平潭縣植物多樣性研究及其植被景觀建模.pdf

1、摘要中文摘要景觀多樣性業(yè)已成為景觀生態(tài)學(xué)在生物多樣性研究中的一個重點，本文依托景觀生態(tài)學(xué)理論，對平潭縣主要植物群落物種多樣性進(jìn)行了野外調(diào)查，并對其進(jìn)行分析，同時結(jié)合衛(wèi)星遙感影像，對研究區(qū)植被景觀多樣性進(jìn)行了研究，運用BP神經(jīng)網(wǎng)絡(luò)建立了研究區(qū)植被景觀人工神經(jīng)網(wǎng)絡(luò)模型，并對研究區(qū)植被景觀多樣性與自然和社會經(jīng)濟因素間關(guān)系進(jìn)行了擬合，在此基礎(chǔ)上對平潭縣植被景觀形成機制進(jìn)行了分析，并提出針對平潭的植被建設(shè)優(yōu)化對策。通過研究得到如下結(jié)論：(1)研究

2、區(qū)主要植物群落的總體物種多樣性較低，以人工植物群落低于半自然和自然灌叢為特征，群落生長型物種多樣性表現(xiàn)為喬木層低于灌木層低于草本層。與物種多樣性有關(guān)的均勻度指數(shù)也反映了相同趨勢的人工林和天然灌叢的差異。(2)平潭縣植被景觀以木麻黃林和次生相思樹林為主要構(gòu)成部分，林地多分布于沿海地帶，景觀斑塊大小以細(xì)粒為主，形狀指數(shù)總體較小，景觀異質(zhì)性較差，各類景觀要素所占比例不協(xié)調(diào)。(3)基于Matlab平臺，對研究區(qū)植被景觀建立了BP神經(jīng)網(wǎng)絡(luò)模型，選

3、用居民點個數(shù)、風(fēng)力級別和距海邊距離三個因素作為輸入變量，植被景觀蔓延度指數(shù)、分維數(shù)和多樣性指數(shù)作為輸出變量，結(jié)果表明，該模型的精度高，平均誤差為744％，最小誤差僅為018％，在此基礎(chǔ)上，將訓(xùn)練好的模型用來對植被景觀與自然和社會經(jīng)濟因素的關(guān)系進(jìn)行擬合，得到了較好的效果，說明該模型對植被景觀格局的定量分析和預(yù)測具有一定的指導(dǎo)意義。關(guān)鍵詞：植物群落物種多樣景觀多樣性BP神經(jīng)網(wǎng)絡(luò)平潭縣AbstractAbstractLandscapedive

4、rsityhasbecomeamajorconcernoflandscapeecologyinbiodiversityresearchRelyingonlandscapeecologytheorythispaperdidfieldinvestigationandanalysisaboutspeciesdiversityofmainplantcommunityinPingtanCountyandcombined誠msatelliterem

5、otesensingimage，vegetationlandscapediversityWasdiscussedinthestudyareaUsingBPneuralnetwork，artificialneuralnetworkmodelofvegetationlandscapeintheareawasestablishedandtherelationshipbetweenvegetationlandscapediversityandn

6、aturalandsocio—economicfactorsarefittedOnthisbasis，formationmechanismofvegetationlandscapeinPingtanCountyWflSanalyzedandsolutionsofvegetationconstructionoptimizationwereputforwardConclusionsareasfollows：(1)Theoverallspec

7、iesdiversit)rofmainplantcommunityislowerintheresearchareacharacterizedbythatartificialplantcommunityislessthansemi—naturalshrubandgrowthtypespeciesdiversityshowstreelayershrublayerherblayerEvennessindexwhichisrelatedtosp

8、eciesdiversityalseflectsthedifferencesbetweenartificialforestsandnaturalshrublandwhichhavethesametrend(2)CasuarinaequisetifoliaforestsandsecondaryAcaciaspparethemaincomponentpartsofvegetationlandscapeinPingtanCountyThewo

9、odlandiswidelydistributedalongthecoastLandscapepatchesaremainlyfine—grained，shapeindexisgenerallysmall，landscapeheterogeneityisworseandvariouslandscapecomponentsarenotinharmoniousproportions(3)BasedonMatlabplatform，BPneu

10、ralnetworkmodelofvegetationlandscapeWaSsetupNumberofsettlements，windscaleanddistancefromthecoastarechosenastheinput“礦ariablesandcontagion(CONTAG)index，dimensionanddiversityindexareoutputvariablesTheresultsindicatethatthe

11、modelhashighprecisionwithaverageerror744％andminimumerror018％ThenthewelltrainedmodelWasusedtofittherelationshipbetweenvegetationlandscapediversityandnaturalandsocio—economicfactorsItdemonstratesthatthemodelmightbeusefulin