J. Field Ornithol., 68(1):98--101 SPIDER WEBS AND WINDOWS AS POTENTIAIJJY IMPORTANT SOURCES OF HUMMINGBIRD MORTALITY DEVON L. G•nqAM Departmentof Biology Universityof Miami PO. Box 249118 Coral Gables,Florida, 33124 USA Abstract.--Sourcesof mortality for adult hummingbirdsare varied, but most reports are of starvationand predation by vertebrates.This paper reportstwo potentiallyimportant sources of mortality for tropical hermit hummingbirds,entanglement in spiderwebs and impacts with windows.Three instancesof hermit hummingbirds(Phaethornis spp.) tangled in webs of the spider Nephila clavipesin Costa Rica are reported. The placement of webs of these spidersin sitesfavored by hermit hummingbirdssuggests that entanglementmay occur reg- ularly. Observationsat buildingsalso suggest that traplining hermit hummingbirdsmay be more likely to die from strikingwindows than other hummingbirds.Window kills may need to be consideredfor studiesof populationsof hummingbirdslocated near buildings. TELAS DE ARAI•A Y VENTANAS COMO FUENTES POTENCIALES DE MORTALIDAD PARA ZUMBADORES Sinopsis.--Lasfuentes de mortalidad para zumbadoresson variadas.Pero la mayoria de los informes se circunscribena inanicitn y a depredacitn pot parte de vertebrados.En este trabajo se informan dos importantesfuentes de mortalidad para zumbadorestropicales (Phaethornisspp.) como los impactoscontra ventanasy el enredarsecon tela de arafia. Se informan tres casosde zumbadoresenredados con tela de la arafia Nephila clavipesen Costa Rica. La construccitn de las telas de arafia en lugares frecuentadospot los zumbadores sugiereque el enredarseen la tela de los arAcnidospudiera ocurrir regularmente.Obser- vacionesen edificiostambitn sugierenque las especiesde Phaethornispudieran estarm•s propensasa motif como resultadode impactoa ventanasque otrasespecies de zumbadores. Se debe considerarla mortalidad causadapot colicionescontra ventanascuando se hagan estudiospoblacionales de zumbadoresen lugarescercanos a edificios. Documentedsources of adult hummingbird (Trochilidae) mortalityin- clude starvation (Stiles1992) and predation by raptors (Beebe 1950, Low- ery 1938, MayT 1966, Peeters1963, Sick 1993, Sprot 1927, Stiles 1978), other birds (Sick 1993, Wright 1962), frogs (Monroe 1957, Norris-Elye 1944), snakes(Sick 1993), manfids (Butler 1949, Hildebrand 1949, Murray 1958), fish (Lockwood 1922), and mammals (Bent 1964, Sick 1993). Most predatorsappear to be opportunistic.However, the Tiny Hawk (Accipiter superciliosus)may specializein hunting hummingbirds(Stiles 1978). I report two additional sourcesof mortality, entanglement in spider webs and window kills, from observationsduring field work (June-July 1990;June 1991-December 1992; March-June 1993) at the La SelvaBi- ological Station, Costa Rica. I suggestthat both sourcesof mortality may be important for some speciesor populationsof hummingbirds. Spiderwebs.--Many hummingbirds regularly use spider webs for nesting material, and forage for insectstrapped in spider webs (Reinsen et al. 1986, Stiles1995, Wagner 1946, Young 1971). In primary forestundersto- ry at La Selva,two speciesof hermit hummingbirds,Long-tailed Hermits (Phaethornissuperciliosus) and Little Hermits (P long'uemareus),visit spi- 98 Vol.6a, •o. I HummingbirdMortality [99 der websincluding those of the large Golden Orb Spider, Nephila clavipes (hereafterNephila) (Stiles1995, Young 1971, pers. obs.). For both her- mits, lek and nestingsites are located in the understory,and foragingand flight paths are largely confined to trails, streamcourses, and other open areasin the understory (Stiles 1992, 1995; pers. obs.). Nephilais conspicuousthroughout its rangedue to its size(1-2 g), abun- dance, and placementof webs acrosstrails, streams,along ridgetops and acrossnatural gaps in the vegetation(Higgins 1987; Lubin 1978,1983; Ryp- stra 1985). Websare large (to 1 m in diameter),frequently aggregated, and surroundedby a tangleof strongnon-sticky threads. Fresh sticky strands are extremelyviscous (Lubin 1978, 1983;Rypstra 1985, Young 1971). On 16 Jul. 1990, two Long-tailed Hermits were found on the ground tangled in a destroyedNephila web, lessthan 50 m from an active lek. The attention of my party (C. Horvitz, J. LeCorff, M. Molina, and I) was drawn by the birds' distresscalls. One bird freed itself as we approached, the other was thickly coated with stickywebs, gluing it to severaldead twigsand leaves.When we removed the webbing,this bird could not fly, but recuperatedafter 2-3 min and two extended drinks of proffered Ga- torade©;it would certainly have died without intervention. In November 1992, L. Vargasand I observeda single Long-tailedHer- mit tangled in a Nephilaweb 4-5 m off the ground. This bird freed itself after about 30 s (assumingit began giving distresscalls the moment it became entangled), and the web was not totally destroyed.The third record of which ! am aware was of a Little Hermit found tangled in a Nephilaweb in 1977 by L. McDade (A. Mack, pers.comm.). This bird was releasedand presumablysurvived. Considering the rarity with which pre- dation events are witnessed, or dead birds encountered in forest habitats, I suggestthat these observationsare significant. For hummingbirds that forage in locationsfavored by Nephila such as forest understoryand edges,entanglement in spider webscould be an underrated or unacknowledgedsource of mortality. During periods of food scarcity,weakened birds might be especiallylikely to die. When hit- ting a web, a bird's momentum may either carry it through the web, or, when websare aggregated,cause it to becomecoated with both the sticky strandsof the inner web and the strong supporting strands.Humming- birdsare much larger than recordedNephila prey items (Higgins1987, Rypstra1985), and it is unlikely that Nephilaprey on birds. Entrapment of hummingbirds in spider webs has been reported repeatedly,but is usuallytreated as a rare, chance occurrence.Accounts of entrapment in Argiopeand Nephilawebs are related by Bent (1964), Baird et al. (1905), McCook (1889), Skutch (1973), Sick (1993), and Teixeira et al. (1991). In addition to death from exhaustion and shock, trapped humming- birds may be subjectto predation. Raptorsare often attracted to mist- netted birds (B. Loiselle, pers. comm.; pers. obs.), and trapped hermits are typicallyconspicuous, struggling vigorously and giving repeated dis- tresscalls. Small trapped birds are alsoopen to attackby other avianand mammalian predators. 100] D.L. Graham J.Field Ornithol. Winter 1997 Windows.--Observationsat La Selvasuggest that window-killsmay be a significantsource of mortalityfor local hummingbird populations.During two yearsat La Selva,birds hit windowsdaily. While some birds suffered glancing blowsand flew awayapparently uninjured, otherswere stunned or killed outright. Basedon dead birds found beneath windows,I estimate that 2-3 birds were killed weekly.Two sidesof a laboratory with large windowswere checked on a near-dailybasis while other locationswere checked irregularly; 2-3 deaths per week may thus be an underestimate. The birds most commonlyimpacting windows were hummingbirds,es- pecially the abundant Rufous-tailedHummingbird (Amazilia tzacatl)and the Long-tailed Hermit. These two speciesaccounted for about one-third of birds observedor known to hit windows.Only two impacts by the Rufous-tailedHummingbird are known to have resultedin death. In con- trast,dead Long-tailedHermits were found beneath windowsat leastonce a month, and stunnedindividuals every 1-3 wk. The massof Long-tailed Hermits (6.0 g; Stilesand Skutch 1989) and long traplining flights may mean that this speciesis more likely to hit windowswith lethal force than many other hummingbird species. Stunnedbirds that eventuallyflew awaymay have sustained injuries fatal in the short term. One Long-tailed Hermit bent half of the lower and upper mandiblesat a 30ø angle to the side,for instance,and Klem (1990a) documentsnon-apparent internal injuries in birds that hit windows.In- jured birds may alsobe vulnerable to predator attack, or be killed by fire ants (Solenopsissp.) which scavangeddead birds. For speciessuch as the Long-tailedHermit, even a mortalityrate of one individual per month might affect local populations.At La Selva, Stiles (1992) estimated that despite their seeming abundance,there may be only a few hundred individualson the 1500 ha property.Window mor- tality could conceivablyaffect demographicsof nearbyleks if, for instance, wider-rangingmales (Stiles and Wolf 1979) are more likelyto impactwin- dows,or if naivejuveniles suffer disproportionatemortality. Given that window-killsmay be the greatestsource of human-relatedbird mortality in the United States(Klem 1990b), and that populationsof dif- ferent bird speciesmay be affectedto different degreesby window-induced mortality,investigation into window-killsat tropical sitesis warranted.Ad- ditionally,research stations, private individualsand othersshould consider methodsof reducingwindow-kill deaths (see Klem 1979, 1990b). ACKNOWLEDGMENTS This paper wasimproved by commentsby C. R. Chandler,E Chavez-Ramirez,I4L Lips, D. McKey,E G. Stilesand one anonymousreviewer. This work wassupported by grantsto the author from the Organizationfor Tropical Studies(Pew Charitable Trust), and the National ScienceFoundation (DissertationImprovement Grant 4/9123052). LITERATURE CITED BAIRD.S. F., T. M. BREWER,AND R. RIDGEWAY.1905. A historyof North Americanbirds. Vol. II, Land birds. Little, Brown, and Company,Boston, Mass. Vol.68, No. 1 HummingbirdMortality [ 101