Burrowing seabirds affect forest regeneration, Rangatira ...tomtit (Petroica macrocephala chathamensis), marinered-crowned parakeet (Cyanoramphus novaezelandiae chathamensis), warbler

208 NEWZEALANDJOURNALOFECOLOGY,VOL.31,NO.2,2007

New Zealand Journal of Ecology(2007)31(2): 208-222©NewZealandEcologicalSociety

Availableon-lineat:http://www.newzealandecology.org/nzje/

Burrowingseabirdsaffectforestregeneration,RangatiraIsland,ChathamIslands,NewZealand

CynthiaM.Roberts1,2*,RichardP.Duncan1andKerry-JayneWilson11Bio-ProtectionandEcologyDivision,POBox84,LincolnUniversity7647,Lincoln,NewZealand2Currentaddress:SchoolofGeographyandEnvironmentalStudies,UniversityofTasmania,PrivateBag78Hobart,Tasmania7001,Australia*Authorforcorrespondence(Email:[email protected])

Publishedon-line:3December2007

___________________________________________________________________________________________________________________________________

Abstract: TheforestsofRangatiraIsland(218ha)intheChathamIslandsareacriticalbreedingsiteforanumberofrareandthreatenedforestbirdspecies,butarealsohometomorethanthreemillionseabirds,whichcouldsignificantly affect forest regeneration processes. We surveyed the forests of Rangatira Island by establishing 40permanentforestplots,estimatedseabirddensitythroughburrowcounts,andanalysedsoilproperties.Todetermineifseabirdswereimpactingonforestregeneration,weestablishedexclosures(0.25m²)in30oftheforestplots,andexaminedtheroleofcanopygapsinforestregeneration.Thetallestcurrentforest(c.15m),dominatedbyPlagianthus chathamicus,hasmostlyregeneratedsincestockwereremovedin1959.Meanburrowdensitywasestimatedtobe1.19persquaremetre,allsoilswerehighlyacidic(pH3.36–5.18),andburrowdensitywaspositivelycorrelatedwithsoilphosphorus.Seedlingdensityofwoodyspeciesinseabirdexclosuresmeasured after 9, 24 and 33 months was significantly higher than in the adjacent non-gap plots, and seedling density was positively associated with reduced canopy cover. Seedling densities were also significantly higher in canopy gaps than in adjacent non-gap plots, but seabird burrow density was significantly lower in gaps. These resultssuggestthatcanopygapsallowforestregenerationdespitethenegativeimpactsofseabirdburrowing.However,thegapmakers,largelysenescingOlearia traversii,areslowlydisappearingfromtheforests.ThecohortofPlagianthusthathasregeneratedfollowingfarmabandonmentmayprogressivelycollapse,allowingregenerationtocontinueinsmallopenings,butthereisalsothepotentialforacatastrophicblowdown.Thismighthaveseriousimplicationsforforest-dwellingbirds,invertebrates,andplants.___________________________________________________________________________________________________________________________________

Keywords: canopygap; forestcomposition; foreststructure;seabird trampling;seedlingdisturbance;soils;treefall

IntroductionRangatira(SouthEast)IslandintheChathamgroup(44°20´ S, 176°10´ W) is a small island (218 ha)that ranks among the Southern Hemisphere’s mostimportant wildlife sanctuaries. Most of the islandis forested and these forests support the largest orthe only breeding populations of several rare andcriticallyendangeredbirdspecies,namely,theblackrobin (Petroica traversii), and the other ChathamIsland endemics the snipe (Coenocorypha pusilla), tūī (Prosthemadera novaeseelandiae chathamensis),tomtit (Petroica macrocephala chathamensis), red-crowned parakeet (Cyanoramphus novaezelandiae chathamensis),warbler(Gerygone albofrontata),andfantail (Rhipidura fulginosa penita) (Nilsson et al.1994).SeveraltreeorshrubspeciesinseriousdeclineornationallyvulnerablealsooccuronRangatiraIsland,

including the Chatham Island tree daisy (Olearia chathamica),ChathamIslandribbonwood(Plagianthus chathamicus)andtheChathamIslandtreehebe(Hebe barkeri)(Walls etal.2003).

RangatiraIslandalsosupportsalargepopulationofburrow-breedingseabirds.Estimatesfromburrowcounts put the population at around three millionbirds,whichequatestoapproximately14000birdsperhectare,or1.4birdspersquaremetreacrosstheentireisland(West&Nilsson1994).Atsuchhighdensities,these birds must have a major influence on the island’s ecology,primarilybytransferringnutrientsfromthemarineecosystemtoland(Smith1978;Siegfried1981;Polis&Hurd1996;Mizutani&Wada1998;Wainright etal.1998;Anderson&Polis1999;Mulder&Keall2001)and throughdisturbanceassociatedwith theirburrowing and trampling activities (Gillham 1960;Campbell1967;Johnson1975;Warham1996;Maesako

209ROBERTSETAL.:BURROWINGSEABIRDSANDFORESTREGENERATION

1999). Disturbance by seabirds has the potential tocause catastrophic vegetation collapse. GrassholmIsland,offthecoastofWales,supportedhalfamillionpuffins (Fratercula arctica)in1890,buttramplingandburrowingbybirdsledtoalmostcompletevegetationlossandextensiveerosion,tothepointthattheislandcould no longer support a large seabird population(Lockley1953).Asimilarscenariowasfoundamongthe islands off WesternAustralia, where burrowingseabirdssoalteredthevegetationofanislandthatitwasnolongersuitableforbreeding,forcingageneralexodustoanotherislandinthegroup.Whentheoriginalisland’s vegetation had sufficiently recovered the burrow-breedingseabirdsreturned(Gillham1961).OnLittleMangereIsland,ChathamIslands,thecollapseoftheforest thatsupportedthelastremainingblackrobinpopulationwasattributedinparttotheimpactofburrowingbysootyshearwaters(Puffinus griseus)(Butler&Merton1992).TheforesthabitatofRangatiraplayedacriticalroleatthattime,andcontinuestodoso,insavingtheblackrobinfromextinction(Butler&Merton1992;DepartmentofConservation2001).

AstheonlyislandintheChathamsgroupwithasizeableforestthatisalsofreeofintroducedmammals,Rangatira Island’s status as a wildlife sanctuary iscriticallydependentonmaintaining its forest cover.However, studies on Rangatira Island show thatinterspecific and intraspecific competition for breeding burrowsisintense(Was etal.2000;Sullivan&Wilson2001), perhaps indicating a lack of suitable habitatelsewhereintheChathamIslandsand/oranincreasein seabird populations with implications for forestregeneration.

Despitethis,wehavelittleunderstandingofforestdynamicsontheisland,andparticularlyhowseabirdsmightimpactonregenerationprocessesandlong-termforestreplacement.Thisstudyhadfouraims:(1) todescribethecurrentcomposition,size-andage-structureofforestsonRangatiraIslandandtoinferfromthispastregenerationprocesses;(2)toquantifyvariationinseabirdburrowdensityacrosstheisland;(3)toexaminetheimpactofseabirdsonseedlingregenerationintheforestunderstorey;and(4)toassesstheimportanceofcanopygapsassitesfortreeregeneration.

StudysiteRangatiraIslandisthesouthern-mostvolcaniccentreoftheChathamIslands(Fig.1),andatabout3.8–4millionyearsolditisamongsttheyoungestoftheChathamIslandvolcanoes.Itiscomposedentirelyofadippingsequence of coarse-grained volcanic sediments thathaveaccumulatednearavolcanicvent(Watters1978).Volcanicbrecciasformthecragswithsofterbasalticlapillituffsnearthecoast(Hay etal.1970;Campbell

Figure 1. Rangatira Island showing location and classification of forest surveyplots (n=40).The shadedarea is forest.Inset:ChathamIslands’locationinrelationtoNewZealand,andRangatiraIsland’spositionwithintheChathamIslandsarchipelago.

etal.1993).AnotablegeologicalfeatureofRangatiraIslandisthewell-preservedmarineterraces,about15mand80mabovesealevel.Itisthoughtthesewereformedwhenthesealevelwashigherduringthewarmerinter-glacialstagesofthePleistocene(Hayetal.1970;Wattersetal.1987).

Temperaturesarecoolwithayear-roundaverageofabout11°C.Frostsareinfrequentandlight,althoughhail showers are common in the winter months.Annualrainfallvariesfrom715to1050mmanddrysummerspellsarecommon,sometimeslastingmorethanamonth,butskiesareoftenovercast(averaging


74%).Summerhumidityishigh,oftenexceeding80%(Thompson1983;Campbell etal.1993).Thealmostincessantwindisreferredtoasthe‘RoaringForties’.FromWoolshedBushinthenorthandTheClearsinthesouth-easttheislandslopesgentlyupwardtothehighestpoint(224m)wherethesouth-facingrampart-likecliffsplungeverticallytothesea(Fig.1).

Land-use history

Moriori periodThere is no record of permanent pre-EuropeansettlementonRangatiraIsland,butMorioriclaimstothe 1870 Land Court hearing confirm the island was significant to them. Moriori probably spent periods ashoreharvestingfood,butthevirtualabsenceofthekopitree(Corynocarpus laevigatus),believedtohavebeenbroughttotheChathamIslandsfromNewZealandbyMorioriandcultivatedfortheirfruits(King2000),suggestsRangatiraIslandwasnotsettledpermanently.There are large kopi tree groves, for example, onadjacentPittIsland(2.5kmaway),whereanestimated300Morioriwerelivingin1790(Richards1972,1982;Wills-Johnson1996;King2000).

European eraIn theearly1800sa sealing stationwasestablishedonRangatiraIslandandpigs(Sus scrofa)andsheep(Ovis aries)wereintroduced(Richards1982).Dogs(Canis familiaris),cats(Felis catus)andrats(Rattus norvegicus, R. rattus)neverestablishedonRangatiraIsland,whichprobablyaccountsforthesurvivalthere,butnotelsewhere,oftheChathampetrel(Pterodroma axillaris)andtheNewZealandshoreplover(Thinornis novaeseelandiae).OnChathamIslandthesepredatorshaveextirpatedmostoftheoriginalavifauna(Atkinson1978;Veitch&Bell1990;Tennyson&Millener1994).Rangatira Island is also the main refuge for otherterrestrialandoceanicbirdspecies.

Whalers,accompaniedbyfarmanimals,includinggoats (Capra hircus), arrived around 1839 andsystematicclearingofRangatira Island’s forests forpastureandpotatocroppingbeganshortlyafter(Ritchie1970; Holmes 1984). Stock numbers fluctuated over the next100years.Thelastlessee(Neilsen),whofarmedtheislanduntilitbecameareservein1954,hadover1000sheepandapproximately20cattle(Bell1953).Apartfromafewstraysshotin1961,allthestockhadbeenremovedby1959(Ritchie1970).

Littlehasbeenpublishedontheearlyvegetationoftheislandorchangessincestockwereremoved.WhilevegetationonChathamandPittislandswasdescribedbyearlybotanists(e.g.Cockayne1902),theydidnotvisitRangatiraIsland.Fleming(1939),inanaccountofhis1937visit,wrote‘bracken“clears”andpasture-landalternatewithbeautifulparklikebushlands’but

gavenodetailsof forest species.However, aphoto(Wotherspoon collection,Auckland Museum) takenonthisvisitintheWoolshedBusharea(Fig.1)showsscatteredOleariatraversii treessetinpasturegrass.AsonChathamIsland(Rekohu),theseOlearia appeartobetheonlysurvivingremnantsoftheoriginalforest,which would have been opened up by fire and stock andthenexposedtowindandsaltdamage.AtthetimeRangatiraIslandbecameareservein1954,theforestwas described as significantly reduced in area and quality(Bell1953);onlyone-thirdoftheisland(TopBush,Fig. 1)was forested,with a heavilybrowsedunderstoreyandasparsecanopyvulnerabletodamagebysalt-ladenwinds.

Sincegrazingceasedtheforesthasregeneratedonboththelower(WoolshedBush)andupper(TopBush)marineterraces.Forestremnants,suchasIke’sBushandIslandBush(Fig.1),appearas‘forestislands’setingrasslandinphotographstakenbyB.D.Bellin1961.However,in2002whiletheseforestislandswerestillvisible,theyweresetinaseaofbracken(Pteridium esculentum) overrun with Muehlenbeckia australis,whichalsocoversmuchoftheforestmargins(Roberts2004).Intheareacalled‘TheClears’(Fig.1),whichadjoinsthesealcolonies,stumpsandlogsareevidenceofaforestthathasbeenkilledbythewindafterthestunted,shelteringmarginhadbeenbrokendownbysheep and cattle (Bell 1953). On this eroding salt-marsh,forestregenerationisslowduetopermanentsoillossinanareaofextremeexposuretosoutherlystorms.Taylor(1991)listedplantspeciesandestimatedtheirrelativeabundance,andWestandNilsson(1994)recordedplantspeciespresence/absenceinassociationwithburrowhabitat,butdidnotmeasurefrequency.Nilssonetal.(1994),whenrecordingbirdlifeontheisland,estimatedtheforesttocover98haor45%oftheislandarea.

MethodsForest samplingTodescribethepresentcompositionandstructureoftheforestsweestablished40permanentlymarkedplots(10×10m)inAprilandMay2002.Plotswerelocatedusing stratified random sampling. We gridded a map of theislandintosquares(100×100m),andnumberedallgridsquareshavingatleasttwo-thirdsforestcover.Of these, we selected at random 10 grid squares inWoolshed Bush (five each on NW and SE sides of the maintrack),and20gridsquaresinTopBush(10onNWand10onSEsidesoftheisland),andassignedaplot to each grid square (Fig. 1). It can be difficult to avoiddamagingseabirdburrowswhenwalkingacrossthe island, and to confine impact, a network of walking trackshasbeenestablished,providingaccesstoalmost


allpartsoftheisland.Tominimisedamageweusedtheintersectionoftwotracksinagridsquareasthestartingpointtolocateeachplot,andduetotheextremefragilityofthesoilonceoffatrack,andinmanycasesevenonthe track, ‘petrelboards’ (plywoodsquaresattachedtosnowboardbindings)wereworn.Plotlocationwaschosenbyheadingarandomdistance(0–100paces)alongoneoftherandomlychosentracks,andthenafurtherrandomdistance(0–20paces)offthetrackintothe forest.An additional 10 plots were subjectivelyplaced in locations identified as having features not represented in the first 30 plots. This included smaller bushpatchessuchasIke’sBushandIslandBush,higheraltitudelocations,apparentlyolderPlagianthusforest,andpatchesofverydenselyregeneratingforest.

Sampling methods were based onAllen (1992,1993).Eachplotwasgriddedintofoursubplots(5×5m).Withineachofthese,wemeasuredthediameteratbreastheight(dbh;1.35maboveground)ofalltreespecieswithdbh>3.0cm,includingdeadstandingtrees.Treeswithtrunksgrowinghorizontallyweremeasured1.35malongthestemfromwheretheywererooted,ratherthanatbreastheight.Formulti-stemmedtrees,wemeasuredthedbhofthelargeststemandrecordedthe number of stems >3.0 cm dbh. Saplings, defined as stems>1.35mhighbut<3.0cmdbh,werecountedineach subplot. We counted seedlings, defined as woody species <1.35 m tall, in eight circular understoreyplots,eachwitharadiusof0.49m(area=0.75m2).Understorey plots were located at the subplot gridintersections.

Compositional analysis of forest plotsWeusedclusteranalysistogroupthe40forestplotson thebasisofsimilarities in theircompositionandsizestructureinordertoidentifyforestcommunities.Todothis,wegroupedtreesinto3-cm-dbhsize-classbins(e.g.3–5.9cm,6–8.9cm,andsoon)upto54cm.Trees ≥54 cm dbh were put in a single bin. For each plot,weusedthenumberoftreesofeachspeciesineachsize-classbintocalculateasimilaritymeasureandperformaclusteranalysis.Thus,plotsthatweregroupedin the cluster analysis shared a similar tree speciescomposition and size structure. We used the Bray–Curtismeasureofsimilaritytoconstructamatrixofpairwisedistancesbetweenplots,andanagglomerativeclusteringtechnique(unweightedpair-groupmethodusingarithmeticaverages–UPGMA)togroupplotsonthebasisoftheirpairwisesimilarities(Baker1992).From the resulting dendrogram, we identified five mainclustersofplots,whichwerecognisedasforestcommunities.Combiningtheplotsineachcluster,treespeciesineachsize-classweretotalledandconvertedintototalstemdensityperhectare.

Bar graphs showing the size-class profile of canopy (Fig. 2a & 2b), subcanopy, and dead trees

(notshown)ineachclusterwereusedtodescribethechiefcharacteristicsoftheforestcommunities(Table1), which were named following Atkinson (1985).Figure1showsthedistributionofforestcommunitiesacrosstheisland.Inusingonlythelargeststeminamulti-stemmed individual for analysis the resultinghistogramofsizeclassescanbebiasedtowardssmallerand thus apparently younger stems. However, heretreesizewasusedasacoarsesurrogatefortreeage,inordertoinferthetypeofregenerationprocessesintheseforests,particularlytolookforalackofsmallindividualsthatmightindicateadiscontinuouspatternof regeneration, as in the case of Plagianthus. Forthispurpose,thelargeststemofeachindividualisasensiblemeasure.

Age structureTo determine the age structure of the forest weincrement-cored the largest tree (with a minimumdbhof8cm)ineachsubplot(5×5m)at0.5maboveground. Following standard dendrochronologicalprocedures,incrementcoresweremountedinwoodenblocks, sanded until tree rings were clearly visible(Stokes&Smiley1968),andthengrowthringswerecountedunderamicroscope.Whencoresmissedthechronologicalcentrebutinnergrowthringarcswerevisible,weestimatedtheageof themissingportionusingthegeometricmodelinDuncan(1989).Wherethecentreofthetreewasrotten,agewasestimatedusingthepartialcorebyassumingthatthechronologicalcentrewasatthegeometriccentre(Norton etal.1987).Theageestimatesreportedinthisstudywerenotcorrectedfortreegrowthtocoringheight.

Environmental variables and site featuresToidentifywhethertherewasarelationshipbetweenenvironmental variables and forest structure andcompositionwerecordedthefollowingateachplot,usingmethodologyadaptedfromAllen(1992):Slope(usingaclinometer);aspect(usingacompasstothenearest1°atrightanglestothegenerallieoftheplottocalculatedegreesfromnorth);andaltitude(ma.s.l.)(fromNZtopographicalmap1:50000(1998)).

InApril 2004, using a corer (5× 10 cm), soilwassampledfromeachforestplot(tothemaximumdepthpossiblewithoutpenetratingburrowchambers).Surface litter was first removed and four soil samples (approximately600ml)weretakenwithina2-mradiusoftheplotcentre.Thefoursamplesweremixedandapproximatelyone-quarterwastakenforanalysis.Inthelaboratorysoilsampleswereair-dried,sievedthrougha2-mmmesh,andanalysedforpH,N,CandP.SoilpHwasdeterminedinasuspensionofsoil(1g)anddistilledwater(2.5g),usingapHmeter(MettlerToledo).Totalnitrogenand total carbonweredeterminedby


Figure 2a. Bar graphs showing density of stems in diameter size-class for two of four main canopy species found in five forest communities identified by cluster analysis on Rangatira Island.


Figure 2b. Bar graphs showing density of stems in diameter size-class for two of four main canopy species found in five forest communities identified by cluster analysis on Rangatira Island.


drycombustion,usingacarbonandnitrogenanalyser(LecoCNS-2000Analyser).Availablephosphoruswasextracted with a solution of 0.03 M ammonium fluoride and0.1MHCI(Bray&Kurtz1945)andwasdeterminedcolorimetricallybythephosphomolybdatebluemethod.WeusedanovaandTukey’sHSDtoidentifyvariablesthatdifferedamongforestcommunities.

Burrow densityThetotalnumberofseabirdburrowentrancesineachplot(10×10m)werecountedasthebestavailablemeasureofburrowingactivity.Wedidnotdistinguishbetween burrow entrances of different size. Forsimplicity,andbecausetheactualnumberofburrowchamberswasnotdetermined,ourmeasureofburrowdensityisthenumberofentrancespersquaremetre.

Seabird exclusion experimentToexaminetheimpactofseabirdburrowingonwoodyseedlingregeneration,weplacedaseabirdexclosurepairedwithacontrolplotinthecentreofeachofthe30 randomly located forest plots (10 × 10 m). Theexclosureswereconstructedfromwiremesh(25×20mm)withdimensionsof0.5×0.5mand0.2mhigh,openonthebottom,andheldinplacewithaluminiumpegsateachcornerandatthemidpointalongeachside.Thepairedplotswerelocatedsothattheydidnotcoverburrowentrancesbutwerenomore than1mapart,andtheexclosureorcontroltreatmentwasrandomlyassigned.TheplotswereputinplaceinApril2002,at which time all woody seedlings in the exclosureandcontrolplotswererecorded.Whentheplotswereremeasured in2003,oneexclosurewasfoundtobeimpactedbyburrowingsothatpairwasdroppedfrom

the2003analysisandtheexclosurerepaired.However,thisexclosurewasincludedinthesamplingafter24months(April2004)and33months(January2005),asachangeinseedlingheightovertimewasnotusedin the final analyses, as in most of the exclosures (with afewexceptions)theseedlingcohortofoneyearwasreplacedwithanewcohortthenext.

InJanuary2003,canopycoverabovethepairedexclosures and controls was estimated by taking adigital photograph at each plot centre using a fisheye lens.ThedigitalimageswereanalysedusingAdobePhotoshop Software (version 7). The Histogramfunctionwasusedtocalculatethenumberofskypixelsintheimage.FollowingSullivan’s(2003)methodsthecolourselectionwassettopurewhiteandthefuzzinessto 150, which was sufficient to distinguish white cloud andblueskyfromfoliage.

Toexplorethevariationinseedlingcountsamongexclosureplots,weexaminedtherelationshipbetweenthe number of seedlings counted in exclosure plotsin 2005 and four explanatory variables: soil N andP, percent of the canopy that was open to the sky(from the fisheye photographs), and density of bird burrows.Wemodelled the relationshipsusinga loglink function assuming the errors follow a negativebinomial distribution to allow for overdispersion inthecounts.Parameterswereestimatedusingmaximumlikelihood.

Regeneration in forest canopy gapsToassesstheimportanceofcanopygapsassitesfortreeregeneration,wesurveyedcanopygapsalong30belt transects (100 × 20 m) located along walkingtrackstominimiseburrowdamage;theirstartlocated

Table 1. Summary of the characteristic features of the five forest communities identified by a cluster analysis (see text) followed by a size-class profile, from 40 forest survey plots located on Rangatira Island. Nomenclature follows Atkinson (1985).___________________________________________________________________________________________________________________________________

PlagianthusForest(12plots).CanopyofPlagianthus chathamicus(ribbonwood)andanunderstorey(ofequaldensity)ofMelicytus chathamicus (māhoe) and Myrsine chathamica (matipou); few other species present. The size-class profile shows fewliveordeadspeciesinthelargerdbhclass,suggestingthisforestisayoungforestcommunity;30%oftheforestisinthisforesttype,whichisfoundinWoolshedBush,IkesBushandTopBush.

Mixed Forest (12plots).Allcanopyspecies(Plagianthus chathamicus, Olearia traversii(akeake),Coprosma chathamica(karamū), and Myoporum laetum(ngaio))andallunderstoreyspecies,includingMacropiper excelsum,present;30%oftheforestisinthisforesttype.Ahigherdbhsize-class(includingdeadtrees)wasfoundinthisforest,suggestingamorematureforest.ThisforestwaspredominantlyfoundinTopBush(11outof12plots).

Plagianthus/Melicytus Forest(10plots).CanopyofchathamicusandunderstoreydominatedbyMelicytus chathamicus;25%oftheforestisinthisforesttype,whichisfoundinWoolshedBush,IslandBushandTopBushandfavoursforestedges.

Plagianthus/Myrsine Forest(3plots).CanopydominatedbyPlagianthus chathamicusandunderstoreybyMyrsine chathamica.BothMacropiper excelsumandOlearia traversiiareabsent;7.5%oftheforestisinthisforesttype.ThisforestcommunityisonlyfoundinWoolshedBush.

Coprosma–Olearia Forest (3 plots). Canopy of Coprosma chathamica and Olearia traversii. These plots are generallycharacterisedbyyoung,denselyregeneratingCoprosma chathamicaforestinthedamperareasofTopBush;7.5%oftheforestisinthisforesttype.___________________________________________________________________________________________________________________________________


by stratified random sampling. Five transects were allocatedtoeachoftheeasternandwesternsidesofWoolshedBush,and10transectstoeachoftheeasternandwesternsidesofTopBush.Atransectstartlocationwaschosenbyrandomlychoosingatrack,astartpoint,andadirection.Fromthestartlocation,wewalkedalongthetrackintheselecteddirectionandlocatedallcanopygaps10meithersideofthetrack.Acanopygapwasdefined as an area ≥25m2(equivalenttoasquare5×5m)createdbyatreeorlimbfall,extendingthroughalllevelsofforestandopentothesky.Eachcanopygapwasdelimitedbyverticallyprojectingtheopeninginthecanopytothegroundsurface(Brokaw1982).Canopygaparea,A,wascalculatedbymeasuringthelength,L,(longestdistancefromgapedgetogapedge)andwidth,W,(longestdistanceatrightanglestothelength),andusing the formulaforanellipse:A =πLW/4(Stewartetal.1991).

Alongthelengthofthegap,welocatedaseriesofcircularunderstoreyplotseachwitharadiusof49cm(area=0.75m²).Thesewerepositionedbylayingameasuringtapealongthelengthofthegapandlocatingaploteverymetrealternatelytotheleftandrightofthetape.Thedistancefromthetapewasrandomlychosen.Becauseplotswerelocatedeverymetre,wemeasuredmoreplotsinlargergaps.Ineachplot,wecountedthenumberofwoodyseedlingsofeachspecies.Wealsocounted the number of seabird burrow entrances inthe rectangular area defined by the length and width ofthecanopygap.

Tocompareregenerationpatterns ingapversusnon-gap areas, we paired each gap with a non-gapsiteofequivalentdimensionsandrepeatedtheentiresamplingprocedure.Eachnon-gapsitewaslocatedarandomdistance(between5and10m)inarandomdirection away from the gap edge. For comparisonbetweengapsandnon-gapsalldatawereconvertedintodensityperhectare.

ResultsComposition and structure of the forestsTheforestsofRangatiraIslandcomprise,invaryingcombinations,eightmajortreespecies(fourcanopyandfoursubcanopy).ThecanopyspeciesarePlagianthus chathamicus, Olearia traversii, Coprosma chathamica and Myoporum laetum.Overtheforestedpartoftheisland,theseformacontinuouscover withamaximumcanopyheightvaryingbetween10and15m.

The four canopy species have different size-frequencydistributions, and this combinedwith therelative density of canopy and subcanopy speciesdistinguishestheforestcommunities(Fig.2a&2b).Overall stem density per hectare was significantly higher in the Coprosma–Olearia Forest than the

Plagianthus Forest (P < 0.05 Tukey’s HSD test),but not significantly different from the other forest communities.BothOleariaandMyoporumoccuratlowdensitiesasscattered,oftenlargetrees,withveryfewstemsinanysizecategory(Fig.2b).Coprosma(Fig. 2a) has a reverse-J-shaped size distribution intwo of the five forest communities, with many small stemsandprogressivelyfewerinthelargersize-classes,while Plagianthus (Fig. 2a) has a bell-shaped sizedistributioninallforestcommunitieswithmoststemsin therange9–24cmdbh.Thefoursubcanopytreespecies,Melicytus chathamicus,Macropiper excelsum,Myrsine chathamica and Pseudopanax chathamicus,allhaveareverse-J-shapedsizedistribution(notshown).Melicytusdominatedthe3–9cmdbhclassesbuttherewerenotreeslargerthan15cmdbh.Myrsineoccursatmuchlowerdensitybutwasfoundinallsizeclassesupto30cmdbh.Deadstandingtreeswithdbh>15cmwerevirtuallyabsentfromeveryforestcommunityexcepttheMixedForest.

The understorey subplots revealed very patchyseedling establishment. Of the canopy species,Plagianthus was the dominant woody seedling (c.4200ha–1),withnoOlearia orCoprosma seedlingsinthesubplotsandveryfewMyoporum.Bysaplingheight,Plagianthusdensityhaddroppedto28saplingsha–1.OfthesubcanopyspeciesMelicytus, MyrsineandMacropiper wereallfoundathighseedlingdensities(c. 3000 ha–1), which dropped to c. 1000 ha–1 forMelicytusandMyrsineandc.855ha–1forMacropiperbythesaplingstage.

Of 153 trees cored, only Plagianthus (n = 75)

Figure 3.Relationshipbetweendiameteratbreastheight(1.35m)andage(years)forcored Plagianthus trees(n=75)in40plotsonRangatiraIsland,ChathamIslands.Theverticallineindicatesstockremoval43yearsago.


Figure 4.Relationshipofseabirdburrowdensitytoenvironmentalvariables,saplingandtreedensity(stemsha–1),andsoilproperties.Seabirdburrowdensityisshownonthey-axisinallgraphsforconsistencybutisnotmeanttoimplythatthevariableonthex-axiscausesvariationinburrowdensity.Inseveralcases,variationinburrowdensityismorelikelythecauseofvariationinthex-axisvariable(e.g.forsoilnutrients).Altitude(masl=metresabovesealevel),DegreesNorth(degreesawayfromnorth).r2 values are the coefficient of determination: * = P<0.05,**=P<0.01,***=P<0.001(withtheP-valuescorrectedformultiplecomparisonsusingtheBonferroniprocedure).

producedringsclearenoughforaccurateageing.ThemajorityofPlagianthustreesagedwerelessthan43yearsold.Assumingthatcoringthelargesttreeoneachplotprovidesanestimateoftheageoftheforest,wesurmisethatmostofthepresentPlagianthus foresthasregenerated since farmingceased in1959andmoststockwereremoved(Fig.3).

Of the environmental variables we measured,onlysoilpHdifferedamongcommunities,withthepHsignificantly lower in both the Plagianthus(4.02±0.12)andPlagianthus/Myrsine(3.36±0.02)(P <0.05Tukey’sHSDtest)comparedwiththeotherforestcommunities

(MixedForest5.18±0.31,Plagianthus/Melicytus4.72±0.37,Coprosma–Olearia5.06±0.99).

Seabird burrow density and soil characteristics Withinplots,seabirdburrowcountsrangedfrom0to223per100m2,withameanburrowdensityof1.19±0.10m–2.Thesoilcharacteristicsofallplots(mean±SE)were:total%C17.74±1.03,total%N1.43±0.08,andPmg/kg1024±110.Allsoilswerehighlyacidic(pH4.52±0.18),andburrowdensitywaspositivelycorrelatedwithsoilphosphorusandnegativelywithaltitude,pH,saplingandtreedensity(Fig.4).


Seabird exclusion experimentThere was no significant difference in woody seedling densitybetweenthe30pairedexclosureandcontrolplotswhentheplotswereestablished(pairedWilcoxonrank sum test, V = 6, P = 0.17).At all subsequentmeasurements, therewerecleardifferencesbetweentreatments with, on average, more seedlings foundinexclosurethancontrolplots(Fig.5).In2005,themeannumberofwoodyseedlingscountedincontrolplotswas1.1±0.50andinexclosureplotswas13±2.9. There was nevertheless substantial variation inseedlingcountsamongexclosureplotsandseedlingspecies composition fromonecount to thenext.Atthe first count of seedlings in exclosure plots in 2003, Plagianthus(44%)andMelicytus(39%)werefoundtobe themostcommon (total seedlings=700).By2005,however,Plagianthus(3%)wererarelyfoundandMacropiper(55%)becamethemostabundant,withMelicytus (39%)remainingunchanged(totalseedlings= 103). By comparison, in the control plots at the first countin2003,Plagianthusrepresentedonly19%oftotal

seedlings(390)whereasMelicytusrepresented69%.By2005Plagianthus (3%)hadvirtuallydisappearedandMacropiper(39%)andMelicytus(54%)werethemostcommon (total seedlings = 33). We found no significant differencesinseedlingdensityorseedlingcompositionamongforestcommunities,andnocorrelationbetweenseedlingdensityandenvironmentalvariables,withtheexception that seedling counts were higher in moreopen-canopiedplots(changeindeviancerelativetoanullmodelofnorelationship=11.6,1d.f.,P<0.001),with a greater number seedlings under higher lightconditions(Fig.6).

Regeneration in forest canopy gapsWesearchedatotalof6haalongtransectsandfound14canopygaps,sixinWoolshedBushandeightinTopBush.Canopygapsweresmall(meanarea<32m2),usuallyhavingbeencreatedbythefallofasingletree.Thetotalareafoundincanopygapswas438.6m2,or0.73%oftheareasurveyed.

F i g u re 5 . F r e q u e n c ydistributions of the totalnumber of seedlings foundin all seabird exclosure andcontrol plots (0.5 × 0.5 m)duringannualmeasurementsfrom2002to2005(exclosureplots were established aftermeasurement in 2002). n =30 exclosure and 30 controlplotsinallyearsexcept2003,whenn=29.Notshownarethree plots in 2003 and oneplotin2005thateachhad>60seedlings.


Figure 7. Density(stemsperhectare)ofthemaintreespecies(includingseedlings)in14canopygapplotscomparedwith14controlplotsonRangatiraIsland.Blackshadedbarsindicatecanopygapplots; grey shadedbars indicate control plots.Errorbarsare±1SE.

Figure 6.Relationshipbetweennumberofseedlingscountedinseabirdexclosureplots(0.5×0.5m)andpercentcanopycover.Thedottedlineisthemaximumlikelihoodregressionlineobtainedusingaloglinkfunctionandassumingnegativebinomial distributed errors to allow for overdispersion inthecounts.

Seventy-onepercentofcanopygapswerecreatedbythefallofOleariatreesand21%byPlagianthus. TheratiooffallenOleariatoPlagianthuscorrespondstotheratioofthelargersizeclassesofthesespecies(Fig.2a&2b).Thedbhofgapmakersrangedfrom20to99cmwithameanof57cm.Mostfallentreeshadbeenuprooted(71%)andtherootsystemofallfallentreeshadbeenheavilyburrowed.

Theseedlingdensityofsevenoftheeightdominantcanopy and subcanopy species was significantly higher incanopygapscomparedwithnon-gapplots(Table2), with regeneration in canopy gaps dominated byPlagianthusandMelicytus (Fig.7).Theretendedtobefewerseabirdburrowsingapsthannon-gapareas(Table2).

DiscussionForest structure and composition

Theforeststhathaveregeneratedsincefarmingceasedarerelativelysimpleinstructureandcomposition,andwhileitispossibletoidentifycommunitiescomprisingdifferentratiosofthespeciesrepresented,thesearenotclearly differentiated by environmental factors.Thepost-farmingforestremnantinTopBushispresumedtobethenearestincompositiontotheoriginalforestontheisland,giventhatthiswasthelargestsurvivingremnant. Our analysis identified this forest type as MixedForestwithallcanopyandsubcanopyspeciespresent.Apartfromoneoutlier(Fig.1)thisforesttypewas foundonly inTopBush. If ourpresumption iscorrect,thentheoriginalcanopyandsubcanopyspeciesmixhasnotregeneratedacrosstheisland.

ThedeciduousPlagianthus chathamicuscurrentlydominates four of the five forest communities (Fig. 2a),occurringinvaryingassociationswiththeothermain canopy species Olearia traversii, Coprosma chathamica and Myoporum laetum. Olearia occursalmostexclusivelyaslargetreesscatteredthroughout

Table 2. Paired t-testscomparingmeannumberofwoodyspecies(n=7)seedlingsin14canopygapsand14non-gapplots,andthemeandensityofburrowsincanopygapscomparedwithnon-gapplotsonRangatiraIsland._______________________________________________________________

n Mean±SE t PWoodyspecies<135cm Gap 14 48.57±5.71 6.689 0.001 Closedcanopy 14 7.28±1.90 No.ofburrows Gap 14 43.00 ± 8.23 −2.939 0.012 Closedcanopy 14 66.64±7.25_______________________________________________________________


theforest,andthepresentlargeOleariatreesappearto have colonised during, or persisted through, thefarmingera indicativeperhapsof their resilience todisturbancecomparedwithotherforestspecies.WhileOleariaisabundantontheforestedges,wefoundnoOlearia seedlings in the exclosures and few in theunderstoreyplotsorcanopygaps.ManyofthelargeOleariatreesarenowsenescing,withOleariabeingthedominantcauseof canopygaps, suggesting thisspecies isdecliningandmaydisappear fromwithintheforestastheremaininglargetreesdie.

Most large Plagianthus trees that were coredhadadatepost-farm-abandonment(Fig.3),andthistogetherwiththeirbell-shapedsizedistributionfoundinallforesttypesimpliesapulseofregenerationpost-farming,followedbyaslowingofregenerationduetocanopyclosure.Oftheothertwocanopyspecies,therewerenoCoprosmaseedlingsineithertheexclosuresor understorey subplots and it was rare in canopygaps(Fig.7).However,Coprosmawasfoundinhighdensitiesassaplingsandyoungtreesinregeneratingareasonthedampernorth-westernsideofTopBush(Fig. 1). Coprosma is not regenerating elsewhere,suggestingatrendawayfromtheoriginalMixedForestcommunitytoamorehomogenousforestdominatedbyPlagianthus,withCoprosmarestrictedtothewesternsideoftheisland.

Bird impacts, canopy gaps and forest regenerationWerecordedameanseabirdburrowdensityof1.19m–2,similartotheburrowdensityof1.39m–2recordedbyWestandNilsson(1994)in1989/90.ThisishighcomparedwithotherseabirdcoloniesinNewZealand.MulderandKeal(2001)recordedadensityof0.84m–2onStephensIsland,CookStrait,Waughetal.(2003)foundaburrowdensityof0.24m–2forWestlandpetrel(Procellaria westlandica),Fukamietal.(2006)recordedadensityof0.36±SEM0.11m–2acrossnineseabirdislands innorthernNewZealand,althoughWarhamandWilson(1982)estimatedsootyshearwaterburrowdensitytobe1.16m–2inOlearia lyalliforestontheSnaresIslands.InacolonyofPuffinus tenuirostrisonCapeQueenElizabeth,Tasmania,burrowdensityvariedbetween0.04and0.74m–2(Walsh etal.1997).

Seabirdburrowdensityvariedacrosstheisland,withfewerburrowsathigheraltitude(>65m)andinareaswithahighdensityofsaplingsandtrees(Fig.4). This most likely reflects a preference by birds for more open-canopied forest because of the difficulty of burrowinginheavilyrootedareas.AreaswithhigherdensitiesofseabirdburrowsalsohadlowersoilpH,consistent with findings in other seabird colonies (Ellis 2005),andhigheravailablephosphoruslevels.ThiswasmostlikelytheresultofseabirdguanoloweringsoilpHthrough nitrification from ammonium to nitrate after

mineralisationoforganicnitrogen,andfromgreaterinputs of P through nutrient transfer in areas moreheavilyusedbyseabirds(Smith1978;Okazaki etal.1993;Anderson&Polis1999).

Exclusionofseabirdsresultedinamarkedincreaseinwoodyseedlingnumbersintheunderstoreyexclosureplots(Fig.5),implyingthatseabirdsinhibitregenerationthrough their trampling and burrowing activities(Gillham1956,1960;Maesako1985,1999;Warham1996;Mulder&Keall2001).Nevertheless,therewasnorelationshipbetweenseabirdburrowdensity(whichwe assume is a surrogate for bird activity) and thenumbersofwoodyseedlingsintheexclosures.Instead,thenumberofwoodyseedlingsregeneratinginseabirdexclosureswaspositivelyrelatedtotheamountofopencanopydirectlyabovetheexclosures,suggestingthatinadditiontoseabirds,seedlingregenerationisstronglylimited by light availability. This finding is reinforced bythecanopy-gapsurvey,whichrevealedhighdensitiesof seedlings regenerating in gaps. Hence, althoughthere is an inhibitory effect of seabirds on seedlingestablishmentandsurvival,thisdoesnotappeartobeabarriertosuccessfulregenerationunderthehigherlight conditions found in canopy gaps. We furtherobservedalowerdensityofburrowsingapsthaninadjacentnon-gapplots(Table2),consistentwithourfinding that seabird burrow density was lower in plots withhighdensitiesofsaplingsandtrees(Fig.4).Birdsmaybeforcedtoabandonburrowsinareasofdenselyregenerating forest (such as in canopygaps)due todifficulties in physically accessing the sites.

Gapsarecentraltoforestdynamics(Runkle1982;Brokaw1987;Schnitzer&Carson2001).However,littleworkhasbeendoneoncanopygapsonseabirdislands with which to compare the findings of this study, thoughJohnson(1982)notedthatinthelow-canopy(6–8mtall)Olearia lyaliiand O. augustifolia forestonPutauhinaIsland(South-WestMuttonbirdIslands),windthrowusually involvedmore thanonetreeandclearingsof10–20macrossweretherule.

The treefall gaps on Rangatira Island covered0.73%ofthe6hasurveyedandweregenerallysmall,with12ofthe14gapshavinganarealessthan31m2.Whilethereareanumberoffactorsthatcanleadtouprooting–windiness,growthform,rigidityandtheheightoftreesabovethesurroundingforest(Peterson2000;Martin&Ogden2006)–anadditionalfactoron Rangatira Island is the burrowing around treeroots,whichloosensthesoil,driesouttheroots,andweakensthetrees.

Future of the forestRangatira Island is an important wildlife sanctuarydependentonthemaintenanceofforestcovertoprovidehabitat for its many rare and endangered species.Our results show that most of the present forest is


young, resulting from regeneration following farmabandonment, and that the forest across the islandcomprisesvariousmixturesoffourdominantcanopyand four dominant understorey species. Seabirdsburrowatveryhighdensitiesintheforest,andseabirdactivity has a significant inhibitory effect on seedling regenerationunderclosedcanopies.Althoughseveralof the understorey species, including Myrsine andMacropiper, canregeneratefromrootsuckersandformthickets, thedominant canopy species,Plagianthus,does not resprout and relies on regeneration fromseed.Ourresultsshowthatcanopygapsarecriticalfor successful canopy tree regeneration, and thatregenerationisdominatedbyPlagianthus,suggestingitwillremainthemajorcanopyspeciesontheisland.The largesenescingOlearia thatcreatemostof thecurrentcanopygapsaregraduallydisappearingfromtheforest.Thiswillmostlikelycauseadeclineinthenumberofpotentialgap-makersuntilthepost-farmingcohortofPlagianthussenesces,orbecomesunstableduetoseabirdburrowing.Thefutureoftheforestatthisstageisunclear.IfthePlagianthuscanopysuffersa progressive collapse over a reasonably long timeperiod,theforestcouldremainlargelyintactwithtreereplacementoccurringinsmallgaps,asappearstobehappeningatpresent.Thereisalsothepossibility,giventhedensityofburrowingseabirds,thattherelativelyeven-age Plagianthus trees could simultaneouslybecomesusceptibletowindfall,andtheforestcouldbecomepronetoamorecatastrophicblowdown.

AcknowledgementsWe thank fieldworkers Lisa Berndt, Roz Heinz and ShonaSamfortheirworkonrespectivevisitstotheisland,RozHeinzforprefabricatingthewireexclosures,andBradCaseformapwork.ThankstotheDepartmentof Conservation (DOC) staff at Te One, ChathamIslands,fororganisingthelogisticsofgettingonandoff Rangatira Island, and HilaryAikman and DaveHouston, DOC Wellington Conservancy, for theirsupportoftheproject.Threeanonymousrefereesmadeconstructivecommentsonthemanuscript.Weespeciallythank all who shared their experiences, knowledgeandphotographsofRangatiraovertheyears–theseconversationsshapedourthinkingandplanningforthisresearch.VegetationplotdatafromthisstudyarestoredintheNationalVegetationSurveydatabank.

ReferencesAllen RB 1992. RECCE: An inventory method for

describingNewZealandvegetation.FRIBulletin181.Christchurch,ForestResearchInstitute.25p.

AllenRB1993.Apermanentplotmethodformonitoringchanges in indigenous forests. Christchurch,LandcareResearch.

Anderson WB, Polis GA 1999. Nutrient fluxes from watertoland:seabirdsaffectplantnutrientstatuson Gulf of California islands. Oecologia 118:324–332.

AtkinsonIAE1978.EvidencefortheeffectsofrodentsonthevertebratewildlifeofNewZealandislands.In:DingwallPR,AtkinsonIAE,HayCedsTheecologyandcontrolofrodents inNewZealandnaturereserves.NewZealandDepartmentofLandsandSurveyInformationSeries4.Pp.7–30.

AtkinsonIAE1985.DerivationofvegetationmappingunitsforanecologicalsurveyofTongariroNationalParkNorth Island,NewZealand.NewZealandJournalofBotany23:361–378.

BakerWL1992.Structure,disturbance,andchangeinthebristleconepineforestofColorado,U.S.A.ArcticandAlpineResearch24:17–26.

Bell LC 1953. Unpublished file note 46/29/90, 15 April. Wellington,DepartmentofInternalAffairs.

Bray RH, Kurtz LT 1945. Determination of total,organic, and available forms of phosphorus insoils.SoilScience59:39–45.

Brokaw NVL 1982. The definition of treefall gap and its effectonmeasuresofforestdynamics.Biotropica14:158–160.

BrokawNVL1987.Gap-phaseregenerationofthreepioneertreespeciesinatropicalforest.JournalofEcology75:9–19.

ButlerD,MertonD1992.Theblackrobin:savingtheworld’smostendangeredbird.Auckland,OxfordUniversityPress.294p.

Campbell DJ 1967. The Trio Islands, MarlboroughSounds: an ecological study of a bird modified island.UnpublishedMAthesis,VictoriaUniversityofWellington,Wellington,NewZealand.216p.

Campbell H, Andrews AG, Beu AG, MaxwellAR, Edwards AR, Laird MG, Hornibrook N,MildenhallDC,WattersWA,BuckeridgeJS,LeeDE,StrongCP,WilsonGJ,HaywardBW1993.Cretaceous–Cenozoicgeologyandbiostratigraphyof the Chatham Islands, New Zealand. LowerHutt,InstituteofGeologicalandNuclearSciencesMonograph2.269p.

Cockayne L 1902. A short account of the plant-covering of Chatham Island. Transactions andProceedings of the New Zealand Institute 34(1901):243–325.

DepartmentofConservation2001.Blackrobinrecoveryplan 2001–2011. Threatened Species RecoveryPlan40.Wellington,DepartmentofConservation.19p.

DuncanRP1989.Anevaluationoferrorsintreeageestimatesbasedonincrementcoresinkahikatea


(Dacrycarpus dacrydioides).NewZealandNaturalSciences16:31–37.

Ellis JC 2005. Marine birds on land: a review ofplantbiomass,speciesrichness,andcommunitycomposition in seabird colonies. Plant Ecology181:227–241.

FlemingCA1939.BirdsoftheChathamIslands,Part1.Emu38:380–413.

FukamiT,WardleDA,BellinghamPJ,MulderCPH,TownsDR,YeatesGW,BonnerKI,DurrettMS,Grant-Hoffman MN, Williamson WM 2006.Above-andbelow-groundimpactsofintroducedpredatorsinseabird-dominatedislandecosystems.EcologyLetters9:1299–1307.

Gillham ME 1956. Ecology of the PembrokeshireIslands IV. Effects of treading and burrowingbybirdsandmammals. JournalofEcology44:51–82.

GillhamME1960.Destructionof indigenousheathvegetation in Victorian sea-bird colonies.AustralianJournalofBotany8:277–317.

GillhamME1961.Alterationofbreedinghabitatbysea-birdsandsealsinWesternAustralia.JournalofEcology49:289–300.

HayRF,MutchAR,WattersWA1970.GeologyoftheChathamIslands.NewZealandGeologicalSurveyBulletin83:1–86.

HolmesF1984.ChathamIslands̀ Rekohu’1791–1984.Waitangi,ChathamIslands:F.Holmes.161p.

Johnson PN 1975. Vegetation and flora of the Solander Islands, southern New Zealand. New ZealandJournalofBotany13:189–213.

JohnsonPN1982.BotanicalnotesonsomesouthernNew Zealand islands. New Zealand Journal ofBotany20:121–130.

KingM2000.Moriori.Apeoplerediscovered.Revisededition.Auckland,Viking.

Lockley RM 1953. Puffins. London, Dent. MaesakoY1985.CommunitystructureofMachilus

thunbergiiforestsdisturbedbybirds(Calonectris leucomelas:streakedshear-water)onKanmurijimaIsland,KyotoPrefecture,Japan.JapaneseJournalofEcology35:387–400[inJapanesewithEnglishsynopsis].

Maesako Y 1999. Impacts of streaked shearwater(Calonectris leucomelas) on tree seedlingregenerationinawarm-temperateevergreenforeston Kanmurijima Island, Japan. Plant Ecology145:183–190.

MartinTJ,OgdenJ2006.WinddamageandresponseinNewZealandforests:areview.NewZealandJournalofEcology30:295–310.

MizutaniH,WadaE1998.Nitrogenandcarbonisotoperatios in seabird rookeries and their ecologicalimplications.Ecology69:340–349.

MulderCPH,KeallSN2001.Burrowingseabirdsand

reptiles:impactsonseeds,seedlingsandsoilsinisland forest in New Zealand. Oceologia 127:350–360.

NilssonRJ,KennedyES,WestJA1994.Thebirdlifeof South East Island (Rangatira), ChathamIslands,NewZealand.Notornis41(Supplement):109–125.

NortonDA,PalmerJG,OgdenJ1987.DendroecologicalstudiesinNewZealand1.Anevaluationoftreeageestimatesbasedonincrementcores.NewZealandJournalofBotany25:373–383.

OkazakiM,OshidaY,MalonyR,WarhamJ1993.EffectsofsootyshearwatersPuffinus griseusonsurfacesoilsonMoutuaraIsland,NewZealand.YamashinaInstituteforOrnithology25:137–143.

PetersonCJ2000.CatastrophicwinddamagetoNorthAmerican forests and the potential impact ofclimatechange.ScienceoftheTotalEnvironment262:287–311.

PolisGA,HurdSD1996.Linkingmarineandterrestrialfoodwebs:allochthonous input from theoceansupports high secondary productivity on smallislandsandcoastallandcommunities.AmericanNaturalist147:396–423.

RichardsR1982.WhalingandsealingattheChathamIslands.Canberra,RoebuckSocietyPublicationNo.21.

RichardsRM1972.AtentativepopulationdistributionmapoftheMoriorisofChathamIsland,circa1790.JournalofthePolynesianSociety81:351–374.

Ritchie IM 1970.A preliminary report on a recentbotanical survey of the Chatham Islands.Proceedings of the New Zealand EcologicalSociety17:52–56.

Roberts CM 2004. The effects of burrow-breedingseabirds on the forests of South East Island(Rangatira), Chatham Islands. UnpublishedMApplScithesis,LincolnUniversity,Canterbury,NewZealand.

RunkleJR1982.Patternsofdisturbanceinsomeold-growthmesicforestsofeasternNorthAmerica.Ecology63:1533–1546.

Schnitzer SA, Carson WP 2001. Treefall gaps andthemaintenanceofspeciesdiversityinatropicalforest.Ecology82:913–919.

SiegfriedWR1981.Therolesofbirdsinecologicalprocessesaffectingthefunctioningoftheterrestrialecosystematsub-antarcticMarionIsland.ColloqueSurLesEcosystemesSubantarctiques,Paimpoint,C.N.F.R.A.51:493–499.

SmithVR1978.Animal–plant–soilnutrientrelationshipsonMarionIsland(Subantarctic).Oecologica32:239–253.

Stewart GH, Rose AB, Veblen TT 1991. Forestdevelopmentincanopygapsinold-growthbeech(Nothofagus) forests, New Zealand. Journal of


VegetationScience2:679–690.StokesMA,SmileyTL1968.Tree-ringdating.Chicago,

UniversityofChicagoPress.Sullivan JJ 2003. Density-dependent shoot-borer

herbivory increases the age of first reproduction and mortalityofneotropicaltreesaplings.Oceologia136:96–106.

SullivanW,WilsonK-J2001.DifferencesinhabitatselectionbetweenChathampetrels(Pterodroma axillaris) and broad-billed prions (Pachyptila vittata):implicationsformanagementofburrowcompetition. New Zealand Journal of Ecology25(2):65–69.

TaylorGA1991.FloraofSouthEastIsland(Appendix3). In:Reporton theChathamIsland taikoandChatham Island petrel recovery programmes(1990/91). Threatened Species OccasionalPublication No. 2. Wellington, Department ofConservation.

TennysonAJD, Millener PR 1994. Bird extinctionsandfossilbonesfromMangereIsland,ChathamIslands.Notornis41(Supplement):165–178.

ThompsonCS1983.TheweatherandclimateoftheChatham Islands. New Zealand MeteorologicalService Miscellaneous Publication 115(13).Wellington,GovernmentPrinter.

VeitchCR,BellBD1990.EradicationofintroducedanimalsfromtheislandsofNewZealand.In:TownsDR,DaughertyCH,AtkinsonIAEedsEcologicalrestorationofNewZealandislands.ConservationSciencePublicationNo.2.Wellington,DepartmentofConservation.Pp.137–146.

WainrightSC,HaneyJC,KerrC,GolovkinAN,FlintMV 1998. Utilization of nitrogen derived fromseabird guano by terrestrial and marine plantsatSt.Paul,PribilofIslands,BeringSea,Alaska.MarineBiology131:63–71.

Walls GY, Baird A, de Lange PJ, Sawyer J2003. Threatened plants of the ChathamIslands.WellingtonConservancy,DepartmentofConservation.

WalshD,Kirkpatrick JB,Skira IJ1997.Vegetationpatterns,environmentalcorrelatesandvegetationchangeinaPuffinus tenuirostrisbreedingcolonyat Cape Queen Elizabeth,Tasmania.AustralianJournalofBotany45:71–79.

WarhamJ1996.Behaviour,populationbiologyandphysiologyofpetrels.London,AcademicPress.

Warham J, Wilson GJ 1982. The size of the sootyshearwaterpopulationattheSnaresIslands,NewZealand.Notornis29:23–30.

Was NW, Sullivan WJ, Wilson K-J 2000. Burrowcompetition between broad-billed prions(Pachyptila vittata)andtheendangeredChathampetrel (Pterodroma axillaris). Science forConservation 167. Wellington, Department ofConservation.41p.

WattersWA1978.NotesonthegeologyofMangereand South-East islands, Chatham Islands. NewZealandGeologicalSurveyreportG22.32p.

WattersWA,CampbellHJ,AdamsCJ1987.NotesonthegeologyofSoutheast IslandandStarKeys,ChathamIslands.In:BeggsJed.Researchnotes1987. Record/New Zealand Geological Survey20.Pp.139–143

WaughSM,CabreraH,WoodGC,DavisLS2003.BurrowoccupancyinWestlandpetrels(Procellaria westlandica).Notornis50:123–127.

West JA,NilssonRJ1994.Habitat use andburrowdensitiesofburrow-nestingseabirdsonSouthEastIsland,ChathamIslands,NewZealand.Notornis41(Supplement):27–37.

Wills-JohnsonTMK1996.Introduction.In:DepartmentofConservationed.TheChathamIslandsheritageand conservation. Christchurch, CanterburyUniversityPress.Pp.11–20.EditorialBoardmember:DavidWardle

Documents

Burrowing seabirds affect forest regeneration, Rangatira ...tomtit (Petroica macrocephala chathamensis), marinered-crowned parakeet (Cyanoramphus novaezelandiae chathamensis), warbler