On the Evolution and Ecology of Island Populations of Crossbills Author(S): Craig W

Home , Scottish crossbill

On the Evolution and Ecology of Island Populations of Crossbills Author(s): Craig W. Benkman Reviewed work(s): Source: Evolution, Vol. 43, No. 6 (Sep., 1989), pp. 1324-1330 Published by: Society for the Study of Evolution Stable URL: http://www.jstor.org/stable/2409369 . Accessed: 25/01/2012 20:28

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Society for the Study of Evolution is collaborating with JSTOR to digitize, preserve and extend access to Evolution.

http://www.jstor.org 1324 NOTES AND COMMENTS

populationthat was formedby a recentbottleneck. WATTERSON, G. A. 1984. Allele frequenciesafter a Genetics 111:675-689. bottleneck.Theoret. Popul. Biol. 26:387-407. NEI, M., T. MARUYAMA,AND R. CHAKRABORTY.1975. WRIGHT, S. 1931. Evolution in Mendelian popula- The bottleneckeffect and geneticvariability in pop- tions. Genetics 16:97-159. ulations. Evolution 29: 1-10. WRIGHT, S., AND W. E. KERR. 1954. Experimental POLANS,N. O., ANDR. W. ALLARD. 1985. Inheritance studies of the distributionof gene frequenciesin of electrophoreticallydetectable variants in rye- verysmall populationsof Drosophila melanogaster. grass. J. Hered. 76:61-62. II. Bar. Evolution 8:225-240. ROBERTSON,A. 1955. Selectioninanimals:Synthesis. Cold Spr. Harb. Symp. Quant. Biol. 20:225-229. CorrespondingEditor: J. L. Hamrick

Evolution,43(6), 1989, pp. 1324-1330

ON THE EVOLUTION AND ECOLOGY OF ISLAND POPULATIONS OF CROSSBILLS

CRAIG W. BENKMAN' Departmentof Biology,Princeton University, Princeton, NJ 08544-1003

Received June 16, 1988. Accepted April 11, 1989

The correlationbetween bill sizes of differentspecies of black spruce (Fig. le), white spruce (Fig. If), and of Eurasian crossbills(Loxia) and the sizes of the co- tamarack on the mainland of North America (Benk- nifercones which crossbills usually exploit is well-known man, 1987a, 1987b). Red Crossbills (L. c. bendirei; (Lack, 1944; Newton, 1967). In North America, there Fig. lb) are typicallyassociated with the large-coned is a similarcorrespondence between bill sizes ofWhite- pines (Pinus strobusand P. resinosa; Fig. Ig, h; Benk- wingedCrossbills (Loxia 1. leucoptera)and Red Cross- man, 1987a, 1987b), yet pines have been uncommon bills (L. curvirostrabendirei) and the sizes of cones on on Newfoundlandfor the last 10,000 years, since the which theyfeed; this correspondenceextends to most glacial retreat(Delcourt and Delcourt, 1987). In gen- subspeciesof the Red Crossbill(Griscom, 1937; Benk- eral, White-wingedCrossbills are more efficientthan man 1987a, 1987b). However,the residentRed Cross- Red Crossbillswhen foraging on sprucecones, whereas bill on Newfoundland(L. c. percna) (Griscom, 1937; the converse is trueon pine cones (Benkman, 1987b). Dickerman, 1986a) is an exceptionto thisgeneral pat- Given thatL. c. percna is adapted forsurviving year- tern. round on Newfoundland(Griscom, 1937), two ques- The subspeciespercna has a more massive bill than tions need to be addressed. First, why has the Red is characteristicof White-wingedCrossbills and a rel- Crossbill differentiatedinto a distinct subspecies on ativelydeep bill compared to other Red Crossbills in Newfoundland?Second, since L. c. percna is a resident North America (e.g., L. c. bendirei;Fig. 1, Table 1), in the spruce and tamarack forestsof Newfoundland, yetit depends on conifersnormally exploited by White- why is L. c. percna, or a similarlysized crossbill,not wingedCrossbills. White-winged Crossbills also occur foundthroughout similar forests on the mainland?No on Newfoundland,but have not differentiatedinto a othercrossbill species or subspecies in North America distinctsubspecies, and theyapparently move on and is restrictedto a small part of the range of a conifer offNewfoundland depending on the size of the cone species or group of species on which theyforage. crop. Spruce (Picea) has been common on Newfound- land for at least the last 8,000 years (Delcourt and Delcourt, 1987), and black spruce (P. mariana), white WhyHas a Large-billedCrossbill spruce(P. glauca), tamarack(Larix laricina), and bal- Evolvedon an Islandwith sam fir(Abies balsamea) are presentlythe predominant Small ConiferCones? conifers(Damman, 1983). The relativelysmall- and Answeringthis question requiresinformation on co- slender-billedWhite-winged Crossbill (Fig. la) spends nifer-ripeningpattems and crossbillecology (see Benk- most of its time foragingon the relativelysmall cones man, 1987a, 1987b). Cone crops fluctuatetremen- dously fromyear to year (Fowells, 1965; Smith, 1970; Janzen, 1971), and duringsome years, thereare few, I Presentaddress: Departmentof Zoology, Univer- if any, seeds available over large areas. On the main- sityof BritishColumbia, 6270 UniversityBoulevard, land, crossbillsare rarelypresent in a local regionfor Vancouver,BC V6T 2A9, Canada. more than a year,and theyoften remain absent from NOTES AND COMMENTS 1325

9~~~~~~~~~

FIG. 1. Head profilesof foursubspecies of crossbills(a-d), drawn froma photographof museum skins,and open conifercones of fourspecies of conifers(e-h). Crossbills: a = North American White-wingedCrossbill (L. 1. leucoptera),b = NorthAmerican Red Crossbill(L. c. bendirei),c = NewfoundlandRed Crossbill(L. c. percna), d = European Red Crossbill (L. c. curvirostra);conifer cones: e = black spruce(Picea mariana), f= whitespruce (P. glauca), g = white pine (Pinus strobus),h = red pine (P. resinosa). Scale bar indicates 1 cm. extensiveareas foreven greaterlengths of time (Benk- just transientsin some years.Crossbills escape local or man, 1987a; also see Lawrence [1949] and Bailey et regionalcone failuresby moving out (Newton, 1972; al. [ 195 3]). Althoughcrossbills may be foundin a given Smith and Balda, 1979; Benkman, 1987a). area forseveral years in succession,they are probably Based on data fromthe mainland(Benkman, 1 987a),

TABLE 1. Bill dimensions (x ? SE) of five crossbill subspecies (N = 10 individuals for each, except for bill depth formegaplaga whereN= 3). Measurementswere taken as follows:upper mandible lengthwas measured fromthe tip of upper mandible to the anteriorend of the rightnares; lower mandible lengthwas measured from the tip of lowermandible to the base whererami meet;bill depthwas measuredat thenares; and upper mandible widthwas measured at anterioredge of nares.

Mandible lengths Upper mandible Crossbill (locality) Upper (mm) Lower (mm) Total bill depth (mm) width(mm) Loxia 1. leucoptera(North America) 13.8 ? 0.21 9.3 ? 0.16 7.8 ? 0.10 5.6 ? 0.09 L. c. bendirei(North America) 15.3 ? 0.33 12.0 ? 0.29 9.3 ? 0.09 7.4 ? 0.08 L. c.percna(Newfoundland) 14.9 ? 0.18 11.0 ? 0.16 10.6 ? 0.10 8.1 ? 0.09 L. c. curvirostra(Eurasia) 15.3 ? 0.18 11.7 ? 0.11 10.7 ? 0.14 8.2 ? 0.06 L. 1. megaplaga (Hispaniola) 16.2 ? 0.17 12.2 ? 0.19 9.8 ? 0.28 7.8 ? 0.07 1326 NOTES AND COMMENTS

0.6 ~ - leucoptera E 0.5 - @- - bendirei I- .4- percna < 0.4-

0.3- I-~~~~~~~~~~~~~~~0

LU w z 0.1 w Y 0.0o , . . CLOSED OPEN W/FEW SEEDS CONE STAGE FIG. 2. Kernel intake rates (x ? SE) fortwo subspecies of crossbillsforaging on cones of black spruce. The data are forL. 1. leucopteraand L. curvirostrabendirei foraging on 59 cones (973 seeds) and 87 cones (1,417 seeds), respectively.Predicted intake rates forL. c. percnaare indicated by the dotted line (see text). whiteand black spruce and tamarack are presumably cone stages(Benkman, 1987b fig.5c). These data were the most importantseed trees for crossbills on New- obtainedfrom captive crossbills foraging on blackspruce foundland.White spruce and tamarackcone crops fluc- cones gatheredfrom Maine. Also shown are the pre- tuate greatlybetween years (Fowells, 1965; Benkman, dicted intakerates of L. c. percna,based on the size of 1987a). Furthermore,the cone crops of these three its bill (Table 1). It should be noted that the precise conifersoften cycle in synchrony(see Smithand Balda, location of the intake rates of L. c. percna cannot be 1979; pers. observ.); thus,in some years,virtually no assigned,but based on the consistentpatterns for the seeds are available over extensiveareas. For example, two mainland species, certain relations seem reason- in 1987, tamaracksproduced few, if any, seeds in New- able. For example, as bill depth and width increase, foundland(pers. observ.), althoughwhite spruces did seeds in closed cones become more accessible. The produce some seeds along Newfoundland'swest coast deep-billedL. c.percna is predictedto be moreefficient (B. Maybank, pers. comm.). Black spruce cone crops, on closed cones thanare White-wingedCrossbills, which in contrast,rarely fail, and duringmost (if not all) years, have theirhighest intake rates just afterthe cones open. thereare at least some seeds available (Fowells, 1965). Althoughscales of these cones are only slightlysepa- Moreover, at least some black spruce seeds are held rated,White-winged Crossbills can slide theirslender year-roundin the cones (Chai and Hansen, 1952). In bills easily between the scales. Red Crossbills, with yearsof whitespruce and tamarackcone failure,a res- wider bills (Table 1), have their highestintake rates ident crossbillwould have to subsist on black spruce afterthe cone scales have spread fartherapart, even seeds throughoutthe year. Indeed, surveysof birds in though some seeds have been shed. L. c. percna is Newfoundlandfrom 1957 to 1963 characterizedthe hypothesizedto be less efficientthan the other cross- Red Crossbillas preferring"the old growth,black spruce bills on open cones because of its wider bill, and its forest"(H. Deichmann, pers. comm.). On the conti- peak intakerate should be on fullyopen cones (shifted nent,White-winged Crossbills usually forageon black even fartherto the rightin Fig. 2). spruce only fromwinter to early summer,when the The size and shape of L. c. percna's bill is similarto cones are open and seeds are most accessible (Benk- that of the Red Crossbill in Eurasia (L. c. curvirostra) man, 1987a, 1987b). White pine (Pinus strobus)is an (Fig. 1, Table 1). This similarityin bill shape may be importantseed treefor crossbills but is uncommon on a resultof the importanceof seed in closed cones for Newfoundland,and by the early 1900's it had been both subspecies. Seeds in closed pine cones are prob- mostlyremoved by logging(Montevecchi and Tuck, ably morecritical, on average,to Old World Red Cross- 1987). Furthermore,it also fluctuatesin cone-croppro- bills, because Old World pines usually hold mature duction(Fowells, 1965), oftenin synchronywith spruce seeds in closed cones fromfall to earlyspring (Newton, and tamarack(pers. observ.). 1972), whereasmost New World pines have cones that Cone failureson islands cannot be escaped by no- open in the fall (Fowells, 1965). Presumably,this is madic movements. Therefore,crossbills confinedto whyOld World populationsof Red Crossbillsthat for- islands should have bills that provide access to seeds age on pine have much deeper bills than those popu- over the complete cone-ripeningphenology. Large- lations in the New World (see Griscom, 1937). Alter- billed crossbillscan extractseeds more rapidlyfrom a natively,L. c. percna may be a recentcolonist from widerrange of cone sizes and, most importantly,from the Old World (J. Groth, pers. comm.), and the sim- more stages of cone ripeningthan can small-billed ilarityof its bill size and shape to thatof L. c. curvirostra crossbills(Benkman, 1987b). Figure2 shows theintake may result from lack of divergence. This cannot be rates of.mainland Red Crossbills and White-winged confirmedwithout genetic analysis. Nevertheless, L. c. Crossbills foragingon black spruce cones at different percna's persistenceon Newfoundlandand its absence NOTES AND COMMENTS 1327 formthe boreal forestsof the mainland needs expla- TABLE 2. Black spruce cone-scale thicknessfor four nation. sites in Canada. Scale thicknesswas measured at the middle of the scale at the anteriorend of the seed scars. Whyis L. c. percnaConfined to Newfoundland? N equals the numberof trees.For each tree,five scales were measured from each of 5-10 cones. Significant The absence of red squirrels(Tamiasciurus hudson- icus) fromNewfoundland may have been essentialfor differencesbetween means, as indicated by the LSD the evolution of L. c. percna. Red squirrelshave been multiple-rangetest (P < 0.05), are indicatedby differ- a major selective forceon cone structure,particularly ent lowercase letters. on conifersthat produce regularcone crops and hold LSD theirseeds forextended periods (Smith, 1970; Elliott, multi- 1974). For example, in westernNorth America, lodge- Scale thickness(mm) ple- range pole pine (P. contorta)has thickercone scales where Site x SE N test Longitude cone crops are more regularand red squirrelpopula- tions are more stable (Smith, 1970). Consequently,the Manitoba 1.04 0.02 5 b 100?W presenceof squirrelsas a selective agent can have in- W. Ontario 1.11 0.02 5 c 91?W directeffects on otherseed consumers(Benkman et al., E. Ontario 1.15 0.02 6 c 82?W 1984). If black sprucecone scales are thinneron New- Newfoundland 0.96 0.02 6 a 58?W foundlandbecause of the absence of red squirrels,then seeds in closed cones may be more accessible to crossbills on Newfoundland than on the mainland (see squirreldensities in black spruce forestsare probably Benkman, 1987b). considerablymore than twice as greaton Newfound- To determinewhether black spruce cones differbe- land as in comparable forestson the mainland. tweenthe mainland and Newfoundland,I gatheredblack The highersquirrel density on Newfoundlandmay spruce cones from Riding Mountain National Park, be because thereare fewerpredators. However, food Manitoba, western and eastern Ontario, and New- supply,and not predators,generally limits densities of foundland. Cone-scale thicknesswas measured with squirrelsin thegenus Tamiasciurus(Smith, 1968, 1970; digitalcalipers at the anterioredge of the seed scars of Kemp and Keith, 1970; Rusch and Reeder, 1978; Sul- five scales fromthe lower half of each of 5-10 cones livan and Sullivan, 1982). from5-6 trees fromeach of the above fourlocations The presenceof red squirrelsmay have precludeda (Table 2). Scale thicknesswas measured because it is residentL. c. percna-typecrossbill on the mainland. the one scale characteristicof spruce in easternNorth Even thoughL. c. percna is presumablymore efficient America that is uniformlycorrelated with the time on closed black sprucecones than are othersubspecies requiredby crossbills to removeseeds fromclosed cones of Red Crossbillsin NorthAmerica, it may not be able (Benkman, 1987b). to survive for extended periods on mainland black Scale thicknesswas significantlyless forcones from spruce,which have thickercone scales. There are no Newfoundlandthan forthose fromthe othersites (Ta- records of Red Crossbills on the Labrador Peninsula ble 2; LSD [least significantdifference] multiple-range adjacent to Newfoundland,where black spruceis abun- test,P < 0.05), as predicted.Cones fromRiding Moun- dant (Todd, 1963). L. c. percna has been collected on tainhad significantlythinner scales thandid thosefrom AnticostiIsland (National Museum of Canada). These the two Ontario sites,and scale thicknessdecreased to birds,however, probably represent eruptive birds (birds the west in the three mainland samples. The signifi- thatemigrate when theirdensities are high relativeto cance of these last two resultsis unknown. the available food supply in their usual habitat) and Red squirrelswere introducedto Newfoundlandin althougheruptive birds may returnto previouslyoc- 1963 and 1964 (Dodds, 1983). The latterintroduction, cupied habitat (Newton, 1972), thereis no indication by theNewfoundland Wildlife Service, was apparently that theymove regularlybetween Newfoundlandand done to provideprey for the pine marten(Martes amer- the mainland (contra Dickerman, 1986b). Moreover, icana), whichhad been overtrapped.Red squirrelsare thereis no evidence that L. c. percna has persistedon now quite common on Newfoundland(Dodds, 1983; the mainland forextended periods. pers. observ.). For example, the average springdensity Alternatively,White-winged Crossbills may com- fortwo three-hectare plots in northwestNewfoundland petitivelyexclude L. c. percnafrom the mainland.This in 1985 was 3.4 individuals per hectare(R. West,pers. alone, however,is not a sufficientexplanation, because comm.). If black spruce on the mainland has more White-wingedCrossbills do not seem any less common defensesagainst vertebrateseed predators,then lower on Newfoundlandthan in comparable forestson the densitiesof red squirrelsare to be expectedthere. The mainland. springdensities of red squirrelsin spruceforests on the Presumably,red squirrelsare now exertinga strong mainlandrange from 2.0 to 2.5 individualsper hectare selectiveforce on black sprucecone structureon New- (six studiessummarized in Rusch and Reeder [1978]). foundland. If cone-scale thickness increases among These values probablyoverestimate red squirrelden- black spruce, then intake rates for L. c. percna will sitiesin black sprucehabitat on the mainland,because decline. This, in combination with cone removal by theyinclude red squirrelsin species other than black squirrels,a competitiveeffect, may cause the decline, spruceand black spruce is not a preferredconifer (see and possible extinction,of the once plentifulL. c. perc- Rusch and Reeder, 1978). Red squirrelswere foundto na (see Austin [1968] forcomments on L. c. percna's lose bodymass whengiven only cones fromblack spruce formerabundance). In fact,there appears to have been fromthe mainland (Alaska), but not when given only a large decline in Red Crossbills on Newfoundland cones of whitespruce (Brink and Dean, 1966). In sum, since 1957-1963, whenit was characterizedas a "fairly 1328 NOTES AND COMMENTS

TABLE 3. Summary of the seven species/subspeciesof crossbills restrictedto islands. Coniferdistribution is based on Critchfieldand Little (1966) and Mirov (1967); all are species of Pinusexcept Picea mariana.Fire- adapted statusis based on the known adaptations to fire,fire frequency (see text),and whetherthe pines occur in savanna-typehabitats perpetuatedby fire(Mirov, 1967). Squirrel presence is based on Walker (1975) and Brink(1977). Bill size is based on Griscom (1937), Knox (1975; pers. comm.), Massa (1987), and Table 1 (see text).

Bill size Conifer relativeto fire- Squirrel mainland Crossbill Island Conifer adapted? present? counterpart L. c. percna Newfoundland Picea mariana yes yesa large L. 1.megaplaga Hispaniola Pinusoccidentalis yes no large L. c. balearica Majorca P. halepensis yes no small L. c. corsicana Corsica P. halepensis,P. nigra,P. pinaster yes no large L. c. guillemardi Cyprus P. nigra yes no large L. c. luzoniensis Luzon P. insularis yes no small L. scotia Great Britain P. sylvestris ? yes small a Recent introduction(1963-1964). regularresident" (H. Deichmann, pers. comm.). Such cones, but these cones often have thick cone scales a decline is confirmedby data fromannual Christmas (Smith, 1970) that are difficultfor crossbillsto open; bird counts in Terra Nova National Park, Newfound- and 3) it is easier to open closed cones on Newfound- land (Fig. 3). In contrast,the status of the White-winged land, where squirrelshave not selected forthick cone Crossbill, which relies more on the fluctuatingcone scales, than on the mainland. Thus, it is predictedthat crops of tamarack and white spruce, appears to have 1) crossbills are more likely to establish themselves remainedunchanged (H. Deichmann, pers. comm.). It and evolve on islands where conifersproduce cone is ironic thatthis introductionto help one species, the crops everyyear and squirrelsare absent and 2) island marten,has thepotential to cause thedemise ofanother. crossbillpopulations should have largerbills thantheir mainland counterpartsforaging on cones of the same Does thePattern Foundfor the Newfoundland species or on cones of similar size. CrossbillRepresent an IsolatedCase, or There are seven species or subspecies of crossbills Arethe Principles General to OtherIsland restrictedto islands (Table 3). Althoughdetailed eco- Populationsof Crossbills? logical informationis not known for most of these To reiteratethe main points: 1) conifersthat produce islands, some general comments can be made con- at least some cones everyyear and hold seeds through- cerningthe coniferson the islands, the presence of out the yearare criticalfor island populations of cross- squirrels,and the size of the bills of the island popu- bills; 2) thereis selectionfor larger bills to open closed lations of crossbills. First,most of the habitats in which island populationsof crossbills occur experience frequent fires (Table 3). For instance,black spruceis mostcommon on New- 5.0- foundlandwhere firesare frequent(Damman, 1983). Fires are frequentand criticalin maintainingP. occi- I 4.0- dentalisas the dominanttree in the mountainsof His- paniola (Fisher-Meero, 1983). On the Mediterranean islandsof Majorca, Corsica, and Cyprus,fire is frequent cc 3.0- and has been a criticalinfluence in the evolutionof the vegetation(Mooney and Parsons, 1973). Fire-adapted Xl 2.0- O conifersare characterizedby littleannual variationin cone-crop size and by the factthat theyhold seeds in 1.0 .P theircones forextended periods (Fowells, 1965; Smith, d %d 1970). Z ' ono , ,,,x Second, tree squirrels (Tamiasciurus and Sciurus) 0.0 i ?I I I I I 67 68 69 70 71 72737475767778798081 828384888788 and other squirrelsthat eat coniferseeds are absent fromall the islands exceptNewfoundland (where they YEAR were introducedonly about 25 years ago) and Great FIG. 3. The number of Red Crossbills observed Britain(Table 3; Walker,1975; Brink,1977). On Great duringthe annual Terra Nova National Park, New- Britain,squirrels (Sciurus vulgaris)feed on Pinus syl- foundland,Christmas bird count from 1968 to 1988 vestriscones (Nethersole-Thompson,1975; Gurnell, (data fromTerra Nova National Park records).A "par- 1987) and appear to have a negative impact on the ty-hour"is one hourof observationby an independent abundance of L. scotia. This is suggestedby thedecline groupof observers.The numberof party-hoursranged in L. scotia's abundance between 1930 and the 1970's from 12 (in 1971) to 88 (in 1984). The census area (Nethersole-Thompson,1975), whenthe abundance of encompasses extensive areas of undisturbed black squirrels(S. vulgaris)in Scotland was increasing(Gur- spruceforest (pers. observ.). nell, 1987). What is probablyessential for the crossbills NOTES AND COMMENTS 1329 to surviveon Great Britainis sufficientgeographic vari- ativelylarge bills compared to theirmainland coun- ation in cone productionso thatduring any given year terpartsthat rely on conesof similarsize. crossbills can find some areas with good cone crops In summary,there are ratherspecific requirements (see Nethersole-Thompson,1975; Knox, 1987). This forthe evolution of islandpopulations of crossbills, is critical,because in any one area in Great Britain,P. includingthe presence of fire-adapted conifers and the sylvestriscone crops vary greatlyin size fromyear to absenceof squirrels.This impliesthat squirrels, as a year (Nethersole-Thompson,1975). In effect,spatial selectiveagent on cone structure,have had a strong variationprovides a keyingredient more characteristic impacton bothisland and continentalcrossbill pop- of continentsand oftenmissing from (or at least more ulations.The extentof currentcompetition between limitedon) islands. squirrelsand crossbillsfor conifer seeds is unknown, The few islands that commonly support crossbills althoughon Newfoundlandit is now likelyto be in- but lack endemic populations eitherare verywet and tense.On theother hand, the long-term effect of squir- lack fire-adaptedconifers or have squirrels.These is- relson crossbillsmay have been to increasethe di- lands include the Queen CharlotteIslands, which lack versityof cone structures and, hence, of crossbill species squirrelsbut are quite mesic, and Japan, which has and subspecies.The selectivefeeding of squirrelson squirrels. conifershas causeda greatervariety of defense mech- Third, fourisland populations have relativelylarge anismsamong conifer species (Smith, 1970), and pos- or stout bills as compared to theirmainland counter- siblyas a result,different bill structuresare required parts,but threedo not (Table 3). For most of the cross- forforaging efficiently on the more distinct cone struc- bill populations,there is no simple mainland counter- tures(Benkman, 1987b). Moreover, the increase in cone part,because coniferson the islands differin specific defensesbrought about from selective feeding by squir- or subspecificstatus fromthose on the mainland and rels(Smith, 1970) mayultimately be responsiblefor may differin termsof cone size. For example, the pine theevolution of crossbillsfrom less specializedCar- on Hispaniola, P. occidentalis,is not foundon thecon- duelisfinches. tinent.However, the cones of P. occidentalisare similar in appearance to those of P. ponderosa,although they ACKNOWLEDGMENTS are smaller(P. occidentaliscone length:x ? SD = 5.6 P. R. Grant,A. Knox, D. Schluter,C. C. Smith,and + 0.6 cm, range4.6-6.6 cm,N= 20 cones; P. ponderosa two anonymous reviewersmade comments that im- cones are usuallybetween 7 and 15 cm in length[Munz, provedthe manuscript, and A. Knox providedinfor- 195 9]). Consistentwith the idea thatisland populations mation on bill sizes. B. Maybank gatheredand sentme have relativelylarge bills, the crossbill subspecies on thecones fromNewfoundland, for which I am grateful. Hispaniola, L. 1. megaplaga,has a similar although I am also gratefulto H. Deichmann forsending me an slightlylarger bill than the crossbillthat normallyoc- unpublishedmanuscript and to R. West forproviding cupies P. ponderosaforests (L. c. bendirei;Table 1) (see squirrelcensus data fromNewfoundland. The conifer Benkman, 1987a; Dickerman, 1986a). conesand crossbillswere graciously drawn by P. King The crossbillon Majorca (L. c. balearica)has a smaller and D. Nash, respectively.J. Ottenwalder, A. Dod, and bill than its mainland counterpartin northernAfrica D. Dod facilitatedmy researchin the Dominican Re- (L. c. poliogyna)(Massa, 1987). Both subspecies of public. Supportduring this project was provided by an crossbillsforage mostly on Pinushalepensis, which is NSF PostdoctoralFellowship in EnvironmentalBiol- the predominantpine on Majorca and the only pine ogy. in the crossbill'srange in northernAfrica. Pinus hale- pensishas a restricteddistribution in Africa(Critchfield LITERATURECITED and Little, 1966), however, so L. c. poliogynamight AUSTIN, 0. L. 1968. Lifehistories of North American be effectivelyan island population;L. c. poliogyna,like cardinals, grosbeaks, buntings,towhees, finches, the Mediterraneanisland populations,is thoughtto be sparrowsand allies, Part I. U.S. Nat. Mus. Bull. sedentary(Massa, 1987). Moreover,there are probably 237:498-500. more seed predators,especially mammals, in northern BMILEY,A.M., R.J.NIEDRACH,ANDA.L. BAILY. 1953. Africathan on Majorca. As a result,P. halepensismay The Red Crossbillsof Colorado. Publ. Denver Mus. have thickercone scales in northAfrica than on Ma- Nat. Hist. 9:1-64. jorca. If this is true, then a larger-billedcrossbill is BENKMAN, C. W. 1987a. Food profitabilityand the expectedin northernAfrica than on Majorca. foragingecology of crossbills. Ecol. Monogr. 57: The crossbill confinedto Luzon (L. c. luzoniensis) 251-267. also has a smaller bill than its mainland counterpart 1987b. Crossbillforaging behavior, bill struc- (L. c. meridionalis)(Griscom, 1937; A. Knox, pers. ture,and patternsof food profitability.Wilson Bull. comm.; Table 3). However, like L. c. poliogyna,L. c. 99:351-368. meridionaliscan be considered to be effectivelyan is- BENKMAN,C. W., R. P. BALDA, AND C. C. SMITH. 1984. land population (see Critchfieldand Little [1966] for Adaptations for seed dispersal and the compro- the distributionof pines), and thus,similar arguments mises due to seed predationin limberpine. Ecology as given above forL. c. balearicaand L. c. poliogyna 65:632-642. applyto L. c. luzoniensisand L. c. meridionalis. BRINK, C. H ., AND F. C. DEAN. 1966. Spruce seed The Scottishcrossbill (L. scotia)is confinedto Pinus as a food of red squirrels and flyingsquirrels in sylvestrisin Scotland (Nethersole-Thompson,1975), interiorAlaska. J. Wildl. Manag. 30:503-512. and it has a smaller bill (Table 3) than its mainland BRINK,F. H. VAN DEN. 1977. A Field Guide to the counterpart,the ParrotCrossbill (L. pytyopsittacus)in Mammals of Britainand Europe. Collins, London, northernEurope (Newton, 1972). This is the one clear U.K. exceptionto the rule of island populations having rel- CHAI,T. S., AND H. L. HANSEN. 1952. Characteristics 1330 NOTES AND COMMENTS

of Black Spruce Seed fromCones of DifferentAges. LACK,D. 1944. Correlationbetween beak and food School of Forestry,Univ. Minnesota, St. Paul. in the crossbill (L. curvirostra).Ibis 86:552-553. CRITCHFIELD,W. B., AND E. L. LITTLE,JR. 1966. Geo- LAWRENCE, L. DE K. 1949. The Red Crossbills at graphicDistribution of the Pines ofthe World. Mis- Pimisi Bay, Ontario.Can. Field-Natur.63:147-160. cellaneous Publication 991. U.S. Department of MASSA,B. 1987. Variation in Mediterraneancross- Agriculture,Wash., DC. bills Loxia curvirostra.Bull. Brit. Ornithol. Club DAMMAN,A. W. H. 1983. An ecological subdivision 107:118-129. of the island of Newfoundland,pp. 163-200. In G. MIRov, N. T. 1967. The genus Pinus. Ronald, N.Y. R. South (ed.), Biogeographyand Ecology of the MONTEVECCHI,W. A., ANDL. M. TUCK. 1987. New- Island of Newfoundland.Junk, The Hague, Neth. foundlandbirds: Exploitation, study, conservation. DELCOURT,P. A., AND H. R. DELCOURT. 1987. Long- Nuttall OrnithologicalClub, Cambridge,MA. termforest dynamics of the temperate zone. Spring- MooNEY, H. A., AND D. J. PARSONS. 1973. Structure er-Verlag,N.Y. and functionof the California chaparral-An ex- DICKERMAN,R. W. 1986a. A review of the Red ample fromSan Dimas, pp. 83-112. In F. di Castri Crossbill in New York state,Part 2. Identification and H. A. Mooney (eds.), MediterraneanType Eco- of specimens from New York. Kingbird 36:127- systems:Origin and Function.Springer-Verlag, N.Y. 134. MUNZ,P. A. 1959. A California Flora. Univ. Calif. 1986b. The "old northeastem"subspecies of Press, Berkeley. Red Crossbill. Amer. Birds 41:188-194. NETHERsoLE-THOMPSON, D. 1975. Pine Crossbills:A DODDS, D. 1983. Terrestrialmammals, pp. 509-550. ScottishContribution. Poyser, Berkhamsted, U.K. In G. R. South (ed.), Biogeographyand Ecology of NEwToN, I. 1967. The adaptive radiationand feeding theIsland ofNewfoundland. Junk, The Hague, Neth. ecology of some Britishfinches. Ibis 109:33-98. ELLIOTT, P. F. 1974. Evolutionaryresponse of plants . 1972. Finches. Collins, London, U.K. to seed-eaters:Pine squirrelpredation on lodgepole RUSCH, D. A., AND W. G. REEDER. 1978. Population pine. Evolution 28:221-231. ecology of Alberta red squirrels.Ecology 59:400- FISHER-MEERO,L. L. 1983. Floristics,ecology, phy- 420. togeography,and historyof botanical exploration SMITH, C. C. 1968. The adaptive nature of social of the Sierra de Bahoruco, Dominican Republic. organization in the genus of tree squirrels, Ta- M.S. Thesis. Univ. Florida, Gainesville. miasciurus.Ecol. Monogr. 38:31-63. FOwELLS,H. A. 1965. Silvics of Forest Trees in the 1970. The coevolution of pine squirrels United States. AgriculturalHandbook 271. U.S. (Tamiasciurus)and conifers.Ecol. Monogr.40:349- Departmentof Agriculture,Wash., DC. 371. GRISCOM,L. 1937. A monographicstudy of the Red SMITH, C. C., AND R. P. BALDA. 1979. Competition Crossbill. Proc. Boston Soc. Nat. Hist. 41:77-210. amonginsects, birds and mammalsfor conifer seeds. GURNELL,J. 1987. The natural historyof squirrels. Amer. Zool. 19:1065-1083. Facts on File, N.Y. SULLIVAN, T. P., AND D. S. SULLIVAN. 1982. Popu- JANzEN,D. H. 1971. Seed predationby animals. Ann. lation dynamics and regulation of the Douglas Rev. Ecol. Syst. 2:465-492. squirrel(Tamiasciurus douglasii) with supplemen- KEMP,G. A., ANDL. B. KEITH. 1970. Dynamics and tal food. Oecologia 53:264-270. regulationof red squirrel (Tamiasciurus hudsoni- TODD, W. E. C. 1963. Birds of the Labrador Penin- cus) populations. Ecology 51:763-779. sula and adjacent areas. Univ. Toronto Press, To- KNox, A. 1975. Crossbill taxonomy,pp. 191-201. ronto,Canada. In D. Nethersole-Thompson(ed.), Pine Crossbills: WALKER,E. P. 1975. Mammals of the World, Vol. A ScottishContribution. Poyser, Berkhamsted, U.K. II, 3rd Ed. JohnsHopkins Univ. Press, Baltimore, 1987. Crossbills: Their general biology and MD. some conservation problems, pp. 64-74. In E. Cameron (ed.), Glen Tanar: Its Human and Natural History. [Intemal report,Scotland Northeast Re- gion].Nature Conservancy Council, Aberdeen, U.K. CorrespondingEditor: K. F. Liem