J Raptor Res. 28(1):9-12 © 1994 The Raptor Research Foundation, Inc. WINTER ROOST-SITE USE BY FEMALE AMERICAN KESTRELS IN LOUISIANA {Falco sparverius) Sean Doody^ J. Museum ofNatural Science, Louisiana State University, Baton Rouge, LA 70803 U.S.A. — Abstract. Roosting ecology ofAmerican kestrels {Falco sparverius) wintering in southcentral Louisiana was studied during the winters of 1988-89 and 1989-90. Twenty-eight roost sites were found for 26 kestrels. Twenty-four (857o) roost sites were man-made structures and four (15%) were natural roosts. Roost times averaged 2.1 ± 0.15 (SD) min before sunset {N = 46), Median height of man-made roost perches was 5.0 m (A'^ = 20, range = 2-50 m); mean height of natural roost perches was 6.3 ± 2.94 m {N = 4, range = 3-10 m). Kestrels did not roost communally; however, a male and a female roosted together for at least 10 d just prior to spring departure. Man-made roosts seemed to be preferred by migrant, female kestrels in southcentral Louisiana, as few females utilized natural roosts. Within areas of sufficient foraging quality, man-made roost sites may be a limiting factor for migrating kestrels. Key Words: American kestrel] Falco sparverius; Louisiana; roost site; winter. Uso de perchas en invierno por hembras de Falco sparverius en Louisiana — Resumen. Le ecologia de perchado de Falco sparverius invernantes en el centro-sur de Louisiana, fue estudiada durante los inviernos de 1988-89 y 1989-90. Se encontraron 28 percas para 26 individuos de F. sparverius. De estas perchas, 22 (85%) correspondian a estructuras hechas por el hombre y cuatro (15%) eran naturales. El tiempo promedio de perchado antes del atardecer {N = 46) fue de 2.1 (SD = 0.15). La altura promedio de las perchas artificiales fue de 5.0 m (A^ = 20, rango = 2-50 m); el promedio de altura de las perchas naturales fue de 6.3 m (SD = 2.94; A^ = 4, rango = 3-10 m). Los individuos de F. sparverius no se ubican en las mismas perchas, sin embargo, un macho y una hembra se percharon juntos por al menos 10 dias, antes de la partida en primavera. Las estructuras hechas por el hombre parecieron ser preferidas por las hembras migrantes de F. sparverius en el centro-sur de Louisiana y pocas utilizaron perchas naturales. En areas de alimentacion de suficiante calidad, los sitios de perchas artificiales pueden ser un factor limitante para la migracion de F. sparverius. [Traduccion de Ivan Lazo] In contrast to more traditional studies that have at dusk, and noted that an important territory re- focused on breeding populations of American kes- quirement appeared to be the availability of a roost. trels, {Falco sparverius) (Brewster 1925, Bent 1938, Quantified descriptions of roost-site characteristics Roest 1957, Willoughby and Cade 1964, Heintzel- are restricted to limited observations in Saskatche- man and Nagy 1968, Mueller 1971), more recent wan, Canada (Bortolotti and Wiebe 1993), and the studies have focused on their winter ecology (Craig- potential importance of roost sites in winter territory head and Craighead 1956, Koplin 1973, Balgooyen acquisition has not been examined. My 1976, Mills 1976, Layne 1980, Bildstein 1987, objective was to document roost-site use of Smallwood 1987). Because kestrels wintering in female kestrels, to comment on the potential advan- Louisiana spend up to 6 mo there each year (Table tages of roosting in man-made structures, and to 1), investigations into their wintering habits are nec- speculate on the potential importance of roost sites essary ifwe are to understand their complete biology. in the winter territory selection of migrant females. Roosting ecology of American kestrels has been little studied, and quantitative data are lacking (Bortolotti Study Area and Methods and Wiebe 1993). Mills (1975) observed kestrels American kestrels were observed on Ben Hur Research entering old buildings and barns and a hollow tree Farms, an agricultural extension of Louisiana State Uni- versity, located in the southwest corner of East Baton Rouge Parish (30°22'N, 91°11'W) during two winters. I ' Present address: Department of Biological Sciences, observed kestrels in the winter of 1988-89 and became Southeastern Louisiana University, Box 814, Hammond, familiar with both the existing habitat and the habits of LA 70402 U.S.A. settled, migrant females. The following winter (1989-90) 9 10 J. Sean Doody VoL. 28, No. 1 I examined and quantified roosting patterns. Kestrels are four (85%) roosts were in man-made structures and not permanent residents on the study area (Peterson 1980, four (157o) were in trees. Trees used included two Van Remsen pers. comm., pers. obs.). The study was water oaks {Quercus nigra) and two black willows limited to females due to the absence of males in the study {Salix nigra). In three of these a cavity was used area. The study area is bordered by the Mississippi River to Mean height of tree roosts was 6.3 ± 2.94 m (range the south, wooded bottomlands to the west, urban areas = 3-10 m). Two of the four tree-roosting kestrels to the north, and open habitat (<1% canopy cover) to the abandoned their respective roosts within 40 d of their east. An area of 7 km^ was surveyed from unimproved roads that served as survey routes between fields. Vege- first sighting. tation was grazed pasture during the winter and early Eleven kestrels were recorded going to roost 46 spring. Woody canopy, consisting of scattered oaks (Quer- times on evenings in January and February 1990. cus spp.), pecans {Carya illinoensis), and willows {Salix Mean roosting time was 2.1 min before sunset (SD nigra), which flank ditches on the site, was estimated at = = <5% 0.15 min; range 0.6-3.8 min). using a spherical densitometer. I searched for birds on weekdays between 0800 and 1000 H and again between Most barns and sheds had up to three open sides H 1500 and 1800 from 1 September 1989-30 March 1990. and roosting kestrels entered through all of these. These times were chosen for two reasons: 1) I was able Roost-perches were usually horizontal wooden raft- to observe kestrels going to roost each evening, and 2) ers but some were metal I-beams (e.g., a tower- kestrels were found to be most active during morning and roosting bird). Median height of man-made roost- late afternoon hours. This I noted while observing kestrels 3 d/wk in February and March 1989. Each survey con- perches was 5.0 m {N = 20, range = 2-50 m). Height sisted of driving the length (3.5 km) of each transect and of roost perch varied with the respective structure. stopping to observe each bird for up to 20 min. Transects One kestrel roosted 50-m high in a radio tower were then run in reverse producing four surveys per day. x2x2m m A Another roosted inal shelter, 2 above Evening surveys included watching kestrels go to roost. vehicle enabled observations of up to five individual kes- ground. Kestrels were easily disturbed when at roost trels going to roost per evening. Birds were viewed through even late at night, and would fly to the nearest tree 7 X binoculars and the few areas that were inaccessible by after being flushed. vehicle were surveyed on foot. Thirteen barns and sheds were determined to be As sunset approached, I observed individual kestrels to determine when and where they roosted, I located roosting favorable for roosting kestrels due to their structure kestrels at night to determine the height of roost-perches. and surrounding habitat. Nine (69%) were used in Daily sunset times were obtained from the Department winter 1988-89, and eight (62%) were used in 1989- of Meteorology, Louisiana State University. Barns were 90. Kestrels used eight barns and sheds both winters, differentiated from sheds in that the latter had at least two while three barns and sheds were vacant both win- open sides and barns generally were higher than sheds. Individual kestrels were distinguishable from one an- ters. Two ofthose three were second structures with- other because they used the same perches repeatedly in a kestrel’s territory. Four females defended areas throughout the winter and were watched until dusk when which encompassed two or more man-made struc- they went to roost. Kestrels could be found near their tures. Three of these chose to roost in only one of respective roosts each day, and moved closer to them as the structures, while the fourth utilized two sheds dusk approached. For example, a 0.5 hr evening transect would typically reveal kestrels near the same roosts, perched as roost sites. One female was believed (due to the — m the same perches (e.g., branches of a tree), or hunting uniqueness of its roost site a henhouse vent) to the same area surrounding those roosts. Two individuals roost in the same structure for two consecutive years. were recognized by their respective breast-feather color Kestrels did not roost communally. anomalies, and were used to test the above criteria. Because birds were not banded or radiotagged, positive Areas (4 km^ total) with no man-made roosts were identification was not assured. On occasion in 1989, in- either avoided or deserted by females, despite being dividual kestrels were watched the entire day and were similar in hunting habitat to areas with such struc- found to stay relatively close to the structure in which they tures. One open area (0.8 km^) in particular was would eventually roost. It is unlikely that kestrels switched roost sites, but females may have displaced other females hunted by females early during both winters but was during the study. Smallwood (1987), however, found no deserted later in the season. such displacement for 650 kestrel territories in peninsular Females who hunted adjacent areas showed no Florida. agonistic behavior, except in late September and Oc- tober when kestrels first began to arrive (Table 1) Results I witnessed no aggressive interactions between fe- Roosts were identified for 14 female kestrels in males (eight different individuals) perched within m winter 1988-89 and for 12 in 1989-90. Twenty- 15 of one another on 27 occasions in 1989-90. March 1994 Kestrel Winter Roosts 11 Table 1. Length of stays of female American kestrels wintering in southcentral Louisiana. Description SD Range N a: Arrival date 30 Sept. 17.0 d 15 Sept.-26 Oct. 11 Departure date 14 Mar. 7.6 d 28 Feb.-28 Mar. 29 Length of winter stay 166 d 126-195 d 29 All of these females were “neighboring females,” or These interactions were the first reported for mi- those hunting adjacent areas, and most perches were grant kestrels around roost sites. Such interactions utility lines. In contrast, 18 high-flying “transient” were commonplace in my study area, particularly kestrels were harassed by kestrels who had already in the early weeks of arrivals (Table 1). However, chosen a roost. In six of these cases, two adjacent “neighboring” females (migrants which have hunted females simultaneously attacked “high-flying tran- near one another) defending adjacent areas often sients.” In eleven instances I observed females dis- perched next to one another for extended lengths of playing in a series of climbs and dives at heights of time without agonistic behavior. The abundance of 5-15 m. The displays were continuous as the females prey and apparently suitable habitat might repress circumnavigated areas encompassing man-made the need for territory defense (chasing, contact) roosts. The displays were often interrupted by chases against a “familiar” female by a kestrel who does initiated by the displaying female and were usually not wish to displace its neighbor. accompanied by a series of “klee” notes. Nineteen Man-made structures as roosts may be advanta- chases were directed toward “unfamiliar” females geous to wintering kestrels. The introduction of nest (females who did not hunt in adjacent areas), and boxes was probably responsible for the increase of one female briefly chased a “neighboring” female. kestrels nesting in Pennsylvania (Nagy 1963) and in Wisconsin (Hamerstrom et al. 1973). Areas en- Discussion compassing preferred roost sites could similarly at- Wintering kestrels seemed to prefer man-made tract higher densities of kestrels than areas lacking structures as roosts in southcentral Louisiana. All such structures. The man-made structures on my females in my study that had man-made structures study area allowed females to roost close to appar- in their territories roosted in them. Most of these ently suitable foraging habitat (e.g., open pasture structures were barns or sheds. Three of four tree- with scattered trees). roosting kestrels had no man-made structures in their Alternatively, female kestrels may choose man- territories. The fourth had access to one building, made structures as potential roost sites due to their but it contained no sheltered perch, which seemed visibility; man-made structures may simply be easier to be a requirement for roosting kestrels. Roost sites to locate than natural cavities. I noted that while were extremely variable in structure and height. It none of the three tree cavities were used in consec- would appear, therefore, that any sheltered perch utive winters, eight buildings were utilized both win- might suffice, regardless of height above ground as ters. Another advantage for kestrels roosting in man- evident in the tower-roosting bird. made structures may be protection from severe Kestrels I observed generally went to roost around weather. Kestrels in Saskatchewan, Canada roosted sunset, but Miller (1954) observed one female going in spruce trees (Bortolotti and Wiebe 1993). Bal- to roost an average of 13 min after sunset. Bortolotti gooyen (1976) postulated that roosting in conifers and Wiebe (1993) observed kestrels roost between (vs. deciduous trees) was thermally advantageous for 5-15 min after sunset. Female kestrels on my study kestrels, as did Warkentin and West (1990) for Mer- area did not actively hunt just prior to roosting, but lins (Falco columbarius). generally perched close to the roost, as reported by Smallwood (1988) suggested that a kestrel’s ar- Bortolotti and Wiebe (1993). rival date on wintering areas was the principal de- Bortolotti and Wiebe (1993) observed seven ag- terminant of the foraging quality of habitats still onistic intraspecific reactions between kestrels of both available for occupancy. In my study a lack of kes- sexes in a short time period in Saskatoon, Canada. trels within seemingly favorable areas existed 12 J. Sean Doody VoL. 28, No. 1 throughout each winter. These areas, however, dif- 1973. Nest boxes; An effective management tool for fered from areas hunted by kestrels in that they did kestrels. J. Wildl. Manage. 37:400-403. not contain man-made roosts. Heintzelman, D.S. and A.C. Nagy. 1968. Clutch sizes, Where a surplus of apparently suitable hunting hatchability rates, and sex ratios of sparrow hawks in habitat exists, roost sites may be important for fe- eastern Pennsylvania. Wilson Bull. 80(3):306-31 1. Koplin, J.R. 1973. Differential habitat use by sexes of males in the location of their winter territories. Ad- American kestrels wintering in North Carolina. Raptor ditional observations, however, particularly at dif- Res. 7:39-42. ferent locations with variable roost-site availability, Layne, J.N. 1980. Trends in numbers of American are needed to reveal the ultimate importance ofroost- kestrels on roadside counts in southcentral Florida from site selection in winter territory establishment. Roost- 1968 to 1976. Fla. Field Nat. 8(l):l-36. site use of wintering males, which have been shown Miller, L. 1954. A sparrow hawk’s roosting schedule. to utilize different habitats than wintering females Condor 56:230-231. A (Koplin 1973, Mills 1976, Stinson et al. 1981), also Mills, G.S. 1975. winter population study of the should be investigated. American kestrel in central Ohio. Wilson Bull. 87:241- 247. Acknowledgments 1976. American kestrel sex ratios and habitat . I am grateful to J.V. Remsen for his encouragement separation. Auk 93:740-748. and advice in the initial stages of this study; the 1988-89 Mueller, H.C. 1971. Displays and vocalizations ofthe data were collected as part of a research project for Rem- sparrow hawk. Wilson Bull. 83(3):249-254. sen’s ornithology class. I also thank the Ben Hur Research Nagy, A.C. 1963. Population densities of sparrow hawks Station for allowing me access to the farms. The Louisiana in eastern Pennsylvania. Wilson Bull. 75(1):93. StateUniversity Department ofMeteorology provided daily Peterson, R.T. 1980. A field guide to the birds of sunset times. eastern and central North America. Houghton-Mifflin, MA Literature Cited Boston, U.S.A. Balgooyen, T.G. 1976. Behavior and ecology of the Roest, A.I. 1957. Notesonthe American sparrow hawk. Auk 74:1-19. American kestrel (Falco sparverius) in the Sierra Ne- Smallwood, J.A. 1987. Sexual segregation by habitat vada of California. Univ. Calif. Publ. Zool. 103:1-87. in American kestrels {Falco sparverius) wintering in Bent, A.C. 1938. Life histories ofNorth American birds southcentral Florida: Vegetative structure and re- of prey, part 2. U.S. Natl. Mus. Bull. 170, Washington, DC sponses to differential prey availability. Condor 89:842- U.S.A. Bildstein, K.L. 1987. The behavioral ecology of red- 849. 1988. A mechanism of sexual segregation by tailed hawks, rough-legged hawks, northern harriers, . and American kestrels in southcentral Ohio. Ohio Biol. habitat in American kestrels {Falco sparverius) winter- Surv. Notes No. 18. Ohio State Univ., Columbus, OH ing in southcentral Florida. Auk 105:36-46. Stinson, C.H., D.L. Crawford and Lauthner. 1981 U.S.A. J. Bortolotti, G.R. and K.L. Wiebe. 1993. Roosting Sex differences in winter habitat of American kestrels American kestrels {Falco sparverius) during migration in Georgia. J. Field Ornithol. 52:29-35. Warkentin, LG. and N.H. West. 1990. Ecological in Saskatchewan. Raptor Res. 27:47-49. J. Brewster, W. 1925. The birds of the Lake Umbagog energetics of wintering merlins {Falco columbarius). Physiol. Zool. 63:308-333. region of Maine. Bull. Mus. Comp. Zool. 66(2):211- 402. Willoughby, J. and T.J. Cade. 1964. Breeding be- Craighead, J.J. and F.C. Craighead, Jr. 1956. havior of the American kestrel (Sparrow Hawk). Liv- ing Bird 3:75-96. Hawks, owls, and wildlife. Stackpole Co., Harrisburg, PA U.S.A. Hamerstrom, F., F.H. Hamerstrom and J. Hart. Received 24 February 1993; accepted 4 October 1993