Change in Marine Communities
An Approach to Statistical Analysis and Interpretation, 3rd edition
by
K R Clarke, R N Gorley, P J Somerfield & R M Warwick
(2014)
Introduction and acknowledgements
0.1 Introduction
Third edition The third edition of this unified framework for non-parametric analysis of multivar...
0.2 Acknowledgements
Any initiative spanning quite as long a period as the PRIMER software represents (the first recog...
0.3 Citing this book
Please use the following to cite this book or any of its content: Clarke KR, Gorley RN, Somerfiel...
Chapter 1: A framework for studying changes in community structure
1.1 Introduction
The purpose of this opening chapter is twofold: a) to introduce some of the data sets which are u...
1.2 Univariate techniques
For diversity indices and other single-variable extractions from the data matrix, standard statis...
1.3 Example: Frierfjord macrofauna
The first example is from the IOC/GEEP practical workshop on biological effects of pollutants (),...
1.4 Distributional techniques
Table 1.3. Distributional techniques. Summary of analyses for the four stages. A less condense...
1.5 Example: Loch Linnhe macrofauna
Table 1.4. Loch Linnhe macrofauna {L}. Abundance/biomass matrix (part only); one (pooled) set o...
1.6 Example: Garroch Head macrofauna
describe the sampling of a transect of 12 sites across the sewage-sludge disposal ground at Garr...
1.7 Multivariate techniques
Table 1.5 summarises some multivariate methods for the four stages, starting with three descripti...
1.8 Example: Nutrient enrichment experiment, Solbergstrand
Table 1.7. Nutrient enrichment experiment, Solbergstrand mesocosm, Norway {N}. Meiofaunal abund...
1.9 Summary
A framework has been outlined of three categories of technique (univariate, graphical/distributio...
Chapter 2: Simple measures of similarity of species ‘abundance’ between samples
2.1 Similarity for quantitative data matrices
Data matrix The available biological data is assumed to consist of an array of p rows (species) a...
2.2 Example: Loch Linnhe macrofauna
A trivial example, used in this and the following chapter to illustrate simple manual computation...
2.3 Presence/absence data
As discussed at the beginning of this chapter, quantitative uncertainty may make it desirable to ...
2.4 Species similarities
Starting with the original data matrix of abundances (or biomass, area cover etc), the similarity...
2.5 Dissimilarity coefficients
The converse concept to similarity is that of dissimilarity, the degree to which two samples are ...
2.6 More on resemblance measures
On the grounds that it is better to walk before you try running, discussion of comparisons betwee...
Chapter 3: Clustering methods
3.1 Cluster analysis
The previous chapter has shown how to replace the original data matrix with pairwise similarities...
3.2 Hierarchical agglomerative clustering
The most commonly used clustering techniques are the hierarchical agglomerative methods. These u...
3.3 Example: Bristol Channel zooplankton
perform hierarchical cluster analyses of zooplankton samples, collected by double oblique net ha...
3.4 Recommendations
Hierarchical clustering with group-average linking, based on sample similarities or dissimilari...
3.5 Similarity profiles (SIMPROF)
Given the form of the dendrogram in Fig. 3.3, with high similarities in apparently tightly define...
3.6 Binary divisive clustering
All discussion so far has been in terms of hierarchical agglomerative clustering, in which sample...
3.7 k-R clustering (non-hierarchical)
Another major class of clustering techniques is non-hierarchical, referred to above as flat clust...
Chapter 4: Ordination of samples by principal components analysis (PCA)
4.1 Ordinations
An ordination is a map of the samples, usually in two or three dimensions, in which the placement...
4.2 Principal components analysis
The starting point for PCA is the original data matrix rather than a derived similarity matrix (t...
4.3 Example: Garroch Head macrofauna
Fig. 4.1 shows the result of applying PCA to square-root transformed macrofaunal biomass data fro...
4.4 PCA for environmental data
The above example makes it clear that PCA is an unsatisfactory ordination method for biological d...
4.5 Example: Dosing experiment, Solbergstrand mesocosm
An example of this final point for a real data set can be seen in Fig. 4.2. This is of nematode ...
Chapter 5: Ordination of samples by multi-dimensional scaling (MDS)
5.1 Other ordination methods
Principal Co-ordinates Analysis The two main weaknesses of PCA, identified at the end of Chapter ...
5.2 Non-metric multidimensional scaling (MDS)
The method of non-metric MDS was introduced by and , for application to problems in psychology; ...
5.3 Diagnostics: Adequacy of MDS representation
Is the stress value small? By definition, stress reduces with increasing dimensionality of the ...
5.4 EXAMPLE: Dosing experiment, Solbergstrand
The nematode abundance data from the dosing experiment {D} at the GEEP Oslo Workshop was previous...
5.5 Example: Celtic Sea zooplankton
In situations where the samples are strongly grouped, as in Figs. 5.4 and 5.5, both clustering an...
5.6 Example: Amoco-Cadiz oil spill, Morlaix
Benthic macrofaunal abundances of 251 species were sampled by at 21 times between April 1977 and ...
5.7 MDS strengths and weaknesses
MDS strengths MDS is simple in concept. The numerical algorithm is undeniably complex, but it ...
5.8 Further nMDS/mMDS developments
Higher dimensional solutions MDS solutions can be sought in higher dimensions and we noted previo...
5.9 Example: Okura estuary macrofauna
describe macrofauna samples from the Okura estuary {O}, on the northern fringes of urban Aucklan...
5.10 Example: Messolongi lagoon diatoms
sampled 17 lagoons in E Central Greece for diatom communities (193 species), and also recorded a...
5.11 Recommendations
Non-metric MDS can be recommended as the best general ordination technique available (e.g. ). Im...
Chapter 6: Testing for differences between groups of samples
6.1 Univariate tests and multivariate tests
Many community data sets possess some a priori defined structure within the set of samples, for e...
6.2 ANOSIM for the one-way layout
Fig.6.3 displays the MDS based only on the 12 samples (4 replicates per site) from the B, C and D...
6.3 Example: Frierfjord macrofauna
The rank similarities underlying Fig. 6.3 are shown in Table 6.2 (note that these are the similar...
6.4 Example: Indonesian reef-corals
examined data from 10 replicate transects across a single coral-reef site in S. Tikus Island, Th...
6.5 ANOSIM for two-way layouts
Three types of field and laboratory designs are considered here: a) the 2-way nested case can ari...
6.6 Example: Clyde nematodes (2-way nested case)
analysed meiobenthic communities from three putatively polluted (P) areas of the Firth of Clyde ...
6.7 Example: Eaglehawk Neck meiofauna (two-way crossed case)
An example of a two-way crossed design is given in and is introduced more fully here in Chapter ...
6.8 Example: Mesocosm experiment (two-way crossed case with no replication)
Although the above test may still function if a few random cells in the 2-way layout have only a ...
6.9 Example: Exe nematodes (no replication and missing data)
A final example demonstrates a positive outcome to such a test, in a common case of a 2-way layou...
6.10 ANOSIM for ordered factors
Generalised ANOSIM statistic for the 1-way case Now return to the simple one-way case of page 6.2...
6.11 Example: Ekofisk oil-field macrofauna
studied the soft-sediment macrobenthos at 39 sites at different distances (100m to 8km) and diff...
6.12 Two-way ordered ANOSIM designs
Under the non-parametric framework adopted in this manual (and in the PRIMER package) three forms...
6.13 Example: Phuket coral-reef time series
These data are discussed more fully in Chapters 15 and 16; sampling of coral assemblages took pla...
6.14 Three-way ANOSIM designs
Table 6.4 details all viable combinations of 3 factors, A, B, C, in crossed/nested form, ordered/...
6.15 Example: King Wrasse fish diets, WA
We begin the 3-factor examples with a fully crossed design A$\times$B$\times$C of the composition...
6.16 Example: NZ kelp holdfast macrofauna
We now consider the fully nested design, C(B(A)). In north-eastern New Zealand, examined assembl...
6.17 Example: Tees Bay macrofauna
The final example in this chapter is of a mixed nested and crossed design B$\times$C(A), for a to...
6.18 Recommendations
For typical species abundance matrices, it is much preferable to use a non-parametric ANOSIM-ty...
Chapter 7: Species analyses
7.1 Species clustering
Chapter 2 (page 2.4) describes how the original data matrix can be used to define similarities be...
7.2 Type 2 and type 3 SIMPROF tests
describe in full detail a range of useful SIMPROF tests, which they classify as Types 1, 2 and 3...
7.3 Example: Amoco-Cadiz oil spill
A second example of deriving sets of coherent species curves, this time temporal rather than spat...
7.4 Shade plots
An alternative to line plots, and a technique that can often be even more useful, in terms of the...
7.5 Example: Bristol Channel zooplankton
This example, last seen in Chapter 3, consists of 24 (seasonally-averaged) zooplankton net sample...
7.6 Example: Garroch Head macrofauna
An example where the biotic sample axis could have sensibly been ordered according to an a prior...
7.7 Example: Ekofisk oil-field macrofauna
The 39 sites sampled for benthic infauna at different distances from an oil-field in the N Sea we...
7.8 Species contributions to sample (dis)similarities – SIMPER
Dissimilarity breakdown between groups The fundamental information on the multivariate structure ...
7.9 Example: Tasmanian meiofauna
Another clear generalisation is to a 2-way rather than 1-way layout, illustrated by the 16 meiofa...
7.10 Bubble plots (plus examples)
Bubble plots Abundance (or density, biomass, area cover etc) for a particular species can be show...
Chapter 8: Diversity measures, dominance curves and other graphical analyses
8.1 Univariate measures
A variety of different statistics (single numbers) can be used as measures of some attribute of c...
8.2 Graphical/distributional plots
The purpose of graphical/distributional representations is to extract information on patterns of ...
8.3 Examples: Garroch Head and Ekofisk macrofauna
Plots of geometric abundance classes along a transect across the Garroch Head {G} sewage-sludge d...
8.4 Examples: Loch Linnhe and Garroch Head macrofauna
ABC curves for the macrobenthos at site 34 in Loch Linnhe, Scotland {L} between 1963 and 1973 are...
8.5 Multivariate tools used on univariate data
Ekofisk macrofauna: testing dominance curves Fig. 8.5b compares the averaged community samples fo...
8.6 Example: Plymouth particle-size data
Fig. 8.15 is from Coulter Counter data of particle-size distributions for estuarine water samples...
8.7 Multiple diversity indices
A large number of different diversity measures can be computed from a single data set and it is r...
Chapter 9: Transformations and dispersion weighting
9.1 Introduction
There are two distinct roles for transformations in community analyses: a) to validate statistica...
9.2 Univariate case
For purely illustrative purposes, Table 9.1 extracts the counts of a single Thyasira species from...
9.3 Multivariate case
There being no necessity to transform to attain distributional properties, transformations play a...
9.4 Recommendations
The transformation sequence in a multivariate analysis, corresponding to a progressive downweight...
9.5 Dispersion weighting
There is a clear dichotomy, in defining sample similarities, between methods which give each vari...
9.6 Example: Fal estuary copepods
and present biotic and environmental data from five creeks of the Fal estuary, SW England, whos...
9.7 Variability weighting
describe a similar idea to dispersion weighting for use when the data are continuous biological ...
Chapter 10: Species aggregation to higher taxa
10.1 Species aggregation
Fig. 10.1a repeats the multivariate ordination (nMDS) seen in Fig. 1.7 for the macrofaunal data f...
10.2 Examples
Multivariate examples Nutrient-enrichment experiment In the soft-bottom mesocosms at Solbergstr...
10.3 Recommendation
Clearly the operational taxonomic level for environmental impact studies is another factor to be ...
Chapter 11: Linking community analyses to environmental variables
11.1 Introduction
Approach In many studies, the biotic data is matched by a suite of environmental variables measur...
11.2 Example: Garroch Head macrofauna
For the 12 sampling stations (Fig. 8.3) across the sewage-sludge dump ground at Garroch Head {G},...
11.3 Linking biota to univariate environmental measures (and examples)
Univariate community measures If the biotic data are best summarised by one, or a few, simple uni...
11.4 Linking biota to multivariate environmental patterns
The intuitive premise adopted here is that if the suite of environmental variables responsible fo...
11.5 Further ‘BEST’ variations
Entering variables in groups In some contexts, it makes good sense to utilise an a priori group s...
11.6 Linkage trees (and example)
The idea of linkage trees¶ is most easily understood in the context of a particular example, so F...
11.7 Concluding remarks
For this chapter as a whole, two final points need to be made. The topic of experimental and fiel...
Chapter 12: Causality - community experiments in the field and laboratory
12.1 Introduction
In Chapter 11 we have seen how both the univariate and multivariate community attributes can be c...
12.2 `Natural experiments’
It is doubtful whether so called natural experiments deserve to be called ‘experiments’ at all, a...
12.3 Field experiments
Field manipulative experiments include, for example, caging experiments to exclude or include pre...
12.4 Laboratory experiments
More or less natural communities of some components of the biota can be maintained in laboratory ...
Chapter 13: Data requirements for biological effects studies - which components and attributes of the marine biota to examine?
13.1 Components
The biological effects of pollutants can be studied on assemblages of a wide variety of marine or...
13.2 Plankton and fish
Plankton The advantages of plankton are that: a) Long tows over relatively large distances result...
13.3 Macrobenthos and meiobenthos
Macrobenthos The advantages of soft-bottom macrobenthos are that: a) They are relatively non-mobi...
13.4 Hard-bottom epifauna and hard-bottom motile fauna
Hard-bottom epifauna The advantages of using hard-bottom encrusting faunas, reef-corals etc. are:...
13.5 Attributes and recommendations
Attributes Species abundance data are by far the most commonly used in environmental impact studi...
Chapter 14: Relative sensitivities and merits of univariate, graphical/distributional and multivariate techniques
14.1 Introduction
Two communities with a completely different taxonomic composition may have identical univariate o...
14.2 Examples 1, 2 and 3
Example 1: Macrobenthos from Frierfjord/Langesundfjord, Norway As part of the GEEP/IOC Oslo Works...
14.3 Examples 4, 5, 6 and 7
Example 4: Fish communities from coral reefs in the Maldives In the Maldive islands, compared ree...
14.4 General conclusions and recommendations
General conclusions Three general conclusions emerge from these examples: The similarity in com...
Chapter 15: Multivariate measures of community stress and relating to models
15.1 Introduction
We have seen in Chapter 14 that multivariate methods of analysis are very sensitive for detecting...
15.2 Meta-analysis of marine macrobenthos
This method was initially devised as a means of comparing the severity of community stress betwee...
15.3 Increased variability
noted that, in a variety of environmental impact studies, the variability among samples collecte...
15.4 Breakdown of seriation
Clear-cut zonation patterns in the form of a serial change in community structure with increasing...
15.5 Model matrices & ‘RELATE’ tests
The form of the seriation statistic is simply a matrix correlation coefficient (e.g. equation 11....
15.6 Examples
Example: Tees Bay macrofauna Fig. 15.7 shows the nMDS plot for the inter-annual macrofauna sample...
Chapter 16: Further multivariate comparisons and resemblance measures
16.1 Introduction
To motivate the first method of this chapter look again at the analysis of macrobenthic samples f...
16.2 Matching of ordinations
The BEST (Bio-Env) technique of Chapter 11 can be generalised in a natural way, to the selection ...
16.3 Example: Amoco-Cadiz oil spill
Applying this (BVStep) procedure to the 125-species set from the Bay of Morlaix, a smallest subse...
16.4 Further extensions
Both BEST Bio-Env and BVStep routines can be generalised to accommodate possibilities other than ...
16.5 Second-stage MDS
It is not normally a viable sampling strategy, for soft-sediment benthos at least, to use BVStep ...
16.6 Comparison of resemblance measures
S Tikus Island coral cover The use of second-stage MDS plots can be extended to also include the ...
16.7 Second-stage interaction plots
Phuket coral-reef times series A rather different application of second-stage MDS¶ is motivated b...
16.8 Example: Algal recolonisation, Calafuria
An example of this type (though not a classic BACI situation) is given by , for a study by . Sub-...
Chapter 17: Biodiversity and dissimilarity measures based on relatedness of species
17.1 Species richness disadvantages
Chapter 8 discussed a range of diversity indices based on species richness and the species abunda...
17.2 Average taxonomic diversity and distinctness
Two measures, which address some of the problems identified with species richness and the other d...
17.3 Examples: Ekofisk oil-field and Tees Bay soft-sediment macrobenthos
The earlier Fig. 14.4 demonstrated a change in the sediment macrofaunal communities around the Ek...
17.4 Other relatedness measures
The remainder of this chapter deals only with data in the form of a species list for a locality (...
17.5 ‘Expected distinctness’ tests
Species master list The construction of taxonomic distinctness indices from simple species lists ...
17.6 Example: UK free-living nematodes
examined 14 species lists from a range of different habitats and impacted/undisturbed UK areas (...
17.7 Example: N Europe groundfish surveys
An investigation of the taxonomic structure of demersal fish assemblages in the North Sea, Englis...
17.8 Variation in taxonomic distinctness, $\Lambda ^ +$
VarTD was defined in equation (17.7), as the variance of the taxonomic distances {$\omega _ {ij}$...
17.9 Joint (AvTD, VarTD) analyses
The histogram and funnel plots of Figs. 17.7 and 17.8 are univariate analyses, concentrating on o...
17.10 Concluding remarks on taxonomic distinctness
Early applications of taxonomic distinctness ideas in marine science can be found in for demersa...
17.11 Taxonomic dissimilarity
A natural extension of the ideas of this chapter is from $\alpha$- or ‘spot’ diversity indices to...
17.12 Examples
Example: Island fish species lists Fish species lists extracted from FishBase for a selection of ...
Chapter 18: Bootstrapped averages for region estimates in multivariate means plots
18.1 Means plots
Several examples have been seen in previous chapters of the advantages of viewing ordination plot...
18.2 Example: Indonesian reef corals, S. Tikus
The point is made here in Fig 18.1 for the Shannon diversity of coral community transects (% cove...
18.3 ‘Bootstrap average’ regions
The idea of the (univariate) bootstrap () is that our best estimate of the distribution of values...
18.4 Example: Loch Creran macrobenthos
collected a set of 256 soft-sediment macrobenthic samples along a transect in Loch Creran, Scotl...
18.5 Example: Fal estuary macrofauna
The soft-sediment macrobenthic communities from five creeks of the Fal estuary, SW England, {f} w...
Appendices
Appendix 1: Index of example data
The following is a list of all (real) data sets used as examples in the text, where they are refe...
Appendix 2: Principal literature sources and further reading
A list of some of the core methods papers was given in the Introduction, and the source papers fo...
Appendix 3: Bibliography
Addison & Clark (1990) Addison, R.F. and Clarke, K.R. (1990) ‘Biological effects o...