UNUSUAL PREDATORY AND CACHING BEHAVIOR OF AMERICAN KESTRELS IN CENTRAL MISSOURI BRIAN TOLAND ABSTRACT- Caching behavior of the AmericanKestrel (Falco sparver/us) was studied 1981-1983 in BooneCounty, Missouri.Both wild prey and quarry thrown from car windowswere cached.Kestrels cached food 116 timesand retrievedit 77.5%of thetime. Males cached food in elevatedsites (at least 4 m high)64% of thetime while females did so only20%. During spring and summer, 93% of the preyitems were cached uneaten. During fall andwinter, only 42% of the foodcached was uneaten. When a surplusof preywas created by releasingseveral mice at a time,kestrels killed them whileflying to theircache sites. These prey items were stored in the samecache site. Apparently, caching behavior of AmericanKestrels is not directlycorrelated with the lengthof time betweenfeedings, and cachingbehavior operates independentlyof fooddeprivation, especially in springand summer. Many reptorshave been observed caching prey. cache,weather conditions, time of day, and durationof caching Mueller(1974) provided a reviewof foodstoring in sequence.Additional live prey was thrown from my carwindow to kestrelsperched within 25 m of roads.The prey thrown were severalcaptive species.Among the falconiforms, brown,gray, white and blackHouse Mouse (Mus musculus) and preycaching seems to be mostdeveloped and wide- HouseSparrows (Passer domesticus) with several primaries pulled to spreadin falcons.Published records of cachingin- ensuretheir captureby kestrels. clude those for the Merlin (Falco columbarius) RESULTS AND DISCUSSION (Greaves1968; Oliphant and Thompson 1976; Pitcheret al. 1982),American Kestrel (F. sparverius) During the 2 yr study 1210 h of observationof (Pierce 1937; Tordoff 1955; Roest 1957; Stendell kestrels were made and 30 kestrels were seen cach- and Waian 1968; Balgooyen1976; Collopy1977), ing prey a total of 116 times. They subsequently Prairie Falcon (F. mexicanus)(Peterson and Sitter retrievedfood successfully77.5% of the time. Prey 1975; Oliphant and Thompson 1976), Peregrine cached were 95% rodents and 5% birds. Both wild Falcon (F. peregrinus)(Beebe 1960; Brown and and providedprey were cached.Kestrels captured Amadon 1968; Nelson1970; Cade 1982), Gryfal- 95% of the prey thrown from car windowsand con (F. rusticolus)(Jenkins 1978; Cade 1982) and cached46 (48%). The remaining 70 (60%) prey many others. cached consisted of 55 Prairie Vole (Microtusoc- hrogaster),6 housemice, 3 White-footed Mouse STUDY AREA AND METHODS (Peromyscusleucopus) 2 Western Harvest Mouse Data reportedhere are from a 48 km: area in BooneCounty, (Reithrodontomysmegalotis), 2 House Sparrows,1 Missouri,where farmlandsare interspersedwith woodlots,old Grasshopper Sparrow (Ammodramussavannarum) fields, meadows and residential areas. I observedkestrels from September1981 through August 1983 and 1 EasternMeadowlark (Sturnella magna). usinga 30x spottingscope and 9x binocularsat distancesof 5-200 Cache sites were of 8 kinds (Table 1). Males m. For eachobservation I recordedspecies cached, location of cachedprey in elevatedsites significantly more Table 1. Cachesites used by AmericanKestrels in BooneCounty, Missouri1981-83. LOCATION AND HEIGHT OF CACHES Hollow Grass railroad Tree Fence Building Tree limbs Tops of clumps ties roots Bushes posts gutters and holes power poles Sex (0-0.1 m) (0-0.1 m) (0-0.1 m) (0.5-1.0 m) (1.0 m) (4.0 m) ( 4.5 m) (10.0-20.0 m) Total M 16 3 1 1 6 2 44 3 76 F 27 0 0 2 3 0 8 0 40 107 RAPTORRESEARCH 18(3):107-110 108 BRIANTOLAND VOL. 18, NO. 3 often than did females(Table 1). Malescached prey whichcached prey used objects nearby to memorize 4 m or more abovethe ground 64% of the time, the location of the cache site. whilefemales d•d soonly 20% (X2 = 22.16, P<0.01, When retrievingprey, kestrelsflew to the cache df= 1). siteand if unableto find storedprey, hoveredabove During the breedingseason kestrels have special the area or walked to adjacent grass clumps to courtshipfeeding ceremonies(Fox 1979; Cade search. In several instanceskestrels appeared to 1982). Foodtransfer often beginsas remotefood give up their searchwhen they flew to a nearby passing(Nelson 1977)when the male deliberately perch,only to return and resumethe search.One cachesprey within view of the female. When he female kestrel spent 15 min investigatinggrass leaves,she fliesto the cache,retrieves a0d eatsthe clumpsboth on foot and in brief hoversbefore food. I observedthat all copulationand courtship givingup. activitiesincluding hitched wing-displays,food Becausemy fieldobservations were evenly distri- begging,courtship feeding and remote food pas- butedthroughout the day, I assumedthat there was singoccurred at elevatedsites, on or near favorite no differencein the probabilityof observingeither .pluckingor huntingperches. Because males alone preystoring or retrieval(see Collopy 1977). On this cacheprey at theseelevated sites during courtship assumption,I consideredmy cachingdata asa rep- theymay be predisposedto cachein elevatedsites resentativesample of the relativefrequency of prey during the rest of the year. cachingand retrieving and calculateda recovery Partiallyeaten carcasses were alwaysdecapitated efficiencyof about 78%, similar to the 70% re- beforebeing cached and only the posterior2/3 to ported by Collopy(1977). 1/2 of the body of the prey wasplaced in the cache. I observedthat presentingkestrels with live prey However, kestrels cached 78% (36 of 46) of the stimulateda responsethat simulatesthe reactions presentedprey and 69% (48 of 70) of wild prey of kestrelsto natural prey. The typicalresponse was completely uneaten. Of all prey items cached, to fly from the perchtoward the prey within20 sec only28% (32 of 116) were decapitated.In contrast of its detection. Kestrels then would either (1) bind Stendelland Waian (1968) reported that 14 of 15 to the quarryon the groundand kill it with a biteto small mammalscached by a singlefemale kestrel the neck immediately (or after having flown to a were decapitatedand Collopy(1977) reportedthat nearby perch) or, (2) snatch the prey from the 10 female kestrelsdecapitated 13 of 17 (76.5%) groundwithout landing and fly to a nearbyperch to small animals cached. In Missouri kestrels cached dispatchit with a bite to the neck. more (58%, 27 of 48) partly-eatenprey during the On 6 March 1982 at 1430H I observed a female fall and winter than the 7% (5 of 68) during spring through binocularsfrom a distanceof 100 m. The and summer,which may be explained by the lackof weatherwas 38øC, calm and clear. I approachedin hunger during the breedingseason due to greater my car to within 25 m, threw a white mouseout the availabilityof food and warmer temperatures. window and waited. Within 2 min the kestrelap- Kestrelshiding food approachedthe cachesite proachedwithin 4 m of my car, hovered,and then furtively, then thrust, pushed,or nudged the prey retreated to a wire 20 m away. I then threw out 2 into positionwith the beak (see Balgooyen1976; more miceand backedthe car 25 m away.At 1440H Collopy 1977). Sometimessites apparently were the kestrel again flew toward the mice but after selectedbefore the flight started,and kestrelsflew hoveringabove them and lookingat my car, again directlyto the cachespot. Prey was not placedin any retreated to the wire only 15 m distance.I then preferentialposition such as belly-down(as if to presented4 morewhite micefor a total of 7, all of take advantageof the prey'sprotective coloration) whichwere conspicuousagainst a recentlymowed as reported by Balgooyen(1976) and Collopy lawn. At 1445 H the kestrel flew to a wire only 5 m (1977). I found itemslying on backor sidesas often from the miceand after hesitatingfor 15 or 20 sec ason the belly.Kestrels sometimes did makeseveral flew down and captured a mouse. However, she attemptsat repositioningprey until it was better immediatelyflew west75 m during which flight I concealed.In all caseswhen prey wasstored on the saw her bend over several times in midair to bite the groundin grassclumps, kestrels chose sites next to neckof the mouse.She immediately landed on the fenceposts, utility poles,sign posts, or other mar- groundand cachedthe prey in a grassclump at the kers.Tordoff (1955)observed that a captivekestrel baseof a fence post. She quickly returned to cap- FALL 1984 AMERICANKESTREL CACHING BEHAVIOR 109 ture and dispatchin flight the remaining mice in attemptingto hurriedlyexploit suddensurpluses in rapid succession.All 7 micewere cachedin 1 or 2 prey availability.Because I sawkestrels caching 7 grassclumps 1 m apart. None of the 7 micewere prey items together in a 5-min period, I cannot eaten at this time. supportMueller's (1973) statementthat "excessive A month later at the sametime of daya maletook killing resultedfrom the falcon"forgetting" that it 7 micein the samefashion, killing them midair asit had cachedfood when it wasexposed to the prey flew to the cache site in a white oak tree (Quercus stimulus."Kestrels cache several consecutive prey alba). itemsin the samespot in a periodof minutesor I wasable to elicitthe capture and cachingof as days,and later retrievethem (Stendelland Waian many as 10 mice in sequenceby both captiveand 1968). wildkestrels when presenting them with preyone at ACKNOWLEDGMENTS a time, over 2 to 8 h periods.Nunn et al. (1976) reported that 1 wild female took 20 white mice I appreciatethe guidanceof William H. Elder. Tim Haithcoat and DaveScarbrough provided valuable field assistance.Thomas thrown from a car window one at a time, over an S. Baskettand CurticeGriffin