Habitat Heterogeneity and Life-History Variation of Mediterranean Blue Tits (Parus Caeruleus)

The Auk 110(3):511-520, 1993

HABITAT HETEROGENEITY AND LIFE-HISTORY VARIATION OF MEDITERRANEAN BLUE TITS (PARUS CAERULEUS)

JACQUESBLONDEL, PAULA CRISTINA DIAS, MARIE MAISTRE, AND PHILIPPE PERRET Centred'Ecologie Fonctionnelle et Evolutive,CNRS, B.P. 5051, 34033 Montpellierc•dex, France

AI•STRACr.--Theextensive variation in breeding traits of Mediterranean Blue Tits (Parus caeruleus)is hypothesizedto result from large differencesin the timing and abundanceof food resources,depending on whether the dominant tree speciesof the habitat is deciduous (Quercuspubescens) or evergreen(Q. ilex). Data were collectedon tree phenology,food abundance(caterpillars), and breeding traits of tits in one mainland deciduoushabitat, two mainland evergreenhabitats, and one evergreenhabitat on the island of Corsica.In the mainland deciduoushabitat, an early breeding time and a large clutch size were associatedwith an early and abundantfood supply. In the evergreenhabitats (both on mainlandand on Corsica), the leaflng processoccurred three weeks later and the abundanceof caterpillarswas much lower. On Corsica,tits startedto lay three weekslater than in the deciduousmainland habitat and laid about 30% fewer eggs.As in the deciduousmainland habitat, the breeding process seemedto be adjustedto the patterns of food availability. In the two mainland evergreen habitats,tits started to breed earlier and laid more eggs than expectedfrom the patternsof food availability so that they mismatchedthe bestperiod to raisetheir young. The apparent realadaptationof tits in theseevergreen mainland habitatsis hypothesizedto result from an asymmetricgene flow between rich deciduoushabitats (source), where well-adapted birds producemany fledglings,and poor evergreenhabitats (sink), where the densityis maintained through immigration from rich habitats. Received4 February1992, accepted20 November1992.

MANY PARTSof the Mediterranean region are northern limit of the Mediterranean, show some "checkerboard"landscapes with a large variety regular trends (Table 1): (1) three populations of habitatsdue to small-scaleheterogeneity in in deciduousoakwoods have the largestclutch such factors as altitude, geological substrate, size and start to breed early; (2) populationsin isolation, and especiallyvariation in dominant evergreen oakwoods have a smaller clutch size tree species (deciduous broad-leaved, ever- and breed later in season;and (3) somepopu- green broad-leaved, coniferous). In addition, lationsin evergreenoakwoods breed extremely land-use practiceshave contributed greatly to late and havea very smallclutch size, especially the mosaiclike pattern of many Mediterranean those that are isolatedon islands(Canary Is- landscapes.Because the Blue Tit (Parus caeru- lands, no. 13; Corsica, no. 22). There is an in- leus)occurs in many forest types and has been verse relationship between laying date and extensively studied acrossits range, this species clutch size (r = -0.729, P = 0.003, n = 15). is a good model for evaluating the extent to Becausefood supply has consistentlybeen which life-history traits are adaptedto local en- shownproximately and ultimatelyto determine vironments.Large variation occursin breeding important breeding traits such as laying date traits of Blue Tits over their range (Fig. 1). On and clutchsize (Lack1966, Klomp 1970,Drent average,birds start to breed earlier and have a and Daan 1980, Ricklefs 1983, Martin 1987, Per- lower clutch size (with a larger among-habitat rins and McCleery 1989, Zandt et al. 1990), we variation of these traits) in the Mediterranean hypothesizethat the large among-habitatvari- region than in centraland northern Europe(Fig. ation of breeding traits in the Mediterranean 1). For example, the Mediterranean includes region representsa responseto large variation populations that have the earliest (21 March; in the food supply. population 14 on Fig. 1) and among the latest Evergreen trees renew but a fraction of their (11 May; population 22) onsetsof breeding in foliage (ca. 30%) yearly, since their leavesare Europe.Patterns of variation in laying dateand photosyntheticallyactive for severalyears (three clutch size of the Blue Tit at 12 Mediterranean years in the Holm oak, Quercusilex; Floret et al. study sitesin the Mediterranean bioclimaticre- 1989). By comparisonwith deciduousoaks that gion, ranging from the Canary Islands to the renew their whole foliage each year, this pat- 511 512 BLONDELE? AL. [Auk, Vol. 110

•cs tt.$ 11.$ tl.6 lO. 7 L•I lO.6 .--I Z 1o.g 1t.5 to.8 Z '•- lO.O lO.$ Z tt.o 51 '------.•o to.$ 48 .f2

C S 4.8 7.9 69 10.0 8.1 10.9 8.8 80 7.9 6.2 LO 63 2t 50 43 51 41 56 59 63 72

•r MEDITERRANEAN

Fiõ. 1. Mean clutchsizes (CS _+1 SD) and layinõ dates(LD _+1 SD) oœBlue Tits (Paruscaeruleus) in different localitiesof WesternPalaearctic. Dates in March dates(32 = 1 April). Sources:(1) Hild•n 1990;(2, 3) K•illander 1983; (4) Frederiksen et al. 1972; (5) Lack 1950; (6) Perrins 1965; (7) Berndt et al. 1983; (8) Delm•e et al. 1972; (9) Zang 1980;(10) Neub 1977;(11) Leclercq1977; (12) Glutz von Blotzheim1962; (13) B. Schottlerpers. comm.; (14-18, 20) Isenmann 1987; (19, 22) Blondel 1985; (21) Orsini pers. comm. tern has two important consequencesfor phy- chemicalrepellents (Feeny 1975, Crawley 1983), tophagous animals. First, since most (70%) of there should be less production of insects in the photosyntheticsystem is activeyear-around, evergreen than in deciduoustrees in which all a large part of the energy available in spring is the leavesproduced in spring are available to used by this system so that only a fraction is the insects.Therefore, the spring bloom of de- allocated to the production of new leaves. Ac- ciduous vegetation should be sharper and fast- cordingly, the leating processshould occur later er, with a higher peak of productionresulting and more slowly in evergreen than in decidu- in a sharper, but shorter, pulse of arthropods, ous trees in which a much larger part of the than in evergreenvegetation. Such a pattern is available energy is allocatedto the production generally true in highly seasonalenvironments of new leaves.Second, since most phyllopha- when compared with less-seasonalones (Lack gous insects, such as caterpillars, cannot eat 1950, Rabenold 1978). If these hypothesesare leavesmore than a year old becausethey are true, the lower amount of food available late in too hard and include tannins and other bio- the seasonin evergreen as comparedwith de- July 1993] BlueTits in MediterraneanHabitats 513

TABLE1. Clutch size and laying date of the Blue Tit in 12 Mediterranean habitats.a

Clutch size Laying date Latitude (øN) No. Deciduous Evergreen 1 March-14April 15April-30 April 1 May-15 May 44ø07'b 19 -- 8.8 -- 25 April -- 43058' 16 10.0 -- 13 April -- -- 43o54' 17 -- 8.1 -- 20 April -- 43053' 18 10.9 -- 10 April -- -- 43051' 23 -- 8.6 -- 21 April -- 43040' 24 10.0 -- 7 April -- -- 43ø22'b 20 -- 8.0 -- 20 April -- 43015' 21 -- 7.9 -- -- 2 May 42ø34'c 22 -- 6.2 -- -- 11 May 37050 ' 14 -- 7.9 21 March -- -- 34002' 15 -- 6.9 -- 19 April -- 28ø20'c 13 -- 4.8 -- -- 3 May ßBold figures indicate habitats considered in this study.Numbers in secondcolumn refer to populationnumbers of Figure1 (populations23 and 24 not included in Fig. 1). • More than 800 rn above sea level. ß Island. ciduous habitats should result in differences in for routine inspectionthroughout the breeding sea- how well birdsare adapted to their localhabitat. son. The aim of our paper is two-fold. First, we Climaticdata were obtained(years 1975-1991) from try to identify proximate and ultimate factors official meteorologicalstations (Service Technique d'Etudes des Facteurs Climatiques de l'Environne- that determinethe timing of breedingand clutch ment, INRA) located inside or close to the study areas. size of Blue Tits in four Mediterranean habitats Mean ambienttemperature and rainfallfor the months that differ in their dominant tree species and Januaryto April are 7.3 to 8.0øCand 241 to 250 mm, degree of isolation. Second,we investigate to respectively,in the three mainland habitats,and 9.3øC what extent local populationsliving in habitat and 378 mm in the Corsican habitat. patches, which differ in the timing and the The springdevelopment of oak leaveswas checked abundanceof their food supply, have evolved from 1987 to 1991 using the procedureof Du Merle locally adapted traits. Answering these ques- and Mazet (1983). Random sampleswere taken from tions requires a large body of data on life-his- 25 oaksevery three to four days at different heights (2 to 20 m) in the tree in five areaswithin each habitat. tory traits, food resourcesand the phenology In each area, five trees have been sampled and six of the dominant trees. apical buds per tree have been checked.Bud burst and the developmentof the leaveswere assignedto STUDY AREAS AND METHODS one of sevencategories ranging from stage1 where the bud is completelyclosed to stage7 where the new Three habitatsare located near Montpellier, south- leaf is fully developed(see Fig. 2). During each visit, ern France: a deciduous habitat (hereafter referred as the number of buds at each state was multiplied by MD; no. 48 on Fig. 1 and Table 1) where downy oak the number of that stageand then weighted to 100 (Quercuspubescens) is the dominant tree species;and so that the final figure for eachdate givesthe average two evergreen habitats (MES and MEP; no. 17 and stageof the leating processin the habitat. Data were no. 23) where Holm oak (Q. ilex) is the dominant tree fitted to a logisticequation (Fig. 2). This procedure species.The fourth habitat is an evergreenoakwood gives the timing and the duration of the spring de- dominatedby Holm oak on the islandof Corsica(IE; velopment of the tree. no. 22). Most of the vegetation on Corsica is ever- The mostimportant food item for tits is caterpillars green, especiallyat low altitudes where the Blue Tit (Perrins 1965, Van Balen 1973, Zandt et al. 1990). These is most abundant. The three mainland habitats (MD, are always the preferred prey when availableand by MES and MEP) are part of the same landscape(same far the mostimportant on a biomassbasis in all areas altitude, substrateand climate so that the only factor including the Mediterranean(Zandt et al. 1990,Blon- in which they differ is dominant tree species).More del et al. 1991a). We estimated the abundance of cat- details about the habitats may be found in Blondel erpillars by collectingtheir droppingsfalling from (1985) and Isenmann et al. (1987). Nest boxes (50 to the foliage of the trees. Data were collected in two 140 dependingon yearsand sites)have been erected evergreen oakwoods(IE and MEP) in the years 1988- in each habitat at a density of two nest boxes per 1991 and in the deciduous mainland oakwood (MD) hectare. Nest boxes were visited at least once a week in 1991 using the method of Tinbergen (1960; see 514 BLONDELET AL. [Auk, Vol. 110

16 22 50 58 46 54 62 70 78 86 94 402 '1'10 4'18

i,i

March date

Caterpillarabundance

20-% 4-'

16 22 •0 • •6 5• 62 70 78 86 94 402 •0 4•8 March date Fig. 2. Upper:Leating phenology of deciduousoaks (habitat MD on mainland) and evergreenoaks (MES, MEP on mainland;IE on Corsica).Vertical axis indicatesstages of developmentof young leaves.Asterisks indicatelaying date of Blue Tits. Lower:Percent clutches initiated and variationin caterpillarabundance with date. Symbolsto right of arrows indicate mean date when nestlingsare on average10 daysold. The lengths of arrowsdiffer becauseof differencesin clutchsize. Datespresented as March dates(32 = 1 April).

Zandt et al. 1990). Droppings were collectedtwice a Data on breeding traits are available for the years week over the breedingseason on 0.25-m2 trays erect- 1985-1991 in the three mainland habitats and the ed under the canopy of the oaks. We sampled 10 to years 1976-1991in Corsica.Laying date is the mean 25 oaks using one tray per tree. Abundanceof cat- of the dateson which eachpair laid its first egg.Since erpillars in the habitat was expressedas an index Blue Tits usually lay one egg per day, the date on ranging from 0 to 5 (see Fig. 2). which the first egg was laid has been extrapolated July1993] BlueTits in MediterraneanHabitats 515

TABLE2. Breedingdensities (pairs/nest box), laying date _+1 SD, clutch size _+1 SD, breeding success,and timingof the mainevents of breedingcycle of BlueTits in deciduousmainland oakwood (MD) andevergreen oakwoods(MEP and MES = mainland;IE = island).Sample size in parentheses.Data for MES from Isenmann et al. (1987).

Breeding site MD MEP MES IE Breedingdensity 0.45 0.07 0.10 0.60 Laying date 10 April + 6.7 (81) 21 April + 3.9 (49) 19 April + 4.3 (88) 11 May + 6.8 (417) Clutch size 9.8 + 1.7 (81) 8.5 + 2.1 (49) 7.8 + 1.5 (85) 6.6 + 1.1 (417) Breeding success(no. fledglings/egg) 0.60 (64) 0.43 (40) 0.54 (85) 0.65 (380) Bud stageat laying 3.2 1.4 1.4 3.9 Date of bud opening 14 April 6 May 12 May 8 May No. days between laying date and bud opening 3 -15 -25 -3 Caterpillar peak date 30 April 13 June 3 June No. days between bud opening and caterpillar peak date 16 38 26

from the weekly visits to the nest boxes.Clutch size (MD), caterpillarsappeared at the beginning of is the mean of the completed clutcheslaid by each April, quickly increased in abundance and female. We have excluded those clutches where the peaked at the end of April (peak date 61 = 30 female never got as far as incubating,and clutchesof April; Fig. 2, Table 2). In the two evergreen three or fewer eggs becausesuch small clutches are oakwoods for which data are available, cater- probably incomplete.Only first clutcheswill be con- pillars began to emergethree to four weeks later sideredin this paper. Becauseyear-to-year variation in the phenology of the trees and breeding traits of (peak date 105 = 13 June in MEP, and 95 = 3 the tits was much less at the within-habitat level than June in IE), but they were available over a lon- at the between-habitat level, data for each habitat ger period of time, presumablya consequence were averagedover the yearsof the study. of the longer leating processof the Holm oak. Peakabundance of caterpillarsoccurred 16 days RESULTS after bud burst in the deciduousoakwood (MD), and 38 and 26 daysafter bud burst in the main- Springdevelopment of vegetationand caterpillar land (MEP) and the island (IE) evergreen oak- abundance.--Theleating processof the oaksdif- woods,respectively (Table 2). As expected,more fered strongly between deciduous and ever- caterpillars (index of abundance = 5.0) were green trees.Bud burst (stage3.5 of leating pro- found in deciduousthan in evergreenoaks (3.9 cess; see Fig. 2, Table 2) occurred three weeks and 2.6 on the mainland and in Corsica, re- earlier in the deciduous habitat (MD, "March- spectively).Here again, the pattern was similar date"45 = 14April) than in the evergreen(MEP, on the mainland and on the island, although 6 May; MES, 12 May; IE, 8 May). The overall the peak date for caterpillarsoccurred slightly pattern in evergreen oaks was similar in the earlier on Corsica than on the mainland. In two mainland sites and on Corsica. This is sur- summary,the combinationof earlier and faster prising becauseCorsica enjoys a warmer, moist- development of new leaves in deciduousthan er climate. in evergreen oaks,coupled with a parallel trend The seasonalvariation of caterpillar biomass in arthropod production, supportsthe predic- was closely related to the development of oak tions made in the introduction. leaves, which was expected since caterpillars Breedingtraits of Blue Tit.--Population sizes, are dependent on young leaves. For example, expressedas the number of breeding pairs per young caterpillarsof Tortrixviridana, an impor- nest box (excluding repeat and second clutch- tant prey speciesfor tits, cannot enter oak buds es),varied amonghabitats. Densities were fairly until they burst, which occursthree weeks later high in the deciduous mainland habitat (0.45 in the Holm oak than in the downy oak (Du breeding pair per nest box) and in the ever- Merle 1983). In the deciduous mainland habitat green island habitat (0.60), but much lower in 516 BLONDELET AL. [Auk, Vol. 110

RICH POOR DISCUSSION

The main features of life-history variation among these four habitats are the combination APRIL NI A Y J U NE of: (1) large clutch size and early laying date in 10 2O // 10 20 the deciduous oakwood; (2) small clutch size 12 ' ' ' • i ' ' and late laying date in the Corsican evergreen oakwood; and (3) intermediate values for these traits in the two mainland evergreen oakwoods (Fig. 3). Assuming that birds that do best are those that start to breed as early as possiblein habitats where food is plentiful (Perrins 1965, 1970, Van Balen 1973), habitats may be sepa- rated into richer habitats (i.e. deciduous habitat [MD] on mainland), and poorer habitats(i.e. two

RICH POOR POOR mainland evergreen habitats [MEP and MES] and evergreen habitat in Corsica;see Fig. 3). MAINLAND CORSICA The first three habitats (MD, MES, MEP) are Fig. 3. Adapredness(to rich mainland habitat and parts of the samelandscape over which tits free- poor Corsican habitat) and maladaptedness(to poor ly disperse.In contrast,the insular evergreen mainland habitat) of clutch size and laying date of habitat (IE) is isolatedfrom any rich deciduous Blue Tit. Breeding traits in poor mainland habitat habitat. As a result, nearly all the habitats with- approachthose of rich habitat (dashedarrow) because in the dispersalrange of this CorsicanBlue Tit of asymmetricgene flow from latter to former. population are of poor quality. Becauseterritory size is to some extent related to the abundance of resources(Lack 1968, Cody and Cody 1972, the evergreen mainland habitats (0.10 in MES Nilsson 1976, Brown 1982, Stampsand Tolle- and 0.07 in MEP). A higher occupancyrate of strup 1984), densities are much higher in the nest boxes in Corsica is consistent with mea- deciduous habitat than in the evergreen habi- sured higher natural densitiesof Blue Tits on tats on the mainland. However, Blue Tit den- Corsica than on the mainland (Blondel et al. sities on Corsicaare very high, presumably be- 1988). cause of density compensation (or density Laying started earliest (mean laying date of inflation) in this depauperateinsular commu- all the pairs of the population) in the mainland nity (for discussion,see Blondel et al. 1988). deciduousoakwood (10 April; Table 2), but one Several studies in temperate Europe have month later in the evergreen oakwood of Cor- shown that Great Tit (Parusmajor) and Blue Tit sica (11 May). In the two mainland evergreen laying dates are negatively correlated with habitats,the onsetof breedingwas intermediate springtemperatures (Kluyver 1951,Perrins 1965, (19 April in MES; 21 April in MEP). Since the Van Balen 1973, Slagsvoid 1976, Schmidt 1984, three mainland habitats are parts of the same Blondel 1985,Perrins and McCleery 1989). Such landscapeand with the sameclimate, these dif- a relationshiphas been found in Mediterranean ferences in the onset of breeding mean that habitats on the mainland, but not on Corsica factorsother than temperature play an impor- (Blondel 1985). Because Corsica is further south tant role in determining laying date. Blue Tits and has a much warmer climate than the main- startedto lay within a few daysof bud burst in land site (ambient temperatures• = 9.3øCfor the mainland deciduous habitat and in Corsica the months January-April vs. 7.3øC on the (Fig. 2), but long before budsopened in the two mainland), the extremely late breeding date of mainland evergreenoakwoods (MES, MEP). At Corsican tits was unexpected. these two sites, which match the Corsican hab- Temperature acts mainly through the spring itat where tits were expected to show the same development of the leaves and the associated general pattern in breeding traits, they started arthropodfauna the birds feed upon (Lack 1966, to lay 10 to 11 days later than in the deciduous Van Balen 1973, Perrins and McCleery 1989, mainland habitat but three weeks earlier than Zandt et al. 1990). Laying date has been found on Corsica. to be significantlycorrelated with bud burst in July1993] BlueTits in MediterraneanHabitats 517

deciduous habitats for Blue Tits by Clamens development of the deciduous vegetation on (1988, 1990) in the Mediterranean region and the mainland, laying date is alsodetermined by by Jones (1972) at Oxford for Great Tits. Bud factorsrelated to the food supply, sinceit nicely burst is an important event for tits becauseit is coincideswith the period of caterpillar abun- the time when first-instar caterpillarsinvade dance (Figs. 2 and 3). Resultsof our study sup- the budsand beginto eatthe young leaves(Var- port the prediction that selection would favor ley and Gradwell 1968, Perrins 1973). Young earlier breeding and larger clutchesin rich de- tits must be in the nest as early as possiblein ciduous mainland habitats, and smaller clutches the periodof caterpillarabundance, because the and later laying dates in poor evergreen island most successfulyoung are those that are about habitats(Fig. 3). Similar effectsof current food 10 days old when caterpillar numbers peak supply on clutch size have been reported by (Perrins 1965, 1990). In order to have 10~day- Van Balen (1973), Klomp (1970), Arcese and old young in the nest at the peak of caterpillar Smith (1988), Perrins and McCleery (1989) and numbers,egg developmentmust begin 35 days Perrins (1990), who found a correlation be- beforefor a clutchof eight eggs(4 daysto form tween clutch size and annual caterpillar den- an egg, 7 days to complete the clutch, 14 days sity. Combining the data of the two Mediter- incubatingplus a nestling period of 10 days). ranean habitats (MD and IE), where there is To do this, birds must anticipate the most fa- good correspondencebetween laying date and vorableperiod and have to make use of appro- food availability, with five other data setsin the priate proximate clues.Tits begin laying assoon Netherlands and Great Britain (see Zandt et al. as they are able to store enough reserves to 1990), laying date is significantly correlated (r produce eggs (Perrins 1970, Martin 1987). = 0.863, P < 0.05) with the peak date of cater- Therefore, the onset of breeding varies with pillar abundance. food availability becausetits cannot store more Laying date and clutch size in the two main- than a tiny proportion of the reservesneeded land evergreen oakwoodswere expected to be for the whole clutch.This has been shown by similar to thoseon Corsicasince the processof correlations between laying date and food leaf development and the variation of food availability at both the among-(Greenwood and abundance are similar. However, tits in these Hubbard 1979, Dhondt et al. 1984) and within- habitatsbred too early so that they missedthe habitat level over several years (Perrins and peak of caterpillarabundance, and laid too many McCleery 1989, Zandt et al. 1990), as well as eggs in relation to the number of young they food supplementationexperiments where sup- could successfullyraise (Table 2, Fig. 3). There- plemented pairs bred earlier than unfed control fore, breeding success(no. fledglings per egg) pairs (K•llander 1974, Von Br6mssenand Jans- was lower in these habitats than in either the son 1980, Smith et al. 1980, Ewald and Rohwer mainland deciduous oakwood or in the Corsi- 1982, Clamens and Isenmann 1989). can evergreen oakwood (Table 2; Dias et al. The relationshipbetween laying dateand cat- 1994). Their timing would have been correct erpillar abundance indicates that Blue Tits in and the tits would have produced the right the mainland deciduous habitat (MD) and in number of eggshad they been breeding in the the evergreenhabitat of Corsica(IE, Fig. 2) time neighboring deciduous oakwoods. Moreover, their laying so that their young are in the nest laying date was not correlated with the bud around the peak of food abundance.Blue Tits burst of evergreenoaks in their habitat,but was lay four weeks later in Corsicathan in MD be- significantly correlated with the bud burst of causeof the three- to four-weektime lag in the deciduousoaks in surroundinghabitats (r = 0.85, leating patterns and the development of cat- P < 0.01; Clamens 1988, 1990). It appearsthat erpillars in this habitat. the leating processof deciduousoaks predicts In summary,in the mainland deciduoushab- laying date in both evergreen and deciduous itat, the egg laying by Blue Tits, bud opening oakswithin the sameneighborhood. of oaks, and emergence of caterpillars were all The relevant questionraised by our study is closely correlated with spring temperature as not so much why Corsican tits breed so late, in other deciduoushabitats in Europe (Slags- but why tits in mainland evergreen habitats void 1976, Schmidt 1984). On Corsica, where breed so early and do not shift their breeding evergreen trees begin development at temper- traits in order to match the local variation in aturesmuch higher than thosethat activatethe food resources.One explanation for the poor 518 ]3LONDELET AL. [Auk, Vol. 110 timing of Blue Tits laying in evergreen main- shown by somepopulations that fledge only a land habitatsis the gene-flowhypothesis (Blon- few low-quality young (Dias et al. 1994)predicts del et al. 1991b). Birds breeding in rich decid- that suchprocesses should occur (PullJam 1988). uous habitatsproduce more fledglings of good For a speciesliving in landscapeswhere habi- quality than thosebreeding in poor ones(Table tats are not equally productive and in which 2; Dias et al. 1994). Thus, a large proportion of there are density-dependentcomponents in re- the birds settling in the poor-quality habitats production, habitats with an overproduction of are immigrantsfrom the good-qualityhabitats individuals may act as "sources"from which and time their breedingin the poor habitatsas individuals will emigrate to habitatswith pro- if they were in rich ones.Hence, their timing ductivity lower than that necessaryto maintain is incorrect presumably because asymmetric the population size. Such "sink" populations gene flow from deciduousoakwoods prevents will persist becauseof immigration from the them becoming adapted to evergreen wood- source(Blondel et al. 1991b). Third, what makes land. On Corsica,where nearly all the vegeta- an individual decide to leave the habitat where tion is evergreenin the landscapeunder study, it was born or to stay is unknown. Morpho- gene flow must be small. Therefore, the pop- metric data indicate that individuals that settle ulation (which belongs to the subspeciesP. c. in the poor mainland evergreen habitats are ogliastraeHartert) has been able to evolve life- smaller(shorter wing and tarsus)and have low- history traits fairly well adjustedto the local er breeding successthan thosethat breed in the environment. The gene-flow hypothesis is rich deciduoushabitat (Dias et al. 1994). Similar supportedby the demonstrationof a genetic differences found by Van Balen (1973) for componentof laying date (Blondel et al. 1990, breedingsuccess and by Lemel (1989) and Ulf- Lambrechtsand Dias 1993),which confirmspre- strand et al. (1981) for body size in the Great vious studieson the heritability of this trait in Tit support the despotic-distribution model of the Great Tit (Van Noordwijk et al. 1980). The Fretwell and Lucas (1970), whereby dominant slightly delayedlaying date(10 days)and small- individuals breed in the habitats richest in re- er clutch size (2.3 eggs smaller) in mainland sources.However, since body size is not nec- evergreen as compared with deciduousoak- essarilyrelated to an individual's ability to out- woodssuggest that sometuning is proximately competean opponent (Lemel 1989, Lambrechts achieved through a phenotypicadjustment of and Dhondt 1986), additional studies are re- both laying date and clutch size to food avail- quired to assessthe body condition, history, ability. Similar casesof poorly timed breeding social status and genetic constitution of indi- by tits in habitatsof differing quality have been viduals accordingto their decisionto stay or to reportedby Van Balen(1973), Perrins and Moss move. (1975) and Lemel (1989) for Great Tits, and by Dhondt et al. (1990) for Blue Tits. In a similar ACKNOWLEDGMENTS context, Emlen (1978) gave evidence of the swamping effect on incipient local adaptation Logisticalsupport was provided by the Association of several bird speciesin the southern part of pour l'Etude Ecologiquedu Maquis(APEEM). We owe special thanks to D. 0_ridG. Viale. Financial support the Florida peninsulaby gene-flow from pop- was provided by the CNRS, the Ministery of Envi- ulationsof the centralpart. Correspondingpop- ronment (SRETIE), and the JNICT (Programa Ci•n- ulations in similar environments on the island cia). The CRBPO provided us with rings. We thank of Grand Bahama evolved appropriate adapta- P. Isenmann, A. Clamens, B. Schottler, H. Zandt and tions becausegene flow is prevented by the sea many studentsfor help with severalaspects of this barrier. study.Olavi Hildfin, MarcelLambrechts, Christopher Our study raisesthree points that should be Perrins, Michel Raymond, Gary D. Schnell, and Johrt clarified by studiesin progress.First, we do not Wiens made useful comments on earlier versions of know which factorsproximately determine lay- the manuscriptand improved the English. ing date in mainland evergreen oaks.Cues must be different on the mainland and in Corsica LITERATURE CITED

since the birds begin laying at very different ARCESE,P., ANDJ. N.M. SMITH. 1988. Effectsof pop- stagesin the spring developmentof the trees. ulation densityand supplementalfood on repro- Second, gene flow and asymmetric dispersal duction in SongSparrows. J. Anim. Ecol.57:119- amonghabitat patches on the mainlandstill has 136. to be demonstrated.However, the poor timing BERNDT,R., W. WINKEL,AND H. ZANG. 1983. 0ber July1993] BlueTits in MediterraneanHabitats 519

Legebeginn und Gelegest•irke von Kohl- und pubescent,du ch&nevert et de Tortrixviridana L. Blaumeise(Parus major, P. caeruleus)in Beziehung (Lep., Tortricidae).Mise en &videncechez l'in- zur geografischen Lage des Brutortes. Vogel- sectede deuxpopulations sympatriques adapt&es warte 32:46-56. chacune /• l'un des ch&nes. Acta Oecol./Oecol. BLONDEL,J. 1985. Breeding strategiesof the Blue Tit Appl. 4:55-74. and the Coal Tit (Parus) in mainland and island DU MERLE,P., AND R. MAZET. 1983. Stadesph&no- Mediterranean habitats:A comparison.J. Anim. logiqueset infestationpar Tortrixviridana L. (Lep., Ecol. 54:531-556. Tortricidae) des bourgeonsdu ch&nepubescent BLONDEL,J., D. CHESSEL,AND B. FROCHOT.1988. Bird et du ch&nevert. Acta Oecol./Oecol.Appl. 4:47- species impoverishment, niche expansion and 53. density inflation in Mediterranean island habi- EMLEN,J. 1978. Density anomaliesand regulatory tats. Ecology 69:1899-1917. mechanismsin land bird populationson the Flor- BLONDEL,J., A. DERVIEUX,M. MAISTRE,AND P. PERRET. ida peninsula. Am. Nat. 112:265-286. 1991a. Feedingecology and life historyvariation EWALD,P. W., AND S. ROHWER.1982. Effectsof sup- of the Blue Tit in Mediterranean deciduous and plemental feeding on timing of breeding, clutch sclerophylloushabitats. Oecologia 88:9-14. sizeand polygyny in Red-wingedBlackbirds (Ae- BLONDEL,J., P. PERRET,AND M. MAISTRE. 1990. On galiusphoeniceus). J. Anim. Ecol.51:429-450. the genetical basisof the laying date in an island FEENY, P.P. 1975. Biochemical coevolution between population of Blue Tit. J. Evol. Biol. 3:469-475. plants and their insect herbivores. Pages3-19 in BLONDEL,J., P. PERRET,M. MmSTRE,AND P. C. DIAS. Coevolutionof plants and animals (L. E. Gilbert 1991b. Do harlequin Mediterranean environ- and P. H. Raven, Eds.). Univ. Texas Press, Austin. ments function as source sink for Blue Tits (Parus FLORET,C., M. J. GALAN, E. LE FLOC'H,F. LEPRINCE, caeruleusL.)? LandscapeEcol. 6:213-219. ANDF. ROMANE.1989. Pages9-97 in Plant phe- BROWN,J. L. 1982. Optimal group size in territorial nomorphological studies in Mediterranean-type animals. J. Theor. Biol. 95:793-810. ecosystems(G. Orsban, Ed.). Kluwer Academic CLAMENS,A. 1988. D•terminants •cologiquesde la Publishing, Dordrecht, The Netherlands. date de ponte et de la grandeur de ponte de la FREDERIKSEN,K. S., M. JENSEN,E. H. LARSEN,AND V. M•sangeBleue (Parus caeruleus): Le casde la ch•- H. LARSEN. 1972. [Some data illustrating time naie de Chines verts. Ph.D. dissertation, Univ. of breeding and number of eggsin tits Paridae]. Montpellier, France. Dan. Ornithol. Fbren. Tidsskr. 66:73-85 [in Dan- CLAMENS,A. 1990. Influenceof oak (Quercus)leating ish]. on Blue Tits (Paruscaeruleus) laying date in Med- FRETWELL,S. D., AND H. L. LUCAS. 1970. On territorial iterranean habitats. Acta Oecol. 11:539-544. behaviour and other factors influencing habitat CLAMENS,A., ANDP. ISENMANN.1989. Effect of sup- distribution in birds. Acta Biotheor. 19:16-36. plemental food on the breedingof Blueand Great GLUTZYON BLOTZHEIM,U. N. 1962. Die Brutv•3gel tits in Mediterranean habitats. Ornis Scand. 20: der Schweiz. Verlag Aargauer Tagblatt AG, Aa- 36-42. rau, Switzerland. COD¾,M. L., AND C. B. COD¾. 1972. Territory size, GREENWOOD,I. J. D.,AND S. F. HUBBARD. 1979. Breed- clutch size and food in populations of wrens. ing of Blue Tit in relation to food supply. Scott. Condor 74:473-477. Birds 10:268-271. CRAWLEY,M. J. 1983. Herbivory. The dynamics of HILDfi'q, O. 1990. Long-term study of a northern animal-plant interactions.Studies in Ecology,vol. population of the Blue Tit Paruscaeruleus. Pages 10. Blackwell, Oxford. 65-75 in Population biology of passerinebirds, DELMf•E, E., P. DACHY, AND P. SIMON. 1972. Contri- an integrated approach (J. Blondel, A. Gosler, J. bution/• la biologie des M•sangesen milieu fo- D. Lebreton, and R. McCleery, Eds.).NATO ASI restier. Aves 9:1-79. SeriesG, vol. 24. Springer-Verlag,Berlin. DHONDT, A. A., F. ADRIAENSEN,E. MATTHYSEN,AND B. ISENMANN,P. 1987. Geographicalvariation in clutch- KEMPENAERS.1990. Nonadaptive clutch sizesin size:The example of the Blue Tit (Paruscaeruleus) tits. Nature 348:723-725. in the Mediterranean area. Vogelwarte 34:93-99. DHONDT,A. A., R. EYCKERMAN,R. MOERMANS,AND J. ISENMANN, P., P. CRAMM, AND A. CLAMENS. 1987. HUBL•. 1984. Habitats and laying date of the Etudecompar6e de l'adaptationdes m6sanges du Great and Blue tit, Parusmajor and P. caeruleus. genre Parusaux diff&rentesessences foresti&res Ibis 126:388-397. du bassin m6diterran6en occidental. Rev. Ecol. DtAS, P. C., F. MEUNIER, S. BELTRA, AND M. CAR- Suppl. 4:17-25. TAN-SON. 1994. Blue Tits in Mediterranean mo- JONES,P. J. 1972. Food as a proximate factor regu- saics.Ardea. In press. lating the breeding seasonof the Great Tit (Parus DRENT,R. H., ANDS. DAAN. 1980. The prudent par- major). Pages 657-658 in Proceedings XVth In- ent: Energeticadjustments in avian breeding.Ar- ternationalOrnithological Congress (K. H. Voous, dea 68:225-252. Ed.). The Hague, 1970. E. J. Brill, Leiden, The DUMERLE,P. 1983. Ph&nologiescompar6esduch&ne Netherlands. 520 BLONDELET AL. [Auk, Vol. 110

K•LLANDER,I-I. 1974. Advancementof laying of Great PERRINS,C. M., AND D. MOSS. 1975. Reproductive Tits by the provisionof food. Ibis 116:365-367. rates in the Great Tit. J. Anim. Ecol. 44:695-706. K*LLANDER,H. 1983. Aspectsof the breeding biol- PULLIAM,H.R. 1988. Sources,sinks, and population ogy, migratory movements,winter survival, and regulation. Am. Nat. 132:652-661. population fluctuations in the Great Tit (Parus RABENOLD,K.N. 1978. Foragingstrategies, diversity, major)and the Blue Tit (Paruscaeruleus). Ph.D. and seasonalityin bird communitiesof Appala- dissertation, Univ. Lund, Sweden. chian spruce-fir forests.Ecol. Monogr. 48:397- KLOMP, H. 1970. The determination of clutch size 424. in birds. A review. Ardea 58:1-124. RICKLEFS,R.E. 1983. Avian demography.Curr. Or- KLLP•ER,H.N. 1951. The populationecology of the nithol. 1:1-32. GreatTit, Parusm. majorL. Ardea 39:1-135. SCHMIDT,K. H. 1984. Fr/ihjahrstemperaturenund LACK,D. 1950. The breeding seasonsof European Legebeginnbei Meisen (Parus).J. Ornithol. 125: birds. Ibis 92:288-316. 321-331. LACK,D. 1966. Population studiesof birds. Claren- SLAGSVOLD,T. 1976. Annual and geographicalvari- don Press. Oxford. ation in the time of breeding of the Great Tit, LACK,D. 1968. Ecologicaladaptations for breeding Parusmajor, and the Pied Flycatcher,Ficedula hy- in birds. Methuen, London. poleuca,in relation to environmentalphenology LAM•REcFrrs, M., AND A. A. DHONDT. 1986. Male and springtemperature. Ornis Scand.7:127-145. quality, reproduction,and survival in the Great SMITH,J. N. M., R. D. MONTGOMraUE,M. J. TAn'r, AND Tit (Parusmajor). Behar. Ecol. Sociobiol. 19:57- Y. YoM-Tov. 1980. A winter feeding experiment 63. on an island sparrowpopulation. Oecologia 47: LAMBRECHTS,M. M., AND P. DIAS. 1993. Differences 164-170. in the onset of laying between island and main- STAMPS,J. A., AND K. TOLLESTRUP.1984. Prospective land Mediterranean Blue Tits: Phenotypic plas- resourcedefense in a territorial species.Am. Nat. ticity or geneticdifferences? Ibis. 135:451-455. 123:99-114. LECLERCO,B. 1977. Etude ph•nologique des para- TINBERGEN,L. 1960. The natural control of insects m•tres li6s • la reproductiondes M6sangesen in pinewoods.I. Factorsinfluencing the intensity futaie de Chines. Rev. Ecol. 31:599-619. of predationby songbirds.Arch. N(•er.Zool. 13: LEMEL,J. 1989. Habitat distribution in the Great Tit 265-336. Parusmajor in relation to reproductive success, ULFSTRAND,S., R. V. ALATALO, A. CARLSON,AND A. dominance and biometry. Ornis Scand.20:226- L•JNDBERG.1981. Habitat distribution and body 233. sizeof the GreatTit Parusmajor. Ibis 123:494-499. MARTIN,T.E. 1987. Food asa limit on breeding birds: VAN BALEN,J. H. 1973. A comparativestudy of the A life-historyperspective. Annu. Rev. Ecol.$yst. breedingecology of the Great Tit Parusmajor in 18:453-487. different habitats. Ardea 61:1-93. NmJB,M. 1977. Evolutions6kologischeAspekte zur VAN NOORDWIJK,A., J. H. VAN BALEN, AND W. Brutbiologie yon Kohlmeise, Parus major, und SCHARLOO.1980. Heritability of ecologically Blaumeise, Parus caeruleus. Ph.D. dissertation, importanttraits in the Great Tit. Ardea 68:193- 203. Univ. Freiburg-im-Brisgau,Germany. VARLEY,G. C., ANDG. R. GRADWELL.1968. Population NILSSON,S.G. 1976. Habitat, territory size, and re- modelsfor the winter moth. $ymp. R. Entomol. productivesuccess in the Nuthatch (Sittaeuro- Soc. Lond. 4:132-142. paea).Ornis Scand.7:179-184. VON BR6MSSEN,A., AND C. JANSSON.1980. Effectsof PERruNS,C. M. 1965. Population fluctuations and food addition to Willow Tit Parus montanus and clutch-sizein the Great Tit (Parusmajor). J. Anim. Crested Tit P. cristatusat the time of breeding. Ecol. 34:601-647. Ornis Scand. 11:173-178. PERruNS,C. M. 1970. The timing of birds' breeding ZANG, H. 1980. [The altitudinal influence on the seasons. Ibis 112:242-255. populationdensity and on the breedingbiology PERruNS,C.M. 1973. Some effectsof temperature on of Ficedulahypoleuca, Parus palustris, P. caeruleus, breedingin the Great Tit and Manx Shearwater. P. majorand P. ater in the Harz mountains].J. J. Reprod.Fert. Suppl.19:163-173. Ornithol. 121:371-386 [in German]. PERruNS,C.M. 1990. Factorsaffecting clutch-size in ZANDT, H., A. STPaJKSTRA,J. BLONDEL, AND J. H. VAN Great and Blue tits. Pages121-130 in Population BALEN. 1990. Food in two Mediterranean Blue biology of passerinebirds, an integratedap- Tit populations:Do differencesin caterpillar proach(J. Blondel, A. Gosler,J. D. Lebreton,and availabilityexplain differencesin timing of the R. McCleery,Eds.). NATO ASI SeriesG, vol. 24. breeding season?Pages 145-155 in Population Springer-Verlag,Berlin. biology of passerinebirds, an integrated ap- PERRINS,C. M., AND R. H. McCLEERY.1989. Laying proach(J. Blondel,A. Gosler,J. D. Lebreton,and dates and clutch size in the Great Tit. Wilson Bull. R. McCleery,Eds.). NATO ASI SeriesG, vol. 24. 101:236-253. Springer-Verlag,Berlin.