A comparison of three methods of multivariate analysis of upland grasslands in North Wales

Abstract. In a study of the relationships between vegetation and environment in North Snowdonian grasslands, 166 quadrats on 10 altitudinal transects were analysed with Detrended Correspondence Analysis, Detrended Canonical Correspondence Analysis and Fuzzy Set Ordination. These techniques provide consistent results. DCCA and FSO, both using floristic community data and environmental information, simplify the procedure of vegetation-environment analysis. FSO analyses and synthesizes ecological information; it may yield more reasonable and interpretable results.     

北京八达岭地区灌木群落空间分布格局与土壤环境因子的关系分析

于2007年8月,在北京八达岭地区野外实地踏查的基础上,按海拔梯度每升高100 m布设10个典型灌木样地,共40个样地,分别进行灌木群落调查.采用TWINSPAN方法、依据各样地灌木各物种的重要值对研究区灌木群落进行分类,采用DCCA方法对分类结果进行验证,并选取代表地形因素和土壤因素在内的环境因子共16个指标进行相关分析.结果显示:(1)北京八达岭地区灌木群落共划分为10个类型.(2)北京八达岭地区灌木群落与各环境因子之间有较好的独立性,土壤环境因素对灌木空间分布格局影响较大(特征值为0.90);土壤环境因素中,灌木群落格局与土壤容重因子具有极显著的相关性(P＜0.01).(3)土壤环境因子对灌木群落格局变异的解释能力达到36%;而单纯由地形环境因子的解释能力只有9%,二者交互解释能力为3%,说明土壤环境因素是影响北京八达岭地区灌木群落格局的主导因素. 研究结果表明,北京八达岭地区灌木群落分布格局与土壤容重和海拔因子间存在极显著相关性.     

Influence of ionic strength and conductivity on benthic diatom communities in a tropical river (Monjolinho), S?o Carlos-SP, Brazil

Benthic diatoms are important indicators of ecological conditions in lotic systems. The objective of this study was to elucidate the confounding effects of eutrophication, organic pollution and ionic strength and conductivity on benthic diatom communities. Benthic diatoms and water quality sampling was done at 10 sites during summer base flow period (2008 and 2009). Detrended correspondence analysis (DCA) and canonical correspondence analysis (CCA) were used to determine environmental gradients along which species vary with respect to ionic strength and conductivity and other environmental variables. Using variance partitioning, we assessed the individual importance of a set of environmental variables (eutrophication and organic pollution) versus ionic strength and conductivity on diatom community structure. The effects of ionic strength and conductivity and organic pollution, eutrophication and other environmental variables were integrated into overall resultant benthic diatom communities. Through partial CCA, we partitioned the variance in diatom data between two sets of exploratory variables, i.e. ionic strength and conductivity (26.9%); other variables, particularly eutrophication and organic pollution (23.0%); shared variance (11.3%) and unexplained variance (38.8%). Due to the interaction of the effects of ionic strength and conductivity and other variables in this study, laboratory experiments must be performed to confirm the observed effects of ionic strength and conductivity.     

戴云山黄山松群落与环境的关联

基于戴云山黄山松群落类型31个样地(20 m×20 m)野外调查,筛选4个地形因子和11个土壤环境因子,采用除趋势典范对应分析法(DCCA)探讨黄山松群落分布格局与环境之间的关联,定量分析环境因子间相互关系及其对黄山松群落格局的影响。结果表明:(1)DCCA第一排序轴主要反映黄山松群落的海拔变化,第二轴主要反映坡向变化,即沿第二轴从上到下,坡向越朝向阳坡,黄山松群落分布越明显。(2)DCCA表明第一轴与海拔的相关系数达0.5570,即海拔是黄山松群落分布起着决定性作用的环境因子,呈现海拔越高,黄山松优势种群越显著,在海拔1400-1600 m表现突出。(3)黄山松群落主要物种在DCCA排序图的相对位置,反映坡向、坡位、有机质、人为干扰因素是影响黄山松群落分布的重要因素。(4)DCCA排序图中黄山松群落种类排序轴与环境排序轴的相关系数高于DCA,且消除CCA的 "弓形效应",更能凸显海拔、有机质等环境因子对黄山松群落的影响,即黄山松群落与环境因子之间关联以DCCA排序方法为最佳。     

Vegetation-environment relationships on a species-rich coastal mountain range in the fynbos biome (South Africa)

The species-rich fynbos of the southern Langeberg Mountains, South Africa was studied along three transects (a) to evaluate the compatibility of a floristic classification of the southern Langeberg vegetation with a fynbos biome-wide structural classification of mountain vegetation, (b) to describe the environmental gradients to which the vegetation responds and (c) to investigate the relationship between the vegetation and the abiotic environmental variables which determine the pattern of distribution of the fynbos communities on the southern Langeberg.Principal Components Analysis (PCA) was used to determine correlations between environmental variables independent of vegetation data. Similarities between the 46 communities (determined by floristics) from the three transects were determined using cluster analysis and grouped into 14 higher-level units. Detrended Correspondence Analysis (DCA) was then used for indirect gradient analysis after which Canonical Correspondence Analysis (CCA) was used in a direct gradient analysis of the vegetation with the environmental variables.Compatibility between the floristic and structural classification of the vegetation was analysed. The PCA principal gradient was defined as one from sites with high rock cover, shallow soils and north aspects to those with low rock cover, deeper soils and south aspects. The second gradient is most strongly positively correlated with percentage organic carbon and most strongly negatively correlated with soil clay content. In contrast to the PCA, the DCA showed that the principal gradient is a precipitation gradient, with the response of the vegetation dominated by the change from wet to dry conditions and from low to high winter incoming radiation. The CCA showed that the variation in the mountain habitats to which the vegetation responds can be predicted from a combination of a few environmental variables. The principal gradient was one of change from high to low mean annual precipitation with an opposite change in winter incoming radiation. The second gradient was described by percentage surface rock cover and soil clay content. A simple model using the environmental factors selected in the CCA was proposed for predicting the distribution of floristically determined community groups in the fynbos vegetation of the Langeberg and the southern Cape coastal mountains in general.     

Environmental relationships of vegetation patterns in saltmarshes of central Argentina

We describe vegetation-environment relationships in the saltmarshes of central Argentina. Gradient analysis (Detrended canonical correspondence analysis, DCCA) was performed involving 14 parameters of the groundwater that account for most of the variation in plant communities. We used a stepwise multivariate procedure to classify the vegetation data into 8 clusters, named according to the most abundant characteristic species:Chloris canterai, Cynodon dactylon, Distichlis, spicata, Spartina densiflora andPaspalum vaginatum clusters, containing relevés of tall grassland communities, andAtriplex undulata, Cyclolepis genistoides andHeterostachys ritteriana clusters, containing relevés from scrub. Our interpretation of DCCA ordinations suggests that vegetation pattern is primarily related to a salinity-moisture gradient. There is a strong relationship between vegetation type and the amount of salt in the groundwater and the pattern of its variation during the year. The depth of the groundwater and the conditions of submersion are also related to the compositional variation of the vegetation. Although flooding causes some differences between sites, the most important discriminant variable, and therefore the best predictor of floristic variation, is salinization.     

Drivers of plant species composition in alder-dominated forests with contrasting connectivity

The effects of local and regional environmental variables as well as spatial gradients on the plant species composition of two types of alder-dominated forests (riparian forests and alder carrs) with contrasting connectivity were studied across the Western Carpathians from Hungary through Slovakia to Poland. We used large vegetation (240 sampling plots) and environmental (24 variables) datasets, which were accompanied by spatial variables represented by principal coordinates of neighbour matrices. Canonical correspondence analysis (CCA) of the two datasets revealed 13 and 29 variables with significant effects on variation in species composition of alder carrs and riparian alder forests, which jointly explained 41.2% and 36.4% of the variability, respectively. Altitude was the most important factor explaining 7.7% of the variability in the species composition of alder carrs and 8.2% in riparian alder forests. Variation partitioning in CCA revealed that local variables were crucial drivers for species composition patterns in alder carrs, while spatial processes unrelated to the measured environmental variables shaped the vegetation structure of riparian forests.

     

协惯量分析与典范对应分析在植物群落排序中的应用比较

采用协惯量分析(PCA-CA COIA)和典范对应分析(CCA)两种排序方法, 对北京小龙门林场的黄檗 (Phellodendron amurense)群落进行了分析, 并用Spearman秩相关系数检验了对应排序轴的相关性。两种排序方法得出的结果基本一致, 两者的第一排序轴都反映了海拔高度和坡向对群落分布的影响, 而各自第二、第三排序轴所代表的环境意义有所差异, 并出现了交叉, 但是两者的前3个排序轴均反映了海拔、坡位、土壤厚度和凋落物层厚度的变化趋势, 说明在环境因子个数较少或共线性效应不明显的情况下, 协惯量分析也能达到CCA的分析效果, 并且在排序轴特征值解释量上高于典范对应分析。     

山西高原植被与气候的关系分析及植被数量区划的研究

本文利用去势典范对应分析和数量区划的方法,研究了山西高原植被与气候之间的关系,并进行了数量区划。排序的结果表明：DCCA的第—轴代表山西高原植被和气候梯度的纬向性,热量梯度是决定植被分布最主要的气候因子,水分梯度中的年降水量也对第—轴有较大的影响,由于山西高原南北跨度大,植被与气候因子表现出明显的纬向性;DCCA第二轴代表山西高原植被和气候梯度的经向性,与DCCA第二轴相关性较大的是水分因子中的年降水量、年蒸发量,由于山西高原东西跨度不大,而且大部分地区处在吕梁山脉和太行山脉之间,东西向的气候变化幅度不大,所以植被与气候梯度的经向性不明显。植被数量区划的结果表明：山西高原可划分为17个植被区,用图示的方法确定山西高原大致有三个极点和—个中心。     

排序和广义线性模型与广义可加模型在植物种与环境关系研究中的应用

排序和广义线性模型(Generalized Linear Model,GLM)与广义可加模型(Goneralized Additive Model,GAM)是研究植物种与环境间关系的重要方法。基于线性模型的排序方法应限定于环境梯度较短的植被数据。而基于单峰模型的排序方法更适用于梯度较长的情况。PCA、CA／RA系列和CCA系列是常用的排序方法。同时进行环境数据和植被数据分析的CCA系列,能清楚地得出植物种与环境间的关系。CCA改进后的DCCA和PCCA,是现今较理想的排序方法。GLM和GAM实质上是用环境变量的高阶多项式来拟合植物种与环境变量的关系。GLM和GAM扩展了植物种与环境变量之间的关系模型,能深入地探讨植物种与环境间的关系。GLM主要是模型决定的,而GAM主要取决于原始数据。一般来说,排序能得出研究区域的主要环境梯度,提供了物种聚集和植物群落的概略描述。GLM与GAM对于深入研究单个植物种与环境间的关系具有优势。在实际研究中,两种方法结合使用能互补不足。     

The Finnish forest site type approach: ordination and classification studies of mesic forest sites in southern Finland

A.K. Cajander's forest site type classification system is based on definition of plant communities typical to certain climatical and edaphical site conditions, but the structure and composition of the tree stands in Finland are considered sensitive to random variation and are therefore not used as primary classification criteria. The system has often received criticism, usually that the effects of the tree stand and successional stage of the stand have been underestimated. Most of the present-day forest stands in Finland represent young successional stages and are subjected to intensive management. This should result in an additional difficulty in the application of the forest site types in the field.The present study is based on three independent data sets representing forests on mineral soil in southern part of Finland. TWINSPAN classification, DCA ordination and canonical correspondence analysis (CCA) techniques were applied in successive stages of the data analysis. It was found that the definition of the intermediately fertile, mesic site types was clearly confused by the effects of the tree species and age of the stand. The analyses also revealed that the succession pathways on mesic forest sites are largely determined by the tree species composition. In stands dominated by Pinus sylvestris, the succession follows the competitive hierarchy model, whereas in stands dominated by Picea abies, severe shading of the tree canopy governs the development of understorey vegetation.Abbreviations CCA Canonical correspondence Analysis - DCA Detrended correspondence Analysis - TWINSPAN Two-way indicator species analysis     

Mountain summer farms in Røldal,western Norway –; vegetation classification and patterns in species turnover and richness

This paper discusses vegetation types and diversity patterns in relation to environment and land-use at summer farms, a characteristic cultural landscape in the Norwegian mountains. Floristic data (189 taxa) were collected in 130 4-m² sample plots within 12 summer farms in Røldal, western Norway. The study was designed to sample as fully as possible the range of floristic, environmental, and land-use conditions. Vegetation types delimited by two-way indicator species analysis were consistent with results from earlier phytosociological studies. Detrended correspondence analysis and canonical correspondence analysis show that rather than being distinct vegetation types, the major floristic variation is structured along a spatial gradient from summer farm to the surrounding heathland vegetation. Species richness (alpha diversity) was modelled against environmental variables by generalized linear modelling and compositional turnover (beta diversity) by canonical correspondence analysis. Most environmental factors made significant contributions, but the spatial distance-to-farm gradient was the best predictor of both species richness and turnover. While summer farms reduce mean species richness at the plot scale, the compositional heterogeneity of the upland landscapes is increased, thereby creating ‘ecological room’ for additional vegetation types and species. Within an overall similarity across scales, soil variables (pH, base saturation, LOI, phosphate and nitrogen) differed considerably in their explanatory power for richness and turnover. A difference between ‘productivity limiting’ factors and ‘environmental sieves’ is proposed as an explanation. Species turnover with altitude is relatively low in grasslands as compared to heaths.     

Vegetation composition and its relationship with the environment in mallines of north Patagonia,Argentina

“Mallines” are characteristic Patagonian wet meadows. The objectives of this study were to describe plant community composition in the main mallines in northern Patagonia and to determine the influence of selected environmental variables on the distribution of vegetation. Fifty-two sites were selected for vegetation surveys and measurements of water table (WT) depth, soil pH, electric conductivity (EC), and mean annual precipitation. We performed cluster analysis for vegetation classification and correspondence analysis (CA) and canonical correspondence analysis (CCA) for vegetation ordination. Plant composition was mostly related to both environmental variables and longitude and that it was not possible to disentangle the two (i.e. the vegetation was spatially structured). We defined three plant communities that differed along two main environmental gradients. The main gradient operates on a regional scale and is determined, from west to east, by a decrease in mean annual precipitation and an increase in the depth of the WT, soil pH, and EC. The secondary gradient operates on a site scale and is determined by topographic features inside the mallín and their influence on the hydrological regime (increasing moisture from the border towards the center). This second gradient allowed us to distinguish two plant communities, one of wet characteristics in the centers of the mallín, and another of mesic characteristics along the borders of the mallín.     

Relationships between vegetation and climate on the Loess Plateau in China

The Loess Plateau is one of the most environmentally sensitive regions in China. This study addresses the relationships between vegetation and climate of this area quantitatively at a large-scale, in order to determine the factors that control vegetation distribution. The Loess Plateau, located at 101°01′–155°10′ E and 34°02′–40°40′ N, covers an area of 52 million hectares. Vegetation data were collected from the vegetation map (1:500,000) and the Landsat Thematic Mapper scenes of the Loess Plateau. The Loess Plateau was divided into small districts of 30′ latitude by 30′ longitude on the vegetation map. In each district, areas with different vegetation were measured and used as vegetation data. The climatic data were average values of county meteorological records in each district in the past 25 years. GIS, TWINSPAN and canonical correspondence analysis (CCA) were employed for analysis. 257 small districts were clustered into 7 groups using TWINSPAN, representing 7 vegetation regions or subregions. The first three CCA axes had significant correlations with climate. The first CCA axis represented the variation of vegetation and climate along the latitude gradient, while the second CCA axis the variation along the longitude gradient. The distribution pattern of 171 vegetation formations on the CCA plot is identical to that of vegetation regions (districts). The spatial distribution of vegetation is closely related to climate variables on the Loess Plateau. Water variables and temperature are important in both latitude and longitude gradients, while the sunshine hours, accumulated temperature and wind speed are more important than water variables and temperature in longitude gradients.     

Improving the interpretation of fuzzy partitions in vegetation science with constrained ordinations

Classification and ordination techniques based in fuzzy set theory are now being commonly used in vegetation studies. However, several problems have been detected in spite of the significant theoretical advantages of the theory. In this paper we have improved the interpretability of fuzzy partitions by combining fuzzy partitions with correspondence analysis (CA) and detrended canonical correspondence analysis (DCCA) in an analysis of the beech forests of Basque Country, northern Spain. Our results seem to overcome difficulties in the interpretation of multi-group partitions.     

红花尔基地区沙地樟子松群落及其与环境关系研究

通过野外调查,采用双向指示种分类（TWINSPAN）和除趋势典范对应分析（DCCA）方法。对内蒙古红花尔基地区沙地樟子松群落进行了分类和排序,并根据DCCA排序结果对樟子松群落和物种空间分布格局及其与环境因子之间关系进行定量分析。结果表明,应用TWINSPAN方法将该区沙地樟子松植被划分为6个类型。DCCA分析表明,海拔高度、地貌类型、群落盖度、土壤总碳含量、土壤总氮含量、乔木胸面积等因子对植物群落和物种分布格局影响明显。在DCCA排序图上,樟子松群落及物种的空间分布呈明显的聚集格局,可划分出不同的类群,并反映与环境因子之间具有密切的关系。对各类环境因子解释植被分布格局的作用进行了定量分解,指出地形因素是解释作用最强的变量,对未能解释部分的原因进行了分析。     

Development of plant communities after restoration of the Antaibao mining site,China

To investigate the dynamic changes in the artificial vegetation in an abandoned mining site,we analyzed the relationships among community types,environmental variables and community structure in the process of vegetation restoration in the Antaibao mining site,China by survey of the communities and use of biological dating methods.By means of the quantitative classification method (two-way indicator-species analysis,TWINSPAN) and the ordination technique (de-trended correspondence analysis,DCA; and de-trended canonical correspondence analysis,DCCA),the plant communities were classified into seven groups:community Ⅰ,Robinia pseudoacacia + Pinus tabulaeformis-Caragana korshinskii-Agropyron cristatum; community Ⅱ,Robinia pseudoacacia-Hippophae rhamnoides-Artemisia capillaries; community Ⅲ,Ulmus pumila-Elaeagnus angustifolia-Artemisia capillaries;community Ⅳ,Caragana korshinskii-Agropyron cristatum+Artemisia capillaries;community Ⅴ,Hippophae rhamnoides-Elymus dahuricus;cornrnunity Ⅵ,Elaeagnus angustifolia+Hippophae rhamnoides-Brassica jucea;community Ⅶ,Hippophae rhamnoides+Elaeagnus angustifolia-Salsola collina.We conclude that the community types and diversity are mainly influenced by the succession time and the soil organic matter content.The forest community is more adaptable to the special inhabitation than the shrub community.     

Floristics and environmental relations in two extreme desert zones of western Egypt

• 1 The present study provides quantitative estimates of soil, vegetation structure and species distribution in 144 stands along two transects in two extreme desert zones of western Egypt. This study had two aims: (1) to describe the floristic composition of the accidental type of vegetation growing in parts of the Western Desert of Egypt; and (2) to analyse the distribution of species in relation to certain environmental factors by multivariate data analysis.
• 2 Eight environmental variables were recorded: salinity, pH, calcium carbonate, soil moisture, organic matter content, and percentages of sand, silt and clay. A total of 60 species belonging to 19 families of the angiosperms and Ephedra alata was recorded. Chamaephytes and therophytes were the most frequent, denoting a typical desert life‐form spectrum, mainly of Saharo‐Arabian distribution.
• 3 The samples were classified into eight twinspan groups. Detrended correspondence analysis (DCA) showed that these groups could be distinguished by the first two DCA axes. They could also be effectively interpreted and explained with canonical correspondence analysis (CCA). Based on forward regression, salinity, fine sediments, organic matter and soil moisture content related closely to the first three canonical axes, and accounted for 84.1% of the species–environment relationship among the stands.
• 4 Both classification and ordination resulted in a clear demonstration of the vegetation pattern in the study area. Other twinspan groups were characterized in addition to those identified previously. In conclusion, the study area has a transitional character where the Nubian Desert and the Northern Mediterranean vegetation meet.

     

Canonical correspondence analysis of early volcanic succession on Mt Usu,Japan

Canonical correspondece analysis (CCA) was applied to explore revegetation patterns during early succession on Mt Usu. Vegetation was buried by deposits of ash and pumice from 1 to 3 m in depth from the 1977–78 eruptions. Three habitats were selected: tephra, tephra in gully and original surface. Plant density and plant cover data were analyzed separately. Environmental factors consisted of five quanticative variables (organic matter, elevation, distance from colonizing source, erosion and deposition of volcanic deposits) and three nominal variables (habitat types: tephra, tephra in gully and original surface). Canonical correspondence analysis showed that the original surface played a special role in vegetation development because the old topsoil supplied both nutrients and seed-bank species. The CCA also suggested that the environmental factors that influence plant density and cover differ. Distance from colonizing source affected plant density while erosion affected cover. Using CCA, factors could be distinguished that influenced seedling establishment from vegetation expansion and vegetation recovery dynamics could also be more clearly interpreted.     

Structure of herbaceous plant assemblages in a forested riparian landscape

We assessed patterns of herbaceous and woody species richness, plant-environment interactions, and correspondence between the herb and tree layer in a riparian landscape (the Ozark National Scenic Riverways, Missouri, USA). A total of 269 herb and 70 tree species were identified on 94 sample plots. Gradient analysis revealed that environmental variables and vegetation were influenced by a strong elevation gradient. However, high variability in environmental variables (pH, elevation, slope, sand, clay, organic matter) indicated a high level of substrate heterogeneity across the riparian landscape. We were unable to predict the composition of the herb understory from the canopy trees with any detailed accuracy and no clear characterization of herb species assemblages was found using cluster analysis or ecological land type (ELT) classifications. Canonical correspondence analysis (CCA) results for both tree and herb plots showed that elevation (height above river) and pH were the dominant environmental gradients influencing vegetation patterns on the first CCA axis while soil particle size exhibited the strongest correlation with the second CCA axis. Secondary gradients of importance included slope, soil container capacity, and organic matter. No significant linear or quadratic correlation was found between elevation and herb or woody species richness. Environmental variables alone or in combination, were weak predictors of herb and woody species richness, despite the patterns observed in the gradient analysis and the correlations observed in the CCA results. Ecotonal analysis showed that the herb layer exhibited a high species replacement rate at the lower elevations most susceptible to flooding (0–3 m). Above the flooding zone, there was more or less continuous species replacement, suggesting the presence of a gradual ecotone/ecocline. The tree layer exhibited much stronger discontinuities than the herb layer in the lower elevations along the height gradient (0–10 m). Recognizing the limitations of classification techniques for riparian herb assemblages and the importance of scale and heterogeneity in vegetation layers is especially important in light of mandates to preserve, protect, and manage for plant diversity.