farther offshorein two distance zones fairly near the island, andresulted in a new movement away from the islandduring tailwinds. There were significantly morevertices (course changes)/km near the island thanfar from it when the lights were on, but the meannumber of vertices/km did not differ significantlybetween lighting settings in the zone nearest theisland, and the lights had no effect on original vs.actual passing distance in the zone closest to theisland. Hence, the natural avoidance response ofthese birds was enhanced only intermittently. Onthe other hand, there was no evidence that thelights were disrupting any aspect of migration.

The clearest effect of the anti-collision lightswas on spatial distribution, with a positive slope tothe regression involving net changes in movementdensity after the lights were turned on, indicating ashift in the spatial distribution of birds away fromthe island (net avoidance out to ~1,550 m) whenthe anti-collision lights were on; however, thiseffect was seen only when ice was absent. Giventhe lack of effect of the anti-collision lights onoverall flight directions, this change may occur at asubstantial distance from the island (i.e., off of theradar screen); there also were more coursechanges/km as birds approached the island, so theymay represent part of the subtle shift that wasoccurring. The slope of the regression line(0.0004), however, was small, resulting in anincrease of only 0.4 “eider” targets/km²/h for every1,000 m of increasing distance from the island.Further, the anti-collision lights explained only 4%of this change in spatial distribution, suggestingthat other factors have a much larger effect onspatial distribution than turning on theanti-collision lights did. (On the other hand, as wehave already shown, several other environmentalfactors cause variability in migratory attributes, soit is not surprising that the R² is so small; however,we were unable to incorporate all of the otherfactors in this type of analysis.) Hence, theanti-collision lights caused avoidance of the island,but the spatial repulsion that they caused averagedonly 56 m. It is possible that the avoidanceresponse was nonlinear (i.e., strong avoidance nearthe island but no response at some distance awayfrom it), but no such response was visible on theregression plot. It is possible, however, that acurvilinear response occurred at distances farther

than we were able to sample or that it was maskedby the substantial among-cell variation. Webelieve, however, that this shift away from theisland is the same as that seen for birds thatoriginally were going to pass closer to the islandbut clearly shifted farther away instead (Fig. 24).

The lights*time of day interaction for velocityindicated that mean velocities at night weresignificantly lower when the anti-collision lightswere on than when they were off; this behavior isgood, for it suggests that “eiders” slow down atnight when the anti-collision lights are on, therebyreducing their chances for collision by giving themslightly more time to react and respond bychanging course. It is possible that “eiders” maybecome confused by these lights during thedaytime or that the higher power of the daytimelights (20,000 candela versus 2,000 candela atnight) may cause some of this difference inbehavior. Alternatively, this inconsistent responsemay be a statistical artifact that resulted from an αof 0.05 (i.e., errors introduced 5% of the time withthis α level).

Another facet of collision avoidance is evidentfrom the high-resolution spatial analyses. “Eiders”clearly exhibited increased awareness and behaviorthat suggests natural collision-avoidance behavioras they approached the island. They exhibitedgreater numbers of course changes as theyapproached the island and a net movement awayfrom the island after course changes for birds thatwould have passed in the innermost zone; however,lights had an inconsistent effect on these behaviors.Although these results suggest pronouncedawareness and avoidance of the island, theanti-collision lights clearly did not cause a strongand consistent increase in this avoidance behavior.There was a slight increase in the innermost zone,but the increase was significant in only one case.Hence, these birds were responding to the islandnaturally, but the anti-collision lights caused aresponse that appeared to be intermittent, ratherthan consistent. We believe that the combination ofthe light buildings and the anti-collision lightsclearly are making the birds aware of the island,and they are responding to it to avoid collisions inmost cases. Most encouragingly, we saw noindication that the lights caused confusion inmigrating birds or disrupted migration by being tooeffective.

Discussion


“NON-EIDER” SPECIESThe anti-collision lights at Northstar Island

caused “non-eiders” to respond in few significantand meaningful ways that we were able to detect(Table 29), and some of the significant patterns areinteresting but provide few insights into collisionavoidance; they also may simply be spuriousstatistical artifacts. The lights caused velocities todrop in calm winds and tailwinds, they decreasedthe proportion of non-directional behavior duringpoor visibility and during crosswinds and tailwinds(when birds might hit the island more easily), andthey increased the island-passing distance incrosswinds (again when birds might hit the islandmore easily). Otherwise, the lights had no effectson migrating “non-eiders.” The lights*time of dayeffect on velocity indicated that these birds flewfaster during the daytime than at night with thelights were on—but there was no difference atnight between lighting settings, indicating that theyhad no significant effect on slowing velocities atnight. In addition, the spatial analysis suggestedthat “non-eiders” were attracted to the island inboth periods, although the attraction was notstatistically significant. We did not do ahigh-resolution analysis of behavior of flightlines,similar to that for “eiders,” so we cannot commenton the effects of lights on that topic.

CONCLUSIONS AND RECOMMENDATIONSOverall, the anti-collision lighting system

generally resulted in some significant and someweak or intermittent responses by eiders that couldaid in collision avoidance. The spatial analyses andsome of the high-resolution analyses showed a netmovement away from the island—the bestevidence for response caused by the anti-collisionlights. On the other hand, these analyses wereunable to incorporate other environmental factorsthat clearly elicited a much stronger effect oneiders than the lights do, so it is possible that thisapparent response is an artifact of the simpleanalyses. These lights also caused migrating eidersto slow down significantly at night, which ishelpful in decreasing the probability of collision byincreasing the length of time in which birds canmaneuver to avoid collision. The other responsestended to be either inconsistent or ones that haveno known effect on collision avoidance that we candetermine (e.g., the interactions involving lights

and some wind directions). The high-resolutionanalyses showed a natural anti-collision responseby eiders to the island, but that response wasincreased only intermittently by the anti-collisionlights. Thus, we conclude that these anti-collisionlights caused avoidance of Northstar Island byeiders but that the amount of response was notdramatic; nevertheless, the lights did cause birds toknow that the island was there and, to some extent,enhanced their ability to respond naturally to theisland. Further, these lights did not disruptmigration in any measurable way, which isbeneficial. In contrast, these lights clearly did notcause non-eider species to avoid the island andactually appeared to cause attraction at times.

We recommend the continued use of thislighting system at night during the fall migration ofeiders. It may be possible to modify the existinglighting system to enhance the avoidance responseof eiders; considerations for possible modificationof this existing system include flashing rate,flashing synchrony, and light intensity. AlthoughManville (in press) recommended that exteriorlights be white rather than red, thatrecommendation was based on studies ofattractiveness of colored vs. white lights topasserines, rather than to waterfowl or, morespecifically, eiders; there also are criticisms aboutthe original study design. In addition, availableinformation from experts on avian vision suggeststhat avian eyes are not good at discerning differentwavelengths at night (Appendix 9), so changingwavelength may not be particularly effective ataffecting these birds’ behavior. We believe,however, that, if additional structures are built inthe Beaufort Sea, and if a light-based bird-deterrentsystem is installed, research on captive animalsshould be undertaken to maximize the deterrentsystem’s effectiveness.

Another issue that should be raised is the issueof gas flaring. The data from 2002 suggest thatlarge gas-flaring events should be avoided (whenand if possible) at night and during periods oflimited visibility, to minimize the attraction ofbirds to the island. This attraction to the islandduring large gas-flaring events was greatest fornon-eider species than for eider species, but we donot have enough data to conclude thatflaring-caused attraction is not a problem formigrating eiders.

Discussion


One final issue that should be consideredinvolves the possible effects of sea ice on eidermovements near Northstar Island. In 2001, sea icewas present near the island at all times, and werecorded 690 “eider” flocks flying near the island.In 2002–2004, however, the ice edge was 75–300mi offshore, and we recorded a total of only238 “eider” flocks flying near the island; hence, weadded only 34% more radar targets of “eiders”during these 3 years combined. During these threeyears, we also lost a substantial number ofsampling sessions because the large fetch createdby the open water allowed seas to build easily inthose three years, thereby decreasing the amount oftime we could sample. Hence, we believe that,

during periods of extensive open water, most eiderswere migrating farther offshore than NorthstarIsland lies, plus we were able to spend less timesampling those eiders that did pass near the islandbecause of heavy sea clutter’s effects on the radar.Unfortunately, the electronic removal of sea clutteron the radar also removes bird echoes, so seaclutter is a serious and (at this time)insurmountable obstacle to sampling. Hence, webelieve that retreating sea ice is moving the eidermigration corridor offshore from Northstar Island,which apparently was only at the inner edge of thecorridor in the 1970s, when sea ice wasconsiderably heavier than it is now (Fig. 26).

Figure 26. Eider migration corridor in the Alaska Beaufort Sea in the 1970s, as determined by Divoky (1984a).

NorthstarIsland

O I L F I E L D S

N O R T H S L O P E O F A L A S K A

Be a u f o r t S e a

Barrow

Kaktovik

155°W

155°W

150°W

150°W 145°W

145°W

70

°N

70°N

71

°N

71

°N7

2°N

25 0 25 50 75

Kilometers

4

ABR File: Divoky_Eider_Corridor_fig_05-153.mxd; 15 April 2005

Eider Migration Corridoras reported by Divoky (1984)

Literature Cited


LITERATURE CITED

Alerstam, T. 1990. Bird migration. CambridgeUniversity Press, Cambridge, UnitedKingdom. 420 pp.

Alerstam, T., C.-A. Bauer, and G. Roos. 1974.Spring migration of Eiders Somateriamollissima in southern Scandinavia. Ibis 116:194–210.

Alerstam, T., and G. A. Gudmundsson. 1999a. Birdorientation at high latitudes: flight routesbetween Siberia and North America across theArctic Ocean. Proceedings of the RoyalSociety of London (B) 266: 2499–2505.

Alerstam, T., and G. A. Gudmundsson. 1999b.Migration patterns of arctic birds:tracking-radar observations along theNortheast Passage. Arctic 52: 346–371.

Anderson, D. R., K. P. Burnham, and W. L.Thompson. 2000. Null hypothesis testing:problems, prevalence, and an alternative.Journal of Wildlife Management 64: 912–923.

Anderson, W. L. 1978. Waterfowl collisions withpower lines at a coal-fired power plant.Wildlife Society Bulletin 6: 77–83.

Anonymous. 2001. Birds choose to fly aroundturbines—Enron offshore study. WindpowerMonthly News Magazine (June 2001): 40.

Bailey, A. M. 1948. Birds of arctic Alaska.Colorado Museum of Natural History, PopularSeries, No. 8. 317 pp.

Beal, K. G., and H. J. Khamis. 1991. A problem instatistical analysis: simultaneous inference.Condor 93: 1023–1025.

Bergman, R. D., R. L. Howard, K. F. Abraham, andM. W. Weller. 1977. Water birds and theirwetland resources in relation to oildevelopment at Storkersen Point, Alaska. U.S.Fish and Wildlife Service, ResourcePublication No. 129. 38 pp.

BP (BP Exploration [Alaska] Inc.). 2003.Biodiversity Action Plan for BP operations onAlaska’s North Slope. BP Exploration(Alaska) Inc., Anchorage, AK. 26 pp.

Burnham, K. P., and D. R. Anderson. 1998. Modelselection and inference: a practicalinformation–theoretic approach. Springer-Verlag. New York, NY. 353 pp.

Cochran, W. W., and R. R. Graber. 1958. Attractionof nocturnal migrants by lights on a televisiontower. Wilson Bulletin 70: 378–380.

Cooper, B. A., and R. H. Day. 1998. Summerbehavior and mortality of Dark-rumpedPetrels and Newell’s Shearwaters at powerlines on Kauai. Hawaii. Colonial Waterbirds21: 11–19.

Cooper, B. A., and R. H. Day. 2003. Movement ofHawaiian Petrels to inland breeding sites onMaui Island, Hawaii. Waterbirds 26: 62–71.

Cooper, B. A., R. H. Day, R. J. Ritchie, and C. L.Cranor. 1991. An improved marine radarsystem for studies of bird migration. Journalof Field Ornithology 62: 367–377.

Cooper, B. A., M. G. Raphael, and D. E. Mack.2001. Radar-based monitoring of MarbledMurrelets. Condor 103: 219–229.

Cooper, B. A., and T. J. Mabee. 1990. Radarstudies of bird movements over power lines inthe Monte Vista National Wildlife Refuge,Colorado. Unpublished report prepared forEdison Electric Institute, Washington, DC,and the Wildlife Research Institute, Moscow,ID, by Alaska Biological Research, Inc.,Fairbanks, AK. 29 pp.

Cooper, B. A., and R. J. Ritchie. 1995. The altitudeof bird migration in east-central Alaska: aradar and visual study. Journal of FieldOrnithology 66: 590–607.

Crawford, R. L. 1981a. Bird kills at a lightedman-made structure: often on nights close to afull moon. American Birds 35: 913–914.

Crawford, R. L. 1981b. Weather, migration, andautumn bird kills at a North Florida TV tower.Wilson Bulletin 93: 189–195.

Day, R. H., and B. A. Cooper. 1995. Patterns ofmovement of Dark-rumped Petrels andNewell’s Shearwaters on Kauai. Condor 97:1011–1027.

Literature Cited


Day, R. H., B. A. Cooper, and R. J. Blaha. 2003a.Movement patterns of Hawaiian Petrels andNewell’s Shearwaters on the island ofHawai’i. Pacific Science 57: 147–160.

Day, R. H., B. A. Cooper, and T. C. Telfer. 2003b.Decline of Townsend’s (Newell’s)Shearwaters (Puffinus auricularis newelli) onKauai, Hawaii. Auk 120: 669–679.

Day, R. H., A. K. Prichard, and J. R. Rose. 2004a.Migration and collision avoidance of eidersand other birds at Northstar Island, Alaska,2001–2003. Unpublished report prepared forBP Exploration (Alaska) Inc., Anchorage,AK, by ABR, Inc.—Environmental Researchand Services, Fairbanks, AK. 156 pp.

Day, R. H., R. J. Ritchie, and J. R. Rose. 2000a.Avian studies along the proposed NorthernIntertie powerline project, spring 2000.Unpublished report prepared for GoldenValley Electric Association, Inc., Fairbanks,AK, by ABR, Inc., Fairbanks, AK. 72 pp.

Day, R. H., R. J. Ritchie, and A. M. Wildman.2000b. Avian studies along the proposedNorthern Intertie Powerline Project, fall 1999.Unpublished report prepared for GoldenValley Electric Association, Inc., Fairbanks,AK, by ABR, Inc., Fairbanks, AK. 100 pp.

Day, R. H., J. R. Rose, and B. A. Cooper. 1998.Evaluation of ornithological radar formonitoring eider migration at Point Barrow,Alaska. Unpublished report prepared for U.S.Fish and Wildlife Service, Fairbanks, AK, andNorth Slope Borough, Barrow, AK, by ABR,Inc., Fairbanks, AK. 42 pp.

Day, R. H., J. R. Rose, B. A. Cooper, and R. J.Blaha. 2001. Migration rates and flightbehavior of migrating eiders near towers atBarrow, Alaska. Unpublished report preparedfor U.S. Fish and Wildlife Service, Fairbanks,AK, and North Slope Borough, Barrow, AK,by ABR, Inc., Fairbanks, AK. 40 pp.

Day, R. H., J. R. Rose, A. K. Prichard, R. J. Blaha,and B. A. Cooper. 2004b. Environmentaleffects on the fall migration of eiders atBarrow, Alaska. Marine Ornithology 32:13–24.

Dick, M. H., and W. Donaldson. 1978. Fishingvessel endangered by Crested Aukletlandings. Condor 80: 235–236.

Dirksen, S., A. L. Spaans, and J. van der Winden.1997. Nocturnal collision risks of birds withwind turbines in tidal and semi-offshore areas.Pages 1–10 in G. Solari and C. Ratto (eds.).Proceedings of the International Workshop onWind Energy and Landscape. Balkema,Rotterdam, The Netherlands.

Divoky, G. J. 1984a. The pelagic and nearshorebirds of the Alaska Beaufort Sea. OuterContinental Shelf Environmental AssessmentProgram, Final Reports of PrincipalInvestigators 23: 397–513.

Divoky, G. J. 1984b. The pelagic and nearshorebirds of the Alaskan Beaufort Sea: biomassand trophics. Pages 417–437 in P. W. Barnes,D. M. Schell, and E. Reimnitz (eds.). TheAlaskan Beaufort Sea: ecosystems andenvironments. Academic Press, Orlando, FL.

Eastwood, E. 1967. Radar ornithology. Methuenand Co., Ltd., London, United Kingdom.278 pp.

Fischer, J. B., and W.W. Larned. 2004. Summerdistribution of marine birds in the westernBeaufort Sea. Arctic 57: 143–159.

Flock, W. L. 1973. Radar observations of birdmovements along the arctic coast of Alaska.Wilson Bulletin 85: 259–275.

Gavin, A. 1977. Wildlife of the North Slope: theislands offshore Prudhoe Bay/the Snow Geeseof Howe Island—the seventh year of study.Atlantic Richfield Company, Los Angeles,CA. 70 pp.

Hamer, T. E., B. A. Cooper, and C. J. Ralph. 1995.Use of radar to study the movements ofMarbled Murrelets at inland sites.Northwestern Naturalist 76: 73–78.

Hohenberger, C. J., W. C. Hanson, and E. E.Burroughs. 1994. Birds of the Prudhoe Bayregion, northern Alaska. Western Birds 25:73–103.

Hope Jones, P. 1980. The effect on birds of a NorthSea gas flare. British Birds 73: 547–555.

Literature Cited


Jehl, J. R., Jr. 1993. Observations on the fallmigration of Eared Grebes, based on evidencefrom a mass downing in Utah. Condor 95:470–473.

Johnson, L. L. 1971. The migration, harvest, andimportance of waterfowl at Barrow, Alaska.M.S. thesis, University of Alaska, Fairbanks,AK. 87 pp.

Johnson, S. R., and D. R. Herter. 1989. The birds ofthe Beaufort Sea. BP Exploration (Alaska)Inc., Anchorage, AK. 372 pp.

Johnson, S. R., and W. J. Richardson. 1982.Waterbird migration near the Yukon andAlaskan coast of the Beaufort Sea: II. Moultmigration of seaducks in summer. Arctic 35:291–301.

Kerlinger, P., and S. A. Gauthreaux, Jr. 1984.Flight behavior of Sharp-shinned Hawksduring migration: I. Over land. AnimalBehaviour 32: 1021–1028.

Kerlinger, P., and S. A. Gauthreaux, Jr. 1985.Flight behavior of raptors during springmigration in South Texas studied with radarand visual observations. Journal of FieldOrnithology 56: 394–402.

Kramer, Q. 1948. Bird tragedy in a fog. Cassinia37: 21–22.

Larned, W., R. Stehn, and R. Platte. 2005. Eiderbreeding population survey, Arctic CoastalPlain, Alaska, 2004. Unpublished reportprepared by U.S. Fish and Wildlife Service,Soldotna, AK, and Anchorage, AK. 48 pp.

Mallory, M. L., H. G. Gilchrist, S. E. Jamieson, G.J. Robertson, and D. G. Campbell. 2001.Unusual migration mortality of King Eiders incentral Baffin Island. Waterbirds 24: 453–456.

Manville, A. M., II. In press. Bird strikes andelectrocutions at power lines, communicationtowers, and wind turbines: state of the art andstate of the science—next steps towardmitigation. In Proceedings of the ThirdInternational Conference of Partners in Flight,Asilomar, CA. 24 manuscript pp.

Mosteller, F., and J. W. Tukey. 1977. Data analysisand regression: a second course in statistics.Addison-Wesley, Reading, MA.

Overing, R. 1936. The 1935 fall migration at theWashington Monument. Wilson Bulletin 48:222–224.

Overing, R. 1938. High mortality at theWashington Monument. Auk 55: 679.

Petersen, M. R., W. W. Larned, and D. C. Douglas.1999. At-sea distribution of SpectacledEiders: a 120-year-old mystery resolved. Auk116: 1009–1020.

Quakenbush, L. T., R. H. Day, B. A. Anderson, F.A. Pitelka, and B. J. McCaffery. 2002.Historical and present breeding seasondistribution of Steller’s Eider in Alaska.Western Birds 33: 99–120.

Quakenbush, L. T., and R. S. Suydam. 2004. Kingand Common Eider migrations past PointBarrow. University of Alaska Coastal MarineInstitute, Annual Report No. 10: 60–69.

Richardson, W. J. 1978. Timing and amount of birdmigration in relation to weather: a review.Oikos 30: 224–272.

Richardson, W. J. 1990. Timing of bird migrationin relation to weather: updated review. Pages78–101 in E. Gwinner, ed. Bird migration.Springer-Verlag, Berlin, Germany.

Richardson, W. J., and S. R. Johnson. 1981.Waterbird migration near the Yukon andAlaskan coast of the Beaufort Sea: I. Timing,routes, and numbers in spring. Arctic 34:108–121.

Sage, B. 1979. Flare up over North Sea birds. NewScientist 81: 464–466.

Schamel, D. 1977. Breeding of the Common Eider(Somateria mollissima) on the Beaufort Seacoast of Alaska. Condor 79: 478–485.

Schamel, D. 1978. Bird use of a Beaufort Seabarrier island in summer. CanadianField-Naturalist 92: 55–60.

Suydam, R. S., D. L. Dickson, J. B. Fadely, and L.T. Quakenbush. 2000a. Population declines ofKing and Common eiders of the Beaufort Sea.Condor 102: 219–222.

Literature Cited


Suydam, R. S., L. T. Quakenbush, D. L. Dickson,and T. Obritschkewitsch. 2000b. Migration ofKing, Somateria spectabilis, and Common, S.mollissima v-nigra, eiders past Point Barrow,Alaska, during spring and summer/fall 1996.Canadian Field-Naturalist 114: 444–452.

Suydam, R., L. Quakenbush, M. Johnson, J. C.George, and J. Young. 1997. Migration ofKing and Common eiders past Point Barrow.Pages 21–28 in D. L. Dickson, ed. King andCommon eiders of the western CanadianArctic. Canadian Wildlife Service, OccasionalPapers, No. 94.

Telfer, T. C., J. L. Sincock, G. V. Byrd, and J. R.Reed. 1987. Attraction of Hawaiian seabirdsto lights: conservation efforts and effects ofmoon phase. Wildlife Society Bulletin 15:406–413.

TERA (Troy Ecological Research Associates).1999. Spectacled Eiders in the Beaufort Sea:distribution and timing of use. Unpublishedreport prepared for BP Exploration (Alaska)Inc., Anchorage, AK, by Troy EcologicalResearch Associates, Anchorage, AK. 19 pp.

Thompson, D. Q., and R. A. Person. 1963. Theeider pass at Point Barrow, Alaska. Journal ofWildlife Management 27: 348–356.

Troy, D. 1985. Birds of Prudhoe Bay and vicinity.Sohio Alaska Petroleum Company,Anchorage, AK. 36 pp.

Troy, D. 2003. Molt migration of Spectacled Eidersin the Beaufort Sea region. Unpublishedreport prepared for BP Exploration (Alaska)Inc., Anchorage, AK, by Troy EcologicalResearch Associates, Anchorage, AK. 17 pp.

Tulp, I., H. Schekkerman, J. K. Larsen, J. van derWinden, R. J. W. van de Haterd, P. vanHorssen, S. Dirksen, and A. L. Spaans. 1999.Nocturnal flight activity of sea ducks near thewindfarm TunØ knob in the Kattegat.Unpublished report prepared for Novem,Utrecht, The Netherlands, by BureauWardenburg bv, Culemborg, The Netherlands.69 pp.

USFWS (U. S. Fish and Wildlife Service). 1996.Spectacled Eider recovery plan. U. S. Fish andWildlife Service, Anchorage, AK. 100 pp. +appendices.

USFWS (U. S. Fish and Wildlife Service). 2002.Steller’s Eider recovery plan. U. S. Fish andWildlife Service, Anchorage, AK. 27 pp.

Verheijen, F. J. 1980. The moon: a neglected factorin studies on collisions of nocturnal migrantbirds with tall lighted structures and withaircraft. Die Vogelwarte 30: 305–320.

Verheijen, F. J. 1981a. Bird kills at lightedman-made structures: not on nights close to aFull Moon. American Birds 35: 251–254.

Verheijen, F. J. 1981b. Bird kills at tall lightedstructures in the USA in the period 1935–1973 and kills at a Dutch lighthouse in theperiod 1924–1928 show similar lunarperiodicity. Ardea 69: 199–203.

Wallis, A. 1981. North Sea gas flares. British Birds74: 536–537.

Wiens, J. A., and K. R. Parker. 1995. Analyzing theeffects of accidental environmental impacts:approaches and assumptions. EcologicalApplications 5: 1069–1083.

Wiese, F. K., W. A. Montevecchi, G. K. Davoren, F.Huettmann, A. W. Diamond, and J. Linke.2001. Seabirds at risk around offshore oilplatforms in the North-west Atlantic. 2001.Marine Pollution Bulletin 42: 1285–1290.

Woodby, D. A., and G. J. Divoky. 1982. Springmigration of eiders and other waterbirds atPoint Barrow, Alaska. Arctic 35: 403–410.

Zar, J. H. 1984. Biostatistical analysis. Second ed.Prentice-Hall, Englewood Cliffs, NJ. 718 pp.

Appendices


Appendix. 1. Statistical review of the Draft Report and the response to that review.

Appendices


Appendix 1. Continued.

��

��

��

�

� �� !"�#��

�

� $�%�&�$��'�

�

()*��#)��+�)�

��,-.�/)�#�/0��"1�� 02*�

�023/)�4��

�

��

()*�,5�%6).3&�/7�#3��80�+�)"�#&�/7��"1�� "��!��#/�5/�/0�&��26)"/)&�"#�#�"#�2��)�+�� /7�#3��

)�./)#�!�2�6"��/7�#��#�#�/0"9�3/ �+�)��3��)��5�#3��%�#3/5"��05�,-�26#�+��6��)&�2�)�76��&��05�

)��5�"/��/7�#3��"6�#"*��,5��5�0#�7��5�# /��./)#�0#�.��2�"�7/)��.)/+��0#�#3�#� ��4)�� #3��#�

��"#��0�.)�02�.��:!6#� �#3�"/��2�+��#"*�

��)"#��3��"644�"#�5�#3�#��.��)�5�"��.��)�05/��!�/21�5�"�40� /6�5�!��/)��./ �)76��0��&"�"�

#3�0�#3��)�05/��;�5�!�/21�5�"�40�#3�#� ��6"�5*��<��!��+��#3�#�3��"�2/))�2#��#3/643�#3�)��)��"/��

2�+��#"*��=3��7�)"#�2�+��#��"�#3�#� �� /6�5�!��5�2)��"�04�#3��/+�)��"��.��"�;��5)��#�2��&��7� ��

!�/21�5�!&�3/6)��#� /6�5�5�2)��"��!�2�6"��/7�#3��3/6)��)�#3�)�#3�0�3��7�3/6)��!�/21�04��!�2�6"��

��0&�3��7�3/6)�.�)�/5"��)��0/#��#23�5��!�2�6"��0&�3/6)�&�!�/21"�5/�0/#�3�+��5�0#�2��

�0+�)/0��0#��2/05�#�/0"� �#3�0�#3��0#�)��3/6)��05�!�2�6"��0&�3/6)�&�!�/21"�5/�0/#�3�+��!�)5"��0�

!/#3�3��7�3/6)�"�""�/0"��./)#�0#�7/)�)�"./0"��+�)��!��"�#3�#��+��6�#��05�+�56��#�)4�#"�*��>�02��/6)�

�0�#��"��.��"�;��/7��"�""�/0"��05��5�)�#�)4�#"� /6�5�5)/.�#/�'��!�/21"�7/)��/+��0#�)�#�"�

�05�/0�&�$�!�/21"�7/)��5�)��05��'�!�/21"�7/)�0/0��5�)�.�""�04�5�"#�02�"�/)�+��/2�#��"*��=36"��#��"�

4/�04�#/�)�?6�)��4)��#��02)��"��0�./ �)�#/�2/�.�0"�#��7/)�#3�#��""�+��/""��0�"��.��"�;�*��6)#3�)��

#3�"��..)/�23� /6�5� /)1�!�"#�/0�&�7/)��/+��0#�)�#�"��7/)� 3�23��0��!"�02��/7�!�)5"��*�*��;�)/��"�

)��9�3/ �+�)�� "6".�2#�#3�#��43#"�3�+��##��77�2#�/0��/+��0#�)�#�"*��"�2/05�2�+��#��"�#3�#�,5�

"644�"#�5�#3�#� ��+�)�4��/04�#�)4�#"� �#3�0��"�""�/0�#/�4�#��0�)�"./0"��+�)��!��#3�0�6"��#3�#�

��0�06�!�)�7/)�#3�#�"�""�/0� �#3�0�#3�#��3/6)�!�/21*��<��3�+��)�"�)+�#�/0"��!/6#�6"�04��0"�7/)�

#�)4�#�".�2�7�2��##)�!6#�"��05��)��".�2��&�2/02�)0�5��!/6#�6"�04�#3��7/)�3�43�&�+�)��!��5�#��"623��"�

.�""�04�5�"#�02�*��!+�/6"�&��#3��226)�2&�/7�"623��0��"#��#�� #3�0��"�""�/0� /6�5�!��3�43�&�

�77�2#�5�!&�#3��/+�)��+�)��!��#&�/7�#3��)�"./0"��+�)��!��05�#3��"��.��"�;�� #3�0�#3�#�"�""�/0*��

>�02�� )46��4��0"#�6"�04�#3��.��)�5�"��.��)�05/��;�5�!�/21�5�"�40�7/)��0&��0��&"�"��#�#3�"�#��*�

,5@"�"�2/05�"644�"#�/0�/7�0/#��"�#3�#� ��2/0562#�./ �)��0��&"�"�!&��43#"�/77A��43#"�/0�#/�"��

3�#3�)� ��2#6��&�2�0�5�#�2#��5�77�)�02��2�6"�5�!&�#3��0#��2/��"�/0��43#"*��<��3�+��2/0562#�5�

%/0#��B�)�/�"��6��#�/0"�#3�#�6"��#3��2#6��C��43#"�/77C�.�""�04�5�"#�02��5�#�� #3�+�)�/6"�5�"#�02�"�

�55�5��*4*��"3�7#"�/7��$��7/)�$��#�)�#�/0"�/7��"��.��#�#�"#*��02�� )��D6"#�

5/�04�#�#�"#"��05�0/#��02�65�04�/#3�)�"/6)2�"�/7��0+�)/0��0#��+�)��#�/0��0�#3�"��0��&"�"��#3��)�"6�#�04�

./ �)��"#��#�"��)��1��&�605�)�"#��#�"*��<��"/�3�+��2��26��#�5�#3��.)/!�!��#&�/7�5�#�2#�04��".�#��

"3�7#�/7�2�)#��0��0�5�"#�02�"��$��#�5�77�)�0#�"��.��"�;�"��4�+�0�#3��+�)��!��#&� ��

�)��02/60#�)�04��0�#3��C��43#"�/77C�.�""�04�5�#�*��<��3�+��02�65�5�#3�"��07/)��#�/0��0�#3��)�./)#*��

>/ �+�)�� !��+��#3�#�#3��2��26��#�/0"�"3/6�5�!��5/0��/0�&�7/)�.�""�04�5�"#�02��"�02��#3�#��"��

+�)��!��#3�#� �"�/7�4)��#�!�/�/4�2��"�40�7�2�02��!�2�6"��/7�#3��+�)&�)��./""�!��#&�/7�2/��"�/0*��<��

"6".�2#�#3�#��7�#3�)��2#6��&��"��0��77�2#�/7��43#"��#��1��&�/226)"�/0�&�605�)��".�2�7�2�"6!"�#�/7�

��#3�)�2/05�#�/0"��1�04��#��/)��5�77�26�#�#/�2��26��#��./ �)��226)�#��&��05�#/�5�#�)��0�� 3�#�"��.��

"�;��"�0�2�""�)&�#/�5�#�2#��#)6��77�2#�/7�#3��0#��2/��"�/0��43#"*�

Appendices



��

��

��

�

� �� !"�#��

<��"/�3�+��5�"26""�5� �#3�,5�3�"�"644�"#�/0"��!/6#�6"�04�2/0#�06/6"�+�)��!��"�"623��"� �05�

�05�#�1�04��43#��"6)��0#"�#/�4�#�2/0#�06/6"�+�)��!��"*��)/��/7�/6)��0��&"�"�/7�)�5�)�5�#��05�

!�)5��/+��0#"�/+�)�#3��&��)"�� 3�+��7/605�#3�#� �� &"�4�#��623�!�##�)�)�"6�#"��7� ��!�/21�#3��

�05��0#/�2�#�4/)��"��"623��"�#3��A3�2�#�4/)��"� ��6"��)�#3�)�#3�0�6"�04��#��"��2/0#�06/6"�+�)��!��*��

<�#3�)�".�2#�#/�#3��43#��"6)��0#��""6�� 2�00/#�6"��43#��#�)�#/��"6)��!��0#��43#�

!�2�6"��43#"�/0�#3��"��05��1��"623��"6)��0#"��./""�!��#/�#�1�9��0��55�#�/0��/"#�/7�/6)�5�#��

�)��2/��2#�5� 3�0�#3�)��"��0��!��0#��43#�/7��6-*��>��4)��5� �#3�!/#3�/7�#3�"��./�0#"*�

>�02�� 4)��0�.)�02�.�� #3�#3�"��# /��0�./�0#"�/7�,5@"��#3/643� ��5/��.3�"�;��#3�#�

��3�+��)�"�)+�#�/0"��!/6#�#3��)�� /)�5��77�2#�/7�"623��!�/21�04�5�"�40*��/)�#3��)��"/0"��!/+��05�

�".�2��&�!�2�6"��/7�#3��4)��#��/""�/7�"��.��"�;��2�6"�5�!&�#3��!�/21�04�� )46��4��0"#�#3��)�+�"�5�

�0��&"�"�7/)��0&#3�04�/#3�)�#3�0��/+��0#�)�#�"��05�?6�"#�/0� 3�#3�)��0&#3�04� ��!��4��0�5�!&�5/�04�

�+�0�#3�#��0��&"�"*�

�

��02�)��&�&/6)"��

�

�

�/!�)#�>*�(�&��3*(*�

��0�/)��2��0#�"#�

Appendices


Appendix 2. Movement density (targets/km²/h) of “eiders” on ornithological radar near Northstar Island, northern Alaska, fall 2002, by anti-collision lighting setting.

Lights Off

4ABR file: Eiders_2002_05_153.apr; 21 April 2005

"Eiders"

Northstar

Island

Northstar

Island

500 0 500 1,000 1,500 2,000 Meters

2002


2/h)

0.0 – 0.5

0.5 – 1.0

1.0 – 1.5

1.5 – 2.0

2.0 – 3.0

3.0 – 4.0

> 4.0

Lights On

Appendices



Lights Off


"Eiders"

Northstar

Island

Northstar

Island

500 0 500 1,000 1,500 2,000 Meters

2003


2/h)

0.0

0.0 – 0.5

0.5 – 1.0

1.0 – 1.5

1.5 – 2.0

2.0 – 2.5

> 2.5

Lights On

Appendices



Lights Off


"Eiders"

Northstar

Island

Northstar

Island

500 0 500 1,000 1,500 2,000 Meters

2004


2/h)

0.0 – 0.5

0.5 – 1.0

1.0 – 1.5

1.5 – 2.0

2.0 – 2.5

2.5 – 3.0

> 3.0

Lights On

Appendices


Appendix 5. Movement density (targets/km²/h) of “non-eiders” on ornithological radar near Northstar Island, northern Alaska, fall 2002, by anti-collision lighting setting.

Lights Off

4ABR file: NonEiders_2002_05_153.apr; 21 April 2005

"Non-Eiders"

Northstar

Island

Northstar

Island

500 0 500 1,000 1,500 2,000 Meters

2002


2/h)

0.0 – 2.0

2.0 – 3.0

3.0 – 4.0

4.0 – 5.0

5.0 – 6.0

6.0 – 8.0

> 8.0

Lights On

Appendices



Lights Off


"Non-Eiders"

Northstar

Island

Northstar

Island

500 0 500 1,000 1,500 2,000 Meters

2003


2/h)

0.0 – 3.0

3.0 – 6.0

6.0 – 9.0

9.0 – 12.0

12.0 – 15.0

15.0 – 18.0

> 18.0

Lights On

Appendices



Lights Off


"Non-Eiders"

Northstar

Island

Northstar

Island

500 0 500 1,000 1,500 2,000 Meters

2004


2/h)

0.0

0.0 – 0.5

0.5 – 1.0

1.0 – 1.5

1.5 – 2.0

2.0 – 2.5

> 2.5

Lights On

Appendices


Appendix 8. Numbers of flocks and numbers of birds recorded during visual sampling near Northstar Island, northern Alaska, fall 2001–2004.

Species Number of flocks Number of birds

Unidentified goose 1 23 Tundra Swan 1 3 American Wigeon 2 3 Northern Pintail 3 10 Steller's Eider 2 2 King Eider 9 41 Common Eider 10 389 Unidentified eider 88 2,126 Long-tailed Duck 179 4,247 Unidentified duck 158 6,879 Unidentified waterfowl 29 900 Red-throated Loon 2 7 Pacific Loon 71 104 Yellow-billed Loon 1 1 Unidentified large loon 6 6 Unidentified loon 135 303 Pelagic Cormorant 2 2 American Golden-Plover 2 4 Semipalmated Sandpiper 2 5 Unidentified sandpiper 10 22 Unidentified phalarope 3 3 Unidentified shorebird 52 173 Parasitic Jaeger 2 2 Unidentified jaeger 7 9 Herring Gull 1 6 Glaucous Gull 196 364 Black-legged Kittiwake 35 190 Unidentified gull 20 78 Common Murre 1 2 Unidentified murre 3 3 Black Guillemot 9 16 Horned Puffin 1 1 Snowy Owl 1 1 Common Raven 4 4 Unidentified passerine 11 21 Unidentified bird 16 1,090 Total 1,075 17,040

Appendices


Appendix 9. Memorandum about avian vision at night.

MEMORANDUM

Date: 8 September 2003

To: Bill Streever, BP Exploration (Alaska), Inc., Anchorage

Ted Swem, USFWS, Fairbanks

From: Bob Day, ABR, Fairbanks

Re: Avian vision

During our last conference call (late July), I was tasked to track down information on

avian vision, with implications for light sensitivities of birds at night. I had Lauren Attanas of

our office conduct a search for information on avian vision and to compile a list of the major

references. I will send that bibliography if you are interested. Lauren also tracked down the

expert on Avian Vision from Yale University; his name is Dr. Timothy Goldsmith, and he just

retired. He was very helpful with understanding the issues. I enclose below a copy of the emails

between us. In essence, one cannot conduct the experiments to determine sensitivities of birds

without sacrificing the bird in the lab itself; hence, one cannot take carcasses and do the

experiments. He also indicated that there are now only a couple of labs in the entire world where

such work can be done, now that he has retired. So, that part of our interest seems to be a dead-

end.

Dr. Goldsmith also indicated that, since we are interested in birds' vision at night, the

questions become much simpler, in that cone (color) vision is irrelevant and rod (black/white)

vision is the dominant mode of vision at that time. In that case, wavelength of light probably is

irrelevant, whereas the simple presence/absence of light is most important.

Lauren also helped me track down Dr. Robert Beason of APHIS in Sandusky, OH. He

knows quite a bit about avian vision, too. He was able to send one additional paper citation that I

did not have.

I enclose a copy of the email correspondence below, in chronological order.

Appendices



At 01:06 PM 8/8/03 -0800, you wrote:

Dr. Goldsmith:

I am trying to locate information on avian vision, especially that of diving ducks and

procellariiform birds, for two projects on bird collision and collision avoidance that I am

involved with. Your name has been passed on to me as that of an expert on avian vision. I am

interested in learning about bird eyes and, in particular, what wavelengths of light that these

birds can and cannot see. If you have any publications or references that may be of use in this

line of research, I would appreciate your help.

Along these lines, you are the person who has been indicated to be able to examine eyes of a

particular species of bird and tell which wavelengths that bird can and cannot see. Is this true?

If it isn't true, do you know anyone who might be able to determine this sensitivity? I might be

able to secure funding for such testing on some carcasses, with the hope that there would be

something publishable from the testing.

Anyway, I apologize if you are not the avian-vision expert, but, if you are, I hope that we can

correspond further.

Thank you for your help.

Bob

=====================================================================

From: Timothy H. Goldsmith [mailto:[email protected]]

Sent: Saturday, August 09, 2003 10:28 AM

To: Bob Day

Subject: Re: hello, questions

Bob Day,

The short answer to your question is that birds have 4 cone pigments instead of the three that

humans possess. They see all wavelengths that we do, and in addition, their visible spectrum

extends into the near UV where ours does not. I know of no study of procellariiform vision per

se, but I have no reason to expect any differences from species that have been studied. (They are

reported to have better olfactory sense than most other birds, but this is presumably for detecting

up-wind sources of oily surface plankton from a distance.) The data that are available on the

cone pigments of ducks are unexceptional, but they are from dabblers.

Data on avian photopic sensitivity can be found in Goldsmith, T.H. and Butler, B.K. (2003) "The

roles of receptor noise and cone oil droplets in the photopic spectral sensitivity of the budgerigar,

Melopsittacus undulates." J. Comp. Physiol. A 189:135-142. The experiments described in that

paper are behavioral, but the data on receptor spectral sensitivities were obtained by other means

and are typical of most birds that have been examined. The behavioral data I therefore believe

also to be widely applicable.

Appendices



You imply, but do not make explicit, that at least one of your issues is vision under water. As

animals move far from the surface, available light decreases, and rod vision becomes more

important than cone vision. Penguins have been examined, and if memory serves me, their rod

vision is somewhat blue-shifted, as is characteristic of diving mammals and deep water fish. In

general, however, rhodopsin (the visual pigment of rods) has been highly conserved in evolution.

Adaptive variation is modest, with species differences in maximum sensitivity differing by no

more than about 25nm (birds, fish, mammals, frogs).

There is no way to examine visual sensitivity in carcasses. The biophysical techniques that are

needed require specialized equipment, fresh material, and considerable experience.

Collision avoidance during nocturnal migrations is a serious issue. Birds have not evolved with

the need to contend with the hazard of tall, man-made structures. I would anticipate that certain

underwater objects could pose a similar problem. In your case, I am curious to know what

environmental issue are under consideration. My response might be better tailored with that

information.

T.H. Goldsmith

=====================================================================

At 07:54 AM 8/11/03 -0800, you wrote:

Dr. Goldsmith:

Thanks for the prompt and informative answer. I am most interested in above-water vision, in

that I have two groups of birds (tubenoses and seaducks) that may collide with objects during

daily movements at colonies (tubenoses) or during migration (seaducks). In the former case, I

am interested in knowing that the tubenoses are not missing some wavelengths, so we could do

experiments to see what wavelengths of street lights might not attract them. In the latter case, we

are trying to develop an anti-collision lighting system for migrating eiders and want to ensure

that they can see wavelengths that we may want to use in the lights.

If I understand you correctly, we would need live birds for experiments. Would the birds be

injured during those experiments? Would the birds have to be studied at your lab, or could the

important equipment be moved to a location closer to the birds?

Thanks again for the information. I will look up your paper.

Bob

=====================================================================

Appendices



Bob,

As I said before, these birds are likely to see any lights that we can see, but it sounds like you are

concerned about collisions at night. In that case cone vision is likely irrelevant. Rod vision of

birds is basically the same as mammalian.

I think the problem for migrating birds is that they are evolved to fly at night at altitudes where

there are no natural obstacles. Towers and such are therefore a hazard that their brains are not

evolved to deal with. There is currently controversy about the installation of windmills for

generating electric power in Nantucket Sound. There is a lot of opposition from many quarters,

one of which is concern for the fall shorebird migration, as the off-shore sandbars and mud flats

around Chatham are a major refueling stop for many species.

When I was just starting graduate school I spent a week on an island off the coast of Maine

monitoring boroughs of breeding Leach's petrel that were being used in homing experiments.

The little rascals seemed to be able to fly in at night and find their hole near the edge of a spruce

forest. But again, this is a task that nature has prepared them to accomplish.

As for determining the visual sensitivity of birds there are several choices, none easy.

1) Behavioral, involving operant conditioning with food reward. Long, tedious, probably

possible with ducks, unlikely with tubenoses. Not relevant with rod vision.

2) Microspectrophotometry of rod and cone outer segments. Requires retinas removed from

birds that have been dark-adapted for hours before sacrifice. Difficult and requires specialized

equipment, for which there are only a few iterations in the world. The person with the most

experience applying this technique to birds is Jim Bowmaker in England. What one gets for the

effort are noisy absorption spectra from single rod and cone outer segments, from which one can

infer how much of the wavelength band birds can see. Hardly worth doing for rods; they just

don't vary.

3) Spectral sensitivity of eyes in anesthetized birds or in open eye cups form dark-adapted,

freshly sacrificed birds. Again, requires specialized equipment and experience. Depending on

how the experiment is done one gets a spectral sensitivity function for the retina or just the

receptor layer, which in either event is likely to reflect the relative numbers of different cones

present rather than the photopic luminosity function characteristic of a living bird.

I am not in a position to do any of these things. I retired several months ago, and my lab is a pale

shadow of what it once was.

Let me know if I can be of further "help."

Tim Goldsmith

migration and collision avoidance of eiders and …

Documents