SHORT COMMUNICATIONS

Breedingof Darwin's Finchesat an UnusuallyEarly Age in an E1Nifio Year

H. LISLEGIBBS, PETER R. GRANT, AND JON WEILAND Divisionof BiologicalSciences, The University of Michigan,Ann Arbor,Michigan 48109-1048 USA

The age at which an organismfirst breedshas an skull characters,that passerinebirds canbreed in the important effecton its relative fitness. wild when about 3 months old (see also Orr 1945). usually breed for the first time at an age of 1 yr or Early breeding may be common in regions where more. In arid regionsof the world they may breed suitableconditions for breedingpersist for more than in the year of their birth. This is suggestedby labo- 3 or 4 months, but the only record we have been able ratory studiesof reproductivematuration (Marshall to find of a banded in the wild breeding at a and Disney 1957;Sossinka 1980a, b) and by occasion- comparablyearly age is of a female Zebra Finch (Po- al observations of birds in the wild (Immelmann 1963, ephilaguttata) in Australia, which laid her first egg Davies 1977, Kikkawa 1980). Here, we report the when she was 86 daysold (Immelmann 1963). breeding of two speciesof Darwin's fincheson the Geospizafortis individuals breeding in their year of GalfipagosIslands at an age of 3-6 months.This is birth were younger, on average, than their G. scan- the first demonstrationof unusuallyearly breeding denscounterparts, as is revealed by Mann-Whitney and its short-termconsequences in wild birds whose U- and t-tests (data In-transformed to correct for agesare preciselyknown. The resultshave important skewness, t•ss= 2.11, P < 0.05; see also Table 1). This implicationsfor the of maturation ratesin maybe relatedto their different growth rates;asymp- birds. totic size is reachedin lessthan 60 days in G. fortis, Populationsof the Medium Ground Finch (Geospi- but some dimensions in G. scandens increase for 80 za fortis)and CaEtusGround Finch (G. scandens)on days (Boag 1983). Isla Daphne Major have been studiedevery year since The parents of these birds bred for the first time 1975,and almostall individuals have been uniquely when they were 22-24 months old on average(i.e. colorbanded. Breeding of the finchesis governedby nearly 2 yr old); the two speciesdid not differ in this rainfall and the ensuing production of plants and respect(ts0 = 0.90, P > 0.1; Table 1). Despite the great arthropods(Grant and Boag1980). From 1976to 1982, difference between the 1982-1983 cohort and their the maximum annual total of rain was 137 mm, which parents, there is a significant correlation between fell over a 4-month period in 1978 (Boagand Grant them in age at first reproduction.We show this for in press).In late November 1982, rain began failing G. fortisby using mid-parentvalues (i.e. the average heavily, associatedwith an E1 Nifio event that has of the agesof motherand fatherwhen they firstbred) beendescribed as the mostsevere of the century(Cane and family mean values for the offspring born in 1983, Kerr 1983). Rain continued until mid-July 1983. 1982-1983 who subsequentlybred. The parametric The rainfall total on I. Daphne for the 8-month pe- correlation coefficient for 15 families is 0.534 (P < riod was 1,359mm, i.e. 10 times the previousannual 0.05). Correlations based on father alone (0.464) or maximum. mother alone (0.496) are not quite significant. The Finchesstarted laying eggs in late November and significant correlation is consistent with the possi- early December 1982. The last nestling fledged in bility that the relative age at first reproductionis early September 1983. During this 9-month period, partly under genetic control. individual pairs bred as many as seven times; the More female (106) than male (39) G. fortisbred in previousmaximum had been five clutchesproduced their firstyear, but the majorityof both sexesdid not by a single pair of G. scandensin 1978.We found all breed.In the absenceof externaldistinguishing fea- the nests,uniquely banded the chicks at 8 days of turesof the sexesof theseyoung birds, we assumea age with one metal and three color bands,identified 1:1 sex ratio of young available to breed. Proportion- the parentsby their bands,and monitored the suc- ally more femalesbred than would be expectedby cessof eachbreeding attempt. chance (x2x= 27.7, P < 0.001). Six male and 39 female From April 1983 onwards the breeding popula- G. scandensbred in their first year, the proportions tions of both specieswere augmentedby birds born againdiffering from an expected1:1 ratio (x2•= 24.2, a few monthsearlier. Ages at firstreproduction (clutch P < 0.001).In addition, G. fortisfemales, with an av- initiation) are displayedin Fig. 1. The youngestG. erage age at first reproductionof 109.2 + 21.0 (SD) fortisattempting to breed was 81 days old, and it days,bred earlier than males(131.9 + 24.3days; t,43 fledged two young. The youngestG. scandenswas 89 = 5.50, P < 0.001). Geospizascandens females and males, daysold, and it fledgednone. The youngestG. scan- however, did not differ in age at first reproduction densto breedsuccessfully laid her first egg when she (t43= 0.13, P > 0.1). was 97 days old. Thesevalues confirm the supposi- Thesedifferences between sexes are not likely to tion, based on estimatesof age from and be causedby intrinsicdifferences in maturationrates,

872 October 1984] Short Communications 873

30

• 18 . % 6

80 100 120 140 160 180 200 • 12 AGEINDAYS

• G.SCANDENS 4

Fig. 1. Ages at first reproduction of female (solid bars) and male (open bars) finchesborn in the 1982- 1983breeding season. Ages are from birth (hatching)to the day the first egg was laid.

because other studies have shown that male birds Both G. fortis and G. scandensbred successfullyin generallyproduce gametes at an earlier age than do their first year when paired with older birds (Table females(Immelmann 1963, Serventy 1971, Sossinka 2). To compare their performance with the perfor- 1980b).Rather, they probablyreflect greater oppor- mance at the same time of older birds of the same tunities for the breeding of females than of males, speciesthat had bred previously,we first combined resulting from a male-biased sex ratio in the adults clutchesinto three groupsof comparablesample sizes: of both species.The bias arosethrough differential 1-2 eggs,3 eggs,and 4-6 eggs.For G. fortis,old fe- mortality during the drought of 1977 and has per- malespaired with old maleslaid proportionallymore sistedever since (Boagand Grant 1981, Price 1984). large clutchesthan did young femalespaired with There were more mates available for females than for either old males(X22 = 7.78, P < 0.02) or young males males, and males had the additional task of establish- (X22= 13.22,P < 0.001).We then comparedpairs that ing territories. hatched at least one egg (Table 2) but fledged no offspring with those that fledged one or more off- spring. Again, the old pairs of G. fortisdid propor- tionally better than either the mixed pairs (Fisher's TABLEl. Mean agesat firstreproduction in days(+ 1 ExactProbability = 0.0004)or young pairs (P = 0.029). SD) for birds breeding in their year of birth and In contrastto G. fortis,there were no detectabledif- for their parents.Ages were determined from the ferencesin breeding successbetween young and old dateon which the first egg was laid by the birds G. scandens(see also Table 2). or by their mates.Only data from parents(•90%) The relatively low reproductive successof young of exactlyknown agesare included.Males and fe- G. fortisappears to be due to their youth rather than maleshave been combined.The range of agesare shown in parentheses;n is samplesize. to their inexperienceat breeding. We reach this con- clusionfrom the absenceof any detectabledifference The 1982-1983 in reproductivesuccess between G. fortis females (n = Species cohort Their parents 18) born in 1981 and breeding for the first time in December1982 or January1983 and older (experi- G. fortis 115.3 + 24.12 745.4 + 247.7 enced)females (n = 41) breedingin the samemonths (81-183) (365-1,154) n = 145 n = 68 (P > 0.! in each case).This does not preclude the possibilitythat the low reproductivesuccess of young G. scandens 125.8 + 27.7 680.6 +_ 231.8 G. fortisin 1983 was attributable,in part, to a con- (89-203) (415-1,157) n = 45 n = 14 trolled, sub-maximal,reproductive effort (cf. Curio 1983). 874 ShortCommunications [Auk,Vol. 101

T^•I•E2. Breedingsuccess of pairs containing0, 1, or 2 birds born in 1982-1983(Young) or in 1981and earlier (Old) and that hatchedat leastone egg. No pairscomprised young malesand old females,and none containedbirds born in the normal breedingseason of 1982(they had all died). Data for old pairswere restrictedto the period when young birds were breeding (n is samplesize).

Pair type Young • x Young • Old • x Young • Old • x Old • G. fortis n 10 32 123 Mean clutch size + SD 3.40 + 0.5 3.70 + 1.7 4.0 + 1.2 Mean number of fledglings + SD 0 0.75 _ 1.2 0.94 + 1.3 G. scandens

n 33 102 Mean clutch size + SD 3.76 + 1.6 3.81 + 1.0 Mean number of fledglings + SD 0.93 + 2.1 0.99 + 1.3

In only 1 yr out of 8 was the wet seasonlong CANE,M.A. 1983. Oceanographicevents during E1 enough to permit finches to breed in the seasonof Nitto. Science 222: 1189-1195. their birth. This frequencymay be misleadinglyhigh. CuRio,E. 1983. Why do young birds reproduceless Since 1950 (excluding 1983), the wet seasonhas last- well? Ibis 125: 400-404. ed for more than 4 months only three times in the DAVIES,$. J. J.F. 1977. The timing of breeding by coastalareas of Islas San Crist6bal and Santa Cruz, in the Zebra finch (Taeniopygiacastanotis) at Mileu- other words about once a decade. Individual finches ra, Western Australia. Ibis 119: 369-377. of both speciescan live for more than 10 yr in the GRANT, P. R., & P. T. BOAG. 1980. Rainfall on the wild (unpubl. obs.) The chief implication of these Galhpagos and the demography of Darwin's resultsis that rare eventsare important selectivede- finches. Auk 97: 227-244. terminants of maturation. The breeding of young IMMELMANN,K. 1963. Drought adaptationsin Aus- birds paired with old birds was remarkably success- tralian desert birds. Proc. 13th. Intern. Ornithol. ful, but whether or not it extracted a cost in terms of Congr. 1962: 649-657. lowered probability of survival during the non- KERR,R.A. 1983. Fading E1 Nitto broadening sci- breeding dry seasonor subsequentlyremains to be entists' view. Science 221: 940-941. seen. KIKKAWA,J. 1980. Seasonalityof nestingby Zebra We thank RosemaryGrant, Trevor Price,and Dolph Finches at Armidale, NSW. Emu 80: 13-20. Schluter for advice. The work was carried out with MARSHALLßA. J., & H. J. DE$. DISNEY. 1958. Exper- the permission of the Direcci6n General de Desarro- imental induction of the breeding seasonin a 11oForestal, Quito, and with the help of the Charles xerophilous bird. Nature 180: 647-649. Darwin ResearchStation. It was supported by NSF ORR,R.T. 1945. A study of captiveGal•pagos finch- grant BSR-82-12525to P.R.G. es of the genusGeospiza. Condor 47: 177-201. PRICE,T. D. 1984. Sexual selection on body size, territory and plumage variablesin a population LITERATURE CITED of Darwin's Finches. Evolution 38: 327-341. $ERVENTY,D. 1971. Biology of desertbirds. Pp. 287- BOAG,P.T. 1983. The heritability of externalmor- 339 in Avian biology (D. S. Farher and J. R. King, phology in Darwin's Ground Finches() Eds.). New York, Academic Press. on Isla Daphne Major, Galfipagos.Evolution 37: $OSSINKA,R. 1980a. Reproductivestrategies of es- 877-894. trildid finches in different climatic zones of the ß & P. g. GRANT. 1981. Intense natural selec- tropics: gonadal maturation. Proc. 17th. Intern. tion in a population of Darwin's Finches (Geo- Ornithol. Congr. 1978: 493-497. spizinae) in the Galfipagos.Science 214: 82-85. 1980b. Ovarian developmentin an oppor- , & --. In press. Darwin's Finches (Geo- tunisticbreeder, the Zebra Finch (Poephilaguttata spiza)on IslaDaphne Major, Gal•pagos: breeding castonotis).J. Exp. Zool. 211: 225-230. and feeding ecologyin a climaticallyvariable environment. Ecol. Monogr. Received10 February1984ß accepted I May 1984.