Effects of Window Angling, Feeder Placement, and Scavengers on Avian Mortality at Plate Glass

2004·3.6 MB·English

by  Daniel Klem

Wilson Bulletin, 1 16(1), 2004, pp. 69—73
EFFECTS OF WINDOW ANGLING, FEEDER PLACEMENT, AND
SCAVENGERS ON AVIAN MORTALITY AT PLATE GLASS
DANIEL KLEM, JR.,' ’ DAVID C. KECK,'-’ KARL L. MARTY,'-’
AMY J. MILLER BALL,'-^ ELIZABETH E. NICIU,' s AND CORRY T. PLATT' <>
—
ABSTRACT. Extensive observations and experiments suggest that collisions with plate glass result in more
avian mortalities than any other human-associated factor. We tested the effects of window angling and the
distance of bird feeders from windows on bird-glass collisions. Strike frequency differed among windows ori-
ented vertically (control) and those angled 20 and 40 degrees from vertical; as the angle oforientation increased,
strikes and fatalities decreased. Strike frequency and fatalities at windows also increased as the distancebetween
bird feeders and the glass surface increased. No fatalities were recorded when feeders were located within 1 m
ofa window, but a marked increase in mortality occurred when feeders were placed 5 and 10 m from the glass.
Most glass-collision victims may go unnoticed, hidden by vegetation where they remain out of view or are
removed by scavengers. We found that scavengers frequently removed baits from beneath windows at six
buildings, but no baits were taken from a site without windows that served as a control. The importance of
window strikes as an avian mortality factor, and the likelihood that it will increase over time, compel us to
recommend a reevaluation of the Migratory Bird Treaty Act (MBTA). Angling panes in new and remodeled
buildings and placing bird feeders closerto windows can potentially reduce avian mortality. Received3 Septem-
ber 2003, accepted 6 April 2004.
Bir—ds are vulnerable to collisions with win- magnitude less than those known to occur at
dows from small panes to walls of glass glass (Banks 1979, Klem 1991, Shire et al.
covering entire buildings (Klem 1989, 1991). 2000, Erick.son et al. 2001, John.son e—t al.
Extensive observations and several experi- 2002). Only predation by domestic cats es-
ments reveal that birds apparently cannot rec- timated a—t 1 billion birds per year in North
ognize clearor reflective panes ofglass as bar- America results in comparable mortality
riers to be avoided (Klem 1990b). Conserva- rates.
tive estimates of annual avian mortality from Glass as a mortality factor for specihc spe-
collisions with glass for the U.S. alone and for cies is generally unknown. One exception is
the entire North American continent range the Sw—ift Parrot (Lathamus discolor) of Tas-
from approximately 100 million to 1 billion mania 1.5% of the entire population (l.OOO
birds, respectively (Klem 1990b, 1991; Dunn breeding pairs) is killed annually by colliding
1993), representing from 0.5 to 5% of the fall with windows (R. Brereton pers. comm.). Le-
bird population (American Ornithologists’ thal collisions have been recorded whenever
Union 1975). Comparative hgures for other and wherever both birds and glass occur.
human-associated bird mortalities (collisions
Moreover, the fittest indi\iduals of popula-
with vehicles, communication towers, wind tions are known to be as vulnerable as any
turbines, power lines, or nocturnal strikes at other (Klem 1990b).
multistory buildings) are at least an order of Although bird-glass collisions occur in e\-
ery season, the general impression is that most
' Dept, of Biology, Muhlenberg College, Allentown, occur during fall and spring migration. In fact,
PA 1X104, USA. systematic monitoring of houses suggests that
I DH^C1u9r9r7e1n,tUadSdAre..ss: X Chatham Rd., Rehoboth Beach, most birds in North America arc killed during
I ^Current address: lU). Box 103, East Dennis. MA winter, when many are attracted to bird feetl-
I' 02641, USA. I ers (Klem 1990b. Dunn 1993). During 2001.
‘Current address: 57 Kiah's Way, Pxist .Sandwich. 54 million U..S. residents (257( of the popu-
j MA 02537, USA. lation >16 years of age) participated in feed-
I 'Current address: 4X0 Kings Rd., Athens. (lA ing birds and other wikllife (U..S. Department
.30606, USA.
C'urrent address: 104 liagles Nest Ct., Cary. NC of Interior and U.S. Department ofC'ommcrce
27513, USA. 2002). Ivxaccrbating our uiulcrstanding of the
^CorrescHinding author; e-mail: klem(«'muhlenberg.edu problem is the practice of jilanting \cgetation
1
70 THE WILSON BULLETIN • Vol. 116, No. I, March 2004
around human dwellings, which often hides consisted ofcut corn stalks and grasses, a uni-
glass-strike casualties from human attention form habitat with no obvious flight paths to
(Klem 1990a). Scavengers are known to re- attract birds to or away from the windows.
move dead birds (Klem 1981), and they are Each window measured 1.4 m wide X 1.2 m
believed to kill and remove injured birds from high, and was mounted 1.2 m above ground.
window-collision sites. However, no experi- Wire-mesh trays were placed under each win-
ments have specifically addressed the removal dow to catch casualties. Tinted and clear win-
of dead or injured birds by scavengers as a dows were placed in each of three orienta-
possible explanation for why collision casu- tions: vertical (serving as the control), and an-
alties are not discovered beneath windows gled downward from vertical at 20 and 40 de-
more often. grees. Each window was constructed in such
Here, we examine the effects ofwindow an- a way that it could be placed at all orienta-
gling on bird-glass collisions, and attempt to tions. All panes remained in the same position
determine where bird feeders should be placed throughout the experiment, and each day the
to reduce or eliminate window hazards. Ad- three pane orientations were assigned random-
ditionally, we present experimental results that ly to the tinted and clear panes; panes were
indicate scavengers regularly patrol areas near checked and changed 30 min before last light.
windows and remove evidence that avian fa- The parameter measured was the number of
talities have occurred. detectable bird strikes. A strike was registered
when either a dead or injured bird was found
METHODS
beneath a window, or when fluid or a blood
We conducted the window angling and bird smear, feather, or body smudge was found on
feeder placement experiments at the 15-ha the glass. All window-killed casualties left ev-
Muhlenberg College Raker field site south of idence of a strike on the glass. Our data are
Germansville, Lehigh County, Pennsylvania likely to be incomplete and conservative be-
(40° 41' N, 75° 42' W). Land cover is 40% cause some strikes (e.g., a glancing blow) may
second-growth deciduous woodland, 35% fal- not have left evidence of a collision. In ad-
low field, 15% wetland, 5% open water (Jor- dition, predators and scavengers may have re-
dan Creek), and 5% man-made structures. The moved some injured or dead birds that we did
scavenger experiment was conducted at Muhl- not detect.
—
enberg College (40° 35' N, 75° 30' W) and Feeder placement experiments We con-
.
Cedar Crest College (40° 35' N, 75° 31' W). ducted two feeder placement experiments,
The two campuses are adjacent to one another from 31 October to 17 December 1991 and
in suburban, west Allentown, Lehigh County, from 24 January to 29 February 1992. The
PenWnisynldvoawniaan.gling experiment.—We conduct- pfilrasttfeoxrpmerfeiemdeenrts t1e,s5t,edantdhe10effmecftsroomfcpolnavceinn-g
ed the window angling experiment from 20 tional, vertically oriented panes. The second
January to 17 May 1991. The basic design tested the effects of placing feeders 2, 3, and
was the same as reported previously (Klem 4 m from windows. In the 1-, 5-, and 10-m
1989, 1990b), consisting of six wood-framed experiment, we used the same six windows
picture windows, simulating those in houses; that were used in the angling experiment, but
all were placed in the same habitat and faced repositioned them so that they were 55 m
the same direction along the edge of a mixed apart. Each feeder placement was tested si-
deciduous forest and open field (Klem 1989: multaneously at one clear and one tinted pane.
Fig. 1). Three ofthe windows were tinted dark Windows were positioned farther apart to re-
gray and three were clear. Tinted panes alter- duce the possibility that birds attracted to one
nated with clearpanes across the six positions, feeder placement might strike a window as-
which were separated from one anotherby 43, sociated with another. In the 2-, 3-, and 4-m
15, 24, 18, and 20 m. Distances between win- experiment, we replaced the tinted panes with
dows were selected to simulate the construc- clear panes because the quality of reflection
tion practice ofbuilding homes in rows, in this from the tinted glass (salvaged) was not sharp,
case adjacent to one another along a tree line and we suspected that this effect contributed
facing a field. The vegetation cover in the field to the lower number of strikes at tinted panes
Klein et al. • BIRD KILLS AT PLATE GLASS WINDOWS 71
in our angling experiment. Each feeder place- TABLE 1. The number of bird strikes (fatalities)
ment was tested simultaneously at two clear decreases with increased window angling. Data are
panes. Bird feeders were flat trays measuring from field experiments at Germansville, Lehigh Coun-
30.5 X 61.0 cm, and placed so that they were ty, Pennsylvania, 1991.
centered and level with the bottom of each
Angleoftilt
windowpane. Feed consisted of a 1:1 mixture (orientation)
of black-oil sunflower seeds and either Glasstype Vertical 20degrees 40degrees Total
cracked corn or white millet. Each feeder was
Tinted 9 (0) 6 (0) 2 (0) 17 (0)
lightly covered with the same feed mixture 1 Clear 21 (7) 9 (4) 6 (1) 36 (12)
hr before sunrise each day. We recorded strike Total 30 (7) 15 (4) 8 (1) 53 (12)
frequencies in the same manner as in the an-
gling experiment.
—
Scavenger experiment. The scavenger ex- Chickadee (Poecile atricapillus). Northern
periment was conducted from 25 January to Cardinal {Cardinalis cardinalis). White-
10 April 1992. Seven locations, distributed throated Sparrow (Zonotrichia albicollis).
among four buildings on the campuses of Song Sparrow (Melospiza melodia), and
Muhlenberg and Cedar Crest colleges, were Dark-eyed Junco {Jiinco hyemalis). Total
selected as experimental sites. Six of these number of strikes differed significantly across
sites were below windows at which lethal bird window angle, with 57% at the vertical con-
strikes were known to have—occurred. A build- trol, 28% at the 20-degree angle, and 15% at
ing wall without windows but which faced the 40-degree angle = 14.3, df = 2, P =
habitat and had hu—man passage similar to that 0.001). Similar strike differences occurred
of the other sites served as a control. Ap- across window angle for both the clear panes
i proximately 30 g of chicken breast meat was (X" = 10.5, df = 2, P = 0.005) and tinted
used to simulate a window-killed bird approx- panes (y^ = 4.4, df = 2, P = 0.1 1). Irrespec-
imately the size ofa Hermit Thrush (Catharus tive of window angle, strike frequency dif-
giittatus). At each location bait was placed out fered between clear (36) and tinted (17) panes
I of sight from human passersby. From 25 Jan- (Binomial test, Z = —2.47, 2-tailed, P =
uary to 8 March, baits at each site were 0.013).
: —
checked every 12 hr (10:00 and 22:00 EST), Feeder placement experiments. During
1I and then once every 24 hr (22:00) from 9 the 1-, 5-, and 10-m experiment we recorded
< March to 10 April. Baits were replaced after 105 strikes, 50% of which were fatal. Dead
4 days if no disturbance occurred, or during birds were: Blue Jay {Cyanocitta cristata).
I
' bait checks if baits had been removed or Black-capped Chickadee, Tufted Titmouse
moved from their original location. The {Baeolophus hicolor). White-breasted Nut-
ground areas on which baits were placed were hatch (Sitta carolinensis), American Robin
f finely raked so as to record the tracks of po- (Tnrdits migratorins). Common Yellowthroal
tential scavengers. The parameter measured (Geothlypis trichas). Northern Cardinal,
was the number of baits taken or moved from White-throated Sparrow, Dark-eyed Junco,
i
their original locations by scavengers each House Finch (Carpodacns mexicanns). and
day at each site. American Goldfinch {Cardiielis tristis). Total
We used SPSS (SPSS, Inc. 2002) for all sta- number of strikes differed significantly across
tistical analyses. Chi-square goodness-of-flt placements, with 25 (24%) at 1 m. 29 (28%)
was used to evaluate experimental results ex- at 5 111, and 51 (48%) at 10 in (y- = 1 1.2, df
cept for one dichotomous comparison in = 2. P = ().()04). Fatal strikes also differed
which a 2-tailed binomial test was more ap- significantly across feeder placement, with 0
propriate. We considered test results to be sta- (0%) at 1 111, 17 (339f ) at 5 in, and 35 (67%)
tistically significant when P < 0.05. at 10 111 (y- = 33.0, df = 2, P < O.OOl): all
I but four fatalities occurred at clear panes.
I RKSUI.TS During the 2-, 3-, aiul 4-ni cxiieriment we
—
Window angling exjieriment. We recortled recortlctl 197 strikes. 21 (11%) of which were
, 53 strikes, of which 12 (239f) were fatal ( fa- fatal. Dead birds were: luftetl ritmouse.
'I ble 1 ). Dead birds included: Black-capped Northern C'ardinal, C’hipping .Sparrow (Sfiizel-
\
72 THE WILSON BULLETIN • Vol. 116, No. 1, March 2004
ing the ground instead ofthe surrounding hab-
itat and sky (Klem 1990b). Our experiments
revealed that window-strike fatalities can be
reduced significantly if panes are angled
downward 20° and 40° from the vertical. An-
gled glass also may reduce the force with
which birds in horizontal flight strike panes.
Although glass orientation does not eliminate
the lethal hazard ofwindows, it is an effective
bird-strike deterrent and should be considered
by architects and others involved in planning
new structures or in remodeling existing ones.
Feederplacement The effectiveness of window angling is sub-
FIG. 1. The proportion of bird fatalities (%) at stantial and is likely to become practical in
windows increases as bird feeders are placed farther one-story structures or at ground level in mul-
fromthe glass surface. Dataarefromfieldexperiments tistory buildings.
at Germansville, Lehigh County, Pennsylvania, 1991- Results of the angled window and feeder
1992. placement experiments in which tinted glass
was used indicate that tinting may afford some
la passerina). Field Sparrow {S. pusilla). protection for birds. Although fewer strikes
White-throated Sparrow, Dark-eyed Junco, were recorded at tinted panes, the numbers of
and House Finch. Total number of strikes dif- strikes at tinted and clear windows had similar
fered significantly across placements, with 46 trends in the angling experiment. We believe
(23%) at 2 m, 91 (46%) at 3 m, and 60 (31%) that the lower number of strikes at our tinted
at 4 m (x^ = 16.2, df = 2, P = 0.001). The panes is explained best by the relatively poor
number of fatal strikes also differed signifi- reflective quality of the salvaged glass we
cantly across feeder placement, with 1 (5%) used. In a previous experiment, clear and tint-
a(tX^2=m,8.90,(d4f3%=)2a,tP3 =m,0.a0n1d8)1.1T(h5e2%c)omabti4nemd e1d989p)a.neHsowweevreer,eqtuhaellhyazlaertdhsaloftovabriyrdisng(tKylpeems
results of both feeder placement experiments of clear and tinted glass need further study to
revealed a marked increase in the proportion determine their specific risks.
of fatal strikes as distance between feeder and Results of previous experiments have doc-
window increased (Fig. 1)—. umented the effectiveness of two other meth-
Scavenger experiment. During 77 days, ods in eliminating window kills, but these
scavengers found and disturbed 69 (13%) of techniques are often unacceptable to home
539 baits. At the six sites below windows, the owners and managers ofcommercial buildings
number of disturbed baits was 17, 12, 12, 8, because they are costly, impractical, or aes-
9, and 1 1, respectively. Tracks revealed that thetically unpleasing (Klem 1990b, 1991).
the following scavengers found bait: 15 (22%) One method is to place a physical barrier (net-
squirrel, 31 (45%) cat, 3 (4%) dog, 4 (6%) ting, awning) in front of glass to prevent bird
bird spp., and 16 (23%) unidentified. Baits strikes. Screens for this purpose are now com-
were found by the same type of scavenger at mercially available. Another method is to ap-
the same site on 6 (cat), 8 (cat), and 2 (squirrel ply opaque or translucent objects (hawk sil-
and dog) consecutive days, suggesting that the houettes, geometric shapes, other creative pat-
same individuals may have returned to a lo- terns), separated by 5-10 cm, to the outside
cation where food was found previously. At surfaces of windowpanes (Klem 1990b). Ob-
the control-site building with no windows jects of any shape that visibly contrast with
there was no evidence that any scavenger dis- glass allow birds to recognize and avoid win-
covered bait during the entire experimental dows. The use of ultraviolet (UV) patterns to
period.
deter bird-glass collisions is currently under
DISCUSSION study in our laboratory. The expected princi-
UV
Preliminary observations had indicated that pal advantage of deterrents is that they
window angling might protect birds by reflect- will be invisible to humans. Another potential
Klem et al. • BIRD KILLS AT PLATE GLASS WINDOWS 73
long-term solution is the manufacture of a LITERATURE CITED
novel type of sheet glass, especially for mul- American Ornithologists’ Union. 1975. Report of
tistory, glass-covered buildings (Klem 1991); theadhoc committeeon scientificandeducational
this glass would provide an unobstructed view use of wild birds. Auk 92(suppl.):1A-27A.
from the inside, but, when viewed from the Banks, R. C. 1979. Human related mortality of birds
in the United States. U.S. Pish and Wildlife Ser-
outside, creative designs (dots, lines, variously vice Special Report 215:1-16.
shaped objects separated by 5-10 cm) would Corcoran, L. M. 1999. Migratory Bird Treaty Act:
be visible to birds and direct them away from strictcriminal liabilityfornon-huntingcausedbird
deaths. Denver University Law Review 77:315-
the hazard.
358.
U.S. courts have established strict liability Dunn, E. H. 1993. Bird mortality from striking resi-
for unintentional avian mortality associated dential windows in winter. Journal of Pield Orni-
with pesticides and power lines pursuant to thology 64:302-309.
the Migratory Bird Treaty Act of 1918 Erickson, W. R, G. D. Johnson, M. D. Strickland,
(MBTA), as amended, orthe Endangered Spe- D. P. Young, Jr., K. J. Sernka, and R. E. Good.
2001. Avian collisions with wind turbines: a sum-
cies Act (ESA) of 1973 (Corcoran 1999); mary ofexisting studies and comparisons to other
however, the courts have not established strict sources ofavian collision mortality in the United
liability for fatalities associated with vehicle, States. National Wind Coordinating Committee,
tower, or glass collisions. Our results suggest Washington, D.C.
Johnson, G. D., W. P. Erickson, M. D. Strickland,
that bird kills at glass are substantial, foresee- M. E Shepherd, D. A. Shepherd, and S. A. Sar-
able, and avoidable (Klem 1989, 1990b, 1991; APPO. 2002. Collision mortality of local and mi-
Corcoran 1999) and we suggest that birds grant birds at a large-scale wind-power develop-
merit consideration for protection from glass ment on Buffalo Ridge, Minnesota. Wildlife So-
MBTA ciety Bulletin 30:879-887.
collision under the purview ofthe and Klem, D.,Jr. 1981. Avian predatorshuntingbirdsnear
ESA. windows. Proceedings ofthe Pennsylvania Acad-
Avian injury and mortality from collisions emy of Science 55:90-92.
with glass can be reduced worldwide by those Klem, D., Jr. 1989. Bird-window collisions. Wilson
who feed birds. Our results showed an in- KlemB,ulDl.e,tiJnr.110919:06a0.6-B6i2r0d.injuries, cause ofdeath, and
crease in window fatalities when bird feeders recuperation from collisions with windows. Jour-
are placed 2-10 m from a glass surface, with nal ofPield Ornithology 61:1 15-1 19.
marked increases at 5 and 10 m. Feeders Klem, D., Jr. 1990b. Collisions between birds and
m windows: mortality and prevention. Journal of
placed within 1 ofa pane led to no fatalities Pield Ornithology 61:120-128.
and offer the most protection for birds, espe- Klem, D., Jr. 1991. Glass and bird kills: an overview
cially at residential buildings and visitor cen- and suggested planning and design methods of
ters of local, state, and federal parks and other preventing a fatal hazard. Pages 99-104 in Wild-
recreational facilities. Wli.feAcodnasmersvaatnidonDi.n Km.etrLoepeodlyi,taEndse.n)v.irNoantimoennatls(ILn.-
stitute for Urban Wildlife Symposium Scries 2.
ACKNOWLEDGMENTS National Institute for Urban Wildlife, Columbia,
Maryland.
We thank Muhlenberg College and, especially, 1. R. ShireC,omGm.unGi.,catK.ionBrtoowwne,rs:anaddeGa.dlyWinhia-z;agrrdadt.o b2i0r0d0s..
Schmoyer, J. R. Vaughan, and J. C. Weston for advice Special Report, American Bird Conservancy,
and financial support. D. J. McCutheon donated bird Washington, D.C.
aseneddanaanldysiEs.LW.eLetwhiasn,k 1P1.1,G.asSsiasetnegderinfodratoabsceorlvlaetcitoinosn SPSSI,nci.NCC.hi2c0a0g2o.,SIPlSliSnoifso.r Windows, ver. 1 1.5. SPSS.
andcommentsthat improvedourwork. Wearegrateful U.S. Di;parimi;nt oi Intirior and I’.S. Diparimini
to PL. H. Dunn, K. L. Bildstein, and an anonymous ()i CoMMLRCi;. 2002. 2001 natit)nal survey offish-
referee for helpful comments and suggestions on the ing, hunting, ami wildlile-associateil recreation.
manuscript. Washington, D.(’.