Use Of CART Analysis To Differentiate Pollen Morphology Of Common Woody Plants In Changbai Mountains And Taiyue Mountains

Changbai Mountains and Taiyue Mountains are rich in plant species,which can provide abundant materials for pollen morphology research.Pollen morphological characteristics of species from the Changbai Mountains(238 species of 163 genera belonging to 41 families),and from the Taiyue Mountains(153 species of 119 genera belonging to 48 families),are observed under light microscope(LM),a total of 391 species of 213 genera belonging to 71 families.This study provides modern pollen morphology data for recovering the history of vegetation evolution,reconstructing paleoclimate,and revealing environmental changes by using corresponding Stratigraphic fossil pollen.In order to adapt to the research trend of pollen morphology,this study uses the Angiosperm Phylogeny Group(APG)to classify plant pollen in order to solve the problem caused by pollen morphology research due to different classification systems.In this study,pollen morphology of some common woody plants were described and illustrated,including: Alnus Mill.(Betulaceae),Betula L.(Betulaceae),Lonicera L.(Caprifoliaceae),Syringa L.(Oleaceae),Pinus L.(Pinaceae),Rhamnus L.(Rhamnaceae),Salix L.(Salicaceae)Acer L.(Sapindaceae).The results showed that the pollen morphology of the same genus but different species is different,but it is difficult to distinguish by numerical statistics and text description.We used classification regression tree model to differentiate pollen morphology of the above-mentioned woody plant species.The main results are as follows: Pollen morphology of the same genus but different species is different.Pollen can be identified to species by using pollen morphological variables.Pollens of different genera have different morphological variables,so it is necessary to establish their own regression tree model.The identification results of CART model are as follows:(1)The classification regression tree used one morphological variable(pore depth)to distinguish two species pollen of Alnus Mill.(Alnus mandshurica.and Alnus sibirica).The pollen with a smaller pore depth(<2.605?m)were identified by the CART analysis as Alnus mandshurica,large grains(?2.605?m)were identified as Alnus sibirica.(2)Pollen grains of Betula dahurica were significantly larger than those of Betula platyphylla for grain diameter.The pollen with large grain diameter(?25.025?m)were Betula dahurica,smaller grain diameter(<25.025?m)were Betula platyphylla.(3)The classification regression tree used four morphological variables(the equatorial axis width,the ratio of P and E,exine thickness,the polar axis length)to distinguish four species pollen of Lonicera L.(Lonicera ferdinandii,Lonicera maackii,Lonicera ruprechtiana and Lonicera tatarinowii).(4)The classification regression tree used there morphological variables(the polar axis length,the equatorial axis width,colpus width)to distinguish four species pollen of Syringa L.(Syringa dilatata,Syringa oblata,Syringa pubescens subsp.microphylla and Syringa reticulata var.amurensis).(5)The classification regression tree used four morphological variables(grain length,corpus height,corpus breadth,bladder width)to distinguish there species pollen of Pinus L.(Pinus bungeana,Pinus sylvestris var.sylvestriformis and Pinus tabuliformis).(6)The classification regression tree used there morphological variables(colpus width,exine thickness,the equatorial axis width)to distinguish five species pollen of Rhamnus L.(Rhamnus arguta,Rhamnus davurica,Rhamnus parvifolia and Rhamnus utilis).(7)The classification regression tree used there morphological variables(the equatorial axis width,the polar axis length,exine thickness)to distinguish five species pollen of Salix L.(Salix gracilistyla,Salix integra,Salix polyadenia,Salix siuzevii and Salix viminalis).(8)The classification regression tree used there morphological variables(exine thickness,the ratio of P and E,colpus width)to distinguish four species pollen of Acer L.(Acer ginnala,Acer mono,Acer negundo and Acer ukurunduense).By using the method of repeated sampling to test the 8 developed models,correctly predicted the identity of all genera and species is more than 80%,which shows that the classification regression tree model has a certain reference in pollen morphological differentiation.In this paper,the pollen morphology of the same plants in Changbai Mountains,Taiyue Mountains and Northern Europe area were compared.Temperature and precipitation may be the main factors that cause the change of pollen morphology and grain size,which needs further observation and confirmation.