Seriation in Paleontological Data Using Markov Chain Monte Carlo Methods 英文参考文献.docVIP

SeriationinPaleontologicalData UsingMarkovChainMonteCarloMethods Kai Puolama¨ki1*,Mikael Fortelius2,Heikki Mannila3 1LaboratoryofComputerandInformationScience,HelsinkiUniversityofTechnology,Espoo,Finland,2DepartmentofGeologyandInstituteofBiotechnology,Universityof Helsinki,Helsinki,Finland,3HIITBasicResearchUnit,UniversityofHelsinkiandHelsinkiUniversityofTechnology,Helsinki,Finland Givenacollectionoffossilsiteswithdataaboutthetaxathatoccurineachsite,thetaskinbiochronologyistofind goodestimatesfortheagesororderingofsites.Wedescribeafullprobabilisticmodelforfossildata.Theparameters of the model are natural: the ordering of the sites, the origination and extinction times for each taxon, and the probabilities of different types of errors. We show that the posterior distributions of these parameters can be estimatedreliablybyusingMarkovchainMonteCarlotechniques.Theposteriordistributionsofthemodelparameters canbeusedtoanswermanydifferentquestionsaboutthedata,includingseriation(findingthebestorderingofthe sites)andoutlierdetection.Wedemonstratetheusefulnessofthemodelandestimationmethodonsyntheticdataand on real data on large late Cenozoic mammals. As an example, for the sites with large number of occurrences of commongenera,ourmethodsgiveorderings,whosecorrelationwithgeochronologicagesis0.95. Citation:Puolama¨ki K,ForteliusM,MannilaH(2006)SeriationinpaleontologicaldatausingMarkovchainMonteCarlomethods.PLoSComputBiol2(2):e6. available.Theseincludethegraph-theoreticalunitaryassoci- ationsmethodbyGuexetal.[16–18],parsimonyanalysis[19– Introduction Seriation, the task of temporal ordering of fossil occur- rences by numerical methods, and correlation, the task of determining temporal equivalence, are fundamental prob- lems in paleontology. Fossils have been used for both tasks since the very beginnings of modern paleontology [1,2]. However, the recent advent of large fossil databases [3] and theincreasedemphasisonquantitativeanalysisofbiological patternsindeeptime(e.g.,[4–7])