278 SHORTCOMMUNICATIONS VOL. 38, NO. 3 j RaptorRes. 38(3):278-282 ¸ 2004 The Raptor Research Foundation, Inc. PREMIGRATORY COMMUNAL ROOSTS OF THE LESSER KESTREL IN THE BOREAL SUMMER PEDRO P. OLEA 1 AND RUBI•N VERA Dpto.de Biologia Molecular y Cienciasdel Medio Ambiente, Faculty of Experimental Sciences, SEK University, CampusSanta Cruz la Real, SegoviaE-40003 Spain d•GEL DE FRUTOS2 AND HUGO ROBLES Dpto.de Biolo•a Animal,Fac. de C.C. BiolSgicasy Ambientales, University of Letn, Letn E-2407l Spain KEYWORDS: LesserKestrel; Falco naumanni;premigratory roosts during the summer in the Palearctic comes from aggregations;communal roosts. Spain and Italy, and consistsof anecdotal records of two roosts(Tejero et al. 1982 and Aparicio1990, respectively) The LesserKestrel (Falconaumanni) is a small migra- and on the monitoring of birds at four roostsduring a tory falcon breeding in the Palearctic and wintering single summer (Palumbo 1997, Olea 200lb, Ursfia and mainly in Africa (Siegfriedand Skead 1971, Cramp and Tella 2001). Here, we examine the use of communal Simmons 1980). This speciesis consideredto be a threat- roostsby Lesser Kestrelsin northwesternSpain during ened specieslisted as Vulnerable (BirdLife International three summers. We describe roost site characteristics and 2002). Although many aspectsof their breeding biology examine numbers and temporal variation of roosting have been studied (e.g., Negro et al. 1992, Donfizar et al. Lesser Kestrels. 1993, Hiraldo et al. 1996, Tella et al. 1996a, 1996b, 1997) METHODS there is little known about the post-fledging period of this species(but see Bustamanteand Negro 1994, Garcia Communal roosts were located in the province of 2000, Olea 2001a). This period can be particularlycritical Le6n, northwestern Spain, at about 800 m above sea lev- for the survivalof the juvenile falconsbecause recently el, between the towns of Santas Martas and Sahagfin fledged birds have little experience in foraging, and in (42ø25'-42ø17'N, and 5ø23'-5ø06'W; 11.7øGannual mean avoiding predators and human activities (Anders et al. temperature and 486 mm rainfall). The studyarea covers 1998). During the post-fiedging period, Lesser Kestrels 384 km2 devotedto cereal crops.This area wasselected because communal roosting was observed previously seem to increasesignificantly in their abundancein some (Olea 200lb; pers. obs.). In 2002, we documented 23 areas in northern Spain, where breeding populations are breeding pairs of Lesser Kestrelsin the study area. The small or nonexistent (e.g., Llamas et al. 1987, Jubete landscapewas fiat and open, with practicallyno trees. 1997, Romfin 1998, Olea 200lb) and gather in commu- The area is broken up by a number of small seasonal nal roosts (Olea 200lb, Ursfia and Tella 2001). streams flowing in a north-south direction. When this Avian communal roosting is thought to provide surviv- study was conducted, the fields had been partially har- al benefits (Blanco and Tella 1999) in terms of decreased vestedand the area wasmostly dominated by stubbleand predation risk and increasedforaging efficiency (Beau- fallow fields (74% of the area). The area is also crossed champ 1999). Some evidence also suggeststhat such by a recently-built highway. Roost Site Detection. An effort was made to find all roosts could facilitate social relationshipswith implica- roostsin the study area. We looked for roost sitesbetween nons in the regulation and viability of bird populations 20-26 August 1998. In 2001 and 2002 the search for (Blanco et al. 1993, Blanco and Tella 1999) and in the roostsbegan on 25 July and 1 July, respectively.We first conservation of some raptor species (Donfizar et al. checked the roosts used by Lesser Kestrel in previous 1996). To manage and conservea bird population, it is years. Next, we searched for new roostsin the area until necessaryto understand the requirements and the lim- the end of the summer. Several observers with binoculars rang factors of a population throughout the annual pe- (8 X) and telescope (20-60 X) were distributedacross the riod (Steenhoff et al. 1984). For Lesser Kestrels, infor- area at the best vantage points. Roostswere located dur- mation on summer communal roostsis incomplete. The ing late afternoon, becauseduring this period the Lesser specieshas been studied during winter in Spanishareas Kestrelstended to gather around "staging" areas near (Negro et al. 1991) as well as in Africa (austral summer; roosts. These aggregationswere fairly conspicuousand easy to detect. From these sites, kestrels progressively Siegfried and Skead 1971, Kopij 2002). To our knowl- moved toward the roost sites about an hour before sun- edge, the only published information on LesserKestrel set. Monitoring of RoostingBirds. We countedbirds at the roostsbetween 26 August and 26 September in 1998 (ev- E-mail address:[email protected] ery 3-7 d, 7 monitoring d), between 25 July and 10 Oc- SEPTEMBER 2004 SHORT COMMUNICATIONS 279 Table 1. Characteristicsof the premigratorycommunal roostsof LesserKestrels during summersof 1998, 2001, and 2002 in northwestern Spain. Study periods: 26 August-8 Septer•nber 1998, 25 July-10 October 2001, and 1 July-9 October 2002. ROOSTS C E BLACK SEVERAI, A B LOCUSTS AND D UTILITY F G EVERGREEN PYLONS COMMON POPLAR POI.ES PYLON POPLAR SUBSTRATE O^K (380 kV) PF•R PL2•NT^TION (<45 kV) (380 kV) PL2•N'I^TION Distance to power- 30m (45 0m (380 20m (<45 40m (380 0 (<45kV) 0 (380kV) 40m (<45 line (kV) in m kv) kV) kV) kV) kV) Max. no. of kestrels 1998 355 (29 Au- 262 (31 Au- 96 (2 Sep- 106 (8 Sep- gust) gust) tember) tember) 2001 200 (12 Au- 750 (31 Au- 27 (31 Au- 7 (6 Sep- 85 (18 Sep- gust) gust) gust) tember) tember) 2002 360 (4 Sep- 641-666 (2 tember) August) tober in 2001 (every 1-17 d, 14 monitoring d) and be- RESULTS tween 1 July and 9 October in 2002 (every 1-11 d, 23 Description of the Roost Sites. In 1998, four commu- monitoring d). In 1998 and 2001, we began to count nal roosts of Lesser Kestrels were found. Roost site A was birds at roostson 26 Augustand 25July,respectively, after the LesserKestrels began using roosts.For this reason, located in a small evergreen oak forest, where LesserKes- in 2002 we initiated monitoring severalweeks before the trels used a group of five scatteredevergreen oaks 5-8 m previous years to determine accurately the date when high, and close to a road (<150 m; see Table 1). Roost Lesser Kestrels started to use the roosts. B, waslocated on three utility pylons (380 kV). More than Except in 1998, we tried to censusLesser Kestrels using 90% of the individualsroosted on a singlepylon. Roost all roosts simultaneously.Only one roost was monitored C was on three trees (two black locusts[Robinia pseudoa- in 1998; in 2001 and 2002, the two largest and more- cacia]and common pear [Pyruscommunis]) 6-7 m high stableroosts were monitored during the entire studype- Roost D was in a poplar (Populussp.) plantation of 612 riod. Observationsites were at ground level and 300-600 m awayfrom the roosts.Individuals were counted using m2 with poplars20-28 m high. binocularsand 20-60X spottingscopes. To avoid bias in All four roostswere in the vicinity of powerlines (<50 the counting between observers,the same personsmon- m), ranging from <45-380 kV (Table 1), where the birds itored the same roosts.Although this does not eliminate grouped, rested and preened before going to the roost. potential biasesamong roosts,counts of Lesser Kestrels The shortest distance between roost sites was 6 km (be- in the roostswere facilitated by their behavior. Specifi- tween roostsA and C), and the longest, 17 km (between cally, before roosting, kestrels perched on powerlines, roosts A and D). where time was spent preening, or bathed in the sand. In 2001, we located four roosts, two of them were the Thus, we believe that the census of kestrels was reliable. same as in 1998 (A and D). The two new roosts were on We counted individuals going to roosts in 2.5-5 min intervals from 30 min before sunset until all individuals electric poles (E on several <45 kV poles and F on 380 entered the roost. The highest evening count was used kV poles). In 2002, LesserKestrels used roostsA and D, as the roost count for that day. In roost A, we censused but in roost A the birds used a larger number of ever- LesserKestrels when perching on powerlinesclose to the green oaks in the forest than in 1998 (see above). smalland scatteredgroup of evergreenoak trees (Quercus Change of Roost Siteswithin the Year. On 10-12 Sep- ilex), which they used as roost sites (Table 1). At this tember 1998 some changesin the roost siteswere ob- roost, birds flew as singles,in pairs, or in small groups served:kestrels stopped using roost B and roost A was from the wires to the trees, but not in the opposite di- moved 0.5-1.5 km from the previousroost location. On rection. This made counting kestrelsrelatively easy. 17-18 September2001, the kestrelsin roost D split into In 2002, we could not obtain accurate counts of Lesser Kestrelsentering one of the roosts on severaloccasions three groups, one stayedin roost D, another moved to (6 of 23 censuses)because roosting activitywas erratic. roost G, and the third group of kestrelsleft and were not Therefore, kestrelswere counted immediatelyfollowing located. In 2002, the kestrelsin roost D split temporally sunrise leaving the roost (from 30 min before sunrise (between27 and 29 August) in three groups,using two until after all the kestrels left the roost). other nearby plantations of poplars (not included in Ta- 280 SHORT COMMUNICATIONS VOL. 38, NO. 3 and 200 birds in roost A (12 August). The number of LesserKestrels decreased markedly after 7 Septemberin • Roost D both roosts.