TESTINGFOREVIDENCEOFHIERARCHICALSYSTEMSTRUCTUREIN THEMANITOBANBOREALFOREST By PAULC.MARPLES ADISSERTATIONPRESENTEDTOTHEGRADUATESCHOOL OFTHEUNIVERSITYOFFLORIDAINPARTIALFULFILLMENT OFTHEREQUIREMENTSFORTHEDEGREEOF DOCTOROFPHILOSOPHY UNIVERSITYOFFLORIDA 1998 ACKNOWLEDGMENTS Thiseffortwouldhavebeeninnowaypossiblewithouttheadvice, support,andinspirationprovidedbyC.S."Buzz"Holling. Buzzhastherare giftofbig-pictureperspective. Ithasbeenmyprivilegetopulloutafew questionsfromthebodyoftheoryheproposesandperformteststhat % challengesomebasictheoreticalassumptionsinecology. CarmineLanciani, LarryMcEdward,ColinChapman,andHenryGholzmustbethankedfor theirfairandopen-mindedguidancethroughoutthisendeavor. Asmy committeemembers,theirsuggestionsregardingthedesignofthestudy, analysisofdata,andeditingthemanuscripthavebeenabsolutelyessentialfor success. Allthehelpprovidedbythefaculty,staff,andgraduatestudentsof thezoologydepartmenthasbeenindispensable. IamdeeplygratefulforalltheassistanceprovidedbyCraigAllen,Toni Carter,KatherineCullison,PaulDrewa,JoseFragoso,LanceGunderson,Mark Hostetler,GarryPeterson,DanPickford,KenPortier,RustyPrichard,Carla Restrepo,JeffSailer,andJanSendzimir. KatherineCullison,inparticular, providedpersonalguidance,support,andassistancewithoutwhichIwassure togoastray. Nowwecanbeginworkonthedomegreenhousemadeoutof home-culturedbeeswaxandtonguedepressorswealwaysdreamedof. My goodfriendMark"Hos"Hostetlersometimesstoppedbuggingmelong enoughformetogetsomethingdone. Seriously,hehasagiftforconverting beerandfrustrationintosolutionsandproductivity. Inapplicationofthe generaltheory,Inevercouldneverhavebattledthe"devilinthedetails" 11 withouthim. Besides,Iswear,InevercheatedatRisk. Toallmyfriendsand family,averyfondthankyou. HeyJoey,guesswhat.I'maprofessional! IhaveGarryPetersontothankfortheuniqueinsightthatmodeling fireprovides. Wildfireactsoverfarlargerscalesinspaceandtimethan peoplecanordinarilyperceive. Amodelworldallowsoneto"see"how patternsanddynamicscanemergeinintriguingvariety. There'snoquestion thatmyfruitfulcollaborationwithGarryhelpedguidethisdissertation throughout. NoneoftheworkonfireinManitobacouldhavebeenbegunwithout thevolumesofdataandkeyconsultationprovidedbyManitobaNatural Resources,ForestryBranch. Whocanbeateffective,immediate,professional helpwithafriendlysmile? GlennPeterson,inparticular,wentoutofhisway toprovideanyassistancehecould. ManythankshavetoalsogotoGerald Becker,NanTuinoff,andBillMeddforalloftheirhelp. JanSendzimirandI willneverforgetthejoyofflyingovertheborealforestinahelicopter. We haveBillMeddtothankforarrangingthat. AsIpictureitinmymind,Ihave toparaphrase,YuriGagarinandsay"Iamdragonfly." Onlyflyingcreatures canseehowfirereworksentirelandscapes. Onlybywalkingabout,canonegetaninklingastohowtheboreal forest"livesandbreathes." I'meversothankfultoJamesEhnesandNorm KenkelattheUniversityofManitobaforwalkingaboutwithmeintheboreal forest. IdoubtI'dhavemuchtosayofinteresttoanyoneiftheyhadn'ttaken thetimetoshowmethenaturalhistoryofSoutheastManitoba. Iamalso gratefulfortheassistanceprovidedbyWilliamO.PruittandHugoVeldhuis attheUniversityofManitoba,HaroldPeacockandPeterClarksonatAbitibi- Price,andRoyDixonandHartleyPokrantattheRemoteSensingCenter. in IamgratefulforthefundingsupportprovidedbyArthurR.Marshall, Jr.EndowedChairinEcologicalSciences. Iamalsogratefulforthefunding supportprovidedbyNASA/EOSgrantnumbersNAGW2524aspartofthe InterdisciplinaryScientificInvestigationsprogramandNAGW3698 aspart oftheTerrestrialEcologyprogram. IV TABLEOFCONTENTS ACKNOWLEDGMENTS ii LISTOFTABLES viii LISTOFFIGURES ix ABSTRACT xii 1 INTRODUCTIONANDOVERVIEW 1 ThreeTheoreticalViewsofEcosystems 2 ThreeTheoreticalViewsofWildfireintheMid-Continental BorealForest 4 AlternativeTheoreticalViewsoftheMammalComponentof BorealForestCommunities 6 Overview 7 2THEEFFECTOFLANDSCAPE-SCALEVEGETATIONPATTERNON BOREALFORESTWILDFIREINSOUTHEASTMANITOBA, CANADA Introduction 9 Wildfire,theBorealForest,andClimateChange 10 AHierarchicalViewoftheMid-ContinentalBorealForest 11 StudySite 13 Methods 15 Acquisitionofdata 15 Creationofmasterfiremap 15 Characterizingfirehistory 16 Calculationofburnhazard 17 Results 19 FireHistory 19 BurnHazard 23 Discussion 26 FireHistory 26 GeneralIndicatorsofBurnHazardfromFireHistory 29 DirectMeasureofBurnHazardanditsImplications 30 Conclusions v 3DETECTIONOFCOMPLEXMULTIMODALITYINECOLOGICAL DATA Introduction 34 StatisticsandtheDetectionofComplexMultimodality 35 Lump Simulation 39 ABriefOverview 39 ExperimentalDesign:anExample 40 CalculationofObservedMismatch 42 DeterminationoftheSignificanceofMismatch 48 SearchingforConsistentPattern 52 PowerandBehavior 56 TestingforComplexMultimodalityinMammalBodyMassData 61 ExperimentalDesign 62 Results 64 Discussion 66 Asignificantbutimperfectmatch 66 Datasharing 69 Power 70 Implications 71 Summary 72 4TESTINGFOREVIDENCEOFCOMPLEXMULTIMODALITYINFIRE SIZESASEVIDENCEOFHIERARCHICALSYSTEMSTRUCTURE INTHEBOREALFORESTOFMANITOBA,CANADA Introduction 74 Wildfire,theBorealForest,andClimateChange 75 MechanismsUnderlyingHierarchi—calSystemStructureinthe Mid-ContinentalBorealForest EvidenceandAlternatives 76 StudySite 80 Methods 83 Digitization,Mapping,andGeneralCharacterizationofData 83 Digitizationofamapofecoclimaticregions 83 Mappingwildfire 83 Measuresoffireregime 84 TestingforMultimodalityinFireSizes 85 Results 88 GeneralFireRegime 88 TestsforMultimodalityinFireSizes 93 Discussion 93 GeneralFireRegime 93 MultimodalityandTheoreticalSystemStructureoftheMid- ContinentalBorealForest 94 Conclusions 97 vi APPENDIX MAMMALSOFTHEBOREALFOREST 99 LISTOFREFERENCES 102 BIOGRAPHICALSKETCH 112 Vll LISTOFTABLES Table page 3- 34--1. Searchresultsobtained(whenSccv=0.75)intheexperimentexamining setsA,B,C,andD 51 3-2. SearchresultsobtainedbytheLSexperimentexaminingborealforest mammals 65 3. Shareddata(orpairwisesetintersection)ofborealforestmammals 70 1. ResultsofLumpSimulationexperiments 92 vm ) LISTOFFIGURES Figure page 2-1. ThelocationofthestudysitewithinSoutheastManitoba,Canada 14 2-2. Thestandoriginmapindicatingthedatesofthemostrecentbumasof 1990throughoutthestudyareainSoutheastManitoba 20 22--3. Theproportionofthestudyareaburnedannually:(a)inchronological order,(b)rankordered 21 3- 2-4. Thetime-since-firedistribution 22 2-5. Theburncountmap. Thenumberoftimeseachportionofthe SoutheastManitobastudyareaburnedbetween1929and1990isshown...24 2-6. Theburncountdistributionforthestudyareaasawholeandeach individual subregion 25 7. Theburnhazarddistribution 25 1. 1,2,and10-modenormalkerneldensityestimatesofdatasetsA,B,C, andD 36 3-2. Samplingdistributionsusedintheexampleexercises 41 3-3. Thebest-fit,2-mode,normalkerneldensity-estimateofsetAwithlump intervals identified 42 3-4. Ofthreenominaldatasequences,whichpairingismoreclosely correlated? ThephicoefficientgivesaanindexofcorrelationforXvs.Y andXvs.Zof0.45. Myeyes,however,tellmethatXvs.Zshouldbemore correlated 43 3-5. Thedistance-from-nearest-lump-edgemeasureforsetAwith2lumps .44 3-6. Thedistance-from-lump-edgemeasuresforsetAwith2lumpsandset Bwith1lump 44 3-7. Thefoursubcomponentsneededtocalculatemismatch(m 46 IX 3-8. Theobservedmismatch(m,y)ofsetsAandBoverallresolutionsgiving upto10modes 48 3-9. ThematchsignificanceforallcomparisonsinalltestsintheLS experimentexaminingsetsA,B,C,andD 50 3-10. Theexperiment-wise,typeIerrorrate(Pvalue),broadpower,and narrowpowercalculatedastheexperimentexaminingsetsA,B,C,andD proceedsthrougheachofits6tests 55 3-11. The2-modekernelestimatesofsetsA,B,C,andD 55 3-12. Theexperiment-wise,typeIerrorrate(Pvalue),broadpower,and narrowpowerobtainedbythe10-modeexperiments, (a)whensamplesize equals3nNAbm/and(b)whensamplesizeequals6nNAbm 58 3-13. ThematchsignificanceforallcomparisonsinalltestsintheLS experimentsexaminingsetsA,BA,C,andDA 60 3-14. Theexperiment-wise,typeIerrorrate(Pvalue)andtheprobabilityof 3-findingvariouskindsofmultimodalityinexperimentsshiftingbimodal distributionsout-of-phase 61 43--15. Thethreeborealforeststudyareasfromwhichmammalbodymass datawerecollected:EasternNorthAmerica,NorthernEurope,andEastern Russia 63 3-16. ThematchsignificanceforallcomparisonsinalltestsintheLS experimentexamining the mammals 65 3-17. Theexperiment-wise,typeIerrorrate(Pvalue),broadpower,and narrowpowerobtainedbythemammalexperiment 66 18. The10-modekernelestimatesofmammalsofEasternNorthAmerica, NorthernEurope,andEasternRussia. Alsoshownarethecorrespondent lumpintervals 67 1. Thelocationofthestudyarea:Manitoba,Canada 81 4-2. EcoclimaticregionsofManitoba 82 4-3. Thetotalareaburnedannuallywithin:(a)thehigh-boreal,and(b)the mid-borealecoclimaticregionsofManitoba(1976-1991:includingthe accuratelymappeddataonly) 86 x