How do cuckoos find their hosts? The role of habitat imprinting

ANIMAL BEHAVIOUR, 1998, 56, 1425 1433 Article No. r980931 How do cuckoos find their hosts? The role of hbitt imprinting YVONNE TEUSCHL, BARBARA TABORSKY & MICHAEL TABORSKY Konrd Lorenz-Institut für Vergleichende Verhltensforschung (KLIVV) (Received 20 June 1997; initil cceptnce 26 November 1997; finl cceptnce 10 June 1998; MS. number: 5663) ABSTRACT Although number of hypotheses hve been proposed for how Europen cuckoo, Cuculus cnorus, femles my find hosts belonging to their foster prents species, cler evidence is lcking for ny of them. Here, we propose hbitt imprinting s n lterntive mechnism for host selection nd provide evidence tht cuckoos re ble to remember cquired informtion bout fmilir hbitt. We hnd-rered seven cuckoos in one of five different rtificil hbitts nd tested them s dults in hbitt choice experiments. In ech test hbitt, pir of zebr finches, Teniopygi guttut, ws presented s hosts. We tested cuckoos of both sexes becuse the genotype of mles my influence egg colour, nd therefore egg mimicry; lterntively, hbitt imprinting my be generl mechnism existing in both sexes but ffecting egg mimicry only vi femles. Test cuckoos spent significntly more time looking t their respective fmilir hbitts thn t other hbitts in 1 of 2 test yers. How long cuckoos were rered in the rtificil hbitts correlted positively with how long they spent in this hbitt during the choice experiments. Additionlly, test cuckoos remined longer with zebr finches tht showed more nestbuilding behviour but hd lower levels of generl ctivity, nd they lso observed these hosts more frequently. If cuckoos choose to breed in hbitts resembling those on which they were imprinted nd serch rndomly for hosts in these hbitts, they would increse their probbility of prsitizing nests of their foster species. We propose tht host specificity would be strengthened, however, if cuckoos use sequence of severl mechnisms, rther thn just one, to find their hosts. 1998 The Assocition for the Study of Animl Behviour Adpttions nd counterdpttions of vin brood prsites nd their hosts re importnt prdigms of coevolution (e.g. Rothstein 1990). Hosts hve developed different mens of egg rejection s dpttion ginst prsitism (e.g. Dvies & Brooke 1988). The egg mimicry shown by some brood prsites is counterdpttion to this behviour (Brooke & Dvies 1988; Rothstein 1990). Egg mimicry in Europen cuckoos, Cuculus cnorus, is prticulrly surprising becuse this brood prsite is generlist using wide rnge of host species. Cuckoo eggs hve been found in nests of more thn 100 different species. Eleven min host species nd similr number of secondry ones hve been identified (Wyllie 1981). Egg colours nd ptterns differ gretly between these host species, nd the cuckoo eggs vry ccordingly (Brooke & Dvies 1988; Moksnes & Røskft 1995). Field observtions suggest tht ech femle cuckoo prsitizes only one host species nd lys eggs of only one type (reviewed in Wyllie 1981; Dröscher 1988). Four hypotheses hve been proposed to explin how cuckoo femle finds the nests of her host species: (1) inherited Correspondence: B. Tborsky, Konrd Lorenz-Institut für Vergleichende Verhltensforschung, Svoyenstr. 1, A-1160 Vienn, Austri. (e-mil: b.tborsky@klivv.oew.c.t). preferences (Newton 1893); (2) host imprinting (cuckoo chicks re imprinted on their foster species; e.g. Lck 1968); (3) ntl philoptry (cuckoos return to where they were born nd choose host rndomly; Brooke & Dvies 1991); (4) nest site choice (cuckoo femles choose group of host species with similr eggs nd nest sites nd serch for nests t rndom within these groups; Moksnes & Røskft 1995). In n experimentl study, Brooke & Dvies (1991) found no evidence supporting hypotheses 1 or 2. Hypothesis 3 hs been supported by observtions of philoptry in cuckoos (Seel 1977; Wyllie 1981). In fvour of hypothesis 4, Moksnes & Røskft (1995) found tht 77% of ll cuckoo eggs were found either in nests of the host species with mtching egg type, or of species with similr nest sites. However, hypothesis 4 does not propose mechnism for how femles should find groups of host species with similr eggs nd nest sites, nd is hence not fully comprble to the other three hypotheses. Here we propose hbitt imprinting (hyposthesis 5) s mechnism in Europen cuckoos for finding n pproprite host. We ssume tht cuckoo chicks lern the chrcteristics of their hbitt in which they grow up by n imprinting-like process. We further ssume tht such chrcteristics of the ntl hbitt re the predominnt 0003 3472/98/121425+09 $30.00/0 1425 1998 The Assocition for the Study of Animl Behviour

1426 ANIMAL BEHAVIOUR, 56, 6 vegettion type nd structure in the re round their nest such s reed beds, medows or open bushlnd. For reproduction, we propose tht cuckoos estblish their home rnges in res consisting minly of hbitt tht resembles the hbitt they hd experienced s nestlings. When they prsitize the hosts vilble in this hbitt, there is n enhnced chnce of encountering nests of the host species by which they were rered. Hypotheses 1 nd 2 would llow for strong host specificity nd hence good egg mimicry. In contrst, hypotheses 3 5 would predict tht the levels of host specificity nd egg mimicry should be fr from perfect, especilly in heterogeneous environments where different host species live in close proximity. In ccordnce with this expecttion, Moksnes & Røskft (1995) found only moderte mimicry with only 30% of the eggs clssified s good or perfect mimics in study of 12 000 prsitized clutches of Europen psserines contined in museum collections. To test whether hbitt imprinting my be involved in the nest selection process, we performed preference experiments in lrge indoor viry with cuckoos tht hd been hnd rered in vrious, rtificil hbitts. We tested both mle nd femle cuckoos, s the genotype of both sexes my influence the egg colour of dughters. In this cse, egg mimicry could evolve only if cuckoos mte ssorttively with mles crrying genes for the sme egg type. Hbitt imprinting hs been proposed s mechnism tht my led to ssorttive mting (Southern 1954; Wyllie 1981). Alterntively, hbitt imprinting my be generl mechnism in cuckoos perhps serving different purpose, but ffecting egg mimicry only vi femles. METHODS Experimentl Rering Conditions We collected seven cuckoo chicks in their nests (three femles nd two mles in 1992, two femles in 1993) from reed wrbler, Acrocephlus scirpceus, nd mrsh wrbler, A. plustris, hosts 6 8 dys fter htching (i.e. before their eyes opened). They were hnd-rered for periods of 28 53 dys in one of five rtificil hbitts (Fig. 1): boxes (35 40 35 cm) lined with crdbord of highly contrsting colours on which were mounted structures tht differed in shpe, colour nd mteril. The chicks remined in their ntl nests which were mounted on flowerpot in the middle of the box, llowing them to see the surrounding structures. The following hbitts were used: pink plstic blls on white (hbitt H O/H ); blue ribbons on yellow (H L/P ); pine cones on red (H N ); commercil wicker nests s used for breeding zebr finches, Teniopygi guttt, on green (H F ); nd nturl spruce brnches on violet bckground (H I ). The strong differences were chosen to fcilitte the cuckoos recognition of specific hbitts, even if they perceive their environment using different criteri from ours. Furthermore, we imed to void structures tht might hve been prt of n innte preference mechnism of the cuckoos. After the chicks fledged, we trnsferred them to cges (0.7 0.7 1.20 m) contining the sme structures nd coloured crdbord s before nd some brnches nd twigs to perch on. The crdbord ws mounted on three sides nd on the bottom of the cges, nd the structures were distributed ll over the cges, hnging from the top, ttched to the sides nd to the twigs. The young cuckoos were kept in these cges until independence. The cuckoos were exposed for different periods to the rtificil hbitts (see Tble 1), s the ge t independence vried between individuls. Then cuckoos were moved to indoor viries (17 m 3 ) contining only twigs nd brnches for perching nd vrious ground covers, but lcking the rtificil hbitt structures tht hd been used during imprinting. In summer, the birds could move freely between the indoor nd outdoor viries (12 m 3 ) which were surrounded by trees nd buildings. Here the cuckoos were kept until the experiments, between the different test series nd between the yers of experiments. After the experiments the birds were kept in cptivity for further behviourl studies, since they were probbly unble to migrte or survive successfully in the wild. Choice Tests The seven cuckoos were tested for their hbitt preferences t the ge of 1 yer. Cuckoos re known to reproduce in their first yer (Seel et l. 1981; Dröscher 1988), but some my not strt until 2 yers of ge (Glutz von Blotzheim & Buer 1980). Therefore, we tested the five cuckoos rered in 1992 gin in 1994. The choice experiments were performed in n experimentl hll of 10 9 10 m with Plexigls ceiling providing nturl dylight from bove. The five rering hbitts nd one hbitt mde of reed stems representing the ntl hbitt of ll experimentl birds were presented simultneously to ech cuckoo (Fig. 1). Hbitts were rrnged in circle round neutrl zone in the centre providing food nd wter on the ground, three perches (one 3 m high nd two 2 m high) nd cge in which the cuckoos were initilly hbituted to the experimentl set-up (see Fig. 1). When sitting on the poles or nywhere else in the neutrl zone the cuckoos could not see ny of the experimentl hbitts (ech 2 2 1.4 m), which were seprted from ech other nd from the centrl re by opque screens 2.6 m high. During their egg-lying phse, cuckoo femles re known to sit in trees for extensive periods while wtching potentil hosts nd nests (e.g. Wyllie 1981). The experimentl set-up llowed similr behviour to occur nd for us to use this s choice criterion: in front of ech hbitt we plced perch 3 m high, llowing the cuckoo to wtch the hbitt from bove. A bird sitting on this perch could see only this hbitt nd the other perches, but not ny other hbitts. Ech hbitt contined n rtificil, open cup-shped nest with one zebr finch egg nd cge with pir of zebr finches, T. g. cstnotis, to represent potentil hosts. We chose zebr finches to exclude the possibility tht n innte host preference my superimpose on potentil hbitt preferences nd becuse this species shows relible nd frequent nest-building behviour in smll cges. Also, zebr finches re not disturbed

TEUSCHL ET AL.: HABITAT IMPRINTING IN CUCKOOS 1427 Tble 1. Totl time cuckoos spent in different hbitts in the three test series, excluding sunshine hours Dte Dys of tril Bird Sex Age (yers) Rering hbitt Dys in rering hbitt Time spent in hbitt (min) H I H F H L/P H O/H H N Reed Time in centre (%) 1993 2 9 August 6 I F 1 H I 42 581 240 192 95 136 254 44.7 28 June 5 July 7 F F 1 H F 49 103 169 144 208 37 155 75.2 21 27 July 6 L M 1 H L/P 49 214 184 545 343 159 277 36.3 11 18 July 7 O M 1 H O/H 53 234 121 80 523 60 71 66.6 10 16 August 6 N F 1 H N 34 60 2 45 5 11 21 95.0 1994/1 7 11 My 4 I F 2 H I 42 552 89 137 247 138 74 39.8 23 27 My 4 F F 2 H F 49 180 97 131 635 100 103 39.6 2 6 My 4 L M 2 H L/P 49 194 67 46 72 241 176 58.6 18 23 My 4 O M 2 H O/H 53 97 176 382 164 153 110 52.3 13 17 My 4 N F 2 H N 34 89 7 5 13 15 131 88.1 29 My 2 June 4 H F 1 H O/H 28 174 38 76 71 82 72 75.4 3 7 June 4 P F 1 H L/P 28 488 405 187 156 247 291 23.8 1994/2 19 23 June 4 I F 2 H I 42 310 279 273 597 165 247 19.1 7 11 July 4 F F 2 H F 49 3 4 9 82 0 15 94.1 13 17 June 4 L M 2 H L/P 49 146 241 50 199 167 123 52.3 2 5 July 4 O M 2 H O/H 53 8 4 99 6 6 5 94.1 26 30 June 4 N F 2 H N 34 5 8 67 104 10 21 88.8 12 16 July 4 H F 1 H O/H 28 1 97 40 5 4 66 89.5 17 21 July 4 P F 1 H L/P 28 97 195 214 182 113 235 50.1 Bold nd underlined numbers represent the fmilir nd the most frequently chosen hbitts, respectively.

1428 ANIMAL BEHAVIOUR, 56, 6 Nest with egg Reed Zebr finch cge H N Door b Cge for ctching nd hbitution H O/H Wter Food H I H F H L/P morning, 1 h t noon, 2 h in the evening; 1994: 3 h strting from sunrise, 2 h in the evening). In 1994, we lso recorded hed nd body postures of test cuckoos. Hed postures recorded were either inclined to the left or right, looking up or down or turned to the left or right, wheres body posture ws described s lert, norml, or with ruffled fethers. Postures nd loction were recorded once every 30 s. We recorded the behviour of ech pir of zebr finches for 5 min twice dy directly before or fter the cuckoo observtion period to be ble to check for possible influence of zebr finch behviour on the test cuckoo. Mle nd femle pir members were observed in turn, ech continuously for 30 s ech. As mesure of ctivity, we noted the number of chnges in position. We quntified nest-building behviour s the number of times zebr finch entered the nest with nesting mteril in its bill. by the presence of n observnt cuckoo, probbly becuse they lck common evolutionry history. We strted the experiments when the cuckoos begn clling, tht is, were reproductively motivted. They were tested seprtely one fter nother. A bird ws trnsferred from the holding viries into the cge in the neutrl zone of the experimentl hll in the erly fternoon nd relesed fter 3 4 h of hbitution. Recordings strted t dusk the next morning. Tble 1 gives the dtes nd durtion of the test series. In 1993, the cges with zebr finches were rotted between trils. In 1994, we did not chnge the experimentl set-up between trils with different test birds, but did so between the two test series involving the sme birds. The positions of the hbitts nd those of zebr finch cges were chnged independently of ech other to llow both position effects nd individul host preferences to be tested. After ech individul test ny eggs lid by the zebr finches were removed to ensure tht the ltter remined in the lying phse. Behviourl Recordings The cuckoo s position ws continuously recorded from bove by video cmer mounted in the ceiling of the experimentl hll. We recorded ctivity by direct observtion through one-wy mirrors. The observtion periods corresponded to periods when egg lying nd mle femle interctions were most common in wild cuckoos (Wyllie 1981; personl observtion; 1993: 2 h in the c 1 m Figure 1. Experimentl set-up: six test hbitts (2 2 m nd 1.4 m high) rrnged round neutrl re in the centre; : perches 3 m high; b: perches 2 m high; c: opque screens 2.6 m high; different symbols in the hbitts represent the structures used. Anlyses From the continuous video recordings, we nlysed the positions of test cuckoos for the totl time between the first chnge of position t dwn nd the lst chnge t dusk. The time spent on the night roosts ws not included in the nlyses. For the hbitt preference tests we used only the dt from 1993 nd those of the first test series in 1994, becuse in the second test series in 1994 four of seven cuckoos spent most of their time in the neutrl re (>88%; see Tble 1). This coincided with the onset of moult nd reduced clling ctivity in severl birds, suggesting tht these birds were in nonreproductive stge. In experiments 1993 nd 1994/1 one very inctive bird lso spent most of its time in the neutrl re (95% nd 88%, respectively; bird N in Tble 1). It ws omitted from the hbitt preference tests, becuse it ws doubtful whether its occsionl presence in one of the test hbitts would give relible preference mesure. The hbitt preference tests were nlysed seprtely for the yers 1993 nd 1994 becuse of differences in the experimentl procedure. The cuckoos tested in 1993 were hnd rered for longer periods thn those tested in 1994, which ment longer exposure to the rering hbitt. In 1993, the experimentl hbitts were rerrnged rndomly between subsequent tests, wheres in 1994 they were rerrnged only once between the two test series with the sme birds. Additionlly, the birds were exposed to externl stimuli tht vried between yers, including sunlight (see the significnt preference for sunlit perches reported below), nd coustic disturbnce by nthropogenic ctivities in the vicinity of the experimentl hll. We used both test series of 1994 to test for the effects of hbitt position nd sunlit perches. An nlysis restricted to three birds tht spent more thn 40% of their time in the experimentl hbitts in both runs of 1994 reveled no preferences for specific positions in the experimentl hll (Ludwig s (1962) exct subgroup tu test, one tiled: K=3, N=6, NS). However, the presence of sunlit perches significntly influenced the hbitt choice of test birds (binomil test: N=12 experiments involving five birds,

TEUSCHL ET AL.: HABITAT IMPRINTING IN CUCKOOS 1429 50 45 40 35 30 25 20 15 10 () * * continuously on perch. In short stys the probbility of observing ny specific behviour ws low compred with longer stys. Therefore, we used only observtions mde in long stys (>3 min) to check for reltionships between cuckoo nd zebr finch behviour. The men occurrence of potentil ttention towrds zebr finches in the different test cuckoos rnged from 39 to 62% of time during long stys. For ech individul cuckoo test period, we clculted the medin zebr finch ctivity of ll 5-min records mde during this time. As nest-building behviour of zebr finches ws observed only rrely, we rnked the six zebr finch pirs ccording to the totl number of observed nest-building behviours for ech of the two independent test series used for the hbitt preference tests (1993, 1994/1). Time in hbitt (%) 5 0 20 18 16 14 12 10 8 6 4 2 0 (b) Fmilir Fmilir Ntl Ntl Hbitt type Others Others Figure 2. Time spent in the three hbitt ctegories () in 1993 nd (b) in 1994 (medins±qurtiles). Fmilir is the hbitt in which the cuckoos were rered; ntl is the nturl reed hbitt from which the cuckoos were tken; others is for ech bird the men of the other four hbitts in the test. *P<0.05, see text. P<0.001). Therefore, we excluded from further nlysis ll hours of the test dys when direct sunlight ws shining into the experimentl hll, tht is, between 0945 nd 1600. From the behviourl recordings of cuckoo postures we clssified different sttes. Here only the stte potentil ttention towrds zebr finches is relevnt. This combines ll observtions in which hed postures llowed the test cuckoo to look t the zebr finch cge with one eye. We clculted the reltive frequency of this stte for ech sty, one sty being the time cuckoo spent Sttistics We used nonprmetric sttistics throughout, s the dt differed significntly from norml distributions or the smple sizes were too smll to test for devitions from norml distributions. We used two-tiled tests unless otherwise stted. We tested for hbitt preferences by Friedmn two-wy ANOVA by rnks. The rering (=fmilir) hbitt is different for ech cuckoo; therefore we tested for preferences of one of the three ctegories fmilir, ntl nd others, others being the men proportion of time spent in the four hbitts other thn the rering nd the reed hbitt. For multiple comprisons between the single hbitt types in 1993, we used n extension of the Qude test, two-wy ANOVA by rnks which gives more weight to nimls tht mke clerer choice between hbitts (Conover 1980). For multiple comprisons, the Qude test hs smller type I nd lrger type II errors thn the Friedmn test nd is therefore more conservtive (Aldredge & Rtti 1986). For correltion nlyses we used prtly subgroup tu correltions: Torgerson s subgroup tu test (TSTT; Torgeson 1959; Lienert 1986), or for smller dt sets, Ludwig s exct subgroup tu test (LESTT; Ludwig 1962; Lienert 1986). These tests llowed us to clculte n overll reltionship between the tested vribles for ll cuckoos, using severl subgroups of the dt set. RESULTS Preference for Rering Hbitt The time spent in prticulr hbitt ws used s n indictor of hbitt preference. In the 1993 test series, the test cuckoos styed significntly longer in their respective fmilir hbitts thn expected by chnce (Friedmn two-wy ANOVA: K=3, N=4 (two mles, two femles), P<0.05; Fig. 2, Tble 1; multiple comprisons Qude test, fmilir other: P<0.05; fmilir ntl: P<0.05; other ntl: NS). No preference ws detected in test series 1 of 1994 (Friedmn two-wy ANOVA: K=3, N=6 (two mles, four femles), NS; Fig. 2b); only one out of six cuckoos spent most time in its fmilir hbitt (Tble 1).

1430 ANIMAL BEHAVIOUR, 56, 6 Time in rering hbitt during experiment (%) 50 40 30 20 10 0 25 H P 30 35 40 45 50 Period in rering hbitt during ontogeny (dys) Durtion of Imprinting Period nd Hbitt Preferences One importnt chrcteristic of imprinting is the existence of sensitive period, restricted period during which the niml estblishes n ttchment to the imprinting object (reviewed in Bolhuis 1991). If cuckoos lern their ntl hbitt by such mechnism the timing or the totl length of exposure to the respective hbitts should influence their lter hbitt preference. In our experiments the ges of independence nd hence the lengths of the imprinting period vried between cuckoos. Cuckoos tht hd been exposed to their fmilir hbitt for longer periods spent significntly more time in this hbitt during the experiment when tested t 1 yer of ge (Spermn rnk correltion nlysis, one tiled: r S =0.76, N=7, P<0.02; Fig. 3). The dt do not llow us to distinguish between n effect of the length of exposure nd possible limited sensitive period occurring only very lte in the rering period. Influence of Activity of Hosts N Figure 3. Reltionship between the length of the imprinting period nd hbitt choice. The letters stnd for the individul cuckoos. Cuckoos my prefer nests with low host ctivity, s this my indicte tht nest is currently unttended. Alterntively, cuckoos my observe nests with very ctive hosts more often becuse these re esier to detect. In 1993 nd 1994, there ws no significnt correltion between medin zebr finch ctivity nd the proportion of time cuckoo spent in the respective hbitts if ll cuckoos were combined (TSTT: 1993: K=4, N=6, P=0.58; 1994/1: K=6, N=6, P=0.82). However, host observtion is minly importnt for cuckoo femles. In 1993, the two femles tested spent more time in hbitts with less ctive zebr finches (LESTT: K=2, N=6, P=0.05), while there ws no significnt reltionship in 1994 (TSTT: K=4, N=6, I L F O 55 P=0.78). Cuckoos looked reltively more often in the direction of less ctive zebr finches (TSTT: K=5, N=6, P=0.03). This correltion ws not significnt for femles lone (LESTT: K=3, N=6, NS). Influence of Nest Building by Hosts Field observtions suggest tht cuckoos my find nests for lter egg deposition during the nest-building stge (Chnce 1922, 1940). Altogether, test birds did not spend more time ner zebr finches tht showed more nestbuilding behviour (TSTT, one tiled, 1993: K=4, N=6, P=0.46; 1994/1: K=6, N=5, NS). Only femles in 1994 tended to spend more time in hbitts in which zebr finches performed nest building more often (TSTT, one tiled: K=4, N=6, P=0.056). Cuckoos looked more often towrds ctive nest-building zebr finches (TSTT, one tiled: K=5, N=5, P=0.05). Femles lone showed no significnt reltionship (LESTT, one tiled: K=3, N=5, NS). Reproductive Motivtion of Cuckoos In 1993, the two mles nd one of the femles were herd clling in their viry during the experiments, which suggests tht they were reproductively motivted. In 1994, ll 2-yer-old birds nd 1-yer-old femle clled in the viry nd in the experimentl hll. As the mjority of the clls in the hll were given in the centrl re, we could not test sttisticlly whether specific hbitts were preferred for clling. The tking of host eggs, nother potentil indictor of reproductive interest in cuckoos, occurred once (femle P, hbitt H F ). There ws positive correltion between the totl time spent in hbitt during the whole test period nd the proportion of time the test cuckoo observed the zebr finch pir in tht hbitt (TSTT: K=5, N=6, P=0.003; Fig. 4). When tested for femles lone, the correltion ws lso significnt (LESTT: K=3, N=6, P=0.001). This my suggest tht cuckoos spent more time in certin hbitts becuse of reproductive interest. DISCUSSION Evidence for Hbitt Imprinting The hbitt preferences found in 1993 indicte tht cuckoos were ble to remember the hbitt in which they hd been rered. Erly experience in prticulr environment determines subsequent hbitt preferences in severl bird species nd cn even reduce or chnge genetic preferences for specific hbitts (Klopfer 1963, 1965; Grünberger & Leisler 1990, 1993b). In our experiments, mles nd femles did not differ in their preferences for the fmilir hbitt, suggesting tht this my be generl mechnism in cuckoos s well. It my be responsible for the mintennce of egg mimicry only vi the femle sex, or vi both sexes if the genotypes of mles influence their dughters egg phenotypes.

TEUSCHL ET AL.: HABITAT IMPRINTING IN CUCKOOS 1431 Time looking towrds zebr finches (%) 90 80 70 60 50 40 30 20 10 0 10 20 30 40 Time in hbitt (%) Femle I Mle L Mle O Femle F Femle P Figure 4. Reltionship between the percentge of totl time cuckoos spent in hbitt during the 1994/1 test series nd the men percentge of time they spent looking in the direction of zebr finches in tht hbitt. When the cuckoos were tested gin in 1994, they did not spend most of the time in their respective, fmilir hbitts s in the previous yer. There re severl possible explntions for this discrepncy. (1) Experience my chnge hbitt preferences, especilly if the hbitt lerned erly in life is poorly structured (Hildén 1965; Glück 1984; Grünberger & Leisler 1990). (2) Hbitt preferences my need to be reinforced in the first yer to prevent forgetting of the lerned hbitt fetures. In severl bird species, reproductive success nd experience with loclity or hbitt in previous nesting ttempts cn influence hbitt choice (Srgent 1965; Ctchpole 1972) nd breeding site fidelity in subsequent nesting ttempts (e.g. Greenwood & Hrvey 1982; Oring et l. 1983; Bollinger & Gvin 1989). (3) The vrition in experimentl procedure between yers, especilly the time spent in the rering hbitt, might hve cused the birds to rect differently. This possibility is supported by the fct tht the two yerlings in 1994, which hd been rered for shorter periods in their experimentl hbitts thn ll other birds, did not prefer their fmilir hbitts. The positive correltion between the length of the imprinting period nd the subsequent preference for the fmilir hbitt (Fig. 3) supports the hypothesis of hbitt imprinting in cuckoos, but it my require longer period thn we llowed for the two chicks tested in 1994. This result my suggest tht, like other birds, wild cuckoo chicks lern the chrcteristics of their hbitt minly fter fledging (Srgent 1965; Ctchpole 1972; Glück 1984), which is fter 17 24 dys (Wyllie 1981). After fledging, young cuckoos remin dependent on their foster prents for bout 2 weeks but begin to explore the nest vicinity, whereby they my increse their knowledge of their ntl hbitt. 50 60 A preference for stying in hbitt does not necessrily men tht cuckoo femle would lso choose this hbitt for egg lying. However, in linnets, Acnthis cnnbin, preference for stying in n experimentl, fmilir hbitt correlted with preference for nesting there (Glück 1984). In our experiments, the cuckoos indicted reproductive motivtion: they clled nd one femle te n egg. Also, the cuckoos clerly rected to zebr finch behviour which suggests tht they regrded them s potentil hosts. The time spent in hbitt correlted positively with the proportion of time zebr finches were observed by the respective cuckoos. Cuckoos observed zebr finches more often if they showed more nest building nd less generl ctivity. Additionlly, in 1993 the two femles tested spent significntly more time in hbitts with less ctive zebr finches nd in 1994, femles tended to spend more time in hbitts with nest-building zebr finches. These results suggest tht (1) potentil host nests t which there is less generl ctivity re more ttrctive to cuckoos, perhps becuse this would indicte reduced host ttentiveness (prsitic eggs re more likely to be rejected if the host observes the cuckoo lying in its nest, e.g. Dvies & Brooke 1988), nd (2) femle cuckoos preferentilly observe nests in the nest-building stge probbly becuse this my fcilitte the optiml timing of prsitism. If cuckoos preferred to sty nd thereby gin informtion in hbitts tht pper fmilir to them, this would increse the chnce of detecting nd prsitizing hosts in those hbitts. Such preference could rise s result of neophobic voidnce of strnge hbitts, which hs been suggested to be n importnt component of hbitt selection (e.g. Grünberger & Leisler 1993). Linnets, when given the opportunity to see the fmilir hbitt, reduced their hert bet frequency (Gssmn 1991), suggesting tht these birds felt more secure. Comprison of Hypotheses The evolution of egg mimicry requires certin degree of host specificity. This is possible if cuckoo femles ly the mjority of eggs in nests of the host species tht rered them. Indeed, using rdiotelemetry, individul femles were found to ly eggs in nests of only one species; few eggs were lid in nests of congener (Wyllie 1981; Dröscher 1988; unpublished dt). Also, cuckoos seem to void host species tht re unsuitble for rering cuckoo chicks (Moksnes & Røskft 1995). The pros nd cons for ech potentil nest selection mechnism (except hbitt imprinting) tht my led to host specificity hve been discussed in detil by Brooke & Dvies (1991) nd Moksnes & Røskft (1995). However, the following brief comprison shows tht in ech cse, either supporting evidence for the mechnism is lcking or the mechnisms re not specific enough to led to host specificity, t lest in frgmented lndscpes. Philoptry Wyllie (1981) found only four out of 60 birds (6.7%) s dults in the sme re where they hd been mrked s

1432 ANIMAL BEHAVIOUR, 56, 6 nestlings. This proportion is very low, even if mortlity is ssumed to be high. Seel (1977) found 32% of cuckoos recovered s dults tht hd been ringed s nestlings within 20 km, nd 55% 21 200 km, from the nests in which they hd been rered (recoveries within the breeding seson from April to July only, N=31). While such rther imprecise philoptry my ccount for certin degree of host specificity in lrge res of uniform hbitts (Brooke & Dvies 1991), it is probbly not sufficient to mintin egg mimicry in tody s frgmented lndscpes. Hbitt imprinting Our study provides evidence for hbitt imprinting, even though it showed clerly in only 1 of 2 test yers. Our preliminry field dt of cuckoo femles fitted with rdiotrnsmitters re comptible with the hbittimprinting hypothesis but dt collection nd the nlyses hve not yet been completed. If hbitt imprinting were the only mechnism for finding host nests, cuckoos would serch rndomly for host nests in their ntl hbitt. In mny cses they would end up lying their eggs in nests of the most bundnt nd/or conspicuous host of this prticulr hbitt, which would lso be, with incresed probbility, the species by which they were rered. Still, errors could occur frequently, especilly in highly structured hbitts where mny potentil host species my coexist, or if the bundnce or conspicuousness of hosts differs between res contining the sme hbitt type. The observtion tht egg mimicry is better in res with homogeneous thn with frgmented hbitts (Moksnes & Røskft 1995) supports the hbitt-imprinting hypothesis. In comprison with ntl philoptry, mechnism such s hbitt imprinting would mintin egg mimicry nd llow young cuckoos to disperse nd colonize res with fewer cuckoos nd less competition for hosts. Under ntl philoptry, dispersl would probbly be mldptive. Host preference No evidence ws found for host preference, either inherited or imprinted, in n experimentl study by Brooke & Dvies (1991), but their test series hd smll smple sizes s well. In comprison with the other proposed mechnisms of finding hosts, host preference could result in very good egg mimicry. However, the nlyses of Moksnes & Røskft (1995) render it unlikely tht cuckoos hve perfect knowledge of their host species nd strictly ly their eggs in nests of this species. Evidence provided by Bker (1942) could point towrds host preference. His observtions suggested tht femles with overlpping territories my prsitize different host species. Nest site selection It is uncler whether nd how nest site selection could mintin egg morphs. Moksnes & Røskft (1995) found tht 77% of cuckoo eggs were in nests of hosts with nesting sites similr to those of the min host but (1) they did not propose mechnism for how femles could recognize or detect similr nest sites nd (2) for the dt nlyses, they defined very corse ctegories of similrity, such s nest sites in trees, nest sites in low vegettion, nest sites on the ground nd nest sites in holes. It ppers unlikely, for exmple, tht preference of cuckoos for nest sites in low vegettion could contribute much to the mintennce of egg morphs, s 42 of the prsitized bird species exmined in Moksnes & Røskft s study nest exclusively or prtly in low vegettion but their eggs correspond to wide rnge of different cuckoo egg types. A Hierrchicl Decision Process for Nest Selection The bove discussion suggests tht single mechnism mintining different egg morphs within popultion is unlikely. A solution to this problem could be stepwise decision process tht combines severl of the hypothesized selection mechnisms in sequence. (1) On their spring migrtion from the Africn wintering grounds, cuckoos return pproximtely to the geogrphicl region where they were born by using n innte or lerned migrtion route (ntl philoptry). This enhnces the probbility tht they encounter their ntl hbitt type. (2) Cuckoos serch for hbitt tht they hd been imprinted on s nestlings or fledglings. (3) Within this hbitt they serch for suitble nests. For tht purpose, they my use finer selection criteri such s gross chrcteristics of the host species or of the microhbitt round the nest which they might hve lernt when young. This ssumed temporl sequence of decision processes is prlleled by sptil hierrchy from brod to fine scle. Sequentil nd sptilly hierrchicl decisions hve lso been proposed s the centrl mechnism in hbitt selection of nonbreeding migrtory lnd birds (Hutto 1985) nd of rptor species (Gmuf 1988). The reltive importnce of hbitt (step 2) versus potentil finer microhbitt nd/or host chrcteristics (step 3) for nest selection my be tested in n re contining t lest two distinct hbitts, ech of them inhbited by host tht is specilized in one of them, nd by host species tht uses both hbitts like (generlist). Acknowledgments We thnk Hns Winkler for suggesting hbitt s potentilly importnt component of cuckoo ontogeny, Christ nd Hns Hudde for help with rering cuckoos, Wolfgng Pegler nd the lte Fritz Spitzhofer for help with building the experimentl set-up, Ev Skubic nd Michel Ritzmeier for fruitful discussions, Sigl Blshine- Ern, Nick Dvies, Richrd Lnctot, Arnon Lotem, Arne Moksnes, John Reynolds nd n nonymous referee for constructive comments on the mnuscript, nd the Deutsche Ornithologen Gesellschft (DO-G) nd the Ethologische Gesellschft for finncil support.

TEUSCHL ET AL.: HABITAT IMPRINTING IN CUCKOOS 1433 References Aldredge, J. R. & Rtti, J. T. 1986. Comprison of some sttisticl techniques for nlysis of resource selection. Journl of Wildlife Mngement, 50, 157 165. Bker, E. C. S. 1942. Cuckoo Problems. London: Witherby. Bolhuis, J. J. 1991. Mechnisms of vin imprinting: review. Biologicl Reviews, 66, 303 345. Bollinger, E. K. & Gvin, T. A. 1989. The effects of site qulity on breeding-site fidelity in bobolinks. Auk, 106, 584 594. Brooke, M. de L. & Dvies, N. B. 1988. Egg mimicry by cuckoos Cuculus cnorus in reltion to discrimintion by hosts. Nture, 335, 630 632. Brooke, M. de L. & Dvies, N. B. 1991. A filure to demonstrte host imprinting in the cuckoo (Cuculus cnorus) nd lterntive hypotheses for the mintennce of egg mimicry. Ethology, 89, 154 166. Ctchpole, C. K. 1972. A comprtive study of territory in the reed wrbler (Acrocephlus scirpceus) nd sedge wrbler (A. schoenobenus). Journl of Zoology, 166, 213 231. Chnce, E. P. 1922. The Cuckoo s Secret. London: Sidgwick & Jckson. Chnce, E. P. 1940. The Truth bout the Cuckoo. London: Country Life. Conover, W. J. 1980. Prcticl Nonprmetric Sttistics. 2nd edn. New York: J. Wiley. Dvies, N. B. & Brooke, M. de L. 1988. Cuckoos versus reed wrblers: dpttions nd counterdpttions. Animl Behviour, 36, 262 284. Dröscher,L.M.1988. A study on rdio-trcking of the Europen cuckoo (Cuculus cnorus). In: Current Topics in Avin Biology: Proceedings of the Interntionl Centennil Meeting of the DO-G (Ed. by R. vn den Elzen), pp. 187 193. Bonn: Verlg der Deutschen Ornithologen-Gesellschft. Gmuf, A. 1988. Hierrchische Ordnung in der Whl der Nistpltzund Jgdhbitte dreier symptrischer Greifvogelrten (Buteo buteo, Pernis pivorus, Accipiter gentilis). Ph.D. thesis, University of Vienn. Gssmnn, H. 1991. Herzfrequenznlyse und Hbittwhl beim Hänfling (Acnthis cnnbin). Ph.D. thesis, Technische Hochschule, Achen. Glück, E. 1984. Hbitt selection in birds nd the role of erly experience. Zeitschrift für Tierpsychologie, 66, 45 54. Glutz von Blotzheim, U. N. & Buer, K. M. (eds) 1980. Hndbuch der Vögel Mitteleurops. Vol. 9. Frnkfurt m Min: Akdemische Verlgsgesellschft. Greenwood, P. J. & Hrvey, P. H. 1982. The ntl nd breeding dispersl of birds. Annul Review of Ecology nd Systemtics, 13, 1 21. Grünberger, S. & Leisler, B. 1990. Angeborene und erfhrungsbedingte Komponenten der Hbittwhl der Tnnenmeise (Prus ter). Journl für Ornithologie, 131, 460 464. Grünberger, S. & Leisler, B. 1993. Auswirkung der Umwelterfhrung uf die Neophobie der Tnnenmeise (Prus ter). Journl für Ornithologie, 134, 352 355. Grünberger, S. & Leisler, B. 1993b. Die Ausbildung von Hbittpräferenzen bei der Tnnenmeise (Prus ter): genetische Prädisposition und Einfluß der Jugenderfhrung. Journl für Ornithologie, 134, 355 358. Hildén, O. 1965. Hbitt selection in birds. A review. Annles Zoologici Fennici, 2, 43 75. Hutto,R.L.1985. Hbitt selection by non-breeding, migrtory lndbirds. In: Hbitt Selection in Birds (Ed. by M. L. Cody), pp. 455 476. New York: Acdemic Press. Klopfer, P. H. 1963. Behviorl spects of hbitt selection: the role of erly experience. Wilson Bulletin, 75, 15 22. Klopfer, P. H. 1965. Behviorl spects of hbitt selection: preliminry report on stereotypy in folige preferences of birds. Wilson Bulletin, 77, 376 381. Lck, D. 1968. Ecologicl Adpttions for Breeding in Birds. London: Methuen. Lienert, G.A. 1986. Verteilungsfreie Methoden in der Biosttistik. Vol. 1. 3rd edn. Königsstein: Hin. Ludwig, O. 1962. U}ber Kombintion von Rngkorreltionskoeffizienten us unbhängigen Meßreihen. Biometrische Zeitschrift, 4, 40 50. Moksnes, A. & Røskft, E. 1995. Egg-morphs nd host preference in the common cuckoo (Cuculus cnorus): n nlysis of cuckoo nd host eggs from Europen museum collections. Journl of Zoology, 236, 625 648. Newton, A. 1893. A Dictionry of Birds. Prt 1. London: Blck. Oring, L. W., Lnk, D. B. & Mxson, S. J. 1983. Popultion studies of the polyndrous spotted sndpiper. Auk, 100, 272 285. Rothstein, S. I. 1990. A model system for coevolution: vin brood prsitism. Annul Review of Ecology nd Systemtics, 21, 481 508. Srgent, T. D. 1965. The role of experience in the nest building of the zebr finch. Auk, 82, 48 61. Seel, D. C. 1977. Migrtion of the northwestern Europen popultion of the cuckoo Cuculus cnorus, s shown by ringing. Ibis, 119, 309 322. Seel, D. C., Wlton, K. C. & Wyllie, I. 1981. Age of first breeding in the cuckoo. Bird Study, 28, 211 214. Southern, H. N. 1954. Mimicry in cuckoo s eggs. In: Evolution in Process (Ed. by J. Huxley, A. C. Hrdy & E. D. Ford), pp. 257 270. New York: Collier Books. Torgeson, W. S. 1959. A non prmetric test of correltion using rnk orders within subgroups. Psychometrik, 21, 145 152. Wyllie, I. 1981. The Cuckoo. London: Btsford.