基于Sentinel-2数据的草地植物功能多样性遥感反演及其与生产力的关系

生物多样性与生态系统功能的关系是当前生态学研究的焦点和难点。植物功能多样性是影响生态系统功能的重要指标, 开展植物功能多样性的研究对了解生物多样性与生态系统功能之间的关系有着重要意义。传统的草地植物功能多样性研究多以实地调查为主, 不仅费时费力, 而且由于受到时空的限制, 很难拓展到大尺度的研究中。遥感技术的发展为评估草地功能多样性提供了一种经济、有效的手段。该研究选取内蒙古自治区锡林郭勒盟乌拉盖管理区草甸草原为研究区, 利用Sentinel-2卫星影像和野外实测数据, 选取了波段及植被指数等46个特征变量, 探讨了逐步回归、偏最小二乘法(PLSR)和随机森林(RFR)等3种不同方法对草地植物功能丰富度(FRic)、功能均匀度(FEve)和功能离散度(FDiv)的反演精度, 并基于PLSR反演草地地上生物量, 进一步分析了研究区功能多样性与生产力的关系。研究结果表明: (1)波段B11、优化型土壤调节植被指数(OSAVI)、水波段指数(WBI)对FRic解释度最高; 波段B6、B10、B12、类胡萝卜素反射指数1 (CRI1)、双峰光学指数(D)、归一化差值指数45 (NDI45)等6个特征变量对FEve解释度最高; 波段B5、B9、B10、B11、加权差分植被指数(WDVI)、凸包面积等对FDiv解释度最高; (2)基于十折重复交叉验证, 利用逐步回归估算的FRic和FEve反演精度远高于其他两种回归方法, R²分别为0.52和0.44; 而利用PLSR方法估算的FDiv反演精度最高(R² = 0.61); (3)群落地上生物量反演精度为R² = 0.61; FRic与地上生产力的关系最好(R² = 0.40), 其次为FDiv (R² = 0.28)和FEve (R² = 0.27)。研究发现, 基于Sentinel-2卫星影像能较好地反演草地功能多样性和生产力, 为下一步能在大尺度上进行草地功能多样性估算及其与生产力关系研究提供了参考和依据。     

基于物候相机归一化植被指数估算高寒草地植物地上生物量的季节动态

准确评估地上生物量对优化草地资源管理和理解草地碳、水和能量平衡具有重要意义。该文通过近地遥感归一化植被指数(NDVI)构建最优经验模型, 对青藏高原高寒草地地上生物量进行估算。该文利用2018-2019年5-9月野外实测的地上生物量和植物冠层光谱仪(RapidSCAN)测定的NDVI_RS数据, 构建了生长季不同时期地上生物量的估算模型; 并结合2018年NetCam物候相机测定的NDVI_Cam时间序列数据, 实现地上生物量季节动态的模拟。主要结果: (1) NDVI_Cam、NDVI_RS与地上生物量具有相似的单峰型季节变化格局, 但NDVI峰值出现的时间(7月)较地上生物量(8月)更早; (2)基于NDVI的生物量估算最优经验模型在5、7和9月是幂函数, 在6和8月是二次多项式, 估算精度为0.29-0.77; (3)基于NDVI_Cam时间序列数据, 生长季不同时期建模(R² = 0.91)较单一时期(9月)建模(R² = 0.49)对地上生物量季节动态的估算更为准确。这些结果表明, 近地遥感是估算高寒草地植物地上生物量的有效手段, 开展季节性植物生长调查将有助于准确评估草地资源。     

黑河中游荒漠草地地上和地下生物量的分配格局

草地生态系统中地上和地下生物量的分配方式对于研究生态系统碳储量和碳循环有着重要的意义。为了解黑河中游荒漠草地的地上和地下生物量分配格局, 从群落和个体两个水平对黑河中游的地上和地下生物量进行了调查。结果表明: 群落水平上地上生物量介于3.2-559.2 g·m^-2之间, 地下生物量介于3.3-188.2 g·m^-2之间, 个体水平上地上生物量介于6.1-489.0 g·株^-1之间, 地下生物量介于2.4-244.2 g·株^-1之间, 群落水平上的根冠比(R/S)为0.10-2.49, 个体水平上为0.07-1.55, 地下生物量均小于地上生物量, 群落水平上R/S值大于个体水平。群落和个体水平地上和地下生物量的拟合斜率分别为1.1001和0.9913, 与1没有显著差异, 说明地上与地下生物量呈等速生长关系。群落和个体水平土壤表层0-20 cm和0-30 cm的根系生物量分别占全部根系生物量的89.81%、96.95%和81.42%、93.62%, 表明地下生物量主要集中在0-20 cm和0-30 cm土壤表层。     

草地植物群落地上生物量非破坏性估测方法的探讨

 羊草群落的高度和盖度与地上生物量存在良好的复相关关系(R=0.9316),所获多元回归方程可用于该群落地上生物量的估测。通过对羊草群落、羊草+杂类草群落和贝加尔针茅群落植冠红光(0.63—0.69μm)和近红外辐射(0.76—0.90μm)反射率实测数据的分析,表明这些群落的光谱反射率比IR、IR/R和VI与地上生物量呈高度的指数相关关系,其中由VI和IR/R所获各群落不同生长时期地上生物量回归模型的估测效果较好。     

一种基于数码相机图像和群落冠层结构调查的草地地上生物量估算方法

草地地上生物量是影响其生态系统功能最重要的因素之一, 也是草地生态学研究中不可或缺的监测指标。草地地上生物量监测多采用收割法进行, 但这种破坏性取样方法会对研究区域带来巨大干扰, 尤其是面积较小的长期定位监测或者控制实验研究样地, 从而使得地上生物量监测的频次受到很大限制。因此, 通过获取某些原位易测变量, 建立地上生物量的估算方法具有重要意义。该研究依托内蒙古典型草地刈割控制实验平台, 通过数码照片获取不同土地利用方式下的植被覆盖度, 并对样方内的叶面积指数、植被高度、物种多样性等参数进行了测定, 最后利用一元回归模型、逐步回归模型和随机森林模型对地上生物量进行估算。结果表明, 植被覆盖度、叶面积指数、植被平均高度、植被最大高度和物种丰富度是影响地上生物量的主要驱动因素。通过构建适宜于本地的逐步回归模型, 可以实现草地地上生物量的准确预测。在该研究区域中, 预测模型的决定系数(R²) = 0.91, 均方根误差(RMSE) = 35.60 g·m^-2。该研究提供了一种快速、准确且非破坏性测定草地地上生物量的方法, 可作为传统收割法的有效补充。     

三江平原湿地小叶章生产力模拟模型

利用气象台站的常规观测资料,依据植物生长模拟理论,以d为步长,建立了湿地小叶章(Deyeuxia angustifolia)植被生产力动态模拟模型。该模型包括3个子模块:1)光合作用与呼吸作用;2)干物质积累;3)同化物分配,主要考虑了温度和积水因子对植物生长的影响。并利用实测资料对该模型进行了检验,结果表明:小叶章地上活体、枯落物、茎、叶各器官枯落物的模拟值与实测值之间均呈极显著的线性相关(R²分别为0.98、0.99、0.99和0.92)。在相邻区域的检验结果也表明,季节性积水沼泽化草甸小叶章的地上生物量明显高于常年积水沼泽。两类湿地小叶章地上生物量的模拟值与实测值之间均呈极显著线性相关(R²分别为0.66和0.79)。相近区域长期定位观测点连续2年的模拟结果与实测值之间也具有极显著的线性相关(R²分别为0.97和0.76)。     

不同地形条件下植被盖度信息提取技术研究

为系统地研究特定区域的植被盖度信息提取技术, 在不同的地形条件下, 比较了目前流行的多种高光谱遥感植被盖度提取方法。结果表明: 最优高光谱归一化植被指数(NDVI₁)的建模和验证精度均高于其他两种归一化植被指数(NDVI), 直接采用NDVI建立的回归模型对研究区植被盖度的估测能力低于像元二分模型; 阴坡的最佳模型为基于一阶微分的偏最小二乘回归模型(PLSR模型), 其建模决定系数(R²)为0.810, 均方根误差(RMSE)为6.29, 验证R²为0.773, RMSE为8.85; 阳坡的最佳模型为基于二阶微分的PLSR模型, 其建模R²为0.823, RMSE为6.04, 验证R²为0.801, RMSE为7.35; 平原的最佳模型为全受限的线性光谱混合分解模型(FCLS), 其验证R²为0.852, RMSE为5.86。     

黄土高原典型草原地上生物量估测模型

为了寻求有效的草地地上生物量估测方法和精确估测黄土高原典型草原草原地上生物量。于2014年8月中旬,在黄土高原典型草原草原地上生物量达到最大值,分别从单株水平和种群水平进行野外调查。以株高(H)和盖度(C)的复合因子(C×H)为自变量,通过回归分析,建立地上生物量估测模型,采用留一法对其精确性进行评估;并通过校正系数以及群落总生物量估测值和实测值比较单株水平和种群水平所建模型的精确性。结果表明:黄土高原典型草原草地,无论在单株水平还是种群水平,线性和幂函数对该区域生物量的拟合效果更好。估测模型检验结果表明,在单株水平各个物种的生物量估测值与实测值相关性较好,均达到了显著水平(P0.05),其r值均大于0.6,总相对误差RS均小于10%,平均相对误差绝对值RMA(average absolute value of relative error)均小于30%,总生物量的实测值与估测值比较接近,校正系数均接近1;而在种群水平上,虽然各物种的生物量估测值与实测值相关性均达到了显著水平(P0.05),但多数物种平均相对误差绝对值RMA大于30%,总相对误差RS(total relative error)均大于10%,总生物量的估测值均大于实测值,校正系数均偏离了1,说明在黄土高原典型草原通过单株水平建立的物种生物量估测模型的精度优于种群水平建立的物种生物量估测模型的精度。     

青藏高原高寒草地地上植物碳积累速率对生态系统多功能性的影响机制

在全球气候变暖的背景下, 草地作为陆地生态系统碳库的重要组成部分, 其较小幅度的波动, 会影响整个陆地生态系统碳循环和生态系统多功能性(EMF)。地上植物碳积累速率(CAR)表示从生长季初始到生长季生物量峰值的群落地上部分碳累积速率, 能够很好地表征固碳功能、固碳潜力和效率。因此, 植物CAR的变化会改变地上和地下群落维持EMF的能力。目前EMF的相关报道多探讨地上群落多样性和EMF的关系, 而缺乏高寒草地生态系统植被地上CAR对EMF的影响机制研究。该研究目的是探究高寒草地群落CAR对EMF的调控作用、机理和过程, 这将对草地生态系统管理提供理论支持, 并推进对生态系统多功能性维持机制的理解。2015年7-8月, 在青藏高原地区进行草地样带调查, 共计取115个样点。综合土壤有机碳、全氮、全磷、地上和地下生物量以及微生物生物量碳等13种生态系统参数计算生态系统多功能性指数(M)。利用归一化植被指数(NDVI, 1982-2013年)计算并提取2015年物候数据, 最终获得CAR。采用薄盘光滑样条插值法插值气象数据, 提取样点2011-2015年年降水量和年平均气温, 以供分析CAR对EMF的调控机理。主要结果: 地下生物量、土壤有机碳、全磷和微生物生物量碳含量对CAR和M有较高的权重(0.58、0.80、0.83和0.79; 1.05、0.98、1.02和0.97), CAR和M呈显著线性正相关关系(R² = 0.45, p < 0.01)。在降水和气温要素的影响下, 植物地上群落和地下土壤要素的协同作用, 影响植被CAR, 进一步调控EMF。     

氮添加对内蒙古退化草地植物群落多样性和生物量的影响

人为干扰及气候变化导致内蒙古草地发生了大面积退化, 氮添加是促进退化草地生产力恢复的一项重要措施。该文基于2011年建立的氮肥添加实验平台, 以3个不同退化程度(中度退化、重度退化、极度退化)草地群落为研究对象, 设置对照、10、20、30、40和50 g·m ^-2·a ^-1 6种氮添加处理, 分析氮添加对退化草地恢复过程中群落多样性和生物量的影响。结果表明: (1)氮添加降低了中度、重度退化草地恢复进程中物种丰富度和多样性, 对极度退化草地恢复进程中物种丰富度和多样性无明显影响。(2)氮添加促进了3个不同退化程度草地恢复进程中群落地上生物量的增加。(3)氮添加显著增加了群落中禾草的地上生物量及其在群落地上生物量中所占的比例, 降低了杂类草在群落地上生物量中的比例, 但对杂类草地上生物量无显著影响。研究表明在利用施肥措施治理退化草地的过程中, 需要充分考虑草地退化程度以及由氮添加引起的群落多样性和生产力的改变对草地生态系统功能的影响。     

Estimation of biomass allocation and carbon density of Rhododendron simsii shrubland in the subtropical mountainous areas of China

Aims As an important potential carbon sink, shrubland ecosystem plays a vital role in global carbon balance and climate regulation. Our objectives were to derive appropriate regression models for shrub biomass estimation, and to reveal the biomass allocation pattern and carbon density in Rhododendron simsii shrubland.
Methods We conducted investigations in 27 plots, and developed biomass regression models for shrub species to estimate shrub biomass. The biomass of herb and litterfall were obtained through harvesting. Plant samples were collected from each plot to measure carbon content in different organs.
Important findings The results showed that the power and linear models were the most appropriate equation forms. The D and D²H (where D was the basal diameter (cm) and H was the shrub height (m)) were good predictors for organ biomass and total biomass of shrubs. All of the biomass models reached extremely significant level, and could be used to estimate shrub biomass with high accuracy. It was more difficult to predict leaf and annual branch biomass than stem biomass, because leaf and annual branch were susceptible to herbivores and inter-plant competition. The mean biomass of the shrub layer was 20.78 Mg·hm^-2, in which Rhododendron simsii and Symplocos paniculata biomass accounted for 93.63%. Influenced by both environment and species characteristics, the biomass of the shrub layer organs was in the order of stem > root > leaf > annual branch. The root:shoot ratio of the shrub layer was 0.32, which was less than other shrubs in subtropical regions. The relative higher aboveground biomass allocation reflected the adaptation of plants to the warm and humid environment for more photosynthesis. The mean total community biomass was 26.26 Mg·hm^-2, in which shrub layer, herb layer and litter layer accounted for 79.14%, 7.62% and 13.25%, respectively. Litter biomass was relatively high, which suggested that this community had high nutrient return. There were significant correlations among aboveground biomass, belowground biomass and total biomass of shrub layer and herb layer. The mean biomass carbon density of the community was 11.70 Mg·hm^-2 and the carbon content ratio was 44.55%. The carbon density was usually obtained using the conversion coefficient of 0.5 in previous studies, which could overestimate carbon density by 12.22%.     

海南热带雨林国家公园不同植被类型的大型真菌多样性

为揭示海南热带雨林国家公园大型真菌多样性及不同植被类型对真菌群落的影响, 本研究于2020年和2021年湿季对海南热带雨林国家公园内7个管理局辖区开展了大型真菌多样性调查, 比较了不同植被类型(山地雨林、低地雨林、低地雨林次生林、人工林)的大型真菌生活型(共生型、腐生型)组成差异。从设置的58条1 km长的样带内采集到1,869份子实体标本, 根据子实体形态与ITS rDNA序列分析, 从中鉴定出562种真菌, 涉及17目64科174属, 其中80%以上的物种由伞菌目、牛肝菌目、红菇目、多孔菌目、鸡油菌目、锈革孔菌目和炭角菌目构成。大型真菌的营养型以腐生型(占48.2%物种)和共生型(44.8%)为主。每条样带的平均物种丰富度和多度以中海拔的山地雨林最高, 分别为28 ± 5种和33 ± 6个, 而人工林最低, 分别为11 ± 1种和11 ± 2个。植被类型主要影响共生型大型真菌物种丰富度(P = 0.026)和子实体多度(P = 0.019)及Shannon-Wiener多样性(P = 0.028), 但对腐生型大型真菌的影响并不显著。多响应置换过程(multiple response permutation procedure, MRPP)检验结果表明, 不同植被类型对共生型与腐生型大型真菌群落物种组成均有显著影响(腐生型: P = 0.004, 共生型: P = 0.041)。冗余分析(redundancy analysis, RDA)的结果表明, 植被类型对腐生型和共生型真菌群落物种组成差异的解释度均较低(共生型: R² = 0.068, P = 0.004; 腐生型: R² = 0.067, P = 0.004)。海拔仅对腐生型真菌群落物种组成产生微弱影响(R² = 0.029, P = 0.001), 而对共生型真菌影响不显著(R² = 0.024, P = 0.072)。在不同保护地之间, 共生型(R² = 0.148, P = 0.001)与腐生型(R² = 0.123, P = 0.002)真菌物种组成均具显著差异; 基于样带‒真菌矩阵的网络图显示, 海南热带雨林国家公园内尖峰岭、霸王岭、五指山等国家级自然保护区的山地雨林是共生型大型真菌多样性较高区域, 应作为共生型真菌与宿主的优先保护区域。     

基于无人机的冬小麦拔节期表层土壤有机质含量遥感反演

快速监测大面积分布的盐渍化麦田土壤有机质含量,可为推进盐渍土改良和促进碳循环研究提供数据支撑。通过野外采样与获取无人机遥感影像,分别基于裸土和植被情况,采用多元线性回归(MLR)、偏最小二乘回归(PLSR)和支持向量机回归(SVR)3种方法,建立区域有机质含量遥感模型,并进行检验和对比,确定最优的土壤有机质含量反演模型;最后基于最优模型进行研究区表层土壤有机质的反演,并与插值结果进行比较。结果表明: 经5×5的中值滤波处理后的光谱与土壤表层有机质对应最优;3种模型中,SVR模型的预测精度最高,PLSR次之,MLR效果最差。对比两种变量的建模效果,基于植被的SVR建模效果最好,其建模决定系数(R²)、均方根误差(RMSE)分别为0.89、0.20,验证R²、RMSE分别为0.82、0.24;基于裸土的建模效果不理想,最优的也是SVR模型,其建模R²、RMSE分别为0.63、0.26,验证R²、RMSE分别为0.61、0.25。根据最优模型反演得到该区域有机质含量为17.51～22.53 g·kg^-1,平均值为19.51 g·kg^-1,与实地调查结果较为一致;插值结果与反演结果相比,精度受到限制。综上,基于无人机多光谱可以对盐渍土冬小麦拔节期土壤有机质含量进行快速、大范围精准估测。     

刈割、围封、放牧三种利用方式下草原生态系统的多功能性与植物物种多样性之间的关系

随着全球变化对生物多样性的影响不断加剧, 生物多样性与生态系统功能之间相互关系(BEF)的研究显得极为重要。过去的20多年, BEF的研究大多集中在对物种多样性与单一或少数生态系统功能之间关系的探讨, 但生态系统最为重要的价值是同时维持多种服务和功能的能力, 基于此, 该文首次在国内引入近年来不断完善的生态系统多功能性(multifunctionality)的概念, 并对目前主流的评价方法进行了改进, 从而对内蒙古三种利用方式(刈割、围封、放牧)下的草地群落进行了多功能性评价, 并探讨了多功能性与物种多样性之间的关系。结果显示本研究改进的方法和目前主流方法评价得出的多功能性指数在样方和样地尺度上都有很高的相关性(R² = 0.6956, p < 0.0001; R² = 0.9231, p < 0.0001), 表明该文作者改进后的方法是可靠的。重度放牧的草地群落物种多样性水平最低, 绝大多数土壤功能指标较差, 表现出退化特征; 7年的围封和刈割群落均有较高的物种多样性水平和改善的土壤功能指标; 三者的多功能性指数为刈割(0.2178) >围封(0.0704) >放牧(-0.8031)。植被样方主要沿水肥梯度分布; 多样性指数中, 均匀度指数(Pielou index)和丰富度指数(Margelf index)对多功能性的影响作用最大, 均为样方尺度(R² = 0.1871, p < 0.0001; R² = 0.1601, p < 0.0001)小于样地尺度(R² = 0.5921, p = 0.0093; R² = 0.7499, p = 0.0007), 有尺度依赖性; 多功能性在样方和样地尺度上均与物种均匀度呈线性正相关关系, 而与物种丰富度呈单峰曲线关系。该文研究结果表明, 相对于重度放牧和围封, 刈割更有利于维持该地区生态系统的多功能性; 物种丰富度适中且物种分布均匀的生态系统可能有更好的多功能性。     

Relationship between ecosystem multifuntionality and species diversity in grassland ecosystems under land-use types of clipping,enclosure and grazing

Aims Over the past twenty years, most biodiversity and ecosystem functioning (BEF) research has focused on the effects of species diversity on single or just a few ecosystem functions. However, ecosystems are primarily valued for their ability to maintain multiple functions and services simultaneously (i.e. multifunctionality here- after). This paper first introduced the constantly perfected concept of “multifunctionality”, and then tried to make some modifications to the current mainstream quantitative method in order to evaluate the multifunctionality of grassland communities with the management of clipping, enclosure and grazing in Inner Mongolia, investigating the relationship between the multifunctionality and species diversity. Methods In free grazing grassland, four sites were set and each site was divided into two parts to conduct enclosure and clipping management respectively. After seven years, 15 quadrats (1 m × 1 m) were established for each type of management in each site (total 60 quadrats for each type) using the regular arrangement method; as a control, we also established 20 quadrats (two sites) in grazing grassland. For each quadrat, we carried out plants census and collected soil mixture sample, measuring 16 soil variables, and then calculated the biodiversity indices and multifunctionality index (M-index) by means of factor analysis. Important findings The results showed that M-indexes by the two evaluation methods were strongly correlated at both quadrat and site scale, suggesting that our modified method was reliable. Over-grazed communities had the lowest biodiversity indices and their most soil indicators were also low, showing obvious degradation features. Enclosure and clipping communities (seven years) had higher biodiversity and better soil indicators. The rank of M-indexes was clipping community (0.2178) > enclosure community (0.0704) > grazing community (-0.8031). The vegetation was distributed mainly along the gradients of water and fertility. Among the biodiversity indices, evenness (Pielou) index and richness (Margelf) index were most strongly correlated with multifunctionality, and their explanatory power (R²) for M-index were higher at site scale (R² = 0.5921, p = 0.0093; R² = 0.7499, p = 0.0007) than at quadrat scale (R² = 0.1871, p < 0.0001; R² = 0.1601, p < 0.0001), indicating study scale played an important role in the determinants of multifunctionality. At both quadrat and site scales, M-indexes is a linear positive function with species evenness and a hump-shaped function of species richness. Therefore, in contrast to enclosure, clipping was more conducive to maintain the ecosystem multifunctionality in this region, and the ecosystem with moderate specie richness, where these species are evenly distributed might have better multifunctionality.     

基于贝叶斯似乎不相关回归方法的天然蒙古栎生物量模型构建

以胸径和树高作为自变量,基于多元似然分析、似乎不相关回归等方法研建了黑龙江省天然蒙古栎可加性生物量模型系统。结果表明: 树高显著提高了树干生物量模型的效果,决定系数(R²)从0.953提高到0.988,均方根误差(RMSE)减小14 kg,对树枝、树叶和树根生物量的影响并不显著。单变量(仅含胸径)和双变量(胸径、树高)幂函数形式的生物量模型系统的误差结构均为相乘型,表明对数转换后的线性模型形式更合适。树干、树枝、树叶、树根生物量模型的R²分别为0.953～0.988、0.982～0.983、0.916～0.917、0.951～0.952,RMSE分别为13.42～27.03、6.84～7.00、1.95～1.97、9.71～9.84 kg。与广义最小二乘法(FGLS)相比,贝叶斯估计产生了相似的模型拟合效果,却提供了不同变异大小的参数估计值。FGLS各参数标准误为0.054～0.211,而使用Jeffreys不变先验的两种贝叶斯估计方法(DMC和Gibbs1)产生相似的参数变异(标准差为0.055～0.221);使用均值向量为0、方差为1000且协方差为0的多元正态先验(Gibbs2)和使用来自栎属树种生物量模型历史研究汇总的先验(Gibbs3)产生了更大的变异(标准差为0.080～0.278),使用自身数据获取的先验(Gibbs4)估计得到的各参数变异小于其他方法(标准差为0.004～0.013)。当使用Gibbs4法建立模型时,两类模型不仅能提供最窄的95%预测区间,还能产生更小的预估偏差,树干、树枝、树叶、树根和总生物量在单变量模型中的平均绝对偏差百分比(MAPE)分别为19.8%、24.7%、24.6%、29.0%和13.1%,树干和总生物量在双变量模型中的MAPE分别减小到10.5%和9.8%,其他组织MAPE未改变,表明Gibbs4法能提供更准确的生物量预测值。与传统回归方法相比,准确的先验信息使贝叶斯统计在估计稳定性和不确定性减小方面具有优势。     

基于神经网络的马尾松叶绿素含量高光谱估算模型

分析不同生长期的马尾松冠层反射光谱特征与相应叶绿素含量的相关关系.利用36个红边参数逐一筛选,最终确定7个与叶绿素含量相关性较高的红边参数作为光谱特征参数,分别应用逐步分析法与BP神经网络构建叶绿素含量的高光谱估算模型;同样,筛选出4个植被指数作为光谱特征参数,同时,将对原始光谱进行主成分分析降维后的前4个主成分作为BP神经网络的输入变量,分别应用逐步分析法与BP神经网络构建叶绿素含量的高光谱估算模型.结果表明: 将红边参数作为输入变量建立的逐步回归模型和BP神经网络模型的决定系数(R²)分别为0.5205、0.7253,均方根误差(RMSE)分别为0.1004、0.0848,相对误差分别为6.3%、5.7%.将植被指数作为输入变量建立的逐步回归模型和BP神经网络模型的R²分别为0.5392、0.7064,RMSE分别为0.0978、0.0871,相对误差分别为6.2%、6.0%.基于主成分分析的BP神经网络模型的预测效果最好,R²为0.7475,RMSE为0.0540,相对误差为4.8%.     

Biomass estimation models and allocation patterns of 14 shrub species in Mountain Luya,Shanxi, China

Aims Shrub species have evolved specific strategies to regulate biomass allocation among various organs or between above- and belowground biomass and shrub biomass model is an important approach to estimate biomass allocation among different shrub species. This study was designed to establish the optimal estimation models for each organ (leaf, stem, and root), aboveground and total biomass of 14 common shrub species in Mountain Luya, Shanxi Province, China. Furthermore, we explored biomass allocation characteristics of these shrub species by using the index of leaf biomass fraction (leaf to total biomass), stem biomass fraction (stem to total biomass), root biomass fraction (root to total biomass), and root to shoot mass ratio (R/S) (belowground to aboveground biomass).
Methods We used plant height, basal diameter, canopy diameter and their combination as variables to establish the optimal biomass estimation models for each shrub species. In addition, we used the ratios of leaf, stem, root to total biomass, and belowground to aboveground biomass to explore the difference of biomass allocation patterns of 14 shrub species.
Important findings Most of biomass estimation models could be well expressed by the exponential and linear functions. Biomass for shorter shrub species with more stems could be better estimated by canopy area; biomass for taller shrub species with less stems could be better estimated by the sum of the square of total base diameter multiply stem height; and biomass for the rest shrub species could be better estimated by canopy volume. The averaged value for these shrub species was 0.61, 0.17, 0.48, and 0.35 for R/S, leaf biomass fraction, stem biomass fraction, and root biomass fraction, respectively. Except for leaf biomass fraction, R/S, stem biomass fraction, and root biomass fraction for shrubs with thorn was significantly greater than that for shrubs without thorn.     

Effects of grazing intensity and grazing exclusion on litter decomposition in the temperate steppe of Nei Mongol,China

Aims Grazing intensity and grazing exclusion affect ecosystem carbon cycling by changing the plant community and soil micro-environment in grassland ecosystems. The aims of this study were: 1) to determine the effects of grazing intensity and grazing exclusion on litter decomposition in the temperate grasslands of Nei Mongol; 2) to compare the difference between above-ground and below-ground litter decomposition; 3) to identify the effects of precipitation on litter production and decomposition. Methods We measured litter production, quality, decomposition rates and soil nutrient contents during the growing season in 2011 and 2012 in four plots, i.e. light grazing, heavy grazing, light grazing exclusion and heavy grazing exclusion. Quadrate surveys and litter bags were used to measure litter production and decomposition rates. All data were analyzed with ANOVA and Pearson’s correlation procedures in SPSS. Important findings Litter production and decomposition rates differed greatly among four plots. During the two years of our study, above-ground litter production and decomposition in heavy-grazing plots were faster than those in light-grazing plots. In the dry year, below-ground litter production and decomposition in light-grazing plots were faster than those in heavy-grazing plots, which is opposite to the findings in the wet year. Short-term grazing exclusion could promote litter production, and the exclusion of light-grazing could increase litter decomposition and nutrient cycling. In contrast, heavy-grazing exclusion decreased litter decomposition. Thus, grazing exclusion is beneficial to the restoration of the light-grazing grasslands, and more human management measures are needed during the restoration of heavy-grazing grasslands. Precipitation increased litter production and decomposition, and below-ground litter was more vulnerable to the inter-annual change of precipitation than above-ground litter. Compared to the light-grazing grasslands, heavy-grazing grasslands had higher sensitivity to precipitation. The above-ground litter decomposition was strongly positively correlated with the litter N content (R² = 0.489, p < 0.01) and strongly negatively correlated with the soil total N content (R² = 0.450, p < 0.01), but it was not significantly correlated with C:N and lignin:N. Below-ground litter decomposition was negatively correlated with the litter C (R² = 0.263, p < 0.01), C:N (R² = 0.349, p < 0.01) and cellulose content (R² = 0.460, p < 0.01). Our results will provide a theoretical basis for ecosystem restoration and the research of carbon cycling.