THE BREEDING OF AUDUBON'S (PUFFINUS LHERMINIERI) IN THE GALAPAGOS•

D.W. SNow

FROMFebruary 1963 to April 1964, in the courseof studieson the be- havior and breedingseasons of Galapagossea-, my wife and I made fortnightlyvisits to SouthPlaza Island, an islet just off the east coastof Santa Cruz Island. Most time was devoted to the Fork-tailed Gull (Creagrusfurcatus), but a considerablenumber of nests of Audubon's Shearwater(Puffinus lherminieri) were also found and regularly checked, and the adult and nestlingshearwaters were bandedwhen they were ac- cessible.I am gratefulto my wife for much help in the field, and to the CharlesDarwin Foundationfor the Galapagosand UNESCO for making the work possible. As regardsbreeding periodicity, the mostinteresting result of thesevisits was the discoverythat both the Fork-tailed Gull and Audubon'sShear- water have non-annualcycles, though this is not obviousin either species becausebreeding is continuous.The presentpaper deals with our observa- tions on Audubon'sShearwater; the data on the Fork-tailed Gull will be dealt with more fully in a separatepublication. In the Galapagos,Audubon's Shearwater (belonging to an endemicsub- species,P. 1.subalaris) is a residentand abundantspecies. It is very much a coastalbird, beinguncommon far from land. When sailing amongthe islands,one frequentlypasses groups of them feedingon shoalsof small .The birdsare in constantcirculation, flying upwind,landing in the water amongthe shoal,ducking the head and front part of the body under water as they lungeat their prey, or submergingcompletely, then fluttering on a few yards to alight again, and so on until they reach the windward edgeof the shoal,whereupon they fly round and comeup again on their prey from downwind. These feeding groups are often accompaniedby Noddies (Anousstolidus), which flutter amongthem and dip down to take the sameprey. Accordingto Loomis (1918), who gives a goodaccount of the Galapagossubspecies (quoted by Murphy, 1936), crustaceansare also taken, and at times the birds dive for as long as a minute. Audubon'sShearwater has previouslybeen reportedas breedingin the Galapagosin several different months. Loomis (1918) concludedthat somebreeding went on throughoutthe year, as also did Lack (1950), and this was confirmedby L•v•que (1964). In generalnesting behavior the Galapagosbirds do not appear to differ much from their closerelatives in the . A singleegg constitutesthe clutch.

t Contribution No. 21, CharlesDarwin Foundation for the Galapagos.

591 The , 82: 591-597. October, 1965 592 S•ow,Audubon's Shearwater in the Galapagos [ Vol.Auk 82 ,2!

,RSRJ:R:R:I

MJJAS OND'JFMA 1962 1963 1964

Figure 1. Breeding periodicity of Audubon's Shearwater at South Plaza Island, November, 1962, to April, 1964. The complete histogram shows the months in which the egg was laid in all nests recorded. Nests which were first used in December 1962- April 1963 and then re-used are shaded distinctively for each month: those marked R are nestsin which it was proved by banding that one or both of the original pair took part in the second nesting.

On South Plaza Island most of the nestsare in holes,recesses, or short tunnelsin the brokenlava cliff-face, from a few feet abovehigh tide level to the cliff-top (about 60 feet). There are somenests among the boulders at the foot of the cliffs, and a few were found in holesa few feet back from the top of the cliffs. Many of the nestsare inaccessible. The South Plaza shearwatersare preyed upon throughoutthe year by the Galapagosowl, Asio fiammeusgalapagoensis. During our visits single owls visited the island at dusk from the neighboringNorth Plaza Island 200 yards away: previously,we heard, a pair had nestedon South Plaza Island itself. In the courseo.f 15 months,the wings of over 10.0 freshly killed shearwaterswere collected,as well as smaller numbersof Ocea- nodroma castro, the only other breedingtubenose. In view of this heavy predationby owls, it is of interest that Audubon'sShearwater is strongly diurnal in the Galapagos,though clearly not exclusivelyso as the owl does not hunt by day.

THE BREEDING SEASON Figure 1 showsthe monthly distributionof 96 nestsstarted from Novem- ber, 1962, to April, 1964. The numbersrecorded from November,1962, to January,1963, are relativelytoo low, as our visitsdid not beginuntil Febru- 1965Oct. ] S•ow,Audubon's Shearwater in the Galapagos 593

TABLE 1 IIgTERVALS BETWEEIg LAYIIgP. DATES AT 30 NEST SITES

Interval in Months Total 6 7 8 9 10 11 12

After successful nesting - 1 3 11 3 2 - 20 After unsuccessful nesting attempt 1 2 2 3 - 1 1 10 Total 1 3 5 14 3 3 1 30 ary, so that of the nestsbegun earlier we couldrecord only thosethat had surviveduntil February. In addition, sincenew holescontinued to be dis- coveredthroughout the courseof the work, the numbersrecorded early in the period are somewhattoo low relative to the numbersrecorded at the end of the period. Thesesources of bias do not, however,invalidate the conclusionthat breedingdoes indeed continue throughoutthe year (the gap in June, 1963,being dearly due to chance),but that there were, in this 18-monthperiod, major "peaks" and "troughs"of egglaying. There was no apparentconnection between the amount of breedingand the state of the weather, which was fairly typical of the average. From February to May the weatherwas calm, hot, and sunny. June marked the transition to the "garua" season,characterized by much stronger winds and more overcastskies and a reductionin sea temperature. December becamehot again, and generallyhot weather, with weak winds, continued until the study ended. Rainfall in the low-lyingPlaza Islandsis extremely slight and can hardly be of importance.The peaks and troughsof egg laying may have had someconnection with changesin the availability of food, but we have no data on this. The most striking aspectof the breedingperiodicity is, as already men- tioned, that it is non-annual: individual birds have a cycle of around 9 months. There were 20 nest siteswhich were re-usedafter a previoussuc- cessfulnesting. The intervals between the two laying dates ranged from 7 to 11 months,with a pronouncedpeak at 9 months (Table 1). In 9 out of these20 casesbanding showedthat at least one member of the origi- nal pair was involvedin the secondnesting, and there was no case of a ringed adult, which had bred successfully,renesting after any different interval. These resultsare also showngraphically in Figure 1, from which it can be seenthat the pronouncedpeak of breedingfrom October,1963, to January, 1964, was due in largepart to the renestingof birds whichhad bred successfullyin the lesswell pronouncedpeak of Januaryto April, 1963. 594 S•ow, Audubon'sShearwater in the Galapagos [ Auk L Vol. 82

There were 10 nest sites in which an unsuccessfulnesting attempt was followedsome time later by another nesting. In 8 of these 10 casesthe intervals betweenthe two laying dates were from 6 to 9 months; the other two werelonger, 11 and 12 months(Table 1). In onecase• banding showed that the same individual was involved in the two nestings (interval, 8 months). There was no evidencefor renestingafter any shorter intervals. There were, however,two caseswhere the intervals between the laying dates of successiveeggs in the same hole were unusuallyshort, but they were exceptionswhich prove the rule. In the first case,with a two-month interval betweeneggs (the first egghaving disappeared),it was provedby bandingthat a different pair had taken over the nest hole. In the second casethe nest hole was used four times, with unusuallyshort intervals be- tweenall the layings. The dateswere, however,exactly those that would have been expectedif two pairs were alternating at this site. Unfortu- nately, this couldnot be proved,as the hole was too deep for the birds to be banded. The incubationand fledgingperiods are long, about 50-53 and 70-80 days respectively(see next section). Nevertheless,the average interval betweensuccessive layings appearsto be little different whether the first nestingattempt fails or whetherit is successful.Hence the internal process leadingto regressionof the gonads(and presumablyalso the molt) must be initiated very early in the nestingcycle and then proceedunaffected by the subsequentfate of the nestingattempt. Indeed,since eggs that were lost were never found to be replaced,it is probable that the onset of re- gressionof the gonads is determined at the time of laying of the egg. However, other factorsmust also be involved in determiningthe date of the next breedingattempt, since the intervals betweenlaying dates are quite variable, ranging from 6 to 12 months, and moreover seem to be more variable when following unsuccessfulthan when following successful nestings(Table 1). External, and perhapsalso social, factors probably play a part. Other- w/seit wouldbe very difficult to accountfor the markedlyuneven distribu- tion of nests throughoutthe year; instead, whatever the original pattern, after a certain length of time a more or less even distribution of nests throughoutthe year would surely result. Figure 1 showsthat 17 success- ful nestings,distributed somewhat unevenly in the five months from De- cember,1962, to April, 1963, were followedby a more compactgroup of 17 renestingsin the four months from October, 1963, to January, 1964. This suggeststhat, although the secondpeak was, as mentionedabove, in part a simple consequenceof the fact that there had been a high level of breedingsome 9 months earlier, it was accentuatedby someother factor. For a few weeksbefore the main peaks of egg laying, we noted a marked 1965Oct. ] S•ow,Audubon's Shearwater in the Galapagos 595 increasein a kind of aerial display. Groupsof adults would spendmuch time whirlinground near the cliff face, flying up to it as if to enter their nestsand then wheelingaway, callingexcitedly the whole time. This be- haviorwas not givendetailed study, but it seemspossible that it may lead to somedegree of synchronizationbetween pairs that are about to breed and so contributeto the unevendistribution of neststhroughout the year.

INCUBATION AND FLEDGING PERIODS Sinceour visitswere fortnight]y,and on eachvisit we stayedfor only one to three days,it was not possibleto obtain very precisefigures for the length of the incubationand f]edgingperiods. At a few nests,however, where the hatching date was obtained fairly exactly, the limits of error were reducedsufficiently for the figuresto be of somevalue. The incubationperiods so obtainedwere: 48 q- 7, 49 ñ 7, 56 q- 7, and 56 ñ 8. These are a]] consistentwith a period of 48-55 days, with the averagefailing probably between50 and 53 days, which agreeswith the 51-day incubationperiod of P. 1. lherminieri in Bermuda (Wingate in Palmer, 1962). The very similar (P. assimilis)has an incubationperiod of 52-53 days (G]auert, 1946), the larger Manx Shear- water (P. puffinus) an averageperiod of 51 days (Lock]ey, 1942), and the Mutton- (P. tenuirostris) a period of 52-55 days (Serventy, 1958). The youngbird becomesmore mobile in the last week or two before it leavesthe nest, and in somedeep holes,or holeswith ramifications,the young tend to retreat out of sight and it is sometimesdifficult to be certain whetheror not a young bird is presentaround the time when it is nearly ready to leave. For this reasonit is possiblethat one or two of the re- cordedfledging periods may be too short. The following twelve periods were recorded: 68ñ7 71ñ 7 76ñ7 69ñ7 73ñ 7 83ñ? 70ñ6 75ñ 10 83ñ8 70ñ6 76ñ 7 88ñ8

Clearly the period must be variable, with a minimum rangeof from 75 to 80 days; probably the great majority fall between 70 and 80 days. Thus it appearsto be a little longer than that of P. 1. lherminieri in Bermuda (about 72 days), P. assimilis(70 days), and P. puffinus (average,72 days). In severalcases, one or both adults were found in the hole on the last visit beforethe youngbird left the nest,whereas adults were very seldom found in the nest with half-grown young. And occasionallyone or both adults were found in the hole on the visit after the young bird had de- parted. Hence it seemsthat the adults' tendencyto spendtime in the nest 596 Snow,Audubon's Shearwater in the Galapagos [ Vol.Auk 82 hole at the end of the nestingcycle is to someextent independentof the presenceof their offspring.

NESTING SUCCESS

Of 53 neststhat could be analyzedfor success(those found with eggs, or beforethe eggwas laid, excludingthose in which the egg was laid after December,1963), 31, or 59 per cent,were successful. The causesof failure could not be properly ascertained;in somenests the egg was deserted,in othersthe eggor nestlingdisappeared, and in three neststhe nestlingwas found dead. Marine iguanas (Amblyrhynchuscristatus), which scrambleall over the cliffs of the Plaza Islands and shelter in the crevices,may have caused somelosses: they were sometimesfound lurking in nest holes that had been occupiedby shearwaterson the previous visit. Other nestsmust be lost when one of the breedingadults is taken by an owl. In addition to the lossesof known nests, six dead chicks were found lying at the foot of the cliffs, two with their headsbitten off, and two more chicks were found, still alive, crouchingat the foot of the cliffs. It may well have been that one of the parents of these chickshad been taken by an owl, and that hungerhad driven them to leave the nest. The two that were decapitatedhad probablybeen found by an owl after they had fallen to the foot of the cliff.

DISCUSSION

Whenevera bird populationis found to breed more or less evenly throughoutthe year, thereis a primafacie case for suspectingthat the in- dividuals'breeding cycles may be non-annual.For if seasonalvariation in the environmentalfactors that affect breedingsuccess is so slight, or un- predictable,that selectionhas producedno restrictedbreeding season, thereis no reasonwhy individualsshould not revert to an internallycon- trolled cycle. SinceAudubon's Shearwater in the Galapagoswas already knownto breedin all months,it is thus not surprisingto find that its cycle is less than annual. It should be stressedthat the reasonfor the non-annualcycle must be connectedwith the lack of any effective"ultimate" factors (those factors in the externalenvironment that affect breedingsuccess). There is no lack of "proximate"factors which could serve to time the onseto.f breeding. Most Galapagosseabirds have normal annual breedingseasons, and in some,such as the ,Diomedea irrorata, and the ,Pterodroma phaeopygia,the startingdate apparentlyshows little variationfrom year to year. 10•6t•] SNOW,Audubon's Shearwater inthe Galapagos 597

For the present,however, it is quite unclearwhy Audubon'sShearwater, and alsothe Fork-tailed Gull, whosefood and feedinghabits are quite dif- ferent, shoulddiffer in this way from mostother Galapagosseabirds. The answerwill come only from an intensiveinvestigation of seasonalchanges in the marine environmentof the Galapagos,the study of which has only just begun.

SUMMARY

Regular checkswere made on the nestsof Puffinus lherminieri on South Plaza Island, Galapagos,from February, 1963, to April, 1964. Breedingoccurs throughout the year, but individual birds were found to have a cycle averagingabout 9 months. The averagelength of the in- tervals betweensuccessive nesting attempts did not appear to be affected by successor failure of the first attempt, but intervals followinga failure were more variable. There was in additionsome degree of synchronizationbetween pairs, of unknown cause, leading to peaks and troughs of breeding in different months. Knowledgeof seasonalchanges in the marine environmentof the Gala- pagosis insufficientto showwhy P. lherminieri,in contrastto most other Galapagosseabirds, should be emancipatedfrom a fixed breedingseason.

LITERATURE CITED

GLAUERT,L. 1946. The Little Shearwater's year. Emu, 46: 187-192. LACK, D. 1950. Breeding seasonsin the Ga]apagos. Ibis, 92:268 278. L•V•QUE, R. 1964. Notes sur la reproduction des oiseauxaux I]es Galapagos. AIauda, 32:5 44, 81-96. LOCKLE¾,R.M. 1942. . London, J. M. Dent. Loo3,flS,L.M. 1918. A review of the ,, and diving petrels. Proc. Calif. Acad. Sci., ser. 4, 2: 1-187. MVRP•¾, R. C. 1936. Oceanic birds of South America. New York, Amer. Mus. Nat. Hist. PA•VrER,R. S. (ed.). 1962. Handbook of North American birds. Vol. 1. New Haven, Connecticut, Yale Univ. Press. British Trust for Ornithology, Witherby House, Tring, Hertfordshire, England.