【作者】 孙超；

【导师】 李月芬；

【作者基本信息】 吉林大学 ， 土地资源管理， 2012， 硕士

【摘要】 大安市姜家店草场位于松嫩平原中部低平原，该区生态环境比较脆弱。近年来，受到全球变化和人为因素的影响，草地面积逐渐减少，植被覆盖度降低，草地沙化、盐碱化、退化三化现象严重，草地生态环境不断恶化，已经开始制约当地社会经济的发展。研究该区退化草地的相关问题对科学保护草地资源、抑制草地退化、修复草地生态环境具有重要的意义。本文以植被-土壤作为系统研究，基于生态化学计量学及元素限制性理论，运用SPSS、Matlab等软件分析羊草叶片碳氮磷生态化学计量特征与土壤营养元素、化学性质间相互关系，应用RBF人工神经网络建立羊草叶片碳氮磷含量的预测模型，并采用K均值聚类分析方法对草地退化程度进行评价，进而分析不同退化程度土壤化学性质及羊草叶片生态化学计量特征。研究结果表明：1.研究区土壤有机质和全氮分别低于全国土壤平均水平(31.00mg·g^-1、1.60mg·g^-1)，根据全国第二次土壤普查分级标准，土壤有机质及全氮含量属三级中等水平，土壤全磷含量属六级极贫乏水平(小于0.4mg·g^-1)；土壤C/N值、C/P值和N/P值分别为9.81、41.72和4.39均低于我国平均值；土壤速效氮、速效磷较贫乏，速效钾较丰富；土壤既存在盐化又存在碱化，可溶盐总量和碱化度与土壤营养元素含量显著负相关。2.大安市姜家店草场羊草叶片碳氮磷的含量分别为443.75mg·g^-1，18.70mg·g^-1，1.30mg·g^-1，皆低于全球陆生植物叶片碳氮磷的平均含量(464mg·g^-1，20.6mg·g^-1，2.0mg·g^-1)，并且氮、磷低于我国植物叶片氮、磷平均含量；羊草叶片碳的空间差异性明显低于氮和磷；羊草叶片C/N值、C/P值、N/P值皆高于全球陆生植物叶片C/N值、C/P值、N/P值；羊草叶片14＜N/P＜16，研究区羊草的生长同时受N、P的限制，N、P含量显著正相关；叶片C与土壤碳氮磷正相关，叶片N与土壤碳氮磷含量相关性不显著，叶片P与土壤氮相关性显著。3.基于人工神经网络的工作原理，采用土壤营养元素及相关化学性质作为输入层，羊草叶片碳氮磷含量作为输出层，利用Matlab软件建立RBF人工神经网络模型用于模拟预测羊草叶片碳氮磷含量与土壤化学性质之间的关系是可行的，经检验相对误差均在20%之内，预测精度较高。4.通过K均值聚类分析法利用土壤和植物指标将研究区85个采样点划分为四种类型，并与野外调查实际情况相结合确定采样点草地退化程度。结果表明研究区轻度退化草地占5.88%；中度退化草地占9.41%；重度退化草地占35.29%；极度退化草地占49.41%，研究区草地退化情况严峻。5.随着草地退化程度加重研究区土壤全碳、全氮含量逐渐降低，全磷含量变化幅度较小；土壤C/N值、C/P值和N/P值总体呈现出下降的趋势；速效氮含量中度退化＞轻度退化＞重度退化＞极度退化，速效磷含量逐渐下降，速效钾含量从轻度退化阶段(214.98mg·kg^-1)急剧下降到中度退化阶段(144.63mg·kg^-1)，随后保持相对稳定；pH值、可溶盐总量和碱化度随草地退化程度的加剧，逐渐升高，草地退化程度状况与土壤盐碱化程度状况具有正相关关系。6.羊草叶片C含量表现为中度退化＞轻度退化＞重度退化＞极度退化；叶片N含量表现为中度退化＞极度退化＞轻度退化＞重度退化；羊草叶片P含量表现为先降低后增加；C/N值呈波状下降，C/P值、N/P值均为先上升后下降；轻度退化阶段羊草生长同时受N、P的限制，在中度退化阶段相对于N主要受到P的限制，在重度退化和极度退化阶段相对于P主要受N的限制；除营养元素以外羊草的生长速率还受其它土壤理化性质的影响。更多还原

【Abstract】 Jiangjiadian meadow of Da’an city locates at the low plain in the middle of SongnenPlain. The entironment here is weaker. In recent years, influenced by global changing andhuman reason, the grassland area is reducing gradually; meanwhile, vegetation coveragelevel is depressing. Desertification, salinization and degeneration of the grassland aremore serious and its entironment is getting worse. All these have restricted thedeveloping of local society and economy. The research of grassland degenerationquestion in this area has important meaning to protecting grassland resource scientifically,to controlling grassland degeneration, and to repairing grassland entironment.This paper takes Plant-Soil as a system to research. Based on ecologicalstoichiometry and element theory of constraints, using software such as SPSS and Matlab,I analyzed the interaction between the ecological chemical measurement features of C, Nand P in L. chinensis leaves and soil element and its chemical character. Using RBFArtificial Neural Networks （ANNs）, I built the forecast model of the content of C,N andP elements in L. chinensis leaves.Using K value clustering analysis method, Ianalyze thedegeneration level of the grassland. After that, I analyze the soil chemistry character ofdifferent degeneration levels, and ecological chemical measurement features of L.chinensis leaves. The result of the research proves:1. The contents of soil gein and TotalN in research area are all lower than nationalaverage. Compared with national second time soil survey standard, the contents of soilgein and TotalN belong to middle third level, and the content of soil TotalP belongs to extremely poor level (lower than0.4mg·g^-1). In the soil, the value of C/N, C/P and N/Pare9.81、41.72、4.39, all lower than national average. The contents of available N andavailable P are poor, available K is richer. There are both a soil salinization and alkalineexistence. Soluble salt and alkaline level show significant negative correlation with thecontent of soil nutrient elements.2. The contents of C, N and P in L. chinensis leaves at Da’anJiangjiadian meadoware443.75mg·g^-1、18.70mg·g^-1、1.30mg·g^-1. They are all lower than global averagecontents in terraneous plant leaves (464mg·g^-1、20.6mg·g^-1、2.0mg·g^-1). And thecontent of N and P are even lower than national data. The space difference of CinL.chinensis leaves is obviously lower than N and P. The values of C/N, C/P and N/P inL.chinensis leaves are all higher than those in global terrestrial plant leaves. InL.chinensis leaves,14＜N/P＜16. The growth of L.chinensis in research area islimited by both N and P.The content of N and P show significant positive correlation; Thecontent of C inleaves show significant positive correlation with content of C, N, P andavailable P in soil. There is no significant correlation between the content of N in leavesand C, N, P in soil.3. Based on the principle of ANNs, I take soil nutrient elements and relevantchemistry character as input layer. The contents of C, N and P in L. chinensis leaves areregarded as output layer. It’s feasible to create RBF ANNS model with Matlab and use itfor calculate the interaction between the contents of C, N and P in L. chinensis leaves andsoil chemical nature. After checking, the relative error is within20%. The forecastingaccuracy is higher.4. Through K value clustering analysis method, using soil and plants as index, Idivide the85sampling points in research area into four types of degeneration level. Andconfirm the grassland degeneration level combined with the physical truth of field survey.The result shows that in research area, grassland of low-grade degeneration takes5.88%,mid-grade takes9.41%, high-grade takes35.29, and extremely high-grade takes49.41%.The regeneration of grassland in research area is severe.5. With the aggravation of grassland degeneration degree, the contents of TotalC and TotalN get lower gradually, but the changing of TotalP content is smaller. As a whole,the values of C/N, C/P and N/P all show decreasing trends. In different degenerationlevels, the content sequence of available N in L. chinensis leaves is mid-grade＞low-grade＞high-grade＞extremely high-grade. The content of available P declinesgradually with the aggravation of grassland degeneration degree. And at the same time,the content of available K declines sharply from low-grade of degeneration(214.98mg·kg^-1)to mid-grade (144.63mg·kg^-1), then remains comparatively steady.PH, total soluble salt and basification degree ratchet up with the aggravation of grasslanddegeneration. Grassland degradation degreeand soil salinity level condition is correlated.6. In different degeneration levels, the content sequenceof C in L. chinensis leavesis mid-grade＞low-grade＞high-grade＞extremely high-grade.The content sequenceof Nis mid-grade＞extremely high-grade＞low-grade＞high-grade. Except extremelyhigh-grade degeneration stage, the variation trend of N in leaves is same as available N.The content of P in L. chinensis leaves shows decrease then increase. C/N value declineswavily. In low-grade degeneration stage, the growth of L. chinensis is limited both by Nand P. In mid-grade degeneration stage, it is limited mainly by P. In high-grade andextremely high-grade degeneration stages, it is limited mainly by N. Except nutrientelement, the growth speed of L. chinensis is still influenced by other soil physical andchemical properties.更多还原