Defining and Classifying Ecological Communities

• It is good to write your name on a piece of papers to get ready to do questions

Defining Communities

• Communities can be statistically defined by patterns of association among species
• The Bray-Curtis similarity index can define similarity
• Cluster analysis and ordination can define similarity
• Cluster analysis produces dendrograms
• Gauch(1982) states that ordination primarily represents sample and species relationships as faithfully as possible in a low-dimensional space

Ordination

• Ordination is an exploratory multivariate technique
• Ordination systematically "orders" the objects
• Similar objects are near one another
• Dissimilar objects are far from one another
• Two samples with similar species composition/similarity are considered ecologically similar
• Two samples which share few species/dissimilar characteristics are considered to be ecologically distant
• Relationships among objects are shown along axes
• Relationships among objects are characterized graphically and numerically

Ordination Techniques

• Ordination techniques include distance-based approaches
• Ordination techniques include eigenanalysis-based approaches
• Ordination techniques include informal techniques
• Ordination techniques include direct gradient analysis

Indirect vs Direct Gradient Analysis

• Indirect gradient analysis utilizes only the species by sample matrix
• Any information about the environment is used after indirect gradient analysis, similar to an interpretative tool
• In indirect gradient analysis distance equals dissimilarity
• Species arrange into groups that are most similar
• Direct gradient analysis utilizes external environmental data in addition to the species data
• Direct gradient analysis is a regression technique
• Direct analysis shows if species composition is related to measured variables
• Direct analysis works best if key gradient are measured

Indirect Analysis - NMDS

• NMDS represents Nonmetric multidimensional scaling
• NMDS produces an ordination based on a distance or dissimilarity matrix
• NMDS attempts to represent pairwise dissimilarity between objects in 2D space
• NMDS routines start with random placement of data objects in ordination space
• The goal of NMDS represents the original position of data in multidimensional space (a=3D) as accurately as possible using a reduced number of dimensions (b=2D)
• 'Stress' measures mismatch between the rank order of distances in the data (a) and the rank order of distances in the ordination (b)
• Steps are repeated until 'stress' reaches a minimum
• The final configuration of points may be rotated if desired

Reading a NMDS Plot

• NMDS plots are straightforward: objects closer to one another are likely to be more similar than those further apart
• The scale of the NMDS plot axes is arbitrary
• Solutions with higher stress values (usually above 0.20) should be interpreted with caution
• Solutions with stress above 0.30 are highly suspect
• If a cluster of points is dissimilar from other clusters, the cluster arrangement may be not be meaningful
• Re-running an NMDS with only objects in a cluster reveals more informative patterns

Multivariate Analyses

• Multivariate analyses explain what is seen using one variable at a time
• Considering the weather today, it can only be explained or predicted with one variable?
• Species and ecological assemblages are the result of complex interactions
• Using multivariate approaches helps determine relationships and co relationships (multi-variable hypotheses)
• A model is a summary of knowledge of a system at the time of investigation
• A model helps to predict future events confidently
• A multivariate model shows the influence that types of variability have on a system
• A multivariate model helps better understood variability
• A multivariate model may improve system if required

Univariate vs Multivariate

• Univariate approaches may provide an over simplistic and optimistic assesment of data
• Univariate approaches fail to detect relationships between the variables being studied
• These approaches treat all variables as independent of each other

Regression

• Multivariate data can be used to explain one or more continuous response variables
• A response variable is an abundance of a particular plant species
• Explanatory variables can be soil texture, pH, altitude and if the sample is in direct sunshine or not
• Regression studies and quantifies the relationship between a plant species and variables

Direct Gradient Analysis

• In direct gradient analysis (DGA), species are directly related to measured environmental
• There are two sets of variables
• Species response variables
• Explanatory variables
• DGA can be as simple as a scatterplot of species abundance as a function of position along a measured gradient
• Community data typically have many species and multiple gradients
• A few major gradients that explain much of the variability of the dataset can be extracted to find ecological processes

DGA

• DGA equals Constrained ordination
• Constrained means to compel or force a particular course of action
• Order plots are made along a measured environmental gradient
• The position of the 'dependent' variables is constrained to be a function of environmental variables

Linear vs Unimodal

• A Linear response model is Redundancy analysis
• A Unimodal Response Model utilizes Canonical correspondence analysis
• Similarity under natural conditions is measured over a sufficient range

RDA vs CCA

• The two most commonly used constrained ordination techniques are Redundancy Analysis (RDA) and Canonical Correspondence Analysis (CCA
• RDA is inappropriate under the unimodal model
• CCA is preferred for most ecological data sets since unimodality is common

Canonical Correspondence Analysis

• Canonical Correspondence Analysis is described as a triplot
• CCA includes samples as points
• CCA includes species as points
• CCA includes environmental variables as arrows or points
• Simply put, Canonical Correspondence Analysis is the marriage between CA or PCA and multiple regression
• CCA finds which environmental variables are responsible for structuring the entire community
• CCA finds which gradients are related
• This is shown by angles between arrows
• CCA explains community variation across the gradients

What process to follow

• Select which species shall be for analysis and select environmental parameters
• Transform and or standardise species data
• Determine the appropriate response model
• Select independent variables
• Conduct ordination
• Interprets results