群落相似性和聚类分析.ppt

《群落相似性和聚类分析.ppt》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。

1、EstimatingCommunityParametersCommunityecologistsfaceaspecialsetofstatisticalproblemsinattemptingtocharacterizeandmeasurethepropertiesofcommunitiesofplantsandanimals.Onecommunityparameterissimilarity.Speciesdiversityisanotheroneofthemostobviousandcharacterist

2、icfeaturesofacommunity.1.MeasurementofSimilarity2.SpeciesDiversityMeasures第四章群落相似性和聚类分析第一节相似性测量在群落研究中，生态学家经常会得到某一群落的物种组成和数量。例如在保护区研究中，我们经常要回答的问题是这几个保护区他们在区系组成上有什么不同？哪些更相似，哪些差异较明显？要回答群落分类的这样复杂问题，我们先以测量两个群落的相似性着手。4.1.1BinaryCoefficients4.1.2DistanceCoefficient

3、s4.1.3CorrelationCoefficients4.1.4Morisita’sIndexofSimilarityBinaryCoefficientsThesimplestsimilaritymeasuresdealonlywithpresence/absencedata.Thebasicdataforcalculatingbinary(orassociation)coefficientsisa2×2table.SampleANo.ofspeciespresentNo.ofspeciesabsentab

4、cdSampleBNo.ofspeciespresentNo.ofspeciesabsentWherea=NumberofspeciesinsampleAandsampleB(jointoccurrences)b=NumberofspeciesinsampleBbutnotinsampleAc=NumberofspeciesinsampleAbutnotinsampleBd=Numberofspeciesabsentinbothsamples(zeromatches)where=Jaccard’ssimilar

5、itycoefficient=Asdefinedaboveinpresence/absencematrixBinaryCoefficientsThereisconsiderabledisagreementintheliteratureaboutwhetherdisabiologicallymeaningfulnumber.Therearemorethan20binarysimilaritymeasuresavailableintheliterature(CheethamandHazel1969),andthey

6、havebeenreviewedbyCliffordandStephenson(1975)andbyRomesburg(1984).CoefficientofJaccardThecoefficientofJaccardisexpressedasfollows:where=Euclideandistancebetweensamplesand=Numberofindividuals(orbiomass)ofspeciesinsample=Numberofindividuals(orbiomass)ofspecies

7、insample=TotalnumberofspeciesEuclideanDistanceThisdistanceisformallycalledEuclidiandistanceandcouldbemeasuredfromFigure11.2witharuler.Moreformally.Euclideandistanceincreaseswiththenumberofspeciesinthesamples,andtocompensateforthis,theaveragedistanceisusually

8、calculated:where=AverageEuclideandistancebetweensamplesjandk=Euclideandistance(calculatedinequation11.5)n=NumberofspeciesinsamplesBothEuclideandistanceandaverageEuclideandistancevaryfrom0toinfin